Investigation of Saline Drainage Water Reuse Effect on Soil Hydraulic Properties Using Inverse Method (Case Study: Moghan Plain)

1. The shortage of water resources of good quality is becoming an important issue in the arid and semi-arid zones, hence using of water resources of marginal quality such as drainage water has become an important consideration. In this study, saline drainage water reuse effect on soil hydraulic properties was investigated in Moghan plain. This objective was examined by Levenberg-Marquart optimization algorithm [1] for inverse modeling to estimate some hydraulic properties of soil in transient condition along with forward model (HydroGeoSphere) as a numerical code to simulate water flow in unsaturated porous media based on Richard's equation. 2. Methodology2.1. Study area and experimental device: Study area is located at Moghan plain in northwest of Iran. Undistributed soil samples were taken by cylinders with inner diameter of 30 cm and height of about 50 cm. Treatments were T1 (irrigation with canal water), T2 (irrigation with mixture of 30% drainage water and 70% canal water) and T3 (irrigation with mixture of 50% drainage water and 50% canal water) at three replications in a completely random design. Artificial rainfall experiments were conducted on top of the columns and free drainage from the bottom of columns was measured in the laboratory. Rainfall intensity was controlled by a rotary pump connected to a raindrop maker that produces water drops. 2.2. Numerical model description: HydroGeoSphere solves the three dimensional modified form of Richard's equation for variably-saturated flow using a Galerkin finite element approach [2]. The finite element grid was generated automatically using the pre-processor GRID BUILDER [3]. Grid independency was implemented and the discretization in the vertical direction was yielded 1.5 cm. The intent of this discretization was to be able to resolve vertical water flow with precision. In this study, we developed an inverse method for estimating parameters based on Levenberg-Marquardt minimization algorithm in C++ programming language. Inverse model was used to estimate the saturated hydraulic conductivity (Ks), and soil water retention function parameters of van Genuchten (α, β, and θr) which were unknown parameters in the unsaturated porous media using measured artificial rainfall and free drainage.3. Calibration periods showed that the estimated free drainage using the optimized parameters exhibits a good match with the observed free drainage for all treatments. Hence one could conclude that Richard's equation along with van Genuchten's retention functions can successfully describe the unsaturated water flow in the treatments. Statistical analysis showed that the calibration is sensitive to β more than the other parameters. The Ks decreased significantly in T3 (P < 0.05) but there was no significant difference (P < 0.05) between T1 and T2. The results showed that parameters of van Genuchten have not been changed significantly in all treatments with application of drainage water reuse at probability level of 50% (Table 1).Table 1. Estimated hydraulic properties of treatments (Ks: saturated hydraulic conductivity, θs: saturated water content, θr: residual water content, α and β: shape parameters)β α(cm-1) θr(cm3 cm-3) Ks(cm min-1) Treatment1.21a 0.005a 0.143a 0.220a* T11.20a 0.005a 0.140a 0.226a T21.22a 0.006a 0.150a 0.166a T3* Same letter in column indicates no significant difference (P < 0.05)4. Due to hysteresis effect, soil water retention curve has two desorption and sorption branches. Ordinarily, the desorption curve is measured by gradually and monotonically extracting water from initially saturated samples in the laboratory [4]. But the sorption curve is essential for modeling water and solute transport in unsaturated porous media. To overcome these problems indirect methods such as inverse method can be used to identify the basic flow and transport parameters. Hydraulic properties were estimated by inverse method at three replications in a completely random design. Estimated hydraulic properties succeeded to reproduce the observed free drainage in the transient condition, indicating van Genuchten functions along with Richard's equation can be used to simulate water flow in the treatments. The results illustrated that irrigation with the mixture of 30% of drainage and 70% canal water (T2) did not affect soil hydraulic properties significantly (P < 0.05) in the fine texture soil of study area after one year.