measurement and simulation of herbicide transport in macroporous soils

Pestic. Sci. 1998, 52, 241È250

Measurement and Simulation of HerbicideTransport in Macroporous SoilsC. Florian Stange,1* Bernd Diekkru� ger1 & Henning Nordmeyer21 Technische Universita� t, Braunschweig, Germany2 Biologische Bundesanstalt fu� r Land-und Forstwirtschaft, Braunschweig, Germany

(Received 16 January 1997 ; revised version received 11 September 1997 ; accepted 28 October 1997)

Abstract : Lysimeter experiments were carried out to study pesticide transportthrough macroporous soils. In order to di†erentiate between the e†ects of soilstructure and chemical behaviour, the leaching experiments were conductedusing disturbed and undisturbed soil samples. Two herbicides with di†erent sorp-tion behaviours, and bromide as tracer were applied. The results were used tovalidate a dynamic simulation model which considers bypass Ñow in macropores.The simulation results show that the model is able to reproduce the soil suctionwithin the soil as well as the spatial distribution of bromide and the herbicides.The continuity of the macropores is most important for the efficiency of bypassÑow. The results indicate that cultivation practices like ploughing signiÐcantlyinÑuence the temporal and spatial distribution of the macropores. 1998 SCI.(

Pestic. Sci., 52, 241È250 (1998)

Key words : herbicide leaching ; macropores ; bypass Ñow; lysimeter experiments ;simulation model

1 INTRODUCTION

Groundwater is an important source of drinking water,supplying 75% or more of the drinking water inGermany. During recent years, contamination ofgroundwater by pesticides has generated environmentalproblems.

When pesticides reach the soil, they may undergomicrobial and/or chemical degradation, photo-decomposition, volatilisation, plant uptake and adsorp-tion. Furthermore, losses can occur by surface run-o†and leaching through the soil proÐle. Water that inÐl-trates through the soil may carry pesticides through andbelow the root zone, possibly reaching groundwater.Many pesticides with di†erent physical-chemicalproperties have been found in groundwater.1 In manycases the reasons for such contamination are not clear.In the past, the description of Ñow processes in the soilproÐle considered only matrix Ñow. However, the

* To whom correspondence should be addressed at : Fraun-hofer Institut Atmospha� rische Umweltforschung, D-82467Garmisch-Partenkirchen, Germany

results of transport experiments can often not beexplained solely by the convective-dispersive theory.2,3Only the presence of preferential pathways can explainfast vertical transport.

A number of Ðeld and laboratory studies have shownthat preferential Ñow is an important mechanism in themovement of pollutants to groundwater.2,4 PreferentialÑow can be deÐned as Ñow through macropores (i.e.earthworm and root channels, shrinking cracks), trans-port through zones with high conductivity5 and Ðnger-ing as a result of Ñuid instability. In particular, macro-pore Ñow takes place only in a small part of the soil6and can be initiated under a number of di†erent condi-tions. The transport of water and pesticides is a†ectedalso by rainfall intensity.7 Soil tillage is also an impor-tant factor for macropore Ñow. The phenomenonof macropore Ñow is summarised by Van Genuchten etal.8

In this study, soil column experiments were carriedout to examine the transport of water, bromide andselected herbicides in a well-structured, water-unsaturated loess soil. The data were used for the vali-dation of a simulation model.

2411998 SCI. Pestic. Sci. 0031-613X/98/$17.50. Printed in Great Britain(

242 C. Florian Stange, Bernd Diekkru� ger, Henning Nordmeyer

2 METHODS AND MATERIALS

The studies were carried out using small-scale lysi-meters with loess soil (soil type : Gleyic Luvisol, siltloam). The soil was taken from an agricultural site(Neuenkirchen) located in Lower Saxony in the regionof Braunschweig, Germany. Details of the investigatedsite are given by McVoy et al.9 The most important soilparameters are shown in Table 1. It is a highly struc-tured soil with continuous earthworm channels down toa depth of 1 m. Undisturbed soil samples were taken insoil column cylinders (diameter 0É30 m, length 0É85 m).

