Early detection of climate impacts on crop production
More than 60% of farmland in India is rainfed, resulting in high volatility in yields in line with variability in monsoon rainfall. Even more critically, rainfed farming dominates the poorest regions of the country, severely affecting the well-being of some of the most vulnerable communities. Experience and anecdotal evidence suggest that some of the most debilitating impacts can be traced back to deviations from the normal in very early stages of the crop cycle – before, during, and up to four weeks after sowing. Presumably, adequate rainfall during germination and vegetative growth enables crops to better withstand impacts such as long dry spells when they occur later in the crop cycle. This project will use a plot-level data on major crops in India to estimate the impact of early-season rainfall deficits and develop a model to predict crop losses using rainfall data from the first 30 days of the crop season. We will build a model for predicting crop yields and use climate variables to estimate losses in bad years relative to years with normal monsoon rainfall. The dataset represents >300,000 observations from ~45,000 farmers across India over 2008-2020, with information on village location, labour and machine costs, costs of inputs such as seeds, fertilizers, and pesticides, as well as irrigation.
