Evaluation of three simple hydrosalinity models applied to citrus orchards in the Lower Coerney River irrigation area, Eastern Cape, South Africa

Abstract

Three hydrosalinity models of different complexity were evaluated against data from selected citrus orchards in the Lower Coerney River irrigation area. These models were statistically compared with field data, and found to vary in accuracy of output predictions. The simplest model, the LEACHING REQUIREMENT (LR) model, has the lowest data input requirements and produces output predictions that correlate with up to 93% of the variance in measured data. SODICS, which is theoretically more detailed and requires a greater level of detail in input, produces predictions of an accuracy equivalent to the LR model. The PEAK model was assessed in two parts, the PEAKM module simulating soil moisture and soil moisture movement was able to predict variations in moisture up to 99% of the time. The PEAKD module, which predicts solute concentrations was less effective in replicating real world conditions