Вы находитесь на странице: 1из 10

The effect of soil electrical conductivity on moisture

determination using time-domain reflectometry in


sandy soil
Z. J. Sun, G. D. Young, R. A. McFarlane, and B.M. Chambers
E.S.I. Environmental Sensors Inc., 100-4243 Glanford Avenue, Victoria, British Columbia, Canada V8Z 4B9.
Received 1 December 1998, accepted 4 September 1999.

Sun, Z. J., Young, G. D., McFarlane, R. A. and Chambers, B. M. 2000. The effect of soil electrical conductivity on moisture
determination using time-domain reflectometry in sandy soil. Can. J. Soil Sci. 80: 13–22. A series of laboratory experiments
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

was conducted, in order to systematically explore the effect of soil electrical conductivity on soil moisture determination using
time domain reflectometry (TDR). A Moisture Point MP-917 soil moisture instrument (E.S.I. Environmental Sensors Inc.,
Victoria, BC, Canada) was used to measure propagation time (time delay) of a step function along a probe imbedded in fine sand
with different moisture and salinity. The volumetric soil water content was independently determined using a balance. With the
help of the diode-switching technique, MP-917 could detect the reflection from the end of the probe as the electrical conductivity
of saturated soil extract (ECe) increased to 15.29 dS m–1. However, the relationship between volumetric soil water content and
propagation time expressed as T/Tair (the ratio of propagation time in soil to that in air over the same distance) deviated from a lin-
ear relationship as the conductivity exceeded 3.72 dS m–1. At the same water content, the time delay in a saline soil was longer
than that in a non-saline soil. This leads to an over-estimation of volumetric soil water content when the linear calibration was
applied. A logarithmic relationship between volumetric soil water content and T/Tair has been developed and this relation includes
soil electrical conductivity as a parameter. With this new calibration, it is possible to precisely determine the volumetric water con-
tent of highly saline soil using TDR.
For personal use only.

Key words: Time domain reflectometry, time delay, bulk electrical conductivity (σ), volumetric soil water content (θ), relative
permittivity or dielectric constant (εr), propagation velocity Vp

Sun, Z. J., Young, G. D., McFarlane, R. A. et Chambers, B. M. 2000. Effets de la conductivité électrique des sols sablonneux
sur les measures d’humidité par la méthode de reflectométrie en domain temporel. Can. J. Soil Sci. 80: 13–22. Une série
d’expériences en laboratoire a été conduite dans le but d’explorer de façon systématique l’effet de la conductivité électrique du sol
sur la détermination du taux d’humidité du sol par Réflectométrie en Domaine Temporel (TDR). Un instrument de mesure d’hu-
midité du sol, Moisture Point MP-917 (E.S.I. Environmental Sensors Inc., Victoria, CB, Canada) était utilisé pour mesurer les
temps de propagation d’une impulsion carrée le long d’une sonde (ligne de transmission) immergée dans du sable fin, à différents
niveaux d’humidité et de salinité. L’humidité volumétrique était déterminée de facon indépendante par mesures gravimétriques à
l’aide d’une balance. A l’aide de la technique brevetée MoisturePoint utilisant des rupteurs à diodes, MP-917 pouvait détecter la
réflexion à la fin de la sonde, jusqu’à un niveau de conductivité électrique de 15.29 dS m–1. Cependant, la relation entre humidité
volumétrique et temps de propagation (exprimé sous la forme de T/Tair: le rapport du temps de propagation dans le sol sur le temps
de propagation dans l’air, pour la même distance et le même signal) s’écartait d’une relation linéaire pour les valeurs de conduc-
tivité supérieures à 3.72 dS m–1. A teneur en humidité égale, le temps de propagation dans une eau saline était plus long que dans
une eau non saline. En conséquence, l’humidité volumétrique était sur-estimée quand l’équation linéaire habituelle était appliquée.
Une relation logarithmique entre teneur en eau et T/Tair a été établie, et cette relation varie avec la conductivité électrique. Avec
cette nouvelle calibration, il est possible de déterminer avec précision l’humidité volumétrique de sols relativement salins par la
méthode de TDR.

Mots clés: Refléctométrie en Domaine Temporel, temps de propagation, conductivité electrique apparente (σ), humidité
volumétrique (θ), permittivité relative ou constante, diélectrique (εr), vitesse de propagation Vp

