Ecological Engineering 117 (2018) 1–17

Contents lists available at ScienceDirect

Ecological Engineering
journal homepage: www.elsevier.com/locate/ecoleng

Modelling soil suction changes due to mixed species planting T

a b,c b,⁎
J.J. Ni , A.K. Leung , C.W.W. Ng
a
Department of Civil and Environmental Engineering, The Hong Kong University of Science and Technology, Hong Kong Special Administrative Region
b
Civil and Environmental Engineering, The Hong Kong University of Science and Technology, Hong Kong Special Administrative Region
c
Division of Civil Engineering, School of Science and Engineering, University of Dundee, Dundee, Scotland, UK

A R T I C LE I N FO A B S T R A C T

Keywords: Vegetation-induced changes of soil hydrology (i.e., infiltration rate and soil matric suction) has important im-
Soil bioengineering plication to the stability of earthen infrastructure. Predicting the plant hydrological effects is challenging when
Mixed plantations multiple species are present. Most existing models consider only single-species due to lack of reliable and re-
Suction levant experimental data and difficulty to partition solar radiant energy into individual species in a mixed-
Infiltration
species condition. This paper aims to develop an improved hydrological model that can determine the water
infiltration rate and suction responses under both single- and mixed-species (one tree and one grass species)
condition. The model is verified by double-ring infiltration tests and field monitoring conducted in bare, grass-
only, tree-only and mixed tree-grass plots. The model was subsequently used for parametric analyses. The mixed
species plot had the lowest infiltration rate and preserved the greatest suctions because the tree-grass compe-
tition for water created high soil moisture deficit. Parametric study identifies a threshold ratio of tree and grass
leaf area index of 2.0, beyond which the effects of grass root-water uptake on water competition with the
adjacent tree were negligible. Induced suction in the mixed tree-grass plot was close to that in a tree-only plot.
Contribution of tree root-water uptake to induced suction in a mixed species plot did not increase further when
the leaf area index ratio exceeds 5.0.

1. Introduction extra slope stability that can be gained through transpiration-induced

Soil hydrology changes by evapotranspiration (ET ), a combined
process that removes soil moisture from a given area, and during a
specified period of time, by evaporation from the bare soil surface and
transpiration from plants (Soil Science Society of America, 2008). This
natural process reduces soil moisture or increases matric suction
(Rahardjo et al., 2014; Garg et al., 2015; Smethurst et al., 2015; Leung
et al., 2015a; Ng et al., 2016a). It is well-known that an increase in
matric suction would not only increase soil shear strength but also re-
duce soil hydraulic conductivity (Alonso et al., 2010; Springman et al.,
2013: Gadi et al., 2017). The ET -induced changes in soil suction and
consequently the soil behaviour have important implication to the
water balance and soil stability of some earthen infrastructure such as
cuttings and embankments (Briggs et al., 2013; Smethurst et al., 2015;
Das et al., 2017; Garg et al., 2017) as well as landfill covers (Feng et al.,
2017), where vegetation is normally found.
Indeed, field studies reported by Rahardjo et al. (2014), Simon and
Collison (2002) and Ni et al. (2017) showed that suction could be
preserved after rainfall, providing substantial stabilisation to soil
(Boldrin et al., 2017a). Recent centrifuge model tests have quantified
hydrological reinforcement (Ng et al., 2016b; Leung et al., 2017), in
addition to mechanical root reinforcement (Fan and Su, 2008; Boldrin
et al., in press; Liang et al., in press). Rock et al. (2012) also showed the
importance of taking ET into account in water balance calculation to
more correctly assess the amount of water percolation in landfill covers.
Mixed plant species of different functional groups (e.g., tree and
grass) are often found in the field, instead of one. Ecological engineers
who adopt soil bioengineering technique would want to, on one hand,
enhance the stability of the infrastructure and on the other hand to
grow multiple species for maximising ecological restoration effects and
biodiversity to the surrounding built environment (Lamb, 1998; Hooper
et al., 2005). In a mixed tree-grass condition, for examples, trees could
facilitate grass growth due to increased nutrient availability (Ludwig
et al., 2004). From engineering perspective, grass has shallow root
system that can reduce soil erosion effectively, while tree has stronger
and deeper root system that can provide tensile strength as mechanical
reinforcement.
Most existing studies measured suction induced by grass-only soil
(e.g., Ng et al., 2013; Rahardjo et al., 2014) or tree-only soil (e.g.,
Smethurst et al., 2015; Ng et al., 2016a). There were a few exceptions

⁎
Corresponding author.
E-mail address: cecwwng@ust.hk (C.W.W. Ng).

https://doi.org/10.1016/j.ecoleng.2018.02.023
Received 2 August 2017; Received in revised form 30 January 2018; Accepted 24 February 2018
0925-8574/ © 2018 Elsevier B.V. All rights reserved.
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

List of notations ρa air density at constant temperature

G soil heat flux
ag extinction coefficient for radiant energy intercepted by h water pressure head
grass foliage Θ degree of saturation
at extinction coefficient for radiant energy intercepted by AE actual evaporation
tree canopy AEV air entry value
hfc water pressure head corresponding to the field capacity ET evapotranspiration
h wp water pressure head corresponding to the permanent LAI leaf area index
wilting point PE potential evaporation
Rg radiant energy intercepted at the grass foliage PET potential evapotranspiration
Rs radiant energy intercepted at the bare soil surface PT potential transpiration
Rt radiant energy intercepted at the tree canopy R incoming solar radiation energy
Rv root volume ratio RH relative humidity
Vr total volume of roots S sink term
Vs soil volume SWRC soil water retention curve
cp specific heat of moist air a fitting parameter for Eq. (13)
ea actual vapor pressure b fitting parameter for Eq. (13)
es saturated vapor pressure d root depth
ks saturated hydraulic conductivity e void ratio
m1 fitting parameter for Eq. (11) g distribution of root volume ratio along depth
m2 fitting parameter for Eq. (11) k soil hydraulic conductivity function as a function of
m3 fitting parameter for Eq. (11) pressure head
m4 fitting parameter for Eq. (11) s soil matric suction
ra aerodynamic resistance for water vapor transfer from ca- t time
nopy to air u wind speed
rs stomatal resistance z vertical co-ordinate
γw unit weight of water Δ rate of change of saturated vapor pressure with time
εg Albedo of grass foliage α transpiration reduction function
εs Albedo of ground surface γ psychrometric constant
εt Albedo of tree canopy θ volumetric water content
θr residual volumetric water content λ latent heat of water vaporization
θs saturated volumetric water content

such as February and Higgins (2010) and Ni et al. (2017) who focused
on the hydrological responses of mixed tree-grass soil. In particular, Ni
et al. (2017) found that tree spacing controls the amount of ET -induced
suction. To-date, it is unclear from these limited field data that (i) how
much each individual species contribute to the overall suction induced
in a mixed-species soil, and (ii) in what circumstance(s) the contribu-
tion from one species might dominate the other. These are two im-
portant aspects that assist ecological engineers to apply the soil
bioengineering technique in terms of the selection of plant species and
long-term vegetation management.
Many models (e.g., Feddes et al., 1976, 2001; Lai and Katul, 2000;
Fatahi et al., 2010; Moene and van Dam, 2014) have been developed to
incorporate a sink term into a water balance equation (e.g., Darcy-Ri-
chards equation) to capture root-water uptake. Although these models
appear to be quite successful in estimating the ET -induced changes in
suction of a single plant species (Allen et al., 1998; Feddes et al., 2001),
further examination is needed for mixed species case. In this more
complicated case, intra- or inter-species competition for sunlight (or
solar radiation) and soil water would occur.
Some studies have been conducted to model the partition of soil
evaporation and plant transpiration for a given ET (Ritchie, 1972;
Mahat and Tarboton, 2012; Kool et al., 2016). Among these, Ritchie
(1972), which satisfies the Beer-Lambert’s law, has been widely used.
Beer-Lambert’s law is a concept from the spectroscopy that is used to
quantify the transmittance and attenuation of solar radiation in the
atmosphere. Ritchie (1972) considered plant canopy/foliage as a
“layer”, through which the solar radiation could penetrate and at-
tenuate. The degree of attenuation by a “layer” depends on the plant
leaf area index (LAI ; ratio of the total green leaf area of a plant species
to the projected canopy area on plan). While the energy attenuated at
the “layer” would be used for plant transpiration, the remaining energy
fallen on the soil surface would be used for soil evaporation. However,
this kind of model considers a relatively simple single-plant system,
where the single-“layer” assumption (Massman, 1992) would not be
applicable to multiple species cases.
This study aims to develop and verify a new soil hydrological model
that can be used to investigate the effects of mixed tree-grass planta-
tions on (i) the partition of solar radiation and (ii) inter-species com-
petition for soil water and associated soil suction change.
2. Soil hydrological model for mixed plantation
The model was derived based on radiant energy partition, water
balance and their coupling in a soil-plant system defined in Fig. 1(a).
The proposed model structure is shown in Fig. 1(b). The system in-
cludes a bare ground (i.e., without vegetation) or a vegetated ground
with single or multiple plant species.
2.1. Partition of solar radiant energy
A major source of energy for ET is the incoming solar radiant en-
ergy, R . The model applies the Beer–Lambert’s law to capture the
partition of R in the above-ground soil-plant system (Fig. 1). For a bare
ground, the radiant energy that is not reflected (due to albedo) would
be absorbed by the soil for evaporation mainly. For a grass- or tree-only
ground, on the other hand, a portion of the energy would be intercepted
by the tree canopy (or grass foliage), leaving the rest fallen on the
ground surface. For mixed species ground with one tree and one grass
species, energy partition would occur first at the tree canopy, followed
by the grass foliage and finally the ground surface. Hence, the energy
partition equations in a mixed tree-grass case can be expressed as:
Rt = R·(1−εt )·[1−exp (−at ·LAIt )] (1)