Five undisturbed (S1ÈS5) and one reference (S6) soilcolumns were investigated. The reference soil (S6) corewas prepared by sieving (2 mm) and repacking the soilof a disturbed soil sample. The resulting bulk density inthe subsoil (0É30È0É85 m) was slightly lower than that ofthe undisturbed soil cores (S6 \ 1É34 g cm~3, undis-turbed soil mean 1É47 (^0É03) g cm~3).

Soil sampling was carried out using hydraulic pres-sure equipment.10 After screwing two anchors into thesoil, they were connected with a horizontal anchorbeam. The pressing of the cylinders into the soil wasdone by a hydraulic jack placed between the cylinderand the anchor beam.

After sampling, the soil monoliths were installed inthe experimental set-up (Fig. 1). A sprinkling device wasset on top of the column. It ensured a good spatial dis-tribution of the irrigation water by supplying it from 89hypodermic needles. The irrigation cycles were set by atime-limit relay which controlled a pump and a mag-netic valve. This equipment allowed quasi-continuouswatering at low Ñow rates.

The bottom of the soil column was installed on aspecial plate, the outlet from which ended in a low-pressure chamber. Permanent suction of [20 hPa wasapplied during all experiments to avoid soil water satu-ration in the lowest soil layer. A fraction collector wasused to sample the percolating water. To prevent theintake of air into the soil column, an air-impermeablelayer was put between soil and the plate. This micro-porous diaphragm acted in a way similar to a ceramicplate.

Fig. 1. Experimental set-up for testing leaching behaviour ofherbicides in unsaturated, structured soil.

Tensiometers with electronic pressure transducerswere installed horizontally (0É25 m and 0É75 m abovethe bottom). Measurements were made continuouslyand stored every two minutes on a personal computer(A/D interface card). The experiments were carried outin an air-conditioned chamber at controlled tem-perature from 7 to 14¡C.

Three irrigations (60 min.) were applied with a 48-hinterval between each. The rainfall intensity was25 mm h~1. This value represented the intensity of aheavy rainfall. The percolate collection was done at36-min intervals. Before starting leaching experiments,all soil samples were irrigated (1 mm h~1 until outÑowoccurred) to ensure the same initial water content. Afterthat, the lysimeters drained for two weeks to reach Ðeldcapacity. Two herbicides, chlorotoluron (CT) and metha-benzthiazuron (MBT), were applied at 10 kg ha~1 and14 kg ha~1. The solute was pipetted in a Ðne grid togive a good spatial distribution.

After the leaching experiments, the soil was removedfrom the bottom of the columns in 5-cm segments.Every 5 cm the occurrence of macropores (pores with adiameter [1 mm) was visually examined and trans-ferred over a transparent foil. The macropores weredivided into three classes (from 1 to 4 mm, from 4 to8 mm and larger than 8 mm).

TABLE 1Soil Characteristics

T opsoil Subsoil

Soil depth (m) 0É00È0É30 0É30È0É50 0É50È0É75 0É75È0É85Soil texture (%)

Sand 2É1 3É7 2É7 1É9Silt 78É8 70É4 78É7 82É1Clay 19É1 25É7 18É6 16É0

pH value (CaCl2) 6É9 7É1 7É5 7É5Organic carbon (%) 1É3 0É5 0É3 0É2Bulk density (g cm~3) 1É38 1É44 (0É30È0É85 m)

Simulation of herbicide transport in macroporous soils 243

TABLE 2Estimated Parameters of the Water Retention Curve of Van Genuchten.12 The standard deviations of these estimates

are shown in brackets.

Sample O1 O2 O3 O4 O5 O6 Mean Median

hs (cm3 cm~3) 41É256 45É583 46É701 44É718 50É501 46É449 45É868 46É016(0É652) (0É836) (0É617) (1É288) (2É199) (0É831)

a (hPa~1) 0É0064 0É0093 0É0146 0É0228 0É0450 0É0124 0É0184 0É0135(0É0019) (0É0029) (0É0030) (0É0105) (0É0287) (0É0036)

n ([) 1É243 1É242 1É227 1É196 1É181 1É230 1É220 1É229(0É027) (0É029) (0É017) (0É029) (0É034) (0É025)

