Dynamic Irrigation Management under Temporal and Spatial Variability

Recent studies report that agricultural productivity must improve significantly in the near future to meet the increase in demand due to rapid growth of the world population. One way to improve agricultural productivity is to increase irrigation. However, fresh water scarcity, increasing production costs, and the climate change necessitate sustainable and efficient methods for irrigation, which accounts for nearly seventy percent of global fresh water usage. This paper studies dynamic irrigation management under uncertainty taking into account temporal variability and spatial soil heterogeneity of the field. We show that an (s, S)-type irrigation policy is optimal for the soil structure that is radially homogeneous. Building on this result, we also derive performance bounds and effective heuristic policies for the general case that allows radial soil heterogeneity. Our numerical study reveals that the proposed dynamic irrigation policies can improve farmers' expected net benefit (yield revenue net of irrigation cost) and water savings significantly over the commonly used irrigation policies in practice.