2
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

PET R

PTt Rt Tree canopy

PTg Rg Grass foliage

PE Rs Ground surface
(a)

Soil-plant-atmosphere interaction and modelling

Atmosphere Penman-Monteith equation PET

to obtain PET (Equation 4)
Interaction Energy partition equation Radiant
(Equations 1 – 3) energy
Penman equation to obtain PE
PE (Equations 5 – 6)
Properties Plant sink term
Plant (Equations 8 and 9) PT = PET – PE

Plant effects
Interaction (Equation 10) Water balance equation
(Equation 7)

Properties Soil hydraulic properties

Soil Model output:
(Equations 11 – 13) Soil pore-water pressure/suction
(b)
Fig. 1. (a) Simplified diagram idealizing the energy partition in a mixed tree-grass plot. Legend: PE is potential evaporation; PET is potential evapotranspiration; PTt
is potential transpiration by tree; PTg is potential transpiration by grass; R is incoming solar radiant energy; Rt is radiant energy intercepted by tree canopy; Rg is
radiant energy intercepted by grass foliage; Rs is radiant energy absorbed by the bare ground surface. (b) Structure of the modelling of soil-plant-atmosphere
interaction.

Rg = (R−Rt )·(1−εg )·[1−exp(−a g ·LAIg )] (2) canopy, grass canopy and ground surface, respectively. at and a g are the
extinction coefficient for radiant energy intercepted by the tree and
Rs = (R−Rt −Rg )·(1−εb) (3) grass, respectively; and LAIt and LAIg are the LAI of tree and grass,
respectively. Allen et al. (1998) reported that the range of albedo (i.e.,
where Rt , Rg , Rs are the radiant energy intercepted at the tree canopy, the ratio of reflected to incoming energy) for a fully vegetated cover
grass foliage and bare soil surface, respectively; εt , εg , εs are the albedo with green leaves is 0.20–0.25. Hence, both εt and εg was taken to be an
(i.e., the ratio of reflected to incoming radiant energy) of the tree average value of 0.23. An average εs of 0.12 (Taha et al., 1988) was

3
J.J. Ni et al.
considered for the ground surface. Note that Eqs. (1)–(3) ignored (i) the
energy used to heat up the air due to low specific heat capacity of air
and (ii) the energy absorbed by plant chlorophyll for the use of pho-
tosynthesis (i.e., generally less than 2%; Salisbury and Ross, 1992).
By knowing the amount of radiant energy intercepted via Eqs.
(1)–(3), the potential transpiration (PT ; maximum amount of water
that plants could extract water from the soil; Feddes et al., 1978) of
each plant may be determined by the modified Penman-Monteith
equation (Ozier-lafontaine et al., 1998), as follows:
Δ(Ri−G ) + ρa cp
(es − ea)
rai
λ·PTi =
Δ+γ 1+ ( rsi
rai ) (4)
where i is equal to g or t, if referring to PT of the grass and tree, re-
spectively; λ is the latent heat of water vaporization, Δ is the rate of
change of saturated vapor pressure with time, which is a function of
temperature, Ri is the intercepted radiant energy by layer i (obtained
from Eqs. (1) and (2)), G is the soil latent heat flux, ρa is air density at
constant temperature, cp is the specific heat of moist air, es is the sa-
turated vapor pressure, which is a function of temperature; ea is the
actual vapor pressure, γ is the psychrometric constant; rai is the aero-
dynamic resistance for water vapor transfer from canopy to the air for
each component; rsi is the stomatal resistance of each component.
After knowing Rs from Eq. (3), the potential evaporation (PE ;
maximum amount of water leaves the soil surface as vapor under un-
limited water supply condition; Penman, 1948) of bare soil surface can
be estimated using the Penman equation (Penman, 1948):
ΔR s
+ γEa
PE = λ where e0 is the initial void ratio before root permeation (i.e., bare soil);
Δ+γ
where Ea .
u Θ(s ) =
Ea = 0.165∗ (es−ea) ⎛0.8 + ⎞ θs−θr
⎝ 100 ⎠
where u is the wind speed. Hence, the potential evapotranspiration
(PET ; maximum amount of ET under unlimited water supply condition;
Rosenberg, 1974) of a vegetated cover could be estimated by the sum of
PTi and PE .
2.2. Soil water balance
Considering one-dimensional (1-D) transient seepage in unsaturated
soil planted with mixed grass-tree species, the conservation of water
mass may be described by the following modified Darcy-Richards
equation:
dθ (h) d ⎡ dh
= k (h) ⎛ + 1⎞ ⎤−Sg (h,z )−St (h,z )
dt ⎢
dz ⎣ ⎝ dz ⎠⎦⎥
where θ is the volumetric water content (VWC); t is the elapsed time; z
is the vertical co-ordinate; h is the water pressure head; k (h) is the soil
hydraulic conductivity function as a function of h or matric suction
(s = −hγw , where γw is the unit weight of water); and Si (h,z ) is the sink
term, which represents the volume of water transpired by a plant i over
the entire root zone for a given time interval (Feddes et al., 1976).
Mathematically, Si (h) may be expressed as (Varado et al., 2006):
α (h,z )·gi (z )
Si (h,z ) = ·PTi
∫di α (h,z )·gi (z )·dz
where di is the root depth of plant i ; gi is considered to be the dis-
tribution of root volume ratio Rv (the total volume of roots (Vr ) per unit
volume of soil (Vs )) of plant i ; PTi is calculated from Eq. (4); and α (h,z ) is
the so-called transpiration reduction function (Feddes et al., 2001),
which can be expressed as:
Ecological Engineering 117 (2018) 1–17
⎧ 0 h ⩽ h wp
⎪ h − hwp
α (h,z ) = h − h h wp ⩽ h ⩽ hfc
⎨ fc wp
⎪ 1 h ⩾ hfc
⎩ (9)
where h wp and hfc are the water pressure head corresponding to the
permanent wilting point and field capacity, respectively. Physically,
Eqs. (8) and (9) mean that for any h higher than hfc , the ability of root-
water uptake is maximum. As the soil dries and h reaches between hfc
and h wp , the ability of root-water uptake decreases linearly with a de-
crease in h . Plant wilts and stops extracting water when soil is too dry to
have reached a h lower than h wp .
In all cases, actual evaporation ( AE ) from the bare soil surface is
considered to be equal to PE obtained from Eq. (5). It is deemed a
reasonable assumption because the amount of suction investigated in
this study is not higher than 3000 kPa, which is the threshold value
beyond which AE drops below PE (Wilson et al., 1997).
2.3. Hydraulic properties of unsaturated rooted soil
The two soil hydraulic properties required to solve the water bal-
ance Eq. (7) are soil water retention curve (SWRC; θ (s ) ) and hydraulic
conductivity functions k (s ) . Ng et al. (2016c) proposed a SWRC model
that could consider the effects of root occupancy on the changes in void
ratio, e , and hence SWRC of rooted coarse-grained soils. The following
equation is used to model the root-induced changes in e :
e0−Rv (1 + e0)
e=
1 + Rv (1 + e0) (10)
(5) Rv is the root volume ratio. Eq. (10) is then coupled with the void ratio-
dependent SWRC equation proposed by Gallipoli et al. (2003):
m −m1
θ (s )−θr se m4 ⎞ 2 ⎤
= ⎡1 + ⎛⎜ ⎟
⎢ ⎝ m3 ⎠ ⎥ (11)
(6) ⎣ ⎦
where θs is the saturated volumetric water content; θr is the residual
volumetric water content; m1, m2 , m3 and m4 are the model parameters.
m1 and m2 control the shape of SWRC, while m3 and m4 relate to the air-
entry value ( AEV ) of the soil.
By knowing a θ (s ) , k (s ) of the same soil may be predicted by the
equation proposed by van Genuchten (1980), as follows:
1 m1 2
⎣ (
k (s ) = ks·Θ(s)0.5 ⎡1− 1−Θ(s) m1 ) ⎤
⎦ (12)
where ks is the saturated hydraulic conductivity, which is a function of
e . Eq. (13) considers the e -dependency on ks of coarse-grained soil ac-
cording to (Yin, 2009):
ks = a·exp(b·e ) (13)
(7) where a and b are the fitting parameters.
Eqs. (1)–(13) were solved simultaneously by an author-developed
script under Matlab programming environment. An implicit finite dif-
ference method (Celia et al., 1990) was used to solve the Darcy-Ri-
chards equation in Equation (7). For each time step, the Picard iteration
technique was used.
3. Materials and methods
3.1. Field testing and monitoring
(8)
To evaluate the model, a comprehensive field testing and mon-
itoring programme was implemented at the Hong Kong University of
Science and Technology (HKUST) Eco-Park. HKUST Eco-Park (latitude
of 114°15′ and longitude of 22°15′) is located at the southeast of Hong
Kong and near the Victoria Harbour. There is a 2 m-high compacted
embankment, and this field study was conducted at the top flat surface
4
J.J. Ni et al.
of the embankment. The groundwater table was located at 4.5 m depth
from the top of the embankment. The field monitoring lasted for
30 days from 12th Dec 2014 to 12th Jan 2015. Fig. 2 shows some cli-
mate data of the site measured by an automatic weather station in-
cluding a rain gauge manufactured by Davis Instruments Ltd. (model
Vantage Pro 2). During this monitoring period, the site temperature was
fairly constant, fluctuating slightly about an average value of 15.5 °C
Fig. 2. Climate data during the monitoring period from 12th Dec 2014 to 12th Jan 2015: (a) temperature; (b) air relative humidity; (c) solar radiation; (d) wind
speed; and (e) rainfall intensity.
5
Ecological Engineering 117 (2018) 1–17
(Fig. 2(a)). During the first 10 days, the relative humidity fluctuated
greatly between 25% and 95%, after which an average level of 72% was
maintained (Fig. 2(b)). On the other hand, solar radiation (Fig. 2(c))
varied between 50 and 450 W/m2, from which the low values corre-
sponded to the cloudy days when rainfall occurred (Fig. 2(e)). The
maximum rainfall amount received during the monitoring period was
67 mm between 8th Jan 2015 to 9th Jan 2015. The wind speed
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