R2 0É9921 0É9913 0É9962 0É9841 0É9742 0É9928

Sample U1 U2 U3 U4 U5 U6 Mean Median

hs (cm3 cm~3) 44É445 43É296 41É596 42É055 41É136 43É840 42É728 42É676(0É454) (0É602) (0É740) (0É637) (0É787) (1É215)

a (hPa~1) 0É2229 0É0557 0É0325 0É0287 0É0328 0É0651 0É0730 0É0443(0É0326) (0É0122) (0É0098) (0É0075) (0É0109) (0É0291)

n ([) 1É134 1É155 1É164 1É167 1É158 1É144 1É154 1É157(0É004) (0É009) (0É014) (0É127) (0É015) (0É016)

R2 0É9990 0É9970 0É9938 0É9953 0É9926 0É9885

Herbicides were extracted and enriched from 10 mlpercolate by using a Baker-10 Extraction System (C18-

phase extraction, SPE). After elution, they weresolidanalysed by high performance liquid chromatography(HPLC) with UV-detector. The quantiÐcation was donewith regard to peak height and internal standard. Therecovery rate for CT in the subsoil (0É30È0É85 m) as wellas in the topsoil (0È0É30 m) was more than 110% for allfour concentrations (see also Table 2). For MBT therecovery increased with concentration and also from thetop to the subsoil in the range from 88 to 98%.

Additionally a tracer (bromide) was applied at200 kg ha~1 with the irrigation water. In the Ðrst andthe second columns (S1 and S2) 150 kg ha~1 bromidewas applied. The bromide was analysed by ionexchange chromatography after Ðltering the watersamples (membrane Ðlter, 0É45 mm). In order to charac-terise the soil physical properties, six samples of thetopsoil (O1ÈO6) and six samples of the subsoil (U1ÈU6)were taken from the Ðeld to measure retention curves inthe laboratory. The results of the experiments wereÐtted to the retention function of Van Genuchten bymeans of the statistic program Statgraphics. The esti-

mated values of the saturated moisture content andhsthe empirical constants a and n are shown in Table 2.

The values of saturated hydraulic conductivity are inthe range typical for a structured soil. The results arefrom 13É2 to 308É2 cm day~1 in the topsoil and 16É4 to653É3 cm day~1 in the subsoil (Table 3). In many cases(e.g. Hartge and Horn11) it has been shown that thesaturated conductivity follows log normal distribution.The mean of the logarithms of the saturated conductivi-ty values was therefore used for the simulation (topsoil :34É8 cm day~1, subsoil : 65É1 cm day~1). The unsatu-rated conductivity was derived from the estimatedvalues for a and n according to the formula of VanGenuchten.12

3 SIMULATION MODEL

In this study, the model SIMULAT13,14 was employed.SIMULAT is a system for simulating water and solutetransport, soil temperature, dynamics of nitrogen andsulphur, and the dynamics of pesticides and theirmetabolites. For calculating the fate of pesticides,SIMULAT includes a model bank for degradation,

TABLE 3Saturated Hydraulic Conductivity

Soil sample O1 O2 O3 O4 O5 O6 L og-mean Median

Conductivity (cm day~1) 18É9 308É2 78É0 20É7 14É3 13É2 34É8 19É8

Soil sample U1 U2 U3 U4 U5 U6 L og-mean Median

Conductivity (cm day~1) 167É0 30É4 653É3 61É1 16É4 22É8 65É1 45É8


Fig. 2. Distribution and size of the macropores in the undisturbed soil columns S1ÈS5.

sorption and dependencies on environmental variables.It considers di†erent sorption isotherms, biotic(metabolic and cometabolic) and abiotic degradation,and di†erent approaches for calculating temperatureand moisture dependencies. In addition to the one-dimensional water and solute transport, evapotranspi-ration, interception and plant growth is simulated.Because in this study only water and solute transport instructured porous media are investigated the modeldescription given here is limited to these processes.

