Using HYDRUS-2D model to simulate the water flow and nitrogen transport in a paddy field with traditional flooded irrigation.

In recent years, agricultural non-point source pollution (ANPSP) has become increasingly prominent, and nitrogen plays an important role in ANPSP. Therefore, we carried out traditional flooded irrigation (TFI) experiments in the paddy field, and applied HYDRUS-2D model to simulate the nitrogen transport in this study. Three observation points A1, A2, and A3 were arranged on the diagonal of the paddy field. We observed ponding water depth on soil surface and nitrogen concentrations in ponding water and soil water at 0.1 m, 0.2 m, and 0.3 m below soil surface. HYDRUS-2D model was proved to be effective in simulating the ponding water depth with root mean squared error (RMSE) = 0.717 cm and Nash-Sutcliffe coefficient (NSE) = 0.805 for the simulated and measured ponding water depth. The simulated and measured NH-N concentrations at different depths below soil surface at point A1 basically had the same trend, and the simulated NH-N concentrations in ponding water had better agreement with the measured data with RMSE = 1.323 mg/L, and NSE = 0.958. The measured NH-N concentrations at depths of 0.1 m, 0.2 m, and 0.3 m below soil surface at point A2 were larger than the simulated values, but they had the same trend on the whole. The simulated NH-N concentrations at different depths below soils' surface at point A3 did not fit well with the measured values. The overall trend of the simulated and measured NO-N concentrations in ponding water on soil surface at point A1 was consistent, but the peak values of the simulated NO-N concentrations were larger than the measured ones. The simulated and measured NO-N concentrations at different depths below soil surface at points A2 and A3 did not agree well although they had the same trend, which became worse with the increase of soil depth. This indicated that the HYDRUS-2D model was effective in simulating water flow and nitrogen transport in TFI paddy fields. Sensitivity analysis suggested different simulated nitrogen concentrations in different water depths at different time were sensitive to different model parameters.