This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).
Surfaceapplied chemicals move through the unsaturated zone with complex flow and transport processes due to soil heterogeneity and reach the saturated zone, resulting in groundwater contamination. Such complex processes need to be studied by advanced measurement and modeling techniques to protect soil and water resources from contamination. In this study, the interactive effects of factors like soil structure, initial soil water content (SWC), and application rate on preferential flow and transport were studied in a sandy loam field soil using measurement (by time domain reflectometry (TDR)) and modeling (by MACRO and VS2DTI) techniques. In addition, statistical analyses were performed to compare the means of the measured and modeled SWC and EC, and solute transport parameters (pore water velocity and dispersion coefficient) in 12 treatments. Research results showed that even though the effects of soil structural conditions on water and solute transport were not so clear, the applied solution moved lower depths in the profiles of wet
Flow and transport processes in the unsaturated or vadose zone control the time and degree of groundwater pollution because the surfaceapplied chemicals need to pass this zone first to be able to reach groundwater. The factors or processes taking place in this zone are numerous and complex due to the soil heterogeneity. Advanced experimental or modeling tools are required to understand the mechanisms of such complex flow and transport processes. Such tools allow us to develop good management practices to protect soil and groundwater from contamination because of the landapplied chemicals, like agricultural fertilizers and pesticides.
Water and solutes can move through the vadose zone along preferred pathways, such as soil cracks, worm holes, and root channels [
Since the transport of water and solute in the vadose zone is not uniform due to soil heterogeneity, classical models using the Richards Equation do not produce acceptable results [
Enabling to measure both SWC [
Studies focusing on the understanding of the mechanisms of preferential flow and transport processes are still being intensively conducted. Even though several studies have been conducted on the investigation of individual factors like soil texture, structure, initial SWC, and application rate on preferential flow and transport in the laboratory conditions [
The objectives of this study are to investigate the individual and interactive effects of factors like soil structure, initial SWC, and application rate on the extent of preferential flow and transport in a sandy loam field soil by using TDR to measure SWC and EC; modeling the TDRmeasured SWC and EC data by MACRO and VS2DTI; comparing the treatment means of SWC and EC, and solute transport parameters (pore water velocity and dispersion coefficient), obtained by fitting the TDR measured breakthrough curve (BTC) data to the onedimensional convectiondispersion equation (CDE) in the CXTFIT program, by means of statistical analyses.
Field experiments were conducted on a sandy loam soil located about 15 km far from the center of Kahramanmaraş City in Turkey (31°55′28″E and 41°54′54″N) between August 11 and September 17, 2007. The area was located on a private farm, where mostly vegetables and corn were grown, but it was not planted in the year experiments were conducted. The study area is under the Mediterranean climatic conditions, characterized mainly by hot and dry summers and warm and rainy winters. The mean annual temperature was 16.3 °C and mean annual rainfall was 708.1 mm. The mean highest evaporation was 333.3 mm.
Some physical and chemical properties (
Soil texture was sandy loam throughout the profile except the last layer which was sandy (
A salt (CaCl_{2}) solution with a concentration of 3,200 mg·L^{−1} was applied by a rainfall simulator, having dimensions of 1.5 m by 1.0 m by 0.30 m, being made of an aluminum sheet with the thickness of 1 mm, and including 150 injectors (10 by 15 with 10 cm space) at the bottom of its reservoir which was 20 cm above the soil surface. Once, the solution concentration to be applied was determined as 3,200 mg·L^{−1} by considering the current soil EC, the applied water EC, and the maximum EC measuring value (4.5 dS·m^{−1}) of the CS630 model TDR probes. Then the simulator was calibrated to obtain 2.962 and 4.060 cm·h^{−1} for the low and intermediate solution application rates by applying the same amount (12 cm) of solution within 4.051 and 2.956 h, respectively, in sandy loam soil.
