Sei sulla pagina 1di 13

HYDROLOGICAL PROCESSES

Hydrol. Process. 19, 1653– 1665 (2005)


Published online 22 December 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hyp.5594

A novel hysteresis model in unsaturated soil


Han-Chen Huang,1 Yih-Chi Tan,1 Chen-Wuing Liu1 * and Chu-Hui Chen2
1 Department of Bioenvironmental Systems Engineering, Hydrotech Research Institute, National Taiwan University, Taipei 106, Taiwan
2 Department of Civil Engineering, Chung Kuo Institute of Technology, Taipei 106, Taiwan

Abstract:
This study presents a novel hysteresis model based on van Genuchten’s soil-moisture relationships. The proposed
model yields a series of closed-form relationships in which two shape factors ˛ and  are determined from the main
drying and wetting curves. Experimental and literature-cited data were used to assess model accuracy. The proposed
model was also compared with the Scott and KP models. Analytical results indicate that the present model is simple,
accurate and effective in constructing the series of wetting and drying scanning curves. Notably, the proposed model
outperforms the Scott and KP models in terms of model accuracy. Moreover, the novel model eliminates the pumping
effect and has perfect closure at scanning curve reversal points. Copyright  2005 John Wiley & Sons, Ltd.

KEY WORDS soil-moisture model; hysteresis; pumping effect

INTRODUCTION
The water content of unsaturated soil depends on complicated hydrological processes, such as infiltration,
evaporation and evapotranspiration. The dynamic change of soil water content generally does not follow the
same path as soil water retention curves. Besides the main drying and main wetting curves, there are primary
drying curves (from the wetting state to the drying state) and primary wetting curves (from the drying state
to the wetting state), and subsequently interactive exchange of drying and wetting states occurs to form a
series of scanning curves (Figure 1). The collective soil water retention curves of a given soil describe the
corresponding hysteresis properties.
Considerable effort has been focused on developing models that describe the hysteresis of the soil water
retention function  . The theory of independent domains was applied first (Poulovassilis, 1962), and
later the theory of dependent domains was developed (Mualem and Dagan, 1975; Kool and Parker, 1987).
The independent domain method assumes that all pore domains drain independently. The statistical method
evaluates the effect of pore radius on the drying and wetting processes (Topp, 1969; Mualem, 1974; Parlange,
1976, 1980). In reality, only pores with free access to the outside air can drain (Kutilek and Nielsen, 1994).
Moreover, this access depends on whether the surrounding pores are water or air-filled. To explain this
dependence, the domain-dependent model uses a domain-dependent factor to search its relationship with 
(Mualem and Dagan, 1975; Kool and Parker, 1987). Calibrating the domain-dependent model requires at
least a set of main wetting and main drying curves. The calibrated model generally yields accurate prediction
scanning curves for different media, although those models may be difficult to use.
Jaynes (1984) assessed four hysteresis methods (point, linear, slope and model II of Mualem and Miller,
1979) that were relatively easy to use, had been used previously, and only required knowledge of main soil
wetting and drying curves. Of the four methods, the model II and linear methods (Hanks et al., 1969) were
constructed to avoid pumping problems during cyclic change in soil water pressure. Meanwhile, the other

* Correspondence to: Chen-Wuing Liu, Ground Water Research Division, Dept of Bioenvironmental Systems, National Taiwan University,
No. 1, Section 4, Roosevelt Road, Taipei, Taiwan ROC. E-mail: lcw@gwater.agec.ntu.edu.tw

Received 28 March 2003


Copyright  2005 John Wiley & Sons, Ltd. Accepted 25 February 2004
1654 H.-C. HUANG ET AL.