In the classical RichardsÏ equation it is assumed thatwater transport in porous media can be described by amacroscopic approach. The relationship between soiltension and water content is often described by a uni-modal pore-size distribution. This approach fails in allsituations where microscopic structures like fractures,Ðssures, aggregate pores and macropores inÑuence thewater Ñuxes signiÐcantly, resulting in transport behav-iour not explainable by the one-domain approach(preferential Ñow). Therefore, in the past, models forwater transport in structured porous media have beendeveloped which describe the water transport in a two-domain approach. While most of the models concen-trate on matrix and macropore Ñow15,16 only a fewapproaches assume two (or more) capillary Ñowdomains.17,18

If macropores exist, the water Ñow cannot bedescribed by the capillary two-domain concept. In thiscase RichardsÏ equation for one-dimensional water Ñow

in the soil matrix (mi)

LhmiLt

\ LLzAKmi(tmi)

ALtmiLz

[ 1BB

[ Sw (1)

is coupled to the water Ñow in the macropore system(ma)

LhmaLt

\ LqmaLz

] Sw (2)

by the exchange term Sw

Sw \ Klattmi [ tma

*x(3)

in which h is the water content, t the soil suction,K \ K(t) the unsaturated hydraulic conductivity, Klatthe lateral saturated hydraulic conductivity, t the timeand z the vertical coordinate. Assuming Ðlm Ñow in themacropores *x of eqn (3) is given as the half-(tma \ 0),width of the matrix pore system. According to Germannand Beven15 the gravity Ñow of water in the macro-pores can be described by the following relationshipbetween water Ñux, and the macropore moistureqma ,content hm

qma\ Ks maA hmahs ma

Bama(4)


Percolate and breakthrough curves of bromide and chlorotoluron at the undisturbed soil column S4.Fig. 3. (K) (=) (|)

in which is an empirical constant and theama hs mamacropore volume. Water Ñux into the macroporesystem occurs when the rainfall rate r minus the(Ima)interception f is larger than the inÐltration capacity ofthe soil matrix In this case the soil suction of the(Imi).upper computational layer n is Ðxed to an e†ectivemaximum soil suction and the inÐltration rate canteffbe calculated from DarcyÏs law. According to theseassumptions the upper boundary condition of thismodel is given as

Imi\ r [ fIma \ 0

Ht

n\ teff

Imi \ qinf ilIma \ r [ f [ qinf il

Ht

n\ teff (5)

qinf il\ K(t)Ateff [ t

n~1dz

] 1B

Only few models consider that is not a constanths mabut may vary with soil structural changes due to swell-ing and shrinking.16,19 In SIMULAT it is assumed thatthe soil is rigid.

Solute transport in the soil matrix is computed usingthe classical convection dispersion equation (CDE)

LLt

(hc] os) \ LLzAhDh

LcLz

[ qcB

] Qs] Qma (6)

in which c is the solute concentration, S the sorbed con-centration, o the bulk density, and the hydrody-Dhnamic dispersion.20 The term summarises all sourcesQsand sinks of the substances, i.e. all processes creatingand consuming the substance and the exchangeQmaterm with the macropore system. The solute transportin the macropores as well as the solute exchangebetween soil matrix and macropores is assumed to bepure convection without any dispersion. Due to the

typical Ñow velocities in macropores, sorption and deg-radation was neglected in the macropore model. At theupper boundary it was assumed that the concentrationof the inÐltrating water was in equilibrium with theImaconcentration of the upper computational layer of thesoil matrix.

4 EXPERIMENTAL RESULTS

4.1 Visual examination of macropores

In all soil columns S1ÈS5, a distinctive macroporesystem was observed (Fig. 2). Clear di†erences betweenthe columns S1ÈS5 were found in the topsoil (0È0É30 m). On average, the number of macropores in thesubsoil was higher than in the topsoil. From depth0É45È0É50 m the number of macropores decreased withdepth in all columns except column S2. Also in S2 thenumber decreased from depth 0É6 m to 0É75 m, but thehighest quantity was in depth 0É8 and 0É85 m. Onaverage, the number of macropores in the topsoil was106 pores m~2, in the subsoil 161 pores m~2. Particu-larly in columns 4 and 5, only few macropores werefound in the topsoil. This is a result of soil tillage. Byploughing, macropores are cut o† and the soil is mixedin the plough-layer causing a change of water conduc-tivity.21

4.2 Herbicide leaching

Figure 3 shows the percolate and the breakthroughcurves of bromide and chlorotoluron (CT). There arehigh concentration peaks of bromide and chlorotoluronafter the irrigation. This result is typical for macroporeÑow.22 Bromide was detected in columns S2ÈS5 (Table4). Because the breakthrough curves are all similar tothose for column S4, we decided to show only onecolumn as example. We found CT only in the percolate


