test of a simple model for estimating evaporation from bare soils in different environments

Ecological Modelling 182 (2005) 91–105

Test of a simple model for estimating evaporation frombare soils in different environments

Mehmet Aydina,∗, Sheng Li Yangb, Nurten Kurta, Tomohisa Yanoc

a Faculty of Agriculture, Department of Soil Science, Mustafa Kemal University, Antakya 31040, Turkeyb Faculty of Agriculture, Tottori University, 4-101 Koyama-cho Minami, Tottori 680-8553, Japan

c Arid Land Research Center, Tottori University, 1390 Hamasaka, Tottori 680-0001, Japan

Received 16 June 2003; received in revised form 15 June 2004; accepted 12 July 2004

Abstract

A simplified model originally proposed by Aydin [Aydin, M., 1998a. A new model for predicting evaporation from barefield soil. In: Proceedings of the International Symposium and second Chinese National Conference on Rainwater Utilization,Xuzhou-Jiangsu, China, pp. 283–287] for estimating actual evaporation from bare soils was tested under different environmentalconditions. Field experiments were carried out on clay soils in a semi-arid region of Turkey. A sandy soil column-experiment ina drying chamber and a study with the same sand media in a greenhouse were conducted at Arid Land Research Center, TottoriUniversity, Japan, in order to test the performance of the model.

The model is based on the relations among potential and actual soil evaporation and soil–water potential at the top surfacelayer of the soil, with some simplifying assumptions. Input parameters of the model are simple and relatively obtainable viz.c ce.

ll sites. Thisa urement ofs©

K

1

t

f

-aridutesom

, butlds,be

0

limatic parameters for the calculations of potential soil evaporation and matric potential measured near the soil surfaDespite some differences between calculated and measured soil evaporation, the agreement was reasonable at a

greement seems to support the model assumptions, and the model is potentially valuable, but the objective measoil–water potential near the surface of the profile is difficult, especially for a drier upper layer.2004 Elsevier B.V. All rights reserved.

eywords: Soil evaporation; Soil–water potential; Modeling

. Introduction

Evaporation from a bare soil surface is the evapora-ion of water surrounding the soil particles as thin films

∗ Corresponding author. Tel.: +90 326 245 58 36;ax: +90 326 245 58 32.

E-mail address:[email protected] (M. Aydin).

and filling the pore spaces between them. In semiregions, evaporation from the soil surface constita large fraction of the total water loss not only frbare soils, but also from cropped fields (Jalota andPrihar, 1998; Wallace et al., 1999). Transpirationthrough crops is regarded as a beneficial depletionthe evaporation from the bare soil in irrigated fiewith a partial canopy cover or from weeds, can

304-3800/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.ecolmodel.2004.07.013

92 M. Aydin et al. / Ecological Modelling 182 (2005) 91–105

considered as a nonbeneficial depletion (Droogers andBastiaanssen, 2002). The transpiration from the canopylayers, and evaporation from the soil can be separatelycalculated using equations of the Penman–Monteithtype (Brisson et al., 1998; Zhang et al., 1996).

In general, the models of soil evaporation haveexpressed the loss rates from cropped areas rather thanjust those from bare soils. However, the evaporationfrom bare soils is an important process. In regionswhere summer fallow is practiced, direct evaporationfrom the soil accounts for about 50% or more oftotal precipitation (Hanks, 1992; Hillel, 1980; Hillel,1998). Parameterization of evaporation from a non-plant-covered surface is also very important in thehierarchy strategy of modeling land surface processes(Mihailovic et al., 1995; Mihailovic and Ruml,1996).

There are numerous methods for direct determina-tions of evaporation (Boast, 1986). However, instead ofworking with a method for directly measuring evapora-tion, many researchers prefer to use a practical model,which estimates actual evaporation. The main purposeof developing models is not merely to describe what iscurrently occurring, but to make predictions about thefuture. If models have a sound root in physical sciencesso that the processes involved are accurately depictedwithin the model, then it is possible to make predictionsby using models with different input scenarios (Kite etal., 2001). However, we must handle the problems ina variety of ways, and at a variety of degrees of thec area thatw an-a uiree ls ared an bem mayd odel-i ua-t r ofs od-e theis a-t ail-a erys bea

Many analitical solutions with simplifying assump-tions and initial and boundary conditions (Black et al.,1969; Gardner, 1959; Gardner and Hillel, 1962) andnumerical solutions under various initial and boudaryconditions (Hanks and Gardner, 1965; Hillel, 1975;Van Keulen and Hillel, 1974) are available which givethe reasonable predictions of evaporation from baresoils. In many practical situations, however, the de-tailed information, such as the hydraulic conductiv-ity function and water characteristic relation, neces-sary for these solutions, is not available. In these sit-uations, much simpler but not necessarily less precisemodels are required. Many simple models seeking rela-tions between the potential and actual evaporation, andsome soil property or surface vapor pressure have beentried with variable degrees of success (Alvenas andJansson, 1997; Brisson and Perrier, 1991; Campbell,1985; Hillel, 1975; Jackson et al., 1976; Katul andParlange, 1992; Liu et al., 1998; Malik et al., 1992,Ritchie, 1972; Staple, 1974). However, the choice ofan evaporation model is not simple. The complexitiesof the process of evaporation and its interaction withsoil properties are paramount (Beven, 1979).

The most commonly used model to predict directevaporation of water from bare soils is based onRitchie(1972)approach, which considers evaporation to occurin two distinct phases. During the first stage, surfaceresistance is zero, and the evaporation from the soilproceeds at the potential rate. During the falling ratestage, with a dry layer at the surface, the evaporationr anDe ilys vid-i verh ths(

plera dailyb ichp ationm ree-m asinge tuale tial.S fort tiale

omplexity, depending on the kind of problem wettempting to address. Thus, we always observehen the models are to be included in soil–water mgement, they need to be relatively simple and reqasily available parameters, whereas when modeeveloped to understand the basic physics, they cade as complicated as we deem necessary andemand many parameters. On the other hand, m

ng, as always is a matter of not just getting the eqions together in a computer program, but a mattekill, experience and judgement on behalf of the mller to match the right degree of complexity to

ssues addressed (Armstrong et al., 1993). There is noingle way that is likely to be applicable to all situions. Depending on the amount of information avble for input and required from output, either a vimple model or else a very complex model mayppropriate.

ate is reduced (Radersma and de Ridder, 1996; Vam et al., 1997; Wallace and Holwill, 1997). How-ver,Ritchie (1972)approach is unable to predict daoil evaporation accurately, but is capable of prong good estimates of cumulative soil evaporation oydrologically significant periods, i.e. weeks–monWallace et al., 1999).