Time-domain reflectometry has been used to measure volu- mittivity of the medium. Since water has a larger relative
metric water content of soil and other porous materials since permittivity or dielectric constant (εr), which is close to 80
1980 (Topp et al. 1980, 1982, 1994; Topp and Davis 1985; at 20°C, than that of other soil constituents (εr = 1 for air and
Heimovaara and Bouten 1990; Wraith and Baker 1991; εr = 2 to 5 for dry soil), the time delay will be extremely sen-
Whalley 1993; Zegelin and White 1994; Nielsen et al. sitive to the amount of water present. Generally, soil texture
1995). Time-domain reflectometry measures round-trip and composition do not have large influences on time delay
propagation time (time delay) of a step function propagating (Topp et al. 1980). Time-domain reflectometry has been
down and back along a probe buried in the medium. The widely accepted as an accurate, quick and non-destructive
time delay is converted to volumetric water content. The method to measure soil moisture. It has been found that
propagation velocity of a step function is related to the per- there was a linear relationship between volumetric soil
13
14 CANADIAN JOURNAL OF SOIL SCIENCE

water content and time delay when the latter is expressed as and the phase velocity of the wave along the direction of
T/Tair, where T is the time delay in soil and Tair is that in air propagation is given by
with the same distance (Hook and Livingston 1996).
Soil salinity, commonly expressed in terms of electrical Vp = ω/β (3)
conductivity of soil solution (ECw), or electrical conductiv-
ity of saturated soil extract (ECe), or bulk soil electrical con- From the propagation velocity of sinusoidal signals, the time
ductivity (σ) will affect TDR moisture determination in taken for a single-frequency sinusoid to travel a length L is
several ways. First, it attenuates the signal due to energy dis- given by
sipation by current flow; second, it causes a signal disper-
sion, resulting in a longer rise time; third, it will increase the Lβ L µε ′
2
L  σ − ωε ′′ 
apparent dielectric constant (Topp et al. 1994), resulting in t= = = 1+ 1+   (4)
decrease of propagation velocity, and consequently, a longer Vp ω 2  ωε ′ 
time delay. These effects will cause a considerable mea-
surement error if the soil electrical conductivity exceeds a where all of the factors are specific electrical properties of
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

certain threshold. Malicki et al. (1994) reported that the rela- the medium:
tionship between volumetric water content and permittivity µ, permeability, it equals µoµr, where µo is the permeability
of a silty loam, which was determined by time delay mea- of free space and µr is the relative permeability. For non-fer-
surement, departs from normal when the ECw of the silty romagnetic such as sandy soil, µr = 1.
loam was above 10 dS m–1. The frequency bandwidth of the ε′, real part of permittivity, it equals εoεr, where ε0 is the per-
TDR instrument also has a profound influence on measure- mittivity of free space and εr is the relative dielectric con-
ment error caused by soil electrical conductivity. Shaun et stant.
al. (1995) used a high-bandwidth (20 GHz) TDR system, ε″, imaginary part of permittivity,
and found that moisture measurements were independent of σ, bulk electrical conductivity
bulk electrical conductivity of the media, which was wetted And ω is the angular frequency of the sinusoid (ω = 2πf,
with up to 0.5 M concentration of KBr solution. However, where f is the frequency).
the common TDR systems have a bandwidth less than 2.5
For personal use only.

GHz, and ECw of soil in many areas can be higher than 10 2. The Effect of Electrical Conductivity on Time
dS m–1 because of prolonged drought and/or abusive irriga- Delay
tion practice, such as in a greenhouse environment. (a) For non-conductive medium (σ = 0)
The paper presented here explores the effect of soil elec- The term ε″ in Eq. 4 represents the energy dissipation due to
trical conductivity on water content determination for sandy the relaxation of the molecules of a medium. Free water
soil using TDR. The study of the same effect in clayey soil molecule has a relaxation frequency of approximately 15
is currently under investigation. GHz; however, bound water molecules have a much lower
relaxation frequency in the range of several hundred MHz
THEORY depending on the binding force. For TDR instruments, the
operation frequency is below 2.5 GHz. Therefore, for sandy
1. Propagation velocity
soil or other soils with low clay contents, ε″ can be neglect-
The description of the propagation of the TDR step pulse
ed. However, if the soil contains large amounts of clay, there
along the parallel-conductor transmission line of the probe
are numerous water molecules being bound to the surface of
embedded in a material medium may be approached analyt-
clay particles, then, the effect of ε″ has to be taken into con-
ically with Maxwell’s equations (Rao 1987).
sideration. For non-conductive sandy soils, both ε″ and σ
By using the phasor technique, the solution of Maxwell’s
equal zero and µr = 1, then Eq. 4 can be simplified as:
equation is given in terms of exponential functions whose
complex exponent — the “propagation constant” — cap-
tures the signal behaviors of interest for TDR purposes: the t = ( L / c) ε r (5)
attenuation and velocity of propagation. The attenuation can
be determined by using so-called attenuation constant α: where c is the propagation velocity in free space, and it
equals 1/ µ 0 ε 0 .
ω µε ′  σ − ωε ′′ 
2 This is the equation that has been commonly used in TDR-
α= 1+   −1 (1) based moisture meters to determine the dielectric constant
2  ωε ′ 
of the medium from measured time delay. To convert the
dielectric constant to volumetric water content of the medi-
and velocity can be determined by using so-called phase um, an empirical relationship is used. In non-conductive
constant β: medium, there is no signal attenuation since α = 0 when σ
= 0 according to Eq. 1.
ω µε ′  σ − ωε ′′ 
2
(b) For conductive medium (σ ≠ 0)
β= 1+   +1 (2)
2  ωε ′  In conductive medium where σ cannot be neglected, then
SUN ET AL. — ELECTRICAL CONDUCTIVITY AND TDR MEASUREMENT 15