Table 1 these plots. All plants were grown for three months at the site before the
Summary of parameters used for calculating energy partition. start of the field test.
Parameter Value A vertical array of jet-fill tensiometer was installed at 100, 250 and
400 mm depths to measure the spatial and temporal variations of matric
Extinction coefficient for net radiation ar 0.75 for tree and 0.5 for suction at the middle of each plot. After installation, the gap between
grass
−2 −1
the wall of each tensiometer and the surrounding soil was backfilled
Solar radiation, R (MJ m d ) Refer to Fig. 2(c)
Net solar radiation, Rn (MJ m−2 d−1) 0.77R
with cement to prevent from any preferential flow.
Temperature, T (°C) Refer to Fig. 2(a)
Slope of vapor pressure curve, Δ (kPa °C−1) Δ(T) , following Allen et al. 3.2. Soil and plants
(1998)
Radiant Energy intercepted by each component, Eqs. (1)–(3)
The soil tested in this study was completely decomposed granite
Ri (MJ m−2 d−1)
Soil heat flux density, G (J m−2 d−1) Assumed negligible (CDG), which is commonly found in south China (including Hong
Latent heat of vaporization, λ (MJ kg−1) 2.45 Kong), parts of the Southeast Asia like South Korea. Based on sieve
Psychrometric constant, γ (kPa °C−1) 0.0674 analysis, the average contents of gravel, sand, silt and clay of the CDG
Relative humidity, ea/ es Refer to Fig. 2(b) were 9.5%, 83.1%, 5.1% and 2.3%, respectively. According to the
Saturated vapor pressure, es (kPa) es (T ) , following Allen et al.
Unified Soil classification System (USCS; ASTM, 2010), CDG is classi-
(1998)
Actual vapor pressure ea (kPa) e fied as well-graded sand with silt (SW-SM). The plastic limit and liquid
es (T ) ∗ ( a )
es limit were 26 and 44%, respectively. Based on the standard Proctor
Wind speed, u (m/s) Refer to Fig. 2(d)
tests, the maximum dry density and optimum water content (by mass)
Air density, ρa (kg m−3) 0.001195
Specific heat of moist air, cp (kJ kg−1 m−3) 1.013 were 1870 kg/m3 and 13%, respectively.
Stomatal conductance, rs (s m−1) 67 for tree and 83 for grass Schefflera heptaphylla (a tree species; also known as Ivy tree) and
Aerodynamic conductance, ra (s m−1) 180/u for tree and 208/u for Cynodon dactylon (a grass species; also known as Bermuda grass) were
grass selected for testing. They were selected due to their commonness in
many parts of Asia and their ability of drought tolerance (Hau and
Corlett, 2003). These species have been used for ecological restoration
fluctuated between 0.5 and 2.5 m/s (Fig. 2(d)). of man-made slopes in subtropical regions, such as Hong Kong and
In total, 12 test plots with 1.5 × 1.5 m2 plan area each were pre- some parts of the Southeast Asia (Hau and Corlett, 2003).
pared for field testing and monitoring. Four treatments were con- At the end of the field test, root traits, including root depth and root
sidered, namely bare plot, grass-only, tree-only and mixed tree-grass volume, were measured, following the procedures described by
plot. Three replication plots were tested for each treatment. To avoid Reubens (2010). Soil block samples that contained root samples were
any interference between neighbouring plots, the minimum distance excavated from the site and then brought to the laboratory to wash
between two plots was 2 m. Grass was first germinated from the nursery using a hydroneumatic-elutriation based apparatus (Smucker et al.,
and then transplanted to the site. Tree individuals (purchased from 1982). Each root zone was divided into several depth ranges, in which
Tung Kee Garden Horticulture Ltd, Hong Kong) were transplanted to the average root volume was determined using image-based analyses
each tree-only and mixed-species plots with a spacing of 180 mm (i.e., (Himmelbauer et al., 2004). The root volume ratio, Rv , was then ob-
1.5 times of the maximum lateral spread of tree roots). Hence, for the tained by normalising the root volume by the soil volume of that depth
given tree spacing, 65 tree individuals were transplanted to each of range.

Fig. 3. Distributions of root volume ratio along depth. Data represents mean value obtained from three replicated pots, while the error bars represent standard error
of mean (n = 3).

6
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

Table 2
Summary of tree and grass traits (mean ± standard error of mean, n = 3).
Plant trait Mixed-species plot Tree-only plot Grass-only plot

Tree Grass Tree Grass

Before transplantation Shoot height: mm 604 ± 24 24 ± 11 601 ± 35 24 ± 6

Leaf area index 0.70 ± 0.07 0.36 ± 0.06 0.73 ± 0.11 0.35 ± 0.05
Root depth: mm 113 ± 11 33 ± 10 116 ± 15 32 ± 6

After three months of growth (i.e., start of monitoring) Shoot height: mm 711 ± 26 108 ± 11 713 ± 25 112 ± 11
Leaf area index 2.4 ± 0.06 1.2 ± 0.08 2.31 ± 0.11 1.30 ± 0.12

After four months of growth (i.e., end of monitoring) Shoot height: mm 745 ± 32 128 ± 9 742 ± 28 137 ± 12
Leaf area index 3.0 ± 0.10 1.6 ± 0.11 3.2 ± 0.11 1.8 ± 0.12
Root depth: mm 188 ± 11 96 ± 10 195 ± 11 110 ± 9

1001 made. As outer ring contributes to lateral water flow, the inner ring is
90
0.9 shielded from lateral water flow and hence contributes mostly to
downward flow. Field experiments conducted by Chowdary et al.
80
0.8 (2006) show that the presence of the outer ring could reduce the lateral
0.7
70 component of infiltration by three to six times when compared with the
Degree of saturation (%)

case without the outer ring. The test setup was similar to that described
0.6
60
by the ASTM Standard D3385 (ASTM, 2009). The only difference was
0.5
50 that grass or/and trees were transplanted inside both the inner and
0.4
40 outer rings.
Measured (bare CDG) Each infiltration test was commenced by applying a constant water
30
0.3
Fitted (bare CDG) head of 100 mm inside both the inner and outer rings using a Mariotte
0.2
20 Predicted (Rv=0.025 m3/m3) bottle. During testing, the variations of suction were recorded, while
10
0.1 Predicted (Rv= 0.045 m3/m3) the amount of water infiltrated to the ground was recorded by the
Mariotte bottle. Each test was completed when all the measurements
00
reached a steady-state condition, which took less than two hours.
0.1 1 10 100
After the infiltration tests, all 12 plots were left to expose to the
Suction (kPa)
climate for a month. Response of suction was continuously recorded
Fig. 4. Soil water retention curves of CDG with and without plant roots. during the entire monitoring period.
Legend: Rv is root volume ratio.
3.4. Statistical analysis
1.E-05
Statistical analysis was performed using Microsoft Excel. Significant
Saturated hydraulic conductivity (m/s)

Bare CDG differences were assessed with one way-ANOVA, followed by post hoc
Rooted CDG Fisher’s least significant difference test. Results were considered sta-
tistically significant when p-value ≤0.05.

