Impact of Soil Water Flux on Vadose Zone Solute Transport Parameters

Pedosphere 10(4): 323~332, 2000323ISSN 1002-0160/CN 32-1315/P@ 2000 SCIENCE PRESS, BEIJINGImpact of Soil Water Flux on Vadose Zone SoluteTransport Parameters*1CHEN XIAOMINI, M. VANCLOOSTER2 and PAN GENXING11 College of Resources and Environmental Sciences, Nanjing Agricultural University, Nanjing 210095 (China)352 Department of Enrironmental Sciences and Land Use Planning, Universite Catholique de Louvain, B-1348Louuain-la-Neuue (Belgium)(Receive July 30, 2000; revised August 23,200)ABSTRACTThe transport proceses of solutes in two soil columns flled with undisturbed soil material collectedfrom an unsaturated sandy aquifer formation in Belgium ubjected to a variable upper boundary conditionwere identified from breakthrough curves measured by means of time domain relectometry (TDR). Solutebreakthrough was measured with 3 TDR probes inserted into each soil column at three different depths ata 10 minutes time interval. In addition, soil water content and pressure head were measured at 3 diferentdepths. Analytical solute transport models were used to estimate the solute dispersion coffcient and averngepore-water velocity from the obeerved breakthrough curves. The results showed that the analytical solutionswere suitable in ftting the observed solute transport. The dispersion coeficient was found to be a function ofthe soil depth and average pore water velocity, imposed by the soil water fux. The mobile moisture contenton the other band was not correlated with the average pore water velocity and the dispersion coeficient.Key Words: dispersion ceficient, pore water velocity, solute transport, time domain rflectometry (TDR)INTRODUCTIONAppropriate management of the soil and water resources requires a good knowledge ofthe transport processes and fate of dissolved chemicals in the soil water. Water fow imposesa convective movement on the dissolved chemicals, while local variations in the fow velocityinduce hydrodynamic dispersion. The process of difusion, induced by the difterences in thesolute concentration, enhances solute displacement. Dissolved chemicals in the soil waterare further subjected to chemical and physical interactions between the soil fuid phase andthe soil matrix, either enhancing or retarding the transfer of the substances dissolved in thewater phase ( Vanclooster et al, 1995). Therefore, the characterization of the chemical solutetransport in the soil has become an active feld of environmental research.Given the intrinsic variability of the soil physical parameters, the characterization ofsolute transport at the field scale is a complicated task. In recent years, the time domainrefectometry (TDR) has been widely used in soil physical research. It has been considered asan attractive tool with a wide range of applicability since it is nondestructive, high in accuracy*1Projcct supported by the European Economic Community Research Program STEP.中国煤化工MYHCNMHG324x. M. Chen et al.and lower in labor requirement compared with other methods (Kim et al, 1998; Wallach andSteenhuis, 1998). In addition, the TDR allows to measure two important soil physical statevariables: the soil moisture content and the total electrical conductivity, and this with ahigh spatio temporal resolution. The TDR technology allows therefore to measure the watertransport and solute breakthrough at the timne when the experiment is being conducted anddoes not require assumptions about the local fux density of water (Kachanoski et al, 1992).Use of TDR in estimation of the concentrations of conservative solutes was initiated byDalton et al. (1984), based on signal attenuation of a voltage pulse propagating along thetransmission line, which serves as a measure of electrical conductivity in a bulk soil. Van-clooster et al. (1993) and Mallants et al. (1994) used horizontally installed TDR probes inlaboratory-colunn experiments taken along a field transect. Horizontal positioning of theTDR probes enables the sampling of a larger area