View article

The paper presents a robust Darcy‐based one‐dimensional soil water balance model (SAWAH) for simulating riceland hydrology. Experimental data from an upland rice toposequence showing highly variable groundwater and volumetric soil water contents θ are used to calibrate the model and to validate it by a modified cross‐validation method. Analysis of variance is applied to estimate the mean squared prediction error and to split it into errors in predicting horizontal, vertical, and temporal gradients of θ. Root mean squared error of prediction (RMSEP) of soil water content at strip‐depth‐time points was about equal to the standard deviation of unreplicated measurements. Measured strip and field means were more accurate than predicted means, but RMSEP remained below 0.03 m³ m⁻³. Ranking of crop and soil parameters for their effects on root zone water content and seasonal flux totals depended strongly …