3.5. Numerical back-analysis

1.E-06

1.E-07
0.4 0.45
Void ratio
Fig. 5. Relationship between saturated hydraulic conductivity and void ratio of
bare and rooted CDG (data from Leung et al. 2015b).
3.3. Test procedures
Before field monitoring, infiltration characteristics of all 12 plots
with four different vegetation cover conditions were determined
through double-ring infiltration tests. Each test was performed at the
middle of the plot. Each setup consisted of an inner ring with a diameter
of 300 mm and an outer ring with a diameter of 600 mm, which were
inserted into the ground by 75 and 150 mm depths, respectively. The
height of the outer and inner rings was 425 mm and 350 mm, respec-
tively. The outer ring (also known as buffer ring) is used to lessen the
lateral water movement from the inner ring, in which measurement is
y = 6E-10e14.8x Two series of analyses were carried out to simulate both the (i)
R2 = 0.91 infiltration tests and (ii) one-month field monitoring. In all analyses, a
1-D CDG soil column with 4.5 m depth was considered. When modelling
vegetated cases (either grass-only, tree-only or mixed-species), two soil
domains were created, one modelling soil in root zone (depending on
the rooting length) and the other for bare soil below the root zone.
0.5 0.55 0.6 The first series of analyses (Series 1) is concerned on the double-ring
infiltration tests. These analyses aim to validate the soil water balance
model (Eq. (7)) as well as the effects of root-induced change in soil
hydraulic properties (Eqs. (10)–(13)). Since all infiltration tests con-
ducted were complete within two hours, the total amount of PT during
this relatively short period of testing is expected to be minimal. Indeed,
the maximum PT experienced in this field study was no more than
1.5 mm/day (shown later). This means that the potential amount of
water being transpired by plants would be less than 0.125 mm, which is
practically negligible compared to the amount of water infiltrated to the
soil. Hence, energy partition (i.e., Eqs. (1)–(6)) and plant root-water
uptake (i.e., Eqs. (8) and (9)) during these tests were not considered. In
each numerical model, a constant water head of 100 mm was specified
at the top boundary for two hours. A static water table was applied at
the bottom of the model to mimic the field condition. The initial suction
distribution followed the measurements made by the three tensiometers
before the start of each test. For soil within the root zone, root-induced

7
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

Fig. 6. Variation of infiltration rate with time for the

four different vegetation cover conditions. Legend: ks
is saturated hydraulic conductivity. Letters in the
legend, B, T, G and M refer to the bare, grass-only,
tree-only and mixed tree-grass plots. Data represents
mean value obtained from three replicated pots,
while the error bars represent standard error of mean
(n = 3).

Table 3
Infiltration rates and suction at the end of each infiltration test (mean ± standard error of mean, n = 3). Letters indicate significant differences as tested using one-
way ANOVA followed by post hoc Fisher’s least significant difference test. N.R. denotes not relevant.
Parameter Depth (mm) Bare Grass-only Tree-only Mixed p-value

−6
Infiltration rate (10 m/s) N.R. 1.51 ± 0.12a 1.36 ± 0.18ab 1.05 ± 0.22b 0.37 ± 0.16c 0.001
Matric suction (kPa) 250 7.63 ± 5.43a 21.66 ± 1.93b 22.67 ± 3.21b 28.56 ± 2.86c 0.021

boundary that represents rainfall infiltration during the monitoring

period was specified at the top boundary, following the recorded hourly
rainfall (Fig. 2(e)). At each time step, when the rainfall intensity is
lower than the ks of the CDG, the flux boundary applied would be re-
mained. On the contrary, when the rainfall intensity was higher, the
flux boundary would be switched to a pressure boundary with a pres-
sure head of 1 mm. This numerical treatment means that any rainwater
that could not be infiltrated into the soil would exist in form of a small
water ponding head less than 1 mm. The bottom boundary condition
was identical to Series 1.

3.6. Parameterization

The solar radiant energy (R ) measured during the monitoring period

Fig. 7. Suction distributions before and after ponding. Letters in the legend, B,
T, G and M refer to the bare, grass-only, tree-only and mixed tree-grass plots.
Data represents mean value obtained from three replicated pots, while the error
bars represent standard error of mean (n = 3).
changes in both SWRC and k (s ) were considered, whereas the hydraulic
properties of the bare soil were specified for the soil below the root
zone.
Series 2 simulated the one-month field monitoring. In this series, all
energy partition and water balance equations (i.e., all Eqs. (1)–(13))
were used. The climate data shown in Fig. 2 was used to determine PE
and PTi for each vegetated case, and these terms were applied to the top
boundary of the model for calculation. On the other hand, a flux
(Fig. 2(c)) was input into Eqs. (1) to (3) to calculate the energy partition
to the tree canopy (Rt ), grass foliage (Rg ) and the ground surface (Rs ).
The extinction coefficient at of the tree is taken to be 0.75, following the
calibration performed by Garg (2015). The extinction coefficient of the
grass a g , on the other hand, was calibrated to be 0.50 by best-fitting the
field measured suction in the grass-only plot. Other parameters re-
quired for calculating energy partition are summarised in Table 1.
The parameters, m1, m2 , m3 and m4 , for the SWRC model (Eq. (11))
were calibrated based on the field measured SWRC of CDG. For the
known distribution of Rv (Fig. 3), SWRC of rooted CDG at any depth
within a tree or grass root zone could be predicted by Eqs. (10) and
(11). Note that for mixed grass-tree case, an equivalent Rv was obtained
by summing up Rv of tree and grass at a given depth. The parameters, a
and b , for the k (s ) model (Eqs. (12) and (13)) were calibrated using the
field data reported by Leung et al. (2015b). They measured ks of the
same type of CDG vegetated with the same tree species (Schefflera
heptaphylla) as in this study.

8
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

0 120 240 360 480 600 720

Time (hour)
(a)

0 120 240 360 480 600 720

Time (hour)
(b)

0 120 240 360 480 600 720

Time (hour)
(c)

0 120 240 360 480 600 720

th
th
(12 Dec 2014) Time (hour) (12 Jan 2015)
(d)
Fig. 8. Partition of solar radiation in (a) the bare plot; (b) the grass-only plot; (c) the tree-only plot and (d) the mixed tree-grass plot.

3.7. Parametric study dn = dg / dt ; representing the below-ground growth condition). For a

One important engineering question in relation to the soil bioengi-
neering application is how the relative growth of grass and tree in
mixed-covered ground would affect the partition of solar radiant energy
and consequently the magnitude and distribution of suction for a given
soil and climate condition. The model, after field verification, was used
to carry out an extensive parametric analysis to answer this question.
Two growth indices are defined, namely (i) ratio of LAI of tree and
grass (i.e., LAIn = LAIt / LAIg ; representing the above-ground growth
condition) and (ii) ratio of rooting depth of grass and tree (i.e.,
given growth condition, the relative effects of ks and PET (i.e., ks / PET )
on suction response were investigated. Hence, two series of steady-state
analyse were carried out to study the effects of (i) above-ground and (ii)
below-ground growth conditions. In the first series, LAIn varied from 0
to 10, while keeping dn to be a constant at 0.2. The second series varied
dn from 0.2 to 1.0, whereas LAIn was fixed at 2.5. Both series of analyses
considered two ks of 1 × 10−6 and 1 × 10−8 m/s, which represents
sand-like and clay-like soil, respectively. Identical geometry and
boundary conditions to that used in the validation analyses were used.
All analyses considered the same (i) climate condition, with solar

9
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

Fig. 9. Calculated PE , PT and PET for (a) the bare plot; (b) the grass-only plot; (c) the tree-only plot and (d) the mixed tree- grass plot.

