Вы находитесь на странице: 1из 246

UNIVERSITI PUTRA MALAYSIA

EVALUATION OF TECHNIQUES FOR DETERMINATION OF


SATURATED HYDRAULIC CONDUCTIVITY IN THE VADOSE ZONE

ABDOLHAKEM O MOHAMED.

FK 2004 53

EVALUATION OF TECHNIQUES FOR DETERMINATION O F


SATURATED HYDRAULIC CONDUCTIVITY IN THE VADOSE ZONE

BY
ABDOLHAKEM 0 MOHAMED

Thesis Submitted to the School of Graduate Studies, Universiti Putra


Malaysia in Fulfilment of the Requirements for the
Degree of Master of Science

June 2004

Abstract of thesis presented to the Senate of Univeresiti Putra Malaysia in fulfilment


ofthe rcquirement for the d e s e e of Master of Sciencc
EVALUATION OF TECHNIQUES FOR DETERMINATION OF
SATURATED HYDRAULIC CONDUCTIVITY
IN THE VADOSE ZONE

ABDOLHAKEM 0 MOHAMED
June 2004

Chairman: Professor Ir. Mohd. Amin Mohd. Soom, Ph.D.


Faculty:

Engineering

Saturated hydraulic conductivity of a soil (Ks) is a measure of a soil's ability to


transmit water in a water-saturated state. Infiltration, drainage, and groundwater
pollution are strongly influenced by the magnitude and spatial distribution of the
vadose zone field saturated soil hydraulic conductivity (Kfs). There are numerous
methods of estimating K, rangmg from direct measurement in the laboratory or in
situ to models that use only basic soil data (e.g. soil textural classes, bulk density,

Db, organic matter, OM, or porosity, E,). However, the results from different
measuring techniques vary under different field conditions. In this study of
Serdang Series soils found in the Universiti Putra Malaysia (UPM) campus, soil K,
values were collected at different depths using three direct methods. Estimation of

K, were done using six empirical models. The direct methods were in situ
techniques of Guelph Permeameter (GP) and double ring infiltrometer (DRI), and
constant head pcrrneameters (SCHP), a laboratory technique on intact soil cores

extracted from the same site at different depths. Predictive models included models
of Cosby et al. ( 1 984); Brakensiek et al (1984); Saxton et a1 (1986); Vereccken et
al. (1990); Sabro (1992) and Amin et al. (1997). In this study of K, in the vadose
zone, the focus was towards comparison of measurements in the field to those of
extracted samples from the same site, but determined by laboratory testing, under
controlled condition, and those estimated from empirical models. In addition, a
model was developed for determining K, values based on seven basic soil
properties (sand, silt, clay, Db, moisture content (MC), E and OM). The results of
the comparison showed that the geometric mean of K, values obtained by the three

~ 1.315 x 1 0-2cm
experimental methods varied from 7.333 x 1o - to

S-I

(6.34 x 1 o-'

m I day to 11.36 m 1 day). The GP method yielded the widest range from 7.333 x
1 0-8to 1.654 x 1 0-3cm s-' while the SCHP yielded the narrowest range from 4.4 x
10" to 1.315 x

cm s-'. Geometric mean K, values were 27 to 360 times greater

for the SCHP compared to the GP method and were significantly different at all
depths. Measurements of Ks for the soil under consideration indicate that the DRI
and GP methods provided reasonable similar values at the topsoil layer (0-1 5 cm).
While the geometric mean Ks values measured by the DRI method was statistically
different from those obtained by SCHP method at 0-1 5 cm depth.

The laboratory technique yielded greater standard deviation (SD) at the 30 cm and
60 cni depths. Some soil cores may have more macropores than others. whereas the
coefficient of variation values were greater for the GP method. The GP produced
in situ calculation of Kfs in a relatively short time (25 to 90 minutes for a single
measurement) compared to DRI (1 20-1 80 minutes) and SCHP (1 500-1 660
minutes).

The results of the multiplc regression analysis indicated that the significant Inter-

correlations limitcd the numbcr of useful functional relationships that could be


der~vedfrom the seven variables (Textural classes, Dh, MC, E, and OM). The
results of regress~onfor full data set showed that only simple function based on silt
content and OM gave a si~gnificantrelationship with K, at 0.05 level, but only 10.5
n/o

of variability in K1, was explained by those variables. There was a sipificant

relationship between K, and the input variables at each depth. These relationships
however were different at each depth. The best models found from this study at
depth of 0- 15 cm, have silt. sand, E, and MC; at depths of 15-30 cm have silt. sand,
and E;at depths of 30-60 cm have clay, sand, OM, and MC: and at depths of 60-90
cm silt. Db and E with values of R'= 0.57, 0.50, 0.41 and 0.74, respectively.

In this study the geometric mean error ratio (GMER) and geometric standard
deviation error ratio (GSDER) were used to evaluate the applicability of the
selected empirical models. The results showed that model o f Amin et a1 (1997)
produced noticeably best results with GMER closest to 1 (0.54) and the lowest
GSDER (7.64) of the models tested here. This is followed by the Jabro (1992)
model with GMER (0.43) and GSDER (10.22), then Brakensiek et a1 (1984) with
GMER (0.43) and GSDER (15.6). It consequently appeared, at least for this soil
(Serdang Series), that of the six models compared in this study. the Amin et a1
model was the model of choice for the prediction of K,. The second best model was
labro model whereas the model of Brakensiek et al. ranked third.

%fif'USTAKAAN U T A F I

WVE4lW WTRA L
U
U
W

Comparison between the methods was hampered by a number of factors. It was


difficult to discriminate between spatial variables of Ks and errors related to the
methods. Different sample volumes and sample numbers were used. Comparisons
made between different K, measurements in the field are subject to natural soil
variations that may be larger than the differences between methods. Findings of
this study can be used as a guideline for application of these methods particularly
to the same soil type and depth setup. The correct use of any of these methods for
one of the most extensive and productive soils in Selangor (Serdang Series) could
be highly beneficial to the agricultural sector.

Abstrak tesis yang dikcmukakan kepada Senat Universiti I'utra Malaysia sebagai
memenuhi keperluan untuk 1-jazah Master Sains

PENINAIAN TEKNIK UNTUK MENENTUKAN KEBERKONDUKAN


HIDRAUL TEPU DALAM ZON VADOS
Oleh

ARDOLHAKEM O MOHAMED
Jun 2004

Pengurusi:

Profesor Ir. Mohad. Amin Mohd. Soom, Ph.D.

Fakulti:

Kejuruteraan

Ketertelapan tepu tanah (K,) adalah ukuran kemampuan tanah untuk mengalirkan
air dalam keadaan tepu. Penyerapan, saliran dan pencemaran air bawah tanah
sangat dipengaruhi oleh magnitud dan taburan spatial dilapaugan ketertelapan tepu
tanah (K,,) dalam zon vados. Ada beberapa kaedah bagi menganggarkan K,
antaranya ukuran terus dalam makmal atau di situ dan menggunakan data asas
bahan organik (OM) dan Keliangan tanah
tanah seperti tekstur, ketumpatan (Db),

(E). Walau bagaimanapun. keputusan yang diperolehi adalah berbeza hasil


daripada perbezaan teknik pengukuran dan perbezaan keadaan lapangan. Dalam
kajian ini, ujikaji dan ramalan K, siri tanah Serdang yang terletak di kampus UPM
telah dibuat mengikut kedalaman tiga kaedah term dan enam model ramalan.
Kaedah terus n~engikutGuelph permeameter (GP), infiltrometer gegelung kembar

(DRI) suatu teknik di situ, dan permeameter turus tetap (SCHP), suatu teknik
makmal ke atas teras tanah jrang tidak diganggu dan diperolehi di kawasan yang
sama pada kedalaman yang berbeza. Sementara model ramalan termasuk model

Cosby et al. (1984), Brakensiek et a]. (1984), Saxton et al. (1986), Vereecken ct al.
( 1990), Jabro et a]. ( 1992) dan Amin et al. ( 1997). Objektif utama kajian kc dalam

zon vados ini ialah untuk membandingkan K, di lapangan yang disctkan dcngan
sampcl yang diperolehi di lapangan yang sama, tctapi ditentukan oleh u-jian
makmal, dan juga

ramalan model cmpirikal. Selain daripada itu. untuk

membangunkan sebuah model bagi rnenentukan nilai K, berdasarkan tujuh data


asas tanah (pasir. kelodak, tanah liat, ketumpatan pukal, kandungan kelembapan,
keliangan dan bahan organik). Keputusan perbandingan uj i kaji menunjukkan nilai
purata geometrik K, oleh tiga kaedah eksperimen berbeza dari 7.333 x
1.3 15 x 10

' sm s-' (6.34 x 10"

hingga

m/hari hingga 1 1.36 mlhari). Kaedah GP pula

menghasilkan jarak nilai paling kecil iaitu 7.33 x lo-' hingga 1.654 x 1 o4 sm s",
sementara SCHP jarak nilai paling besar iaitu dari 4.4 x 10" hingga 1.31 5 x 10

'

sm s-I. Purata nilai geometrik Ks adalah 27 hingga 360 kali lebih besar bagi SCHP
berbanding kaedah GP dan adalah berbeza secara bererti untuk semua kedalaman.
Kajian ini juga menunjukkan nilai K, yang diukur oleh kaedah DRI adalah tidak
berbeza secara statistik dengan nilai yang diperolehi oleh kaedah GP pada
kedalamam 0- 15 sm tetapi berbeza secara statistik dengan yang diperolehi daripada
kaedah SCHP.

Teknik makmal menghasilkan sisihan piawai (SD) yang lebih besar pada
kedalaman 30 dan 60 sm. Kemungkinan ada teras tanah mempunyai lebih banyak
rongga daripada yang lain, sebaliknya nilai pekali perbezaan (CV) adalah lebih
besar untuk kaedah GP. Kaedah GP menghasilkan anggaran pengiraan Kl., di situ
dalam masa yang singkat secara relatif

(35 hingga 90 ininit untuk satu

pengwkuran) berbanding dengan DRI ( 1 20- 1 80 minit) dan SCHP ( 1 500- 1660
minit).

Kcputusan analisis lebi h daripada satu regresi lnenunjukkan keberertian saling


pcrhubungan rnenghadkan nilai fungsi perhubungan yang mungkin timbul daripada
tujuh pembolehubah (kelas tekstur, Dh, MC, E, dan OM). Keputusan regresi untuk
semua set data menunjukkan hanya satu fungsi mudah berdasarkan kandungan
kelodak memberikan hubungan bererti dengan Ks pada tahap 0.05 tetapi hanya
10.5 % daripada keberubahan dalam Kl., yang diperihalkan oleh keberubahan itu.
Keputusan analisis regresi juga menunjukkan ada keberertian perhubungan antara
K, dan input keberubahan setiap kedalaman. Perhubungan ini walau bagaimanapun
adalah berbeza inengikut kedalaman. Model terbaik yang ditemui dalan kajian ini
pada kedalaman 0-15 sm, mempunyai kelodak, pasir, E dan MC; pada kedalaman
15-30 sm mempunyai kelodak, pasir, dan E, pada kedalaman 30-60 sm mempunyai
tanah hat, pasir, OM, dan MC;pada kedalaman 60-90 sm mempunyai kelodak, Db
dan E dengan masing-masing nilai ~ ' = 0 . 5 7 ,0.50, 0.41 dan 0.74.

Dalam kajian ini purata kadar ralat geometri (GMER) d m kadar ralat sisihan
piawai (GSDER) digunakan untuk inenilai kebolehgunaan kaedah model ramalan
yang telah dipilih. Keputusan kajian ini menunjukkan model Amin et al. (1997)
menghasilkan keputusan terbaik dengan GMER menghampiri 1 (0.54) dan GSDER
(7.64) terendah diikuti oleh model Jabro (1992) dengan GMER (0.43) dan GSDER
(1 0.22) dan seterusnya Brakensiek et al. ( 1 984) dengan GMER (0.43) dan GSDER
(15.6). Dalan kajian ini setelah perbandingan dibuat ke atas enam model untuk
tanah siri Serdang, model Amin adalah model yang dipilih untuk menganggarkan

nilai K,. Model kedua terbaik adalah model Jabro sementara model Brakensiek
adalah yang kctiga.

Pcrbandingan antara kaedah tclah dihalang oleh beberapa factor. Adalah susah
untuk mernbezakan keberubahan K, secara spatial dan ralat yang berkait dengan
kaedah yang digunakan dan perbezaan isipadu sampcl serta bilangan sampel.
Perbandingan dibuat antara perbezaan ukuran dalam lapangan adalah bergantung
kepada perbezaan semulajadi tanah yang mungkin lebih besar daripada perbezaan
antara kaedah. Keputusan daripada kajian ini memberikan panduan awal
menggunakan kaedah-kaedah tersebut khasnya pada tanah yang sama dan
kedalaman yang ditentukan. Kaedah yang paling sesuai untuk menentukan K, bagi
sejenis tanah yang paling prodiktif di sekitar Sin Serdang amatlah berguna kepada
sektor pertanian.

ACKNOWLEDGEMENTS

First and foremost, all praise be directed to Allah (SWT) for making all
things possible, Alhamdulillah.

I would like to express my sincerest thanks to my supervisor, Professor Ir.


Dr. Mohd. Amin Mohd. Soom, for his invaluable help, guidance, advice, support
and encouragement throughout this work. I also wish to thank Dr. Abdul Aziz
Zakaria and Associate Professor Kwok Chee Yan for serving as members of the
supervisory committee and for their guidance and valuable suggestions.

Special thanks are extended to Mr. Ghazali Kassium (technicians of Soil


and Water lab) for his constant assistance. I wish to express my gratitude to my
friends in the faculty of engineering for their help and support, especially
Associated Prof. Dr. Abdel Maged Hamuda , Mr Meh Awang, Mr. Johari , and
Mr Tajul Ariffin Tajuddin . Special thanks are extended to my best friends Dr.
Radim Dadang and Dr. Ahmed Ganfoud for their support.

I am grateful to the People of Libyan Arab Jamahiriya and Omer Almukhtar University who provided me the scholarship for pursuing the Master
degree at Universiti Putra Malaysia. Also, I would like to express my gratitude to
Univeristi Putra Malaysia for some financial support to the study.

I would like to express my heartfelt thanks to all members of my family for


their encouragement and overwhelming support to complete this study.

I certify that an Examination Committee met on 14Ih June 2004 to conduct the final
examination of Abdolhakem 0. Mohamed on his Master of Science thesis entitled
"Evaluation of Techniques for Determination of Saturated Hydraulic Conductivity in
the Vadose Zone" in accordance with Universiti Pertanian Malaysia (Higher Degree)
Act 1980 and Universiti Pertanian Malaysia (Higher Degree) Regulations 1981. The
Committee recommends that the candidate be awarded the relevant degree. Members
of the Examination Committee are as follows:

Ir. Lee Teang Shui, Ph.D.


Associate Professor
Faculty of Engineering
Universiti Putra Malaysia
(Chairman)

Ir. Mohd. Amin Mohd. Soom, Ph.D.


Professor
Faculty of Engineering
Universiti Putra Malaysia
(Member)

Abdul Aziz Zakaria, Ph.D.


Faculty of Engineering
Universiti Putra Malaysia
(Member)
Kwok Chee Yan
Associate Professor
Faculty of Engineering
Universiti Putra Malaysia
(Member)

ProfessodDeputy ~ e g n
School of Graduate Studies
Universiti Putra Malaysia
Date:

2 6 AUG 2004

This thesis submitted to the Senate of Universiti Putra Malaysia and has been
accepted as fulfilment of the requirements for the degree of Master of Science. The
members of the Supervisory Committee are as follows:

Ir. Mohd. Amin Mohd. Soom, Ph.D


Professor
Faculty of Engineering
Universiti Putra Malaysia
(Chairman)
Abdul Aziz Zakaria ,Ph.D.
Senior Lecturer
Faculty of Engineering
Universiti Putra Malaysia
(Member)

Kwok Chee Yan


Associate Professor
Faculty of Engineering
Universiti Putra Malaysia
(Member)

AIM IDERIS, Ph.D.


Professor 1 Dean
School of Graduate Studies
Universiti Putra Malaysia

Date:

xii

1 0 SEP 2004

DECLARATION
I hercby declare that the thesis is based on my original work exccpt for quotations
and citations which have been duly acknowledged. I also dcclarc that it has not
bccn previously or currently submitted for any other d e g c e at Univcrsiti Putra
Malaysia or othcr institutions.

Date:

56

TABLE OF CONTENTS
Page

ABSTRACT
ABSTRAK
ACKNOWLEDGEMEKTS
APPROVAL
DECLARATION
LIST OF TABLES
LIST OF FIGURES
LIST OF PLATES
LIST OF SYMBOLES
CHAPTER

INTRODUCTION
General
Statement of the Problem
Objectives of the Study
Thesis Organisation
LITERATURE REVIEW
Introduction
In Situ Vadose Zone Methods
Guelph Perrneameter
Double Ring Infiltrometer, DRI
Laboratory Method, SCHP
Comparison of Methods
Empirical Models
Theoretical considerations of the SCHP, DRI, and
GP methods
Laboratory method using SCHP
Ring Infiltrometer method, DRI
Guelph Permeameter method, GP
Factors Affecting Soil K- Values
Soil Texture and Structure
Soil Heterogeneity
Smearing, remolding and siltation
Air Entrapment
Capillarity
Solution analysis methods
Summary

x
xi

...

Xlll

xvii
xix
xxi
xxii

I11

MATERIALS AND METHOD


Site Description
Experimental Design
Determination of Soil Properties
Determination of Soil Water Moisture Content
Soil Texture
Bulk Density
Calculation of Total Soil Porosity
Determination of Organic Matter (OM)
Determination of Saturated Hydraulic Conductivity
Guelph Permeameter (GP)
Double Ring Infiltrometer (DRI)
Determination of Ks Using SC-HP
Ks Estimation Methods
Jabro Model
Saxton et a1 Equation
Vereecken et a1 Equation
Cosby et a1 Equation
Brakensiek et a1 Equation
Amin et a1 Equation
Statistical Analysis
Statistical Distribution
Comparison Method
RESULTS AND DISCUSSION
Field Site Characteristics
Results of Soil Properties and Infiltration Rate
Soil Moisture Content
Organic matter
Bulk Density
Porosity
Sand
Silt
Clay
Soil Texture
Infiltration Rate
Results of Soil Saturated Hydraulic Conductivity, Ks,
Model Analysis and empirical Models Evaluation
Soil saturated hydraulic conductivity
Statistical Distribution of Ks parameter
Double Ring Infiltrometer
Constant Head Method
Guelph permeameter
Failure and Time Requirement
Comparison of the Methods
K, Model Analysis
Empirical Models Evaluation

V1

SUMMARY AND CONCLUSION

REFERENCES
APPENDICES
BIODATA OF THE AUTHOR

xvi

LIST OF TABLES

Table
2.1

Brief description ol'method for measuring the soil satusatod


hydraulic conductivity. l'ypc. applicability. main direction o l ' h
measurement and approximate sample volume.

2.2

Hydraulic conductivity values of sat~lratedsoils.

2.3

Classification of soil hydraulic conductivity values.

2.4

Approximate relationships between texture. structure. and


hydraulic Conductivity.

3.1

Porous media categories used for estimating a* in


the single-head well permeameter analysis.

4. 1

Soil physical properties of the study area. OM organic matte.


MC moisture content, E porosity and Db bulk density

4.2

Physical properties of Serdang Series soil at the experimental


site across at each depth
Results of Infiltrometer measurement and predictioli

4.4

Equations of infiltration rates and cumulative infiltration

4.5

Descriptive statistics of parameters of Serdang Series soil


measured at the experimental site across all depths.

4.6

The total numbel- of observation. percentage of f a i l ~ ~ratio


re
(FR), percentage of negative KI, values and time requircnient
to complete one set of measurement

4.7

Comparison of K,measurements obtained using the double sing


148
infiltrometer (DRI). Guelph permeanieter GP). and intact soil corc
(SCHP) methods K\Gnlis geometric mean K, valuc: K,I\IIIli 4 the
mininiuni K, value. K,M,I, is the ~ i i a x i m ~K,
~ m\ al~le:SD 1s tlic
standard deviation: CV is the coefficient of variation. ncgatibc
value . and N is the nuniber of mcas~~rements.

4.8

Correlation n~atsisof parameter nieas~~red


at study area

4 .

Derived regression models where Ks-Kli in cni s" arc measuscd

102

saturated hydraulic conductivity using constant head pern1canictcr


(SCI-II') and Guelpli ermeametcr (GP) methods respectively. Dl,
bulk density in g cm- . I=, porosity %I. MC moisture content %,

4.10

Summary of'statistics ol'saturated hydraulic conductivity


for Serdang Series. sandy clay loam soil. obtained bl threc
methods and predicted by the six selected modcls mcasurc

4.1 1

Geometric mean crror ( G M E R ) and geometric standard deviation


of error ratio (GSDER) calculated with six selected models using
the data set of the Serdang Series soil of study area comparcd to
measurements val ucs.

sviii

170

LIST O F FIGURES

2.1

Inliltration ratc ol'dry and wet soil.

2.2

Schematic tcst setup for constant head and lalling head lest.\;

2.3

Approximation of steady state flow out of a well situutcd in a


homogenous isotropic porous medium.

2.4

Illustrates the constant head borehole test (GP) in vaciose


(~lnsaturated)zone using two heads. 14 1 and H2. Whcre. a is
radius of the well and G P represents Guelph Permearnetel-.

3.1

Soil map showing the location of Serdang series at U P M


campus.

3.2

Contour map of the Field Research Area Faculty of Engineering


showing the stud!. area (Basin irrigation area)

3.3

Syninietric plan showing the location of saturated hydraulic


conductivity test and samples collection.

3.4

Shape factors (C) for use in the constant-head well


pernieameter method.

3.5

The components of laboratory constant head permeameter.

4.1

Textural distribution for study area data set

3.2

Soil profiles description of study area located at Facult!


of engineering research area DBAE Field Station

4.3

Textural distributions fbr the data set at each depth.

3.4

Infiltration rate and cumulative infiltration versus elapsed


time for point No. Serdang series soil

4.5

A log-log of infiltration rate and cun~ulativeinfiltration


versus elapscd time for point 110.7Serdang series soil

4.6

Frecluency histogram and fitted distribution f~~nctions


l'or
laboratory evaluated saturated hydraulic conductivity (K,)
using constant head (SCI-1P) method

4.7

Frequency histograni and fitted distribution functions Ihr


field cvaluatcd saturated hydraulic conductivity (I(,)

using Guelph pcrn~camcter(GP) method.


Normal probability plot for field-evaluation logarithmic
saturated hydraulic conductivity (Kt,) values for sand)
clay loam soil
4.9

Normal probability plot for lab-evaluation logarithmic


saturated hydraulic conductivity (K,) values for sandy
clay loam soil

4.1 0

Linear regression betwecn field (GP) and laboratory (SCI 11')


evaluated saturate l~ydrauliccond~~ctivity
(K,) values li)r
Serdang series sandy
clay loam soil.

4.1 1

Measured Ks vs. predicted for coniplete date set of the stud) area. 160

4.12

Saturated hydraulic conductivity. measured vs. estimated


results for the models with the best fit for Serdang Series soil
at each depth

4.13

Measured versus estimated Ks for the six niodels tested li,r


sandy clay loam soil.

;I

LIST OF PLATES

J'late

1';1gc

3.1

The study area at DBAE Field Station. UPM

X4

3.2

Guelph Permeanleter ((31') kit and GP Set up in the s t ~ ~ dill.eil.


y

01

Infiltration rate measurement using double ring i n f i l t ~ . o ~ i i ~ t ~ ~ . .

3 . 3

3.4

View of undisturbed soil core samples collected for bull,


density and saturated hydraulic in the laboratory.

4.1

Plate 4.1 through 4.4 show soil profiles at the study area

SSI

LIST O F SYMBOLS

Description

unit

Bulk Density
Real Density
Liquid density
Dynamic viscosity
Steady Discharge
Steady Discharge at hydraulic head ( H I )
Steady Discharge at hydraulic head (Hz)
Hydraulic conductivity
Intrinsic permeability
Saturated hydraulic conductivity
Field Saturated Hydraulic Conductivity
Geometric mean saturated hydraulic conductivity
Maximum saturated hydraulic conductivity
Minimum saturated hydraulic conductivity
Horizontal saturated hydraulic conductivity
Vertical saturated hydraulic conductivity
The hydraulic or potential gradient
Acceleration due to gravity
Sorptive number
Matric flux potential
Infiltration flux through (x, y, and z) directions
Depth of water level in the borehole
Hydraulic Head
Depth of water ponded at soil core surface
Infiltration Rate
Final infiltration rate
Measured cunlulative infiltration
Organic Mattes
Soil moisture content
Porosity
Number of samples or observations
Length of the intact sample
Cross-section area of the core or brass ring
Internal diameter of the brass ring
Radius of the brass ring
Volume of the cylindrical core
length of the cylindrical core
Time
Weight of the air dry soil
Rate of fall of' water level in GI' rcservois
Coefficient of \,ariation
Standard deviation

g Clll
g cm-'
g c111-'

ssii

-i

poisc
c111.;sen1 ; s- I
cn1-;s- I
cm s-'
3

emem s - '
cm s- I

Clll s-I

cm-

em's-'
Clll

c ni

cI l l

cI l l

-I

R?

v
S
Si
C'

SE
SI-I
GI1
C1-1WP

SCHP
AH
DRI
RI
I PM
DTM
WP
SWPT
SI
TI
FI
G1
PM
PTF
Hz02
GMER
GSDER
PSD
RMSE
RMSR
RMSD
UPM
FA0
USBR
LJSDA

Coefficient of n~ultipledcterniination
Correlation cocl'ficient
variance
Sand
Silt
Clay
Simultaneous 1:quation analysis
Single Head analysis
Guclph permeameter method
Constant Head Well Perniean~etermethod
Constant Head I'ernieametcr method
Auger hole method
Double Ring Inliltrometer method
Ring Infiltronieter
I~istantaneousprofile method
Double Tube Method
Wcll Permeameter
Shallow Water Pernieameter Technique
Sprinkler infiltrometer
Tension infiltrometer
Furrow infiltrometer
Guclpl~infiltrometer
Predictive Model
Pedotransfer function
H! drogen peroxide %
Gcometric mean error ratio
Gcometric standard deviation of error ratio
Pal-iicle-size distribution
Root mean squared error
Root mean squared of residuals
Root mean squared deviations
Ll~liversitiPutra Malaysia.
Food Agricultural Organization
U~litcdStates Bureau of Reclamation
llliitcd States Department of Agriculture

ssiii

CHAPTER 1

INTRODUCTION

General

Water movement in soils whether under the saturated or unsaturated


conditions is highly dependent on the hydraulic conductivity (K) of the soil. For a
given soil, K is defined as a constant that relates the rate of water transport in that
soil to the hydraulic gradient or driving force causing water to move. Under
saturated c ondition it i s called saturated hydraulic conductivity and generally is
denoted by Ks, or

while under unsaturated condition it is referred to as

unsaturated hydraulic conductivity. Qualitatively, K is the ability of the soil to


transmit water and generally speaking, is a maximum at saturation but under
unsaturated condition its value however, have been found to decrease dramatically
with decreasing water content.

Vadose zone soil saturated hydraulic conductivity, K,, is the volume of


water, which will pass through a unit cross-sectional area of a soil above the water
table in unit time, given a unit difference in water potential. Its behavior plays a
crucial role in modeling water flow and chemical transport in the saturated media.
It is perhaps one of the most important hydraulic properties used by hydrologist,
water resources engineers, and environmental soil scientists to solve many
agricultural and hydrological and environmental problems. Soil's K, values have
an important application in areas ranging from the analysis of any saturated-soil

water flow system such as various kind of drainage systems, infiltration,


groundwater recharge, runoff, and irrigation practice. Furthermore, K, is often
required as an input parameter to several hydrological, a ~ c u l t u r a lenvironmental
,
and chemical transport models that allows estimation of the fluxes of water,
pesticides and other hazardous industrial and agricultural chemicals leaving the
soil profile and contaminating on long term groundwater. In addition its values are
also of use to give indirect information about the structure and structural stability
of soils.

Reliable application of most soil water flow theory and the reliability of the
results with numerical models developed for simulating water flow and chemicals
transport in the vadose zone is chiefly linked to the accuracy with which soil
hydraulic properties are determined. Therefore, accurate determination of soil
hydraulic properties such as K,, is considered crucial for successful groundwater
development, designing drainage and imgation systems, and management
practice. Obviously demand for accurate K data for field soils has increased as
related environmental issues have gained prominence, as the use of soil water
simulation models has also increased.

During the past years evaluation of soil K, has occupied the attention of
many researchers and considerable research has addressed on developing and
testing various methods and techniques for determining its value. They are various
methods or techniques used to measure or estimate soil K. They can be classified
as direct and indirect techniques. The direct methods fall into two types depending

on whether the soil was tested in situ or whether the soil samples were transported
to the laboratory for testing by either constant head or falling head technique.

Comprehensive overviews of direct methods of measuring K,, is given by


Klute and Dirksen (1986) and Reynolds (1993a) for laboratory methods and by
Bouwer and Jackson (1 974); Amoozegar and Warrick (1986); Youngs, (1991);
Reynolds (1 993b) and Oosterbaan and Nijland (1 994) for field methods.

The laboratory methods used in measuring K, required that, a soil sample


be taken from the field and placed in one of several kinds of apparatus for test.
The disadvantages of the laboratory methods are that, the soil sample is of
necessity relatively small and that, in obtaining the sample from the field and
preparing it for the apparatus, its structure maybe considerably change. Core
methods, however, can be used to measure the soil K in any direction depending
upon the direction in which they are taken and how they are encased for water
flow through them. Also they can be used to determine the K, of a soil when field
determination is not possible. Core methods may be used as a standard of
comparison with other field method or theories (Hillel, 1980; Reynolds and Elrick
1985) for instance, with Instantaneous Profile (IP) method and the Guelph
Permeameter (GP) method (Paige and Hillel, 1993.) Furthermore, Laboratory
methods are often preferred to in-situ techniques as they permit measurements to
be made simultaneously on many samples in a controlled environment, although
the small size of soil samples used may alter the effects that soil macropores or
structural units may have on soil K (Bouma, 1991).

In order to obtain more accuracy values of soil K,, field techniques are
preferable because the sensitivity of the soil K, to soil structure and generally the
soil sample volume is larger and less disturbed than laboratory techniques. Soil K
is determined in situ by means of a variety of techniques based on forcing water
into the soil (vadose zone area) or removing it under control (below water table).
The most accurate methods are those used in the field in the presence of a water
table such as auger hole (AH) method which is the only method that is possible to
be used in the field on all types of Malaysian soils. However, its implementation
for K, measurements rely on the presence of a shallow water table. Hence, it could
not be used to investigate soil K,, of the vadose zone area (deep water table) which
is needed to plan and design works for the future where the groundwater level is
expected to rise. Therefore, the AH method was not used in this study which
focused on the field measurements of Ks in the vadose zone.

Usually for field methods in the absence of shallow water table (vadose
zone region), a portion of the soil must be wetted before a measurement of Ks can
be done. These include ring infiltrometer methods (RI), double tube method (DT),
air-entry permeameter method (AEP), and various well permeameter methods
(WP). These have been described in several monographs by Bouwer and Jackson
(1 974); Amoozegar and Warrick (1986), and Oosterbaan and Nijland (1994).

Each of these techniques is based on certain approach and has a common


technical limitation such as the equipment and tool required is expensive,
cumberson~e,time-consuming, and large volumes of water are required. For

.~ER~USTAKAANN T W M@lk
m m WTW LUUy-

instance, air entry permeameter is the most rapid method for the measurement of
K, of the v adose zone soil. However, this method tends t o become tedious for

deeper 1 ayers. The double ring infiltration m ethod (DM) h as b een used o n s oil
surface and subsoil, but it requires excavation of the upper soil layers (or
construction of a pit). Less destructive techniques such as DT (Bouwer, 1962) and
the percolation test (PT) (Kessler and Oosterbaan, 1974) have been used to
measure K,of subsoil in situ. However, measurements at depths greater than the
iength of the apparatus require construction of a pit. These methods also result in
fair to poor accuracy in the K, values due to largely undefined boundary
conditions.

The constant head well permeameter, CHWP, is the only field technique
that can be used to evaluate in situ I& of the vadose zone at deep depths. The early
designs of this method have been criticized for its serious drawbacks in the
techniques, and equipment used were subject to large operation and human errors,
time and water requirement and the theoretical background for calculation. So it
has been very little application in soil science over the past 30 years. However,
modified procedure and improvement apparatus, recently called Guelph
permeameter, GP, have increased the soil Ks reliability in less than one hour and
using little water.

The GP method was one of the methods considered in this study.


Eventhough it overcame the limitations to the CHWP method, unfortunately, it has
been shown in various locations or places that the GP soil field saturated hydraulic
conductivity, Ke, values vary highly with different soils especially, in

heterogeneous soils. The drawback of this method, measurement of Ks in loam and


clay soils have been found by various researchers to be statistically low, higher
than or equivalent to those obtained from the intact soil core (SC), AH, DRI, and
velocity permeameter (VP) methods (e.g. Lee et al. 1985; Reynolds and Elrick,
1985; Kanwar, et al., 1989; Dorsey et al., 1990; Mijat and Kanwar, 1994; Mohanty
et a l., 1994;). In addition G P method has been found in loam and clay soils to
yield lower estimate of Ks than pumping test (Dorsey et al., 1990) IP method
(Paige and Hillel, 1993) disk permeameter and DT methods (Mohanty et al., 1994)
and DRI method (Amin et al., 2001). But also yield equivalent or higher estimate
of K, than the AEP method (Lee et al., 1985) and DRI method (Gupta et al., 1993).
It also has been found in texturally uniform silty clay soil at 0.5 m depth to give
equivalent estimates of K, to the AH method (Reynolds and Zebchuk, 1996).

The GP method, however, is the most useful field method where the water
table is deep (vadose zone case). It is widely used and the most promising
approach for assessment of soil's K s near saturation. Moreover, it is simple, easy
to employ, portable, durable, repeatable and allows rapid Kfs calculation. In
addition, the method also yields a combination of vertical Kfs and horizontal Kf,
representing the actual situation in the field and suitable for various types of soils
(Ragab and Cooper, 1990).

The most obvious method or technique to obtain the soil K is by


experimental methods. Unfortunately, determining this soil property in situ or in
the laboratory is often time-consuming, costly, and laborious. Moreover, the
results are often unreliable because of experimental shortcomings and high spatial

and temporal variability. In spite of some of these methods have been


improved for rapid measurements of K,, and easy to use; these methods have been
used with various degrees of success. Variations in soil K as a function of method
of measurement have becn observed by several authors (Lee et al, 1985; Reynolds
and Elrick, 1985: Talsma, 1987).