Therefore, there is a demand for such simpproach to estimate actual soil evaporation on aasis. To our knowledge, it is still unknown as to wharameters can universally be used in evaporodels for bare soils. Although no complete agent exists among researchers, there is incre

xperimental and circumstantial evidence that acvaporation depends on the soil–water poteneveral simplified models have been advocated

he relation between the ratio of actual to potenvaporation and soil–water potential (Kijne, 1974).

M. Aydin et al. / Ecological Modelling 182 (2005) 91–105 93

For example,Ehlers and van der Ploeg (1976), Aydin(1994), andZhu and Mohanty (2002)showed evapora-tion from bare soils as a function of soil–water potential(matric potential) at the surface layer, neglecting theinfluence of the hydraulic gradient. Similarly,Beeseet al. (1977), and Huwe and van der Ploeg (1990)presented relations among the potential evaporation,actual evaporation and water potential at the soilsurface.

Therefore, we validated a simplified model origi-nally proposed byAydin (1998a), and also applied thismodel for the estimation of actual evaporation fromdifferent soils. The advantage of the model presentedis that it is successful in estimating actual evapora-tion on a daily basis for a range of soil types withonly simple parameter requirements such as matricpotential measured only near soil surface, and easilyobtainable climatic data. It is not difficult to use themodel for row crop plots with a reliable estimation,and to compare different alternatives and see whichones will be most likely to be acceptable to differentusers.

2. Theory

2.1. Basic principles

The atmospheric conditions that govern evaporationf froms ribet po-t atedu ;A ro( 996;S 97;W

E

w≈ orp tr x( -c r

pressure deficit of the air (kPa);ra, the aerodynamicresistance (s m−1); �, the latent heat of vaporization(MJ kg−1); �, the psychrometric constant (kPa◦C−1),and; 86.4, the factor for conversion from kJ s−1 toMJ d−1.

The evaporation from bare soils depends not only onthe atmospheric conditions but also on soil properties.The textbook information on evaporation from baresoils indicates that in the absence of a shallow watertable, evaporation following wetting is a variable pro-cess. It takes place in three stages—constant-rate stage,falling-rate stage and low-rate stage (Hillel, 1980; Idsoet al., 1974; Jalota and Prihar, 1998). In the contant-rate stage, soil evaporation is largely determined bymeteorological conditions (Dierckx et al., 1986; Ehlersand van der Ploeg, 1976; Jensen, 1974; Parlange andKatul, 1992; Ritchie, 1972). In the falling-rate stage,the evaporation is mainly determined by the hydraulicproperties of the soil. Theoritical consideration, andlaboratory and field observations have shown that cu-mulative soil evaporation during the falling-rate stageis proportional to the square root of time (Hillel, 1980;Monteith, 1981; Ritchie, 1972; Wallace et al., 1999).In the low-rate stage, a few centimeters of soil surfacedry out to air-dry wetness and water is lost slowly asvapor movement through the dry layer (Ehlers and vander Ploeg, 1976; Hillel, 1980; Jalota and Prihar, 1998).However, there is a question as to whether the low-ratestage of drying needs to be (or can be) defined (Alvenasand Jansson, 1997; Campbell, 1985).

stf topf ndt oil)t ver,t telyc mep ovet andt -t uslyt on-d sest for ad in hy-d uc-t udyt en-

rom a free water surface also govern evaporationoil. Many equations exist in the literature to desche evaporative demand of the atmosphere. Dailyential evaporation from bare soils can be calculsing the Penman–Monteith equation (Monteith, 1965llen et al., 1994) with a surface resistance of ze

Brisson et al., 1998; Radersma and de Ridder, 1aunders et al., 1997; Wallace and Holwill, 19allace et al., 1999; Van Dam et al., 1997).

p = �(Rn − Gs) + 86.4ρcpδ/ra

λ(� + γ)(1)

hereEp is the potential soil evaporation (kg m−2 d−1

mm d−1); �, the slope of saturated vapressure–temperature curve (kPa◦C−1); Rn, the neadiation (MJ m−2 d−1); Gs, the soil heat fluMJ m−2 d−1); ρ, the air density (kg m−3); cp, the speific heat of air (kJ kg−1 ◦C−1 = 1.013);δ, the vapo

After rainfall or wetting soil, water is rapidly lorom the upper portion of surface layer. Hence, theew centimeters of the soil dry out more rapidly ahis dry layer acts as a barrier (especially in tilled so both liquid and vapor flow to the surface. Howehe liquid flow across surface zone is not compleut off. In fact, sub-layer of the soil may for some tiermit upward capillary flow to some distance ab

he interface, depending upon the water contentype of the soil (Jalota and Prihar, 1998). The evaporaive water loss from upper layer reduces simultaneohe water content, water potential and hydraulic cuctivity of the layer, but at the same time it increa

he hydraulic gradient. In other words, the reasonecrease in actual soil evaporation is the decreaseraulic conductivity of the soil, and hydraulic cond

ivity is related to soil–water potential. One can sthe effect of lower potential on evaporation using P


man equation (Monteith and Unsworth, 1990). Also, amajor requirement for studying vertical soil–water flowis the numeric solution of Richards equation, subjectto specified initial and boundary conditions and withknown relations among the volumetric water content,soil–water potential and hydraulic conductivity (Beeseet al., 1977; Van Dam et al., 1997; Warrick et al., 1971).

However, under normal field conditions, neithersoil–water potential nor water content at the soil sur-face as well as the evaporative flux through soil surfaceis known. Assuming that the water potential at dry soilsurface is at equilibrium with the atmosphere, the min-imum water potential can be derived from the Kelvinequation (Brown and Oosterhuis, 1992; Feddes et al.,1978; Kirby and Ringrose-Voase, 2000).

Ψad = RgT

mgln Hr (2)

whereΨad is the water potential for air-dry conditions(cm of water);T, the absolute temperature (K);g, theacceleration due to gravity (981 cm s−2); m, the molec-ular weight of water (0.01802 kg mol−1); Hr, the rela-tive humidity of the air (fraction), and;Rg, the universalgas constant (8.3143× 104 kg cm2 s−2 mol−1 K−1).

