Modeling irrigated area to increase water, energy, and food security in semiarid India

Because of declining public investments in irrigation projects in India, the growth of irrigated agricultural production has increasingly become reliant on unsustainable allocation of groundwater. As a result, groundwater resources are increasingly depleted and their role in buffering climate variability is lost. Given future climate and food supply uncertainty under mounting population pressure, it is vital that the connections between climate variability, unsustainable irrigation practices, and their impacts on regional-scale agricultural production are quantified. Here, the focus is on rice and maize production in the semiarid Telangana region in Andhra Pradesh, where the advent of inexpensive pump technology in the late twentieth century, coupled with governmentally subsidized electricity, has allowed year-round planting of water-intensive crops. Using a 35-yr climate and agricultural dataset from Telangana, nonlinear Gaussian process district-level regression models are developed to model dry-season irrigated area, which is a proxy for total groundwater use, in the function of climate-related predictors. The resulting models are able to accurately reproduce dry-season cropped area in most districts. Interannual climate variations play a significant role in determining groundwater use for irrigation. Nonlinear interactions between selected climate features are likely to influence irrigation water use significantly. These results suggest that the authors' modeling approach, combined with monsoon predictions, allow the forecasting of cropped area and agricultural water requirements at seasonal time scales within the bounds of uncertainty. The usefulness of such data to decision makers and stakeholders is discussed, as they attempt to use scarce surface and subsurface water resources more efficiently and sustainably.