Modeling percolation losses from a ponded field under variable water-table conditions

A numerical simulation model was developed to predict the vertical and lateral percolation losses from a ponded agricultural field. The two-dimensional steady-state unsaturated/saturated flow equation was solved using the finite-difference technique. A constant ponding depth was maintained at the soil surface with different water table conditions in an application of the model for rice fields bordered by bunds. Field experiments were conducted for two different water table depths to collect data on the spatial distribution of volumetric soil-moisture content for model verification. The measured soil-moisture content values were found to be in close agreement with those predicted by the model. The sensitivity analysis of the model with selected hydrologic conditions shows that the model is most sensitive to the values of saturated hydraulic conductivity, but relatively less sensitive to water table depth, ponding depth, and evaporation rate from the soil surface. It implies that, in a ponded rice field condition, the lateral and vertical percolation losses are mostly governed by the hydraulic conductivity of the soil. The vertical percolation losses were almost equal to the saturated hydraulic conductivity values and, in most cases, these losses increased with deeper water table depths. The lateral percolation losses also increased with deeper water table depths; however, these losses were relatively small in comparison to the vertical percolation losses. The vertical and lateral percolation losses increased with the increase in ponding depths. The lateral percolation losses through the bund decreased when the evaporation losses increased from the soil surface. The results of this study indicate that the percolation losses from a ponded field may be predicted accurately for a wide range of soil and hydrological conditions when the values or hydraulic conductivity, evaporation rate, depth of ponding, and water table depth are accurately known.