2.2. Description of the model

The model used to estimate actual evaporation frombare soils is based on approach originally proposed byAydin (1998a). We assumed that initially evaporationf til at ctu-a holdi po-t holdp ra-t holdp rly,t dry-i ver,t inga1 ayd as ab r,1

ationi the

threshold potential (Ψtp) near the soil surface isreached. Beyond the threshold potential, the evap-oration falls progressively below the potential rate.When the top surface layer of the soil dries out toits final air-dry value, or the water potential (matricpotential) at this dry layer reaches the minimumvalue, the so-called air-dryness (Ψad), the soil thenno longer evaporates at a considerable rate, exceptthe neglected water losses by the slow process ofvapor diffusion.Campbell (1985)has also reportedthat the simplest option, and one which is often usedin modeling of water flow, is to neglect vapor flowfor most applications. On the other hand, the resultsobtained byBeese et al. (1977), andHuwe and van derPloeg (1990)showed clearly that the ratio of actualto potential evaporation could be linearly correlatedto the log-transformed soil–water potential. It is nowpossible to define the slope of the straight line inFig. 1.

f = Ea/Ep

log|Ψ | − log|Ψad| (3)

RewritingEq. (3)yields whenEa/Ep = 1.0

f = 1

log|Ψtp| − log|Ψad| (4)

wheref is the slope of the solid line (Fig. 1); Ea andEpare actual and potential evaporation rates, respectively;Ψtp, the absolute values of soil–water potential (matricpotential) at which actual evaporation starts to drop be-l terp eo itub fw

f

F

E

(m a pa-r m ab es ace,

rom a wet soil proceeds at the potential rate unhreshold of water potential at the soil surface (ally near the surface) is reached. When this thres

s reached, the soil evaporation starts to drop belowential one. Thus, we called this threshold as thresotential. During the initial stage, the rate of evapo

ion can remain nearly constant. Beyond the thresotential, the evaporation falls progressively. Simila

he soil surface becomes progressively drier, andng front moves into the soil. Sooner or later, howehe soil surface approaches equilibrium with overlytmosphere, i.e. becomes approximately air dry (Hillel,980). With time, the top surface layer of the soil mry out to air-dry wetness, and this dry layer actsarrier to the water flow to surface (Jalota and Priha998).

Hence, it was assumed that the actual evapors at most equal to potential evaporation until

ow potential one;Ψad, the absolute values of soil–waotential at air-dryness, and;Ψ is the absolute valuf soil–water potential to be determined in setweenΨtp andΨad. The values of allΨ are in cm oater.Combination ofEqs. (3) and (4)results in:

= Ea/Ep

log|Ψ | − log|Ψad| = 1

log|Ψtp| − log|Ψad| (5)

inally, Ea can be obtained fromEq. (5)

a = 1

log|Ψtp| − log|Ψad| (log|Ψ | − log|Ψad|)Ep (6)

Thus, it was assumed that the water potentialΨ )easured near the soil surface can be used as

ameter in the estimation of actual evaporation froare soil (Aydin, 1998a). If it is difficult to measure thoil–water potential (matric potential) near the surf


Fig. 1. Schematic representation of the relation between the ratio of actual to potential evaporation and matric potential (modified fromAydin,1998a).

one can use the potential observed at deeper section ofthe top layer for calculations. However, the model leadsto overestimation of soil evaporation during a dryingperiod when soil–water potential measured at the lowersection of the surface layer, which is wetter than thatjust below the surface, is used. On the other hand, theuse of the water potential of lower sections of the layermay result in underestimation of evaporation duringa slightly re-wetting cycle (Aydin, 1998b). In conse-quence, when soil–water potential at most upper sec-tion of the surface layer (i.e. 1 cm depth) could not bemeasured, we assumed that with use of the potential at5 and/or 10 cm depths, the evaporation rate was givenby the same formula but multiplied by a factor. Alter-natively, for the calculations, the soil–water potentialcan be parameterized as an average term obtained bysumming the weighted soil–water potential at different

depths (i.e. from soil surface to 20 cm)

Ψ =∑z

0

ΨiDi

z(7)

in which Ψi is the soil–water potentials of theith soillayer;Di, the thickness of theith soil layer, and;z, thedepth from the surface to selected deepest layer.

3. Materials and methods

3.1. Experimental sites and measurements

3.1.1. Experiment 1This experiment was conducted at the experimen-

tal farm of the Agricultural Faculty, Mustafa KemalUniversity in Antakya (36◦18′ N, 36◦11′ E), Southern


Table 1Some properties of the soils at the experimental sites

Depth (cm) Particle size distribution Dry bulkdensity(g/cm3)

Porosity(%)

Saturated hydraulicconductivity (cm/h)

Volumetric water content (%) at

Sand (%) Silt (%) Clay (%) Field capacity(−0.033 MPa)

Wilting point(−1.5 MPa)

Experiment 1 (Antakya)0–15 29.2 14.7 56.1 1.48 55.2 0.19 42.7 33.915–30 30.8 12.3 56.9 1.50 53.1 0.16 40.1 35.330–60 30.5 13.2 56.3 1.51 53.2 0.15 41.8 34.9≥60 Not sampled

Experiment 3 (Adana)0–15 22.6 23 54.4 1.29 59.9 1.43 40.7 29.915–30 23.9 22.8 53.3 1.33 59.6 1.43 40.0 30.130–60 24 21.3 54.7 1.35 58.8 1.44 41.6 29.560–90 22.1 22.9 55 1.38 58.5 1.39 42.9 3290–120 22.2 22.8 55 1.42 58.4 1.22 43.3 31.7120–150 21.2 21.5 57.3 1.43 58.1 1.1 43 32.2

Turkey. A 10 m by 10 m plot chosen on a Chromic Ver-tisol was used. The colour of moist soil is dull yellowishbrown throughout the profile. The terrain is nearly flat,with no appreciable slope. The soil has no water ta-ble and salinity problems. Some physical properties ofthe soil are shown inTable 1. The climate of the re-gion is semi-arid Mediterranean with mild rainy win-ters and hot dry summers. Some monthly climatic datafor the period of the experiment are given inTable 2.The weather parameters were obtained from the An-takya meteorological station near the test plot.