When the infiltration capacity of the soil is less than the solution application rate, ponding occurs on the soil surface, resulting in preferential flow and transport automatically. The application rates were defined based on these considerations so that no ponding occured on the soil surface. However, the application durations and rates changed among the treatments even though the same amounts (12 cm) of solutions were applied for high application rate as flooding.
Although no ponding was observed on the soil surface for low and intermediate application rates during the experiments, ponding was inevitable in the treatments with high application rate as expected.
The main factors, subfactors and their interactions produced 12 (2 × 2 × 3) experimental treatments (
The TDR100 (Campbell Scientific, Logan, UT, USA) was used in all experiments of this study. The TDR system comprised a TDR, datalogger and multiplexers. The TDR can be controlled by using either a datalogger or a computer with Windows software (PCTDR). While a datalogger with the TDR and multiplexers is used for automatic and multiprobes measurements, PCTDR is used for either manual measurements or setup and troubleshooting. A detailed theoretical background of TDR can be found in [
The apparent dielectric constant or relative permittivity of the medium (κ) can be used to calculate the volumetric SWC (θ) by using the universal calibration equation developed by [
The calibration equation (
A soil profile was dug along 1.5 m of the experimental plot (1.5 m by 1.0 m) to be able to horizontally insert three CS630 model TDR probes (3rod, 15 cm long, and 0.318 cm in diameter) for measuring SWC and EC in soil depths of 10, 20, 30, 40, 50, 60, and 75 cm using a probe insertion guide. A sketch of the experimental design for a plot of a treatment is illustrated in
The TDR measured EC and especially SWC had some instabilities with time, resulting in the values having no physical sense such as the negative or >1 values of SWC. To resolve the problem, the main trends of SWC and EC with time were visually determined on graphs and then the outliers were removed. For a given depth by averaging three probe readings, the SWC and EC values were determined. The final SWC and EC values for a specific depth of a treatment were obtained by finding the average of the values of six probe readings (three probes and two replicates for a depth). In addition, soil bulk EC (BEC) readings of the TDR were corrected as:
In MACRO model the total soil porosity is divided into macropore and micropore flow regions and each flow region has its own saturation rate, hydraulic conductivity, flux, and solute concentration [
Application of the models which have ability to simulate steady and/or nonsteadystate flow and transport that requires different modeling approaches (single or dualpermeability) helps the user to determine the extent of preferential or nonuniform flow and transport in a soil as in this Solute transport parameters can be determined by inversely fitting the onedimensional CDE to the BTCs data obtained from the experimental measurements using the CXTFIT model developed by [
MACRO and VS2DTI models were utilized to simulate SWC and EC measured along seven layer profiles of field sandy loam soil every 15 min during around 24 h using the TDR. The models had different parameter values for each soil layer because each layer had its own physical and hydraulic properties. Therefore, the determination of the model parameter values is explained as follows.
After solute transport parameters and measured physical and chemical properties of seven soil layers were input to MACRO, initial and boundary conditions and management parameters of the study area were set in the model. The constant hydraulic gradient was set as the bottom boundary condition. The management options such as notillage (for longterm), solution applied by irrigation, no drainage system and crop production in the study area, and nonreactive solute were selected in the model. The values of measured initial SWC and EC, solution application parameters (day of irrigation, amount of irrigation (120 mm), duration of irrigation (varying), and the concentration (3,200 mg/L) of solution applied by irrigation), and solute transport parameters were input to the model. The values of soil hydraulic parameters (θ_{r}, θ_{s}, α, n) used in the model were determined by fitting the [
After the geometry and layer thicknesses of the soil profile were drawn by using the drawing tools in VS2DTI, the options of flow and transport model, iteration, and output were defined. Then the values of flow and transport parameters of each layer were determined. The values of soil hydraulic parameters (θ_{r}, θ_{s}, α, n) used in the model were determined by fitting the van Genuchten hydraulic function to the SWRC data using the RETC programme [
The options of the stochastic equilibrium CDE model and the fieldscale resident concentration were set in the inverse modeling with CXTFIT. The applied solution concentration (3,200 mg·L^{−1}) and its duration were determined after setting the option of pulse input at application time t as the boundary condition in the model. The measured initial concentrations were input to the model as initial condition after setting the constant initial concentration. The values of solute transport parameters (pore water velocity, v and dispersion coefficient, D) were obtained after inputting the measured BTC data and then running the model.