The difficulty to obtain a reliable and representative estimate may be due


to the spatial variability in the field. It is worthwhile, therefore, to consider the
possibility of predicting this property from easily measured soil properties.
Therefore, empincal models have been used extensively to predict soil K from
easily obtained or already available information soil properties such as soil particle
size distribution (PSD), soil structure, soil bulk density (Db),
and organic matter
(OM). These models are simple and take less time to compute Ks. They can
provide reasonable estimates of soil Ks with considerably less effort and expense.
However, because of heterogeneity, irregular geometry, soil structure, the
predictable methods may yield values that are not representative for field
conditions. Also these predictive estimate methods are not depended on the
created ideal experimental conditions. However, usefulness of these models
depends on the reliability of the correlation and the availability and accuracy of the
data used for the parameter in the model and Ks values may well be accurate
enough for a variety of applications (Jabro, 1992: Amin et al., 200 1).

The accuracy of K, estimates determined by the various methods depends


upon the degree to which steady-state flow is attained, and upon other ability of
each method to address the various theoretical and practical constraints involved

in the representing the physical system with the partial differential equations
selected. Furthermore, the method must optimize several interrelated factors,
including accuracy, speed, simplicity, portability, manpower, capital cost etc. (Lee

The criteria most often used for assessing the relative merits of a method
are the accuracy of the parameter estimates used, soil type, and various practical
constraints on the investigation. Another consideration would be the repeatability;
spatial re-saturation and non-destructiveness, ease of operation, and time required
making several sets of K, measurements.

Statement of the Problem

From the above discussion it appears that different measuring methods


and predictive models may produce different Ks values which may in turn lead to
discrepancy in the prediction of water flow and chemical transport in subsurface
environment. Currently, there is 1 ittle i nformation a vailable t o provide a dequate
guidelines for adjusting the estimated K,. The soil researchers must use best
judgement based on experience and the observed behavior of the particular soil. At
present, there is no universal method that is suitable for all soils and circumstances.
Obviously, the choice of method depends largely on site conditions, the required
type and accuracy of the K measurement, soil type, repeatability, simplicity,
portability, manpower, capital costs and time required making several sets of K,.
On the other hand, selection of the prediction models generally depends upon the
availability and the level of information on physical properties and hydraulic

properties of the soil. Obviously. many of these prediction models have been
derived from regression analysis of data obtained from soils of a limited region:
therefore applying these models for predicting soil hydraulic conductivity in other
area is not always reliable and should be verified.

Objective of the Study

Since there is limited information on the validity of these methods and


models for Malaysian soils. a series of experimental studies were camed out with
the following objectives:
( 1 ) To evaluate the performance of three field methods for measuring the K of

saturated soil. using the GP method. DRI method and SC method.

(2) To develop empirical model that describes the relationship between Ks and
other easily measured soil physical properties for Serdang Series soil representing
sandy clay to sandy clay loam range of soil physical properties.
(3) To evaluate existing empirical relationships for determining Ks by investigating

the consequences of different variables among clay, silt. sand, OM, Db and E on the
estimation of K, using different empirical models.

Thesis organization

The organization of this dissertation is arranged in five chapters. Chapter I1


represents a detailed literature review on the soil K,. The review includes direct
and indirect methods and techniques of determination of soil K,. However, since
9

this work concerned with the vadose zone soil K, the review is more concentrated
on the methods based on this approach.

Chapter I 1 1 discusses the methods and materials used in this study. This
chapter starts with a brief description of the study site and its soil type. Also, it
shows a brief description of principles and application of the GP method. CHWP
method, and DRI method.

Chapter IV presents and discusses the results of the study. It shows the
results of K, obtained directly using GP, SC and DRI methods or indirectly by the
empirical models. Also determination of the soil properties using different methods
is given. In addition. a comparison between the results of all methods and empirical
models for soil Ks and an extensive analysis of the result is performed in this
chapter. The summary and some conclusions from the work described in the
dissertation is given in chapter V.

CHAPTER I1
LITERATURE REVIEW

Introduction

The terms hydraulic conductivity, "permeability", and coefficient of


permeability", although having different technical meanings, are often used in the
same sense (Soil conservation Service, 1972). By definition the Coefficient of
permeability is defined as the rate of flow of water through a unit cross-sectional
area of soil in unit of time, given a unit difference in water potential. The term
permeability is used in a general sense to indicate the relative ability of soils to
conducts or transmits water. It is expressed by the hydraulic conductivity, which
under saturated condition may be defined (Reynolds, 1993a) as a measure of the
ease (or difficulty) with which water can be transported through a saturated porous
matrix. It has the dimensions of velocity (i.e., cm s-', cm h-', m d-', etc) and
represents the soil's ability to transmit water. Its values depend on both medium
and fluid properties. As used herein, it refers to movement of water in a particular
soil under specified conditions. It is defined as the coefficient K in Darcy's
equation which for one-dimensional vertical flow, may be written in the form V =
K i where V is the velocity of the seepage, and i is the hydraulic head gradient

(dimensionless) in the porous medium.

Among all the physical properties of the soil, saturated hydraulic


conductivity (K,) is one of the most important properties for flow and transport
related phenomena in soils and frequently used in water flow studies in soils, such

as imgation and drainage system design (Bouwer and Jackson, 1974). In many of
the current models that s imulate movement of herbicides, pesticides, fertilizers,
water and salts through soil to groundwater, the soil media are assumed to be
homogeneous and isotropic in their flow behaviour. However, most of field soils in
nature are seldom homogeneous, and their K, is one of several properties which are
strongly influenced by spatial and temporal variability (Warrick and Nielsen, 1980;
Lee et al, 1985; Kanwar et al, 1989; Mirjat and Kanwar 1994; Mohanty et a1
1994).

Over the past few decades, much research work have been done on
developing various methods and techniques used to measure or estimate soil
hydraulic conductivity (K). Soil K can be made by two ways, either directly by
passing a test fluid through the soil (hydraulic methods) or indirectly from easily
measured soils properties data that are related to soil K (correlation methods)
techniques (Oosterbaan and Nijland, 1994). Brief description of these methods,
their relative advantages and disadvantages, their applicability, main direction of
K, measurement and the approximate sample volume are presented in Table 2.1. In
general, direct methods and techniques fall into two types depending on whether
the soil was tested in situ (Boersma 1965; Bouwer and Jackson 1974) or whether
the soil samples (disturbed or undisturbed) were transported to the laboratory for
testing (Klute and Dirksen, 1986). The direct in situ assessment of soil Ks can be
done either with small-scale methods or with large-scale methods. Within the types
of methods there are numerous variants that differ in suitability, equipment, water
consumption and time requirement.

1 able 2.1: Uriei description of methods for measuring the soil saturated hydraulic conductivity. Type. applicability. main direction
of K measurement and approximate sample volume (Adapted from .Tenssen. 1 990).
Method

Field

1,ab

I+

Direction
V
UD

Zone
measurenient

Time requirement
per test (h)

Amount of Water
used per test

Volume of sril
sample (cm')

Ilckrences

100 1, -1000 1,

I3ouver and Jackson 1974

AH

Nl A

I3or1ner and Jackson 1974

GI'

< 1, -2 1,

R e y o l d s and Elrick 1985

DT

1-c\\. L

Anioozager and N'arrick 1986

I'M

NIA

Anioo7egar and \\ arrick 1986

PTM

NIA

Bouucr and Jackson 1974

CMTIIP

vP

I I,

IR

101,-50L

Arnoozegar and \\'arricl, 1986


Amozegar and Karrick 1986

AEP

10 l,

Amoozegar and Warick. 1986

DOM

NIA

IJouner and Jackson 1974


and Oosterbaan and Ni.jland
1994

SCHI'

Klute 1986: and Reynolds 1993a

SFHP

Klute 1986: and Ke!nolds 1993a

PM

Oosterbaan and Nilland. 1994

I{= horizontal: V= vertical: UD= undefined direction: S= saturated zone (below water table): IJ= unsaturated zone (above water table): CHWP= constant-head well
permeameter method; AH= auger hole method; GI' = Guelph pernieameter method: DT= Double tube niethod: PM= piezometer method: PTM= pump out method:
VP= velocity permeameter method; IK= infiltrometer ring method; MI'= air-entry pernieameter method: I>OM= drainage-outflon niethod: SCHP= constant-head soil core
permearneter method: SCFP= Falling-head soil core pernieameter method: and I'M= Predictive models.

Certain large-scale field methods based on measurements made in the field


where a suitable drain exists have been proposcd for evaluating soil hydraulic
conductivity property. For example a long ditch maybe cut alongside a pond and
the seepage to the ditch measured to determine K o f t h c soil between the pond and
the ditch: or a system of drain lines maybe installed and the rate of discharge
measured. The tile outflow method to estimate effective K in the stratified and
anisotropic soils has been used by (Hoffinan and Schwab 1964; Lembke, 1967).
They pointed out that field measurements of soil K are better than the estimates
obtained by the AH or SC techniques. Bouwer (1 969) reported that, one or more
tile lines maybe installed to obtain soil

K from measurements of the tile outflow

and water table levels near the experimental tile line. However, he claimed that to
obtain sufficient resolution in layered profiles where a number of measurements at
different locations and at different depths will usually be required to adequately
characterize the hydraulic conductivity profile, techniques that measure K from a
relatively small soil regon might be preferable. Hoffman and Schwab (1 964) and
Skaggs (1 976) proposed methods of calculating soil K and drainable porosity from
drain outflow and water table drawdown measurements.

According to Buckland et al., (1986) the best estimates of soil K, values


of soil at a large-scale are those derived from the performance of a drainage system
installed in a portion of the area requiring drainage. These methods however. not
practical for routine drainage investigations. They require existing drain and the
disadvantage of those methods, assuming that correct theoretical formulas are at
hand to relate the permeability to measured quantities; it is the labour involved;

and further, such installations should ordinarily not be made without first making
penncability measurements.

Various small-scale tield methods can be used for experimentally


determining in situ the soil K of saturated zones with respect to the water table
depth. At present. some estimating methods and numerous laboratory and field
methods for direct measurement of soil K, are available (Klute 1986; Gupta et al..
1993; Paige and Hillel. 1993). Soil K, is determined in situ by means of a variety
of techniques based on forcing water into the soil or removing it out of a borehole
under control (Jenssen, 1990). They can be grouped into two groups: The first
group include those methods applicable to sites where water table present (below
water table) and the other group include those that are applicable to sites where
there is no water table (i.e. above the water table). More specifically, these groups
are applicable to sites located, respectively, in saturated and unsaturated (vadose)
zones of the soil. In either group the determination of K, is obtained from applying
an appropriate formula based on Darcy's law after measuring the gradient of the
hydraulic head at the site and the resulting soil water flow rate. Furthermore. with
all the techniques in both groups, the soil should be as close to saturation as
possible for the K measurement, and cloggmg of pores which enters the soil should
be minimized. Numerous methods that are used for determination of K, include
auger hole (AH) method. piezometer method, the multiple-well technique (MWT),
the pumped well (PW) method, the shallow well pump-in technique (SWPT), the
cylinder pennca~neter(CP) method, the infiltration gradient technique (IGT). the
air-entry (AEP) method, the double tube (DT) method. Table 2.1 shows the
methods and some indication of amount of water required and time involved.

In situ measurements of soil K, below the water table provide the most
reliable values for usc in calculations of ground water flows (Youngs, 199 1). Thcy
usually employ unlined or lined cylindrical holes made below the watcr table and
involve measurements of' flow into or out the boreholcs when the water levels in
them are perturbed from the equilibrium. Several certain reliable field methods
have been developed over the past 40 years or more for in-situ measurement of K,
in the saturation zone. where the soil strata is located below a water table. The AH
and piezometer methods are the most common methods to determine the soil K,
(Amoozegar and Warrick, 1986). Field methods other than the two methods
mentioned before such as two wells flow method, multiple well point and the
pumped well method have been reported (U.S. Bureau of Reclamation, 1965 and
1978; Bouwer and Jackson, 1974; Kessler and Oosterbaan. 1974; and Amoozegar
and Wamck, 1986).

According to Bouwer, (1969) these methods have been listed in order of


increasing magnitude of the soil region from which soil Ks is determined as
following, the piezometer method, the tube method, the AH method, the MWT
method, and the PW method. They provide the most reliable Ks values when they
sample large volumes of soil (Youngs, 1991). The AH method is based on an
unlined cylindrical well bored into the soil to a certain depth below the water table.
During the past years, AH method for measuring soil K, has been evaluated. It is
simple, and the most reliable method in studies involving in situ measurement of K
of saturated soils.

The use of this test however, is limited to areas where a high goundwater
tablc occurs and to soil whcre a boring of' known shape can be maintained
throughout thc dcpth. Therefore, the method could not be used to investigate K, of
decp water table (the vadose zone arca). In addition, the soil K, measured by AH
method can be taken as average K, for the full depth of the hole being tested. In
reality, however. the calculated soil K, value reflects the K, value of the most
penneable layers. Despite the AH method samples a fairly large soil volume, but in
highly layered soils gives little information about individual layers (Rogers, et al.,
1985). Unfortunately, the reliability of the results from different measuring
methods vary under different field conditions (Mohanty et a]., 1994) and this
problem had been addressed by many researchers (e.g. Paige and Hillel, 1993;
Gupta et a]., 1993; Gallichand et al., 1990; Kanwar et al., 1989 and Lee et al.,
1985).

The purpose of this chapter is to review the fundamental concepts and


present a brief review of various operational methods and techniques that have
been used in determining soil K,: as applied in irrigation and drainage planning. In
addition, the theoretical background of the different techniques for determining
soil K, and the major soil properties affecting soil K is described. Since this study
is based on evaluation of the soil K, of the vadose (unsaturated) zone using the
direct and indirect techniques, the review will be more related to field and
laboratory methods and empirical models based on those approaches. Bouwer and
Jackson ( 1974); Kessler and Oosterbaan, (1 974); Amoozegar and Warrick (1 986);
Klute and Dirksen (1986) Youngs, (1 99 1); Reynolds (1 993a.b) and Oosterbaan

and Ni-jland. (1994) reviewed thc significant contributions made by early workers
and gave detailed descriptions of the procedure for applying these methods.

In situ vadose zone methods

The common field methods that have been used for measuring K, of soil in
the vadose zone above the water table include ring infiltrometer (RI) methods,
double cylinder (DC) method, double tube (DT), air-entry permeameter (AEP),
and various well permeameter (WP) methods. Each of these techniques is based on
certain approach and has some advantages and disadvantages. These techniques
have been described in several monographs by U.S. Bureau of Reclamation
(1978), Bouwer and Jackson (1974), Amoozegar and Warrick (1986), Youngs
( 1 99 1 ), and Oosterbaan and Nij land.(] 994). Many of these classical procedures

either do not provide realistic data or have been applied with varying degrees of
success. However. it is not obvious that any of these approaches will work under
all circumstances.

Bouwer, (1 969) examined some field methods and their applications. He


pointed out that the commonly used methods have a common technical limitation,
such as necessity of creating saturated soil condition during the measurements. The
equipment and tool required is expensive, cumbersome, or difficult to obtain, timeconsuming, large volumes of water are required for deep water table conditions.
The DT method (Bouwer,1962) and AEP method (Bouwer, 1966; Topp and
Binni.1976) have been proposed for ~neasuringK, in soil above the water table
(vadose zone). They have been used successfully in sandy soils however. problems

are encountered in loamy or clayey soils with macropores (Bouma, 1983). AEI'
method is designed to determine K, in vertical direction (Ksv) and has been
recommended as the most rapid and reliable method for surface and shallow
subsurface soil layers by Bouwcr and Jackson (1974). This method, however,
tends to become tedious for deeper layers and, additionally, it relies on the
presence of distinct air-entry point on the wetting soil moisture characteristics
(Talsma and Hallam, 1980). Therefore, the method cannot be used for subsurface
K, determination. Less destructive techniques such as DT method and the PT

technique (Kessler and Oosterbaan, 1974) have been used to measure in situ soil K
of subsoil above the water table. The measurements at depths greater than the
length of the apparatus, however. require the construction of a pit. These methods
also result in fair to poor accuracy in the K values due to largely undefined
boundary conditions (Bouma et al.. 1982) as cited by Wilson et al. (1989).
Moreover, the DT method K, values influence by K, in both directions (Ksv and

KsH) with more in vertical direction. One of the other methods rather than these
methods has been used for measuring Ks of the vadose zone on soil surface and
subsoil is the double ring infiltration method (DRI). This method, however.
requires excavation of the upper soil layers (or construction of a pit).

Among the various types of WP methods the CHWP is the only


geotechnical field technique that can be used to evaluate the K, at depths
exceeding 2-3

111 and

its Ks values are more influenced by K in horizontal direction

(Kstl)(Stephens and Neuman. 1982a). These direct in situ methods for measuring
soil K, above the water table are relatively simple in concept and are potentially at
least, more reliable (accurate) than laboratory methods. Unfortunately, they also

have a number of limitations that restrict their practical use such as time
consuming, laborious, and expcnsive to perform and all these methods are
applicable in non-stratificd sandy soils, as the clay content increase, several
methods should perferably not be used. Moreover, vadose (unsaturated) soil K is
more difficult to measure. Yet in nature, steady state conditions are seldom
encountered. Many variants of the WP method, which is sometimes called, the
shallow water pump-in technique (SWPT), the dry auger hole, borehole infiltration
test, CHWP method are available both with respect to equipment and model for
calculation of the K, (U.S. Bureau of Reclamation (1 965 and I 978); Stephens and
Neuman, 1982a; and Jenssen, 1990). The simplest variant is measurement of the
rate of decline in water level in an unlined hole (Bouma, 1983). In this study the
attention confines to the CHWP in which the Guelph permeameter (GP) method is
one of its various types.

The CHWP is a simple method and not required any sophisticated or costly
apparatus. It is among the most attractive methods and widely used methods for
assessment K, of vadose (Reynolds et al.. 1983). This method relies on the
interpretation of infiltration theory and assuming that the infiltration rate results are
not influenced by entrapped air (Young 1 99 1 : Talsma 1987; and Amoozegar and
Warrick, 1986). The soil Ks values can be computed from approximated formulas
that are based borehole geometric (radius. length in contact with information).
constant head of water in the borehole. depth to water table, and steady infiltration
rate into the borehole (Talsma and Hallam, 1980).

These types of mcthods and their applications had been examined by


Bouwer (1969). For various significant limitations, however, most of these
methods have not bcen cntircly satisfactory. The early designs of well
permearneter method have a series of drawbacks in their initial theoretical and
practical development. l'he past practical limitations of these methods such as
CHWP procedure include an installation up to several hours, water requirement up
to 1 m3 or more per test, measurement period up to several days, considerable
equipment and apparatus and two operators were required to conduct the test
(Horn 1971: Talsma and Hallam 1980, Stephens and Nueman 1982a,b; and
Reynolds et a1 1983). Furthermore, there is a need of creating a saturated soil
condition when the measurement is made and it is subjected to large operation and
human errors (Bouwer and Jackson, 1974).

Talsma and Hallam. ( 1 980) however, have removed the above practical
limitations. They developed more efficient an in-hole Mariotte type apparatus that
is lightweight, simple, and inexpensive to construct, and can be operated by one
person. In addition, a measurement can be often made within 5 to 30 min, and up
to 2L or less of water were required depends upon soil type and soil condition. The
practical measurement range of Ks of their device, however, restricted to Ix 10" to

3 x 1 04cm s-'. Talsma and Hallam found that steady recharge rate (Q,) for the WP
method was achieved within about 20 min for sandy soils. Reynolds and Elrick.
( 1 985) showed that equilibrium times were about 10 min. 10-60 min, and 1-24

hours in sands, loams, and clays respectively.

In spite the modification increased its reliability and allowed the

measurement of soil KI, less than

cm s-' that increase usefulness of the

method, thc theoretical limitation remained. Reynolds et al., (1983) made some
refinements to this basic equipment and developed a constant-head well
permeameter that regulate the ponded head levels and measure the flux of water
into the soil from a small c~lindricalborehole. They claimed that the CHWP
efficiency and range of operation has been increased. In addition, Kf, from an "inhole" Mariotte type CHWP with these improvements in both theory and field
application studies has been found to be increased by approximately 60 %
(Reynolds et al.. 1983). However the refinements did not extent to account for the
influence of the unsaturated flow in the calculation of K,.

Guelph Permeameter

Reynolds et al. (1 985) at Guelph University advocated the use of a Marriott


siphon to maintain a constant head in the well and carried out the major
development and improvements on the theory and practical application of the
CHWP method for measurement

Ks.

More recently, Reynolds et al., (1985) and

Reynolds and Elrick ( 1 986 and 1987) re-designed this device and called it Guelph
Permeameter (GP). This apparatus has the potential to measure Kfs and matric
fluxes potential ($,,)

of the vadose zone soil (unsaturated zone) and now

commercially available from Soilmoisture Equipment Corp, Santa Barbara.


California (I 987) as shown in Plate 3.3 (Chapter 111).

The G P was designed to infiltrate (slowly add) water into soil hole under
unsaturated soil conditions to provide data suitable for K, estimate near saturation.
It involves drilling a vertical borehole and determining the steady state discharge
of water when the water in the borehole is maintained at constant level. The
principle of the method and equipment required are discussed in materials and
methods chapter (Chapter I I I ) .

Although, the GP method (Reynolds and Elrick, 1985; Soilmoisture


Equipment Corp, 1987) is one of the various borehole permeameter techniques it is
a considerably different apparatus and procedure. Its theory is based on saturatedunsaturated flow that the others are not (Elrick et al., 1987).

From the practical point of view, the advantages of the method include the
apparatus is simple to use, easily to operate by one person, durable and repeatable.
Moreover. it requires less time (5-60 minutes) and small volume of water (0.5 to 2

L) per measurement than other methods. In addition it has an advantage over other
methods for its ability to produce simple, economic and very quick in place
calculation of Ktj from a single determination. Also the method does not greatly
disturb the soil at the measurement site during the augering process ( e g Reynolds
and Elrick 1985; Reynolds and Elrick 1986; Ragab and Cooper, 1990; Elrick and
Reynolds. 1992).

On the other hand, potential limitations, however, include relatively small


sample size, soil heterogeneity within the measured depth, smearing, remolding.
capillarity, siltation, air entrapment and variability of the K, value (Lee et al..

1985; Stephens et al., 1984 and 1987, Talsma, 1987; Elrick et al., 1990). Although
the GP method gave variable values of soil Kss, once the determination are
repeated it is possible to get more reliable values.

Early studies by Lee et al., (1985) on Sour types of soils (sandy loam, fine
sandy loam, silt loam and clay soils) in Southern Ottawa Canada compared GP Kss
values with air-entry and falling head permeameter method. They showed that
there was significant difference between some or all of the methods within each
site. The results also indicated that the Kfs values of GP for loamy sand, very fine
sandy loam, silt loam, and clay were 3.30 m day-', 0.1 14 m day-', 0.1 123 m day-'
and 0.0073 m day". respectively and the CV% values ranged between 75 and

Wilson, (1989) obtained a mean K k value of 18.317 m day-' and 4.787 m


day-' with CV ranged between 268 and 338 % for two forested catchments
dominated by weathered clay subsoil. Reynolds and Elrick (1 987) used the same
method on loamy soil under corn. The average Kfs value was 0.29 m day-' with CV
between 40 and 120 % which was similar to the values presented by Ragab and
Cooper ( I 990).

Talsma (1987) presented field studies to re-evaluate the WP method on 12


soils that resulted in theoretical overestimation of Kss. His study resulted in
theoretical overestimation of KI, by 10 to 40% for boreholes greater or equal to

0.30 m deep and 0.031~1radius. This serious overestimation of KI, may be due to
the neglected capillanty. On the other hand, work by Stephens, ct. al., (1987)
showed that ticld applications of the borehole penneameter tests on sand, loam,
and sand clay soils under deep water table conditions provided useful results over
range of K, from 10"

to lo-' cm s-' (8.64 x lo-' m day-' 8.64 m day-'). He also

pointed out special problen~sdue to air entrapment can occur.

Reynolds and Elrick (1987) found a mean K1.s value of 0.29 m day-' for
loamy soil under corn and CV% between 40 to 120%. Ragab and Cooper (1993)
used the GP method on three lands (arable Land, grassland, and woodland) and
showed that the average values of Kss of each depth were 0.25-0.3 1 m day-', 0.792.52 m day-', and 0.69-1 2 3 m day-' respectively and the average CV values ranged
from 37 to 60%, 50 to 75% and 59 to 13 1 % respectively.

Fieldwork using the GP method by Salverda and Dane (1993) showed that
the GP seems to work best on coarser, uniform soil and on finer uniform soil.
However, the smearing effects during the borehole soil preparation of fine uniform
have kept to a minimum, which is favored during dry soil conditions. They also
found that about 40% of the Ks values obtained by the GP method were negative
(unreasonable). Closed to this result (38.1%) have been obtained by Lee et al.,
( 1 985). Another study using GP in gassland, woodland, and arable land produced

about 55% 39%, and 34% negative values of Kfs respectively was presented by
Ragab and Cooper (1 993).

The negative values for Kf, might resulted from the small valucs of the
ratio between Q,I and

QS2

(i.c, Q , I / Q \ ~is too low) (Reynolds and Elrick 1985) or

macro pores that may intercepted the holc within thc rangc of the large fixed watcr
hcad, the bottom of thc borchole lies on the interface between two layers or thc
holc contains two distmct layers (Ragab and Cooper. 1993) and randomly
occurring heterogeneity (Lee et al., 1985; Reynolds et a].. 1983. Elrick et a]., 1990;
Salverda and Dane. 1993). Negative values of K,, and

4,

obtained by GP using

two-ponded height techniques (Simultaneous Equation analysis, SE.) are a matter


of general occurrence in heterogeneous soils (Lee et a]., 1985; Elrick and
Reynolds, 1992; and Wu et al., 1993).

Reynolds and Elrick (1985) discussed the negative values for Kk obtained
by using GP and they suggested that small head (H) level be used with extensive
heterogeneity and macroporosity soils. Several authors have suggested using single
head analysis as a possible alternative to multiple head analysis technique of the

GP method to avoid these negative values (Stephens and Neuman 1982a,b; Elrick
et al., 1987 and Elrick and Reynolds, 1993). Higher variability of Kss by applying
the SE analysis was found by Kanwar at al., (1989). Same result was also obtained
by other authors (Elrick and Reynolds, 1992 and Bagarello and Provenzano, 1996)
concluded that estimating of Kh variability based on the SE calculations are
generally unrepresentative of true field variability.

Reynolds et al., (1992) and Bagarello and Provenzano (1996) published an


evaluation of this method and showed it widely used measurement and had
potential for field use.

The method also yields a combination of vertical Kls and horizontal Kfs
rcprescnting the actual situation in the field and suitable for various types of soils.
Ragab and Cooper, 1993 pointcd out that the G P method has an advantage over
other methods for its ability to produce both parameters (saturated and unsaturated

K ) from a single detennination. However this method is very sensitive to the


heterogeneity within the measured depth due to the presence of cracks, fissures,
stones and roots. The GP method seems to work best on coarser uniform soils and
on finer uniform soils if smeanng effects during the borehole preparation can be
kept to a minimum (Salverda and Dane 1993).

Gupta. et al., (1993) found that GP took about 15 to 20 minutes to attain


steady-state conditions at a given head and about 60 to 65 minutes to complete one
set of observations. On other hand, he stated that because of the flow of water into
and through the unsaturated zone depends on both the saturated and unsaturated
components of the hydraulic conductivity, steady-state measurements require
anywhere from 10 minutes in highly permeable soils to hours or more in slowly
penneable soil. Reynolds and Zebchuk, (1996) suggested that the magnitude,
range, and pattern of variability of the Kfs measurements were controlled primarily
by the well-developed and stable soil structure at the field site, rather than by
texture, organic carbon content, or surface topography.

Amin et a1 (2001) used the GP method and found that it is not suitable for
measuring K, in clay soils. However, it is a very useful tool in less or nonstructured clay soils, confirming the result that was obtained by Rogers et al.
( 1 957).

According to Ragab, ( 1 998) the GP method was not suitable for hcavy clay
soils with very low hydraulic conductivity values and for coarsc soils with very
high hydraulic conductivity values.

Double Ring Infiltrometer, DRI

Infiltration is defined as vertical movement of water into a soil. The ease or


difficulty with which water can pass into and through a soil profile is important so
as to avoid detrimental effects such as compaction, surface smearing and other
properties that generally lead to structure decline. The rate at which water can
infiltrate soil determines whether rain or irrigation will enter the soil or will run
off. Vertical infiltration into a dry soil starts at a high rate, but decreases over time
until the rate is constant. This final constant rate is expected to be equal to the
saturated conductivity of the soil (see Figure 2.1), if the soil is uniform and does
not develop a surface crust (Landon, 1991 and Kessler and Oosterbaan. 1974).
This is not the case. however, even under ideal experimental conditions, and final
infiltration rates have been reported to be one-half or two-thirds those of soil Ks of
uniform soils (Miyazaki et al., 1993). A volume of entrapped air, a soil crust. a
compact layer, or a textural discontinuity in the profile can strongly affect the
infiltration rate when the wetting front reaches the discontinuity. Although the
final infiltration rate (1) is theoretically not equal to the K,, it nevertheless yields a
fair approximation. The DRI method can therefore be used to determine in order of
ma~gitudeof the soil K, values.

Time (t)

Figure 2.1 : infiltration rate of dry and wet soil. (Adapted from Miyazaki et al. 1993)

Point infiltration measurements are normally made by applying water at a


specific site to a finite area and measuring the intake rate of the soil. A large
numbers of various inliltration measurement methods have been developed during
the last decades in order to determine the soil hydraulic conductivity (Reynolds,
1993b; Y oungs, 199 1 ; Landon 199 1 ; Amoozegar and Warrick, 1986; Bouwer and
Jackson. 1974; and Kessler and Oosterbaan, 1974).

According to Rawls et al., (1 993) there are four types of infiltrometer: the
ponded-water ring or cylinder infiltrometer (RI), sprinkler infiltrometer (SI),
tension infiltrometer or disk infiltrometer (TI), and the furrow infiltrometer (FI).
These methods, however, suffers from potential or practical limitations for
example, Fl method although carried out under representative field conditions, the
method suffers from practical limitations such as large areas of prepared land and
large volumes of water are required (Landon, 1991).

The T1 permeameter can be used to measure hydraulic conductivity under


small positive potential between 0 and 10 mm. It is very easy to set up, operate and
transport. Obtaining data from this permeameter is a lot easier than the double ring
infiltrometer. The T1 method however has potential limitation related to relatively
small sample size. soil disturbance during the ring insertion and possible short
circuit or edge flow along the ring wall (Reynolds, 1993~).Moreover, data analysis
is a lot more complicated than in the case of DRI method due to the flow being in
three dimensions (Reynolds et al., 2000).

On the other hand, ring infiltrometer methods (either a single or double


ring) used in irrigation studies to determine the infiltration rate and cumulative
infiltration, can be used to look at the hydraulic properties of the first layer of
surface soil in the field. In single-ring, the infiltration characteristics are based on
three dimensional flow and properties such as sorptivity and hydraulic
conductivity can be established from this method. In double ring, however, the
flow of water is thus induced between inner and outer rings then the infiltration is
based on one-dimensional (Oosterbaan and Nijland, 1994; Landon, 199 1 and
Bouwer and Jackson, 1974).

Using the DRI method, the infiltration rate is determined by measuring the
time that it takes the level of water sitting on a soil to drop a fixed distance. This
rate changes with time as the soil pore space fills with water and reaches a steady
rate. This process is repeated and a plot of infiltration rate versus time is
constructed. By measuring infiltration over time, Ks can be calculated once steady
state flow has been reached. It is generally assumed that the hydraulic gradient in
transmission zone is asymptotically approaching unity and the final infiltration rate
(1,) equals Ks (Landon, 1991). The procedure for determining soil K, using a DRI
method is described in chapter 111.

The DRI method used as a way of measuring soil K, of the surface layer.
The DRI also has been used at successive depths in a soil pit to study the
difference in K of various layers. Measurements at depths greater than the length
of the apparatus however require excavation of the upper soil layers. Some
drawbacks of the DRI method are that it is very time consuming, particularly when

the soil is very dry and has expanding clay, also it requires large amount of water
depending on the texture and the type of the soil under investigation. In addition,
the practicality of the instr~~ment
is reduced by the fact the rings are extremely
heavy to move. It also rcquires a flat undisturbed surface, which sometimes is not
available. An air entrapment also may occur during the infiltration processes and
thereby reduce the real rate. Further, the DRI method is sensitive to the K, of the
soil in the vicinity of the inner ring where soil disturbance likely to occur during
the insertions of the rings (Youngs, 1991). On the other hand, the DRI method is
usually the most practical and simplest method for the soil surveyor (Landon,
1991). It is inexpensive to construct, easy to make and to operate and only small
area is needed for measurements. In addition, the method does not have water
requirement (Rawls et al., 1993).

Furthermore, having the two rings minimize the lateral spreading of water
originating from inner ring eliminates the problem of overestimating the hydraulic
conductivity in the field due to three-dimensional flow (Landon, 1991). The outer
ring supplies water, which contributes to lateral flow, so as the inner ring is
contributing to the downward flow. Wherever vertical hydraulic conductivity is the
parameter of interest rather than horizontal K, the DRI method is one of the
methods to use.

Laboratory method, SCHP

Measured hydraulic conductivity values for a soil may vary dramatically


with the method ~lsedfor measurement. In contrast to laboratory methods for

measuring conductivity in soil samples, in-situ methods, in general, involve a large


volume of the soil hence the variability should be lower than for the other methods.
Consequently, the results obtained from field methods should reflect the influence
of both the vertical and horizontal directions and should represent an average value
of K. In addition, laboratory determined values rarely agree with field
measurements, the differences often being on the order of 100-fold or more and
field methods generally are more reliable than laboratory methods. However, field
methods are more expensive, time consuming, and laborious.

On the other hand, using the laboratory methods, soil hydraulic


conductivity (saturated or unsaturated) value, K, can be measured on cores that are
taken in the field and transported to the laboratory. Quality of the results depends
very much upon the quality of the sample which should be representative of the site
under investigation (Smedema and Rycroft 1983).