perpendicular to the mean direction ofAow. This is an advantage for estimating solute fuxes, especially in multilayered heteroge-neous soils where horizontal flow at the microscopic scale might inAuence longitudinal solutedispersion. In addition, the borizontal configuration allows measuring solute breakthroughat diferent depths, which is a prerequisite for identifying the governing solute transportmechanism {Jury and Roth, 1990). When solute velocities are independent of solute depthfull mixing of solutes in the complete pore water domain occurs. In this case the convectivedispersive transport model is appropriate, and solute dispersion remains constant with depthchanges. However, if solute velocities vary with the soil depth, a stream- tube concept willbe more appropriate for describing the transport. The identification of the governing solutetransport can be estimated from solute breakthrough curves measured at diferent depths,as presented by Vanclooster et al. (1995), Vanderborght et al. (1996, 1997) and other re-searchers. However, few studies have been reported in literature so far about the impact ofthe fow condition on the governing solute transport concept.The objective of the present study was to determine the governing transport processes assolute moves through undisturbed soil columns and to analyse the dependency of the solutetransport parameters upon water fAux density and soil moisture.THEORYMeasurement of solute brealkthrough by means of TDRAs shown by several researchers (e.g., Kachanoski et al, 1992; Reece, 1998), solute break-through curves may be established from TDR-based estimates of the bulk soil electrical con-ductivities (ECb) during steady state solute transport experiments with salty tracers. Alinear relationship is generally observed between the resident solute concentration of a saltytracer, Cr, and EC, for constant water contents ranging from relatively low to saturationand for salinity levels ranging from 0 to approximately 50 dS m-1 (Ward et al, 1994).C,= a+ βEC% .(1)where a and β are calibration constants. The ECb (dS m-1) can be related to the impedance,中国煤化工MYHCNMHGWATER FLUX IMPACT ON SOLUTE TRANSPORT325Z(8), of an electromagnetic wave that travels through the soil (Topp and Reynold, 1998):EC%=_Ke(2)z- Zeable .where Ke is the cell constant of the TDR probe (m- 1), and Zcable(2) is the resistanceassociated with cable, connectors, and cable tester. Relative solute concentration, C(x, t),can be expressed asC,-C;C(x,t)=Co-Ci(3)27where Co is a reference concentration such as the input concentration during miscible dis-placement, and C{ is the background concentration. Combing Eqs. (1) to (3), we could getz:- z;+C(,)=z51-Z(4)where乙is the impedance before application of the tracer solution, and Zo is the impedanceassociated with the reference concentration, Co. Eq. (4) shows that, under steady flow condi-tions (ie, constant soil water content) the relative solute concentration, C(x,t), at a particular depth, x, and time, t, can be derived from the measured impedance, Zo,, if appropriatevalues of Z; and Zo are available. The values for Zo have been discussed by Mallants et al.(1996).Estimating solute transport parametersAnalytical solutions of the governing steady state solute transport models may be usedin an inverse way to estimate the solute transport parameters (Toride et al, 1995). Usinginverse procedures, breakthrough curves drawn by the observed laboratory or field data arematched to the analytical solutions. Computer codes such as CXTFIT 2.0 (Toride et al.,1995) are ready available to predict solute distributions in time and space for specified modelparameters and to estimate solute parameters in an inverse way.The convection- dispersion equation of the CXTFIT model allows to simulate one-dimen-sional transport of solutes, subject to adsorption, first-order degradation, and zero-orderproduction, in a homogeneous soil. The model is formalized as: .8(0C:+poCc)= 品(D -JwCr) - 0u1Cr - PD14eCc + 0r(x) + PrYe(x)(5)where Cr is the volume-averaged or resident concentration of the liquid phase (gL-1); Cc isthe concentration of the adsorbed phase (g L-1); D is the dispersion coeficient (cm2h- 1);θ is the volumetric water content (cm3 cm- 3); Jw is the volumetric water flux density (enh-1); Pb is the soil bulk density (g cm- -3); μ1 and ho are the first order decay cofficientsfor degradation of the solute in the lquid and adsorbed phases, respectively; 71 (mol cm~h-1) and 7。