During the experimental phase, the test plot andthe surrounding field were kept free of vegetation. Theplot was instrumented with tensiometers and gypsumblocks. Duplicated tensiometers were installed at the5, 10, 20 cm depths and read daily. Matric potential

Table 2Some monthly climatic data of the experimental sites for the studied period

Month Meantemperature(◦C)

Mean relativehumidity (%)

Meanradiation(kJ/m2/day)

Mean durationof sunshine(h/day)

Mean windspeed (m/s)

Rainfall(mm)

Experiment 1 (Antakya)July 28.2 66 18494 12 4.4 –August 27.7 68 17197 11.4 4.1 –September 25.6 64 14080 9.6 3 31.7

Experiment 3 (Adana)April 18 63.4 13984 7.5 2.3 48.5May 22. 4 64.6 14863 9.9 2.2 92June 25.5 67 16784 11.1 2 62.6

lower than −0.08 MPa was measured by gypsumblocks at the same depths by utilizing the electricalresistance–water potential relationship curves ob-tained before plot instrumentation. Soil evaporationwas measured with micro-lysimeters (Boast andRobertson, 1982; Evett et al., 1995). These consistedof open PVC cylinders 57.0 cm long, 10.2 cm internaldiameters and with a thickness of 0.4 cm. The cylin-ders, along with a steel cutter, were filled by pressingthem into wetted soil with a hydraulic arm. They werethen carefully excavated and wire-gauze sieves wereaffixed to the bottom in order to retain soil particleswithout disturbing drainage. The micro-lysimeterswere placed in the field in performed holes lined withopen-ended PVC cylinders of 11.4 cm inside diameter,which served as soil retaining walls. The surface of


the soil inside the micro-lysimeters was flushed withthe surface of the surrounding soil at the beginningof the experiment (Brisson et al., 1998; Yang et al.,2002). Three micro-lysimeters were installed in theexperimental plot. They were weighed every 3–4 daysat the same time with an electronic balance.

Evaporation was calculated from the change inweight. A rain gauge was located near micro-lysimeters. At the beginning and during the experimen-tal period, several times the plot with micro-lysimeterswas watered to observe the changes when soil was re-wetted.

3.1.2. Experiment 2To obtain further information about the rate of evap-

oration from a bare soil, a pot-experiment was carriedout in a growth chamber at Arid Land Research Center,Tottori University, Japan, by controlling light, tempera-ture and humidity. The pots with 337.5 cm2 surface areaand 28 cm long were filled with sandy soil. The soil wasArenosol (silicious sand, Typic Udipsamment) with96% sand. Average field capacity and permanent wilt-ing point of the sand in pots were 0.074 cm3/cm3 and0.022 cm3/cm3, which corresponds to−0.006 MPa and−1.5 MPa, respectively in matric potential (Yang et al.,2002). The average dry bulk density was 1.51 g/cm3.

Three different treatments were applied to pots withthree replications. (I) The soils in the pots withoutdrainage system were kept at saturation throughout theexperiment. A water reservoir device—similar to Mar-i neart wasa ver,w e toc di-r sedo fa-c onw icew nt tos

pedw d atd con-n 1 cm,t intos ten-s cted

by cementing a 2 cm long 0.6 cm outside diameter (OD)porous-ceramic cup to a 0.6 cm OD acrylic tube wereused. After filling with water, a pressure transducer (HI-TECHS Corporation) was connected to the water-filledtubing using a short piece of tygon tubing. Transduc-ers were connected to a 21X data logger (CampbellScientific Inc.) providing output in mV. The outputswere continuously recorded using a computer. FinallymV values were converted to MPa based on calibrationcurve (HI-TECHS Corporation).

3.1.3. Experiment 3A field experiment similar to the experiment 1 was

conducted at the experimental farm of the Agricul-tural Faculty, Cukurova University in Adana (36◦59′N, 35◦18′ E), Southern Turkey. The soil at the site isa clayey Luvisol, which is a dull reddish brown colourthroughout the moist soil profile (Aydin, 1994). Thesoil has no water table and salinity problem, and theterrain is almost flat. Some physical properties of thesoil are given inTable 1. The climate of the regionis subhumid-semiarid Mediterranean with mild rainywinters and hot dry summers. The long term annualmean temperature at the site is 18–19◦C. The aver-age annual precipitation is about 700 mm, and the rel-ative humidity is 66% (Aydin and Huwe, 1993). Themeteorological data were taken from the neighboringAdana meteorological station for the measurement pe-riod (Table 2). In addition, a raingauge was installed onthe plot at 1 m high, and rainfall was recorded daily.

ro-l met-r ndb the3 wasd in3 aterp tedt Ma-t byg tricalr

3ger

r treesi Cen-t

otte bottle—with burette system was connectedhe bottom of pots. The surface level of the waterbout 1 cm below the surface of soil in pots. Howeater films appeared at the surface of the soil duapillary fringe. Thus, potential evaporation wasectly calculated from the amount of water loss ban soil surface area. (II) The pots without drainageility were saturated initially to monitor evaporatiithin a drying period. (III) Pots with drainage devere saturated at the beginning of the experimeimulate field conditions.

Except the first pot-set, the others were equipith tensiometers. The tensiometers were installeepths of 1, 5 and 10 cm below the soil surface andected to pressure transducers. For the depth of

wo replicated micro-tensiometers were installedoil almost laterally. For the other depths, regulariometers were used. Micro-tensiometers constru

In this experimental site, we did not use micysimeters. Soil–water content was measured graviically for 15 cm increments in the top 30 cm layer ay neutron probe for 30 cm increments between0 and 150 cm depths. Thus, the soil evaporationetermined from water depletion of the soil profile-day intervals depending on the weather. Soil–wotential (matric potential) was measured by triplica

ensiometers installed at the 5 and 10 cm depths.ric potential lower than−0.08 MPa was measuredypsum blocks at the same depths using the elecesistance–water potential curves.

.1.4. Experiment 4This study was carried out in the context of a lar

esearch examining evapotranspiration of orangen a greenhouse lysimeters at Arid Land Researcher, Tottori University, Japan (Yang et al., 2002; 2003).