the term σ/ωε′ in Eq. 4 must be taken into account. The time MATERIALS AND METHODS
delay increases as the conductivity σ increases. Only in a The sand used was collected from Cordova Bay area of
high-frequency range, where σ/ωε′ << 1, the effect of σ on Victoria, BC, Canada, and passed through a 2-mm sieve
time delay can be neglected. before a particle size analysis was conducted using hydrom-
The propagation time (t) of an electromagnetic wave eter method. The particle size distribution for sand was:
propagating along a transmission line with length L buried sand, 90%; silt, 8.0% and clay 2.0%. The electrical conduc-
in a conductive medium is given by (where ε″ is negligible): tivity of its saturated extract is 0.34 dS m–1 measured using
K320 Microcomputer Conductometer (Consort pvba,
1 + (σ / ωε 0 ε r ) + 1
2 Turnhout, Belgium), and the pH value for the saturated
t = ( L / c) ε r (6) extract was 6.3. The TDR instrument used was Moisture
2 Point MP-917 (E.S.I. Environmental Sensors Inc., Victoria,
BC, Canada). MP-917 utilises TDR as its baseline technol-
By comparing Eqs. 5 and 6, one may expect that the propa- ogy, but it also employs switching diode technique (Hook et
gation time would increase at least by a correction factor of al. 1992) to produce non-ambiguous time marks. This inno-
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

vation enhanced the ability of MP-917 to detect the highly


1 + (σ / ωε 0 ε r ) + 1
2
attenuated signal in high saline medium when other standard
. TDR instruments failed to show the reflection signal (Hook
2
et al. 1992). The probe used in this experiment consisted of
There are two challenges of directly using Eq. 6. First, TDR two 30-cm rectangular stainless steel bars (1.3 cm in width)
instruments send a step function that consists of a band of separated 1.5 cm by epoxy between them. Switching diodes
frequencies rather than a single one; second, the electrical were mounted at both ends of the probe. The probe was pre-
conductivity σ might also be a function of frequency viously calibrated in air and distilled water. The length of
(Hasted 1973). cable between MP-917 and probe was 2.0 m.
In determining the propagation time using diode differ- The sand was air dried for more than 1 mo and placed in
ence functions (Hook et al. 1992), the intercept of the dif- a half cylinder (52.5 cm in length and 15.2 cm in diameter
ference function base line and the line of the steepest slope PVC pipe) (Fig. 1). The total volume of sand was 4.76 L.
For personal use only.

of the reflection are used. The steepest slope corresponds to The probe was installed in the middle of the half cylinder.
the maximum frequency component in a frequency band. With this configuration, all of the sand was included in the
The maximum frequency (fm) in the frequency band can be sensitive range of the probe. This was further demonstrated
estimated by the rise time of the reflection signal using the by the observation of only 0.0025 m3 m–3 to 0.0038 m3 m–3
simple relation (Oliver and Cage 1971) increase in indicated volumetric water content when a
beaker of water was placed on the top of the half-cylinder.
fm (MHz) = 0.35/rise time (microsecond) (7) That is, bringing a volume of water next to the volume occu-
pied by the experimental sand had negligible effect on the
As a first order approximation, the maximum frequency in a measurement.
band calculated by the measured rise time can be used in Eq. The whole apparatus was set horizontally on a balance
6 to estimate the time delay error caused by electrical con- (Ohaus, Florham Park, NJ, USA). The probe was connected
ductivity of the medium. to an MP-917 and a computer with the balance connected to
The effect of σ on wave propagation properties is gov- a second computer. This set-up supported continuous mea-
erned by the expression (σ/ωε′)2; consequently, at high fre-
surement and logging of the time delay through the moist
quencies, the effect declines and is more important at low
sand and its changes in weight.
frequencies. It is found (Nyfors and Vainikainen 1989) that
At the beginning of the experiment, 1.43 L of distilled
σ is independent of frequency in the range where its effects
are observable. Therefore, it is reasonable to consider that σ water was added to the sand and the volumetric soil water
is a constant with respect to frequency. content measured at 0.30 m3 m–3 using the balance. The
cylinder was then covered with a plastic sheet to prevent
3. Signal Dispersion in Conductive Medium water loss through evaporation. Continuous monitoring of
For a signal comprising a band of frequencies, in non-con- the time delay measurement showed that 5 to 6 h after
ductive medium (σ = 0), different frequency components adding water the reading from MP-917 became stable, indi-
travel at the same velocity according to Eqs. 2 and 3, there- cating that a stable water distribution had been reached.
fore, the waveform shape is maintained during propagation. After an additional equilibrium interval of several hours,
However, in a conductive medium different frequency com- about 200 sequential time delay readings were obtained and
ponents do not maintain the same phase relationships as the average calculated. The plastic sheet was then removed
they propagate in the medium according to Eqs. 2 and 3. to expose the sand to the air, and a fan was used to blow the
Different frequency components travel at different veloci- air across it to facilitate evaporation. Under these condi-
ties. This phenomenon is known as “dispersion”, causing tions, the moisture content of the sand decreased by approx-
distortion of the composite wave shape. The distortion caus- imately 0.03 m3 m–3 in 8 h.
es the reflected signal to have an increased rise time, which The cylinder was then re-covered and the automatic
has been shown to be a cause of increased time delay mea- acquisition and logging of the propagation time and weight
surement error (Hook and Livingston 1996). readings re-started. After an equilibrium interval of more
16 CANADIAN JOURNAL OF SOIL SCIENCE
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