10
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

100
Suction (kPa) B100_Measured
80 B250_Measured
60 B400_Measured
B100_Simulated
40
B250_Simulated
20 B400_Simulated
0
0 80 160 240 320 400 480 560 640 720
Time (hour)
(a)
100
G100_Measured
Suction (kPa)

80
G250_Measured
60 G400_Measured
40 G100_Simulated
G250_Simulated
20
G400_Simulated
0
0 80 160 240 320 400 480 560 640 720
Time (hour)
(b)
100
T100_Measured
Suction (kPa)

80 T250_Measured
T400_Measured
60
T100_Simulated
40 T250_Simulated
T400_Simulated
20
0
0 80 160 240 320 400 480 560 640 720
Time (hour)
(c)
100
M100_measured
Suction (kPa)

80
M250_Measured
60 M400_Measured
40 M100_simulated
M250_Simulated
20 M400_Simulated
0
0 80 160 240 320 400 480 560 640 720
th Time (hour)
(12 Dec 2014) (12th Jan 2015)

(d)
Fig. 10. Variations of suction for (a) the bare plot; (b) the grass-only plot; (c) the tree-only plot and (d) the mixed grass-tree plot. In the legend, letters, B, T, G and M
refer to the bare, grass-only, tree-only and mixed tree-grass plots. Numbers, 100, 250 and 400 refer to the instrument depths. Data represents mean value obtained
from three replicated pots, while the error bars represent standard error of mean (n = 3).

radiation of 1300 W/m2, temperature of 23 °C, relative humidity of
40% and wind speed of 1.6 m/s; (ii) dt of 1 m; (iii) parabolic distribu-
tion of Rv for trees and triangular for grasses (Fig. 3); (iv) transpiration
reduction function (Eq. (9)) and (iv) SWRC of the CDG.
4. Results and discussion
4.1. Plant traits
Fig. 3 shows that in both the tree-only and mixed-species plots, trees
have a parabolic distribution of Rv , peaked at the mid-depth of the root

11
J.J. Ni et al. Ecological Engineering 117 (2018) 1–17

100 species in the growing seasons (such as spring). Once their roots are
established, the tree species could be transplanted afterwards. Other
90
above- and below-ground traits of the grass and trees are summarised in
80 Table 2.
Simulated suction (kPa)

70
60
50
40
30
20
10
0 (2016c). Roots could have occupied some soil pore space and this re-
0 10 20 30 40 50 60 70
Measured suction (kPa)
Fig. 11. Comparison between measured and simulated matric suction at all
instrument depths. In the legend, letters, B, T, G and M refer to the bare, grass-
only, tree-only and mixed tree-grass plots. Numbers, 100, 250 and 400 refer to
the instrument depths. Data represents mean value obtained from three re-
plicated pots, while the error bars represent standard error of mean (n = 3).
zone. The Rv of grass roots distribute almost linearly along depth and
peak near the soil surface. The grass Rv near the soil surface in the
mixed-species plot is 20% smaller than that in the grass-only plot. Inter-
species competition for soil water and nutrient could reduce the pro-
duction of root biomass of each species (Rozados-Lerenzo et al., 2007).
For tree-grass mixed planting, specifically, the shading of sunlight due
to the overlapping of the canopies among neighboring trees would re-
duce grass root growth underneath (Scholes and Archer, 1997; Ludwig
et al., 2004). Hence, it might be advisable to first transplant grass
B100
4.2. Soil hydraulic properties
B250
B400 Fig. 4 shows the field measured SWRC of the bare soil at e0 of 0.52.
G100
G250 After least-square best-fitting procedures, m1, m2 , m3 and m4 are 0.11,
G400 2.5, 0.3 kPa and 3.64, respectively. Two examples of predicted SWRCs
T100 of rooted CDG at Rv of 0.025 and 0.045 m3/m3 (i.e., corresponding to
T250
T400 the two peak Rv shown in Fig. 3). The presence of roots cause an in-
M100 crease in air-entry value, whereas the desorption rate is apparently
M250
M400 unchanged. This agrees with the conceptual model proposed by Scanlan
1:1 line and Hinz (2010) as well as the theoretical model derived by Ng et al.
80 90 100 duces the size of the soil pore throat. Following the capillary law, the
reduction of pore diameter in turn increases the air-entry value. The
root diameter investigated in this study ranges between 0.15 and 2 mm,
implying that the presence of roots affects predominantly low suction
range. It is thus not surprising to see the minimal effects of root on the
desorption rate in relatively high suction range. The higher the Rv , the
greater the increase in air-entry value it is.
Fig. 5 depicts a log-linear correlation between the ks of CDG and e .
This appears to be consistent with previous in-situ test results reported
by Rahardjo et al. (2014) and Jotisankasa and Sirirattanachat (in press).
Based on the ks – e correlation, a and b were calibrated to be 6 x 10−10
m/s and 14.8, respectively, for Eq. (13). Hence, k (s ) at different depths
within a root zone can be estimated by Eq. (12).
4.3. Infiltration tests
Fig. 6 compares the measured and simulated variations of infiltra-
tion rate with time during two-hour constant-head ponding. As

Table 4
Suction (kPa) during one-month monitoring (mean ± standard error of mean, n = 3). Letters indicate significant differences as tested using one-way ANOVA
followed by post hoc Fisher’s least significant difference test.
Time (h) Depth (mm) Bare Grass-only Tree-only Mixed p-value

Selected drying periods

104 100 13.98 ± 2.06a 49.67 ± 5.31b 84.33 ± 6.55c 84 ± 6.16c < 0.001
250 22.33 ± 2.87a 56.21 ± 6.13 cd 50.66 ± 4.92b 63.51 ± 5.35d < 0.001
400 24.46 ± 3.30a 30.69 ± 3.68a 32.25 ± 4.50a 33.01 ± 5.31a 0.252

194 100 20.94 ± 2.05a 27.43 ± 3.68a 56.78 ± 4.49b 67.67 ± 6.94c < 0.001
250 8.51 ± 2.87a 22.87 ± 3.68b 44.21 ± 4.49c 41.58 ± 4.92c < 0.001
400 8.12 ± 2.05a 14.34 ± 2.87ab 19.41 ± 3.68b 27.82 ± 4.11c 0.002

374 100 18.86 ± 2.45a 45.27 ± 4.92b 64.52 ± 5.73c 78.33 ± 6.13d < 0.001
250 10.92 ± 2.45a 13.88 ± 2.46a 30.67 ± 2.87b 42.06 ± 3.30c < 0.001
400 5.00 ± 1.63a 5.45 ± 2.05a 18.98 ± 2.86b 23.42 ± 3.26b < 0.001

544 100 27.25 ± 2.87a 47.37 ± 3.68b 64.31 ± 4.49c 78.33 ± 5.31d < 0.001
250 16.70 ± 2.87a 22.24 ± 2.84a 44.83 ± 4.11b 50.94 ± 4.50b < 0.001
400 5.91 ± 1.63a 20.90 ± 3.30b 21.98 ± 3.68b 32.13 ± 4.08c < 0.001

Selected wetting periods

144 100 2.04 ± 0.82a 12.23 ± 2.05b 12.31 ± 2.87b 16.33 ± 2.78c 0.002
250 5.72 ± 1.63a 9.78 ± 2.87ab 15.17 ± 2.85b 16.23 ± 3.68b 0.020
400 13.06 ± 2.05a 15.18 ± 3.30a 16.43 ± 2.58a 28.72 ± 4.11b 0.005

294 100 1.89 ± 0.61a 6.37 ± 2.05ab 8.66 ± 2.66bc 12.59 ± 2.87c 0.010
250 6.01 ± 0.81a 7.02 ± 3.26ab 12.65 ± 3.31bc 13.58 ± 2.05c 0.040
400 3.69 ± 0.79a 13.95 ± 3.29b 20.26 ± 2.78b 15.41 ± 2.88b 0.002

484 100 5.86 ± 2.06a 13.33 ± 2.11b 26.05 ± 3.26c 27.02 ± 3.66c < 0.001
250 10.47 ± 2.06a 21.04 ± 2.85b 25.07 ± 4.49b 39.07 ± 3.68c < 0.001
400 10.58 ± 2.03a 17.06 ± 2.82ab 20.19 ± 2.86b 28.32 ± 4.49c 0.003

704 100 1.93 ± 0.65a 4.47 ± 2.03ab 8.01 ± 1.63bc 9.47 ± 1.62c 0.006
250 0.60 ± 0.10a 5.21 ± 1.63b 11.32 ± 1.60c 10.85 ± 2.45c 0.001
400 7.56 ± 2.45a 6.81 ± 1.25a 14.24 ± 2.45b 18.89 ± 2.05b 0.002