The main uncertainties in laboratory measurements are concerned with the


representative of the sample (size, density, and structure) and possible interfacial
flow along the tubelsediment interface (Jenssen, 1990). Other problems associated
with laboratory measurements of K on core samples as mentioned by Gumbs
(1974) include soil compaction and smearing, of the soil surface; edge effects;
worm holes; root channels; fissures and other defects that are not representative of
the field soil and the degree of saturation. It is vitally important to minimise soil
disturbance since the K value depends to a large extent upon the soil's pore
geometry (soil structure). The undisturbed cylindrical core is the most
representative and useful form of laboratory measurement of K. Soil K obtained in

the laboratory also is often only an estimate of the actual field conditions (Paige
and I-lillel. 1993). Because of soil K is a highly variable soil property, measured
values easily may vary by 10-fold or more for a particular soil series. In-fact soil K
values measured on soil samples taken within centimetres of one another may vary
by the same fold or more. Thus an extremely large number of core samples would
be required to provide reliable results (Camp, 1977).

In general, the soil core (SC) method is one of the classical techniques for

measuring K,. Like the GP method has potential limitation related to small sample
size, soil disturbance during core collection, and possible short circuit flow through
macropores or along the core wall (Youngs. 1991 and Reynolds, 1993a). The SC
method however, is simple. inexpensive, convenient, and based on a direct
application

of Darcy's law, which defines soil hydraulic conductivity.

Consequently, when the equation of cost becomes decisive, or when actual


representation of field conditions is not of fundamental importance laboratory
methods may be used to determine the soil hydraulic conductivity.

Furthermore, the laboratory methods can be used to determine the K,


values of vadose zone (above the water table) soil in which the field method for
this purpose are almost equally as unsatisfactory as laboratory methods (Smedema
and Rycrofi 1983). The laboratory methods also can be used to compare with other
methods or theories (Hillel, 1980). Moreover, they have been used to separate soil

K horizontally and soil K vertically and used for understanding how salt affects the
soil K and other soil physical properties. The laboratory measurements methods
have been used extensively by many investigators for studying the relationship

between solute. micro-organism and other soil properties with soil hydraulic
conductivity. These methods arc particularly attractive because of the direct
availability of measuring devices and other facility as well as good control of
measuring procedure (Youngs, 199 1 ).

In the laboratory, the K, can be made by several different instruments and


measurement methods such as the permeameter, pressure chamber, and
consolidometer. The measurement techniques, instruments and procedures were
described, among others, by Klute and Dirksen (1986); Youngs (1991); and
Reynolds (1 993a). Laboratory measurements of the K, value may be conducted on
either undisturbed soil samples, soil core, or disturbed samples. K, measurements
on samples in the laboratory usually repeats Darcy's experiment using a constant
head to produce flow or making measurements of flow with a falling head
permeameter. Figure 2.2 gives a schematic test setup for both tests.

Besides these laboratory methods that use completely saturated samples


(SCHP and SFHP, see Table 2.1), other methods wet up samples of dry soil from
saturated surfaces. relying on infiltration theory to give hydraulic conductivity
values of saturated soil such as infiltration method (Youngs, 1991). According to
Childs, (1 969) the K values can be obtained by analyzing vertical one-dimensional
infiltration down columns of soil material. On other hand, the SC method is
attractive because the laboratory measurements are of short duration, they can be
carried in controlling conditions, and they don't require the restrictive boundary
conditions that make many methods slower and expensive. In addition, equipment
for K tests is relatively inexpensive and may be purchased completely or fabricated

/
Falling -head

Over flow device to hold head

Constant head

a- -

Tailwater elevation

Standpipe with area, a -

Soil sample with area, A

Figure 2.2: Schematic test setup for constant head and falling head tests. Use the
right side of the figure to identify terms used for the constant head test,
and the left side for the falling head test (Bowles, 1979).

locally (Oosterbaan and Ni.jland. 1994). Despite potential limitations. SC method


remains one of the most popular means for measuring K, and it is often used as
benchmark for evaluation of other methods (Reynolds at al., 2000; Paige and Hillel
1993; and Reynolds and Elrick, 1985).

Comparison of thc methods

Comparison between several methods of measuring soil Ks are reported in


the literature (Camp 1977; Lee et al., 1985; Rogres et al., 1985; Gallichand et al.,
1990; Gupta et al 1993; Paige and Hillel 1993; Darsey et al., 1995; Bagarello and
Provenzano, 1996; Reynolds and Zebchuk., 1996; Reynolds et.al., 2000 and Amin
et al., 2001)

Previous studies comparing soil Ks values determined from AH and


CHWP methods have reported that those methods are sufficiently reliable in
relatively homogeneous soil, comparing to values determined from the subsurface
drains (Lembke (1 967): Deboer (1 979) and Buckland et al., (1 986). On the other
hand, laboratory determinations of soil K, values from semi disturbed cores have
yielded mixed results when compared with soil Ks value determined from drains
(Lembke 1967; Skaggs et al.. 1977: Camp, 1977). In comparison to field methods,
laboratory methods are particularly attractive because of the direct availability of
measuring devices and other facility as well as good control of measuring
procedure. However, it is hazardous to correlate the K to the soil texture alone,
Childs (1969) and computations of the K, from soil core samples are only an
estimate of the actual field conditions (Paige and Hillel, 1993).

Earlier study by Talsma (1960) compared the CHWP method and AH


method showed that the CI-I W P method tends to underestimate soil K, values by as
much as 61% as compared to soil K, of' A H method. Comparison study of two
methods (CHWP method and soil cores, SC, method) by Gumbs (1974) for soil K,
values determination showed that the mean laboratory measured values were
greater (by a factor of between 1.5 and 15) than the field values and replicates
were more variable.

Camp (1977) evaluated four methods (SC method, auger hole method,
calculation of soil K, value from drain outflow and drawdown measurements, and
calculation of K/f ratio from water table drawdown measurements) for an alluvial
soil in the lower Mississippi Valley. He found that the soil Ks values determined
using SC method (in the laboratory) were much lower and more variable than
those determined by AH method and those calculated from drain flow and water
table drawdown measurement. Moreover, Camp claimed that the drain flow and
water table drawdown measurement method provides the most reliable estimate
value for use in drainage equations. Talsma, (1960,1987) and Bouma, (1983)
reported that there are large variation between field and laboratory tests. Watt et
al., (1982) found that soil Ks values determined on soil cores could be up to 700
times greater than the soil K, determined by in situ techniques. On other hand, Wu
et al., (1993) observed that soil K, values from the detached core samples were
approximately twice that of the field-measured K, values confirming the early
result observed by Bouwer (1 966).

Field investigations of three methods (AH, SWPT, and drain line method)
used to estimate soil Ks were conducted on fine textured, lacustrine soil by
DeBoer (1979). He concluded that the pump-in method soil Ks value was about
one-half the AH method value. I-lowever, a composite pump-in method K, value
for the layered soil profile was about equal to the drain line soil K, value.

Reynolds and Elrick, (1986) by using the numerical solution, found that
the CH WP method produced estimate Ks values of a structureless sandy soil were
statistically equivalent to the estimates Ks value produced by AEP method. Park
and Yu. (1983) compared three methods for measuring soil Ks. There were
inverse auger hole method, infiltrometer method and core sample method. Park
and Yu concluded that the inverse auger hole method was highly correlated with
the infiltrometer method while the core sample method was greatly
underestimated in comparison with the values obtained by the other two methods.

In another study, Rogres et a1 (1985) evaluated four methods (SC method,


AH method. DT method, and drawdown method) for determining K, of an alluvial
soil in the lower Mississippi valley. They indicated that the four methods gave
similar values of soil Ks for a given depth when the normal variation in soils is
considered. Reynolds and Elrick (1 985), compared three methods for assessing
soil K, of four soil types (loamy sand, fine sandy loam, silt loam, and clay). The
methods were instantaneous the GP method, the AEP method, and the laboratory
determination using SC method. Reynolds and Elrick found that the GP method
produced (within-soil type) differences in mean Kk of about a factor of 2

compared to AEP method, and effectively averaged the K, values thosr


determined from the vertically and horizontally oriented soil cores.

Field study to evaluate three methods for determining K, for a loamy sand. a
fine sandy loam, a silt loam, and a clay loam soils in Southern Ontario was
reported by Lee et a1 (1985). The methods used were AEP method, GP method,
and falling head permeameter (SFHP) method. They found that there was
tremendous variability in the results. They indicated that although the techniques
were able to discriminate between the three soils types, the best choice of method
for any particular situation appear dependent on soil type, and the various practical
constraints on the investigator. Buckland et al. (1986) conducted a similar study
with four in situ methods (AH method, CHWP method, in situ Falling-Head
Permeameter. FHP, and from single drain outflow and water recession) in southern
Alberta. They found that the AH and CHWP, and FHP methods provided
sufficiently reliable estimates of K, and they are practical for subsurface drainage
design.

Using the improvements in the field technique and theory of steady out
flow of water ponded to constant head in boreholes Talsma (1987) re-evaluated
the WP method. He found that the soil Ks values obtained by WP method were
between 30% and 60% of the AH method values confirming the early conclusion
of Talsma (1960). Performance of the GP method and VP method for silt loam
soil in central Iowa was investigated by Kanwar et al., (1989). The results showed
that these two methods provided similar Ks values but tended to be much lower
than SCHP (laboratory method) values.

Similar study compared K, values of Ravenne silt loam and Hoyville silty
clay loaln soils obtained from GP method, V P method, a PT procedure, and A H
method was conducted by Dorsey, et al., (1990). The study showed that the GP
method gave significantly lower estimate than the other methods where their
results were within similar ranges. Evaluation study of four methods (GP method,
the SFHP method, and the single AH methods) by Gallichand et al., (1 990) in west
of Montreal, Canada showed that there was tremendous variability in the results.
They found that the SFHP method produced soil Ks values lower than the GP
method and the AH method (in two of three situations) gave higher soil K, values
than the GP method. They gave the following reasons that may are responsible for
the higher K, values of AH method: the large volume of soil sampled by AH
method. the layering of soil profile, and the direction of the flow. On the other
hand, a good agreement has been shown between field measurement of hydraulic
properties using GP method, Guelph Pressure method, tension infiltrometer
attachments and the SC method (Elrick et al., 1990).
Dorsey et al. (1 990) found the GP to yield mean soil Kfs values that were
factors of 4.5 and 83.3 lower than the corresponding AH values in a silt loam and a
silty clay loam, respectively. Another work by Gallichand et al. (1990) showed that
(for various depths) in heavy clay soil that the GP method yielded mean Kf, values
that ranged from factor of 1.43 higher to a factor 24 lower than corresponding AH
values. However, Reynolds and Zebchuk (1996) found the GP method soil Kfs
value was equivalent to corresponding AH values in a textural uniform silty clay
soil with stable but spatial variable structure.

Gupta et al. (1 993) compared the pcrformance of four K, measuring in situ


methods (DRI method, rainfall simulator (RS) method, GP method, and Guelph

infi ltrometer (GI)) in Ottawa Canada. Thcy reported 'that K, values obtained with
GP method and DRI method were statistically the same, but were significantly
lower than those determined by the RS and GI methods. Also Gupta et al. found
that the GP and GI field soils Kt, values were higher variable than those obtained
with DRI and RS methods and then, the GP and IG methods required a larger
number of measurements as compared to the others to achieve a mean Ks values
comparable.

Paige and Hillel (1993) conducted a study in Western Massachusetts to


compare three methods (IP method, the GP method and SC method) for assessing
soil hydraulic conductivity for two soils. They concluded that the field Ks values
determined using the GP method was 1 to 3 orders of magnitude less than the K,
results determined from other methods. On the other hand, Salverda and Dine
(1 993) made a comparison study to re-examine the GP method on a Norfolk sandy

loam, a Lucedale loam. and a Troup loamy sand. Their study. however, hardly
showed any agreement between the GP and the SCHP methods. Mirjat and
Kanwar (1994) conducted similar a study with the two method (GP method and
SCHP method) in loam soil of central Iowa and found the GP method (for various
depths) produced soil Krs values that range from 10 to 130 times smaller than
corresponding SCHP values.

Mohanty et al., (1994), made a comparison study to evaluate the


performance GP method, V P method, disk permeameter method, DP, DT method,

and SCHP method for two soils in a Wisconsin glacial-till soil of the Des Moines
Lobe. They found that the GP method gave the lowest soil Kf., values while the
disk permeameter and the double tube gave the maximum values of soil Ks. They
concluded that lowest values of the GP might be due to the small sample size.

Bagarello and Provenzano (1996) conducted a comparison study to


evaluate the performance of GP method and SCHP method for sandy clay soil at
Palermo University. They concluded that the Ks values obtained by GP method
yielded smaller means and CV values than the Ks values obtained using SCHP
method. However, Bagarello and Provenzano claimed that comparison of results
obtained in the initially dry soil (GP method) and on large cores (SCHP method)
yielded estimates of mean Ks that were not significantly different. Another work
Comparison of GP method and DRI method by Amin, et al (2001) shows that there
was tremendous variability in the results. They found that the GP method gave
lower values for Kfs than the DRI method.

Despite the differences between GP in situ method (Reynolds and Elrick


1986) and Constant-head soil core permeameter (SCHP) laboratory method (Klute
and Dirksen 1986), several researchers compared measurements of these two
methods (Bagarello and Provenzano, 1996; Mohanty, et al., 1994; Paige and
Hillel, 1993; Gallichand. et al.. 1990; Kanwar, et al., 1989; Reynolds and Elrick.
1986; and Lee et al., 1985).

Reynolds et al., (2000); Paige and Hillel, ( 1 993); and Reynolds and Elrick,

(1985) used the soil core method as a standard of comparison with the IP, GP,
Tension infiltrometcr, TI, and single-ring pressure intiltrometer PI, methods.
According to Reynolds et al., (2000) comparing techniques for measuring the K,
of natural soil is an imprecise, perhaps even dubious enterprise because there is no
independent soil K, datum or benchmark upon which evaluations and judgements
can be made. However, it is important to make such comparisons because they
provide one of the few sources of information that researchers can draw upon to
select soil K, measurement methods that are appropriate for their circumstances.

Empirical Models

A broad array of methods currently exists to determine soil hydraulic


conductivity in the field or the laboratory (e.g. Klute, 1986; Reynolds, 1993a,b).
While measurements permit the most exact determination of soil hydraulic
conductivity. they often require a substantial investment in both time and money.
Consequently, it is not always regularly measured in necessary detail for modeling
purposes. Moreover, many vadose zone studies are concerned with large areas of
land and that may exhibit substantial spatial variability in the soil hydraulic
conductivity. It is impossible to perform enough measurements to be meaninghl in
such cases, thus indicating a need for inexpensive and rapid ways to detennine soil
hydraulic conductivity.

Although the most obvious way or technique to obtain soil K, is by


experimental methods, these tend to be difficult, cumbersome, time-consuming,

laborious. costly, and requires a large number of samples. Therefore, considerable


attention has becn given to indirectly predict this parameter from soil variables
routinely measured in thc laboratory or in the field or readily available, such as soil
particlc size distribution, PSD, (% sand, % silt, and ' X ) clay), Dh, E, OM content.
and soil moisture retention. At present numerous relationships have been proposed
that can be used to estimate K, with easily measured or readily available soil
properties (e.g., Rawls and Brakensiek, 1985; Campbell, 1985; Cosby et al. 1986;
Saxton et al. 1986; Ahuja et al., 1989; Vereecken et al., 1989,1990; Uma, et al.,
1989; Jabro, 1992; Schaap et al., 1998; Lin et a]., 1999; Wosten et a]., 1999 and
Amin et al., 2001).

In addition to the direct methods, this study investigates the perfomlance of


a regression of different models in predicting Ks values from soil variables.
According to Klute (1986) selection of soil K prediction technique depends upon
availability and the level of information on physical and hydraulic properties of the
soil.

Many of these empirical models have been derived and have been used
extensively to predict soil K and open promising prospects for estimating soil Ks
over large-scale areas without extensive measuring program. All empirical models
have a strong degree of empiricism in that they contain model parameters that were
calibrated on existing soil hydraulic databases. These predictive estimate methods
are not depended on the created ideal experimental conditions. However,
usefulness of these models depends on the reliability of the correlation, the
availability and accuracy of the parameters data used in the model. Most of these

methods can be classified as pedotransfer functions (PTFs), because they translate


existing surrogate data (e.g. soil PSD, Dh, and OM content) into soil hydraulic
data. A PTF is defined as a function that has as argument basic data describing the
soil and yield as a result the water retention function or (saturated unsaturated)
hydraulic conductivity (Tiet-jeand Tapkenhinrich, 1993).

A PTF can be as simple as a lookup table that gives hydraulic parameters

according to textural class (e.g. Wosten et al., 1995) or include linear or nonlinear
regression equations (e.g. Rawls and Brakensiek, 1985; Vereecken, eta1.1989;
Vereecken, 1995; Minasny et al., 1999). Porosity, PSD, and porous size
distribution are the most commonly used soil characteristics for estimating soil
hydraulic conductivity. Useful prediction of soil K, also was possible using field
texture, grade of structure, dispersion index, and horizon type (Mckenzie and
Jacquier, 1997). and by using initial moisture state, pedability, microporosity, and
root density (Lin et al., 1999).

Generally speaking, these models are either based on physical approaches


or on empirical regression equations of soil textural properties fitted to large data
sets. Tietje and Hennings, (1996) and Zhuang et al., (2000) revealed that, at
present there are three modeling approaches for estimating the soil K, values:
methods of the first approach estimate K, by relating them to non-hydraulic
properties. These models estimate soil K, parameter by (generally, multiple
nonlinear) regression statistics from input easily obtainable soil physical and
chemical properties such as sand, silt, clay, organic matter content, bulk density,
porosity (Rawls et al., 1982; and Vereecken et al 1990). This approach includes

methods from Cosby et al., ( 1 984), Brakensick et al., (1 984), Saxton et al., (1 986),
Vereeckenet al., (1990), Jabro (1992), and Amin et al. (2001). The second
approach based on developing a physico-empirical relationship between the PSD
and soil K, values Campbell (1985) approach. Based on the assumption that PSD is
approximately long normally distributed, Campbell (1985) proposed an empirical
equation to estimate K,. The third approach for estimate soil K, is based on scaling
techniques which can be used to estimate soil K at different locations in a
watershed from measurement of these parameters at one representative location
and limited data at other location (Ahuja et al. 1984).

Because only % clay, % silt, % sand, Db, E, and OM content are routinely
available from soil information systems, in this study soil K, was predicted on
bases of these parameters. Therefore. the last two approaches are not considered in
this study and the attention only is drawn to the first type approach. Since
prediction of soil K, values generally improved if more input data were used,
investigation of six models with the following levels of input data were selected:
( I ) soil PSD (% silt, % clay. and % sand content); (2) % silt. % clay, content and
soil Dh, (3) % sand, % silt, and % clay content and soil E. The six predictive
models Cosby et a1 (1984); Brakensiek, et.al., (1984); Saxton, et al., (1 986);
Vereecken et al., (1990); Jabro, (1992), and Amin et al., (2001) that based on
these parameters were evaluated.

PTF assessment has been made with respect to a data set of measured K, of
soils from Peninsula Malaysia. Except Jabro (1992) and Amin et al., (2001)
models, the above PTFs have yet to be verified against these data. PTFs such as

Vereecken et al., Cosby et al., Brakensiek et.al., Saxton, et al., and Jabro however,
have been derived from regression analysis of data obtained from soils of certain
regions; therefore, using these models for prediction beyond these regions is not
always reliable and should be verified (Ragab and Cooper 1993). Furthermore, an
independent evaluation can only be performed with data sets, which were not
previously used for the development of the PTFs (Schaap and Leij 1998; and
Wagner et al., 2001).

In past decades, several evaluations and comparisons of the performance of


the many PTFs have been carried out. Early studies comparing the PTFs applied to
a common database or consistent set of measurements focused mainly on
evaluation of soil water retention function (e.g. Arya et al., 1999) and saturated
hydraulic conductivity (e.g. Vereecken el al., 1989,1990; Tietje and Hennings,
1996; Schaap et a]., 1998; Wosten et a]., 1999; Schaap and Leij, 2000 Wagner et
al., 1998 and 200 1).

Gumbs (1 974) proposed simple linear regression between the soil I(, and
the sand coarse content (Ks = -0.62+0.11 % of coarse sand) and claimed it should
be adequate for soil survey work and for calculating drain spacing. In another
study, using one type of soil genesis with highly varying texture (1.4-42% clay
content), Puckett et al., (1985) were able to use particle-size fractions (sand, silt
and clay) to predict the soil hydraulic properties. They found that In (Ks) could be
predicted with clay content (r2= 0.77).

Espeby (1990) performed a statistical analysis of 100 soil core samples


taken from six soil pits and Sour different mineral soil horizons in a forested glacial
till slope. I-le found the correlalion of soil K, with other soil properties was highest

with drainable porosity (r2=0.77). Vereecken (I 990) published PTFs for K, with a
coefficient of determination of about 20% using clay and sand content, OM %, and

Db density as input soil data. This may show that PTFs still show a considerable
amount of unresolved variability in the estimates of K, (Vereecken 2002).

Jabro (1992) correlated soil Ks parameter with soil particle size distribution
(% silt and % clay) and soil Db density (as an implicit representation of the soil

structure) and developed multiple linear regression models to predict the soil K,
values. Comparison results with experimental data for Duffield silt loam at
Lancaster Country in Southeastern Pennslvania, Jabro found that the model
performed fairly well and gave a satisfactory validation versus the field measured
soil Ks data using undisturbed core samples. Using sand content, Db and OM
content data enable the estimation of soil Ks values with r2 value = 0.957 and the
models utilizing these easily measurable soil properties can be used for predicting
soil water characteristics curves and K, with a satisfactory level of accuracy (Singh
et al., 1992).

Bonsu (1992) compared values of Ks estimated by texture-based equation


of Rawls et al., (1982) with soil

Ks

values measured by the falling head method.

He concluded that the texture-based equation could predict those measured within
the same order of magnit~ldeonly for the coarse-texture soils. However, for the
fine-textured soils it underestimated the measured values by approximately one

order of magnitude.

Vereecken (1995) examined the quality of I I different

theoretical models to predict the unsaturated hydraulic conductivity for a wide


variety of soils and dcmonstrated that the performance of a PTF depends on the
data sets on which a PTF is calibrated and tested. Study by Tietje and
Tapkenhinirichs (1 993) tested the regression equation of Vereecken et al., (1 989)
and found that model was applicable to many soils and gave the best agreement for
a large data set of German soils.

Tietje and Henniings (1996) tested six PTFs for predicting soil Ks values
using a German database of 1161 samples. They found that none of the tested
PTFs had consistently better prediction for all textural classes. In other study,
Schaap et al., (1998) used a data set of 1209 samples to evaluate neural networkbased PTFs models. Results of this study show that the neural network base PTFs
usually provided better prediction than PTFs published by Rawls and Brakensiek
(1 985); Vereecken et al. (1 989); Saxton et al. ( I 986); and Cosby (1984).

In another study. Wanger et a1 (1998) tested three predictive models for


hydraulic parameters using data sets of six German soils (156 measured water
content vs. matric heads and 130 measured unsaturated hydraulic conductivity vs.
matric heads). They made comparisons of the models of Campbell (1 974, 1985),
regression equations of Vereecken et al., (1989.1990) and the model of Gregson.
Hector. McGowan (GHM model) Gregson et a1 (1987). They found that the GHM
model produced noticeable better results than the other two models and
recommended that hydraulic parameter be estimated from paired measurements of
water content vs. matric head rather than from textural properties of soils.

Arya et al (1999) presented a model to compute K function from the soil


I'SD and D,, data. They found that prediction of the K for a number of soils,
representing a range in texture, Db, and OM content was reasonable. They found
that the root mean square residuals (RMSRs) of log-transformed predicted and
experimental data ranged from 0.487 to 1 S62, the average being 0.878. Schaap
and Lei-j (2000) used a data set of 235 soil samples to test and improve predictions
with the Mualem-van Genuchten (MVG) equation.

Study by Zhuang et al (2000) compared eight models for scaling soil Ks


using a database of 402 data sets showed that models based on the nonsimilar
media concept and the Kozeny-Carman model preformed best for soil K, scaling.
The models of Saxton, Cosby, Vereecken, and complex Brakensiek gave the
second best estimation of soil I(, values. Moreover, they found that all eight
models had smaller estimation deviations for sandy soils than for clayey soils.

Amin et al.. (2001) examined the Jabro model and pointed out that. the
Jabro model K, values were very high compared to the Ks values measured by GP
and DRI methods. Based on similar form of Jabro, Amin et al. used soil PSD (%
silt: % clay) and Db and established an empirical relationship to predict K, values.
They claimed that their model [K, = 5.24 - 2.08(Db)-0.029(% silt)-0.025(% clay) ]
is an advantage over the costly and time consuming nature of field determination
and it is useful for estimating soil K, values for modeling, planning, and
management purposes where soil K, values are not available.

New model for predicting unsaturated or relative hydraulic conductivity


was proposed by Zhuang et al., (2001). Using a data set of 52 soils, they compared
their model with other five widely used models. Zhuang et al found that, although
none of the tested models had a consistently better prediction for all textural
classes, their model generally performed best, having a tendency of being more
suitable to the sand and loam soils. Wanger et al. (2001) reviewed eight
pedotransfer functions used for evaluation of soil K from routinely available soil
data. They pointed out that, the evaluation of soil K by pedotransfer functions
shows, on average, better correlations if the K, (an input parameter) is also
obtained from predictions of pedotransfer functions rather than directly from
experiments. However, Vereecken (2002) disagreed with the statement that
experimental data used for the determination of PTFs might have not been correct.

In general. practical application of most PTFs is often hampered by their


very specific data requirements. Some authors established PTFs that provided the
best results for their data set, which sometimes produced models that require many
input variables ( e g Rawls et al., 1991) or detailed soil PSD (Haverkamp and
Parlange, 1986). However, users of PTFs are frequently confronted with situations
where one or several input variables needed for a PTF are not available. Another
problem is that the PTFs provide estimations with a modest level of accuracy
(Schaap et al., 200 1 ).

On the other hand, the PTFs are quicker. simple to apply, cheaper, and the
input variables can be measured more easily and, hence, are more widely available
than hydraulic properties. Moreover, in the cases of large areas with different soil

types and horizons, more in situ measurements will be required to represent the
spatial variability. Therefore the predictive models might look a more attractive
tool to use. Furthermore, it would therefore be useful if PTFs could accept input
data with varying degrees of detail and if PTF predictions could include reliability
measures.

Theoretical considerations of the GP, SCHP, and DRI methods

One of the most important mass properties of sediment is its permeability,


which describes how easily a liquid will move through a soil. It is also commonly
referred to as the hydraulic conductivity of a soil. Here the permeability depends
on the property of the medium and is independent of the fluid properties. On the
other hand, hydraulic conductivity of a soil represents its average water
transmitting properties. which depends upon the properties of the fluid as well as
the properties of the medium. This relationship can be expressed as:

where K is the hydraulic conductivity (cm sec"), k' is intrinsic permeability


independent of density and viscosity (cm'), g is the acceleration due to gravity
(cm sec-*), and p and q are the mass density (g cm -') and dynamic viscosity
(poise, g cm-' sec-') of the solution respectively.

Liquid in soils tends to flow from a location with high-level energy to low
one. The total energy at any point in a porous media can be represented by the
hydraulic head h, which is, from Bernoulli equation, a function of pressure and

potential energy per u n i t weight of fluid, neglecting the kinetic energy and
assuming isothermal condition. Values of soil hydraulic conductivity, K are
usually calculated assuming that Laplace's equation on saturated conditions, or
Richards' equation in unsaturated soil conditions, with assumption that the soil is
uniform, hon~ogeneous,deep, implicity that Darcy's law describes the flow of
water. In fact there are two difficulties arise when simulating porous media flow,
where both saturated and unsaturated conditions exist. They are, the difficulty
concerns the governing equation, where a unified equation that governs porous
media flow under both saturated and unsaturated conditions is needed, and the
difficulty involves the quantification of a composite hydraulic conductivity for the
porous medium.

The numerical and theoretical development of the measurement techniques


for the K, of vadose zone took place over long time and the physical base of water
flow believed to be fully known since the fundamental work of Richards. Only
recently, new and better measurement technique shows that some practical
considerations on water flow in vadose zone such as air entrapped, soil
heterogeneity, and effect of unsaturated flow need to be re-evaluated.

In recent years, numerous papers and reports on the advances in the


theories underlying the use of SCHP method, GP method and R1 method for the
effective measurement and interpretation of soil Ks have been widely described in
the literatures (e.g. Stephans and Newman, 1982 b,c

Reynolds et al., 1983 and

1985, 1992; Reynolds and Elrick 1985, 1987, 1990; Youngs. 1991; Elrick and
Reynolds 1987, 1992, Erick et al., 1993; Reynolds, 993a,b and Xiang 1994).

Only brief detail of theories of these methods as relevant to this study is repeated
here.

Laboratory method using SCHP

The vertical saturated hydraulic conductivity K, of interest here is a


measure of the ease of water movement in soil, the amount of water that would
move downward through a unit area of saturated in-place soil in unit time under
unit hydraulic gradient. Darcy (1856) proposed an empirical relationship
describing the rate of flow of water through porous media from extensive work on
the flow of water through sand filter beds and derived a constant called it
permeability coefficient (hydraulic conductivity). This relationship has been
known for many years and is the fundamental equation for flow in saturated soil.
Darcy's law states that the volume of water flux, Q, per unit time is directly
proportional to the cross-sectional area, A. of the column and to the hydraulic head
drop (hydraulic gradient), AH, and inversely proportional to the length of the
column. L. The generalized form of Darcy's law for one dimensional vertical
saturated flow in saturated porous media may be expressed as foilows (Klute and
Dirksen 1986) and Reynolds, 1993a):

where Q is the volume of water collected per unit time, cm3 sec", (or the
discharge), i is the hydraulic or potential gradient or driving force, dimensionless,

K, is the sat~~rated
hydraulic conductivity (cm sec-I), and A is the cross-sectional

area of the satnple (cm'). The negative s i g in the equation refers to the flow
moving in thc direction of decreasing head or dccrcasing potential. For a stationary
tlow in a saturated soil the hydraulic gadicnt (driving force), i, is equal to thc
gradient of the sum of the pressure and gravitational potentials so that if the
pressure head h is maintained on the top ot'a soil sample. L. then

and

KS=

OL
At(H, -H,)

In which L is the length of the sample (cm), t is the time (s) and (H2 - H I ) is the
difference in head between the top and bottom of the sample (cm) Q is the volume
of water of water passed in unit time, (cm3 s-I), A is the area of the bed (cm'). L is
the thickness of the bed (cm). H is the height of water on top of the bed (cm) and

K, is the hydraulic conductivity coefficient (cm s-I).

The SCHP test operates in accordance with the direct application of


Darcy's law to a soil liquid confip-ation representing a one-dimensional, steady
tlow of a percolating liquid through a saturated column of soil from a unifonn
cross-sectional area. A constant head difference, Hz - H I , is then applied across the

test sample. To compute hydraulic conductivity water ponded on a saturated soil


core at a constant head, and the resulting steady state flux and head gradient were
measured. With the Darcy's Law cxprcssed in this form and using the simplest
possible experimental set-up as showing in Figure 3.6 chapter I l l , the hydraulic
conductivity of the soil can be determined. By using the sketch drawing of SCHP
(Figure 2.1) the cross-sectional area (A) and the length of the soil column (L), the
height of the water column (H), and the time (t) can all be predetermined. The
amount of water passing through the soil column (Q), in the predetermined time, is
simply measured using a graduated cylinder. The measurements on the soil cores
were analyzed on the basis of the outflow concept of Darcy's law and K,
calculated directly using Eq 2.5.

Ring Infiltrometer method

Infiltration is defined as the process by which water enters the soil. The
infiltration rate 1, of a soil was defined by Miyazaki et al. (1993) as the flux of
water across a land surface into the soil. The mathematical and physical analysis of
the infiltration process developed by Phillip (1957) separates the process into two
components. that caused by the matric potential force (suction) and that caused by
gravity. The early stage is dominated by soil suction, which for a specific soil is
determined by the moisture content. This component tends to vanish at large times
and the infiltration rate becomes equal to the Kfs. Thus as infiltration progresses,
the influence of the initial soil moisture content decreases and eventually becomes
negligible (Phillip, 1957). An ideal infiltration curve, based on theoretical analysis,

is illustrated in Figure 2.2. A given soil ends up with the same infiltration rate
whether the soil was initially dry or moist (Miyazaki et a1 1993).

Infiltration under conditions where free water is present on the ground


surface is referred to as ponded infiltration (Rubin and Steinhardt 1963). They
claimed that in the initial stages of ponded infiltration, the rate of water entry
usually decreases appreciably with time because of the deeper wetting of the soil,
which results in a reduced suction gradient, and the closing of cracks and other
surface connected macropores. Transient ponded infiltration is the stage at which
the ponded infiltration decreases markedly with time. After long continued wetting
under ponded conditions, the rate of infiltration reaches a fairly steady rate. This
stage is referred to as steady ponded infiltration. Assuming the absence of zones of
free water within moderate depths and that surface or near surface features (crust,
for example) do not control infiltration. This infiltration rate is termed infiltration
capacity, I,,. Eventually. steady state infiltration rate is approached. It is generally
assumed that the hydraulic gradient in transmission zone asymptotically approach
unity and the final infiltration rate, infiltration rate capacity, (1,) equals Ks. It is
assumed that the soil is homogeneous and uniformly wetted; the rate of infiltration
is enough to be measured accurately; seepage beneath the inner ring is onedimensional; the wetting front suction was taken as zero; the effect of boundaries
beneath the ring were negligible.

According to Miyazaki et al., (1993) and Rawls et al, (1993) the


mathematical formation of infiltration may be classified as empirical methods or

physical methods. Rawls et al (1993) pointed out that most of the empirical and
approximate models treat the soil as a semi-infinite medium with the soil
saturating from surface down. On the other hand, physically based models specify
appropriate boundary conditions and normally require detailed data input. In
addition to the above strictly infiltration based models many consider rainfall
excess models which lump all losses (infiltration, depression, storage, interception)
together as infiltration models (Rawls et al., 1993). Empirical models generally
relate infiltration rate or volume to elapsed time modified by certain soil
properties. Parameters used in these models are commonly estimated from
measured infiltration rate-time relationships for a given soil condition. According
to Miyazaki et al.. (1993) and Rawls et al., (1 993) the three most well- known
empirical equations are Kostiakov's equation, Horton's equation and Holton's
equation.
1 - Kostiakov model: Kostiakov model is a simple infiltration model relating the

infiltration rate I to time which was presented by Miyazaki as

where B and n are constant which depend on the soil and initial conditions and
may be evaluated using the observed infiltration rate-time relationship (Rawls at
al., 1993).

2- Horton model: Horton (1 940) showed that when rainfall exceeds the infiltration
rate. water infiltrates surface soils at a rate that decreases with time. He presented a

three-parameter empirical infiltration model in which the infiltration capacity I, to


time relationship may expressed as:

wheref is the maximum infiltration rate at time t,

1; is the

final infiltration rate

(the saturated soil infiltration rate) .fo is the initial infiltration rate at time equal
zero, K is a constant defines function f; and t is time. Wide scale application of
Horton's model is limited because of dependence of the parameters on specific soil
and moisture conditions (Rawls et al., 1993).