(qr (1),yr (1))


Negative pressure head

(q 1,y1)
(q 3,y3)
Secondary drying curve qd (y,3)
Main drying curve qd (y,1)
Tertiary wetting curve
qw (y,4)
Primary wetting curve (q 2,y2)
qw (y,2)

Main wetting curve


qw (y,1) (qs (1),ys (1))
0 Water content

Figure 1. Curves given by Equation (1) with r d D r w and s d D s w , where (1 , 1 ), (2 , 2) and (3 , 3) denote reversal points

two methods, point (Dane and Wierenga, 1975) and slope, both displayed pronounced pumping, significantly
affecting algorithm performance when simulating numerous soil wetting–drying cycles. The linear method is
thus superior to the other three methods since it is easier to formulate, requires less computer memory, and
retains much of the accuracy of the other methods.
Kool and Parker (1987) developed a concise closed-form relation for describing hysteretic soil hydraulic
properties based on the van Genuchten (1980) parametric model and the Scott et al. (1983) empirical hysteresis
model (Scott model), but modified to account for air entrapment. The model provided a convenient and simple
method of incorporating hysteretic effects into numerical flow models to significantly improve prediction
accuracy with minimal data requirements. However, the pumping effect occurs when there is cyclic pressure
variation.
This study presents a revised hysteresis model which modifies the Kool and Parker (1987) model, hereafter
called the KP model, for describing soil-moisture retention curves and eliminating the pumping effect. The
accuracy of the proposed model is evaluated by comparing it with the Scott and KP models using data from
the laboratory and the literature.

MODEL DEVELOPMENT
Various models of water retention relationships have been proposed (Brooks and Corey, 1964; King, 1965;
van Genuchten, 1980). The most popular of these models is the van Genuchten model. This model assumes
that the main drying and wetting retention curves can be described accurately by the expression
  r
Se D D 1 C j˛ j m , <0
s  r
Se D 1, ½0 1

where Se denotes the effective saturation,  represents the volumetric moisture content l3 l3 , h is the pressure
head (l), s and r denote the saturated and residual water contents respectively, ˛l1  and  represent the
curve shape parameters and m D 1  1/.

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
HYSTERESIS MODEL IN UNSATURATED SOIL 1655

Additionally, van Genuchten (1980) described a graphical procedure for estimating ˛ and , and found that
s and r are easy to determine using laboratory experimental data. In practice, s and r may be obtained
from data compiled on different soil types (van Genuchten and Nielsen, 1985). In modelling the scanning
curves in the  – relationships, each member of the family of scanning curves is assumed to be described
by Equation (1) with the same shape parameters ˛ and . The following notations are used: the main drying
curve   is denoted by  d  , 1; the main wetting curve is denoted by  w  , 1; the primary wetting
curve is denoted by  w  , 2; the secondary drying curve is denoted by  d  , 3; the tertiary wetting curve
is denoted by  w  , 4; and so on (see Figure 1).
The parameters s , r , ˛ and  in Equation (1) are unknowns to be determined such that the main wetting
curve  w  , 1 is described by the parameters [s w 1, r w 1, ˛w , w ]. Moreover, to eliminate the pumping
effect, the relationships of r d 1 D r w 1 D r and s 1 D s w 1 D s are imposed to close the main
hysteresis loop. Thus the description of the main wetting and drying curves  w  , 1 and  d  , 1 is revised
to [s , r , ˛w , w ] and [s , r , ˛d , d ], respectively. Additionally, ˛w , w and ˛d , d are assigned the same
values in describing various wetting and drying scanning curves.
The reversal point of  d  , 1 occurs at (1 , 1 ) requiring  w  , 2 to satisfy

 w  , 2  r w 2 w w
D 1 C j˛w j m 2
s w 2  r w 2
The primary wetting curve  w  , 2 also passes through the reversal points at (1 , 1) and (s , s ). Substituting
(1 , 1 ) and (s , s ) into (2) yields

1  r w 2 w mw
D 1 C j˛w 1j  3
s 2  r w 2
w

s  r w 2 w mw
D 1 C j˛w sj  4
s w 2  r w 2

Solving s w 2 and r w 2 using Equations (3) and (4) with ˛w and w can describe  w  , 2.
 d  , 3 can be obtained similarly. Thus the relationship of  – for the ith-order scanning curve with no
pumping effect can be obtained using s w i and r w i, or s d i and r d i.