12
J.J. Ni et al.
Fig. 12. Computed variations of the changes in soil water storage (with respect to time on day one of monitoring) for the four different vegetation cover conditions.
Fig. 13. Effects of LAIn on PTi , PE and PET .
expected, the rate of infiltration in all bare and vegetated cases reduced
exponentially with time, due to the continuous reduction of hydraulic
gradient as the soil got wetter. The steady-state infiltration rate in all
cases appears to approach the ks of the bare CDG. The simulation results
have close agreement with the field measurement, especially for the
infiltration rate after 20 min of ponding. The infiltration rate of tree-
only and mixed tree-grass plots is statistically significantly difference
and lower than of the grass-plot and bare plots (p-value = 0.001;
Table 3). The lower infiltration rate in these vegetated plots may be
because of the reduction in k (s ) due to the presence of plant roots.
Among the three vegetated cases, the mixed grass-tree case has the
highest Rv (Fig. 3) and this might have reduced k (s ) (according to Eq.
(10)) and hence infiltration rate (Fig. 6) most significantly. Previous
laboratory tests (Ng et al., 2016a) and in-situ infiltration tests (Huat
et al., 2006; Leung et al., 2015b) also showed the reduction of water
infiltration rate in vegetated plots, especially when the plants are at the
actively growing stage.
Fig. 7 compares the measured and simulated distributions of suction
along depth before and after each infiltration test. Before testing, the
initial suction distributions in different cover conditions were reason-
ably close to each other. The comparison shows that the suction profiles
of both the bare and vegetated cases were well captured by the simu-
lation. While there is no difference in suction between all four plots at
both 100 and 400 mm depths, the suction preserved at 250 mm in all
vegetated cases was higher than in the bare case (p-value = 0.021;
Table 3). This is consistent with the fact shown in Fig. 6 that the ve-
getated cases have lower infiltration rate. Among the three vegetated
cover conditions, the mixed grass-tree case preserved highest suctions
13
Ecological Engineering 117 (2018) 1–17
because of the lowest infiltration rate. The amount of suction preserved
between the grass- and tree-covered plots has no statistical difference.
The comparisons made in both Figs. 6 and 7 show that the model
derived based on soil water balance through Eqs. (7)–(13), is able to
reproduce the field measurements in terms of both the water infiltration
rate and suction, when root-induced changes in soil hydraulic proper-
ties were taken into account.
4.4. Field monitoring
Fig. 8 shows the results of energy partition calculation in terms of
solar radiation intercepted in each cover condition using the climate
data (Fig. 2). For the grass-only plot, approximately 70% of the solar
radiation was intercepted by the grass foliage, while for the tree-only
plot, the percentage of interception was as high as 80% due to the
higher tree LAI . For the mixed grass-tree plot, more than 90% of the
radiant energy was intercepted mainly by the trees and then the grass
underneath. Only less than 10% of energy was received by the bare
surface for evaporation to take place.
Based on the results of energy partition calculation, the PE of the
bare soil surface, the PT of each vegetation component and the PET for
each ground cover condition are calculated (Fig. 9). As expected, PET
of vegetated ground, regardless of single- or mixed-species case, was
always higher than the PE of the bare plot. Moreover, PTt was higher
than PTg because the trees had a greater above-ground biomass (i.e.,
quantified by LAI and height) for intercepting more solar radiant en-
ergy for transpiration. In fact, more than 65% and 75% of PET for the
grass-only and tree-only plots was contributed by PT , respectively. For
the mixed tree-grass plot, PTt dominants the PET , whereas PTg and PE
contributed only negligibly. During rainfall events, there were abrupt
drops of PTi due to high relative humidity (RH ) and cloudy conditions
(i.e., low solar radiation). This climate condition suppressed plant
transpiration (Sidle et al., 1985; Snyder et al., 2003).
Fig. 10 compares the measured and simulated suction variations for
the four different vegetation cover conditions. In general, the simula-
tions have close matches with the measurements for all three instru-
ment depths at 100, 250 and 400 mm. Note that for the tree-only and
mixed plots, measurements of suction higher than 90 kPa were not
possible due to water cavitation of the tensiometers, but the simulations
showed that suction at 100 mm depth could possibly higher than
100 kPa. The close agreement between the field measurements and
numerical simulation is further evidenced in Fig. 11 when all measured
and simulated suction data are compared. The discrepancy is no more
than 25%. The comparison suggests that the model which is derived
based on energy partition and water balance is able to capture the ef-
fects of mixed tree-grass condition on soil suction responses reasonably
well, given the presence of in-situ soil heterogeneity and natural
variability of plants.
During drying periods (i.e., at elapsed times of 104, 194, 374 and
544 h in Fig. 10), the maximum suction induced in shallow depths of
J.J. Ni et al.
Fig. 14. Effects of LAIn on normalized suction in a mixed tree-grass plot for (a) clay-like soil (ks = 1 × 10−8 m/s); and (b) sand-like soil (ks = 1 × 10−6 m/s).
tree-only and mixed tree-grass plots was always higher than in grass-
only and bare plots (p-value < 0.001; Table 4). Except at time 104 h,
suction induced in the mixed tree-grass plot was higher than in the tree-
only ground. At deeper depth (400 mm), the difference of suction be-
tween the four types of plots was little. On the contrary, during wetting
periods (i.e., at elapsed times of 144, 294, 484 and 704 h in Fig. 10), the
minimum suction preserved in the mixed tree-grass plots was always
higher than in the bare plot (p-value < 0.005; Table 4). However, only
a few occasions that the suction preserved in the mixed species plots
was higher than the other vegetated cases are found. When rainfall
amount was less than 8 mm, the minimum suction preserved in the
mixed tree-grass plots was higher than in other vegetated cases by
6–32 kPa. In most of the time suction in the tree-only plot has no sta-
tistical difference with that by the grass-only plot.
It is interesting to see that although trees in the mixed species plot
intercepted most of the radiant energy (Fig. 8), the remaining small
amount of energy intercepted by grass was not negligible as it could
induce suction considerably higher than that in the tree-only plot. In
order to explain this, the calculated change in soil water storage (with
respect to time when the monitoring started) for each vegetation cover
condition is compared in Fig. 12. Note that negative change means net
soil moisture loss to the atmosphere, while positive value means net
storage of soil water. For the bare plot, after the first 120 h of mon-
itoring, the net soil water change is always positive. This means that for
14
Ecological Engineering 117 (2018) 1–17
the entire monitoring period, evaporation was inadequate to release the
soil moisture gained by rainfall infiltration back to the atmosphere. For
the vegetated plots, on the contrary, more soil moisture was lost via
plant transpiration, in addition to evaporation from the bare soil sur-
face. The effects of plant transpiration are more prominent for tree-only
and mixed tree-grass plots, as the soil water storage remained negative
for most of the monitoring period. The presence of plants creates higher
soil moisture deficit, inducing higher suction than the bare plot
(Fig. 10).
5. Further insights to the mixed species effects on soil suction
5.1. Effects of the growth of plant leaves
Fig. 13 shows the competition of grass and tree for energy inter-
ception in a mixed-species plot due to the growth of above-ground plant
biomass. For a given climate condition, an increase in LAIn causes a
significant increase in PTt and a simultaneous decrease in PTg and PE .
When LAIn increases further to values higher than 5.0, all components
have no major changes. The tree dominates the energy interception in
the system. The impact of the relative growth of tree and grass LAI on
steady-state suction, for a given dn of 0.2, is depicted in Fig. 14(a) and
(b). Note that the suction in the y-axis is expressed in two dimensionless
forms. The first one is to normalise suction induced in mixed-species
J.J. Ni et al.
Fig. 15. Effects of dn on normalized suction in a mixed tree-grass plot for (a) clay-like soil (ks = 1 × 10−8 m/s); and (b) sand-like soil (ks = 1 × 10−6 m/s).
plot (sm ) by that induced in tree-only plot (st ) (i.e., sm/ st ). Hence any
change of this ratio reflects the contribution by grass in the mixed-
species plot. By the same token, the second dimensionless suction (i.e.,
sm/ sg ) refers to the contribution of tree in the mixed-species plot. In
clay-like soil which the ks is relatively low (i.e., 1 × 10−8 m/s;
Fig. 14(a)), sm/ sg increases tremendously, especially in shallow depths,
as LAIn increases up to 5.0. This is because of the increasing con-
tribution of tree root-water uptake to sm when the tree with higher LAIt
intercepted more radiant energy. Relatively speaking, sm/ sg at the depth
below the root zone (i.e., 1.5 m) increases less, but this increase is not