3- Holton model: Holton model was developed based on that the soil moisture
storage, surface-connected porosity. and the effect of root paths are dominant
factors influencing the infiltration capacity i,. Holton model may expressed (Rawls
et al., 1993) as:

in which f is the infiltration rate (inlh), GI is the growth index of crop in percent
maturity varying from 0.1 to 1.0 during the season, A is the infiltration capacity (in

'

h-I) per (in)' of available storage and as index representing surface connected

porosity and the density of plant roots which affect infiltration S, is the available
storage in the surface layer in inches and f,is the constant infiltration rate when the
infiltration rate curve reaches steady infiltration rate (Figure 2.2).

On thc other hand, the Green-Ampt cquation and Philip model are the most
used physical modcls (Rawls ct al, 1993). Grecn-Ampt lnodcl is the first equation
of infiltration based on a physical model (Miyazaki ct a1 1993), in which the
infiltration rate is given by:

where K, is the saturated hydraulic conductivity, Ho the pressure head at land


surface (which equal to the depth of ponded water), H I the effective pressure head
(negative value) at the wetting front, and Ls the distance from the land surface to
assumed wetting front. The model assumes that infiltrating water uniformly wets
to a depth and stops abruptly at a front. This front moves downward as infiltration
proceeds. The soil above the wetting front is in the satiated wet condition
throughout the wetted zone. The equation to describe infiltration (Miyazaki et al.
1 993) is:

I:

I = K,t+ Aln l + -

Philip model: Philip ( 1 957) shows that one-dimensional infiltration under ponded
condition would be described by a simple and rapidly converging power series in
t'

'.

Philip equation (1957) is a commonly employed equation for extrapolating

infiltration results to steady state, which refers to the isothermal movement of


water into homogeneous soils in one dimension. Philip (1957,69) showed from

purely theoretical analysis that, for long infiltration times, the transmissivity term
( A ) is thc infiltration model:

where. Z is the measured cumulative infiltration (L). at time (T): S is the sorptivity

(LT' ') and A is the field transmissivity (LT-')that approaches K,. Limitations to
this equation include it is not valid when the water flow is not linear. However.
deviation from linearity may be small in many cases, particularly in the later stage
of infiltration. And the equation can yield sufficiently accurate results (Landon,
1991).

The cumulative infiltration, Z, increases with time as infiltration proceeds.


A consequence of the increase in the cumulative infiltration is that the infiltration

rate, I, decreases with time. As the cumulative infiltration becomes large and the
depth o f wetting considerable, the infiltration rate should approach the value of the
hydraulic conductivity for the satiated condition. Both parameter A and S in the
equation above are functionally related to Ks and these parameter were determined
by fitting Eq. 2.12 to the experimental data. The results, expressed in terms of
accumulated infiltration Z, in mrn, for different time intervals. in minutes, are
plotted on log-log paper with time on the abscissa and of infiltration, Z on the
ordinate axes. If a straight line can be fitted through the points its slope be
determined. This will give the equation:

Lo@

= a log t

+ log

Then

The constant K I can be dctennined by substituting any pair of known values for Z
and t. This is the accumulated infiltration equation. If this function is
differentiated, the rate of infiltration, I, in rnm /h for different values of time can be
determined, as follows:

Guelph permeameter method (GP)

In the CHWP test. such as GP, steady flow out of well into surrounding soil
can be described in terms of pressure and gravity induced fluxes (e.g. Reynolds et
al.. 1985; Reynolds and Elrick, 1983). Refemng to Figure 2.3 water flow out of a
well by radial fluxes pressure (V,)

through the wall, by vertical pressure flux,

(V,,) through the base, and by gravitational flux (V,) through the base. Assuming
the flow out of the well to be steady state flow and the soil around the well to be
homogenous, isotropic. rigid porous medium borehole. Summary of the
approximate open borehole formulas in connection with water table condition can
be found in Stephens and Neuman (1982 a, b).

Fonnulae for estimating the soil K,

[LT-'1

from the steady-state tlow were

derived under the assun~ptionthat all flow in the soil away from the borehole was

saturated flow. These fonnulac arc commonly refcrred to as saturated tlow


equations. Scvcral literature h a w revicwcd thcsc fonnulac ( e . g . US Bureau of
Reclamation 1965. 1978: Bouwcr and Jackson, 1974; Stcphen and Neuman 1982a

b; Amoozcgar and Warrick, 1 986; Elrick et al., 1 993).

Talsma and Hallam ( 1980) used Zangar's equation to calculate the soil K
from thc values of the steady infiltration rates into borchole. Q, the hole radius, r,
and the wetted depth as follows:

The Zangar (1953) equation. based on an assumption that, there is no


impermeable layer at or near the bottom of the borehole as cited by Talsma and
Hallam (1 980) as follows:

K = 3 0 ln ( H / r ) / n H ( 3 H

2s)

[2.17]

in which S is the distance of an impermeable layer to the bottom of the borehole.


Talsma and Hallam (1 980) suggested that, for S=O, calculated K-values using the
Eq. 2.1 7 are about two times those calculated with Eq. 2.16.

According to US Bureau of Reclamation (1978). Glover (1953) solved the


three-dimensional flow problem around a vertical cylindrical source. One of
Glover's solutions developed mostly by the US Bureau of Reclamation has the
form:

Saturated zone
Wetted unsaturated zone

Figure 2.3: Approximation of steady state flow out of a well situated in a


homogeneous isotropic porous medium ( Xiang, 1994)
soil surface

Saturated bulbs

H2

I a1

wetting front

Figure 2.4: Illustrates the constant head borehole test (GP) in vadose unsaturated)
zone using two heads, H I and HZ. Where, a is the radius of the well and
GP represents Guelph Permeameter (Elrick, et al., 1993).

whcre Q is thc steady flow rate into thc borchole (L-'T-I); H is thc constant dcpth
of water in the boreholc (L); and HLl

(LL-I)

is the ratio of water depth to thc

borehole radius r (L).

This solution for calculation of the soil hydraulic conductivity, K,, from
constant head well permeameter methods were developed by ignoring the effects
of unsaturated flow (i.e capillarity). On the other hand, several authors have
reported that the flow rate from a cylindrical borehole above a water table can be
significantly affected by capillarity (e.g. Stephen and Neuman, 1982a; Reynolds et
al., 1985). According to Stephen and Neuman (1 982a, b) the wetted region in the
constant head penneameter test consists of a small-saturated zone adjacent to a
well, surrounded by a much larger unsaturated envelope (see Figure 2.3).
Therefore the realistic description of the steady-state flux out of small well into
the vadose zone must take into account of the flow in the unsaturated zone that
enveloped the saturated zone

Stephen, et al., (1982b,c) developed an empirical solution for deep water


table condition to calculate K, with improved accuracy based on saturatedunsaturated simulations in homogenous, isotropic soils. They pointed that it is
extremely difficult to develop a completely general formula because the effects of
borehole

geometry,

unsaturated

soil

characteristics,

water

table

depth,

heterogeneity and anisotropy. Reynolds et al., (1 983) derived analytical solution


which neglects capillarity (the so-called "Laplace equation" in the early work by
Reynolds and Elrick, 1985). Solution obtained by regession analysis which take

into consideration (account) capillary effects was developed by Stephen et a].,

l~nprovedsolutions that have taken saturated and unsaturated flow into

consideration have recently developed for estimating K, (Stephen and Neuman,


I982 bc; Reynolds ct al., 1985. 1992; Reynolds and Elrick, 1986. 1987; Elrick and
Reynolds, 1986: Elrlck et al.. 1990).

Reynolds and Elrick (1985) expanded earlier work by Reynolds et a]..


(1983) to develop solution of Richards equation which produce more accurate
estimate of Kk and also provides a new method for in situ estimation of the soil
sorptivity (s) and the a-parameter of exponential hydraulic conductivity-pressure
head (Y) relationship. This new method now knows as the GP method (Reynolds
and Elrick 1986, Reynolds 1993b).

GP technique is based on the assumption of three-dimensional steady state


theory of saturated-unsaturated flow from a cylindrical test borehole in the
unsaturated soil. Elrick et al., (1993) schematized the water flow from uncased
cylindrical borehole excavated in unsaturated (vadose) zone in the case of GP in a
way is shown in Figure 2.4.

By the modification of the earlier work of Reynolds et al., (1983) that only
considers saturation flow. Reynolds et a]., (1985) proposed the sequential
measurement of steady state flux out of an borehole, Q [L'T-'I, at two constant
heads levels, H [L] and using Richards equation to simultaneously solve for soil

Kt, [LT"] from thc two measurements. The relationship proposed by Rcynolds

and Elrick ( 1 985) takcs thc form:

where Q [L'T

'1 is the stcady rate of water flow out the penncameter

and into the

soil. H [L] is the stcady depth of ponding (head of water) in the hole. a [L] is the
radius of the test borehole. C is a dimensionless parameter which is primarily a
function of the H/a ratio with second dependence on soil type Q,

[L'T-'1

is the

rnatric flux potentral as defined by Gardner (1 958):

The Eq. [3.19] rearrange

where a*

[L-'1 is the proportionality

constant that is dependent on structure an

texture properties of the soil. Tables, graphs and equations given in Soilmoistu
Equipment Corp, (1 987). Elrick and Reynolds. (1 992). Reynolds et ai. (1 992), a
Reynolds (1993) allow simple evaluations of both C and a* equal to 12 m-'
compute the KI-, was suggested by researchers (e.g. Reynolds and Zabchuk. 1'
Reynolds et al. 1992; Rc,vnolds, 1993b). Values a* may also be obtained by
the siinulteneous equation approach (Elrick, 1988). Steady state rates QI ar
were measured from the same borehole at H I and H 2 and the re:

simultaneous equations were solved for both Kfs and $, Then the value of a was
computed from the relationship:

The steady-state discharge Q, in Eq. 2.21 is usually determined using the


relationships presented in G P method operating instruction book (Soilmoisture
Equipn~entCorp. 1987):

R,

( A L / A ~ ) , = R , = cons tan

and Q ,

1;

i24

R,A

where R [LT-'1 is the rate of fall of water level in the GP reservoir ( R I = transient
rate: R,= constant rate), A [L'] is the cross-section area of the reservoir. and L[L]
is the water level in the reservoir, and the t [TI is time. Steady state flow is
usually assumed when RI is estimated (R,) over four (i=4) consecutive time
intervals (At) (Soilmoisture Equipment Corp, 1987; Kanwar et al. 1989; Wilson et
al. 1989; Campbell and Fitton. 1994; Bagarello and Provenzano. 1996: and
Bagarello, 1997).

The equation proposed by Reynolds et al. (1 985) Eq. 2.19 was derived by
assuming a homogenous. isotrop. rigid soil characterized by a uniform yrl value
(Heinen and Raats. 1990: Reynolds and Elrick. 1987). Moreover. it is a solution of
a three-din~ensional.stead! state infiltration equation in which the first term on

the left side represents the water flow into the soil due to the hydrostatic pressure

of the water in the wcll. The second tenn represents the water flow through thc
bottom of thc well due to g-avitational pull, whilc the third term represents the
water flow out of the well due to the capillary forces in the soll. Both thc first and
thc second tenn In the above equation represent the tield-saturated component o l
flow out the well, whilc the third part represents the unsaturated flow components
(see Figure 2.3). The forces were also observed by Xiang, 1994; Elrick et al..
1990, 1993; Reynolds and Elrick, 1987, 1990: Elrick ad Reynolds, 1986. and
Re,vnolds et a]., 1985.

As both Klgand $, are unknown in the Eq. (2.19), Simultaneous Equations


(SE) based on two or more ponded heads of water in well, Least Squares, and
Single height (SH) analysis procedure were proposed for exact solution for Kfs
(Elrick et a]., 1993: Reynolds, 1993b; Reynolds et al., 1992; Reynolds and Elrick,
1985, 19875 Reynolds et al., 1985, 1992).

In two ponded height (HI and H2) technique (as in the case of the GP
method) producing two steady recharge Q I and Q?; Kcs, ,$,

S, and a* parameters

are computed from the solution of simultaneous equations (SE) as:

where

where Q,, and Qsz are the steady-state flow rate out of the well corresponding to
the steady depth of ponding, H I and H2, respectively, CI and C2 are the values
corresponding to H ,/a and H2ia, respective1y.

However. use of the SE approach by taking steady rate Qs (L'~T-')


measurements at two different heights based on equations 2.26 and 2.27 often
resulted in calculated values of Kk or

4,,,

that are invalid, i.e., negative values

(Reynolds et al., 1996; Wilson, 1989; Elrick and Reynolds 1992, 1993; Reynolds
et al.. 1992; Salverda and Dan, 1993; Wu 1993; Bagarello and Provenzano, 1996).
Soil heterogeneity may be the reason to cause the erroneous (negative) Kt, values.
On the other hand. the simultaneous-equations approach (Elrick and Reynolds,
1986; and Soilmoisture Equipment Corp, 1987) works well only in homogenous
soil (e.g. Reynolds et al., 1983; Elrick and Reynolds 1992). Fortunately, several
approximate solutions that avoid the problem of negative values are available.
Among these solutions are the Laplace analysis (Reynolds and Elrick, 1985), the
regession-based Richards analysis (Reynolds et al.. 1992) and the single-head
Richards analysis (Elrick et al., 1993). The later solution was used in this study.

The single head (SH) approach that proposed by Elrick et al., (1989) for
determining Krs and $,

as cited by Reynolds et al., (1992), and Elrick et al.,

( 1993) involves rewriting Eq. (2.19) in the form:

where H (L) is the constant height of ponded water in the well, a (L") is the well
~ = [ ( 2 n a ~ ' / ~ + n a+
~ 2) na H*/ C ] & ,

radius. C is dimensionless shape parameter corresponds to H/a values, and is


determined graphically for selected soil structural classes (see Figure 3.3 Chapter
I l l and a* (L-I) is the sportive number. In order for the single-head approach to

produce usable results, both the range and sensitivity of a* in porous media
should be small relative to those of Kfs and ,$, (Reynolds et al., 1992).

The regression-based Richards analysis (Reynolds et al., 1992) determining KGp


by applying the relationship:

K, =P K r

where KR (m

S-I)

is the regression-based Richards estimate of KGP,P and o are

dimensionless empirical parameters, and KL is defined as:

where KI (LT-') is the Laplace estimate of Kc,, , Q ( L T~I )

is the steady state rate

of flow out of the well into the soil, H (L) is the steady depth (head) of ponded
water in the well, a (L) is the well radius, and C is dimensionless shape factor that
depend primarily on the H 1 a ratio. The

P and

w values are determined via the

least squares regression relationship between K I and Krs; where Kfs is determined
using the multiple-head procedure of Reynolds and Elrick (Soilmoisture
Equipment Corp. 1987).

Further theoretical details of the mathematical formation and the numerical


simulation regarding the use and applicability of the GP can be found in Reynolds
et al., (1983, 198.5): Reynolds and Elrick, ( 1 986 and l987), Elrick and Reynolds,
(1 986 and 1987).

Factors Affecting Soil K- Values

Soil hydraulic conductivity measured by various approaches direct and


indirect depend on several factors including grain size, grain shape, soil structure,
and degree of cementing (Bagarello et al. 1999; Bagarello and Provenzano, 1996;
Stephens and Newman 1982abc; and Hillel, 1980). Other factors, such as the
dominant ion the exchange complex, presence of chemical cementing agents, and
type of clay minerals present, also influence soil hydraulic conductivity.

1 - Soil texture and structure

Soil texture refers to the relative proportions of the various size groups of
individual soil. Soil structure on the other hand, reflects the way by which these

particles are arranged. The K, values are often well correlated with soil texture and
structure and greatly affected by the presence and characteristics of bio-pores (root
channels, wormholes and other small conduits left by biological processes in the
soil). The values of K, in soils vary within a wide range of several orders of
magnitude. It is perhaps the most important soil characteristic depending on the
soil fractions. Table 2.2 lists the range of expected values of K for various soil
texture several orders magnitude, depending on soil texture.

Table 2.2: Hydraulic conductivity values of saturated soils (after Youngs 1991)
Soil

Soil K, (m s-')

Fine- texture soils

< 0.10

Soils with well defined structure

0.10 -10

Coarse-texture soils

> 10

The magnitude of the K,, which is the measure of soil hydraulic conductivity in
saturation condition generally. is high in sandy soils (Ks = 86.4 to 8.64 m day-'
which is pervious or permeable, and is lower in loam soils, (Ks = 8.64 x lo-' to
8.64 x 1 0-2 rn day -I). In non-agricultural clay soils the Ks may lower than 8.64 x
m day-'. When the soil K, is lower than 8.64 x lo4 m day-' the soil may be
considered impervious or impermeable. F A 0 (1963) rates soils into six hydraulic
conductivity classes as shown in Table 2.3

On the other hand. Cliilds (1969) has shown that the Ks of the soil is
dependent on the soil structure. He concluded that it is hazardous to correlate the
conductivity to the soil texture alone. Furthermore, Reynolds and Zebchul (1996)

stated that the spatial changes in texture and structure could have a large
influence on Kfs.
Table 2.3: Classification of soil hydraulic conductivity values according to F A 0
Hydraulic conductivity ( K )

(m day-')
< 0.2

0.2 - 0.5
0.5 - 1.4
1.4 - 1.9
1.9 - 3.0
> 3.0

conductivity class

( m m h-I)
<8
8 - 20
20 - 60
60 - 80
80 - 125
> 125

Very slow
Slow
Moderate
Moderately rapid
Rapid
Very rapid

A more detailed list of relationships between the soil K, and soil texture, and soil

structure characteristics are presented in Table 2.4

2- Soil heterogeneity

Soil heterogeneity (i.e. soils are variable by nature, composed of various


horizons (layering) among other factors that decrease the accuracy of an in situ Ks
measurement particularly, by constant GP method (Elrick et al., 1990; Wilson,
1989; Lee et al., 1985). Many researchers studied the impact of soil heterogeneity
on the analysis techniques. They pointed out that due to soil heterogeneity
negative values, unrealistic values, of K, and matric potential maybe obtained
frequently by the GP method using the simultaneous equations approach (Lee et
al., 1985; Reynolds et al., 1985; Elrick et al., 1990; Wilson, 1989; Ragab and
Cooper, 1 990; Reynolds et a]., 1992; Salverda and Dane. 1993; and Bagarello and
Provenzano. 1996).

'able 2.4: Annroximate relationshins between texture. structure, and hvdraulic conductivitv (After Landon. 1991)
Soil structure

Soil texture

Coarse sand, gravel

Single grain

Medium sand

Single grain

Loamy sand, fine sand

Medium crumb, single grain

Fine sandy loam , Sandy loam

Coarse, subangular Blocky and granular, Fine crumb

Light clay loam, silt, silt loam, Very fine

Medium prismatic and Subangular blocky

sandy loam, loam


Clay, silty clay, sandy clay, silty clay loam,

Fine and medium prismatic, angular blocky, platy

clay loam, silt loam, silt, sandy clay loam

Clay, clay loam, silty clay, sandy clay loam,

Very fine or fine prismatic, angular blocky, platy

Clay, heavy clay

Massive, very fine or fine columnar

Indicative Soil hydraulic conductivity


(cm h-')
(m day-')

3- Smearing, remolding and siltation

Additional factors include smearing and/or compaction of well walls during


the augering process; siltation of the soil surface during the course of the
measurement (Wu et a]., 1993; Elrick et al., 1990; Heiner and Raats, 1990). Smearing,
remolding and siltation of infiltration surfaces can strongly reduce Q, even to the
point Qs = 0 in extreme cases (Elrick et al., 1990). Asare et a1 (1993) reported that
smearing and compaction in the auger holes are some of the basic problems associated
with use of the GP especially on wetter conditions and in soils with higher clay
content. Another works by Reynolds (1993) and Campbell and Fritton (1994) pointed
out that smearing/compaction of the soil during the augering process may destroy the
soil structure along the well wall and thereby result in the Kfs values that are
unrepresentatively low.

Bagarello and Provenzano (1996) have experimentally showed that the initial
water content significantly affected the GP results. They found that ISfs values
decreased when antecedent soil water content increased. They attributed that to the
smearing and/or compacting of well walls during the augering process.

4- Air Entrapment

lnfiltration into unsaturated soil has been known to entrap air in the pore
space (Reynolds and Elrick, 1985; and Lee et al.. 1985). The presence of

entrapped air has been found to cause well permeameter methods to


underestimate the completely saturated Ks by about a factor of 2 or more
(Bouwer, 1966; and Stephens et a]., 1987). It has been argued, however, that Ks
measured with entrapped air present value of Kr, is more appropriate for
unsaturated (vadose) zone applications than Ks from a completely saturated
condition (Bouwer, 1986; Reynolds and Erick, 1985 and Reynolds et al., 1983).

Using constant head percolation test such as GP method in capillarity


greatly soils (high content-clay) may give overestimated Ks when the capillarity
was not taken into consideration (Jenssen, 1990). Factors decreasing the accuracy
of an in situ

K,measurement by the GP method include, attainment of true steady

state flow rate (Q,). the persistence of capillarity in three-dimensional flow, and
the range and resolution of the apparatus for measuring the flow (Elrick et al.,
1990; Talsma, 1987). Field, laboratory and theoretical works suggested the
neglected capillarity in the GP method analysis cause

K,to be overestimated by

factors ranging from 1.0 to more than 10 depending upon soil properties, soil
wetness, and source geometry (Talsma, 1987; Reynolds and Elrick, 1985; 1987).

6-Solution analysis method

Shamsai and Sitar ( 1 991) stated that underestimation of Ks to the


inadequacy of the field proccdures and some of the assumptions inherent in the
analysis of the shallow-well permeameter tests. Another work by Bagarello and
Provenzano (1996) showed that the high variability of K, was obtained by
applying the simultaneous-equation (SE) analysis. The same result was also
obtained by other authors, conclusion that estimates of Kfsvariability based on the

SE calculation are generally unrepresentative of true field variability (Elrick and


Reynolds, 1992; Kanwar et al., 1989). On the other hand, early study by Elrick et
al., (1 990) reported that for both, the one-dimensional and three-dimensional flow
analysis, however the degree of error introduced in Ks due to overestimating Q, is
small relative to the errors induced by capillarity, soil heterogeneity, smearing,
remolding, siltation, and air entrapment. They also pointed out that the
unreasonable (negative) values of Kfs and $, from the solution of equations used in
two-ponded height technique are a matter of general occurrence in heterogeneous
soils.

Summary

In conclusion, there are various methods and techniques available for


determining a soil hydraulic conductivity. These basically fall into two types of
techniques; either direct or indirect. Numerous direct (field and laboratory)
methods have been developed and improved for rapid and easy to use for K,

measurements. However, it is not obvious that any of these mcthods will work
under all circumstances. Unfortunately, most of them havc bccn applied with
varying degrecs of success and often yicld substantially dissimilar soil K, values,
since the soil K, properties are highly variable. In addition, the problems of
detennmng the soil K by these mcthods, particularly field methods. is
compounded by the expense of experimentally obtained values, the equipment
required, the time involved and a large number of observations are required to

adequately characterize their spatial and temporal variability distribution. As a


consequence, the need to make new measurements continues to be critically
evaluated considering both desired accuracy of the soil Ks and the available
financial means to measure it. In this context two complementary approaches may
be followed to achieve a more rapid and less expensive Ks characterization of soils.
One is to use simple field methods of measuring soil Ks properties. The second is
to estimate soil K, from other easily obtained soil properties or limited data.
Various empirical models have been developed for estimating soil Ks from basic
soil properties. These PTFs methods are simple, cheap and easy to use and take
less time to compute soil K, values. However, because of heterogeneity, irregular
geometry soil structure, the predictable methods may not represent the field values.
In addition, PTFs still show a considerable amount of unresolved availability in
estimating soil K, value. Nevertheless, direct measurements remain indispensable,
because they are only experimental check for model-functions. The need for direct
measurement has stimulated the development and improvement of various
methods to determine soil K,. Currently GP method and SC method for in situ and
Laboratory measurements, respectively, are widely used. Undoubtedly there are
more field, laboratory and predictive models techniques for estimation of soil

saturated hydraulic conductivity present than the ones mentioned here. However,
the success of the individual methods varies. Often a method does fairly well in a
localized area. No one method works really well for all soils. Sometimes,
measurement of the predictor variables is more difficult than measurement of
hydraulic conductivity. At present, there is not much agreement which
lneasuremcnt technique may be the best for a given purpose. Further. it is very
difficult to compare different technologies because there is no independent soil K,
datum or benchmark upon which evaluations or judbments can be made. The soil
researcher must use best judgment based on experience, local conditions and the
observed behavior of the particular soil.

CHAPTER 111
MATERIALS AND METHODS
I -Site description

To meet the objectives of this study, field experiment and laboratory analysis
were carried out. About 1200 m2 experimental area consisting of five adjacent plots
measuring 20 m x 12 m was used for this study. Figure 3.1 and plate 3.1 show the
study area. Figure 3.2 shows the plot indicated as basin irrigation area at the Field
Station of the Department of Biological and Agriculture Engineering UPM.
The study was performed in a Serdang series soil classified according to the
soil taxonomy as a member of the fine loamy siliceous isohyperthermic family of the
Typic Paleudult (Paramananthan 1978). According to Abd. Manan (1 982) the series
was developed on sandstone parent material and occupy an extensive area in the
University farm. The soils of this series is one of the most extensive soils in Selangor
and occur on undulating and rolling to hilly land but mainly on rolling terrain (Wong

The soils have been mapped in every state in Peninsular Malaysia except for
the state of Perlis (Paramananthan 1978). They are characterized by the surface soils
with sandy loam to sandy clay loam in texture while in the subsoil is sandy clay loam
to sandy clay (Abd. Manan, 1982; Wong, 1970). The texture often becomes heavier at
depths around 1 111 (Paramananthan 1978).

KEY:

Figure 3.1 Soil map showing the location of Serdang series at UI'M campus (after Para~nanathanet a1 1979)

Plate 3.1 : The study area at DBAE Field Station, UPM

SKALA

800

Figure 3.2: Contour map of the Field Research Area Faculty of Engineering showing
the study area (Basin irrigation area)

They range in colour from uniform yellowish brown to brownish yellow and with
depths the colours may be strong brown, reddish yellow (Paramananthan 1978).
Figure 3.1 shows the distribution of Serdang Series soils at UPM indicated as
110.3. Four pits about 2.0

111

long, and 1.0 nl wide and I .5 m deep distributed

throughout the study area were excavated for soil description and collecting
undisturbed soil core samples for bulk density and K, measurements, (Figure 3.2).

2-Experimental Design

Three methods were used for field-saturated hydraulic conductivity


measurements. In situ measurements were made using the Guelph penneameter (GP)
and double ring infiltrometer (DRI) methods while the laboratory measurements were
made on undisturbed core soil samples using constant-head core soil penneameter
(SCHP). These measurements were made at five plots (each measuring about 20 m x
12 m) at the study area. The GP wells, soil cores and double ring infiltrometer were

located on the site as indicated in Figure 3.3 For the GP method about 109
measurements at four depths (0.15 m, 0.30 m, 0.60 m, and 0.90 m below the ground
surface) were made. Four measurements were carried out at 3 m intervals in each plot
at each depth (for a total of about 22 measurements site). About 70 undisturbed cores
soil samples at depths of 0.15. 0.3, 0.45. 0.60, and 0.90 nl from the soil profiles were
collected vertically for laboratory measurements of saturated hydraulic conductivity.
Two measurements for double ring were taken within an approximately 4 x 4 m area
on each plot (for a total of 10 measurements). Twelve core samples for bulk density

Figure 3.3: Symmetric plan showing the location of saturated hydraulic conductivity tests and sample collection;
the Guelph Penneameter, GP, measurements; for double ring infiltrometer, DRI, measurements. and
soil profiles location for cores samples collection

for
are

measurements were collected at each depth (for a total of 45 measurements for all
depths). The collected soil samples for bulk density measurements, laboratory
measurements of K, and from preparation of G P wells were used for texture,
porosity. and organic matter determination. The four pits were also used for soil
description.

3- Determination of soil properties

Standard methods of soil analysis were followed for soil moisture content,
soil texture, bulk density, and porosity as recommended by Klute (1 986). The
determination of loss in ipition was used for detennining soil organic matter.
The standard methods were followed as shown in the Appendix A.

4-Determination of soil hydraulic conductivity

In this study, soil hydraulic conductivity. refers to the rate of water moving
downward through a soil as measured by the Guelph permeameter method, GP,
the constant head as a laboratory method. SCHP, using undisturbed soil cores and
the infiltration ring. Also the SCHP method was conducted at depths of 15, 30,45,
60, and 90 cm below the ground surface. Saturated hydraulic conductivity, Ks, was
also estimated fi-om other soil properties using various empirical models

4.1-Guelph permeameter (GP)

Guelph permeameter (GP) is one of various types of the constant head well
penneameter (CH WP) which operates on the Maniott siphon principle to pond the

constant-head of water in the borehole and measure the water rate flowing out thc
borehole (Reynolds and Elrick, 1985). The permeameter is constructed of two
concentric acrylic tubes where the inner air inlet tube provides the air supply, and
the outer tube provides the water reservoir and the outlet into the well. Water flows
out of the outlet tubc through a funnel shaped part located immediately above the
permeameter tip. The GP method provides a quick and simple method for in situ
field measurement of saturated hydraulic conductivity (Kk). It involves augering a
small diameter cylindrical borehole of radius 3 cm to the desired depth and
measuring the steady state flow rate of water into the hole while maintaining a
constant head of water inside the borehole.

In this study GP model 2800 KI permeameter (Soilmoisture equipment Crop;


1987) was used to determine the Kfs for Serdang series soil (Plate 3.2). The
procedure followed was taken from the instruction outlined by Soilmoisture
Equipment Corp, Manual (1 987) that was included with the GP kit. For most of all
the measurements using the GP method, the steady state infiltration rates (Q m3s-')
were measured with 5 cm and 10 cm constant heads levels on the cylindrical
boreholes (3 cm radius). At the points where two heads gave negative values of Kk
only one depth of 5 cm or 10 cm heads was used. As recommended, GP tests have
been conducted in the sunshade. using beach unmbrella, to reduce the influence of
direct sunlight and temperature fluctuations. The GP method tests were made in
the dry period (when there was no rain). The GP cylindrical boreholes were
augered to depth where K measurement was desired (0.15 m, .0.30 m. 0.60 m and
0.90 m). The wells were dug carefully, taking care to limit or to avoid possible
sealing or smearing the well surfaces by the augering process and/or remolding

and compaction of thc borehole walls. To avoid the possibility of alteration in well
characteristics each measurement was carried out immediately after having
prepared the borehole. Soil samples were collected when holes were being
prepared at each site for each depth to determine the soil particle size distribution
(% sand, % silt and % clay).

After the system was assembled, filled with water, placed in the prepared
borehole, and adjusting vertically, reading procedure was carried out first by
slowly raising the air inlet tip (by grasping the upper air tube) to establish the first
prescribed level of 5 cm. The reading of water level in the GP reservoir were taken
at constant time interval of 2 minutes. in some cases, variable time intervals were
used, until steady-state (Q,,) was obtained (i.e. R

= constant

for consecutive 3 to 4

times). The air tube inlet tip was then raised to establish the second head height
(HI) of I0 cm, and the same procedure was repeated again same as described for
the 5 cm head and reading continued until steady-state (Qs2)was obtained. Finally,

K, was calculated by an appropriate equation using Q, H, and r values. In this


study, for each well. results of the two single height analyses were averaged to
obtain an estimate of Kfs using the relationship that as outlined by Soilmoisture
Equipment Corp,(I 987):

(1) When both reservoirs were used:

and
$,,, = (0.0571 ) (A' )

( F ,)

- (0.0237 )

( K)

(X)

Plate 3.2: (A) Guelph Permeameter (GP) kit and (B) GP setup in the study area
.

.-

- -.
-- -

(2) When only the inner reservoir was used:

K 1,

(0 .OO4l ) ( Y )( F 2)

(0.004 ) ( Y )

(q)

and

Where
Kk is the field-saturated hydraulic conductivity (cm s-')

$,, is the matric potential (cm2 sec").

Rland RZare the steady-state rate of fall in the reservoir corresponding to HI and H2
respectively and converted to cmlsec.

X is the reservoir constant, used when reservoir combination is selected, and


corresponds to the cross-sectional area of the combined reservoirs expressed in cm2
Y is the reservoir constant, used when the inner reservoir only is selected, expressed
in cm'.
When a single depth of ponding is used, the single-head analyses (SH),
procedure can be applied to evaluate Kfs, (Elrick and Reynolds, 1992; Reynolds,
1993)

Whcre a* in I/m is defined as:

and where:
Qs is the steady state flow rate corresponding to the water head (H)

Kfs is the field-saturated hydraulic conductivity (cm s-I)

4,

is the matric potential (cm2 sec-I)

A is the well radius (cm)

C is a dimensionless shape factor that depends primarily on the H/a ratio. It


was obtained from Figure 3.4

H is the constant height of ponded water in the borehole (cm)

In this study, also Kfs estimated from the Eq. 3.5 using a* value equal
12 m-' Table 3.1, as suggested by Reynolds et a]., (1992); Reynolds, (1993); and
Reynolds and Zebchuk, 1996). According to Reynolds (1993); Reynolds and
Zebchuk (1996) there was no compensation made for the possible water viscosity
or air entrapped effects on Kfs.

Figure 3.4: Shape factors (C) for use in the constant-head well permeameter
method. H is the depth of water in the well, a is the well radius, 0 :
a* 2 12 cm", x: a* = 0.04 cm-' and H: a* = 0-Olcm
(After
Reynolds, 1993b)

-'

Table 3.1: Porous media categories used for estimating a* in the single-head well
permeameter analysis of Elrick et al (1989) as cited by Reynolds et al

Porous media category

Corresponding a' value


(m-'1

Compacted, structureless clayey materials such as


Landfill caps and liners; lacustrine or marine
sediments. etc.
Soils which are both fine textured (clayey) and
Unstructured.
Most structured soils from clays through loams;
also includes Unstructured medium and fine sands.
The first choice for most soils
Coarse and gravelly sands; may also include some highly
Structured soils with large cracks and macropores.

