Experimental evaluation of different subsurface water movement models in an agricultural field with soil heterogeneity

Abstract

Spatial heterogeneity in soil hydraulic properties poses severe challenges to the modeling of soil moisture dynamics in agricultural fields. For simplicity, 1-D water flow models that ignore soil heterogeneity are commonly used to simulate water movement in agricultural fields. As soil hydraulic properties are one of the most crucial input data, it is particularly desirable to investigate how the simulations of these water flow models compare with experimental observations. This study aims to examine the performances of streamtube, 1-D, and 3-D Richards equation based water flow models with experimental observations. Experiments were conducted at an agricultural plot of 20 m × 30 m located in IIT Kanpur, India, for 1 year under wheat (Triticum aestivum L.) and rice (Oryza sativa L.) crop covers. The site was heavily instrumented to monitor soil moisture variation, weather conditions, root depth, and LAI. Further, 54 undisturbed soil samples were analyzed to determine the heterogeneity in soil hydraulic parameters. The results indicate that the performance of the models varies with land cover, atmospheric boundary condition, and crop growth stages. Besides, the difference between model estimates varies with soil moisture wetness and a maximum relative error of 13% was observed under wetting conditions. The streamtube model showed the best performance under wheat crop cover for all boundary conditions and growth stages, as it considers spatial variability in soil hydraulic properties and irrigation application at each subplot. The numerical-1D model was the best predictor under rice crop cover. Further, the results suggest that an increase in model complexity does not lead to an improvement in water simulation, and accurate representation of soil hydraulic properties plays a more important role.