Fig. 1. The set up of the experiment.


For personal use only.

Fig. 2. The relation between T/Tair and vol-


umetric soil water content when the electri-
cal conductivity of saturated soil extract is
equal to or less than 3.72 dS m–1.

Fig. 3. Volumetric soil water content measured


using balance (Y) vs. that measured using MP-917
(X) when the electrical conductivity of saturated
soil extract is equal to or less than 3.72 dS m–1.

than 24 h, about 200 data points were collected and aver- weight readings obtained from the balance. This procedure
aged. The volumetric soil water content was then calculated was repeated until the volumetric soil water content reached
using the appropriate equation (Hook et al. 1992) from the approximately 0.05 m3 m–3. The time to equilibrium
time delay readings obtained from the MP-917 and the increased as the soil became drier because of the decrease of
SUN ET AL. — ELECTRICAL CONDUCTIVITY AND TDR MEASUREMENT 17

Table 1. The overestimation of water content determined by TDR as soil water content and salinity increase
Electrical conductivity of saturated extract
6.37 dS m–1 9.27 dS m–1 11.73 dS m–1 13.73 dS m–1 15.29 dS m–1
0.040m3m–3 (0.004)z (0.007) (0.009) (0.007) (0.013)
0.082 m3m–3 (0.012) (0.017) (0.016) (0.017) (0.048)
0.120 m3m–3 NA NA NA (0.079) (0.115)
0.126 m3m–3 (0.021) (0.019) (0.032) NA NA
0.160 m3m–3 NA NA NA (0.127) (0.184)
0.165 m3m–3 (0.026) (0.057) (0.066) NA NA
0.200 m3m–3 NA NA NA (0.183) (0.214)
0.210 m3m–3 (0.024) (0.058) (0.067) NA NA
0.250 m3m–3 (0.029) (0.043) (0.071) (0.185) (0.334)
0.270 m3m–3 NA NA NA NA (0.396)
0.300 m3m–3 (0.084) (0.132) (0.148) (0.233) NA
zThe numbers in parentheses are the overestimations of volumetric water content in unit of m3m–3
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21
For personal use only.

Fig. 4. The relationship between


T/Tair vs. volumetric water content
when the electrical conductivity of
saturated soil extract (ECe) exceeds
3.72 dS m–1.

unsaturated soil hydraulic conductivity. For example, when water to the soil. In order to maintain a constant salt quanti-
the soil moisture level was below 0.10 m3 m–3, the time to ty in soil at different water contents, at the beginning of the
equilibrium gradually increased to more than 72 h. experiment approximate 2 g of NaCl was added to 1429 mL
In a sand, pore water hysteresis is greater than in most of distilled water, which would bring the volumetric soil
soils, and this may result in differing spatial moisture distri- water content to 0.30 m3 m–3 (close to its saturation) with
bution. In order to remove the effect of the hysteresis, the 1.54 dS m–1 electrical conductivity for saturated soil extract
above procedure was repeated but in the opposite direction, (ECe ). As soil dried the bulk soil electrical conductivity
from dry to wet. The same sand column was then oven dried decreased; however, the amount of salt in the soil remained
at 105°C for 48 h and the soil moisture was gradually unchanged as well as the electrical conductivity of soil-sat-
increased at increments of 0.05 m3 m–3 to 0.30 m3 m–3 by urated extract.
adding distilled water. At each moisture level, about 200 As the experiment progressed, the same soil sample had
propagation time readings were taken after a stable moisture additional salt added to produce higher ECe levels. For a
distribution had been reached. The difference in measured desired ECe for an increased salinity series, the total salt was
time delay for these two procedures was approximately 170 calculated, then an amount required in addition to that
ps, which translated into 0.01 to 0.02 m3 m–3 in soil mois- already present from the previous series was added.
ture. Normally, the propagation time measurement in the The whole database included eight ECe levels and soil
drying procedure was longer than that in the wetting one. moisture ranged from dry to 0.30 m3 m–3 at each ECe level.
The mean of time delay was used in analysis. The saturated extract was collected by first wetting 0.5 L
The different soil salinity was generated by adding differ- saline soil to saturation, the soil was then placed in a funnel
ent concentrations of salt solution (NaCl) instead of distilled and filtered by a quantitative 1 filter paper. A vacuum pump
18 CANADIAN JOURNAL OF SOIL SCIENCE