4.2-Double Ring Infiltrometer method

Infiltration rate can be measured by observing the fall of water within two
concentric cylinders driven vertically into the soil surface layer. The double ring
infiltrometer method as described by Bouwer, (1986) and Landon (1991), with
measurement confined to the inner ring, was used to estimate Ks. The double ring
infiltrometer cylinders, which are made of galvanized steel with the bottom edge
beveled from the outside to the inside, used had dimensions of inner and outer ring
of 30.5 cm in diameter by 30 cm height and 60.5 cm in diameter by 30 cm height
respectively (Plate 3.3). On each plot two infiltration runs were carried out within
I Om-diameter area (for a total of I0 sets of measurements). The woody materials

were removed from the selected locations. Then the rings were placed
concentrically on the area to be tested. The rings were driven uniformly and
straight down into the soil to approximately 10 cm depth, just far enough to
prevent lateral leakage when water was ponded in them, by a falling weight type
hammer striking on heavy timber placed on the top. The timber was rotated every
few blows and the penetration was then checked (i.e. uniform and vertically). The
inner ring was driven first followed by the outer ring. After the driving was
completed, the disturbed soil ad-jacent to the ring on the inside was stamped firm.
Both rings were filled with water to just above the reference mark; the time was
noted and the water level in the inner ring was measured using a vertical scale
fixed to a leveled beam sitting on the top of the outer ring and then recorded. The
water level measurements were taken inside the inner ring at regular intervals; the
levels were measured and recorded immediately before and immediately after each
refill. Each infiltration test run lasted for 2 to 3 hours by which time the steady
infiltration rates had been attained.

During the experiment, the water in the outer ring was kept as approximately the
same level as in the inner one to eliminate the lateral movement of the infiltrate
water in the soil, thus maintaining a one-dimensional flow condition. The
infiltration rate as a function of time was found using refill of the rings to a fixed
level at suitable time intervals. Usually the measurements were continued for one
hour after steady state appeared to have been reached. Observations of cumulative
infiltration versus time were recorded and subsequently converted to the
corresponding infiltration rates for elapsed times from the start of each experiment.

Plate 3.3 Infiltration rate measurement using double ring infiltrometer

It is generally assumed that the hydraulic gradient in transmission zone is

asymptotically approaching unity and the final infiltration rate (1,) equals K,.

The results, expressed in terms of accumulated infiltration, Z (in mm) for


different time intervals (minutes) were plotted on log-log paper with elapsed time
on the abscissa and of infiltration, Z on the ordinate axes. Those infiltration data
were analysed with Kostiakov's infiltration equation to determined K, values. The
simplest approximation of cumulative infiltration is:

and for infiltration rate is:

Where:

Z is cumulative infiltration in units of volume per unit of area


T is the intake opportunity time
K and a are empirical constants

4.3-Determination of K, using SCHP

The soil saturated hydraulic conductivity (K,) measurements were made on


the cores in the laboratory using the constant-head permeameter (SCHP) similar to
that described by Klute and Dirksen (1986). The core method normally uses a
cylindrical metal sampler that is pressed or driven into the soil to the desired depth

and is carefully removed to preserve a known volume of soil as it existed in site.


Generally cores with diameters of at least 75 cm are the most common practical
samplers used in obtaining soil cores for laboratory measurement (Black, 1 965).
'The core length should be ideally be equal to the diameter. Therefore, samplers
diameter of 7.62 cm by 7.62 cm high were used in this study.

Because of the required size (8.89 cm 0 . D and 8.5 cm I.D. size) of the thin
walled brass steel cylinder were not readily available within the laboratory, the
cylinders samplers tubes therefore. were machined from 8.89 cm (3.5" inch) 0 . D
and 7.62 cm 1.D tubing using an engineers lathe. A sharpened cutting edge was
also machined on one end of the sampling cylindrical metal tube. About 50
undisturbed soil cores were sampled from four freshly dug pits at depths of 15, 30,

45. 60 and < 75 cm. Three replicate samples were collected from two adjacent
faces of the pit at each depth (for a total of 15 cores samples per pit). The sampling
cylinders were lubricated with cooking oil and then inserted into the soil layers by
hammering gently on the top of the cylinder up to the established depth as showed
in the Plate 3.4

As the sampling cylinder was forced into the soil the surrounding soil was
removed to lessen resistance to passage. When the required sample was contained
in the cylinder. the surrounding soil was dug away to a greater depth to allow a
cutting plate to be jacked underneath, separating the cylinder sample from the soil
beneath. All cores were collected from a vertical orientation, and care was
exercised in the collection of the samples to limit the distortion of the natural soil
structure. Furthermore. the soil cores were carefully examined in the field for

cracks and other sampling-induced disturbance and cores with such features were
discarded, only samples free from any visible defects were chosen from the depth
interval. Once extracted from the soil, the soil adhering to the outside of the
cylinders was removed, and the cores were bagged in a plastic bag immediately
and brought to the laboratory.

In the laboratory, the samples were processed according to the procedure


described in Klute and Dirksen (1986). The undisturbed soil cores were leveled
with a small spatula and the upper and lower faces were carefully prepared by
removing any smeared or damaged surfaces to expose fresh surfaces before
saturating them for K, measurements. After cheesecloth (double layers) was placed
at the bottom of each sampler cylinder (using a rubber band to hold the
cheesecloth) to prevent soil loss from the bottom of the sample. the cores were
initially saturated from the bottom by soaking them for more than 24 hours.

In the measurements, the soil was cleaned from around the upper end of the
sample and a cylinder of the same diameter was connected to the top of the core to
allow imposition of the head and a filter paper was placed on the top of the sample.
The core samples were transferred and fixed onto the prepared system (see Figure

3.5). and water was gently added to give a hydraulic head in the extended cylinder.
The constant head above the soil surface was established using an overflow
technique and no significant head fluctuations were observed when the apparatus
was tested. Then the apparatus was allowed to attain steady state flow conditions
for one hour before outflow volumes were measured. The volume of percolation

Plate 3.4: View of the undisturbed soil core samples collected for bulk density, A,
and saturated hydraulic conductivity, B, in the laboratory.

water through the soil core in a set period of time was collected (using funnel) and
recorded. Saturated hydraulic conductivity (K,) for each core samplc was
calculated Srom the flow and the head recorded by the standard constant hcad
equation (Darcy's equation 1856) as:

where V (ml) is the volume of water collected in time t (s). L is the length of
sample (cm). A is the cross-sectional area of the sample (cm'), and AH is the head
difference causing the flow between the top and the bottom of the sample (cm).

5-K, estimation methods

In addition to determining the field Ks several methods of estimating the


saturated Ks based on soil properties were explored. While most of these methods
were designed to estimate values of the K, in this study they were compared to the
measured Kfs. Two approaches may be followed to predict Ks from other easily
obtained soil properties. either estimating Ks from panicle-size distribution (e.g. %
Sand. % Silt. and % clay) using multiple regression equations or based on pore
size distribution. However, selection of K prediction technique depends upon
availability and the level of information on physical and hydraulic properties of the
soil. Hence only soil PSD (% Sand, % Silt, and % Clay). Db . and E are available
for the s t ~ ~ darea.
y Therefore. in this study the first approach was used. and six
empirical models based on these parameters were evaluated.

Acrylic Cylinder

7'

Am/

Water Trough

Graduated Cylinder
Figure 3.5: The components of a laboratory constant head permeameter

Models to predict the Ks investigated here were those required value by


(generally multiple non-linear) regression statistics from input variable, such as
percent of sand; percent of silt, percent of clay, organic matter, and bulk density
include methods from Jabro (1992), Amin et al (2001), Cosby et al (1984) and
Vereech et a1 (1990). The first model, Jabro model is based on an equation
developed by Jabro (1992) which computes an estimate of K, using the following
equation:

5.1-Jabro model

log(ks)

= 9.6 - 0.81 log(% sill)- 1.09 log(% clay)

- 4.64(bulk density )

Where

K, is a soil saturated hydraulic conductivity;

5.2- Saxton et al. (1984) Equation

In the second method Saxon equation in which K, is a function of sand and clay
content, and can be computed using the following equation;

Where S = sand content (%) (50-2000pm); C = clay content (%) (<2pm)

5.3- Vereecken et al (1990) Equation

K , = exp

20 .62 - 0 . 9 6 x l n ( ~ ) - 0 . 6 6 x l n ( ~ )
- 0 . 4 6 x In (m ) - 8.43 xb D

(cm/day

Where S = sand content (%) (50-2000pm); C = clay content (%) (<2pm);

organic matter content (%) and bD= bulk density (g ~ r n - ~ ) .

5.4- Cosby et a1 (1984) Equation

Where S = sand content (%) (50-2000pm); C = clay content (96).

5.5- Brakensick et a1 (1984) Equation

where
X = 19.52348 P-8.9684;LO.O282l3C+O.OOOl8lO?S*S-O.OOO94l26C.C

-8.395215P~P+0.077718S~P-0.00298S.S.P~P-0.01949~C~C.P.P
+0.0000173S.S.C+0.02733C.C.P+0.001434S.S.P-0.0000035C.C.S

Where P is the porosity (m3m3),S is sand content (%), C is clay content (%)

13.131

5.6- Arnin et al (2001) Equation

KS = 5.24

2.08(DB

) -

0.029 ( S t ) - 0.025 ( C ' )

Where
Ks is saturated hydraulic conductivity in cmlh, DB is the bulk density in g/cm3' St
is percent silt, and C is percent clay content.

6-Statistical Analysis

6.1-Statistical Distribution

Before parametric statistical sets can be performed, it is essential to


determine the statistical distributions of these parameters to decide whether
arithmetic or geometric means should be used in the comparison. According to
Warrick and Nielson (1980) the soil physical properties including K, are often
found to be adequately described by either normal or log normal distribution.
Therefore. a log normal distribution was assumed in all statistical analysis and all
statistical tests of the Kfs values results (comparison of means tests, regression)
were carried out using natural logs of the data. Bouwer and Jackson (1974) and
Lee et a1 (1 985) have suggested that the geometric mean should be used, thus all
averages reported in this study are based on the geometric mean, unless otherwise
mentioned.

Where the Kg is the geolnetric mean.

All the data of K1:,and $,, were transformed to logarithms (logloK,,) scale.
Using transformation to a loglo scale rather than a log, scale enables easier
interoperation (a loglo 10

1, loglo 100

2, etc.). The mean (p), standard

deviation (SD), Skewness, and Kurtosis coefficients of the parameters mean field
Kt:,; mean laboratory Ks and mean K, estimated from the models and their logarithmetic transforms were evaluated. The Skewness and Kurtosis coefficients
were used to verify the statistical distribution of the data. The Skewness coefficient
approaches to zero for a normally distributed variable. The kurtosis coefficient, a
dimensionless measure of the flatness of the distribution is 3 for a normally
distributed random variable. Hence, the closer the calculated kurtosis to 3, the
greater is the tendency toward normality. Probability plot was used to determine if
that Kfs was lognormally distributed. All statistical analyses were conducted using
loglo( Kfs). For log-normally distributed data, the F-and t-tests were conducted on
the log-transformed data. However, when two X values were found to be
significantly different the corresponding Ks values were also significantly
different. The F- and t-tests were conducted at the 1 % level or 5 % level of
significance.

Statistics for all data were performed using StatGgraphics (S-Plus) and
SPSS statistical packages for Windows version Release 3 and 1 1 , respectively.
First, average (p) and standard deviation (SD) were calculated for untransfonned
K, values. Then geometric mean value was calculated for a lognormal distribution
of K, from the average of transformed loglo (K,) values. Stepwise regression

procedure was used to dctermine the best relationship between the independent
variables (sand, silt, clay. bulk density, porosity, and organic matter) and
dependent variable (saturated hydraulic conductivity).

6.2-Comparison of methods

The GP, DRI, SCHP methods and empirical models values were
statistically compared for each depth on the basis of geometric mean (K,), range
(R), coefficient of variation (CV), and standard deviation factor (SD) of the K,
values. Statistical procedure using loglo transformed data for soil saturated
hydraulic conductivity, Ks, were compared to determine if there was any difference
according to method and the depth of determination.

The geometric mean KsGmvalues for each of the three methods were
compared on a painvise basis within each depth using two tailed multiple F-and t
tests. GP value of Kfs was compared to those values determined by different
methods using a single factor ANOVA providing the F-statistic was significant,
multiple comparisons were preformed with Tukey's

honestly significant

difference. T-tests were used to compare Kfs between depths within a given
method. A regession analysis was conducted between field-measured loglo (K,)
and laboratory-measured loglo (K,). Additionally, the correlation coefficient (R)
was used to measure the deg-ee of association between measured and predicted
values.

CHAPTER IV

RESULTS AND DISCUSSION

Field Site Characteristics

The field site is principally grassland field provided by surface and


subsurface drains. It was graded several years ago to be a basin irrigation area.
Consequently, the topsoil surface, horizon A, at the site was removed and leveled.
Plates 4.1 through 4.4 show the four pits that were dug at the study area for profile
description and intact soil core samples collection for bulk density and laboratory
determination of saturated hydraulic conductivity.

Core samples were collected at four depths except for one profile where the
core samples were collected from the topsoil only (0.0 to 0.50 m) due to the
presence of gravelly sandy clay layer.

Table 4.1 shows the results of the analyses of the collected soil samples.
Observations in 1.5 m -deep pits revealed that the topsoil and subsoil of Serdang
series in this site is a deep uniform dark yellowish brown (IOYR 414). The soil
structures of the study area are weak to moderate, medium to fine subangular to
angular blocky with abundant fine quartz at the lower parts of the subsoil.

Plate 4.2

Plate 4.3

Plate 4.4

Plates 4.1 through 4.4: show soil profiles at the study area

Table 4. i : Soil physical properties of the study area. OM organic matte, MC


moisture content. E ~orositvand DI, bulk densit;
Soil particle size
3epth distribution (microns) Soil textures OM
Db
E
MC
%
g I cm3
Class
%
O
h
cm Clay
Silt Sand
(USDA)
<2
2-50 >0.50
Plot no.1
SCL
SCL
SCL
SCL
SCL

SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL

27.75

9.00

62.30

SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL

0.5

1.645

37.92

25.58

Plot no.2

SCL
SL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
CL
SCL
SC
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
Plot no.3

SCL
SCL
SCL
SCL
SCL
SCL
SCL

CL
SCL
SCL
SCL
C
SC
SCL
SL
SCL
SCL
SCL
SCL
SCL
SC
SCL
SC

Plot no.4
SCL
SCL
SCL
SCL
CL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SC
SCL
SCL
SCL

SCL
SCL
SCL
SCL
SC
SCL
SCL
Plot no.5

SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SCL
SC
0.15 1.390 47.55 19.00
SC
0.88 1.560 41.13 24.20
SC
1.03 1.510 43.02 24.30
SCL
0.44 1.350 49.06 18.87
SC
1.04 1.522 42.56 18.35
44.00
7.7 48.00
0.95 1.409 46.83 20.50
SC
Soil classification classes according to United states Department of Agriculture
(USDA. C= Clay, CL=clay loam, SC=sandy clay, SL=sandy loam, and SCL=
sandy clay loam

As evidence from Figure 4.1 the soil profiles are characterized by a top 0 to
0.90 m sandy clay loam soil texture in accordance with the USDA soil texture
classification (sand

58.28 * 8.20, clay 28.93

* 5.33 and silt 12.1 f 5.28). Layers,

of sandy clay and clay texture existed as well, however, as evidenced by the large
maximum values for sand and clay content (73.57 % and 44.0 %, respectively).

Minimum, maximum, mean, p, standard deviation (SD) and coefficient of


variation (CV) values of sand, silt, clay, bulk density (Db), porosity (E), organic
matter (0.M) content and soil moisture content (MC) by depth for the pits sampled

in the experimental area are given in Table 4.2

The OM was generally low (mean

1.55 %) as might be expected for the

area in which the top surface have been removed. Other soil profile description
such as, texture colour, root, mottles and fauna are shown in Figure 4.2.

clay

- ,

10

20

30

40

50

60

70

80

% sand

Figure 4.1 Textural distributions for study area data set

90

100

Table 4.2 Physical properties of Serdang Series soil at the experimental site across
at each depth
parameter*

N'

Mean
value

Clay ( < 2pm) %


Silt (2-50 pm) %
Sand (>50 pm) %
OM %
D~ g ~ m ' ~
E%
MC %

19

25.79
1 3.68
60.35
2.002
1.548
41 S 8
17.0

Clay ( < 2pm) %


Silt (2-50 pm) %
Sand (>50 pm) 5%
OM %
Db g cm -3

E%
MC %
Clay ( < 2pm) %
Silt (2-50 pm) %
Sand (>50 pm) %
OM %
D~ g cm-3

E%
MC %

19
19
19
19
19
19

28
28
28
28
28
28
28
21
21
21
21
21
21
27

SD"

Minimum Maximum CV'


Value
Value
%

Depth 0- 15 cm
6.198
18.0
7.08
6.2
12.01
29.89
0.575
0.9
0.0985 1.39
0.852
35.63
3.070
12.23

Depth 15-30 cm
26.82
4.482
19.6
11.31
1.046 2.7
60.55
8.249 40.1 1
1.856
0.562
0.73
1.623
0.131
1.38
38.68
4.917
31.2
18.7
2.124
13.9
Depth 30-60 cm
30.47
3.028
26.35
13.1 1
3.918
8.5
5.444
39.75
55.4 1
1.7
0.69
0.8
1.60
0.097
1.362
39.65
3.66
35.3
20.57
1.979
16.79

Depth 60-90 cm
32.36
4.739
22.74
25
10.91
4.106
2.4
25
56.56
5.373 49.4
25
0.724
0.433
0.26
25
1.554
0.104
1.255
25
41.32
3.93
36.8
5%
25
~ M C%
34
20.51
2.33
13.02
25.58
11.38
+OM =organic matter, Db = bulk density, E = porosity and MC = moisture content.
#Number of measurements.
YStandard deviation
5 Coefficient of variation
31ay ( < 2pm) %
silt (2-50 pm) %
sand (>50 pm) %
3M Yo
lb
g cm-3

Soil series: Serdang series


Classification: fine loamy, siliceous, isohyperthermic, typic Paleudult.
Location: Faculty Of Engineering research area DBAE Field Station.
Physiography: lower mid and footslope of the hill.
Vegetation: shrub and grass
Groundwater: deep

Profile 1

Horizon

BI

Mottles

0 -5 cm
Dark yellowish
brown ( I OY R 414)
Very few

5-100 cm
Dark yellowish
brown ( I OYR 414)
Very few

1 00+
Dark yellowish
brown ( I OYR
Very few

Texture

Sandy clay loam

Sandy clay loam

Sandy clay loam

Weak to moderate
Subangular blocky

Moderate
medium
subangular to
Abundant

Depth (cm)
Colour

1 Structure
Gravel

Very few

Moderate medium
subangular to
angular blockv
Abundant

Fauna

Many pores

Many pores

Few pores

Roots

Many

Few

No

1I

Profile 2

Mottles

I---

-I

Sandy clay loam

Gravel

Weak to moderate
Subangular blocky
Abundant

Roots

Very few
I

Texture

I-

Dark yellowish brown ( I OYR 4/4)

Dark yellowish
brown (1 OYR 414)
Very few

Sandy clay

I Crumbles and angular blocky


Abundant

Macro pores

Many pores

Fine t to medium

Fine

Horizon

1 Depth (cm)

1
0 -1 5 cm

15-55 cm

55+ cm

Dark grayish brown


( I OY R 412)
Very few 1 %

Yellowish brown
( I OYR 518)
Very few 7%

Dark yellowish
brown (I OYR 414)
Iron concentration

Sandy clay loam

Sandy clay loam

Sandy clay loam

Structure

Crumber structure

Subangular
blocky (massive)

Gravel

Many small size

Fauna

No

Colour
Mottles

i
I

Profile 3

I Roots
I

Subangular blocky
structure

No

I Fine to medium

I No

No

I No
I

Profile 4

Horizon

/ Depth (cm)

( 0 -5

, Colour
I
,

cm

1 5-100 cm

loo+

I Dark yellowish

Mottles

Yellowish brown
(1 OYR 518)
Very few

Dark yellowish
brown ( I OYR 414)
Very few

Texture

Sandy clay loam

Sandy clay loam

Sandy clay loam

Structure

Weak to moderate
Subangular blocky

Subangular blocky

Subangular blocky

Few

No

/ No

No

No

No

brown (I OYR 414)


Very few

Fauna

Few
No
No
Roots
1
Figure 4.2: Soil profiles description of study area located at Faculty of engineering
research area DBAE Field Station.
I

Results of Soil Properties and Infiltration Rate

Soil Moisture Content

In the determination of the soil moisture content (MC) in field condition


using gravimetric method, the results showed that the variation of the moisture
content ranged from 12.23 % to 25.58 %, with the mean value of 19.8 %. The
standard deviation was 2.69 %

At the topsoil layer (0 to 0.15 m), the moisture content varied from 12.2 to
22 % with the mean value of 17 % . The standard deviation was 3.1 %.

The moisture content of the layer 0.15 to 0.3 m varied from 13.9 % to 22.3
%. The mean value of the moisture content was 18.7 %. The standard deviation

was 2.12 %.

At the depth 0.3 to 0.60 m, the moisture content ranged from 16.8 % to
23.6 %, with the mean value of 20.6 % and standard deviation of 1.98 %. At the
deep layer (0.6 to 0.90 m), the moisture content varied from 13.02 % to 25.58 %.
And the mean value was 20.5 %, and the standard deviation was 2.38 %.

The organic matter

The results showed that the organic matter (OM) was generally low. It
ranged from 0.26 to 3.46 % and the mean value was 1.55 % with the standard
deviation of 0.76 % and within the expected range for sandy clay loam soil.
At the depth 0.15 In the organic matter content ranged from 0.9 to 3.46 %
with mean value of 2.0 %. The standard deviation was 0.57 %. The organic matter
content of the depth 0.1 5 to 0.30 m varied from 0.73 to 2.87 % with mean value of
1.85 %. The standard deviation was 0.56 % .

At the depth 0.60 m, the organic matter percentage was in the range 0.8 5

to 3.34 %. The mean value and standard deviation were 1.7 % and 0.69 %.
respectively.

The percentage of organic matter of the 0.6 to 0.90 m depth layer varied
from 0.26 % to 2.12 % with the mean value of 0.72 %. The standard deviation was
0.43 %.

The mean percentage of organic matter at each depth was found to be 2.0
%, 1.86 %, 1.69 % and 0.7 % at 0.15 m, 0.30 m, 0.60 m and 0.90 m, respectively

showed that the organic matter content decreased with increasing soil depth.

Bulk Density (Dh)

The soil samples collected using both the brass copper rings with size o f 7 4
mrn diameter and 40 mm high and brass steel cylinders with size of 76.2 diameter
and 76.2 mm high were used for determining the soil bulk density (Db) as
mentioned in the procedures of chapter 11I.

The bulk density obtained in this study ranged from 1.26 to 1.82 g ~ m - ~ .
The mean value was 1.585 g cm" and standard deviation of 0.1 12 g ~ m - ~ .

The bulk density values at the depth of 0.0 to 0.1 5 m ranged from 1.39 to
1.71 g cm-' with standard deviation and mean of 0.096 and 1.548 g cm-' ,
respectively.

At the depth 0.15 to 0.3 m, the bulk density varied from 1.38 to 1.822 g
cm" . with mean value of 1.623 . The standard deviation was 0.13 1 g cm" .

At the depth 0.3 to 0.6 m, the values of bulk density ranged from 1.362 to
1.714. The standard deviation was 0.097 g cm", with the mean of 1.60 g cm-' .

The bulk density of 0.6 to 0.9 m depth varied from 1.255 to 1.675 g ~ m - ~ .
The mean value of the bulk density was 1.554 g cm" with standard deviation of
2.33 g c111-'.

The mean value of each depth showed that the bulk density at the topsoil
(0.0 to 0.15 m ) was the lowest and the highest was at a depth of 0.3 to 0.6 m. The
lowest of the bulk density of the 0.0 to 0.1 5 m layer was attributed to the loose
topsoil.

Porosity (E)

In this Study, the total porosity was calculated from the soil bulk density

(Dh) and the soil particle density (D,) (assume to be 2.65 g cm") as shown in
Appendix A.

The results from the calculation of the total porosity for this study showed
that it ranged from 3 1.2 % to 52.6 % and the mean value was 40.2 % with standard
deviation of 4.27 %.

At the topsoil layer, the total porosity varied from 35.6 % to 47.5 % and the
mean value was 41.6 %, with standard deviation of 0.85 %. The total porosity at
the subsoil layer closed to the topsoil (0.15 to 0.30 m). ranged from 31.2 to 47.9 %.
And the standard deviation was 4.92 %. with the mean of 38.7 %.

At a depth of 0.3 to 0.6 m, the total porosity varied from 35.3 % to 48.6 %
with the mean value of 39.7 %. The standard deviation was 3.7 %. At the subsoil
layer (0.6 to 0.9 m), it ranged from 36.8 to 52.6 %, with the mean value of 41.32
%. The standard deviation was 3.93 %.

Sand (S)

The soil particle size greater than 50 pm was considered as sand (S)
particle as classified by USDA. The results of this study showed that the variation
of the percentage of the sand ranged from 29.9 to 73.57 %. The mean value and the
standard deviation were 58.28 and 8.2 %. respectively.

The percentage of the sand fractions of the topsoil layer varied from 29.9 to
73.6 % and the mean value was 60.35 %, with standard deviation of 12.01 %.

The percentage of sand of the subsoil layer (0.15 to 0.3 m) varied from
40.1 1 to 73.4 %. The mean value and standard deviation were 60.55 and 8.25 %,
respectively.
The soil of the layer at a depth of 0.30 to 0.6 m showed the percentage of
the sand varied from 39.75 to 63.33, with the mean value of 55.41 %. The standard
deviation was 5.44 %.

The variation of the percentage of the sand at a depth of 0.6 to 0.9 m


ranged from 49.4 to 66.22 %. The mean value and the standard deviation were
56.56 and 5.37 %. respectively.

The mean value of each soil depth indicated that the highest sand content
was the topsoil layer (0.0 to 0.1 5 m) and the lowest was the deep layer. The results

of the mean value showed that the percentage of the sand content almost decreased
with increasing soil depth.

Silt (Si)

According to the USDA soil classification, the soil particle between 2 to 50

p n is considered as silts. The results of this study showed that the silt content
ranged from 2.4 to 33.9 %. The mean value and the standard deviation were 12.1
and 5.28 %, respectively.

The variation of the percentage of the silt at a depth of 0.0 to 0.15 m ranged
from 6.2 to 33.98 %, with the mean value of 13.68 %. The standard deviation was
7.08 %.

The soil at the layer 0.15 to 0.3 m showed the percentage of the silt varied
from 2.7 to 23.98 %. The mean value and standard deviation were 11.3 1 and 1.05
%, respectively.

At a depth of 0.3 to 0.6 m, the percentage of the silt ranged from 8.5 to 21.9
%, and the mean value was 13.1 1 % , with standard deviation of 3.92 %. At the

deep layer (0.6 to 0.90 m), the percentage of the silt varied from 2.4 to 18.6 % and
the mean value was 10.9 %. The standard deviation was 4.1 1 %.

As indicated by the mean value of each depth it was found that, the soil of
the top layer (0.0 to 0.15 m) has the highest value of silt compared to the other

layers and the soil at a depth of 0.6 to 0.9 m has the least silt content. The
percentage of si It decreased with increasing soil depth.

Based on the USDA soil classification, the soil fraction less than 2 pm is
considered as clays. The results from the soil particle size analysis for this study
showed that the percentage of clay content varied from 18.0 to 44.0 %, and the
mean value was 28.9 %. The standard deviation was 5.33 %.

At the topsoil layer, the variation of the percentage of clay ranged from 18
to 37.74 %. The mean value and the standard deviation were 25.79 and 6.2 %,
respectively.

The percentage of the clay of 0.-15 to 0.30 m depth varied from


19.6 and 40.5 1 %. The mean value and the standard deviation were 26.82 and 4.48
%. respectively.

At the layer 0.3 to 0.6 m, the percentage of the clay varied from 26.35 to
38.23 %, with the mean value of 30.47 %. The standard deviation was 3.03 %. At
the deep layer (0.6 to 0.9 m), the percentage of clay content varied from 22.7 to 44
%. The mean value and the standard deviation were 32.36 and 4.74 %,

respectively.

As evidence from the mean value at each depth (25.79, 26.82, 30.47, and
32.36 % for 0-0.15, 0.15-0.3, 0.3-0.6, and 0.6-0.9 m depths, respectively) the
percentage of the clay content increased with increasing soil depth.

Soil Textural

The relative proportions of sand, silt, and clay were used to determine the
soil textural classes using soil textural triangle. The particle size distribution (PSD)
was estimated for each sample using the Pipette method. From the PSD results the
soil texture composition was estimated. Despite the vertical soil profiles are not
stratified, a texture analysis at different depths (0-1 5 cm, 15-30 cm, 30-45 cm, 4560 cm, and 60-90 cm) was performed. The soil profiles are characterized by a top
0.9 m consisted mainly of sandy clay loam with small areas of sandy clay and
sandy loam texture in accordance with the USDA soil texture classification.

Figure 4.1 shows the textural distribution of the full data set of the study
area while Figurer 4.3 shows the textural distribution of the soil in each depth. It is
clear from both figures that the texture of the samples in this study was found
mainly sandy clay loam and represents 81.5 % (106 samples) from the total
number of soil samples of 130 samples. The other 18.5 % (24 samples) are
represented by three classes of the textural triangle classification sandy clay, sandy
loam and clay loam (10.8 % (14 samples), 4.6 % (6 samples), and 3.1 % (4
samples) respectively). The results obtained in this study are in agreement with the
work of Wong (1 970) and Paramananthan (1 978).

70

-m0

60

40
30
20
TO
0
0

10

20

30

40

50

60

70

80

90

100

10

20

30

40

% sand

10

20

30

40

50

60

70

Depth 15-30 cm

Depth 0- 1 5 cm

50

% sand

60

70

80

90

100

% sand

Depth 30-60 cm

Depth 60-90 cm

Figure 4.3: Textural distributions for the data set at each depth.

80

90

100

Infiltration rate

Results of 10 infiltration rates studies using DRI were analyzed to establish


a relationship between infiltration rates and the elapsed time. The cumulative
infiltration was also recorded to determine the required amount of water used for
each experimental run. A power function was fitted to the data points. From the
graphs plotted, equations were found to be exponential, and I

=a

T~ and D = c Td

were derived, where 1 is infiltration rate (cmlh), T is elapsed time in (minutes), D


is the cumulative infiltration (cm) and a, b, c, and d are constants of the relations.
The constants a and c are the values when the X-axis is one and constants b and d
are the gradients of the log-log graph. Typical plots are shown in Figures 4.4 and
4.5

The results of curve fitting are shown in Table 4.4. The final infiltration
rates If,,, (cmlh) are found when terminal rates were achieved. This normally is at
100 to 120 minutes. The IfinaIrates from this work varied from 0.02 m day-' to 0.86
rn day-' with one exceptionally high value of 3.36 m day-'.For the infiltration rates
Noriah (1992) reported value of 1.44 m day-' for Serdang Series soils. Amin et al.
(2001) showed infiltration rates values for Serdang Series soils to range from 0.01
to 0.9 m day-', which compares well with the results obtained in this study.

Although it was not possible to accurately determined a steady state


infiltration rate the values which were determined ( albeit not very accurately)

80

100

120

140

160

180

200

Elased time, Minutes


cumullative rate

infiltration rate

-Power (cumullative rate ) -Power (infiltration rate)


Figure 4.4: lnfiltration rate and cumulative infiltration versus elapsed time for
point No. 7 Serdang Series soil

Elapsed Time, Minutes


cumulative rate
A infiltration rate
Power (infiltration rate)
Powes (cumulative rate)

Figure 4.5: A log-log of infiltration rate and cumulative infiltration versus elapsed
time for point no.7 Serdang Series soil

Table 4.3: Results of Infiltronicter measurement and prediction


Locati I - M.C.
I on Cmlhr %

1440

K.,

--

I>

crnlhr mlday

VOILIIII

cm

Table 4.4: Eauations of infiltration rates and cumulative infiltration

Location

D = yf

l=aTh
cni / hr

c nl

Ii

I = 38.72 T-0 295

1I 2

D = 0.05

I
I

= 57.36 T-Oo7

-0 7s 1

0.736 1
TI

does not correspond reasonably to reported values for sandy clay loam soils.
However, closer agreement was found when the infiltration rate was extrapolated
to T = 1440 minutes (ie. one day) as shown in Table 4.3. Multiplying those values
with 0.24 one obtains K, values in m day-'.

Figure 4.5 shows a log-log plot of instantaneous infiltration rates I and


cumulative infiltration D for location no. 7 where Ifinar
0.34. c= 0.9, and

1,440

= 0.90

2.4 cmlh, a= 10.62, b=

cmlh or 0.21 6 ml day. For this particular location, the

cumulative infiltration depth at 100 minutes was 6.52 cm.

Results of soil saturated hydraulic conductivity, K, Model Analysis and


Empirical Models evaluation

Saturated hydraulic conductivity (K,)

To test the validity of the methods outlined in the previous chapter 111, a 12 m
x 100 m field site was prepared for the field experimental work. In this study of

vadose zone saturated soil hydraulic conductivity, the focus was toward comparisons
of measurements in the field to those in the laborato~y.controlled condition, and those
estimated from other soil properties. Comparisons were canied out on Serdang Series
soils between K, determined by means of the GP, SCHP and DRI methods and those
calculated by pedotransfer functions.

The geometric mean Kls and K, values for the three methods regardless of the
soil depth are given in Table 3.5. Also include in Table 4.5 are the arithmetic mean.
maximum, minimum, standard deviation, and coefficients of variation (CV) for the Ks
as well as for the other soil properties.

The following sections well discuss the results of this study of the K, values as
a function of depth and the nlcthod ofdetennination.

Table 4.5: Descriptive statistics of parameters of Serdang Series soil measured at the experimental site across all depths.
Parameter

Units

No of
Mean value$
observations

SD#

Minimum
value

Maximum
Value

Coefficient Skewness1
of variation
%

K~GP
Ksscf-{P
&DRI

Clay ( < 2pm)


Silt (2-50 pm)
Sand (>50 pm)
OM
Db
E
MC

Cm s-'
Cm s-'
Cm s-'
Yo
'Yo

Yo
Yo
g cm-3
'Yo
'Yo

$ Geometric mean for KsGPKsSCHP


and KsDRl; arithmetic mean otherwise.
# Standard deviation of loglo -transformed values for KsGP KcSCllP
and KslJRI; standard deviation of untransformed values otherwise.
7 Skewness of log10-transformed values for KsGpKsSCI1pand KsDRI;skewness of untransformed values otherwise.
$ Kurtosis of loglo -transformed values for K s ~ p KsSCrTpand KsDRI; kurtosis of untransformed values otherwise.

Statistical Distribution of Ks parameter

Before parametric statistical sets can be performed, it is necessary to


determine the statistical distributions of the Ks data to decide whether arithmetic or
geometric means should be used in the comparison. The mean, standard
deviations, skewness, and kurtosis coefficients of the parameter K, and their
logarithmic transforms were evaluated.

