Monitoring and Quantifying the Variability of Monsoonal Rainfall and its Effects on Rice Yield in South Asia

The increasing severity and frequency of extreme weather events due to climate change poses a major threat to the food security of developing countries. Statistical yield models are trained to estimate the effect of extreme weather events on crop yield. The results from such models are available to advisory services and policymakers, to mitigate future yield losses. However , current statistical yield models use a fixed monsoon season to calculate total seasonal precipitation. The accuracy of climate-crop relationships from current models is therefore hindered by misrepresented variability in monsoonal rainfall.

In this study, we focus on the rainfall variability of the Indian monsoon in statistical yield models. Using a historic gridded weather dataset, we estimate the agronomic monsoon onset for 341 districts over the period 1981 - 2016. We find that there is large spatial and temporal heterogeneity in the date of monsoon onset in India at the district level. Failure to account for such heterogeneity leads to errors in estimates of seasonal rainfall totals that are a key input to statistical yield models.

We compare the rice yield variability explained by fixed and variable monsoon onset dates using district-level yield data. Our findings show how the representation of monsoonal rainfall variability in statistical yield models is affected by accurate onset estimation. We highlight the importance of accurately estimating monsoon onset in both monitoring its historic variability and quantifying its effects on future yield.