(mol cm-3 h-1) are the zero order production terms for the liquid and adsorbed中国煤化工MYHCNMHG326x. M. Chen et al.phases, respectively; x is the distance (cm); and t is the time (h). We assumed that μ couldnot be negative. The production functions are given as a function of distance.Solute adsorption by the solid phase is described with a linear sorption isotherm asCc = KsC,(6)where Kd is an empirical distribution constant. Using Eq. (6) and assuming steady-state fowin a homogenous soil, Eq. (5) may be rewritten as_8Cr8C,F:==D8x2-v-μC:+r(x) .(7)where v(= Jw/0) is the average pore water velocity, F is the retardation factor given byF=1+PbKa(8)and μ and γ are, respectively, combined first- and zero-order rate coefficients:PrKdHaμ=μl +θ(9)Pbre(x)r()=n(x)+”日(10)When the frst-order degradation coffcients in the liquid (41) and adsorbed (42) phases areidentical, Eq. (9) becomesμ= u1F(11)MATERIALS AND METHODSThe transport processes of solute in the vadose zone were examined in two undisturbedsoil columns. The experimental soil was collected from a sandy aquifer close to a largelandfll site. The soil columns have a total length of 30 cm and a diameter of 15 cm. Some .characteristics of the soil are given in Table I.TABLE1Selected characteristics of the soil used in the studySoil horizonDepthClaySiltSandBulk densityPorosityOrganic Cg kg~g cm-3gkg-+A10~103206401.4644.9022210~203326261.4545.2820320~30350071.4047.1718The soil colunns were installd in the laboratory. Fig. 1 gives a schematic diagram of thedetermination device for solute transport in the soil column experimental device.中国煤化工MYHCNMHGWATER FLUX IMPACT ON SOLUTE TRANSPORT327Solution input systemPeristaiticc pumStopperReservoirCasingpipeTensiometer31Probe 03X95 mm4Discharge systemH:HComputerLevel change打TDRReference level: h=0zFig. 1 Schematic diagram of the determination device for solute transport in the soil column. To simplifythe drawing, only the probe for 1 measuring depth is depicted.The soil columns were put on filter. A 0.05 m high supporting reservoir was put under theflter layer. A 1.5-cm- diameter drainage tube was embedded in water layer and connected toa water column. The soil columns allow to create a negative pressure head at the bottom ofsoil column and unsaturated fow conditions. Soil water contents,日, and bulk soil electricalconductivities, ECb, were monitored by means of TDR probes installed at three differentdepths (10 cm, 20 cm and 30 cm). Prior to the installation, the TDR probes were calibratedin the laboratory for moisture and impedance measurements. In addition to the TDR probes,three tensiometers were also inserted horizontally at the same depths as TDR probes.The soils were saturated with tap water before the experiments. As the impedance ofdistilled water was higher than 1 ks, which is beyond the measuring range of the cable tester,normal tap water (electrical conductivity = 0.75 dS m- 1) was used. A constant fux of waterwas established by means of a syringe peristaltic pump. Three different fux densities (2.11,3.59 and 4.61 cm h~1) in soil colunn 1 and two different fux densities (2.14 and 3.58 cmh-1) in the soil column 2 were applied. After establishing steady state saturated flow usingtap water, a solute solution containing 1 g L-1 Ca (NO3)2 was added continuously to keepthe soil columns in a state of C = Co. Solute application time (to) ranged from4h to 7 h.Measurements were made with a Tektronix cable tester (Tektronix 1502C, Beaverton, OR).After saturated state conditions were reached, the tap water was applied again until solutes中国煤化工MYHCNMHG328X.M.Chenetal..completely