TABLE 4Bromide, Chlorotoluron and Methabenzthiazuron Residues in Percolating Water and Soil Studies Columns S1ÈS6

L ysimeter S1 S2 S3 S4 S5 S6

Irrigation water (litres) 3É963 3É963 4É954 4É954 4É954 4É954Percolate (% of irrigation water) 60É6 66É2 58É2 55É9 71É5 81É9Bromide in percolate (% of applic. rate) n.d. \0É1 4É9 2É4 0É2 n.d.CT in percolate (% of applic. rate) n.d. n.d. 0É1 0É1 n.d. n.d.MBT in percolate (% of applic. rate) n.d. n.d. \0É1 n.d. n.d. n.d.Bromide in soil (% of applic. rate) 59É3 64É3 84É3 67É2 79É9 86É9CT in soil (% of applic. rate) 64É9 64É0 74É4 77É2 78É3 75É2MBT in soil (% of applic. rate) 74É8 80É6 83É8 85É6 85É8 83É6

Recovery rate (% of application rate)

Bromide 59É3 64É4 89É2 69É6 71É7 86É9CT 64É9 64É0 74É5 77É3 78É3 75É2MBT 74É8 80É6 83É8 85É6 85É8 83É6

Substance share in the topsoil (0È0É30 m) (% of all the substance in the soil)

Bromide 86É3 76É9 59É6 57É0 68É6 62É1CT [99É9 100É0 95É5 98É5 99É9 100É0MBT 100É0 100É0 98É7 100É0 100É0 100É0

Max. detectable moving depth in soil (m)

Bromide 0É85 0É85 0É85 0É85 0É85 0É65CT 0É45 0É25 0É65 0É65 0É45 0É15MBT 0É25 0É20 0É45 0É20 0É30 0É05

n.d.Ènot detectable.

of columns S3 and S4, methabenzthiazuron (MBT) onlyin the percolate of column S3.

Table 4 summarises the results of the soil residueanalysis of CT, MBT and bromide for the six lysimeterexperiments. The recovery rates for bromide in columnsS1, S2, S4 and S5 are rather low. It is possible that, as aresult of the high silt and clay content, a part ofbromide adsorbed to the soil. The substance share inthe topsoil and the maximal detectable moving depthdemonstrates the distinction of the water and mattertransport in the Ðve undisturbed soil columns S1ÈS5.However, the di†erence from the disturbed soil columnS6 is marked. Both bromide and herbicide removeddeeper into the undisturbed soil columns than into thedisturbed soil. To homogenise the samples before herbi-cide analysis the soil must be dried for two days, so

one must consider possible decomposition of CT andMBT when looking at the recovery rate.

5 SIMULATION RESULTS

5.1 Fundamentals

As input parameters, the results of the soil examinationswere used as far as possible. The sorption parameterswere taken from literature23,24 (see Table 5). It shouldbe taken into account that these values are not neces-sarily transferable.

For the calculation of decomposition in SIMULAT, amodel is implemented which is described in detail byRichter et al.20 To describe the dependency of degrada-tion on moisture, we applied the non-linear relationshipof Richter et al.20 The parameters are given in Table 5.

TABLE 5Adsorption and Decomposition Parameters of Chlorotoluron (CT) and Meth-

abenzthiazuron (MBT)a

CT MBT

Ads. parameter topsoil Koc \ 158É723 Kfr\ 6É6822 n \ 1É4722Ads. parameter subsoil Kfr\ 1É9322 n \ 1É2722

kopt (day~1)18 0É0071 0É011hopt (cm3 cm~3)18 0É69 0É553a ( )18 1É514 2É025

a Sorption function : linear sorption, and n, Freundlich.Koc Kfr


Because the temperature was constant in this investiga-tion no temperature dependency was considered. As nodata concerning the macropore model were available,macropore parameters were Ðtted using the experimen-tal results.