Table 2. The comparison of calculated time delay using Eq. 6 and measured one at 0.30 m3m–3 volumetric soil water content with different
electrical conductivity
Difference
Calculated time between
f MAX delay (ns) Measured measured and
σ (dS m–1) Rise time (ns) (MHz) using Eq. 9 time delay (ns) calculated
0.22 2.783 126.00 8.372 8.328 0.014
0.40 3.594 97.40 8.528 9.107 0.579
0.62 4.020 87.07 8.861 9.544 0.683
0.85 6.063 57.73 10.027 10.817 0.790
1.12 7.563 46.28 11.526 12.817 1.291

was used to facilitate the extraction. The electrical conduc- 3.72 dS m–1 could be further demonstrated by the following
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

tivities of the extract ECe were measured using a K320 calculation. Assuming that the volumetric soil water content
Microcomputer Conductometer (Consort pvba, Turnhout, θv is 0.30 m3 m–3, ECe was 1.54 dS m–1, then the calculated
Belgium). They were (at 0.30 m3 m–3 volumetric soil water bulk soil electrical conductivity σ using Archie’s formula
content): 0.34 dS m–1, 1.54 dS m–1, 3.72 dS m–1, 6.37 dS (Ferré at al.1998) would be 0.22 dS m–1. The measured rise
m–1, 9.27 dS m–1, 11.73 dS m–1, 13.73 dS m–1 and 15.29 dS time was 2.783 ns and the calculated fm according to Eq. 7
m–1 generated by adding an appropriate amount of salt. The is 126 MHz. This caused 14 ps longer in time delay (calcu-
time delay in the saline soil was measured using the same lated using Eq. 9) than that in nonsaline soil with the same
MP-917 and probe as mentioned before. water content. The difference was within the accuracy of
All the above experiments were conducted in ESI MP-917.
research laboratory at room temperature of 22 to 23°C.
2. For Soils with High Electrical Conductivity (ECe
RESULTS AND DISCUSSION ≥ 6.37 dS m–1)
For personal use only.

1. For Soils with Low Electrical Conductivity (ECe There was an overestimation of volumetric soil water con-
≤ 3.27 dS m–1) tent using TDR when the electrical conductivity of soil solu-
The results of time delay (T/Tair) and volumetric soil water tion at saturation was above 6.37 dS m–1 and the magnitude
content measurements for soil with low ECe (0.34 dS m–1, of overestimation increased as ECe increased at moisture
1.54 dS m–1, and 3.72 dS m–1) were pooled together, saturation (Table 1).
because the relationship between T/Tair and volumetric Two general patterns are shown in Table 1. First, at each
water content all appeared linear in this electrical conduc- ECe level, the overestimation increased with increase of soil
tivity range (Fig. 2). At next electrical conductivity level water content. Second, at each water content, the overesti-
where ECe = 6.37dS m–1, this relation obviously deviated mation increased with increase of electrical conductivity.
from linear. The slope 0.1195 was smaller than the pub- Both patterns could be explained by an increase in ionic
lished 0.1257 (Sun 1995; Hook and Livingston 1996).
conductivity in soil either by increasing the amount of con-
However, if the point at the upper right corner was exclud-
ductive carriers, and/or by increasing the facility of migra-
ed, then the equation became y = 0.1233x – 0.2030. The
tion of the ions as soil moisture increases. Both mechanisms
slope was very close to the published one. The intercept of
0.2030 was very close to 0.2090 from another experiment would increase σ, resulting in a longer time delay according
by Sun (Sun 1995). The point at the upper right corner rep- to Eq. 6. The overestimation was as large as 0.40 m3 m–3 at
resented the measurement at 0.30 m3 m–3 water content and 0.270 m3 m–3 moisture level (balance measured) when the
3.72 dS m–1 ECe conditions. TDR overestimated approxi- ECe reached 15.29 dS m–1. There were several data in
mately 0.05 m3 m–3 in water content at this condition. This Table1 out of the expected descent/ascent pattern. For
indicated that salt started affecting the measurement at rela- example, at 9.27 dS m–1 conductivity, the overestimation at
tively low concentration if the water content was high. 0.25 m3 m–3 moisture level was less that at 0.21 m3 m–3. At
In general the volumetric water content determined using 0.08 m3 m–3 moisture level, the overestimation at 11.73 dS
MP-917 agreed well with that measured using the balance m–1 conductivity was less than that at 9.27 dS m–1 level. The
(Fig. 3). If the upper right point was excluded, then the pub- cause of the inconsistency might be the results of uneven
lished linear calibration of time delay vs. volumetric water distribution of water and ions, and/or different soil com-
can be used for saline soil when ECe is no more than 3.72 pacting in different measurements since the same soil sam-
dS m–1. As soil dries, the concentration of the soil solution ple had been repacked at each salinity level. However, the
increased; however, the impedance (tortuosity) associated inconsistency was not detectable when the soil water con-
with current flow through the water phase increases as well, tent was above 0.08 m3 m–3 with ECe above 11.73 dS m–1.
and this offset the effect of the increase in ionic concentra- This indicates that at this moisture and salinity levels, the
tion on bulk soil electrical conductivity. overestimation in water content due to electrical conductiv-
The lack of influence of electrical conductivity on time ity overwhelmed the errors caused by other reasons. Clearly,
delay measurement using TDR when ECe was smaller than TDR cannot measure volumetric soil water content correct-
SUN ET AL. — ELECTRICAL CONDUCTIVITY AND TDR MEASUREMENT 19
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