The sandy soil in the greenhouse was the same as thatused for the pot experiment (see above Experiment 2).The daily potential soil evaporation was estimated frompan evaporation based on a relationship obtained be-tween the saturated surface evaporation around treesand pan evaporation. Three pans (20 cm in diameterand 10 cm in depth) located in the greenhouse wereused for this purpose. Direct evaporation of water fromthe soil was measured using three micro-lysimeterslocated around trees (the plant spacing was 1.5 m×1.5 m) in the same greenhouse. The dimensions of themicro-lysimeters were 60 cm long by 13 cm in diam-eter. The micro-lysimeters and surrounding sand wereartificially wetted when trees were irrigated. We did notmeasure soil–water potential of sand media directly.However, we monitored soil–water content with TDRsensors installed at the depth of 7.5 cm, and derivedsoil–water potential from the soil–water characteristicdata for the same sand (Inoue et al., 1984). Climato-logical data were obtained from an automated stationin the greenhouse. Parameters measured and recordedwere air temperature and average relative humidity. Forthe experimental period, mean temperature and relativehumidity in the greenhouse were 29.8◦C and 72%, re-spectively (Yang et al., 2002). Aerial microclimate wasdifferent in the greenhouse from outside. The tempera-ture and relative humidity inside were 2–5◦C and 15%higher, respectively, than those outside (Yang et al.,2003).

3

-t tiale theP del,A ionw -p uredu ttere ial)w om0 out1 . Ford me-t atea ea-s . In

the field experiments, the averaged soil–water potentialmeasured at the 5 and 10 cm depths was used. There-fore, we assumed that with the use of the potentialat the deeper layers, the evaporation rate was givenby the same formula but multiplied by a coefficient(correction coefficient). This coefficient value wasdetermined from linear regression between measuredand calculated evaporation fromEq. (6), by settingintercept as zero. Once the measured and calculateddata had been fitted, the simulation of soil evaporationwas carried out continuously for the entire periodunder consideration, without an additional calibration.The potentials measured at midday were used forcalculations.

4. Results and discussions

The threshold potential value (Ψtp) was firstlydetermined in order to estimate actual evaporation. Asa typical example, the observations for sandy soil (potexperiment) are shown inFig. 2. The ratio of actualevaporation from the soil saturated initially to thepotential evaporation from soil kept at saturation wasplotted against time. The correspondinglog|Ψ | valueswere given in the plot as labels to the points. The actualevaporation was about equal to potential evaporation(Ea/Ep ≈ 1.0) as long as the matric potential measuredat the depth of 1 cm had not reached a value of 15.2 cmof water (log 15.2≈ 1.18). Beyond this threshold, ther lcu-lt d tob hedd tuale facew is inaB ; Zhua oldp et luefR aseI edsaaa as

.2. Calculations

The data necessary forEq. (6) are easily deerminable. Under field conditions, daily potenvaporation from bare soils was calculated usingenman–Monteith equation. In the original moydin (1998a)calculated the potential soil evaporatith Priestley and Taylor (1972)formula. For pot exeriment, potential evaporation was directly meassing continuously saturated soil-pots. In the laxperiment, soil–water potential (matric potentithin the working range of micro-tensiometers (frto −0.08 MPa) installed near the soil surface (abcm) was used for actual evaporation calculationsrier soils, entrapped air restricts the use of tensio

ers. Thus, in the pot experiment, we could not estimctual soil evaporation using soil–water potential mured at the 1 cm depth for the entire drying period

atio ofEa toEp began to decrease. Thus, for the caations, we used 15.0 cm of water asΨtp for sand. Thehreshold water potential for clay soils was assumee equal to approximately 60 cm of water (unpublisata of a parallel study). The dependence of acvaporation on soil–water potential near the suras clearly observed in this study. The tendencygreement with the literature findings (Aydin, 1998b;eese et al., 1977; Huwe and van der Ploeg, 1990nd Mohanty, 2002). On the other hand, the threshotential value (Ψtp) found in this study may giv

he information about corresponding threshold vaor similar soils for theRitchie (1972)approach. Initchie’s model, soil evaporation switches from phto phase II when the cumulative evaporation excevalue,U. The value ofU is soil dependent.Wallace

nd Holwill (1997)reported thatU value was as lows 3 mm in very sandy soils. A value of 6 mm w


Fig. 2. The ratio of observed actual (Ea) to potential (Ep) evaporation vs. time for sandy soil columns in a drying chamber (the values oflog-transformed matric potential measured at the 1 cm soil depth are given to the points as labels).

used by Ritchie for a Plainfield sand, 9 mm for a loamand 12 mm for a clay. A value of 7.5 mm was used byWallace et al. (1999)for a sandy loam soil. Note thatthis aspect was not a part of the model tested in thisstudy.

A comparison of measured and predicted evapora-tion rates from drying soil columns (pots) is shownin Fig. 3, together with potential evaporation. Agood agreement between water losses observed from

Fig. 3. Comparison of calculated (Ec), measured (Em) and potential (Ep) daily evaporation from sandy soil columns in a drying chamber (Ec

was determined only within the working range of micro-tensiometers installed at the 1 cm soil depth).

pot weight changes and evaporation calculated usingsoil–water potential at the 1 cm depth is evident. If thesurface of the soil is wet, the evaporation rate staysnearly constant for some time, and then suddenly de-creases. When the soil dries so sufficiently that watercannot be supplied to the surface fast enough to meetthe evaporative demand, the soil surface dries and theevaporation rate is reduced (Campbell, 1985). Whenthe top surface layer dried out, we could not monitor


the potential at the 1 cm depth due to the limited work-ing range of tensiometers, but we observed it at the5 and 10 cm depths. However, it should be noted thatwhen we calculated the actual evaporation with useof soil–water potential measured at the 5 and/or 10 cmdepths, the results were overestimated, and on average,2–3 times higher than observed evaporation (data notshown). Briefly, for calculations, one can use the poten-tial values measured at the lower section of the surfacelayer, which is wetter than that just below the surfaceduring drying periods, but the estimated evaporationrates should be multiplied with a correction coefficient(cd) to fit the estimations to the observed evaporationdue to differences in potentials at upper and, lowersections of the layer.Aydin (1998b), comparing theevaporation rates calculated by use of soil–water po-tential individually at different depths, i.e. 5, 10, 20 and30 cm under field conditions, also reported that waterpotential near the soil surface was more representative

F poratiT

than that of deeper layers. In sand, high evaporationenhances the rate of water loss from the surface layerof a few centimeters. This reduces the water contentand hydraulic conductivity of the upper layer, but atthe same time it increases the hydraulic gradient. Thelatter tends to sustain the upward flow for some time.Eventually, however, the flow from lower layers fails toreplenish the water loss from the surface because of lowcapillary flow even if lower layers are wet. Thus, thedry surface of sand acts like soil mulch at the surfaceof fine-textured soils created by tillage.