First, the water transport in a homogeneous mediumwas simulated. The nearly homogeneous, packedcolumn, S6, serves as a reference for these simulations.The solute transport was calculated using the watertransport and sorption parameters of the herbicideschlorotoluron and methabenzthiazuron obtained fromthe literature. In the next step, the Ñow in the undis-turbed columns with macropores was simulated. Forthese calculations, the input data and the knowledgegained from the simulation of the reference column S6were used. The aim of adding the macropore part to theprogram was to achieve a good simulation of the mea-sured results in the Ðve undisturbed columns.

5.2 Simulation of the water and solute transport in thepacked column S6

Because the soil structure was destroyed, the high inten-sity of irrigation (3 ] 25 mm) resulted in the develop-ment of a crust at the soil surface. The wateraccumulated at the surface. The inÑuence of a surfacecrust on water transport and the possibility of its simu-lation are described by Diekkru� ger & Bork.25

For the packed column, the conductivity and theretention curve of the undisturbed soil samples wereused as input data without further calibration. With thisparameter vector, the water transport and thus thebromide distribution for the packed column could bewell simulated by means of the convection-dispersionequation using the model SIMULAT. The parameter

Measured and (È) simulated bromide movementFig. 5. (=)in the repacked soil column S6 after 75 mm of irrigation. Thesimulated bromide concentrations refer to the bromide resi-

dues found in this column.

vector well describes the water Ñow in the soil columnand the course of water tension in 25 cm depth, as isshown in Fig. 4. Figure 5 shows the measured and thesimulated bromide concentrations for the packedcolumn.

A better agreement could be obtained by using a two-domain model, because the water within the soil aggre-gates does not take part in the water transport. Hutson& Wagenet26 developed a multi-region model, a two-domain model extended by considering a region of pref-erential Ñow. It is impossible to estimate only with soilcharacteristic data the distribution of water between themobile and the immobile phase. Therefore, the par-titioning coefficient could be obtained by Ðtting themodel to the bromide data. So far, the model

Fig. 4. Simulated and measured water tension at a depth of 0É25 m and 0É75 m in the packed column (S6). (È) Simulated(>) (+)and measured temporal course of the percolation rate.(=)


Fig. 6. Simulated and measured (È; CT and (È È ; i)>)MBT concentrations in the repacked soil column S6.

SIMULAT does not consider immobile soil water. Forthe given soil, the good Ðt to the measured data sug-gests that this is not necessary.

With this parameter vector, and the sorption anddegradation parameters (Table 5), the transport of CTand MBT was simulated. The result of the simulation isshown in Fig. 6. As for the simulation of the watertransport, the parameter vector gives good Ðts. The dif-

ferent sorption behaviour of the herbicide results in dif-ferent leaching depths.

5.3 Simulation of water and solute transport in theundisturbed soil

For the input data of the macropore Ñow (hs ma , Ks ma ,particular features that were noticed whenama , Klat)

visually recording the columns, such as the soil crack incolumn 3 or the smaller number of macropores in thebottom part of the plough horizon, were considered.However, the observed number and size of macroporesgave no correlation with the water and solute transport.Therefore, the parameters of macropore Ñow wereobtained by Ðtting the simulated bromide distributionto the measured distribution. The parameter values ofthe simulations are shown in Table 6.

For the parameter of the macropore Ñow model asensitivity analysis was carried out. It turns out, that themaximum depth of the macropore Ñow depends on theratio between the saturated hydraulic conductivity intothe macropore system and the lateral saturated(Ks ma)hydraulic conductivity the velocity mainly on(Klat),

and the macropore volumeKs ma (hs ma).The results of the simulations are shown for column 4

in Figs 7 and 8. As these Ðgures show, the distribution

TABLE 6Macropore Model Parameters of the Undisturbed Column S4

T opsoil 1 T opsoil 2 Subsoil 3 Subsoil 2(0È0É15 m) (0É15È0É30 m) (0É30È0É60 m) (0É60È0É85 m)

hs ma (cm3 cm~3) 0É2 0É1 0É05 0É01Ks ma (cm day~1) 160 80 80 10ama ( È ) 1É2 1É2 1É2 1É2Klat (cm Pa~1 day~1) 0É3 0É4 0É3 0É3

Fig. 7. (È) Simulated and measured water tension in the undisturbed column S4 at a depth of 0É25 m and 0É75 m and(=) (>) (+)temporal course of the percolation rate.