Fig. 5. The relationship between the normal-


ized rise time (the ratio of rise in saline soil to
that for no saline dry soil), soil water content
and electrical conductivity.
For personal use only.

Fig. 6. The comparison of reflected


signal between ECe = 15.27 dS m–1
and 0.34 dS m–1 at 0.12 m3 m–3 water
content.

ly without a suitable calibration when the soil electrical con- increase in time delay (T/Tair) is no longer directly propor-
ductivity and moisture exceed a certain threshold. tional to the increase in water content. A large portion of the
When the ECe reached 6.37 dS m–1 or higher, the relation increase of time delay is attributed to the soil electrical con-
between T/Tair vs. θv deviated from linear, and the deviation ductivity. Figure 4 shows that at the same soil water content,
increased as the soil electrical conductivity elevated (Fig. 4). the time delay increases with the increase in soil electrical
As conductivity increased, time delay increased significant- conductivity. On the other hand, soils with different water
ly. For example, at 0.20 m3 m–3 volumetric water content, content may correspond to the same time delay when their
the measured time delay expressed as T/Tair increased from electrical conductivity is different.
3.510 when ECe = 6.37 dS m–1 to 4.905 when ECe = 15.29 It is interesting to see whether Eq. 6 is adequate to explain
dS m–1, and this would translate into approximately 0.18 m3 the increased time delay in highly conductive soils when
m–3 increase in displayed moisture reading. using TDR. Calculated and measured time delays for soil
The reason for a non-linear relation at high soil electrical with 0.30 m3 m–3 water content but different electrical con-
conductivity can be explained by the fact that the effect of ductivity are summarized in Table 2. The εr in Eq. 6 can be
addition of salt on time delay increases with increasing soil determined by using the relationship between volumetric
water as the ions become more mobile. In this situation, the water content θV and travel time T for non-saline soils, if θV
20 CANADIAN JOURNAL OF SOIL SCIENCE
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21
For personal use only.

Fig. 7. (a) The relationship between elec-


trical conductivity of saturated soil extract
and coefficient C1. (b). The relationship
between electrical conductivity of soil
solution and coefficient C2.

is known. The first column is the bulk electrical conductiv- TDR trace displayed using MP-917 soil moisture
ity σ calculated using Archie’s formula. The second column instrument) increased with the increase of soil moisture
is the rise time that can be measured using fine scan in and soil electrical conductivity. More research is need-
Moisture.Point MP-917. The third column is the maximum ed to explore the relationship between the rise time and
frequency calculated using Eq. 7. Table 2 shows that using measurement error.
Eq. 6 alone was not adequate to explain the increased time (b) In TDR, the transmitted signal comprises a band of fre-
delay in highly conductive soils. For example, when σ = quencies. According to Eq. 6, the time delay would be
1.12 dS m–1, there was still 1.291 ns increased time delay longer for low frequency components. In our calcula-
unexplainable by applying Eq. 6. The divergence between tion, only one frequency, the maximum frequency was
the calculated and measured time delay increased when soil used, and the effect of the low frequency components
electrical conductivity increased. The remaining overesti- was not considered.
mation might be explained by: (c) The relaxation related dielectric loss ε″ is neglected in
(a) The rise time of the reflection signal increases signifi- the above calculation. The calculated time delay would
cantly when soil electrical conductivity increases. The have been be longer if this term had been included.
slope of the reflection signal becomes so low that iden- Figure 6 shows the superimposed actual reflecting signals
tification of the portion of the waveform with the steep- from sandy soil with ECe of 0.34 dS m–1 and 15.27 dS m–1
est slope becomes very difficult. The commonly used at 0.12 m3 m–3 volumetric water content. It clearly shows
method to identify the time mark by intercepting the increases in rise time, signal attenuation and time delay due
steepest slope and time base line might be not valid to the raise of salinity, if the conventional steepest slope
when the rise time becomes very large. There may be an line method was applied. Even when the water content was
increase of rise time related measurement error as men- the same, the time delay (T/Tair) increased by 0.912; that
tioned before (Ferré et al. 1999). Figure 5 showed that translates into 0.115 m3 m–3 of overestimation of water
the rise time of the reflected signal (estimated from the content.
SUN ET AL. — ELECTRICAL CONDUCTIVITY AND TDR MEASUREMENT 21