The Skewness and Kurtosis coefficients were used to verify the statistical
distribution of the data. The Skewness coefficient close to zero for a normally
distributed variable. The kurtosis coefficient, a dimensionless measure of the
flatness of a distribution, is 3 for a normally distributed random variable. Hence,
the closer the calculated kurtosis to 3, the greater is the tendency toward normality.

The frequency distribution of Ks for the results obtained from the three
methods (GP, DRI and SCHP) were fitted to both normal and lognormal
distributions. Figures 4.6 and 4.7 show the corresponding frequency histograms
and fitted distribution functions.

It is evident from the figures and the values of Kurtosis (Table 4.4) that the

K, values were found to be logarithmically distributed. This distribution which is


common for this soil property has been reported by many researchers to describe
the variability of this parameter, e.g. Warrick and Nielsen (1980), Talsma and
Hallam ( 1 980), Reynolds and Erick (1 985), Lee at a]. ( 1 985), Wilson et a]. (1 989),

and Mohanty et al. (1994). Consequently, all the statistical tests of the K, results
were therefore carried out using the geometric mean of untransformed data
(arithmetic means of logtransformed data) rather than arithmetic mean (Bouwer
and Jackson, 1974; and Lee et al. 1985).

Figures 4.8 and 4.9 present the normal probability plots for log (K ,). These
values approximate a straight line and demonstrate lognormal distribution in both
cases.

Double Ring infiltrorneter (DRI)

As mentioned earlier in chapter IV, the double ring infiltrometer (DRI) was
only used to measure the Ks at the soil surface. The final infiltration rates Ifinal
values were obtained (see Table 4.3).

The geometric mean Ks values obtained by DRI method ranged from 2.78

x 1 o - ~and 1.297 x

cm s-', with the geometric mean value of 1.342 x lo4

cm s-' and the standard deviation of value of 0.7484 cm s-I. The coefficient of
variation (CV) value for the DRI method was 123.97 %.

The results also show that there was statistically significant difference at
0.05 level between Ks obtained from the Ifinalresults and the Kfsobtained by the GP
method but they were in agreement at 0.01 level. While the Ifinalvalues were found
to be similar to those obtained by the SCHP at the 0.05 and 0.01 levels.

0 002500

0.005000

0 007500

Lab- K s (crnk)

Figure 4.6: Frequency histogram and fitted distribution functions for laboratory
evaluated saturated hydraulic conductivity (K,) using constant head
(SCHP) method

0.000800

0.00 1200

0.00 1600

ks-Field (cm /s)

Figure 4.7: Frequency histogram and fitted distribution functions for field
evaluated saturated hydraulic conductivity (K,) using Guelph
permeameter (GP) method.

-7

-6

-5

-4

-3

-2

Log (Ks-Field ) cm 1s

Figure 4.8: Normal probability plot for field-evaluation logarithmic saturated


hydraulic conductivity (Krs)values for sandy clay loam soil

Log (ks- lab) (cmls)


Figure 4.9: Normal probability plot for lab-evaluation logarithmic saturated
hydraulic conductivity (K,) values for sandy clay loam soil

On other hand, the comparison results show that the

1,440

data were not statistically

different from the Kli values of the G P method while they were statistically
different from those found by the SCHP at both levels of significance (0.05 and
0.0 1 ).

Constant Head method (SCHP)

The undisturbed soil core samples collected at different depths were tested
for Ks using constant head method (SCHP) as shown in the previous chapter IV.
The K, values range from 4.4 x 1 0-6 to 1.315 x I 0-2cm

S"

with the geometric mean

value and standard deviation of 4.238 x 10" cm s-' and 1.993 cm s-'.

Table 4.5 also shows that the SCHP method produces the largest KsGm,

KsMaxand KsMlnregardless of soil depth . Measurements at the four depths, 0 to


0.15, 0.15 to 0.30, 0.30 to 60, and 0.6 to 0.90 m were performed during the
experimental work and values of K, equal to, respectively, 1.4312 x I o", 4.795 x
lo", 2.2 1 x 1 0" a and 3.34 x 1 o ' ~cm

S-'

were estimated.

The geometric mean value of Ks of the top layer, measured at depth of 00.15 m ranged from 1.93 x 10" to 1.214 x 1 0-2 cm S-I.The standard deviation of
the log transformed values was 3.1 1 cm s", with the geometric mean value of
1.431 x 10" cm s" .

The K, values measured at the depth of 0.15 to 0.30 m range from 1.43 x
1 0-5to 7.104 x 1 0-3cm s-'. The geometric mean value and standard deviation were

4.7 95 x l o 4 and 6.854 cm s-' . respectively.

The K, values at the depth of 0.30 to 0.60 varied from 7.333 x 1 0-6to 7.57 1

x lo4 cm s-' , with the standard deviation of 6.854 cm s-I. And the geometric
mean for this layer was 2.2 13 x I o4 cm S-I.

And the K, value of the deep layer, measured at 0.60 to 0.90 ranged from
I x 1 0-5to 1.3 I 5 x 1 0-2cm s". The geometric mean and the standard deviation were
3.335 x 1 o - ~and 8.72 cm s-'>respectively.

The geometric mean value of K, at topsoil was generally higher than at the
lower depths. This very likely due to the differences in PSD causing differences in
pore size distribution. There is more coarse sand in the top 30 cm of the profiles at
the site than at the lower depths.

Moreover, some of the cores tested in the laboratory had visible fine roots.
Therefore, the higher KsGm.KsMax,and KsMlnvalues obtained by soil core method
may reflect either natural (macropores, wormholes, root channels, and cracks) or
macropores created by insertion of the ring (although not observed during the
measurements) extended through the entire length of the core (Paige and Hillel,
1 993, Reynolds, 1993b; Bagarello and Provenzano, 1996).

Guelph permeameter (GP)

As mentioned in chapter I l l , the field saturated hydraulic conductivity, KI,,


values were estimated from the steady state infiltration measurement obtained by
the GP method. Table 5.4 shows that the geometric mean Kfs values varied from
7.333 x lo-' to 1.654 x 1 0-3 cm s- with the geometric mean value of 1.268 x1 o ' ~
cm s-' . The standard deviation was 2.075 cm s".

At the topsoil layer (0.0 to 0.15 m), the results of GP measurements


conducted on sandy clay loam showed that Kfs ranged from 1.1 x 1 o

- ~to

1.65 x

cm s-I. The geometric mean value and the standard deviation were 6.424 x
1o

- ~and 4.933 cm s-I, respectively.


At the depth of 0.1 5 to 0.30 m Kfs value varied from 5.87 x loe7to I .903 x

lo4 cm s-I, with geometric mean value of 5.53 x

cm s-I. The standard

deviation was 3.84 cm s-' .

The Kfs value measured at the depth of 0.30 to 0.60 m range from 7.333 x

1 0-8to 7.571 x lo4 cm s-'. The geometric mean and standard deviation values were
1.023 x

and 7.85 cm s-' . At the depth of 0.60 to 0.90 m, the Kfs value ranged

from 1.47 x 1 o ' ~ to 7.4353 x 10" cm

S"

with standard deviation of 9.56 cm s-' .

And the geometric mean value for this layer was 1.099 x

cm

S-I

The results show that the GP has the lowest geometric mean values of Ks.
Moreover, the lowest geometric mean value was found at 0.60 m depth measured
by GP method. The geometric mean value of the four layer depths also showed
that the K, value decreased with increasing soil depth because of the increase in the
clay contents with depth.

It appears from this study that the GP method seemingly performed better
at the shallower depth (0 to 30 m). This could be attributed to the higher sand
content (see Table 4.2), at which the measurement was taken. This was very likely
due to the differences in PSD causing differences in pore size distribution. There is
more coarse sand in the top 0.30 m of the profiles at the site than at the lower
depths.

During the field experimental work, the GP technique gave a number of


negative values. As can be seen in Table 4.6 approximately, 35.5 % of the results
obtained by GP method were negative when the "standard" method of calculation

in the user's manual (Soil Moisture Equipment Corporation, 1987) was employed.

Other researchers reported 26 % for Troup loamy sand soil, 55 % for


Norfolk sandy loam and 44 % for Lucedale loam soils (Slaverda and Dane, 1993).
Negative percentage values of 55 %, 39 %26 % and 34 % obtained from the
permeameter tests in grassland, woodland. and arable land respectively, were
reported by (Ragab and cooper, 1993). The percentage of negative value found in
this study, therefore, is within the range found by other workers.

In addition, the negative values obtained in this study were higher for

the lower depths (0.30 to 0.9 m) which may be attributed to high clay content
(Table 4.2). Thus these negative values may due to random heterogeneity and
smearing andlor compaction of the well walls, especially during wetter soil
conditions which cause changes in wall resistance to water flow.

As pointed out by Reynolds and Elrick, (1985 and 1987), and Lee et al.
( 1 985) the negative values may be attributed to the soil heterogeneity. Whereas the

use of the simultaneous-equations approach in heterogeneous soils can result in a


high percentage of negative Kfs and matrix flux potential values (Elrick, and
Reynolds; 1992).

It should be noted at the outset that non-negative determinations of


hydraulic conductivity were included in the analysis. Consequently, to avoid the
negative results in these cases, the "alternative" method of Elrick and Reynolds
(1 992b), the single head analysis (SH), was used to derive Kfs (see Appendix B).

Failure and Time requirement

The failure attempt was any attempt that did not produce a K, value. It was
calculated by dividing the number of failed attempts to obtain a Ks measurement
by the total number of attempts multiplying by 100. The summary statistics for the
failure ratio, negative values and time required to complete one set of measurement
for the three methods regardless of the depth of measurement, are given in
Table 4.6.

Table 4.6: The total number of observation, percentage of failure ratio (FR),
percentage of negative Kk values and time required (T) to complete
one set of measurement.

Methods

Failure Ratio
YO

No of negative
Kfs values %

T
Minutes

Guelph permeameter

93

14.7

35.5

25 - 95

Constant head permeameter

62

Double ring infiltrometer

10

1500 -1 650
9.1

In 16 out of 109 Kfs measurement was not possible because there was no Ks
values that could not be obtained due to no decline in the GP reservoir after the
starting the experiment. For these wells, the soil was compacted or may be its Kfs
was beyond the lower limit measurable by the GP apparatus (The GP marketed by
Soilmoisture Equipment Corp. can measure a Kfs range of about 1 0-2to I od).

In this study, the GP method has the highest FR while the DRI and SCHP
method were about the same (Table 4.6). Given that soil at lower depths, the clay
content is high, these failures with GP method were undoubtedly due to smearing
of the well surfaces by the auger particularly under wet soil conditions. The
smearing andlor compaction are usually issues when the flows are very slow
(Reynolds 1993a). This partial sealing of the borehole generally results in an
unrepresentatively low steady state flow (Qs) and corresponding Kfs (Reynolds and
Elrick, 1986) and smearing, remolding and siltation of infiltration surfaces can
strongly reduce Q, even to point Qs = 0 in extreme cases (Elrick et al., 1990).

While the SCHP method sometimes failed to produce K, measurements


mainly because of continuous natural macropores or macropores created by the
insertion of the ring vented at both top and bottom of the core. Additionally,
failures may be attributed to some natural physical property of the soil affecting K,
such as cracks, layering, slaking, swelling etc., which is not adequately accounted
for by the measurement technique (Reynolds and Elrick, 1985).

Also presented in Table 5.6 is the time (T) required to obtain a given K,
value by these methods. The average time (T) value was calculated only for the
successful K, measurements and includes the time required to prepare and install
the apparatus as well as time spent taking the reading. It is seen from Table 4.5 that
the time spent preparing, installing, and to complete one set of measurement with
the DRI ranged from 100-180 min, as compared to 25 to 90 minutes for the GP
method and 1500 to 1650 minutes for SCHP method to make one vertical
conductivity measurement.

It is noted that the GP method produced the lowest T values and the SCHP
method consistently produced the largest value (due to more than 90 % of the total
time spent saturating the cores in the laboratory). However as the GP method
required a larger number of tests, resulting in more total time. Therefore number of
measurements and time per run required to achieve an estimate of Ks with selected
level of confidence must be given careful consideration.

Comparison of The methods

The reliability of the Guelph Permeameter method (GP) is shown for


Serdang Series sandy clay loam soil by comparing estimation K, values with data
points obtained via the constant head permeameter (SCHP), and double ring
infiltrometer (DRI) methods. The DRI method is in situ (like the GP method)
while the SCHP method is laboratory determination on an "undisturbed", vertically
oriented soil core taken from the field.

As mention earlier, Kfs data were consequently log-transformed before


statistical comparison; geometric means were calculated rather than arithmetic
means; Kscmis used to represent the geometric means K, value, KsMax
to represent
the maximum Ks value, KsMln
to represent minimum Ks value, CV for coefficient
of variation .

Statistics were performed using Statgraphics software version 3. First,


average and standard deviation were calculated for untransformed Ks values. Then
geometric mean value was calculated for a lognormal distribution of Ks from the
average of transformed loglo (K,) values. The Geometric mean Ks values for each
method were compared on a pairwise basis within each depth to determine if there
were any difference according to the method and the depth of determination using
two tailed, multiple F-and t tests. The F- and t-tests were conducted at the 1 %
level and/or 5 % probability level. Multiple comparison was performed with
Tukey's honestly significant difference (THSD).

GP, DRI and SCHP Ks values were statistical compared within each depth,
on the basis of geometric mean (KsGM),maximum and minimum, coefficient of
CV, and SD of the K, values.

A summary of analytical results, including the number of observation (N),

geometric mean, maximum, minimum, standard deviation and coefficient of


variation for the Ks obtained using the three methods are given in Table 4.7. The

KsGm is used to represent the geometric mean Ks value, KsMaxand KsMlnto


represent the maximum and minimum K, value respectively, SD for standard
deviation and CV for coefficient of variation.

As evidence in the ranges of values obtained, each method of


experimentation yielded Ks estimates varying widely in magnitude; the order of
magnitude of range varied substantially from method to method. The GP method

~ 1.654 x 1o - cm
~
yielded the widest range of values, from 7.333 x I o - to

S"

while

~ 1.315 x lo-' cm S-I.


the SCHP yielded the narrowest range, from 4.4 x I o - to
The results show that there is a statistically significant difference between
the means of the top layers and the lower layers. The loglo Kfs values were found
to be significant at the depths. While only between 0.0 to 0.15 m and 0.30 to 0.60
m depths was significantly different for Ks measured by SCHP method.

Table 4.7: Comparison of K, measurements obtained using the double ring infiltrometer (DRI), Guelph permealneter GP), and intact
soil core (SCHP) methods. KSGmis geometric mean K, value: KsMI,,is tlie minimum K, value, KsMa, is the maximum K,
value; SD is the standard deviation; CV is the coefficient of variation, negative value , arid N is the number of
measurements.
No. of
K*SG~
Ks~in
~ M R X
sD$
CV %
negative Ks
Depth Methods N

cm

GP
SCHP
DRI

25
14
10

6.424 x 10-5a
1.4312 x
1.3426~10~~

15-30

GP
SCHP

25
12

6.1272 x 1 0 " ~
4.7953~10~~

30-60

GP
SCHP

21
24

1.0229 x 10-6a
2.2131 x 1 0 " ~

60-90

GP
SCHP

22
12

1.099 x
3.3351 x

a
a

*Values followed by the same letter are statistically different at the 0.01 probability level for that depth.
$Standard deviation of loglo-transformed values for Krs and K,.

The results of the three measurement techniques showed that there was a
considerable variation in values of Ks. The GP. DRI , and SCHP techniques
produced K, values. The coefficients of variation of K, for sandy clay loan1
obtained from GP were higher than those obtained from laboratory (206.5 to 285
96) with average 240.26 %. While the SCHP method CV values ranged from 123.1

to 186.04 with the average value of 157.5. Lee et al (1 985) reported coefficient of
variation, CV, of Kfs and Ks in the order of 65% for sand soil, 130% for loam soil
and up to 600% for clay, which compares well with the results obtained in this
work using the two methods.

An implication of the differences in CV values is that substantially more


test runs are required when using the GP and SCHP techniques to obtain a mean
value of K, in which one has the same level of confidence (Warrick and Nielsen,
1980). The GP method exhibits a CV that is I .7 times the CV value of the SCHP
method and 2.5 times the CV value of the DRI method. Gupta et a1 (1993) found
that experimental results for GP method were about twice as variable as the results
for the rainfall simulator and the ring infiltrometer.

In situ field measurement of Ks by GP method agreed favorably with those


determined from DRI method for 0 m to 0.15 m depth. Good agreement between
K, determined by GP and DRI was reported by Gupta et a1 (1 993).

At all depths, geometric mean Ks values measured on intact soil cores using
SCHP method differed significantly from those Kfs values obtained with GP
method (Table 4.7). The laboratory K, values were nearly 27 to 360 times greater

than geometric mean Kfs values determined by the GP method. These results are
within the range reported by other researchers. Kanwar et a1 (1989) found that the
mean Ks values of the laboratory method were about I0 times to 800 times higher
than the GP mean Ks values. Also K, values 10 to ,130 times greater for the
laboratory method than for the GP method were found by Mirjat and Kanwar
(1 994).

These results indicate that the relationship between geometric mean Kfsand
K, values is variable. The GP method often yields conductivity values lower than
those determined by intact soil cores and other methods (Reynolds and Elrick,
1985; Lee et a]. 1985; Talsma 1987; Kanwar et al. 1989; Paige and Hillel, 1993;
Gupta et al. 1993 and Bagarello and Provenzano, 1996).

Reasons for these differences amongst the geometric mean values are not
completely obvious. However, there could be several possible reasons for this
disagreement between the GP (in situ) method results and SCHP (laboratory)
method results. The primary reason is the entrapment of air in the soil can lead to
Kfs values that are less than the saturated values (Bouwer, 1966; Talsma and
Hallam 1980; Lee et a1 1985). Some difference could be due to soil anisotropy. As
the undisturbed core samples were taken vertically, the SCHP measured vertical
K,", whereas the GP method delineates the combination of vertical and horizontal
soil Kfs. Stephens and Neuman (1 982a) stated that anisotropy sometimes is leading
to conductivity results 56 % of the rate in an isotropic soil.

It should also recognised that smearing of the well surface during the well
preparation under relatively wet conditions can contribute to low Kfs results
especially in clay rich soils (Reynolds at al. 1985; Talsma 1987; Wilson et al.
1989; Salverda and Dane 1993). However, there was low clay content at the
topsoil layer (0 cm to 15 cm depth) in the tested soil but the clay content increased
at the lower depths (see Table 4.2). Therefore, at the lower depths smearing might
have taken place and results in the low Krs values.

In addition to the air entrapped and anisotropy, the macroporosity (soil


cracks, wormholes, roots, etc) observed in the soil structure could explain part of
the difference between field and laboratory Ks values. It can lead to high Ks
readings in soil cores due to the pipe flow where continuous macropores connect
the end of the core to other (Gumbs 1974). Also macroporosity can cause
anisotropic conditions in the soil and, therefore, affect the flow out of the well. A
possible consequence of discontinuous macropores in the field could be lower
conductivity values for the GP than for the core method (Paige and Hillel 1993).

The results of Ks values given in Table 5.6 reflect the contribution of


failure or success of each method. The validity of each method depends upon the
accuracy and precision of the data collected. On the other hand the failures may be
the result of human errors made in the use of each method and the contribution of
some of natural physical properties of the soil that affect K, values. As pointed out
by Kanwar et al. (1989) the spatial and temporal variability of some of the soil
physical properties, such as macropores, cracks, wormholes, slacking, swelling,

etc., which are not accounted for by any of these techniques could add to the
increase or decrease of K, values.
The results presented in Table 4.7 also show that the SD values at the lower
depths of 0.30 and 0.60 m for the laboratory method are greater than those derived
by the GP method. The high SD values indicate that some soil cores may have
more macropores than others (e.g. Mohanty et al. 1994). Moreover, some of the
cores tested in the laboratory had visible roots. Therefore, the higher KsGm,KsMax,
and KsMlnvalues obtained by soil core method may reflect rapid pipe flow through
wormholes, root channels, and cracks that extend through entire length of the
cores.

As pointed by Kanwar et al. (1989) there is also a possibility that the


vertical macropores may be functioning well under laboratory conditions because
most of the entrapped air is removed gradually by saturating the cores from the
bottom. On other hand, the SD values at the depth of 0.90 m for the SCHP method
is smaller than the SD value for the GP method. Fewer open-ended macropores
were encountered in the sample soil cores.

To further investigate the relationship between the field and laboratory Ks


values, the log Kk was plotted against the log Ks (Figure 4.1 0). It is evidence from
the scatter of the points and the correlation coefficient (r) of 0.1 71 (Table 4.8) that

Log (Ks-Field) (cmls)

Figure 4.1 0: Linear regression between field (GP) and laboratory (SCHP)
evaluated saturated hydraulic conductivity (K,) values for Serdang
series sandy clay loam soil.

the results of the GP method and constant head method showed hardly any
agreement. This lack of the correlation between the results of the two methods
appear to confirm the findings of others (e.g., Bouma, 1980, 1983; Watt et al.
1982; Salverda and Dan, 1993 and Reynolds et al., 2000), that K, is extremely
sensitive to even relatively small differences in sample size, flow geometry and
soil structure.

The results of this study show that Serdang Series generally exhibited
higher Ks at the topsoil than subsoil. The results also indicated high variation in K,
estimates both within and between soil depths.

Despite the GP method was easy to operate, yielded lower estimates of K,


than those from SCHP technique for the all depths however, the GP method was
the most useful field method for vadose zone soils. It yields a combination of
vertical and horizontal Krs, representing the actual situation in the field. The GP
method has an advantage over the two other methods for its ability to produce both
soil saturated conductivity the slope of the unsaturated hydraulic conductivity
function (alpha parameter) from a single determination. Additionally. the most
advantage of this technique is its repeatable and rapid Kfs calculation. However, it
is very sensitive to the soil heterogeneity within the measured depth due to the
presence of cracks, fissures, stones, roots, macropores, etc.

Generally. comparison between the methods is hampered by a number of


factors. First, it is difficult to discriminate between spatial variables of K, and
errors that related to the methods. Second, different sample volumes and sample

numbers were used. Comparisons made between different K, measurements in the


field are subject to natural soil variations that may be larger than differences
between methods.

Differences

in assumptions made in the theoretical

developments suggest that some differences in results should be expected


(Bouwer, 1966)

K, Model analysis

Data of fundamental soil properties of Serdang Series include fractions of'


sand (S) , silt (Si), and clay (C ) ( > 50 pm, 2 to 50 pm and < 2 pm), organic matter
(OM), bulk density (Db ), porosity (E) and moisture content (MC ) were tested for

their ability to explain variation in Ks or log Ks. The Ks or log (K,) was the
dependent parameter and percentage of S, Si, C, MC, OM, E and Db were the
independent parameters.

To facilitate the statistical analysis of the data, all of the parameters were
keyed in into MS Excel and then transferred to the Statgraphics (Version 3, STSC
Plus Ware, Rockville, MD) and SPSS (Version 1 1 .O) software.

Pearson

correlations were used to investigate the relationship between Ks and the other
seven variables mentioned above. The multiple stepwise regression was then used
to select the variables to be included in the model and their coefficients. The
Model Fit Technique was carried out. The model takes the form

where the a0 = the regression constant, giving the intercept of the regression line on
the y axis; a, = the linear regression coefficients, representing the slope of the
regression line: XI = independent variables; and

E =

a random deviation of the y

from the regression line. Greater detail regarding stepwise regression can be found
in Draper and Smith (1 981).

Before deriving the regression equations between the soil properties and
the K, parameter, correlation coefficient (r) were calculated for the soil parameters
incorporated in the model. The results of the first analysis are shown in Table 4.8.

The results of the correlation showed that the KsGPdata was not correlated
with S, Si, OM, Db, E, or MC parameters (Table 4.8). There was also no
correlation between KsSCHP
and S, C, OM, Db, E, or MC (Table 4.8). There were
significant ( P < 0.05) negative correlation, however, KsGPand C (r = -0.2 13), and
positive correlation between KsSCHP
and Si (r = 0.258).

Table 4.8 shows also that the KsGp and KsSCHPdata were found to be
positively correlated (r = 0.172), but not significant at either 0.01 or 0.05 level. The
results of correlation procedure also showed that there were significant strong
negative correlation between C and S (r = -0.768, P < 0.01) and between Si and S
(r = -0.775, P < 0.01). There were also significant positive correlation between C
and Si (r = 0.223 P < 0.05) and between C and MC (r = 0.224, P < 0.05). While the
OM parameter produced a significant (P < 0.01) negative correlation with the C
when r = -0.282.

A routine course for analysing of the data gathered here is to use some

form of multiple linear regression with the soil parameters as explanatory variables
and K, as the response. However, the regression equations sometimes fail to
account for complex interactions between variables. Complex interactions were
expected with the current data set. There are several options for developing
empirical models when the complex interaction is evident. One option to combat

Table 4. 8: Correlation matrix of parameter measured at studv area.


Clay

Silt

Sand

KsGP

K~SCHP
Clay
Silt
Sand
OM
Db

E
MC

0.085
0.103
0.224*
0.030
-0. 148
-0.20 1
-0.007
0.006
1
Correlation is significant at the 0.01 level (2-tailed) and * Correlation is significant at the 0.05 level (2-tailed). KsGp Guelph
permearneter Ks values and KsSC"p constant head Ks value. Db the bulk density, E porosity, OM organic matter content and MC
moisture content.

**

the problem of regression on several highly correlated, so-called multicollinear


predictor variables, is to develop regression models using the Model Fit
Technique. This procedure has taken into account the level of correlation the ease
of' determination and it produces the best single-variable model, the best twovariable model and the best three-variable model, etc. Prediction models for K,
were obtained by regression analysis with each predictor variable tested singly and
also in combination (using stepwise multiple regression). The results of the
multiple regression analysis indicated that the significant inter-correlations limited
the number of useful functional relationships that could be derived from the seven
variables (Textural fractions, Db, MC, E, and OM) in Table 4.9

When the multiple linear regression (BACKWARD variable selection


method) was applied to the complete data for all depths, only simple function
based on Si, S, and E gave a significant relationship with K, at 0.05 level.
However, for full data set of four depths only 10.5 % of variability in Kfs was
explained by soil Si content and OM (Eq. 3 and Table 4.9). In order to improve the
predictor power of the model, attempts to transform the data into several
mathematical

forms were conducted. The results, however, showed no

improvement in the value of RI.

As indicated by the value of the coefficient of determination ( R =


~ 0.105, P
< 0.05) regression presented graphically in Figure 4.1 1 gave:

K, (cm s-') = 1.612 x lo-'

- 1.337 x 1 o3 (silt %) - 9.8 1 x 1 O ~ ( O M


%)

Measured Ks (cmls)

Figure 4.11 : Measured K, vs. predicted for complete data set of the study area.

Shown in Figure 4.1 1 is a scatter diagram of the measured vs. estimated


values for K, for the full data set. It is apparent that a measured and estimated
value for the U s show considerable variation around the regression line, which
indicates that the K, parameter do not adequately describe this relationship. This
may be attributed to the clear trend of U sand other soil parameters, such as C and

S, with depth (Table 4.2). The means of both Ks and Kk and sand % decrease with
depth while the mean of clay % increases with depth. Therefore the data for all
depths probably cannot be regarded as stationary.

Thus the multiple linear regression analysis was performed again separately
to the exactly the same depths as where the measured Ks and K , values were
obtained. The results of the regression analysis indicted that there was a significant
relationship between K, and Kfs and the input variables at each depth. The several
final regression equations with significant level and their coefficient of
determination are presented in Table 4.9

As can be seen in Table 4.9 some improvements in R~ were found when the
regression analysis applied to each depth and when the data transformations were
used. For example, at the depth 90 cm the regression models depicting the
relationships between Ks and the Si, Db and E had relatively high and significant
coefficient of determination ( R =
~ 0.738, a< 0.01) and indicated that variation in Si.
Db. and E accounted for over 73.8 % of the variability in K,. From the results the
regression models with the highest R' were chosen as the best fit at each depth.

Table 4.9: Derived regression models (where K,, Kr, in cm s-' ) are measured sati~ratedhydraulic conductivity using constant head
permeameter (SCHP) and Guelph perrnea~neter(GP) methods respectively. Dbbulk density in g cm-'. E porosity %. MC
moisture content %)
Regression model

coefficient of determination (R'

All soil depths


+ 1.2613 x lo-5(silt %) + 8.687 x 10.' (Sand %) + 8.283 x 10-"(E%)
Kr, = -9.62 x
Log (K,) = - 12.22 + 0.94 log (Silt %) + 3.436 log (Sand %) + 0.67 1 log (OM %)
K, = 1.612 x 10" - 1.337 x 1 0 ' ~(Silt %) - ,931 x
(OM %)
Kry=
K, =
Log (K,) =
Log (KJ =
Log (K, ) =

Depth 0- I5 cm
5.072 x 10." 3.02 x 10" (clay %) + 6.01 1 x 10.' (MC %)
-2.44 x 10-'+5.775 x 10-'(Silt%)+ 3.169 x 10-~(Sand%)
- 61.75 + 0.563 (Clay %) + 0.632 (Silt %) + 0.592 (Sand %)
14.118 + 1.917 log (Silt %) +2.841 log (Sand %) + 3.36 log (MC %)
- 1 1.864 + 2.721 log (Silt %) + 4.103 log (Sand %) - 3.296 log ( E %) + 3.324 log (MC' %)

signilkant le\.el ( a )

[I]
121
(31

0.067
0.089
0.105

a =0.15
a = 0.05
a = 0.05

(41
151
[GI
171
[g]

0.26 1
0.320
0.484
0.52 1
0.568

a =0.100
a =0.150
a = 0.075
a = 0.053
a = 0.08 1

[ 91

[I 01
11 11
1121
[I31

0.131
0.307
0.39
0.479
0.503

a =0.215
a = 0.062
a =0.108
a = 0.055
a =0.116

1141
[I51
1161
1171
[ 181
[191

0.245
0.128
0.318
0.329
0.387
0.408

a =0.179
a = 0.291
a = 0.320
a = 0.073
a = 0.036
a = 0.064

[201
1211
1221
1231

0.356
0.534
0.608
0.738

Ileptli 15-30cm
~
%) - 2.30 x 10'" (Sand %)
Krr = 2.808 x 10.' - 4.58 x 1 0 .(Clay
K, = 1.352 x
- 2.9 x x10" (E %)
Log (K, ) = -16.930 + 0.241 (silt) YO + 0.1 79 (sand %)
Log (KJ = -40.367 + 5.208 log (silt %) + 17.886 log (sand %)
Log (KJ = -36.479 + 5.143 log (silt %) + 18.881 log (sand YO)- 3.458 log (E YO)
Depth 30-60 crn
Kfr = -9.79 x 10'' + 2.709 x 10.' (Silt %) + 1.621 x 10.' (Sand %) - 1.19 x
(OM %)
- 3.03 x (Silt %) - 1.85 x 10.' (Sand %)
K, = 1.523 x
Log (K,) = 6.913 - 0.229 (Silt %) - 0.138 (Sand %)
Log (K,) = -8.98 - 2.99 x 10 -2 log (Silt %) - 0.540 (OM %) + 0.246 (MC %)
Log (KrJ = -20.580 + 3.03 log (Clay % ) - 1.944 log (OM %) + 8,579 log (MC %)
Log (Kr,) = -34.7 10 +6.462 log (Clay % ) + 4.233 log (Sand) - 1.809 log (OM %) + 9.83 1 log (MC %)
Depth 60-90 cm
Krs = 3.82 x 10.' + 3.079 x i t 5 (Clay %) + 4.68 x 10-5 (Silt %) +3.756 x 10-5 (Sand %) + 1.655 x 10-5 (E %)
2.16 x 10-3 (MC %)
K, = - 7.28 x lo-*+ 7.546 x 10.' (Silt %) + 4.426 x 1 0 ' ~ (Sand %) 4-1.025x 10.' (E %)
K, = -0.175 + 1.559 x 10.* log (Silt %) + 6.396 x 1 0 .log
~ (Sand %) +1.211 x
(E %)
K, = 3.022 + 6.988 x 10.' log (Silt %) - 2.40 log (Db)- 1.589 (E %)

= 0.177
a = 0.092
a = 0.048
a = 0.010

Examination of Table 4.9 shows that except for the depth 0.60 m, relating
the K, values with clay, silt, sand, OM,, Dh , E and MC gave the highest R2 values
compared to Kc,. At the 0.60 m depth the log (Kcs) values with C, S, OM, and MC
content gave the highest R' values.

As indicated by the R2 values, the models [8], [12], [I 91 and [23] in Table
4.8 are the best models at the depths 0.15 m, 0.30 m, 0.60 m, and 0.90 m
respectively. The best model at a depth of 0.15 m has Si, S, E and MC content as
input variables and the value of R2 was 0.568. While the best model at a depth of
0.30 m used Si, and sand as inputs with an ultimate value of R2 = 0.479. At a depth
of 0.60 m relating Ks values with C, S, OM, and MC content gave the best model
when R2= 0.408. While relating K, with Si content, Dband E gave at a depth of 0.90

m gave the highest values for the coefficient of determination, R2= 0.738.

Although the regression analysis results showed that there were significant
relationships between the soil saturated hydraulic conductivity and the seven input
variables at all depths, these relationships however were different at each depth. The
best model for the prediction of Ks at 0.1 5 m is

Log (Ks ) = -1 1.864 + 2.72 1 log (Silt %) + 4.103 log (Sand %)


-3.296 log (E ) + 3.324 log (MC)
and (R2= 0.568; a = 0.08 1)

and at 0.30 m the best model is

= -40.367

Log (K,)
and (R'

= 0.479;

+ 5.208 log (silt %) + 17.886 log (sand %)

a = 0.055)

and at 0.6 m the best model is

Log (KfJ = -34.71 0 +6.462 log (Clay %) + 4.233 log (Sand %)


-

I .SO9 log (OM) + 9.83 1 log (MC)

and (R2 = 0.408; a = 0.064)

and at 0.9 m the best model is

~ (Silt %) - 2.40 log (Db) - 1.589 (E )


K, = 3.022 + 6.988 x 1 o - log
and (R'

= 0.738;

a = 0.01)

where
K, in cm s-' is saturated hydraulic conductivity obtained by soil core constant head
method; Kfs in cm s-' is the field saturated hydraulic conductivity obtained by
Guelph permeameter method the laboratory; OM is the organic matter %, Db is the
bulk density in g ~ m ' E~ is; the porosity %; MC is the moisture content %.

The corresponding graphs of the regression models mention above are


shown in Figures 4.12 a to d. It is apparent that there are improvements in the
estimation of Ks.