negligible. Indeed, test results reported by Ng et al. (2016a) showed
that Schefflera heptaphylla could induce substantial amount of suction
even at depths 3.5 times deeper than the rooting depth. On the con-
trary, the contribution of grass root-water uptake to sm is minimal, as
the change in sm/ st are little.
When the same species were planted in sand-like soil with relatively
high ks of 1 × 10−6 m/s (Fig. 14(b)), the magnitude of suction induced
is much smaller. When the soil is more permeable, soil water supply is
greater than the plant water demand. The water flow is fast enough to
replenish the water lost due to plant transpiration. This reduces the
hydraulic gradient between the root zone (where root-water uptake
occurs) and the surrounding soil, hence reducing tree-grass competition
for water and so the amount of suction induced.
Based on the simulation results, some implications that are relevant
to the application of soil bioengineering technique can be obtained.
15
Ecological Engineering 117 (2018) 1–17
Firstly, there appears a threshold LAIn of 2.0, beyond which the con-
tribution of grass root-water uptake to sm is negligible. When young
trees are transplanted and their canopy has not yet developed at the
initial stage (i.e., LAIn smaller than 2), the contribution of grass root-
water uptake to the overall suction induced in the mixed-species plot,
especially in shallower depths, should not be neglected, regardless of
soil texture considered. On the contrary, when tree canopy is more
established (i.e., LAIn larger than 2), the effects of grass root-water
uptake on soil hydrological change could be reasonably ignored. In fact,
results from both Fig. 14(a) and (b) show that sm/ st approaches 1.0 for
sufficiently large LAIn . This means that sm may be conveniently ana-
lysed by using a tree-only plot, which is a much simpler system to
analyse, without the need to consider species-species competition for
water. The threshold LAIn of 2.0 appears to be independent on the soil
hydraulic conductivity.
Secondly, when the tree is more developed and has a LAIt that is
sufficiently higher than LAIg , the contribution of tree root-water uptake
to sm does not increase further. Apparently, the value of this LAIn re-
duces from 5.0 (for the case of clay-like soil) to approximately 2.0 (for
the case of sand-like soil). This suggests that the relative contribution of
grass and tree root-water uptake to the overall suction induced in the
mixed-species plot depends not only on their relative growth but also
on the soil hydraulic properties. This implies that if tree pruning is
required (e.g., due to excessive tree growth and blocking of sight and
road signs on transport embankments), arborists might be advised to
J.J. Ni et al.
maintain the LAIn of the mixed tree-grass plot to be no less than 5.0 for
maximising sm .
5.2. Effects of rooting depth
Fig. 15 shows the effects of relative root length of grass and tree,
dn = dg / dt , on suction, for LAIn of 2.5. Regardless of the soil texture
considered, an increase in dn causes a drop in sm/ sg at all depths because
of (i) the reduction of the contribution of tree root-water uptake and (ii)
increased tree-grass competition for water, as grass roots grow deeper.
The reduction of sm/ sg in shallower depths is more significant as it is the
region where most tree root biomass exists following the parabolic
shape of Rv considered. Although the contribution of tree root-water
uptake to sm reduces with the increase in dn , it is not negligible even the
rooting depth of the grass is as long as that of the tree (i.e., dn = 1). sm in
the top 1 m could be two times of that induced by the grass-only plot.
An implication to soil bioengineering application is that to maximise
the amount of sm in a mixed tree-grass plot, deep-rooted grass species
might be avoided to minimise water competition with the adjacent tree.
Indeed, the maximum root length of Cynodon dactylon is normally no
more than 0.5 m (Skerman and Riveros, 1990). This particular grass
species appears to be a suitable candidate to grow and “work” with
Schefflera heptaphylla for not reducing sm due to species-species com-
petition for water. In fact, if such combination of grass and tree species
could be found to form a mixed plot, it is not unreasonable to estimate
sm through the analysis of a tree-only plot. The result depicted in Fig. 15
shows that sm/ st is apparently independent on dn and it is close to 1.0
for both types of soil texture.
6. Conclusions
This study develops an improved hydrological model that could
quantify (i) intra- or inter-species competition for solar radiation and
soil water and (ii) their effects on soil suction changes. In addition to
plant root-water uptake, the hydrological model also captures the ef-
fects of root-induced changes in soil hydraulic properties on soil suction
changes. Good agreements between field measurements and simula-
tions were obtained, given the presence of in-situ soil heterogeneity and
natural plant variability.
Vegetated plots, regardless of the type and the number of species
present, have lower water infiltration rate than the bare plot. During
most of the drying periods, the mixed-species plot induced higher
suction than single-species and bare plots in shallow depths. During the
wetting periods, in only a few occasions (when rainfall amount was less
than 8 mm), the minimum suction preserved in the mixed tree-grass
plots was higher than in other vegetated cases by 6–32 kPa. With the
aid of the numerical model, it was mainly the tree in the mixed-species
plot to intercept more than 75% of solar radiation for root-water up-
take, whereas grass root-water uptake and surface evaporation played
little role.
There exists a threshold LAIn (i.e., ratio between tree LAI and grass
LAI ) of 2.0, beyond which the contribution of grass root-water-uptake
to steady-state suction induced in the mixed tree-grass plot can be ne-
glected. When the tree in the mixed-species plot is not well-developed
yet (i.e., LAIn < 2.0 and comparable tree root length with the grass’s),
the effects of grass root-water uptake on the overall suction induced is
not negligible. When LAIn is higher than the threshold value, a mixed
tree-grass plot may be simplified as a tree-only plot. In this case, ig-
noring the effects of grass root-water uptake on water competition with
tree would have negligibly small influence on the overall amount of
suction induced in the mixed tree-grass plot. When the LAIn is suffi-
ciently large (> 5.0), the contribution of tree root-water uptake in the
mixed species plot to the suction induced does not increase further.
In this study, only one tree and one grass species were considered in
the mixed-species condition. Although potentially the proposed energy
partition equations can be extended for multiple species, this could not
16
Ecological Engineering 117 (2018) 1–17
be verified yet until more high-quality field data is available. Moreover,
the model considers only the effects of root growth on the changes of
soil hydraulic properties. In order to apply the model for longer-term
vegetation management, effects of root decay, especially on the soil
permeability change, need to be considered in the future.
Acknowledgements
We acknowledge research grant T22-603/15N provided by the
Research Grants Council of the Government of Hong Kong SAR. The
second author acknowledges the EU Marie Curie Career Integration
Grant under the project ‘BioEPIC slope’ and research travel support
from the Northern Research Partnership (NRP).
References
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration. Guidelines
for computing crop water requirements. Irrigation and Drainage Paper 56. FAO,
Rome, Italy.
Alonso, E.E., Pereira, J.M., Vaunat, J., Olivella, S., 2010. A microstructurally based ef-
fective stress for unsaturated soils. Géotechnique 60 (12), 913–925.
ASTM, 2009. Standard Test Method for Infiltration Rate of Soils in Field using Double-
Ring Infiltrometer. ASTM standard D3385. American Society for Testing and
Materials: West Conshohocken, Pa.
ASTM, 2010. Standard practice for classification of soils for engineering purposes (Unified
Soil Classification System). ASTM standard D2487. American Society for Testing and
Materials, West Conshohocken, Pa.
Boldrin, D., Leung, A.K., Bengough, A.G., 2017a. Correlating hydromechanical properties
of vegetated soil with plant functional traits. Plant Soil. http://dx.doi.org/10.1007/
s11104-017-3211-3.
Boldrin, D., Leung, A.K., Bengough, A.G., in press. Root biomechanical properties during
establishment of woody perennials. Ecological Engineering.
Briggs, K.M., Smerthurst, J.A., Powrie, W., O’Brien, A.S., Butcher, D.J.E., 2013. Managing
the extent of tree removal from railway earthwork slopes. Ecol. Eng. 61P, 690–696.
Celia, M.A., Bouloutas, E.T., Zarba, R.L., 1990. A general mass-conservative numerical
solution for the unsaturated flow equation. Water Resour. Res. 26 (7), 1483–1496.
Chowdary, V.M., Rao, M.D., Jaiswal, C.S., 2006. Study of infiltration process under dif-
ferent experiment conditions. Agric. Water Manage. 83, 69–78.
Das, G., Hazra, B., Garg, A., Ng, C.W.W., Lateh, H., Nazi, A., 2017. Bivariate probabilistic
modelling of hydro-mechanical properties of vegetated soils. Adv. Civ. Eng. Mater. 6
(1), 1–23.
Fan, C.C., Su, C.F., 2008. Role of roots in the shear strength of root-reinforced soils with
high moisture content. Ecol. Eng. 33 (2), 157–166.
Fatahi, B., Khabbaz, H., Indraratna, B., 2010. Bioengineering ground improvement con-