The rates of estimated and measured evaporationfrom a vertic soil under field conditions in Antakyaare illustrated inFig. 4. Generally, the simulated soilevaporation matches quite well with the measuredevaporation. In the case of using the average soil–waterpotential of the 5 and 10 cm depths, a cd value of 0.67was fit. The use of water potential at deeper layersleads to overestimation of evaporation during drying

ig. 4. A comparison of calculated (Ec) and measured (Em) daily evaurkey.

on from a bare vertic soil along with potential values (Ep) in Antakya,


period, and its underestimation for a slightly re-wettedsurface. However, due to restrictions of matric po-tential measurements near the surface of a dry soil,one has to use the water potential of a deeper layer,especially when the most upper layer of the soil profileis dry. However, this does not necessarily mean that themodel is not capable of providing good estimates ofactual soil evaporation. This means that the potentialmeasurements as close to the soil surface as possibleare preferable for accurate calculations. Otherwise,estimated actual evaporation should be multiplied bya coefficient experimentally determined. In this study,it was found that the model results, after having beenfitted, matched well with the measured values.

Although estimated and observed evaporations arecomparable, some deviations are evident. A reasonfor differences might be the errors in measurementsof soil–water potentials. Meanwhile, the hysteresisand hydraulic gradient effects were not taken into

F vapora ,T

consideration in the model calculations.Beese andvan der Ploeg (1976)also reported the same effects.In addition, a similar problem may be thought to beresponsible for deviations in determination of potentialevaporation and the water potential for air-dry con-ditions. The other reasons of considerable differencesbetween calculated and measured evaporation ratescould be the complexity of water regime of clay soils.Similar observations for clay soils were also reported inprevious studies (Aydin and Huwe, 1993; Aydin, 1994;Hasegawa and Sato, 1987; Ritchie et al., 1972). Inother words, swelling, shrinkage and cracking proper-ties of the soils may influence the drying and rewettingbehaviors of the soils (Kirby and Ringrose-Voase,2000), and consequently the process of water evapora-tion from our vertic soil. It is a fact that the estimationof water balance components of swelling soils is moretricky to interpret compared with non-swelling soils.On the other hand, there may be another possible error

ig. 5. A comparison of calculated (Ec) and measured (Em) daily eurkey.

tion from a bare clay soil along with potential values (Ep) in Adana


in measurement of soil evaporation by mico-lysimetes.Generally, micro-lysimeters provide good estimates ofsoil evaporation at the wet condition but underestimateevaporation from dry soil, because the evaporationfrom whole profile may be greater than that frommicro-lysimeters (Kirby and Ringrose-Voase, 2000).

The evaporation rates from a clay soil plot in Adanaare presented inFig. 5. The potential evaporation gen-erally was lower in April than in May and June becauseof the lesser atmospheric demand. However, the rate ofactual evaporation was mainly affected by rainfall andpresumably soil wetness, increased with higher waterpotentials. The results presented in this section are com-parable with those found byAydin (1998b)for similarsoil and climatic conditions. The differences betweentwo studies were the formula used for potential evapo-ration calculations, threshold potentials as well as theselected depths at which soil–water potential was mea-sured. For this experimental site, the average soil–waterpotential of the 5 and 10 cm depths was used. A cd value

F tial (Ep) treesi

of 0.8 was suitable to fit the estimations. It can be seenthat the estimated and the observed soil evaporationrates compare reasonably well. The observed evapora-tion might have some errors associated with unknownfraction of water loss due to drainage after rainfall. Theevaporative water loss was determined from soil–waterdepletion of whole profile in this experimental site.

The estimated and observed soil evaporation ratesfrom the sand planted with orange trees in the green-house are compared inFig. 6. A cd value of 0.4 wasused to fit the estimations to the measurements. De-spite some expected errors in derivation of matric po-tential values from the equations obtained previouslyfor water content and potential relationships, and pos-sible systematic errors connected with the TDR mea-surements (Alvenas and Jansson, 1997), the agreementbetween calculated and measured evaporation is partic-ularly well. In addition, possible errors in determina-tion of potential soil evaporation could have influencedthe evaporation calculations. Determination of the in-

ig. 6. A comparison of calculated (Ec), measured (Em) and potenn a greenhouse.

daily evaporation from sandy soil between the row of orange


Fig. 7. Daily calculated (Ec) vs. measured (Em) soil evaporation.

put parameters for testing a model is ideally done bymeasuring them independently. In situ determination ofsoil–water potential near soil surface is very difficult ina drier upper layer. The use of available soil–water char-acteristics and measurement of water content of a uppersoil layer is a possibility to overcome the difficulties inpotential measurements. As reported byKoorevaar etal. (1983), in a bare, dry soil, the limiting factor forevaporation is the supply of the water to the soil sur-face. Water in the soil is conserved by the formation ofa very dry topsoil with a low water conductivity. Theyindicated that the combined liquid and vapor of waterin a topsoil at pF 4 may produce a total flux density ofno more than approximately 1 mm d−1.

In order to evaluate the performance of the modelfor any soil texture, the actual evaporation data fromall sites were pooled; and then the estimated evapora-tion rates were plotted against measured ones (Fig. 7).The scatter of data is high, although there are some

deviations. The model tested in this study appears tobe applicable for a wide range of soils if soil specificparameters are used.

5. Conclusions

It was found that the simple daily time step model,which depends on the period considered for the cal-culation of potential soil evaporation, was capable ofestimating actual evaporation from soils as a functionof matric potential measured at the 1 cm soil depth ac-curately. But it was unable to provide good estimationsin the case of using water potential at deeper layers.

Finally, it can be concluded that as a result of itssimple and practical nature, the tested approach can beused in practice to estimate evaporation from bare soilsas well as soil evaporation from row crop plots, i.e. treeplantations, with some limitations.


References

Allen, R.G., Smith, M., Perrier, A., Pereira, L.S., 1994. An updatefor the definition of reference evapotranspiration. ICID Bull. 43(2), 92.