Although the soil textural classes (S, Si and C), DB, OM, and E have
widely been used to predict K,, the use of these parameters as input variables in
estimating the K, in this study seems weak (Table 4.9). This is also evident from the
goodness of fit of lower information level models which does not appear
satisfactory.

Empirical Models Evaluation

In this study, thc data set obtained li-om particle size distribution. bulk dcnsity.

Dh, organic matter, OM, and porosity E for Serdang Series soil were employed to
evaluate the applicability and validity of the six selected pedotransfer f'unctions,
PTFs.. Sand, silt, and clay, percentage,

Dh,

OM. and E were used as input for the

PTFs. which subsequently provided Ks as output.

In the evaluation, the statistical analysis approach that was used by Tietje and
Hennings ( 1 996) and Wagner et al, (2001) for evaluation of PTFs and K, was used in
this study. The geometric mean error ratio (GMER) and geometric standard deviation
of the error ratio (GSDER) were calculated from the error ratio

of measured (K,,,)

vs. predicted (KSp) values by the following equations as given by Wanger et al, (2001):

GMER = exp

i l l1,
-

ln( zi)

According to Wanger et al., ( L U U I ) the GMER equal to 1 corresponds to


an exact matching between measured and predicted values; GMER < 1 or > 1
indicates that predicted values are generally underestimated or over-prediction,
respectively. The GSDER equal to 1 corresponds to a perfect matching and it
grows with deviation from measured data. Then the best model will, therefore,
give a close to 1 and a small GSDER.

To evaluate the model performance over the whole data set. the saturated
hydraulic conductivities of the Serdang Series soils in investigation were estimated
according to the afore mentioned PTFs (Eqs. 3.1 1 to 3.16) in chapter 111. Then the

-geometric mean error ratio (GMER) and geometric standard deviation of the error
ratio (GSDER) values were calculated using Eq 4.8 and 4.9 respectively. The
Geometric means of the Ks as obtained using the two methods and the six models
are shown in Table 4.10

Table 4. 1 shows the quantitative error criteria estimated using the data set
to evaluate the six investigated PTFs in general. The analysis of the results
obtained shows hat the PTFs of Vereecken et al. (1990), Saxton et al. (1986).
Brakensiek et al. (1984). Jabro (1992) and Amin et al. (1997) tend to
underestimate the values of K, , whereas the Cosby et al. (1 984) model was found
to more likely result in higher values than the obtained by SCHP method
measurement. This is assumed to be the result of the neglect of the soil structural
regime in the formulation of the Model. Soil structures can produce many coarse
pores and, as result, are prime factors influencing Ks (Landon 1991 and Campbell
et al. 1994).

Table 4.10: Sutnmary of statistics of saturated hydraulic conductivity for Serdarig Series. sandy claj, loam soil. obtained
by three neth hods and predicted by the six selected models.
Measured Ks cm s-'
Predicted K, crn s"
GP
SCHP
MI
M2
M3
M4
M5
M6

No. of observation

93

62

90

90

89

89

89

88

Geometric mean* x lo4


Variance x 1 0-8
Standard deviation x 1 0-4

Kurtosis

\coefficient of variation
285.85
161.58 29.25
147.28
42.52
102.76
123.81
34.29
*Values followed by the same letter are not statistically different at the 0.01 level according to the Tukey Honestly
Difference test. Where GP is the Guelph Permeameter, DRI is the double ring infiltrometer method, SCHP is the soil
core constant head permeameter, MI is Cosby at al. (1 984) model; M2 is Brakensiek et al. (1 984) model; Mj is Saxton
et a1 (1 986) model; Mqis Vereecken et al. (1 990) model; M5is Jabro ( 1 992) model and M6 is Amin et al. (1997) model.

Table 4.1 1 : Geometric mean error (GMER) and geometric standard deviation of error ratio (GSDER) calculated
with six selected models using the data set of the Serdang Series soil of study area compared to
measurements values.

K, model

Cosby
et a1 (1984)
(MI)

Brakensiek
el al. (1984)
(M2)

Saxton
el al. (1986)
(M3)

Vereecken
et al. (1990)
(M4)

(' 1 990)

(Mj)

Amin
et al. (1997)
(M6)

7.85

15.6

7.7 1

8.92

10.22

7.64

Jabro

GMER

GSDER

it is noted in Table 4.1 1 , that the GMER values which were calculated with
the Cosby et a1 (1984). Brakensiek et a1.(1984) and Saxton et a1.(1986) models
correlated well with the finding of' Wanger et al. (2001) and Tiet.je and Hennings
(1 996).

Table 4.1 1 shows that the geometric standard deviation error ratio GSDER,
an indicator of data scatter, differs only slightly between the most of the selected
models. Only the model of Brakensiek et al. (1984) stands out because it yields the
largest deviation (1 5.6) between predicted and measured values. The models of
Amin, Saxton, and Cosby exhibit the lowest GSDER (7.64, 7.71 and 7.85
respectively).

For the data set under consideration, it is clear from the Table 4.10 that
results generated by Amin et a1 (1 997), Jabro (1992) and Brakensiek et a1 (1 984)
models provide the closest estimated Ks values to the measured values. It was also
observed from the comparison results (Table 4.10) of the tested models in this study
that the GMER value was improved when more input data such as bulk density.
porosity, and organic matter were used. This holds true for the K, models by Amin.
Jabro, and Brakensiek. However. there was no reducing in the value of the GSDER.
These results support the statement made by Tietje and Hennings (1996) that
integrating additional independent variables such as bulk density and organic matter
content may improve the GMER, but the GSDER is not reduced essentially.

The model of Amin give the best fit to the data set with GMER closest to 1
(0.54) and lowest GSDER (7.64) of the models tested here. Followed by the Jabro

with GMER (0.43) and GSDER (10.22) then Brakensiek with GMER (0.43) and
GSDER (1 5.6). It consequently appears, at least for this soil (Serdang Series), that
of the six models compared in this study, the Amin et al model seems to be the

model of choice for the prediction of Ks. The second best model is Jabro model
whereas the model of Brakensiek et al. rank third.

Shown in Figure 4.13 are scatter diagrams of the measured versus the
predicted values for soil Ks. The solid line in each figure is a regression line for the
best fit. It is apparent that measured vs estimated values for Ks are, in general,
scattered around the line, indicating that the relatively low agreement between
predicted and measured values of Ks. This is in agreement with the results
demonstrated by comparing the GMER and GSDER values for the predictive
models with the measured values (Table 4.1 1).

To check further the validity of the six models in investigation the means Ks
values were compared on a pairwise basis using two tailed multiple F- and t-tests
(using Statgraphics plus version 3 software package) The results of the test at 95 %
significant level is given in Table 4.10.

Coshy (MI ) modcl

I h k c n s l c k ( M 2 ) modcl

-2 9 ,

Ih,plU(c~w\]j
Mcasurcd liydraul~cconduc~~v~ty

Vercechen (M4) model

Measured hydraulic conductivity [loglO(cnvs)J

-55

-50

-45

Jabro

(M5)model

-10

-35

-30

Measured hydraulic c o ~ ~ d u c t ~ v[logl0(c1~1/s)]


~ty

A n i ~ n( M h ) modcl

-25

-20

-15

Measurcd Itydraulic co~tductivity[logl O(cmk)]

Figure 4.13: Measured versus estimated Ks for the six models tested for a sandy clay
loam soil.

The results o f these statistics demonstrate that there were significant


differences between the model prediction K, values and the measured data. Then
multiple range test was conducted to identify which of the models are different.
The results of the Tukcy Honestly Sign~ficantDifference (Tukey HSD) test show
that the geometric mean K, values predicted by all models were significantly
values. While the results of
d~fferentat the 0.05 lcvel from the geometric mean K,,
the cornpanson of geometric means for laboratory

(SCHP) method and the

empirical models indicate that only the models of Cosby et a1 (1 984) and Amin et

a1 (1997) are not significantly different at the 0.01 level (Table 4.9).

When the Tukey HSD test was conducted on untransformed data of the K,
for the measurement methods and the empirical models there was significant
difference at the 0.05 level only between the result of the model of Cosby at al.
( 1 984) and the results the GP method. The results of the comparison also show that

the predicted mean K, values by all the models were statistically significantly
different at the 5 % probability level from the mean K, values obtained by the
laboratory method (see Appendix C)

In conclusion, the results reveal that since all the variables incorporated in
the models (Sand, silt, clay, Db, E, OM, and MC) contained in soil dataset, the
range of applicability is about the same for all the models compared in this study.
It seems that the applicability of different empirical models depend on the set of
soils chosen, which mean, that they may change somewhat for other data set. In
addition, the results show that for the data set of the soil under consideration here
the K, \:alucs obtained using the empirical models of Brakensiek, Sabro and Amin

et al model was in better agreement with the measured

Ks

values compared with the

other models. This suggests that the best performance of Amin model can be
attributed to the fact that, some models perform better in the area where climate and
geology are more similar to the soils used for developing of the respective models
(Wanger el a1200 1 ).

Generally, the prediction of Ks of soils by PTFs using texture soil properties


only as input data are expected to be poorer, when the soil is structured (Tietje and
Hennings, 1996). The poor prediction of Cosby et al (1 984) and Saxton et al (1 986)
models may be attributed to the fact that these models do not account for Db, E or
OM whereas soil structure parameter. In this respect, the models of Amin et al.,
Jabro, Brakensiek et a1 models and Vereecken et a1 seem to be less error prone as
the structure of the soils was accommodated by

Db,

or E or OM.

However, despite the soil structure in Vereecken et a1 (1990) model was


represented by

Db

and OM, the model led to estimates far from the K, values

obtained by SCHP method. The possible reason for the poor performance may be
due to estimation method used within investigation of Vereecken et a1 (1990). As
pointed by Wagner et al (2001) the methods (crust technique combined with the air
hot method) used by Vereecken et al. (1990) tend to be inaccurate and therefore,
experimental data used for the determination of PTFs may have been incorrect.
Contrary to Vereecken et a1 (1 990) the GP method was used by Amin et a1 (1997)
and soil intact core method (laboratory) were applied by Jabro (1992) and
Brakensiek et al (1 984).

Generally speaking, although accuracy of prediction generally increased if


more input data are used, however, there was always a considerable difference
between prediction and measurements.

CHAPTER V
SUMMARY AND CONCLUSION

For the study site chosen, (Serdang Series soils with sandy clay loam soil)
the comparison study of three measuring techniques; GP, DRI and SCHP methods
revealed that no in situ method (GP) and laboratory method (SCHP) are going to
give about the same results. Nevertheless, both methods were able to determine that
the Serdang Series sandy clay loam soil at the topsoil was the most permeable (at
about 5.4 x

cm s-I); that the K, has, in general, a tendency to decrease with

depth; and that the lower depth (60 and 90 cm) was the least permeable (about 1.69

x I o4 cm s-I).

In general, it is concluded on the basis of the results of the comparison in


this study that the laboratory method evaluated geometric mean of K, values (4.238

x lo4 cm s-I) was 33.4 times higher than GP method evaluated Kfs (1.268 x10-~cm
s-I). This may reflect the role of the macropores and the size of core upon the K,
values. However, K, values obtained by GP method was about twofold more
scattered than those obtained by SCHP method, as represented by the CV values.

In addition. this study showed that about 35.5 % of the Kfs values obtained
with the GP method. subjected to multiple head approach, are negative. This is
illustrating the heterogeneity of the soil under consideration. However, to avoid these
negative values, the study recommended using single head method (SH) which was
suggested by Reynolds et al ( 1 992).

Despite the G P method giving a lower KI, than the SCHP method, the GP
method was found to be portable, easily operated, requires only small volume of
water / measurements. repeatable and provides in situ rapid Kl, calculation. In
addition, it yields a combination of vertical and horizontal K, representing the actual
situation in the field, it also provides an estimate of unsaturated hydraulic
conductivity. Therefore, the GP method can be used successfully as a routine
method for determining of Ks in studies of various soil and water management
projects.

The results of the comparison also showed that each method has its own
strength and limitations. The selection of method, therefore, should be based on the
user's requirement, size of the budget, manpower availability, simplicity and ease
of operation, and time constraints.

This study showed that the interactions between the soil properties and the
measurement method are rather complex. Therefore, an accurate comparison
should take into account the factors that affect the Ks values. Leaving the influence
of these factors out of consideration implies that any conclusion on the results of
the comparison is only approximate. Therefore this study suggested that more
consideration be given to the influence of the soil properties on the Ks values when
assigning a comparison study between methods.

This study has outlined and evaluated six models for estimating

K,of sandy

clay loam soil. From the results of the PTFs evaluation study it can be concluded
that the PTFs performed reasonably well if more predictors were used (Texture.
bulk density, OM, etc.). The estimations were less accurate when fewer predictors

were used, however, such prediction by these models may still be useful when no
data are available. The dataset used for this study however, is probably too small to
conclude if the ranking of the different models is of validity. Some models may
perform better in other areas, where climate and geology are similar to the soils
used for determining of the respective PTFs.

The results of this study showed that the attempt to relate the soil K,
parameter to soil mass fractions of clay, silt, and sand separates (PSD), bulk density,
porosity, and organic matter produced very week relationship only about 10 %
percent of variation could be explained by the aforementioned soil properties.
However some improvements in R' were found when the regression analysis
applied cross the individual depths and the best model of this study showed R~ of
0.73 at 0.6-0.9 m depth and was significant at the level 0.01.

Further improvement of the model should be possible by employing a larger


set of data. Validation tests are needed for the model using various ranges of soils and
different K, measuring techniques. Discussion and conclusions also show the trends
and needs of further comprehensive work to develop PTF for K, of Malaysian soils
that can be used in many hydrologic applications in tropical regions and for further
use in modern hydrologic models.

REFERENCES

Abd. Manan, M. M. 1982. Aggregation and stability of Ultisols and Oxisols in


peninsular Malaysia. /'h.D. ~hesis,State University of Ghent.
Ahu-ja, L. R., 0. Wendroth, and D. R. Nielsen. 1993. Relationship between initial
drainage of surface soil and average profile saturated conductivity. Soil
Sci. Soc. Am. .I. 57: 19-25.
Ahu-ja, L.R., D.R. Cassel, R.R. Bruce, and B.B. Barnes. 1989. Evaluation of spatial
distribution of hydraulic conductivity using effective porosity data. Soil
Sci. 148 (2): 404-411
Ahuja, L.R., J.W. Naney, P.E. Green, and D.R. Nielsen. 1984. Macroporosity to
characterise spatial variability of hydraulic conductivity and effect of
land management. Soil Sci. Soc. An7. J. 48: 699-702
Amin, M. S. M., Z. M. Amin, and A. A. Zakaria. 2001. Estimation of saturated
hydraulic conductivity by field methods and by correlation. Pertanika J.
Sci.& Techno].9(1): 1-12
Amoozegra. A. 1993. Comments on Methods for analyzing constant-head well
permeameter data. Soil Sci. Soc. Am. J. 57: 559-560.
Amoozegar, A., and A. W. Warrick. 1986. Hydraulic conductivity of saturated
soils: Field Methods. In Methods of Soil Analysis, part I, Physical and
Mineralogical Methods, (2nd ed.) ed. A. Klute,pp735-768. Madison,
WI: Agronomy No.9. American Society of Agronomy.
Asare, S. N., R. P. Pudra, W. T. Dickinson, and G. J . Wall. 1993. Seasonal
variability of hydraulic conductivity. Trans. ASAE 36: 452-457.
Bagarello, V. 1997. lnfluence of well preparation on field-saturated hydraulic
conductivity measured with Guelph Permeameter. Geoderma 80: 169180.
Bagarello, V., and G. Giordano. 1999. Comparison of procedures to estimate
steady flow rate in field measurement of saturated hydraulic
conductivity by Guelph Permeameter method. J Agric. Engng. Res. 74:
63-7 1
Bagarello, V., and G. Provenzano. 1996. Factors affecting field and laboratory
measurement of saturated hydraulic conductivity. Trans. ASAE 39 (1):
153-159.
Bagarello, V., M. lovino, and W.D. Reynolds. 1999. Measuring hydraulic
conductivity in a cracking clay soil using the Geulph Permeameter.
Transaction qf the ASAE 42 (4): 9.57-964

Black, G. R. 1965. Bulk density, In Methodr of Soil Analysis, partl, Physical and
Mineralogical Methods, C.A. Black, pp374-390 Madison, Wl:
Agronomy No. 9. American Society of Agronomy.
Blake, G. R., and K. H. Hartge. 1986. Bulk density. In Methods of Soil Analy.sii.s,
par//, Physical and Mineralogical Methods, (2nd ed.) ed. A. Klute,
pp363-375. Madison, W1: Agronomy No.9. American Society of
Agronomy.
Bonsu, M. 1992. A study of a texture-based equation for estimating the saturated
hydraulic conductivity of an alfisol in the Sudan Savannah ecological
zone, Ghana. Hydrological Sci. 37: 599-606.
Bouma . J. 1982. Measuring the hydraulic conductivity of soil horizons with
continuous macropores. Soil Sci. Soc. Am J. 46:438-4441
Bouma, J., 1983. Use of soil survey data to select measurement techniques for
hydraulic conductivity. Agric. Water Management 6: 177-190
Bouma, J. 1991. Influence of soil macroporosity on environmental quality. Adv.
Agron. 46:1 -37
Bouma. J., C. Belmans, L.W. Dekker, and W.J.M. Jeurissen. 1982. Assessing the
suitability of soils with macropores for subsurface liquid disposal. J.
Environ. Qual. l2(3): 305-3 1 1
Bouwer, H. 1962. Field determination of hydraulic conductivity above the water
table with the double tube method. Soil Sci. Soc. Am. Proc. 26: 330335.
Bouwer. H 1966. Rapid field measurement of air entry value and hydraulic
conductivity of soil as significant parameters in flow system analysis.
Water Resour. Res. 2: 729-738.
Bouwer, H 1969. Planning and interpreting soil permeability measurements.
Journal ofthe Irrigation and Drainage Division , Proc.Amer.Soc. Civil
Engrs. 95 (IR3): 39 1-402.
Bouwer. H. 1 970. Groundwater Hydrology. Toronto: McGraw-Hill Press.
Bouwer, H. 1986. Intake rate: cylinder infiltrometer. In Methods of Soil Analysis,
part I, Physical and Mineralogical Methods, (2nd ed.) ed. A.
Klute,pp825-843. Madison, W1: Agronomy No.9. American Society of
Agronomy.
Bouwer, H. and R . D. Jackson. 1974. Determining soil properties. In Drainage for
Agriculture, ed. J. van Schilfgaarde,pp611-672. Madison, WI:
Agronomy No. 17. American Society of Agronomy.
Brakensiek, D.L.. Rawls, W.J., and Stephenson, G.R. 1984. Modifying SCS
Hydrologic Soil Groups and Curve Numbers for Rangeland Soils. ASAE
paper no. PNR-84-203, St. Joseph, Mich.

Buckland, G. D., D. B. Harker. and T. G. Sommerfeldt. 1986. Comparison of


methods for determining saturated hydraulic conductivity and drainage
porosity of two southern Alberta soils. Can. .I. Soil Sci. 66: 249-259.
Camp, C. R. 1977. Determination of hydraulic conductivity for a Louisiana alluvial
soi I. In American Society Agriculture Engineering Third Nutionul
Drainage S y q m ~ i u mproceedings, pp I 04- 1 08. Chicago, Illinois.
Campbell, G. S., 1985. Soil physics with basic development in soil science.
Elsevier, Amsterdam Netherlands.
Campbell, G. S., 1974. A simple method for determining unsaturated conductivity
from moisture retention data. Soil Sci. 1 17: 3 1 1-3 14
Campbell. C.M. and D.D. Fritton. 1994. Factors affecting field saturated hydraulic
conductivity measured by borehole permeameter technique. Soil Sci.
Soc. Am. J 5 8 : 4354-1357
Childs, E. C., 1969. An Introduction to the Physical Basis of Soil Water
Phenemona. Wiley, London ,
Cosby, B.J., G.M. Hornberger, R.B. Clapp, and T.R. Ginn. 1984. A statistical
exploration of the relationships of soil moisture characteristics to the
physical properties of soils. Water Resources Research 20(6):682-690.
Day, P.R. 1965. Particle Fractionation and Particle-size Analysis. In . In Methods of
Soil Analysis, part I , Physical and Mineralogical Methods, ed. C. A.
Black, pp234-252. Madison, W1: Agronomy No. 9. American Society
of Agronomy.
DeBoer, D. W. 1979 Comparison of three field methods for determining saturated
hydraulic conductivity. Trans. ASAE. 22: 569-572.
Dorsey, A. D., A.D. Ward, N.R. Fausey and E.S. Bair. 1990. A comparison of four
field methods for measuring saturated hydraulic conductivity.
Transactions ofASAE. 33 (6): 1925-1931
Draper, N., and H. Smith. 1981. Applied regression analysis. John Wiley and Sons,
Inc., New York, N.Y.
Elrick, D. E., and W. D. Reynolds. 1986. An analysis of percolation test based on
three-dimensional saturated-unsaturated flow from a cylindrical test
hole. Soil Sci. 142: 308-32 1.
Elrick, D. E., and W. D. Reynolds. 1992. Methods for analyzing constant-head well
permeameter data. Soil Sci. Soc. Am. J. 56: 320-323
Elrick, D. E., and W. D. Reynolds. 1993. Reply to "comments on 'Methods for
analyzing constant-head well permeameter data'". Soil Sci. Soc. Am. J.
57: 559-560.

Elrick, D. E., and W. D. Reynolds, N . Baumgartner, K. A. Tan, and K. 1.,


Bradshaw. 1987. In-situ measurements of hydraulic properties of soils
using the Guelph permeameter and Guelph infiltrometer. In
Proceedings third international workshop on land drainage, pp G 13G23. Ohio State University, Columbus
Elrick, D. E., W. D. Reynolds, H.R. Geering and K.A.Tan. 1990. Estimating steady
infiltration rate times for infiltrometers and permeameters. Water
Resour. Res. 26 (4): 759-769
Elrick, D. E., W. D. Reynolds, R. G. Kachanoski, and G. W. Parkin. 1993. Field
measurement of water and solute transport parameters in soils. In Water
Flow and Solute Transport in Soils, ed. D. Russo and G. J.
Dagan,ppI 99-21 0. Advanced series in Agricultural Sciences, Vol. 20.
Espeby, B. 1990. An analysis of saturated hydraulic conductivity in a forested
glacial till slope. Soil Sci. 150 (2): 485-493
Haverkamp, R., and J.Y. Parlange. 1986. Predicting the retention curve from
Particle-size distribution: sandy without organic matter. Soil Sci. 142:
325-339
Heinen , M. and P. A. C. Raats. 1990. Evaluation of two models describing the
steady discharge from a constant head well permeameter into
unsaturated soil. Soil Sci. 150 (1): 401-412
Hillel, D. 1980. Fundamentals of Soil Physics. New York: Academic Press,
Hoffman. G. J., and G. 0. Schwab. 1964. Tile spacing prediction based on drain
outflow. Trans. ASAE. 7: 444 - 447.
Horton, R.E. 1940. An approach towards a physical interpretation of infiltration
capacity: Soil Sci. Soc. ofAm. Proce. Vol. 5: 399-4 17.
Gallichand, J., C. A. Madramootoo, P. Enright, and S. F. Barrington. 1990. An
evaluation of the Guelph permeameter for measuring saturated
hydraulic conductivity. Trans. ASAE. 33: 1 179-1 184.
Gergson, K., D.J. Hector and M. McGrowan. 1987. A one -parameter model for the
soil water characteristic. J. Soil Sci. 38: 483-486
Grander, W.R. 1958. Some steady state solutions to the unsaturated flow equation
with application to evaporation from a water table. Soil Sci. 85:228-232
Gumbs, F. A. 1974. Comparison of laboratory and field determined saturated
hydraulic conductivity and prediction from soil particle size. Trop.
Agric. (Trinidad) Vo1.(5 1 )3: 375-382
Gupta, R. K., R. P. Rudra, W. T. Dickinson, N. K. Patni, and G. J. Wall. 1993.
Comparison of saturated hydraulic conductivity measured by various
field methods. Trans. ASAE 36: 51 -55.

Jabro, J. D. 1992. Estimation of saturated hydraulic conductivity of soils from


particle size distribution and bulk density data. Trans. ASAE 35: 557560.
Jenssen, P. D. 1990. Methods for measuring the saturated hydraulic conductivity of
tills. Nordic Hydrology 21, 95-1 06.
Kanwar, R.S., H. A. Rizvi, M. Ahmed, R. Horton, and S. J. Marley. 1989.
Measurement of field-saturated hydraulic conductivity by using
Velocity Permeameters. Trans. ASAE 32 (6): 1 885- 1 890.
Kessler, J. and R. J. Ooserbaan. 1974. Determining hydraulic conductivity of soils.
in Drainage Principles and Applications. ILRI pub]., 16 (3):255-295,
Wagening, Netherlands.
Klute. A. 1986 Methods of soil analysis, part I , Physical and mineralogical
methods, Second Edition. American Society of AgronomyJnc. and Soil
science Society of America, Inc. Publisher. Madison, Wisconsin, USA.
Klute, A., and C. Dirksen. 1986. Hydraulic conductivity and diffusifity: Laboratory
Methods. In Methods of soil analysis, part 1, Physical and
mineralogical methods, (2nd ed.) ed. A. Klute, pp687-732. Madison,
WI: Agronomy No. 9. American Society of Agronomy.
Landon, J . R. 1991. Booker Tropical soil Manual " An handbook for soil survey
and agricultural land evaluation in tropics and subtropics" second
edition. Longman, Scientific and Technical.
Lee, D. M., W. D. Reynolds, D. E. Elrick, and B. E. Clothier. 1985. A comparison
of three field methods for measuring saturated hydraulic conductivity.
Can. J. Soil Sci. 65: 563-573.
Lembke, W. D. 1967. Observed and predicted tile outflow on a lake plain soil.
Transaction ASAE 10: 142- 144
Lin. H.S., K.J. Mclnnes, L.P. Wilding and C.T. Hallmark. 1999. Effect of soil
morphology on hydraulic properties Soil Sic. Soc. Am J. 63: 955-961
Mbagwu, J.S.C. 1995. Saturated hydraulic conductivity in relation to physical
properties of soils in the Nsukka Plains, southeastern nigeria. Geoderma
68: 5 1-66
McKezie, N., and D. Jacquier. 1997 Improving the field estimation of saturated
hydraulic conductivity in soil survey. Aust. J. Soil Res. 35: 803-825.
Mirjat, M.S., and R.S. Kanwar. 1994. A comparison of two saturated hydraulic
conductivity measuring techniques in relation to drain installation
methods. Applied Engineering in Agriculture 10 (1): 65-68
Miyazaki T., S. Hasegawa and T.Kasubuchi. 1993. Water flow in soils. pp15-63,
Marcel Dekker, INC. New York, USA.

Mohanty, B. P., R. S. Kanwar, and C. .I. Evert. 1994. Comparison of saturated


hydraulic conductivity measurement methods for a Glacial-till soil. Soil
Sci. Soc. Am. .I. 58: 672-677.
Noraiah, Abdullah. 1992. Physical properties, infiltration rate and available
moisture of several irrigated soil in Malaysia. Bachelor Project
Report, Univeristi Putra Malaysia
Oosterbaan. R. J.. and H. .I. Nijland. 1994. Determining the saturated hydraulic
conductivity. In Drainage principles and applications. Volume 3, (2nd
ed.) ed. H. P. Ritzema,pp435-475. Wageningen, The Netherlands: Int.
Institute for Land Reclamation and Improvement (ILRI) Publ.16.
Paramananthan, S. 1978. Register of soils peninsular Malaysia. Soils and analytical
services bulletin No. 7, Kuala Lumpur, Ministry of Agriculture.
Paige, G. B. and D. Hillel. 1993. Comparison of three methods for assessing soil
hydraulic properties. Soil Sci. 1 55(3): 1 75- 1 89.
Philip, J. R. 1957. The theory of infiltration.. I. The infiltration equation and its
solution. Soil Sci. 83:345-357
Philip, J. R. 1969. Theory of infiltration . I n advances in Hydroscience, vol. 5. ed
Ven Te Chow, pp. 2 15-296. Academic Press, New York
Puckett, W.E., J.H.Dane, and B.F. Hajek. 1985. Physical and mineralogical data to
determine soil hydraulic properties. Soil Sci. soc. Am. J. 49: 83 1-836
Ragab, R., and J.D. Cooper. 1990. Obtaining soil hydraulic properties from field,
laboratory, and predictive methods. Nirex Safety Studies. Nirex,
Harwell UK.
Ragab, R., and J.D. Cooper. 1993. Variability of unsaturated zone water transport
parameters: implications for hydrological modelling 1 . In situ
measurement. J. of hydrology. 148: I 09- 131
Rawls. W.J., annd D.L. Brakensiek. 1985. Predicting of soil water properties for
hydrologic modeling. In watershed management in the eighties. ed
E.B. Jones and T.J. Ward, pp293-299. Proc. Irrig. Drain. div.. ASCE
Denver, CO. 30 April-1 May. New York.
Rawls, W.J., D.L. Brakensiek, and K.E.Saxton. 1982. Estimation of soil water
properties. Trans. ASAE 25:13 16-13.20,1328
Rawls, W.J., T.J. Gish, and D.L. Brakensiek. 1991. Estimating soil water retention
from physical properties and characteristics. In Advances in Soil
Science. B.A. Stewart (ed) Springer-Verlag, New York.
Rawls, W. J., L. R. Ahu-ja, D. L. Brakensiek, A. Shirmohammadi. 1993. Infiltration
and soil water movement. In Handbook of hjidrology. ed D. R.
Maidment, pp5.1-5.5 1. New York. McGraw-Hill, Inc.

Reynolds,

W. D. 1993a. Saturated hydraulic conductivity: Laboratory


measurement. In Soil sampling and Me1hod.s of Analysis, ed. M . R.
Carter,pp589-598. Boca Raton, FL.: Lewis, Canadian Society of Soil
Science.

Reynolds, W. D. 1993b. Saturated hydraulic conductivity: Field measurement. In


Soil sanzpling and Methods of Analysis, ed. M. R. Carter,pp599-613.
Boca Raton, FL.: Lewis, Canadian Society of Soil Science
Reynolds W. D. 1993c . Unsaturated hydraulic conductivity: Field measurement. In
Soil sampling and Methods of Analysis, ed. M. R. Carter, pp. 633-644.
Boca Raton, FL.: Lewis, Canadian Society of Soil Science
Reynolds, W. D., D. E. Elrick. 1985. Measurement of field-saturated hydraulic
conductivity, sorptivity, and conductivity-pressure head relationship
using "Guelph permeameter". In Proceedings, National Water Well
Association Conference on Characterization and Monitoring of the
vadose zone, pp9-33. Denver, Colorado
Reynolds, W. D., D. E. Elrick. 1986. A method for simultaneous in situ
measurement in the vadose zone of field-saturated hydraulic
conductivity, sorpitivity. and the conductivity-pressure head
relationship. Ground water Moniloring Review 6 : 84-95
Reynolds, W. D., D. E. Elrick. 1987. A laboratory and numerical assessment of the
Guelph permeameter method. Soil Sci. 144: 282-299
Reynolds, W. D., D. E. Elrick. 1990. Ponded infiltration from a single ring: I.
Analysis of steady flow. Soil Sci. Soc. Am. J. 54: 1233-1241.
Reynolds, W. D., W. DZebchuk. 1996. Hydraulic conductivity in a clay soil: two
measurement techniques and spatial characterization. Soil Sci. Soc. Am.
J. 60: 1679-1985.
Reynolds, W. D., D. E. Elrick, and G. C. Topp. 1983. A reexamination of the
constant head well permeameter method for measuring saturated
hydraulic conductivity above the water table. Soil Sci 136: 250-268.
Reynolds. W. D., D. E. Elrick, and B. E. Clothier. 1985. The constant head well
permeameter: Effect of unsaturated flow. Soil Sci. 139:172-180.
Reynolds, W. D., S.R. Vicira and G. C. Topp. 1992. An assessment of the singlehead analysis for the constant head well permeameter. Can. J. Soil Sci.
72: 489-501
Reynolds, W. D., B. T. Bowman, R. R. Brunke, C.F. Drury, and C. S. Tan.2000.
Comparison of Tension infiltrometer, Pressure infiltrometer, and soil
core estimates of saturated hydraulic conductivity. Soil Sci. Soc. Am. J.
64:478-484.

Rogres, L.S., J.L. Fouss and C.E. Carter. 1987. Field experience with a velocity
head permeameter. In drainage design and management. Proc. Of the
5'h national Symposium. Dec. 14- 1 5.
Rogers, J. S., V. McDaniel, and C. E. Carter. 1985. Determination of saturated
hydraulic conductivity of a commerce silt loam soil. Trans. ASAE 28
(4): 1141-1 144.
Rubin, J., and Steinhardt, R., 1963, Soils water relations during rain infiltration; Part
I-Theory. Soil Sci. Soc. of Am. Proce. Vol. 27: 246-251.
Salverda, A. P. and J. H. Dane. 1993 An examination of the Guelph permeameter
for measuring the soil's hydraulic properties. Geoderma. 57: 405-42 1 .
Saxton, K.E., Rawls, W.J., Romberger, J.S., and Papendick, R.I. 1986. Estimating
generalized soil water characteristic from texture. Soil Sci. Soc. Am. J .
5O:lO3 1-1 036
Schaap, M.G., and F. J. Leij, 1998. Database related accuracy and uncertainty of
pedotransfer functions. Soil Sci. 163: 765-779
Schaap, M.G.;and F.J. Leij, 2000. Improved prediction of unsaturated hydraulic
conductivity with the Muale-van Genuchten model. Soil.Sci. Soc. Am.
J.64:843-85 1
Schaap, M.G., F.J. Leij, and M.Th. van Genuchten. 1998. Neural network analysis
for hierarchical prediction of soil hydraulic properties. Soil Sci. Soc.
Am. J. 62: 847-855
Schaap, M.G., F.J. Leij, M.Th. van Genuchten. 2001. ROSETTA: A computer
program for estimating of soil hydraulic parameters with hierarchical
pedotransfer functions. J. of Hydrology 25 1 : 163-176
Singh, R., D. K. Das, A. K. Singh. 1992. Prediction of hydrological characteristics
from basic properties of alluvial soils. J. of Ind. Soc. of soil sci. 40 (1):
180-183
Skaggs. R. W. 1976. Determination of the hydraulic conductivity-drainable porosity
ratio from water table measurements. Trans. ASAE 19 (1): 73-80, 84.
Smedema, L. K.: and D. W. Rycroft. 1983. Land Drainage planning and design o f
agricultural drainage systems'. Ithaca: Cornel l University Press.
Soil Conservation Service. 1972. Drainage of agricultural land. Washington, D.C.
Soilmoisture Equipment Corp. 1987. 2800K1 Operating Instructions. Santa
Barbara, Calif.: Soilmoisture Equipment Crop.
Stephens, D. R., S. P. Neuman. 1982a. Vadose zone permeability tests: Summary.
J. Hydrology Division Proc. Am. Soc. Civil Eng. 108: 623-639.