Bulk electrical conductivity (BEC) was measured directly by TDR as Siemens m^{−1} at different soil layers with time for the treatments. MACRO model produced the results for total solute storage in each layer as g·m^{−3} or ppm, whereas VS2DTI outputs for total solute storage at defined nodes were obtained as mg·cm^{−3}. Then the solute storage outputs of MACRO and VS2DTI models were converted to the BEC to be able to compare with the BEC measured by TDR. The following equation was used in the conversion of model outputs to the BEC as [
Therefore, BEC (assumed to be EC) was preferred for simulations because converting the TDR measurements and model outputs (in different units) to solute concentration, resulting in a 2step conversion rather than a singlestep conversion, increases uncertainty in the measurements and modeling.
The performances of the models in the estimation of the TDR measured SWC and EC were evaluated using some statistical parameters. The comparison of the means of SWC and EC values measured in the treatments was performed by Tukey test using the SPSS statistical programme as well as the means of the measured and estimated (by two models) values of SWC and EC in a treatment. Then they were grouped using the letters. In addition, the values of parameters like the coeficient of model efficiency (CME) and the root mean square error (RMSE) were computed for evaluation of the performances of the models (MACRO and VS2DTI) in estimation of SWC and EC in a soil depth of a treatment as:
The experimental and modeling studies were conducted on the 12 treatments. The results of the spatial and temporal variation of the SWC and EC for the treatments 1 and 3 were visually presented in
In addition, the results of the only first 8 treatments (see
SWC and EC changed until the 5th h of 30 cm of the 1st treatment (
As the initial SWC and EC did not change throughout the profiles of the treatments 1, 2, 4, and 5 in the 1st h of the experiments, they increased up to 40 and 50 cm with time in the treatments 1 and 2 and the treatments 4 and 5, respectively (
The increase in SWC and EC with time through the profiles of the treatments 7 and 8 was more than that of the treatments 1 and 2 under undisturbed soil conditions. Similarly, the applied solution increased SWC and EC with time through the profiles of the treatments 10 and 11 more than that of the treatments 4 and 5 under disturbed soil conditions. The results showed that the wet initial SWC was more effective on water and solute transport than the dry initial SWC condition and it caused water and solute to move deeper depths as reported by [
As long as the other conditions were contant, when the treatments were compared for the effects of application rates among 4 treatment groups (treatments 1, 2, and 3; treatments 4, 5, and 6; treatments 7, 8, and 9; treatments 10, 11, and 12), the solution applied with the higher application rate moved lower depths at shorter durations in each group. Similarly, large rainfall events after herbicides application caused considerable amount of them to leach deeper depths in a sandy loam soil [
In the evaluation of the performances of the models (MACRO and VS2TI) for estimation of SWC and EC with space and time in a treatment, two statistical parameters (CME and RMSE) were calculated and the results were presented on
MACRO simulated SWC and EC better than VS2DTI in three of the seven depths and in six of the seven depths, respectively, in the 1st treatment (
Soil hydraulic parameters are crucial input data in modeling studies on water flow and solute transport in the vadose zone; therefore, their accurate determination through either measurement techniques and/or predictive methods (
The model parameter values used in the modeling studies were presented in
Tukey test was performed to analyze whether the differences among the means of the measured and simulated (by MACRO and VS2DTI models) SWC and EC in 12 treatments were statistically significant or not, and the results were presented in
Solute transport parameters (v and D) were inversely estimated by fitting the BTC data of the treatments to the onedimensional CDE by using the CXTFIT program and the means of these parameter values were compared by the Tukey test. The results were presented in
In this study, individual and interactive effects of different factors like soil structure, initial SWC, and application rate on water flow and solute transport characteristices in a sandy loam field soil were studied by using the TDR measured SWC and EC, modeling the measured parameters (SWC and EC) with MACRO and VS2DTI models, and comparing the treatment means of SWC and EC and solute transport parameters by means of statistical analyses.