3. The Logarithm Calibration ductivity in soil at one to fifty megahertz. Soil Sci. Soc. Am. J. 54:
The results indicate that using the linear relationship 332–341.
between T/Tair and θv plus theoretical correction using Eq. 6 Dalton, F. N., Herkelrath, D. S., Rawlins, D. S. and Rhoades, J.
is not adequate to explain all the increase in time delay in D. 1984. Time-domain reflectometry: Simultaneous measurement
of soil water content and electrical conductivity with a single
highly conductive media. Our experimental data suggest
probe. Science 224: 989–990.
that it is better to use a logarithm calibration (Fig. 5). All the Ferré, P. A., Redman, J. D., Rudolph, D. L. and Kachanoski,
calibrations shown in Fig. 5 were conducted on sandy soil R. G. 1998. The dependence of the electrical conductivity mea-
and the upper limit for ECe was 15.29 dS m–1. The general sured by time domain reflectometry on the water content of a sand.
form of the logarithm calibration is given as Water Resour. Res. 34: 1207–1213.
Ferré, P. A., Hook, W. R., Livingston, N. J. and Bassey, C.
θv = C1 ln(T/Tair) + C2 (8) 1999. Errors in TDR-determined water content in saline sand.
Pages 253–256 in Collection of papers presented at the Third
where coefficient C1 and C2 are functions of electrical con- Workshop on Electromagnetic Wave Interaction with Water and
ductivity of soil saturated extract. The dependence of C1 and Moist Substances. USA Department of Agriculture, Agriculture
C2 on ΕCe is shown in Fig. 7. As the conductivity increases, Research Service, Athens, GA.
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

Hasted, J. B. 1973. Aqueous dielectrics. Chapman and Hall,


C1 declines and C2 increases (less negative).
London, UK. p. 238.
Using Eq. 8 to obtain correct moisture readings in saline Heimovaara, T. J. and Bouten, W. 1990. A computer-controlled
soil requires the information of soil electrical conductivity. 3-channel time domain reflectometry system for monitoring soil
This can be achieved either by a direct measurement using water contents. Water Resour. Res. 26: 2311–2316.
salinity sensors or estimated from a TDR wave trace (Dalton Hilhorst, M. A. and Dirksen, C. 1994. Dielectric water content
et al. 1984; Topp et al. 1988). Nevertheless, Figs. 4 and 7 sensors: Time domain versus frequency domain. Pages 23–33 in
provide useful information that can be used to eliminate the Symposium and Workshop on Time Domain Reflectometry in
discrepancy between the actual soil moisture and that mea- Environmental, Infrastructure, and Mining Applications. United
sured using TDR in saline soil conditions. For the medium States Department of Interior Bureau of Mines, Northwestern
with conductivity above 15.29 dS m–1, the signal will be University, Evanston, IL.
greatly attenuated according to Eq. 1 and final reflection is Hook, W. R., Livingston, N. J., Sun, Z. J. and Hook, P. B. 1992.
For personal use only.