leached out of the soil columns. Values of the total resistance or inpedance, z, ofsoils were obtained from 3 TDR probes inserted into the soil column at three depths of 10, .20 and 30 cm. The travel time of the electromagnetic wave and the impedance were takenfrom the screen of the cable tester.Impedance of the soil, soil water content, and pressure head at the 3 different depths weremeasured every 10 minutes. Topp's equation was used to calibrate TDR for soil moisture.The indirect calibration procedure described by Mallants et al. (1996) was used to calculaterelative solute concentration. The observed breakthrough curves were modeled by means ofthe convection-dispersion equation available in CXTFIT 2.0.RESULTS AND DISCUSSIONThe measured impedance variations with the time after adding Ca(NO3)2 to the soilcolumn 1 are shown in Fig.2. At time = 0 the impedance was about 260 n at the depth of10 cm in the soil column 1 when the solution Aux was 3.59 cm h- 1 and dropped to about 180几about 1 hour after Ca(NO3)2 was added. The value of impedance remained fairly constantaround 180 n during about 4.5 hours, indicating that the TDR signal was not affected by thelocation of the solute mass within the transmission (probe) length. Some minor variations inimpedance were measured, but the variations were much smaller than the impedance increaseafter 4.5 hours.300.52500.49200000.°E0000q吴150Depth=10 cm0.2-十Depth=10 cmDepth=20 cm0, Depth-30cm0.1一Depth=20 cm065101520253035%55101520253035Time nhTime(h)Fig. 2 Relationship between soil impedance and time after adding Ca (NO3)2 in soil colunn 1.Fig. 3 Relationship between moisture distribution in the soil coluan and time measured by TDR.The impedance of middle soil layer dropped to some 170 8 about 2 hours after Ca(NO3)2was added, while that of the bottom layer decreased to 170 8 about 3 hours after. Similarresults were obtained by the other Auxes in two soil columns. However the higher the fux,the lower the impedance was.The value of impedance for almost all the TDR probes did not return exactly to theoriginal value before Ca(NO3)2 was added. After 10 hours of the experiment, the finalimpedance was slightly higher than the initial impedance value. This suggested that theaddition of the solute pulse had in some way altered the baseline ECb (Kachanoski et al,1992). In soils where the original solute concentrations are high, the final impedance readingsmay be higher than the initial readings because of leaching in the soil. The magnitude of中国煤化工MYHCNMHGWATER FLUX IMPACT ON SOLUTE TRANSPORT329the error will depend on the initial solute concentrations of the soil and the leaching water.Fig. 3 shows the soil water distribution in soil column during the leaching experiment.Steady state flow was obtained as observed from the steady-state moisture contents. Thvertical distribution of the water in the soil columns were very nonuniform during the leachingexperiments. Near the reserivoir, the horizontal TDR lines gave water contents about 0.4cm3 cm-3. The distribution of the moisture content tended to be low in soil surface andhigh in the bottom of the soil column due to the hysteresis, caused by the diferent moisturecontents of the three layers at the begining of the experiment (Heimovaara and Bouten,1990) and the infuence of gravitational potential on the water movement.The CXTFIT model (Toride et al, 1995) was used to analyze the solute transport be-haviors based on the one dimensional convection-dispersion equation (Eq. 5) under variousboundary conditions. The model was used to optimize pore water velocity and dispersioncefficient from observed breakthrough curves. Fig. 4 shows the experimental data and theanalytical data with CXTFIT of the two undisturbed sandy soil columns. Good agreementcould be seen between the measured and simulated breakthrough curves of relative concen-trations of solute with time. A perfect convergence of breakthrough curves in the two soilcolumns could be found. This study demonstrated that the analytical model available inCXTFIT was suitable for ftting the observed solute transport. However, this does not meanthat the convection dispersion model is the governing transport mechanism, since this canonly be proofed by analyzing the travel depth dependency of the solute transport parameters(Jury and Roth, 1990).1.2p1.Column 1Column 2.80.