LABORATORY EXPERIMENT
To evaluate the accuracy of the present model, a laboratory experiment was performed to establish the series
of hysteresis scanning curves.
The tank was 15 cm high, 20 cm wide and 6 cm deep. Oven-dried white silica sand with grain diameters
between 0Ð074–0Ð297 mm was placed uniformly in an acrylic tank. The sand was then compacted using a
high-speed vibrator. The packed sand had a bulk density of 1Ð68 g cm3 and a porosity of 0Ð36.
Water pressure was measured using tensiometers manufactured by UMWELTANALYTISCHE MESS-
SYSTEME GMBH. Moreover, moisture content was measured using TDR (time domain reflectometry)
(reflectometer manufactured by IMKO MICROMODULTECHNIK GMBH).
The tensiometer and reflectometer used to measure the negative water pressure head and the moisture
content at the beam position were installed via threaded holes in the sand tank, and were connected to the
transducers directly. A computer data acquisition system was used for automatic data collection and storage.
The measured negative water pressure head and moisture content data were stored electronically over a set
time interval using a computer-driven data acquisition system (Figure 2).
Two experimental sets of hysteresis scanning curves were measured, including one main drying curve
 d  , 1 and its derivative curves  w  , 2,  d  , 3,  w  , 4 and one main wetting curve and its

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
1656

Adjustable
Water Tank
6 cm

Tensiometer Transducer TDR Transducer

Copyright  2005 John Wiley & Sons, Ltd.


Sand Tank

Data logger

15 cm
H.-C. HUANG ET AL.

1 cm Porous Plate

Computer
Control Control
Valve Valve

Figure 2. Illustration of the hysteresis laboratory experiment setup

Hydrol. Process. 19, 1653– 1665 (2005)


HYSTERESIS MODEL IN UNSATURATED SOIL 1657

tension head y j [qs (n ), qr (n ), aw, hw ]

No No
check reversal of check y j + 1 falls on
drying or wetting qw (y, n − 2) curve

Yes
Yes

adjust parameters to
qs (n + 1), qr (n + 1), ad, hd adjust parameters to
qs (n − 2), qr (n − 2), aw, hw

No
check y j + 1 falls on
qd (y, n − 1) curve

Yes

adjust parameters to
qs (n − 1), qr (n − 1), ad, hd

calculate q j + 1 (y j + 1)

Figure 3. Flow chart displaying the evaluation of soil water content from a series of soil water retention curves

derivatives  d  , 2,  w  , 3,  d  , 4. These two sets of data were used to compare the performance
of the present model with the Scott and KP models.

MODEL EVALUATION
The developed hysteresis model needs an efficient algorithm to accurately establish a series of scanning
curves. Soil drying and wetting conditions change dynamically. The present model compares the calculated
negative pressure head of the current and previous time steps according to the hysteresis relationship. If the
pressure head difference between the tension of the current and previous time steps is positive, hysteresis
occurs under continuous drying or wetting and the soil water content can be evaluated using the previously
established soil water retention curve parameters. Meanwhile, if the pressure head difference is negative, the
process is changing from either drying to wetting or wetting to drying. The soil water content of the current
time step can be evaluated using the relationship at the reversal points of Equations (3) and (4) to determine
s n and r n, and update the shape parameters ˛w , w or ˛d , d . Figure 3 illustrates the flow chart for
evaluating soil water content based on a series of soil water retention curves.

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
1658 H.-C. HUANG ET AL.

The present model is compared with the Scott and KP models. The Scott model used r , ˛,  and s d D s w
as variables, the KP model used ˛, s d D s w , r d D r w and d D w , and the present model used ˛, ,
s d D s w and r d D r w . The average absolute deviations, Em , are used to assess the performance of the three
models, defined as
1
n
Em D ji j 5
n i

where i and n denote the ith data point and the total number of data points evaluated in water retention curves,
while  i represents the difference between the measured and fitted soil water contents at the ith point.

Experimental data
Figures 4 to 6 illustrate the hysteresis scanning curves with primary drying or wetting of the experimental
measured data versus those obtained with the Scott, KP and present models. Additionally, Table I lists the
average absolute deviations between the experimentally measured data and the results of the three models.
The three models performed equally well on primary scanning curves, and the deviations among them were
generally within 0Ð0070. For the secondary and tertiary scanning curves, the Em values of the present model
are all below 0Ð0082 and achieve much more accurate water content predictions than the other two models.
The present model imposes a closed condition on the reversal points during the hysteresis of the drying

250 250
: Measured data : Measured data
: Fitted by : Fitted by
Scott model Scott model
: Predicted by : Predicted by
Scott model Scott model
200 200
Negative pressure head (cm)

Negative pressure head (cm)