sidering root water uptake model. Ecol. Eng. 36 (2), 222–229.
February, E.C., Higgins, S.I., 2010. The distribution of tree and grass roots in savanna in
relation to soil nitrogen and water. S. Afr. J. Bot. 76 (3), 517–523.
Feddes, R.A., Kowalik, P., Kolinska-Malinka, K., Zaradny, H., 1976. Simulation of field
water uptake by plants using a soil water dependent root extraction function. J.
Hydrol. 31 (1), 13–26.
Feddes, R.A., Kowalik, P.J., Zaradny, H., 1978. Simulation of field water use and crop
yield, Centre for Agricultural Publishing and Documentation.
Feddes, R.A., Hoff, H., Bruen, M., Dawson, T., de Rosnay, P., Dirmeyer, P., Pitman, A.J.,
2001. Modeling root water uptake in hydrological and climate models. Bull. Am.
Meteorol. Soc. 82 (12), 2797–2809.
Feng, S., Leung, A.K., Ng, C.W.W., Liu, H.W., 2017. Theoretical analysis of coupled effects
of microbe and root architecture on methane oxidation in vegetated landfill covers.
Sci. Total Environ. 599–600, 1954–1964.
Gallipoli, D., Wheeler, S.J., Karstunen, M., 2003. Modelling the variation of degree of
saturation in a deformable unsaturated soil. Géotechnique 53 (1), 105–112.
Gadi, V.K., Tang, Y.R., Das, A., Monga, C., Garg, A., Berretta, C., Sahoo, L., 2017. Spatial
and temporal variation of hydraulic conductivity and vegetation growth in green
infrastructures using infiltrometer and visual technique. Catena 155, 20–29.
Garg, A., Leung, A.K., Ng, C.W.W., 2015. Transpiration reduction and root distribution
functions for a non-crop species Schefflera heptaphylla. Catena 135, 78–82.
Garg, A., 2015. Effects of Vegetation Types and Characteristics on Induced Soil Suction.
(PhD Thesis). The Hong Kong University of Science and Technology, Hong Kong.
Garg, A., Hazra, B., Ng, C.W.W., 2017. Stochastic hydro-mechanical stability of vegetated
slopes: an integrated copula based framework. Catena 160, 124–133.
Hau, B.C.H., Corlett, R.T., 2003. Factors affecting the early survival and growth of native
tree seedlings planted on a degraded hillside grassland in Hong Kong, China.
Restoration Ecol. 11 (4), 483–488.
Himmelbauer, M.L., Loiskandl, W., Kastanek, F., 2004. Estimating length, average dia-
meter and surface area of roots using two different image analysis systems. Plant Soil
260 (1), 111–120.
Huat, B.B.K., Ali, F.H.J., Low, T.H., 2006. Water infiltration characteristics of unsaturated
soil slope and its effect on suction and stability. Geotech. Geol. Eng. 24 (5),
1293–1306.
Hooper, D.U., Chapin, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavorel, S., Lawton, J.H.,
Lodge, D.M., Loreau, M., Naeem, S., Schmid, B., Setälä, H., Symstad, A.J.,
Vandermeer, J., Wardle, D.A., 2005. Effects of biodiversity on ecosystem functioning:
J.J. Ni et al.
a consensus of current knowledge. Ecol. Monogr. 75, 3–35.
Jotisankasa, A., Sirirattanachat, T., in press. Effects of grass roots on soil-water retention
curve and permeability function. Can. Geotech. J.
Kool, D., Kustsa, W.P., Ben-Gal, A., Lazarovitch, N., Heitman, J.L., Sauer, T.J., Agam, N.,
2016. Energy and evapotranspiration partitioning in a desert vineyard. Agric. For.
Meteorol. 218–219, 227–287.
Lai, C.T., Katul, G., 2000. The dynamic role of root water uptake in coupling potential to
actual transpiration. Adv. Water Resour. 23, 427–439.
Lamb, D., 1998. Large-scale ecological restoration of degraded tropical forest lands: the
potential role of timber plantations. Restor. Ecol. 6, 271–279.
Leung, A.K., Garg, A., Ng, C.W.W., 2015a. Effects of plant roots on soil-water retention
and induced suction in vegetated soil. Eng. Geol. 193, 183–197.
Leung, A.K., Garg, A., Coo, J.L., Ng, C.W.W., Hau, B.C.H., 2015b. Effects of the roots of
Cynodon dactylon and Shefflera heptaphylla on water infiltration rate and soil hy-
draulic conductivity. Hydrol. Process. 29 (15), 3342–3354.
Leung, A.K., Kamchoom, V., Ng, C.W.W., 2017. Influences of root-induced soil suction
and root geometry on slope stability: a centrifuge study. Can. Geotech. J. 54 (3),
291–303.
Liang, T., Bengough, A.G., Knappett, J.A., Muir Wood, D., Loades, K.W., Hallett, P.D.,
Boldrin, D., Leung, A. K., Meijer, G.J., in press. Scaling of the reinforcement of soil
slopes by living plants in a geotechnical centrifuge. Ecol. Eng.
Ludwig, F., de Kroon, H., Berendse, F., Prins, H.H.T., 2004. The influence of savanna trees
on nutrient, water and light availability and the understorey vegetation. Plant Ecol.
170 (1), 93–105.
Mahat, V., Tarboton, D.G., 2012. Canopy radiation transmission for an energy balance
snowmelt model. Water Resour. Res. 48, W01534.
Massman, W.J., 1992. A surface energy balance method for partitioning evapo-
transpiration data into plant and soil components for a surface with partial canopy
cover. Water Resour. Res. 28 (6), 1723–1732.
Moene, A.F., van Dam, J.C., 2014. Transport in the Atmosphere-vegetation-soil
Continuum. Cambridge University Press.
Ng, C.W.W., Ni, J.J., Leung, A.K., Zhou, C., Wang, Z.J., 2016a. Effects of planting density
on tree growth and induced soil suction. Géotechnique 66 (9), 711–724.
Ng, C.W.W., Leung, A.K., Yu, R., Kamchoom, V., 2016b. Hydrological effects of live poles
on transient seepage in an unsaturated soil slope: centrifuge and numerical study. J.
Geotech. Geoenviron. Eng 04016106 1.
Ng, C.W.W., Ni, J.J., Leung, A.K., Wang, Z.J., 2016c. A new and simple water retention
model for root-permeated soils. Géotech. Lett. 6 (1), 106–111.
Ng, C.W.W., Woon, K.X., Leung, A.K., Chu, L.M., 2013. Experimental investigation of
induced suction distribution in a grass-covered soil. Ecol. Eng. 52, 219–223.
Ni, J.J., Leung, A.K., Ng, C.W.W., 2017. Investigation of plant growth and transpiration-
induced suction under mixed grass-tree conditions. Can. Geotech. J. 54 (4), 561–573.
Ozier-Lafontaine, H., Lafolie, F., Bruckler, L., Tournebize, R., Mollier, A., 1998. Modelling
competition for water in intercrops: theory and comparison with field experiments.
Plant Soil 204, 183–201.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. Proc. R.
Soc. London 193, 120–146.
Rahardjo, H., Satyanaga, A., Leong, E.C., Santoso, V.A., Ng, Y.S., 2014. Performance of an
instrumented slope covered with shrubs and deep-rooted grass. Soils Found. 54 (3),
17
Ecological Engineering 117 (2018) 1–17
417–425.
Reubens, B., 2010. Woody vegetation for gully rehabilitation in northern Ethiopia: species
suitability, root structure, and seedling establishment, growth and management. PhD
Dissertation, Katholieke Universiteit Leuven, Belgium.
Ritchie, J.T., 1972. Model for predicting evaporation from a row crop with incomplete
cover. Water Resour. Res. 8 (5), 1204–1213.
Rock, S., Myers, B., Fiedler, L., 2012. Evapotranspiration (ET) Covers. Int. J.
Phytoremediation 14 (sup1), 1–25.
Rosenberg, N.J., 1974. Microclimates: The Biological Environment. Wiley, New York.
Rozados-Lorenzo, M.J., González-Hernádez, M.P., Silva-Pando, F.J., 2007. Pasture pro-
duction under different tree species and densities in an Atlantic silvopastoral system.
Agrofor. Syst. 70 (1), 53–62.
Salisbury, F.B., Ross, C.W., 1992. Plant Physiology. Wadsworth Publishing Company,
Belmont – California.
Scanlan, C.A., Hinz, C., 2010. Insight into the processes and effects of root induced
changes to soil hydraulic properties. In: Proceedings of the 19th World Congress of
Soil Science, Soil Solutions for a Changing World, Brisbane, Australia, vol. 2, pp.
41–44.
Scholes, R.J., Archer, S.R., 1997. Tree-grass interactions in savannas. Annu. Rev. Ecol.
Evol. Syst. 28, 517–544.
Skerman, P. J., Riveros, F., 1990. Tropical grasses (No. 23). Food and Agriculture Org.
Sidle, R.C., Pearce, A.J., O’Loughln, C.L., 1985. Hillslope stability and land use. American
Geophysical Union. Water Resource Monographs, 11. Washington DC.
Simon, A., Collison, A.J.C., 2002. Quantifying the mechanical and hydrologic effects of
riparian vegetation on streambank stability. Earth Surf. Proc. Land. 27 (5), 527–546.
Smethurst, J.A., Briggs, K.M., Powrie, W., Ridley, A., Butcher, D.J.E., 2015. Mechanical
and hydrological impacts of tree removal on a clay fill railway embankment.
Géotechnique 65 (11), 869–882.
Smucker, A.J.M., McBurney, S.L., Srivastava, A.K., 1982. Quantitative separation of roots
from compacted soil profiles by the hydropneumatic elutriation system. Agron. J. 74
(3), 500–503.
Snyder, K.A., Richards, J.H., Donovan, L.A., 2003. Night-time conductance in C-3 and C-4
species: do plants lose water at night? J. Exp. Bot. 54, 861–865.
Soil Science Society of America, 2008. Glossary of Soil Science Terms. Soil Science Society
of America, Madison, Wisconsin.
Springman, S.M., Thielen, A., Kienzler, P., Friedel, S., 2013. A long-term field study for
the investigation of rainfall-induced landslides. Géotechnique 63 (14), 1177–1193.
Taha, H., Akbari, H., Rosenfeld, A., 1988. Residential cooling loads and the urban heat
island: the effects of albedo. Build. Environ. 23 (4), 271–283.
van Genuchten, M.T., 1980. A closed-form equation for predicting the hydraulic con-
ductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44 (5), 892–898.
Varado, N., Braud, I., Ross, P.J., 2006. Development and assessment of an efficient vadose
zone module solving the 1D Richards’ equation and including root extraction by
plants. J. Hydrol. 323 (1–4), 258–275.
Wilson, G.W., Fredlund, D.G., Barbour, S.L., 1997. The effect of soil suction on eva-
porative fluxes from soil surfaces. Can. Geotech. J. 34 (1), 145–155.
Yin, J.H., 2009. Influence of relative compaction on the hydraulic conductivity of com-
pletely decomposed granite in Hong Kong. Can. Geotech. J. 46 (10), 1229–1235.