Fig. 8. Comparison of the simulated and measured spatialdistribution of (É É É ; bromide, (È; CT and (È È ; i)=) >)MBT in column S4. The simulated bromide concentrations

refer to the bromide residues found in this column.

of bromide, CT and MBT measured in the lysimeterstudies could be well simulated using the modelSIMULAT. The sudden decrease of the simulated her-bicide concentration from topsoil (0È0É30 m) to thesubsoil (0É30È0É85 m) can be explained on the basis ofthe model by di†erent values.Kd

The course of the water retention in the columns atdepths of 0É25 m and 0É75 m is also well described bythe model. the minor deviations can be explained bysmall-scale variability. In particular, macropore Ñowand the resulting lateral inÐltration out of the macro-pores into the soil matrix is signiÐcantly inÑuenced bythe small-scale spatial variability of soil properties.

6 DISCUSSION

Visual examination showed that none of the Ðve lysim-eters had continuous macropores down to the bottomof the soil column. Therefore, the breakthrough curvesare the results of lateral inÐltration from the macro-pores into the soil matrix and afterwards leachingthrough the soil matrix. Because only small amounts ofthe herbicides were washed out during this experiment,the spatial distribution within the soil column is moreimportant than the breakthrough curves. Because deg-radation and sorption are signiÐcantly reduced belowthe ploughing layer, a large part of the pesticide cross-ing this layer may contaminate the groundwater.27

The concentration proÐles show the continuity of themacropore system is disrupted at ploughing level. Thisleads to lateral inÐltration from the macropores into thesoil matrix, resulting in an increased concentration. Ifsaturated conditions exist, the soil water may enter themacropore system again as is illustrated in Fig. 9. Howmany macropores are active in the subsoil depends onsmall-scale spatial variability.

The simulation results given in Fig. 8 show that themean concentration proÐle is well described, but thetypical e†ect of the ploughing layer is not reproduced.

Fig. 9. Proposed water Ñow in a column at the bottom of theplough horizon.

This is due to the limitations of the model used in thisstudy. The actual version of SIMULAT is able to con-sider one continuous macropore system. In order to beable to consider the e†ect of the ploughing layer, it isessential to complete the model system by another inde-pendent macropore system and to allow lateral exÐltra-tion and inÐltration at the same depth.

ACKNOWLEDGEMENT

This study has been performed within the framework ofthe collaborative research program “Water and MatterDynamics in Agro-ecosystemsÏ which is Ðnanced by theGerman Research Foundation (Deutsche Forschungs-gemeinschaft).

REFERENCES

1. Ritter, W. F., Pesticide contamination of ground water inthe United StaesÈA review. J. Environ. Sci. Health, 25(1990) 1È29.

2. Beven, K. & Germann, P., Macropores and water Ñow insoils. W ater Resour. Res., 18 (1982) 1311È25.

3. Isensee, A. R., Nash, R. G. & Helling, C. S., E†ect of con-ventional vs. no-tillage on herbicide leaching to shallowgroundwater. J. Environ. Qual., 19 (1990) 434È40.

4. Czapar, G. F., Horton, R. & Fawcett, R. S., Herbicide andtracer movement in soil columns containing an artiÐcialmacropore. J. Environ. Qual., 21 (1992) 110È5.

5. Stock, A., Untersuchung der ra� umlichen Variabilita� t desSto†transportes in der wasserunge sa� ttigten Bodenzone.

und Umweltforschung Braunschweig,L andschaftso� kologie23 (1995) 170.

6. Watson, K. W. & Luxmoore, R. J., Estimating macro-porosity in a forested watershed by use of a tension inÐl-trometer. Soil Sci. Soc. Am. J., 50 (1986) 578È82.


7. Edwards, W. M., Shipitalo, M. J., Dick, W. A. & Owens,L. B., Rainfall intensity a†ects transport of water andchemicals through macropores in no-till soil. Soil Sci. Soc.Am. J., 56 (1992) 52È8.

8. Van Genuchten, M. T. Ralston, D. E. & Germann, P. F.(eds), Transport of water and solutes in macropores. Geo-derma, 46 (1990) 1È297.

9. McVoy, C. W., Kersebaum, K. C., Arning, M., Kleeberg,P., Othmer, H. & Schro� der, U., A data set from northGermany for the validation of agroecosystem models :documentation and evaluation. Ecological Modelling,81.1-3 (1995) 83È95.