- + Measured告Simulated人o.2P102030040Time (h)Fig. 4 Measured and simulated breakthrough curves of relative solute concentrations (C/Co) in relation totime in the two undisturbed soil colunns.Estimated transport parameters for three solute Auxes in the soil colurnn 1 and the twosolute fAluxes in the soil column 2 are given in Table II. A good ft between the experimentaldata and the data of analytical model, CXTFIT, could be obtained, as shown by the highR2 values.In the soil column 1, as shown in Fig. 5, the dispersion coefficient of solute decreased withthe soil depth, and regression analysis indicated that the correlation between the dispersioncoefficient and depth was significant at the 5% probability level. The equation is Y =2310.36X1-98 (r= -0.8477, n = 9), where X is the soil depth (cm) and Y is the dispersioncofficient (cm2 h- 1). The decrease of the dispersion coefficient with depth may indicate中国煤化工MYHCNMHG330X. M. Chen et al.that stable solute fow was not reached in the top soil layer.TABLE IICharacteristics of solute breakthrough curves for two undisturbed soil columnsSoil columnFluxDepthRmh-1mcm3 cm-3cm b-1cm2 h-13.5900.296715.7130.120.86540.305512.023.500.9533300.422210.211.800.92592.110.27738.569.900.9719200.29547.394.700.41736.282.300.86314.610.273834.8952.880.91440.290718.009.340.427714.896.060.850922.14l00.28468.323.220.95830.30286.462.440.39636.390.96153.580.25479.494.690.94930.30251.950.94290.37858.719.350.9695In the soil column 2, the relationship of the dispersion coefficient of solute with soil depthcould not be found to have any regularity (Fig. 5)。Especially, the dispersion coefficientincreased at the bottom soil layer, which might be attributed to the difference of texturebetween top and bottom soils.60[50◆Column 1fLx=4.61 cm h1; 40●Column 2.+ C2 lux=3.58cmhtq2o}11:ie40Depth (cm)Fig. 5 Variations of dispersion effcient (D) of solute with the depth in the two soil columns.Fig. 6 Relationsbips between the pore water velocities (v) and the soil depths in the soil colunn 1 (C1) andsoil colunn 2 (C2) at diferent solute fuxes.These observations showed that the adopted size of the experimental device might becritical in obtaining solute transport parameters. This is suggested by the decrease of thehydrodynamic dispersion cofficient with depth in the first column due to unstable fowconditions; and sudden increase of the hydrodynamic dispersion cofficient with depth inthe second column due to structural heterogeneity. We therefore recommend to increase thesampling size of the soil columns in future solute transport studies. Average velocities of---- .中国煤化工MYHCNMHGWATER FLUX IMPACT ON SOLUTE TRANSPORT331pore water in the soil columns were calculated by dividing the Darcian fAux by the averagemeasured volumetric moisture content. The relationships of pore-water velocities with thesoil depths in the two soil columns are shown in Fig. 6. The pore-water velocities were alsoaffected by gravitational potential in homogeneous soils. It was indicated that the pore-watervelocities decreased with the soil depth.By regression analysis, a significantly positive correlation at the 5% probability level couldbe found between the average pore water velocity and the dispersion coeficient of solute inthe soil colunn 1 (Fig.7). But such a correlation could not be found in the soil colunn2, because the dispersion coeficient in the bottom soil layer suddenly increased due to soil43heterogeneity.Op6Cy= 5.453x - 57.15550 y= 3.3433x- 19.1415o y=2.4193x-31.906@R2= 0.92624CR2= 0.962740R7= 0.993353C30920|202C10}BC240 0V(cm h")Fig. 7 Correlations between average pore-water velocities (川) and dispersion coeficients (D) in the soilcolumnl at solute fuxes of3.59 