Predicting soil water balance for irrigated and non-irrigated lucerne on stony, alluvial soils

Efficient use of irrigation is essential for minimising environmental impacts of agriculture, particularly on stony soils which are vulnerable to nitrogen leaching. Models of soil water balance are essential tools for effective agricultural water management. Here we use the soil water flow model HYDRUS-1D to predict evaporation, total 1.5 m soil profile water content, and drainage at 1.5 m soil depth for irrigated and nonirrigated lucerne (Medicago saliva L.) on stony, alluvial soils in central South Island, New Zealand. Modelled estimates were compared against measurements of evaporation by eddy covariance, profile soil water content from soil moisture sensors, and drainage from lysimeters from 1 October 2016 to 1 July 2018 at each site. Cumulative evaporation and drainage were within 4% of measurements and short-term variations were well predicted for non-irrigated lucerne. Under irrigation, modelled cumulative evaporation and drainage overestimated measurements by 6 and 24% respectively. This overestimation must be viewed within the context of uncertainties related to the measurements, particularly given non-closure of the measured water balance related to spatial variability and inherent biases in the measurement methods. Model inputs and concept are likewise sources of uncertainty. Total water content predicted for the 1.5 m depth soil profile was 40-50% greater than the measured volume for both non-irrigated and irrigated lucerne, although short term variations were well predicted, suggesting that further research is needed on estimation of soil hydraulic parameters which represent water storage in these stony, alluvial soils. Our results highlight the value of well-measured sites, such as those of the current study, for evaluating the quality of soil hydrological predictions.