150 150

100 100

50 50

0 0
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
Water content (%) Water content (%)
(a) (b)
Figure 4. Comparison of laboratory measured scanning curves with (a) primary drying and (b) primary wetting versus the Scott model
prediction

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
HYSTERESIS MODEL IN UNSATURATED SOIL 1659

250 250
: Measured data : Measured data
: Fitted by : Fitted by
KP model KP model
: Predicted by : Predicted by
KP model KP model
200 200

Negative pressure head (cm)


Negative pressure head (cm)

150 150

100 100

50 50

0 0
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
Water content (%) Water content (%)
(a) (b)
Figure 5. Comparison of the laboratory measured scanning curves with (a) primary drying and (b) primary wetting versus the KP model
prediction

and wetting processes, which eliminates the pumping effect since negative water pressure varies cyclically
(Figure 6). Notably, the Scott and KP models create a false pumping effect (see Figures 4 and 5). The predicted
water content tends to decrease with changing negative water pressure. The pumping effect is an aberration of
the algorithm and not an actual soil property. If the model requires no specific treatment to generate a series
of closed scanning curves, the resulting pumping effect causes a large error in soil water content prediction
during modelling of periodical wetting–drying cycles.

Literature-cited data
Model performance was compared using hysteresis data of three soils taken from the literature, including
Rideau clayey loam, Rubicon sandy loam (Mualem, 1976) and Dune sand (Gillham et al., 1979). Table II lists
the parameters used in the three models to simulate the hysteresis curves of three soils. s d and s w denote
the measured data, while other parameter values used in the three models were obtained via least square
fitting.
Table III lists the average absolute deviation, Em , of the measured and three model-fitted boundary (main
drying and main wetting) curves for the above three soils. The KP model has the highest Em values among the
three models for all soils. Meanwhile, the present model obtains similar Em values to the Scott model. Both
models produced reasonable and accurate simulated results. Table IV lists the Em obtained by measurement

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
1660 H.-C. HUANG ET AL.

250 250
: Measured data : Measured data
: Fitted by : Fitted by
present model present model
: Predicted by : Predicted by
present model present model
200 200
Negative pressure head (cm)

Negative pressure head (cm)


150 150

100 100

50 50

0 0
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
Water content (%) Water content (%)
(a) (b)
Figure 6. Comparison of the laboratory measured scanning curves with (a) primary drying and (b) primary wetting versus the present model
prediction

Table I. Average absolute deviations (Em ) between the experimentally measured data and the prediction results for the
three models
Scanning curve Model

Scott KP Present

Wetting Drying Wetting Drying Wetting Drying

Main 0Ð0031 0Ð0026 0Ð0105 0Ð0108 0Ð0036 0Ð0025


Primary 0Ð0033 0Ð0070 0Ð0035 0Ð0028 0Ð0035 0Ð0066
Secondary 0Ð0029 0Ð0035 0Ð0054 0Ð0095 0Ð0050 0Ð0031
Tertiary 0Ð0099 0Ð0128 0Ð0130 0Ð0137 0Ð0082 0Ð0042

and the three models when predicting the primary wetting and drying curves of the three soils. The same
hysteresis parameters used were as listed in Table I, and were obtained from the fitted boundary curves. The
largest Em of 0Ð0175 is found in the KP model of the Rubicon sandy loam in the primary drying curve. All
models yield large Em values for the Rubicon sandy loam. This type of soil is a domain-dependent medium.
In domain-dependent media, as in drying from a saturated state, groups of similar size pores (or domains)

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
Copyright  2005 John Wiley & Sons, Ltd.
Table II. Parameters used in the three models for simulating the hysteresis curves of the three soils [˛ cm1 ]

Soil type Measured Scott model KP model Present model

sd sw rd rw ˛d ˛w d w r ˛d ˛w d r ˛d ˛w d w

Rideau clayey loam 0Ð416 0Ð416 0Ð290 0Ð279 0Ð017 0Ð061 3Ð41 1Ð91 0Ð288 0Ð018 0Ð047 2Ð63 0Ð284 0Ð017 0Ð048 2Ð98 2Ð30
Rubicon sandy loam 0Ð381 0Ð381 0Ð176 0Ð140 0Ð011 0Ð052 6Ð31 1Ð87 0Ð170 0Ð013 0Ð037 3Ð30 0Ð165 0Ð011 0Ð040 5Ð18 2Ð43
Dune sand 0Ð301 0Ð301 0Ð098 0Ð093 0Ð030 0Ð054 8Ð90 4Ð26 0Ð101 0Ð030 0Ð052 6Ð77 0Ð097 0Ð030 0Ð052 8Ð59 4Ð95
HYSTERESIS MODEL IN UNSATURATED SOIL

Hydrol. Process. 19, 1653– 1665 (2005)


1661
1662 H.-C. HUANG ET AL.

