Quantifying risk for water harvesting under semi-arid conditions

Risk assessment of maize yield was carried out using a crop growth model combined with a deterministic runoff model and a stochastic rainfall intensity model. These were compared with empirical models of daily rainfall-runoff processes. The combination of the deterministic runoff model and the stochastic rainfall intensity model gave more flexible performance than the empirical runoff model. Scenarios of crop simulation included production techniques (water harvesting, WH, and conventional total soil tillage, CT) and initial soil water content at planting (empty, half and full). The in-field water harvesting technique used in the simulation was a no-till type of mini-catchment with basin tillage and mulching. The lower the initial soil water content at planting, the greater the yield difference between the WH and CT production techniques. With the low initial soil water content at planting, the WH production technique had up to 50% higher yield compared to the CT production technique, clearly thus demonstrating the superiority of the WH production technique. Under all the variations in agronomic practices (planting date, plant population, cultivar type) tested, the WH had a lower risk than CT under these semi-arid climatic conditions (i.e., WH increased the probability of higher crop yields).