Overall the effects of soil structure on water flow and solute transport were not so clear under dry and wet initial SWC conditions, but the applied solution moved slightly lower depths in the undisturbed soil profiles than in the disturbed profiles under the wet initial SWC conditions. The applied solution moved lower depths in the wet initial SWC conditions than in the dry initial SWC condition and this effect of wet initial SWC condition was more distinctive under the undisturbed soil condition compared to the disturbed condition. The solutions applied with the higher application rates moved lower depths provided that the other conditions were contants, but it was difficult to differentiate the effects of the application rates on water and solute transport under different soil structural and initial SWC conditions. The interactive effects on water flow and solute transport may be better differentiated if more field experiments are conducted with the distinct interactive treatments.
Although the models had relatively low performances in the estimation of SWC and EC with space and time, the results showed that MACRO had somewhat better performance than VS2DTI. The differences among the 12 treatments determined using statistical analyses based on SWC, EC, and solute transport parameters were another indication of a certain degree nonuniform or preferential flow and transport in the soil.
This project was funded by The Scientific and Technological Research Council of Turkey (TÜBİTAK) with the Project No: 104Y224. The author would like to acknowledge the helps of Jason Ritter and his colleagues, Campbell Scientific Inc., on the CR1000 datalogger programming. The author would like to thank Iris Vogeler for her help on the experimental design, and Nicholas Jarvis for his help on the modeling.
A sketch of the experimental design. Connection of a single thermocouple to the datalogger and connection of a single probe to the multiplexer are shown in the skecth to make presentation simple.
Spatial and temporal variability of soil water content (SWC) and electrical conductivity (EC) in the treatment of Sandy loam + Undisturbed + Dry + Low. The symbols of ○, ▲, and ■ represent the measured and model (MACRO and VS2DTI; values, respectively. CME and RMSE are the coeficient of model efficiency and the root mean square error, respectively.
Spatial and temporal variability of soil water content (SWC) and electrical conductivity (EC) in the treatment of Sandy loam + Undisturbed + Dry + High. The symbols of ○, ▲, and ■ represent the measured and model (MACRO and VS2DTI; values, respectively. CME and RMSE are the coeficient of model efficiency and the root mean square error, respectively.
The variation of soil water content (SWC) and electrical conductivity (EC) with depth at 1 h of the experiment in the 8 treatments. The initial, measured, and model (MACRO and VS2DTI) values are exhibited by symbols ★, ○, ▲, and ■, respectively. CME and RMSE are the coeficient of model efficiency and the root mean square error, respectively.
The variation of soil water content (SWC) and electrical conductivity (EC) with depth at 3 h of the experiment in the 8 treatments. The initial, measured, and model (MACRO and VS2DTI) values are exhibited by symbols ★, ○, ▲, and ■, respectively. CME and RMSE are the coeficient of model efficiency and the root mean square error, respectively.
The variation of soil water content (SWC) and electrical conductivity (EC) with depth at 10 h of the experiment in the 8 treatments. The initial, measured, and model (MACRO and VS2DTI) values are exhibited by symbols ★, ○, ▲, and ■, respectively. CME and RMSE are the coeficient of model efficiency and the root mean square error, respectively.
Some physical and chemical properties of experimental field soil.