Alvenas, G., Jansson, P.-E., 1997. Model for evaporation, moistureand temperature of bare soil: calibration and sensitivity analysis.Agric. For. Meteorol. 88, 47–56.

Armstrong, A.C., Matthews, A.M., Portwood, A.M., Addiscott, T.,Leeds-Harrison, P.B., 1993. Modelling the effects of climatechange on the hydrology and water quality of structured soils.NATO Advanced Research Workshop, Silsoe, 20–24 September,1993.

Aydin, M., 1994. Hydraulic properties and water balance of a claysoil cropped with cotton. Irrig. Sci. 15, 17–23.

Aydin, M., 1998a. A new model for predicting evaporation from barefield soil. In: Proceedings of the International Symposium andsecond Chinese National Conference on Rainwater Utilization,Xuzhou-Jiangsu, China, pp. 283–287.

Aydin, M., 1998b. Test of a simple model for estimating actual evap-oration from bare soil. In: Proceedings of the MSY InternationalSymposium on Arid Region Soils. International Agrohydrol-ogy Research and Training Center, Menemen-Izmir, Turkey, pp.432–437.

Aydin, M., Huwe, B., 1993. Test of a combined soil moisture/soilheat simulation model on a bare field soil in Southern Turkey. Z.Pflanzenern. Bodenk. 156, 441–446.

Beese, F., van der Ploeg, R.R., 1976. Influence of hysteresis on mois-ture flow in a undisturbed soil monolith. Soil Sci. Soc. Am. J. 40,480–484.

Beese, F., van der Ploeg, R.R., Richter, W., 1977. Test of a soil watermodel under field conditions. Soil Sci. Soc. Am. J. 41, 979–984.

Beven, K., 1979. A sensitivity analysis of the Penman–Monteith ac-tual evapotranspiration estimates. J. Hydrol. 44, 169–190.

B n ofl. Soil

B highis,. 9,

B forora-

B me-ax-07,

B soils. 27

B andome

C dels0.

Dierckx, J., Belmans, C., Pauwels, P., 1986. SWATRER a computerpackage for modelling the field water balance. Laboratory of Soiland Water Engineering, K. U. Leuven, Belgium.

Droogers, P., Bastiaanssen, W., 2002. Irrigation performance usinghydrological and remote sensing modeling. J. Irrig. Drain. Eng.128 (1), 11–18.

Ehlers, W., van der Ploeg, R.R., 1976. Evaporation, drainage andunsaturated hydraulic conductivity of tilled and untilled fallowsoil. Z. Pflanzenern. Bodenk. 139, 373–386.

Evett, S.R., Warrick, A.W., Matthias, A.D., 1995. Wall material andcapping effects on microlysimeter temperatures and evaporation.Soil Sci. Soc. Am. J. 59 (2), 329–336.

Feddes, R.A., Kowalik, P.J., Zaradny, H., 1978. Simulation of fieldwater use and crop yield. Simulation Monographs. Pudoc. Wa-geningen, p. 189.

Gardner, W.R., 1959. Solutions of the flow equation for the dryingof soils and other porous media. Soil Sci. Soc. Am. Proc. 23,183–187.

Gardner, W.R., Hillel, D., 1962. The relation of external evaporativeconditions to the drying of soils. J. Geophys. Res. 67, 4319–4325.

Hanks, R.J., 1992. Applied Soil Physics: Soil Water and TemperatureApplications, second ed. Springer-Verlag, New York/Berlin p.176.

Hanks, R.J., Gardner, H.R., 1965. Influence of different diffusivitywater content relations on evaporation of water from soils. SoilSci. Soc. Am. Proc. 29, 495–498.

Hasegawa, S., Sato, T., 1987. Water uptake by roots in cracks andwater movement in clayey subsoil. Soil Sci. 143, 381–386.

Hillel, D., 1975. Evaporation from bare soil under steady and diur-nally fluctuating evaporativity. Soil Sci. 120, 230–237.

Hillel, D., 1980. Applications of Soil Physics. Academic Press, NewYork, London, p. 385.

Hillel, D., 1998. Environmental Soil Physics. Academic Press, SanDiego, CA, p. 771.

Huwe, B., van der Ploeg, R.R., 1990. Modelle zur Simulationlicherniv.

I ama,Sci.

I me-urveric.

J evap-soil

J ationss,

J ationty of

K vap-132,

lack, T.A., Gardner, W.R., Thurtell, G.W., 1969. The predictioevaporation, drainage and soil water storage for a bare soiSci. Soc. Am. Proc. 33, 655–669.

oast, C.W., 1986. Evaporation from bare soil measured withspatial resolution. In: Klute A. (Ed.), Methods of Soil AnalysPart 1, Physical and Mineralogical Methods. Agr. Monog. No889–900.

oast, C.W., Robertson, T.M., 1982. A ‘microlysimeter’ methoddetermining evaporation from bare soil: description and labtory evaluation. Soil Sci. Soc. Am. J. 46, 689–696.

risson, N., Itier, B., L’Hotel, J.C., Lorendeau, J.Y., 1998. Paraterisation of Shuttleworth–Wallace model to estimate daily mimum transpiration for use in crop models. Ecol. Model. 1159–169.

risson, N., Perrier, A., 1991. A semi-empirical model of bareevaporation for crop simulation models. Water Resour. Re(5), 719–727.

rown, R.W., Oosterhuis, D.M., 1992. Measuring plantsoil–water potentials with thermocouple psychrometers: sconcerns. Agron. J. 84, 78–86.

ampbell, G.S., 1985. Soil Physics with Basic: Transport Mofor Soil–Plant systems. Elsevier, Amsterdam, Oxford, p. 15

des Stickstoffhaushaltes von Standorten mit unterschiedlandwirtschaftlicher Nutzung. Eigenv. Inst. Wasserbau, UStuttgart. Heft 69.

dso, S.B., Reginato, R.J., Jackson, R.D., Kimball, B.A., NakayF.S., 1974. The three stages of drying of a field soil. SoilSoc. Am. Proc. 38, 831–837.

noue, M., Yano, T., Chikushi, J., 1984. Optimum values of paraters in equations for estimating the soil water characteristic cand relative hydraulic conductivity of sandy soils. J. Fac. Ag(Tottori Univ., Japan) 19, 24–34.

ackson, R.D., Idso, S.B., Reginato, R.J., 1976. Calculation oforation rates during the transition from energy limiting tophases using albedo data. Water Resour. Res. 12, 23–26.

alota, S.K., Prihar, S.S., 1998. Reducing Soil–Water Evaporwith Tillage and Straw Mulching. Iowa State University PreAmes (ISBN 0-8138-2857-0), p. 142.

ensen, M.E. (Ed.), 1974. Consumptive use of water and irrigwater requirements. A report published by American SocieCivil Engineers, New York, p. 215.

atul, G.G., Parlange, M.B., 1992. Estimation of bare soil eoration using skin temperature measurements. J. Hydrol.91–106.