Remote diode shorting improves measurement soil water by time


barely detectable unless the TDR probe is coated with elec-
domain reflectometry. Soil Sci. Soc. Am. J. 56: 1383–1391.
trical insulation material with the expense of reduced instru- Hook, W. R. and Livingston, N. J. 1996. Errors in converting
ment sensitivity. It is necessary to conduct a separate time domain reflectometry measurements of propagation velocity
calibration for coated probe. to estimates of soil water content. Soil Sci. Soc. Am. J. 60: 35–41.
Malicki, M. A., Walczak, R. T., Koch, S. and Fluhler, H. 1994.
CONCLUSIONS Determining soil salinity from simultaneous readings of its electri-
The effect of soil electrical conductivity on volumetric soil cal conductivity and permittivity using TDR. Pages 328–336 in
water content determination using TDR has been studied. A Symposium and Workshop on Time Domain Reflectometry in
very large amount of data for soil water content, soil electri- Environmental, Infrastructure, and Mining Applications. United
cal conductivity and time delay for sandy soil have been col- States Department of Interior Bureau of Mines, Northwestern
lected, and the empirical logarithmic calibrations among University, Evanston, IL.
Nadler, A. and Frenkel, H. 1980. Determination of soil solution
these three have been established. The ordinary linear rela-
electrical conductivity from bulk soil electrical conductivity mea-
tion between T/Tair and soil water content θ is not valid as surements by the four-electrode method. Soil Sci. Soc. Am. J. 44:
the electrical conductivity of saturated soil extract (ECe) at 1216–1221.
saturation is above 6.37 dS m–1. The overestimation of soil Nielsen, D. C., Large, H. J. and Anderson, R. L. 1995. Time-
water content is attributed to the increase in soil electrical domain reflectometry measurements of surface soil water content.
conductivity and increase in rise time of the reflected signal, Soil Sci. Soc. Am. J. 59: 103–105.
which lead to a longer time delay. The correction of the soil Nyfors, E. and Vainikainen, P. 1989. Dielectric properties of
water content reading obtained using TDR in saline soils materials. Page 99 in Industrial microwave sensors. Artech House,
can be conducted by using the logarithmic calibration pre- Norwood, NY.
sented, but it requires an independent measurement of ECe Oliver, B. M. and Cage, J. M. 1971. Band width and rise time.
or σ at the same time. Page 372–373 in Electronic measurements and instrumentation.
MaGraw-Hill, New York, NY.
Ramo, S., Whinnery, J. R. and Van Duzer, T. 1994.
ACKNOWLEDGEMENTS Transmission line. Pages 218–219 in Fields and waves in com-
We thank Dr. Nigel Livingston, Biology Department, munucation electronics 3th ed. John Wiley & Sons Inc., New
University of Victoria, for permitting the use of equipment. York, NY.
This research was supported, in part, by grants from the Rao, N. N. 1987. Wave equation and solution for material medi-
Science Council of British Columbia. um. Pages 286–293 in Elements of engineering electromagnetics.
2nd ed. Prentice-Hall Inc., New Jersey.
Campbell, J. E. 1988. Dielectric properties of moist soils at RF Rhoades, J. D. and Van Schilfgaarde, J 1976. An electrical con-
and microwave frequency. Ph.D. dissertation. Dartmouth College, ductivity probe for determining soil salinity. Soil Sci. Soc. Am. J.
Hanover, NH. 40: 647–650.
Campbell, J. E. 1990. Dielectric properties and influence of con- Rhoades, J. D. 1984. Principles and methods of monitoring soil
22 CANADIAN JOURNAL OF SOIL SCIENCE

salinity. Pages 103–142 in I. Shainberg and J. Shalhevet, eds. Soil Topp,G. C. and Davis, J. L. 1985. Measurement of soil water
salinity and irrigation – processes and management. Vol. 5. content using TDR: A field evaluation. 1985. Soil Sci. Soc. Am. J.
Springer Verlag, Berlin, Germany. 49: 10–24.
Rhoades, J. D., Manteghi, N. A., Shouse, P. J. and Alves, W. J. Topp, G. C., Zegelin, S. J. and White, I . 1994. Monitoring soil
1989. Soil electrical conductivity and soil salinity: New formula- water content using TDR: An overview of progress. Pages 67–80
tions and calibrations. Soil Sci. Soc. Am. J. 53: 433–439. in Symposium and Workshop on Time Domain Reflectometry in
Rhoades, J. D. and Loveday, J. 1990. Salinity in Irrigation agri- Environmental, Infrastructure, and Mining Applications United
culture. Pages 1089–1142 in B. A. Stewart and D. R. Nielsen, eds. States Department of Interior Bureau of Mines, Northwestern
Irrigation of agricultural crops. SSSA, Inc., Madison, WI. University, Evanston, IL.
Shaun, F. K., Selker, J. S. and Green, J. L. 1995. Using short Whalley, W. R. 1993. Consideration on the use of time-domain-
soil moisture probes with high-bandwidth time domain reflectom- reflectometry (TDR) for measuring soil water content. J. Soil Sci.
etry instrument. Soil. Sci. Soc. Am. J. 59: 97–102. 44: 1–9.
Sun, Z. J. 1995. Water use by 10 white spruce crosses as deter- Wraith, J. M. and Baker, J. M. 1991. High-resolution measure-
mined by time domain reflectometry. Pages 35–55 in Stable carbon ment of root water uptake using automated tine-domain-reflectom-
isotopes as indicators of increased water use efficiency and bio- etry. Soil. Sci. Soc. Am. J. 55: 928–932.
mass production in white spruce (Picea glauca (Moench) voss) Zegelin, S. J. and White, I. 1994. Calibration of TDR for appli-
Can. J. Soil. Sci. Downloaded from cdnsciencepub.com by 197.25.197.6 on 02/15/21

seedlings grown under different water and nitrogen regimes. Ph.D. cations in mining, grains, and fruit storage and handing. Pages
Dissertation, Biology Department, University of Victoria, Victoria, 115–129 in Symposium and Workshop on Time Domain
BC. Reflectometry in Environmental, Infrastructure, and Mining
Topp, G. C., Davis, J. L. and Annan, A. P. 1980. Applications. United States Department of Interior Bureau of
Electromagnetic determination of soil water content: Mines, Northwestern University, Evanston, IL.
Measurements in coaxial transmission lines. Water Resour. Res.,
16: 574–582.
Topp, G. C., Davis, J. L. and Anna, A. P. 1982. Electromagnetic
determination of soil water content using TDR: I. Applications to
wetting fronts and steep gradients. Soil Sci. Soc. Am. J. 46:
672–684.
For personal use only.

Вам также может понравиться