Physically Based Green–Ampt Model in Polar Coordinate System Predicting Soil Water Transport in Moistube Irrigation: Comparison of Physical, Semi-Physical-Empirical, and Numerical Models Under Varying Working Pressure Heads

Predicting soil–water dynamics in Moistube irrigation (ΜΤΙ) favours understanding ΜΤΙ functioning mechanisms and technical parameter design. This study proposed a physically based infiltration (PH) model extending the Green–Ampt (GA) model to a two-dimensional polar coordinate system. We treated Moistube as a clay and considered the infiltration from internal Moistube to surrounding soils. The performances of the PH model, together with a semi-physical–empirical (PH–EM) model and a numerical simulation (NUM) model, were evaluated based on regulated working pressure head (WPH) experiments. A HYDRUS 2D model was used based on experimental design to reproduce the soil–water dynamics by assigning Moistube and soil two sets of hydraulic parameters. WPH increase or decrease treatments were applied to Moistube. The Moistube discharge rate, infiltration volume, and wetting front (WF) advance were analyzed and predicted by three models. The results showed that cumulative infiltration, Moistube discharge, and effective saturation around Moistube were enhanced or abated under WPH increase or decrease, with WF accelerating or decelerating. The modelled effective saturation varied between 0.45 and 0.70, providing suitable moist conditions for crops. Percentage of bias (PBIAS) and mean absolute percentage relative error (MAPRE) were employed to evaluate model performances. Three models well-predicted infiltration characteristics and WF advance but differed in accuracy. The PH model overestimated and underestimated the Moistube discharge rate in early and later phases. The prediction accuracy in WF varied across infiltration phases and WPH modes. The PH–EM model yielded accurate results due to its empirical attribute. The NUM model produced novel phenomena of infiltration characteristics at WPH adjustment points, i.e., the discharge rate exponentially decreased over time after the WPH increased but presented restraining followed by rebounding trends after the WPH decreased. The NUM model strongly depended on the selection of the Moistube hydraulic parameters. Extending the GA model to a two-dimensional polar coordinate system by treating Moistube as a clay was practicable in modelling soil water dynamics, thereby contributing to designing and optimizing MTI technical indexes.