Table III. Average absolute deviation (Em ) between the measured data
cited in the literature and boundary curves of three soils predicted by
the three models
Soil type Model

ScottŁ KPŁŁ PresentŁŁŁ

Rideau clayey loam 0Ð0032 0Ð0057 0Ð0057


Rubicon sandy loam 0Ð0045 0Ð0130 0Ð0055
Dune sand 0Ð0027 0Ð0080 0Ð0031

Ł Scott model variables: r , ˛,  and s d D s w .


ŁŁ KP model variables: ˛ and  d D  w ,  d D  w , d D w .
s s r r
ŁŁŁ Present model variables: ˛,  and  d D  w ,  d D  w .
s s r r

Table IV. Average absolute deviations (Em ) between measured data cited in the literature and the primary scanning curves
of three soils predicted by the three models

Soil type Model

Scott KP Present

Wetting Drying Wetting Drying Wetting Drying

Rideau clayey loam 0Ð0038 0Ð0050 0Ð0024 0Ð0065 0Ð0034 0Ð0071


Rubicon sandy loam 0Ð0106 0Ð0141 0Ð0118 0Ð0175 0Ð0076 0Ð0105
Dune sand 0Ð0074 0Ð0096 0Ð0096 0Ð0151 0Ð0057 0Ð0096

do not drain independently. Specifically, some similar sized pores may remain water-filled while water drains
completely from surrounding pores, and the pressure head is smaller than the local air entry value.
The Rubicon sandy loam has a narrow pore-size distribution and small pressure head changes cause
significant changes in water content. Domain dependence does not occur during the main drying process,
since the pore volume is drained partially and air can freely enter the pores. An empirical hysteresis model
or domain-independent model based on the simple rescaling parameters of the main drying curve thus cannot
accurately describe the primary drying and subsequent scanning curves of the domain-dependent medium.
However, domain-dependent media with high and well-defined air-entry pressure in the main hysteresis curves
generally do not occur in common soils.
To further assess model performance, Figures 7–9 show the simulated and measured scanning hysteresis
curves for Dune sand obtained using the Scott, KP and present models, respectively. The scanning hysteresis
loop comprises a primary wetting and a secondary drying curve. Notably, the hysteresis scanning loops were
not closed and the subsequent drying curves departed from the main drying curves. The pumping effect is
magnified as the pressure head undergoes cyclic variations.
The scanning curves generated by the present model display a perfect closed loop, i.e. the secondary drying
curve passes through the reversal point as the primary wetting curve departs from the main drying curve.
Moreover, the present model eliminates the pumping effect.

CONCLUSION
This study presents a novel hysteresis model using the van Genuchten relationship. The developed model used
two shape functions ˛ and , determined from the main drying and wetting curves, to construct a series of
closed-form expressions of hysteresis scanning curves. The present model was checked using the experimental

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
HYSTERESIS MODEL IN UNSATURATED SOIL 1663

60
Measured data (drying)
Measured data (wetting)
50 Main drying curve
Main wetting curve
Primary wetting curve
Negative pressure head (cm)

Secondary drying curve


40

30

20

10

0
5 10 15 20 25 30 35 40
Water content (%)
Figure 7. Scanning loop predicted by the Scott model for Dune sand

60
Measured data (drying)
Measured data (wetting)
50 Main drying curve
Main wetting curve
Primary wetting curve
Negative pressure head (cm)

Secondary drying curve


40

30

20

10

0
5 10 15 20 25 30 35 40
Water content (%)
Figure 8. Scanning loop predicted by the KP model for Dune sand

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
1664 H.-C. HUANG ET AL.