Kijne, J.W., 1974. Determining evapotranspiration. In: Drainageprinciples and applications-III, survey and investigations. Inter.Ins. Land Reclamation and Improvement (ILRI), WageningenPublications, III (16) 53–111 (Chapter 19).

Kirby, J.M., Ringrose-Voase, A.J., 2000. Drying of some Philippineand Indonesian puddle rice soils following surface drainage: nu-merical analysis using a swelling soil flow model. Soil TillageRes. 57, 13–30.

Kite, G., Droogers, P., Murray-Rust, H., Voogt, de K., 2001. Model-ing Scenarios for Water Allocation in the Gediz Basin, Turkey.Research Report 50, Colombo, Sri Lanka: International WaterManagement Institute.

Koorevaar, P., Menelik, G., Dirksen, C., 1983. Elements of SoilPhysics. Elsevier, Amsterdam, Oxford, p. 228.

Liu, S., Riekerk, H., Gholz, H.L., 1998. Simulation of evapotranspi-ration from Florida pine flatwoods. Ecol. Model. 114, 19–34.

Malik, R.S., Anlauf, R., Richter, J., 1992. A simple model for pre-dicting evaporation from bare soil. Z. Pflanzenern. Bodenk. 155,293–299.

Mihailovic, D.T., Rajkovic, B., Dekic, L., Pielke, R.A., Lee, T.J., Ye,Z., 1995. The validation of various schemes for parameterizingevaporation from bare soil for use in meteorological models: anumerical study using in situ data. Boundary-Layer Meteorol.76, 259–289.

Mihailovic, D.T., Ruml, M., 1996. Design of land-air parameter-ization scheme (LAPS) for modelling boundary layer surfaceprocesses. Meteorol. Atmos. Phys. 58, 65–81.

Monteith, J.L., 1965. Evaporation and environment. In: Proceedingsof the 19th Symposium of the Society on Experimental Biology.State and Movement of Water in Living Organisms. Swansea,1964. Cambridge University Press, Cambridge, pp. 205–234.

Monteith, J.L., 1981. Evaporation and surface temperature. Quart. J.Royal Meteorol. Soc. 107, 1–27.

Monteith J.L., Unsworth, M.H., 1990. Principles of EnvironmentalPhysics, second ed. Edward Arnold, London/New York, p. 291.

P tiondrol.

P urfacenthly

R irationatial

scales using published parameter values. Agric. Water Manage31, 17–34.

Ritchie, J.T., 1972. Model for predicting evaporation from a row cropwith incomplete cover. Water Resour. Res. 8, 1204–1213.

Ritchie, J.T., Kissel, D.E., Burnett, E., 1972. Water movement inundisturbed swelling clay soil. Soil Sci. Soc. Am. Proc. 36,874–879.

Saunders, I.R., Bailey, W.G., Bowers, J.D., 1997. Evaporationregimes and evaporation modelling in an alpine tundra environ-ment. J. Hydrol. 195, 99–113.

Staple, W.J., 1974. Modified Penman equation to provide the upperboundary condition in computing evaporation from bare soil. SoilSci. Soc. Am. Proc. 38, 837–839.

Van Dam, J.C., Huygen, J., Wesseling, J.G., Feddes, R.A., Kabat, P.,van Walsum, P.E.V., Groendijk, P., van Diepen, C.A., 1997. The-ory of SWAP version 2.0. Simulation of water flow, solute trans-port and plant growth in the Soil–Water-Atmosphere-Plant envi-ronment. Technical Document 45, DLO Winand Staring Centre,Report 71, Department of Water Resources, Agricultural Univer-sity, Wageningen, p. 167.

Van Keulen, H., Hillel, D., 1974. A simulation study of the dryingfront phenomenon. Soil Sci. 118, 270–273.

Wallace, J.S., Holwill, C.J., 1997. Soil evaporation from tiger-bushin south-west Niger. J. Hydrol. 188–189, 426–442.

Wallace, J.S., Jackson, N.A., Ong, C.K., 1999. Modelling soil evap-oration in an agroforestry system in Kenya. Agric. For. Meteorol.94, 189–202.

Warrick, A.W., Biggar, J.W, Nielsen, D.R., 1971. Simultaneous so-lute and water transfer for an unsaturated soil. Water Resour. Res.7, 1216–1225.

Yang, S.L., Yano, T., Aydin, M., Kitamura, Y., Takeuchi, S., 2002.Short term effects of saline irrigation on evapotranspirationfrom lysimeter-grown citrus trees. Agric. Water Manage. 56 (2),131–141.

Yang, S.L., Aydin, M., Yano, T., Li, X., 2003. Evapotranspiration of49.

Z dro-ation,

Z rtiesRes.

arlange, M.B., Katul, G.G., 1992. Estimation of the diurnal variaof potential evaporation from a wet bare soil surface. J. Hy132, 71–89.

riestley, C.H.B., Taylor, R.J., 1972. On the assessment on sheat flux and evaporation using large-scale parameters. MoWeather Rev. 100, 81–92.

adersma, S., de Ridder, N., 1996. Computed evapotranspof annual and perennial crops at different temporal and sp

orange trees in greenhouse lysimeters. Irrig. Sci. 21, 145–1hang, L., Dawes, W.R., Hatton, T.J., 1996. Modelling hy

logic processes using a biophysically based model-applicof WAVES to FIFE and HAPEX-MOBILHY. J. Hydrol. 185147–169.

hu, J., Mohanty, B.P., 2002. Upscaling of soil hydraulic propefor steady state evaporation and infiltration. Water Resour.38 (9) (17)-1-13.

test of a simple model for estimating evaporation from bare soils in different environments

Documents