 

(cm)  (%)  (%)  (%)  (g·cm^{−3})  (%)  (%)  (cm·h^{−1})  (dS·m^{−1})  (%)  
10  77 (3.43) 
15 (2.95)  8 (0.62)  SL  1.54 (0.05)  42 (1.98)  1.25 (0.06)  2.292 (1.70)  7.89 (0.04)  0.688 (0.10)  24.61 (0.68) 
20  78 (8.76)  14 (7.47)  8 (1.30)  SL  1.53 (0.07)  42 (2.71)  1.07 (0.08)  2.931 (0.77)  7.97 (0.12)  0.592 (0.11)  26.08 (1.39) 
30  75 (2.56)  17 (1.53)  8 (1.20)  SL  1.52 (0.13)  43 (5.05)  0.86 (0.05)  2.759 (0.75)  8.00 (0.02)  0.556 (0.04)  28.62 (1.57) 
40  72 (6.63)  20 (5.86)  8 (0.97)  SL  1.36 (0.04)  49 (1.50)  0.70 (0.02)  3.525 (0.48)  8.02 (0.02)  0.530 (0.08)  29.27 (0.59) 
50  77 (3.52)  15 (4.11)  8 (1.53)  SL  1.39 (0.04)  48 (1.57)  0.69 (0.24)  5.942 (2.60)  7.98 (0.04)  0.492 (0.07)  30.69 (0.39) 
60  77 (4.36)  15 (4.09)  8 (0.48)  SL  1.41 (0.05)  47 (1.86)  0.74 (0.19)  6.768 (3.92)  7.95 (0.05)  0.495 (0.02)  28.97 (3.18) 
75  91 (1.63)  2 (1.80)  7 (0.59)  S  1.46 (0.04)  45 (1.54)  0.77 (0.10)  13.761 (4.74)  8.01 (0.11)  0.338 (0.09)  22.91 (5.64) 
The mean (standard deviation) of three replications, SL: Sandy loam, S: Sand, BD: Bulk density, P: Porosity, OM: Organic matter content, Ks: Saturated hydraulic conductivity, EC_{25}: Electrical conductivity at 25 °C, and CaCO_{3}: Calcium carbonate content.
Experimental treatments.
 



 
1  X  X  X  2.962  
2  X  X  X  4.060  
3  X  X  X  ponding  
4  X  X  X  2.962  
5  X  X  X  4.060  
6  X  X  X  ponding  
7  X  X  X  2.962  
8  X  X  X  4.060  
9  X  X  X  ponding  
10  X  X  X  2.962  
11  X  X  X  4.060  
12  X  X  X  ponding 
Unit for application rate is cm h^{−1}, ponding refers to flooding irrigation type of solution application.
Comparison of model characteristics and/or abilities.
Things to simulate  Water flow and movement of nonreactive solutes (Cl, Br), tritium, colloids, and pesticide  Water flow and solute and energy (heat) transport 
State of flow and transport  Steady and nonsteadystate  Steadystate or equilibrium 
Dimension of simulation  Onedimensional  One or twodimensional 
Domain  Macroporous layered soil  Gridded soil domain 
Modeling approach  Single or dualpermeability  Singlepermeability 
Governing equations  Richards Equation for water flow and CDE (convectiondispersion equation) for solute transport  Richards Equation for water flow and CDE (convectiondispersion equation) for solute transport 
Soil hydraulic functions  Van GenuchtenMualem [ 
Van Genuchten [ 
Numerical technique to solve governing equations  Finitedifference  Finitedifference 
Data requirement  Rainfall or irrigation and meteorological parameters (daily min. & max. air temperature, solar radiation, wind speed, and vapor pressure), initial and boundary conditions, and soil hydraulic functions  Geometry of domain, initial and boundary conditions, hydraulic and transport properties of porous medium, soil hydraulic functions, dispersivity, and molecular diffusion 
Model parameters used in the modeling studies.