cmh-1 (A), 2.11 cm h-1 (B), and 4.61 cm h-1(C).CONCLUSIONSThe fAux dependency of the vadose zone solute transport parameters was analyzed bymodeling the TDR measured breakthrough curves in undisturbed small soil cores. Solutetransport parameters were inferred by inverting the analytical solution of the governing clas-sical convection dispersion equation. Breakthrough of a salty tracer (Ca(NO3)2) at differentdepths in the soil colurnns were measured on-line by means of TDR.Notwithstanding the concern about the adopted sample size in our study, clear linearrelationships between the solute dispersion coficient and the pore water velocity were iden-tifed. This confirms our hypothesis that vadose zone solute transport parameters are largelydependent on the flow regime occurring in the porous media. More experimental studiesshould therefore be carried out at the larger scales to further explore the relationships be-tween solute transport parameters and water flow regime imposed by the surface boundaryconditions.REFERENCES ,simultaneous measurements of soil water content and electrical conductivity with a single probe. Science.224: 989~990.2 Heimovara, T. J. and Bouten, W.1990. A computer- controld 36-channel time domain reflectometry systemfor monitoring soil water content. Water Resour. Res. 26: 2311~2316.中国煤化工MYHCNMHG332x. M. Chen et al? Jury, w. A. and Roth, K. 1990. Transfer functions and solute transport through Boil: theory and applications.Birkhauser Verlag, Basel, Switzerland. 232 PP.Kachanoski, R. G, Pringle, E. and Ward, A. 1992. Field measurement of solute travel time using timedomain reflectmetry. Soil Sci, Soc. Am. J. 56: 47~52.5 Kim, D. J.. Vanclooster, M., Feyen, J. and Vereecken, H.1998. Simple linear model for calibration of timedomain reflectmetry measurements on solute concentration. Soil Sci. Soc. Am. J. 62: 83~89.Mallants, D. Vanclooster, M., Meddahi, M. and Feyen, J.1994. Estimating solute transport in undisturbedsoil colunns using time domain rflectometry. Journal of Contaminant Hydrology. 17: 91~109.1 Mallants, D, Vanclooster, M., Toride, N, Vanderborght, J, van Genuchten, M. Th. and Feyen, J. 1996.Comparison of three methods to calibrate TDR for monitoring solute movement in undisturbed soil. SoilSci. Soc. Am. J. 60: 747~754.百Recce, C. F. 1998. Simple method for determining cable length resistance in time domain reftectometrysystems. Soil Sci. Soc. Am. J. 62: 314~317.1Topp, G. C. and Reynolds, w. D.1998. Time domain reflectometry: a seminal technique for measuring massand energy in soil. Soil and Tllage Research. 47: 125~132.oToride, N. Leij, F. J. and van Genuchten, M. Th. 1995. The CXTFIT code for estinating transport param-eters from laboratory or field tracer experiments. Research Report No. 137. U.S. Salinity LaboratoryAgricultural Reaearch Service, U.S. Department of Agriculture, Riverside, California, 121pp."' Vanclooster, M., Malants, D. Diels, J. and Feyen, J.1993. Determining local scale solute transport param-eters using time domain reAectometry (TDR). Jourmal of Hydrology. 148: 93~107.2 Vanclooster, M., Mallants, D., Vanderborght, J, Diels, J.. Van Orshoven, J, and Feyen, J. 1995. Monitoringsolute transport in a multi-layered sandy lysimeter using time domain reflectometry. Soil Sci. Soc. Am.J. 59:; 33713vandnderborght, J.. Malants, D.. Vanclooeter. M. and FPeyen, J. 1997. Parameter uncertainty in the mobile-immobile solute transport models. Journal of Hydrology. 190: 75~101.14Vanderborght, J, Vanclooster. M, Mallants, D, Diels, J. and Feyen, J. 1996. Determining convectivelognormal solute transport parameters from resident concentration data. Soil Sci. Soc. Am. J. 60:1 306~1317i1 Wallach, R. and Steenhuis, T. s. 1998. Model for nonreactive solute transport in structured soils with{、continuous preferential flow paths. Soil Sci. Soe. Am. J. 62: 881~86.b Ward, A. L., Kachanoeki. R. G. and Elrick, D. E. 1994. Laboratory measurements of solute transport usingtime domain reflectometry. Soil Sci. Soc. Am. J. 58: 1 031~1 039.----中国煤化工MYHCNMHG