Stephens, D. R., S. P. Neuman. 1982b. Vadose zone permeability tests: Steady state
results. J. Hydrology Division /'roc. Am. Soc. Civil Eng. 108: 640-659.
Stephens, D. R., S. P. Neuman. 1982c. Vadose zone permeability tests: Unsteady
flow. J. Hydrology Divi.sion Proc. Am. Soc. Civil Eng 108: 660-677.
Stephens, D.B., K. Lamber, and D. Watson. 1984. Influence of entrapped air on
field determination of hydraulic properties in the vadose zone. In
Proceeding conference on characterization and monitoring in the vadose
zone. Pp57-76. National Water Well Association Worthington USA
Stephens, D.B., K. Lamber, and D. Watson. 1987. Regression models for hydraulic
conductivity and field test of the borehole permeameter. Water Resour.
Res. 23 (12): 2207-22 14
Talsma, T. 1960. Comparison of field methods of measuring hydraulic
conductivity. Trans. Congr. Irrig. and Dra. IV(6):C 145-C 156.
Talsma, T. 1987. Re-evaluation of the well permeameter as a field method for
measuring hydraulic conductivity. Aust. J. Soil Res. 25: 361 -368.
Talsma, T., and P. M. Hallam. 1980. Hydraulic conductivity measurement of forest
catchments. Aust. J. Soil Res. 18: 254-261.
Tietje, 0 . and Hennings, V. 1996. Accuracy of the saturated hydraulic conductivity
prediction by pedo-transfer functions compared to the variability within
F A 0 textural classes. Geoderma 69: 71 -84
Tietje, 0. and M. Tapkenhinirichs, 1993. Evaluation of pedotransfer functions. Soil
Sci.Soc. Am. J. 57: 1088-1095
Topp, G. C. 1993. Soil water content. In Soil sampling and Methods ofAnalysis, ed.
M. R. Carter. pp541-557. Boca Raton, FL.: Lewis, Can. Soc. Soil Sci.
Topp, G. C. and M. R. Binni. 1976. Field measurement of hydraulic conductivity
with a modified air-entry permeameter. Can. J. Soil Sci. 56: 139-147.
Uma, K.O.. B.C.E. Egboka and K.M. Onuoha. 1989. New statistical grain-size
method for evaluating the hydraulic conductivity of sandy aquifers. J . of
Hydrology. 108: 343-366.
U. S Bureau of Reclamation. 1965. Earth Manual. Washington, D. C.: U. S.
Department of the Interior, Bureau of Reclamation Superintendent of
Documents, U. S. Government Printing Office.
U. S. Bureau of Reclamation 1978. Drainage Manual. Washington, D. C.: U. S.
Department of the Interior, Superintendent of Documents, U. S.
Government Printing Office.
Reeuwi-jk Van, L. P. 1993. Procedures for soil anal,vsis. Technical paper 110.9.
International Soil Reference and Information Center.

Vereecken, H. 1995. Estimating the saturated hydraulic conductivity from


theoretical models using simple soil properties. Geodernza 65: 81 -92
Vereecken, H. 20002. Comment on the paper, Evaluation of pedo-transfer function
for unsaturated soil hydraulic conductivity using an independent data
set. Geoderma 108:145-147
Vereecken, H., J. Maes, P. Darius and J. Feyen. 1989. Estimating the soil moisture
retention characteristic from texture, bulk density, and carbon content.
Soil Sci. 148: 389-403
Vereecken, H., J. Maes, and J. Feyen. 1990. Estimating of unsaturated hydraulic
conductivity from easily measured soil properties. Soil Sci. 149: 1 - 1 2.
Wagner, B.. V.R. Tamawski, G. Wessolek and R. Plagge. Suitability of models for
the estimation of soil hydraulic parameters. Geoderma 86: 229-239
Wagner, B., V.R. Tamawski, V. Henning, U. M: Her, G. Wessolek and R. Plagge.
2001. Evaluation of pedo-transfer functions for unsaturated hydraulic
conductivity using an independent data set. Geoderma 102: 275-297
Warrick, A. W. and D. R. Nielson. 1980. Spatial variability of soil physical
properties in the field. In application of soil physics, ed D. Hillel
pp.3 19-344. New York; Academic Press.
Watt, F.C., H. F. Huckle, and R. F. Pactzold. 1982. Field vs. Laboratory determined
hydraulic conductivities of some slowly permeable horizons. Soil Sci.
Soc. Am. J. 46: 782-784
Wilson, G. V., J. M. Alfonsi, and P. M. Jardine. 1989. Spatial variability of
saturated hydraulic conductivity of the subsoil of two forested
watersheds. Soil Sci. Soc. Am. J. 53: 679-685.
Wong. 1. F. T. 1970. Reconnaissance soil survey of Selangor. Ministq, of
Agriculture and Lands Malaysia
W:

sten, J.H.M., P.A. Finke, and M.J.W. Jansen. 1995. Comparison of class and
continues pedotransfer functions to generate soil hydraulic
characteristics. Geoderma 66: 227-237

W:

sten, J.h.M., A. Lilly, A. Nemes, and C. Le Bas. 1999 Development and use of
a database of hydraulic properties of European soils. Geodern~an
90: 169-185

Wu, L., J. B. Swan, J. L. Nieber, and R. R. Allmaras. 1993. Soil-macropores and


layer influences on saturated hydraulic conductivity measured with
borehole permeameters. Soil Sci. Soc. Am. J. 57: 9 I 7-923.
Xiang J. 1994. Improvements in evaluating constant-head pernieameter test data.
Journal of Hydrolog)) vol. 1 62(1): 77-97

Youngs, E .G. 1991. Hydraulic conductivity of saturated soils. In Soil analy,~i.~:


Physical methods, ed. K. A. Smith and C. E. Mullinns, ppl80-199. New
York: Marcel Dekker.
Zhanga, Z.F., R.G. Kachanoskic, G.W. Parkinb and B. Sic. 2000. Measuring
hydraulic properties using a line source. I1 Field test. Soil Sci. Soc. Am.
J. 64: 1 563-1 569.

Appendix A
Determination of soil properties

Appendix

A-1.

Determination of Soil Moisture content ( M e ) using the


gravimetric method

Materials

1. Sampling material. Soil cores, auger


2. Soil container with tight fitting lids
3. Oven
4. Desiccator with active desiccant

5. Balance accurate to the nearest 0.001 g accuracy.

Procedure

1. Collect soil samples at the desired depths. Place the soil samples
immediately into dry tared moisture tins of known weight and enclosed by putting
on the tight fitting lids.

2. The moist soil sample was weighed to the nearest 0.00lg. Each tin's lid was
removed from the tin and placed on the bottom of its tin.

3. The tared moisture tins, with the samples place in the drying oven set at 105 f

5 ' for
~ 24 hours until there is no further loss in weight.
4. After 24 hours of drying, remove the tared moisture tins from the oven,

replace the lid firmly on each drying tin, cool in a desiccator, and weigh to the
nearest 0.001 g accuracy.

5 . calculate the soil moisture content (mass basis) as a percentage of the mass of dry
soil is d as follow:

MC' = I00

(mass of moist soil +tin) - (massof dry soil +tin )


mass of dry soil

Appendix A-2. Determination of Soil particle Distribution (PSD) using Pipette


method

The scheme used in this study was that adopted by the United States
Department of Agriculture (USDA) as shown in the Figure 1 Appendix A-2

Apparatus

1. Water bath or sand bath

2. Glass sedimentation cylinders, marked at 1 litre


3. Drying oven
4. Pipette sampling apparatus ( see Plate 1 Appendix A. 1)

5 . Moisture tins

6. Beakers 100, to 1000 ml


7. Evaporation dishes

8. Plunger
9. Electrical stirrer
10. Shaker

0.002
IJritish Standard
Institulion

CIAY
SILT
I

International
Society of Soil
Science

CILAY

United States
Department of
Agriculture

SAND

SILT

I Coarse

Fine

CLAY

SILT

7
GRAVEL

0.002

GRAVEL

SAND

0.05

0.10
Very
line

0.25

Fine

0.5

Med

1.0
Coarse

2.01
Vep
coarse

I1

GRAVEL

SAND

United States
Public Report
Administration

Figure 1 : Soil particle size distribution (Soil Survey Staff, 1975)

GRAVEL

Plate 1 : The pipette method for soil mechanical analysis

1 1 . Set of sieves with cover and pan

12. Stopwatch
1 3. Desiccator
14. Balance

1 5. 50 pm sieve

Reagents

1. Hydrogen peroxide (H2 02), 30%

2. Dispersing agent: sodium hexametaphosphate solution (Calgon)


3. Hydrochloric acid (HCL)
4. Ethanol

Procedure

Oxidation of organic matter

1.

Weight out approx. 20 g of 2 mm dried soil into a 1L beaker.

2.

Add about 50 ml of water and 20 ml of hydrogen peroxide (Hz 02), 30%


to the beaker. Cover beaker with watch-glass and allow to stand overnight.
In case of strong forting place beaker in basin with cold water. Frothing
can be tempered by adding a few drops of Ethanol.

3.

Next day , place the beaker on sand bath (90 "C) and regularly add 5-10
ml HZO230% until decomposition of organic matter is completed (usually
observed by the colour and the rate of the reaction of the sample)

4.

Add water to volume of water about 300 ml.

5.

Place on sand hot plate and carefully boil for 1 hour to remove any
remaining H2 0 2 .

Removal of Carbonate

Because of the carbonate is absent, the procedure for removal of carbonate


was omitted.

Dispersion

1. Transfer suspension quantitatively to a stirrer cup

2. Add 30 ml of Calgon to the suspension and disperse with the soil stirrer for 10
to 15 min. Wash any soil sticking to the stirrer into the cup.
3. Pass the suspension through a 50 pm sieve which placed in a funnel positioned
above a sedimentation cylinder by a retort stand.
4. Make to 1 L mark with water.
5. Wash the sand fraction remaining on the sieve quantitatively into crucible of
knowing weight and oven dry at 105 O C for 24 hrs.

6. Transfer the dried sand to the top sieve of a stacked set of sieves of the
following mesh sizes: 500 pm ,250 pm, 100 pm, and 50 pm. Sieve for 10 min
on the sieve shaker.
7. Empty each sieve into a traded weighing dish. Weigh with 0.01 g accuracy (net
weights WI, Wi, , W,,, and W,, individual sand fractions)

8. Stir the suspension in the sedimentation cylinder for I min with a hand

stirrer, using an up and down motion. Place the cylinder on the table.
Sample the 0-50 pm fraction at the 10 cm sampling depth extract at the
calculated time (Table 1).
9. The aliquot evaporate on sand bath, dried at 105 OC overnight, cold in a

desiccator, and the residue weigh (net weight Wv for fraction <0.02mm)
with 0.00 1 g accuracy.
10. For the clay fraction, the temperature of the suspension was measured and
the procedure described for the 50 pm fraction was repeated at the suitable
sampling time, (Table 1). Transfer the aliquot to a basin, evaporate on a
sand bath and dry overnight at 105 "C, cool in a desiccator, and the residue
weigh (net weight Wvrfor fraction ~ 0 . 0 0 2mm) with 0.001g accuracy
11. To avoid the error that may occur in calculation of the silt and clay
percentages due to the dispersion agent (Calgon), 30 ml of the calgon
solution was dilute in 1L, and 20 ml extract from the blank cylinder as
describe for the silt and clay fractions. Then sample was transfer to basin,
evaporate on sand bath and dry overnight at 105 "C, cool in a desiccator,
and the residue weigh (net weight Wvll for the dispersion agent.) with
0.001g accuracy

The oven-dry sample was weighed after all treatments were taken as the
basis for the calculations to determine the percentage of the total soil in suspension
present in each sample as follows :

Clay (<0.002mm) = (WVIIx 50) - (Wvr x 50)

(wt. A)

Silt (<0.02mm) = (Wvll x 50) - (Wv x 50)

(wt. B)

Sand (>0.02mm) = W, + W,, + W,,, + W,,

(wt. C)

Sample weight

=A

+B +C

where W, through W , , = weight individual sand fraction, Wv

weight 20 ml

pipette aliquot of fraction < 0.02mm, WVI= weight 20 ml pipette aliquot of


fraction < 2 pm Wvll = weight 20 ml pipette aliquot of blank.

Table 1 : Settling times for the soil fractions (C0.02 mm and <0.002 mm) at various
temperatures. Calculated for a depth 10 cm. (after Chopra and
Kanwar; 1976)
Tempeeture

1
I

25

7 hr 5.6 rnin

4 min 15.5 sec


4 min 9.7 sec

6 hr 56.2 rnin

4 rnin 4.1 sec

6 hr 46.9 min

3 rnin 58.8 sec

6 hr 38.0 rnin

3 min 53.6 sec

6 hr 29.3 rnin

3 min 48.5 sec

6 hr 20.9 rnin

3 min 43.7 sec

6 hr 12.8 min

3 min 39. Sec

6 hr 5.0 rnin

The proportional amounts of the fractions was calculated by:

% Clay

( < 0.00

2mm)

A
sample

XI00
wt

YO Silt

( < 0.02mm)

- 2mm)

% Sand

(0.5

% Sand

(0.25

- 0.5mm)

wt

W i

sample

% S a n d ( 0 . 1 - 0 . 2 5 mm ) =

% Sand

B
sample

wt

W ii

sample

wt

iii
sample wt

(0.05 - O.lmm ) =

X 100

X 100

X 100

x 100

W iv
X 100
sample
wt

Use then the DUSA texture triangle, which is shown in Figure 2 to obtain the soil
textural classes by plotting percentages of clay, silt, and sand.

Percent SAND

Figure 2: Soil triangle of the basic soil textural classes (Soil Survey Staff 1975)

Appendix A-3 Determination of Bulk density

In this study, bulk density was determined employing a core method and
following the procedure as described by Blake and Hartge (1 986).

Materials

1 . Cylinders core with 70 mm diameter and 40 mm height and 76mm

diameter and 76 mm height.


2. Calculate the volun~eof the metal cylinder using the equation 1

3. Balance with 0.0 1 g accuracy


4. Drying oven

5. Plastic bags to hold soil cylindrical core.


6. Spatula
7. Disks to cover ends of cores

Procedure

I . Identify and weigh cylindrical cores and label as W1 (g) .


2. Weigh the tins and label weight as W2 (g)

3. Prepare the soil surface by removing vegetation and loss soil. Excavate to
the desired depth.
4. Carefully push the metal cylinder into the soil profile pits at the desired
depths.

5. Excavate around the cylinder with trowel until the cylinder exposed

6. Using the spatula carefully dig out the ring and examine the undisturbed
core for sign of shattering or compression.

7. Cut smoothly at both ends, clean from outside to preserve a known volume
of sample as it exist in situ.

8. Transfer the content of the cylinder into the preweighed weighing tray
which then weighed and label as W3.

9. Place the tray in an oven and dry at 105 OC for 24 hrs. After drying and
cooling in desiccator, weigh the tray plus dry soil plus core cylinder and label
as W5 (g).
10. Calculated soil bulk density using the equation 7 and 8

Where
D,, = the soil bulk density (g,/cm3)
M,

= soil

dry mass (g)

Vb

= the

total soil volume (include the volume of pore space.) (crn3).

V b = totalsoilv olume = core volume

= Irr 2 h

Where
r = radius of core (cylinder) (cm)
h = the height of the core (cm)
7c = 3.142

Appendix A-4 Calculation of total soil porosity from particle and bulk
densities

Porosity can be derived from measurements of soil dry bulk density, Db,
and soil particle density, Ds using the following relationship for determining
porosity by bulk density:

Where

the porosity of the soil

Dh = the soil bulk density


D, = the particle density.

In this study Because the organic matter was less than 1 % (mineral soil),
according to Culley ( 1 993), the porosity was calculated on the basis of an assumed
soil Dh of 2.65 g/cm3 (2.65 Mg m'3).

Appendix A-5 Determination of Organic matter (OM) using determination


of loss on ignition technique

This loss on ignition actually represents the organic fraction present in the
soil; after ignition the as portion of the organic matter still remains.

Apparatus

1. Crucible
2 . Desiccator
3. Drying oven
4. Muffle furnace

5 . Metal tong
6 . Balance with 0.001 (g) accuracy

Procedure

1. Clean and rinse crucibles with distilled water and dry in the oven at 105 O C .
After the crucibles cooled in a desiccator weigh and record the weight as
(WI).
2. Place about 5 g of oven dry soil the previously weighed empty crucible.
3. Cover the crucibles with the soil and place in the muffler furnace. The
temperature was increased gradually to 350 "C and the samples were ignited
for 7-8 hours.
4. Transfer the crucibles from the furnace into the desiccator using metal tong,
allowed to cool and weigh. Label the weight as (W2).
5. Calculate the organic matter by using the following formulae:

% organic matter= -

ws

Where
WI

= Dry

weight of crucible

W2

= Weight

of crucible + dry sample

W,

= Weight

of oven dry soil sample

APPENDIX B

Examples of Ka Calculations
Guelph Permeameter Data Sheet

Investigator: -Abdolhakem 0 . Mohamed


Location: Plot 5
Test No.: 1
Radius: 3 cm (standard calcs assume 3 cm radius)
Depth of hole: 60 cm
Reservoirs used during test (check one):
Combined: ({ ) Inner only: (
)
Reservoir constant used:

Water level in the well = 5 cm

Time
t

~t

water
rate of
level in
change
reservoi ~h
r
(cm) Ah 1 At

Steady rate for 4


consecutive readings (R,):

Water level in the well =

Time

~t

(min) (min

water
level in
reservoi
r

Ah

(cm

Steady rate for 4 consecutive


readings (R2):

rate of
change

l-Guelph Permeameter Kfs Calculations


Reservoir Constant (A): 35.36
Steady flow rate: Q l= A x RJ60 = 35.36 x 0.1 0160 = 0.0589 cm s-'
Q2 =A x R2/60 = 35.36 x O.l25/6O= 0.07367 cm s-'
[The reservoir constant, which has different values depending on whether only one
or both of the instrument reservoirs are used ( It is equal to 35.36 cm' in this
case)3.

Method 1:
Use this first. Assumes 3 cm borehole, 5 and 10 cm heads

Since the result of the both heads in this example gives negative values, the Kn
was then calculated according to the method2
Reservoir constant:
A: 35.36
Radius of borehole :
r: 3 cm
H I : 5 cm
Water depth in borehole during first test
Water depth in borehole during second test
H2: 10 cm
Shape factor for H l/r (obtain from graph)
C1: 0.08
Shape factor for H2/r (obtain from graph)
C2: 1.32
Steady flow rate: Q l = A x R1160 = 35.36 x 0.10160 = 0.0585 cm s-'
Q2 = A x R2/60 = 35.36 x 0.1 25/60 = 0.0737 cm s-'

Based on the soil structural and texture consideration, a constant a* value equal to
0.12 cm-' was assumed for applying the single head level technique (SH) analysis
(Elrick and Reynolds, 1992).
Method 2

Use this if Method 1 gives negative KfS.Estimate a* and calculate for R1 and R2
Results for R1

CAR

K$. =

2xH-+Cnr

2nH

+-

a*

0.08 x.35.36.x 0.0585


= 3.93 x 1 o
2
2 2n5
2 x 5 +O.O8x3 +0.12

- cm
~

1.32 x 3 5 . 3 6 ~0.0737
= 2.89 x 1 o
2
2 2x10
2x10 + 1 . 3 2 ~ 3 +0.12

- ~cm2 sec-'

sec-I

Results for R2

CAR
Kj.s =

2nH +Cnr

2nH

+-

a*

Where

C: dimensionless shape factor from graph on following page


A: Reservoir constant
R: Steady state rate of fall
H: Steady depth of water in boring
r: radius of well
K*: 0.01 compacted structurless, clayey materials-landfill cap, lacustrine, marine
clays0.04 fine-grained, unstructured clay 0.12 structure soil, clays through loams,

unstructured medium to fine sand 0.36 coarse-grained sand and gravel, highly
structured soil with large cracks

2- Example for the Constant Head Permeameter K, Calculation

Soil profile No 2 plot Ill, depth 0-1 5cm


-the inner diameter of the ring (D) = 76.2 mm
-the length of the sample intact (L) = 8.0cn-1,

- the constant depth of water on the soil surface (H) = 4.3 cm,
-the cross sectional area A = 45.58 cm2
- the steady state volume of water pass through the sectional-area A in time t (Q) =

0.3523 cm3s-'
from:

APPENDIX C
SPSS ANA STATGRAPHICS SAMLPES

relation?
CLAY

SILT

SAND

OM

BD

POROSITY

MC

KFS

KS

^t*

SAND
-.768'
,000
93
-.775'
,000
93
1

Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-lailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N

.301
93
,109
,300
93
.224*
,031
93
-.213*
.04 1
93
,058
,656
62

Correlation is significant at the 0.05 level (2-tailed).


Correlation is significant at the 0.01 level (2-tailed)

93
,034
.743
93
,101
,333
93

MC
,224'

I
1

KFS
-.213'

KS
,058

Regression
Variables Enteredl~emovec!
Variables
Removed

Variables
Entered
MC,
PORESIT
Y, OM,
CLAY.
SILT, a
SAND

Model
1

Method

Enter

Backward (criterion: Probability of F-to-remove

CLAY

>= .200)

Backward (criterion: Probability of F-to-remove


>= ,200).
Backward (criterion: Probability of F-to-remove
>= ,200).
Backward (criterion: Probability of F-to-remove
>= ,200).

MC
PORESITY
SAND
a. Tolerance = ,000 limits reached.

b. Dependent Variable: SCKS


Model summary'

Model
1

I
1

R
R Square
.357a (
,128

I
1

Adiusted
R square
.032

Std. Error of
the Estimate
,002635545

a. Predictors: (Constant), MC, PORESITY, OM. CLAY, SILT,


SAND

b. Predictors: (Constant), MC, PORESITY, OM, SILT, SAND


c. Predictors: (Constant), PORESITY, OM, SILT, SAND
d. Predictors: (Constant), OM, SILT, SAND
e. Predictors: (Constant), OM, SILT
f.

Dependent Variable: SCKS

Model
1

Regression
Residual
Total
Regression
Residual
Total
Regression
Residual
Total
Regression
Residual
Total
Regression
Residual
Total

Squares
,000

I
1

df
6

Square
,000

I
1

F
1.341

--

a. Predictors: (Con$ ant), MC, PORESITY, OM, CLAY, SILT, SAND


b. Predictors: (Con: ant), MC, PORESITY, OM, SILT, SAND

c. Predictors: (Constant), PORESITY, OM, SILT, SAND


d. Predictors: (Constant), OM, SILT, SAND
e. Predictors: (Constant), OM, SILT
f.

Dependent Variable: SCKS

,2515~

Model
1

Unstandardized
Coeff ents
B
Std. Error
(Constant)
-.014
,030
CLAY
,000
9.032E-05
SlLT
,000
2.892E-04
SAND
,000
1.454E-04
OM
.001
-.001
PORESITY 5.035E-05
.ooo
MC
,000
4 734E-05
(Constant)
-.006
,007
SlLT
,000
2 121E-04
SAND
,000
6.300E-05
OM
-.001
.001
PORESITY 4.752E-05
.ooo
MC
,000
4.875E-05
(Constant)
-.005
.006
SlLT
,000
2 105E-04
SAND
,000
5.883E-05
OM
-.001
,001
PORESITY 4.843E-05
,000
(Constant)
.005
SlLT
,000
SAND
.ooo
OM
,001
(Constant)
SlLT
OM

a. Dependent Variable: SCKS

Standardized
Coefficients
Beta
,171
,630
,508
-.215
,082
,049

Sig.
,635

Excluded variables'

Model
1
2
3

BDENSITY
BDENSITY
CLAY
BDENSITY
CLAY
MC
BDENSITY
CLAY
MC
PORESITY
BDENSITY
CLAY
MC
PORESITY
SAND

Beta In
19.644a
20.16gb
.17Ib
18.852c
.I
7gc
.05OC
-.078d
.I
34d
.053d
.O7gd
-. 094e
-.118e
.035e
.095e
.22ge

,864
,904
,290
,857
,308
,389
-.615
.232
,409
,620
-.748
-.a37
,274
,753
,925

Slg.
,391
,370
,773
,395
,760
,699
,541
,817
,684
,538
,458
,406
,785
,455
,359

Collinearity
Partial
Correlation
,117
,121
,039
,114
,041
.052
-.081
,031
,054
,082
-.098
-. 109
,036
,098
,121

Tolerance
3.086E-05
3.145E-05
4.589E-02
3.191E-05
4.596E-02
.933
,945
4.663E-02
.934
,944
,966
,762
.954
.966
,249

a. Predictors In the Model: (Constant), MC, PORESITY, OM, CLAY, SILT, SAND
b. Predictors in the Model: (Constant), MC, PORESITY, OM. SILT, SAND

c. Predictors in the Model: (Constant), PORESITY, OM, SILT, SAND


d. Predictors in the Model: (Constant), OM, SILT, SAND
e. Predictors in the Model: (Constant), OM, SILT
f.

Dependent Variable. SCKS

Variables EnteredlRemoved
Variables
Removed

Variables
Model
1
LOGDB,
LOGSILT,
LOGOM,
LOGCLAY.
LOGSAND,
LOGPOROSIT?

Method

Enter

Backward (criterion: Probability of


F-to-remove >= ,200).
Backward (criterion: Probability of
LOGDB
F-to-remove >= ,200).
Backward (criterion: Probability of
LOGMC
F-to-remove >= ,200).
Backward (criterion: Probability of
LOGPOROSITY
F-to-remove >= ,200).
LOGCLAY

a. Dependent Variable: LOG Kfs


Model summary'

Model
1

I
I

I
1

R
R Square
.327= (
.I07

--

1
1
1

Adjusted
R Square
.033

Std. Error of

1 the Est~mate
1 .8700719440

--

a Predrctors (Constant), LOGMC. LOGDB, LOGSILT,


LOGOM, LOGCLAY, LOGSAND, LOGPOROSITY
b Pred~ctors(Constant), LOGMC, LOGDB, LOGSILT,
LOGOM, LOGSAND, LOGPOROS
c. Predictors: (Constant), LOGMC, LOGSILT, LOGOM,
LOGSAND, LOGPOROSITY
d. Predictors: (Constant), LOGSILT, LOGOM, LOGSAND,
LOGPOROSITY
e. Predictors: (Constant), LOGSILT. LOGOM, LOGSAND
f. Dependent Variable: LOG Kfs

Model
1

Regression
Residual
Total
Regression
Residual
Total
Regression
Residual
Total
Regression
Residual
Total
Regression
Residual
Total

Squares
7.699
64.347
72.046
7.698
64.348
72.046
7.695
64.351
72.046
7.427
64 619
72.046
6.391
65.655
72.046

I
I

df

Mean
Square
1.100

I
1

F
1.453

Sig.
.195=

a. Predictors: (Constant), LOGMC, LOGDB, LOGSILT, LOGOM,


LOGCLAY, LOGSAND, LOGPOROSITY
b. Predictors: (Constant), LOGMC, LOGDB, LOGSILT, LOGOM,
LOGSAND, LOGPOROSITY

c. Predictors: (Constant), LOGMC, LOGSILT, LOGOM, LOGSAND,


LOGPOROSITY
d. Predictors: (Constant), LOGSILT, LOGOM, LOGSAND,
LOGPOROSITY
e- Predictors: (Constant), LOGSILT, LOGOM, LOGSAND
f.

Dependent Variable: LOG Kfs

.I
27b

.075=

.046*

.040e

Coefficients?

Model
I

Unstandardized
snts
Std.
B
Error
(Constant)
LOGCLAY
LOGSILT
LOGSAND
LOGOM
LOGDB
LOGPOROS
LOGMC
(Constant)
LOGSILT
LOGSAND
LOGOM
LOGDB
LOGPOROS
LOGMC
(Constant)
LOGSILT
LOGSAND
LOGOM
LOGPOROS
LOGMC
(Constant)
LOGSILT
LOGSAND
LOGOM
LOGPOROS
(Constant)
LOGSILT
LOGSAND
LOGOM

-16.634
,077
,827
3.290
,686
1.591
3.410
-.863
-16.397
.808
3.180
,682
1.657
3.455
-.859
-14.301
.818
3.203
,680
2.317
-.856
-15.903
.871
3.432
,719
2.345
-12.220
,941
3.436
,671

a. Dependent Variable: LOG Kfs

33.969
2.496
,943
4.149
,383
25.838
17.850
1.444
32.889
,690
2.061
,355
25.600
17.686
1.430
5.731
,669
2.018
,352
1.982
1.421
5.057
,660
1.974
,345
1.975
4.003
,659
1.979
.343

Standardized
Coefficients
Beta

t
-. 490

,007
.I86
,245
,207
,057
,177
-.064
.I82
.237
,206
.059
,179
-.064
.I84
,239
,205
.I20
-.064
.I96
,256
.217
.I22
,211
.256
,202

,031
,877
,793
1.790
,062
,191
-.598
-.499
1.171
1.543
1.919
.065
,195
-.601
-2.495
1.223
1.588
1.930
1.169
-.602
-3.145
1.319
1.739
2.084
1.187
-3.052
1.427
1.736
1.952

Sig.

,626
,976
,383
,430
,077
,951
,849
,552

Excluded Variable$

OSITY

, 1 2 2 ~ 1.187

,238

.I26

a. Predictors in the Model: (Constant), LOGMC, LOGDB, LOGSILT,


LOGOM, LOGSAND, LOGPOROSITY
b. Predictors in the Model: (Constant), LOGMC, LOGSILT, LOGOM.
LOGSAND, LOGPOROSITY
c. Predictors in the Model: (Constant), LOGSILT, LOGOM, LOGSAND,
LOGPOROSITY
d. Predictors In the Model: (Constant), LOGSILT, LOGOM, LOGSAND
e. Dependent Variable: LOG Kfs

.972

Comparison between the three methods


Analysis of Variance

...............................................................
Source

Sum of Squares Df

Mean Square F-Ratio

...............................................................

P-Value

Between groups
Within groups

The F-ratio equals 31.7384, is a ratio of the between-group


estimate to the within-group estimate.Since the P-value of
the F-test is less than 0.05,there is a statistically
significant difference between the means of the 3 variables
at the 95.0% confidence level. Multiple Range testes were used
to determine which means are significantly different.
Options.

Multiple Range Tests

Contrast

Difference

+/-

Limits

.............................................................
0.802036
loglO(Kfs 0.15m) - log(Ks 0.15m)
*2.25321
0.88961
loglO(Kfs 0.15m) - loglO(DR1 Ks)
-0.424283
log (Ks 0.15m) - loglO(DR1 Ks)
*-2.67749
0.942799
* denotes a statistically significant difference

Homogenous groups are identified using columns of X's.


Within each column, the levels containing X'sform a group
of means within which there are no statistically significant
differences. The method currently being used to discriminate
among the means is Tukey's honestly significant difference (HSD)
procedure. With this method, there is a 5.0% risk of calling
one or more pairs significantly different when their actual
difference equals 0.

Comparison between Ks values obtalned by GP and SCHP methods

Source

Sum of Squares

Df

Mean Square

F-Ratio

P-Value

..................................................................

Between groups
134.825
7
19.2608
29.78
0.0000
Wlthln groups
95.0606
147
0.64667
...................................................................
Total (Corr.i
229.886
154

The F-ratio equals 29.7845, 1s a ratio of the between-group


estlmate to the withln-group estimate.Since the P-value of
the F-test 1s less than 0.05,there 1s a statlstlcally
slgnlficant difference between the means of the variables at
the 95.0% confidence levei. Multiple Range testes were used
to determine which means are slgniflcantly different.

Multlpie Range Tests

.....................................

........................

Method: 95.0 percenc Tukey HSD


Count
Mean

Homogeneous Groups

.....................................
loglO(Kfs 0.9m)
25
loglO(Kfs 0.3m)
28
loglO(Kfs 0.6m)
21
loglO(Kfs 0.15~1) 19
log10 (Ks 0.6m)
24
log10 (Ks 0.9m)
12
log10 (Ks 0.3m)
12
loglO(Ks 0.15m)
14

-5.46234
-5.27416
-5.20492
-4.29633
-3.65511
-3.47689
-3.31918
-2 .a4443
Difference

loglO(Kfs
loglO(Kfs
loglO(Kfs
log10 (Kfs

0.15m) - loglO(Ks 0.15m)


0.3m) - loglO(Ks 0.3m)
0.6m) - loglO(Ks 0.6m)
0.9m) - log10 (Ks 0.9m)

*-1.4519
'-1.95498
*-I .54982
'-1.98545

Limits
0.870985
0.853208
0.7389
0.86843

* denotes a statistically significant difference

homogenous groups are identified using columns of


X's. Within each column, the levels containing X's
form a group of means within which there are no
statistically significant differences. The method
currently being used to discriminate among the means
is Tukey's honestly significant difference (HSD)
procedure. With this method, there is a 5.0% risk
of calling one or more pairs significantly different
when their actual difference equals 0.

Comparison between measured and predicted Ks values


Analysis of Variance
..................................................................

Source

Sum of Squares

Df

Mean Square

F-Ratio

P-Value

..................................................................

Between groups
Within groups

0.000160654
0.000536946

7
708

.0000229506
7.58398E-7

30.26

0.0000

The F-ratio equals 30.2619, is a ratio of the between-group estimate


to the within-group estimate.Since the P-value of the F-test is less
than 0.05,there is a statistically significant difference between the
means of the 8 variables at the 95.0% confidence level. Multiple Range
tests were to aetermlne which means are significantly different.

RIODATA OF THE AUTHOR

Mr. Abdolhakem 0 Mohamed was born in August 27. 1961 in Tobruk, a small
town about 1300 km east of Tripoli, Libyan Arab Jamahiriya. He obtained his
Bachelor of Science degree from Soil and Water Department, Faculty of
Agricultural, University of Garyunis in July 1983. Upon his graduation, he was
appointed as a demonstrator in the Soil and Water Department, Faculty of
Agricultural, University of Omar Al-Mukhtar. During his stay in University of
Omar Al-Mukhtar he has been involved in the teaching of Irrigation and Drainage
Science, Fundamental of Soil Science, Plant Nutrition and Soil Physics. Moreover,
he was involved in soil physical properties analyses of more than 2000 sample for
Great Man Made River Project. In 1992, he was awarded a scholarship from the
university to pursue his study for Master Degree. In July, 1992 he was enrolled as
full time MS student at the Department of Biological and Agricultural
Engineering, Faculty of Engineering, Universiti Putra Malaysia majoring in Soil
Water Engineering.