10. Nordmeyer, H. & Aderhold, D., Leaching of pesticides insoil macropores as a possibility for ground and surfacewater contamination. Nachrichtenbl. Deut. PÑanzen-schutzd., 47 (1995) 137È43.

11. Hartge, K. H. & Horn, R., Die physikalische Untersuchungvon Enke Verlag, Stuttgart, 1989, 175 pp.Bo� den.

12. Van Genuchten, M. T., A closed-form equation for pre-dicting the hydraulic conductivity of unsaturated soils.Soil Sci. Soc. Am. J., 44 (1980) 892È8.

13. Diekkru� ger, B. & Arning, M., Simulation of water Ñuxesusing di†erent methods for estimating soil parameters.Ecological Modelling, 81.1-3 (1995) 83È95.

14. Diekkru� ger, B., No� rtersheuser, P. & Richter, O., Modelingpesticide dynamics of a loam site using HERBSIM andSIMULAT. Ecological Modelling, 81.1-3 (1995) 111È9.

15. German, P. & Beven, K., Kinematic wave approximationto inÐltration into soils with sorbing macropores. W aterResour. Res., 21 (1985) 990È6.

16. Jarvis, N., MACROÈA model of water movement andsolute transport in macroporous soils. Reports and Disser-tations 9 (1991) 58 pp. Department of Soil Sciences.Swedish University of Agriculture Sciences, Uppsala,Sweden.

17. Othmer, H., Diekkru� ger, B. & Kutilek, M., Bimodal poro-sity and unsaturated hydraulic conductivity. Soil Sci., 152(1991) 139È50.

18. Gerke, H. & van Genuchten, M. T., A dual porositymodel for simulating the preferential movement of water

and solutes in structured porous media. W ater Resour.Res., 29 (1993) 305È19.

19. Bronswijk, J. J. B., A general approach to incorporateswelling and shrinkage processes in soil water transportsimulation models. Modeling of Geo-Biosphere Processes,1 (1992) 253È70.

20. Richter, O., Diekkru� ger, B. & No� rtersheuser, P., Environ-mental fate modelling of pesticides : from the laboratory tothe Ðeld scale. Verlag Chemie, Weinheim (1996) p. 281.

21. Aderhold, D. & Nordmeyer, H., Leaching of herbicides insoil macropores as a possible reason for groundwater con-tamination. Pesticide movement to water. BCPC Mono-graph 62 (1995) 217È22.

22. Meyer-Windel, S. & Lennartz, B., Herbicide transport indi†erently structured soil horizons under constant andtransient Ñow conditions. Pesticide movement to water.BCPC Monograph 62 (1995) 99È104.

23. Brumhard, B., Lysimeterversuche zum Langzeitverhaltender Herbizide Metamitron und Meth-(Goltix})abenzthiazuron in einer Parabraunerde mit(Tribunil})besonderer Beru� cksichtigung der Transport- und Verla-gerungsprozesse unter Einbeziehung von Detailuntersu-chungen. Berichte des Forschungszentrums (1991)Ju� lich235 S.

24. Gottesbu� ren, B., Pestemer, W., Wang, K., Wischnewsky,M.-B. & Zhao, J., Concept, structure, and validation ofthe expert system HERBASYS (Herbicide-Advisorysystem) for selection of herbicides, prognosis of persistenceand e†ects on succeeding crops. BCPC Monograph 47(1990) 129È38.

25. Diekkru� ger, B. & Bork, H.-R., Temporal variability of soilsurface crust conductivity. Soil T echnology, 7 (1994) 1È18.

26. Hutson, J. L. & Wagenet, R. J., Multi-region water Ñowand chemical transport in heterogeneous soils : theory andapplications. Pesticide movement to water. BCPC Mono-graph 62 (1995) 171È80.

27. Pothuluri, J. V., Moormann, T. B., Obenhuber, D. C. &Wauchope, R. D., Aerobic and anaerobic degradation ofalachlor in samples from surface-to-groundwater proÐle.J. Environ. Qual., 19 (1990) 525È30.

measurement and simulation of herbicide transport in macroporous soils

Documents