 

θr 
θs (cm^{3}·cm^{−3})  α (cm^{−1})  n  Ks (cm·h&^{minus;1})  Disp. (cm)  CTEN (cm)  XMPOR (cm^{3}·cm^{−3})  KSM (mm·h^{−1})  ASCALE (mm)  ZN  
10  0.07088  0.36947  0.03932  1.64177  25.710  1  20  0.310  8.55  5  2 
20  0.07046  0.41685  0.06441  1.61921  5.830  1  20  0.310  8.87  5  2 
30  0.07708  0.39738  0.03396  1.70396  27.198  1  20  0.340  7.94  5  2 
40  0.06478  0.46317  0.06360  1.56112  32.878  1  20  0.350  7.11  3  2 
50  0.07102  0.42046  0.03711  1.75928  39.511  1  20  0.350  8.55  5  2 
60  0.06848  0.43008  0.04336  1.65125  38.400  1  20  0.350  8.55  5  2 
75  0.04660  0.23964  0.05583  1.56521  19.193  1  20  0.190  1.48  5  2 
θ_{r}, θ_{s}, α, n, Ks and l: Soil hydraulic function parameters of van Genuchten (1980) and Mualem (1976); θ_{r} and θ_{s}: Residual and saturated SWC, respectively; α and n: Shape parameters of SWRC; Ks and l: Hydraulic conductivity parameters in Mualem (1976) function; Disp.: Dispersivity; CTEN: Boundary soil water tension; XMPOR and KSM: Boundary SWC and boundary hydraulic conductivity, respectively; ASCALE: Effective diffusion pathlength; and ZN: Tortuosity factor in macropores.
Comparison of the means of the measured and simulated SWC and EC, and the solute transport parameters of the treatments using Tukey test.

 

1  SWC 
0.165^{a}  0.227^{c}  0.232^{c}  17.22^{abc}  72.29^{abc} 
EC  0.040^{a}  0.072^{b}  0.085c  
2  SWC  0.170^{a}  0.207^{bc}  0.226^{c}  41.16^{abc}  500.00^{d} 
EC  0.043^{a}  0.068^{b}  0.091^{c}  
3  SWC  0.174^{a}  0.193b  0.204^{b}  41.16^{abc}  500.00^{d} 
EC  0.035^{a}  0.052^{b}  0.089^{c}  
4  SWC  0.194^{ab}  0.212^{bc}  0.221^{c}  21.33^{abc}  357.57^{abcd} 
EC  0.053^{a}  0.069^{b}  0.109^{c}  
5  SWC  0.182^{a}  0.211^{bc}  0.221^{c}  41.16^{abc}  500.00^{d} 
EC  0.052^{a}  0.069^{b}  0.119^{d}  
6  SWC  0.255^{c}  0.198^{a}  0.230^{b}  58.00^{cd}  286.14^{abcd} 
EC  0.077^{b}  0.059^{a}  0.094^{c}  
7  SWC  0.296^{ab}  0.303^{b}  0.283^{a}  37.76^{abc}  78.16^{abc} 
EC  0.080^{a}  0.077^{a}  0.092^{b}  
8  SWC  0.304^{b}  0.337^{c}  0.302^{b}  41.16^{abc}  500.00^{d} 
EC  0.076^{a}  0.083^{a}  0.082^{a}  
9  SWC  0.288^{b}  0.300^{b}  0.299^{b}  34.58^{abc}  475.71^{d} 
EC  0.064^{a}  0.072^{b}  0.099^{d}  
10  SWC  0.294^{a}  0.350^{c}  0.305^{ab}  27.34^{abc}  79.15^{abc} 
EC  0.073^{a}  0.076^{ab}  0.081^{b}  
11  SWC  0.283^{ab}  0.338^{c}  0.268^{a}  9.52^{ab}  12.56^{abc} 
EC  0.066^{a}  0.081^{b}  0.086^{b}  
12  SWC  0.311^{b}  0.314^{b}  0.303^{b}  53.31b^{cd}  289.11^{abcd} 
EC  0.091^{b}  0.074^{a}  0.091^{b} 
SWC: Soil water content, EC: Electrical conductivity, v: Pore water velocity, and D: Dispersion coefficient. The same letters; for a given parameter of Measured, MACRO, VS2DTI, v, or D; indicate the same groups of treatments, implying that there are no statistically significant differences among these groups of treatments in