60
Measured data (drying)
Measured data (wetting)
50
Main drying curve
Negative pressure head (cm)

Main wetting curve


Primary wetting curve
40
Secondary drying curve

30

20

10

0
5 10 15 20 25 30 35 40
Water content (%)
Figure 9. Scanning loop predicted by the present model for Dune sand

and literature-cited data to evaluate model performance. The present model was also compared with the Scott
and KP models. For the experimental data, three models performed equally well on primary scanning curves,
but the present model achieves much more accurate water content prediction than the other two models. For
the literature-cited data, the KP model has the highest Em values among the three models for all soils. The
Scott model yielded slightly lower Em values than the present model. However, the hysteresis scanning loops
of the Scott model are not closed and the pumping effect occurs when there is cyclic pressure variation. The
present model eliminates the pumping effect and achieves perfect closure at the reversal point of the scanning
curve. The comparison results indicate that the present model outperforms the Scott and KP models. The
present model is simple, accurate and effective in constructing a series of hysteresis scanning curves, and
easy to use for practical application.

ACKNOWLEDGEMENTS

The authors would like to thank the National Science Council of the Republic of China for financially
supporting this research under Contract No. NSC 89-2313-B-002-243. Dr Cheng-Kaw Lee and Dr Kuo-Chin
Hsu are appreciated for their valuable comments.

REFERENCES
Brooks RH, Corey AJ. 1964. Hydraulic properties of porous media. Hydrology Paper 3, Colorado State University, Fort Collins, CO.
Dane JH, Wierenga PJ. 1975. Effect of hysteresis on the prediction of infiltration, redistribution and drainage of water in a layered soil.
Journal of Hydrology 25: 229– 242.
Gillham RW, Klute A, Heermann DF. 1979. Measurement and numerical simulation of hysteretic flow in a heterogeneous porous medium.
Soil Science Society of America Journal 43: 1061– 1067.
Hanks RJ, Klute A, Bresler E. 1969. A numeric method for estimating infiltration, redistribution, drainage, and evaporation of water from
soil. Water Resources Research 5: 1064– 1069.
Jaynes DB. 1984. Comparison of soil-water hysteresis models. Journal of Hydrology 75: 287–299.
King LG. 1965. Description of soil characteristics for partially saturated flow. Soil Science Society of America Proceedings 29: 359– 362.
Kool JB, Parker JC. 1987. Development and evaluation of closed-form expressions for hysteretic soil hydraulic properties. Water Resources
Research 23: 105–114.

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)
HYSTERESIS MODEL IN UNSATURATED SOIL 1665

Kutilek M, Nielsen DR. 1994. Soil Hydrology. Catena-Verlag: Cremlingen-Destedt, Germany; 73.
Mualem Y. 1974. A conceptual model of hysteresis. Water Resources Research 10: 514– 520.
Mualem Y. 1976. A catalogue of the hydraulic properties of unsaturated soils. Research Project Report 442, Technion, Israel Institute of
Technology, Haifa.
Mualem Y, Dagan G. 1975. A dependent domain model of capillary hysteresis. Water Resources Research 11: 452–460.
Mualem Y, Miller EE. 1979. A hysteresis model based on an explicit domain-dependence function. Soil Science Society of America Journal
43: 1067– 1073.
Parlange JY. 1976. Capillary hysteresis and the relationship between drying and wetting curves. Water Resources Research 12: 224– 228.
Parlange JY. 1980. Water transport in soils. Annual Review of Fluid Mechanics 12: 77– 102.
Poulovassilis A. 1962. Hysteresis of pore water, an application on the concept of independent domain. Soil Science 93: 405–412.
Scott PS, Farquhar GJ, Kouwen N. 1983. Hysteretic effects on net infiltration. Advances in Infiltration 11–83: 163– 170.
Topp GC. 1969. Soil-water hysteresis measured in a sandy loam and compared with the hysteretic domain model. Soil Science Society of
America Proceedings 33: 645–651.
van Genuchten MT. 1980. A close-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Science Society of
America Journal 44: 892–898.
van Genuchten MT, Nielsen DR. 1985. On describing and predicting the hydraulic properties of unsaturated soil. Annales Geophysicae 3:
615–628.

Copyright  2005 John Wiley & Sons, Ltd. Hydrol. Process. 19, 1653– 1665 (2005)

Potrebbero piacerti anche