Agricultural impacts of monsoon variability across various spatial scales - a case study of groundnut yield in India.

Andrew Challinor, Julia Slingo; Dept of Meteorology, University of Reading; Tim Wheeler; Dept of Agriculture, University of Reading.

The continuing development of numerical crop and weather simulation models presents an opportunity to combine these models into a single forecasting system. For such a system to have any skill, the empirical link between crop yields and weather variables should be detectable at the spatial scale on which the models operate. Hence it should be possible to upscale small-scale yield measurements so that they can be related to the large-scale fields of the type predicted by General Circulation Models (GCMs).

The continuing development of numerical crop and weather simulation models presents an opportunity to combine these models into a single forecasting system. For such a system to have any skill, the empirical link between crop yields and weather variables should be detectable at the spatial scale on which the models operate. Hence it should be possible to upscale small-scale yield measurements so that they can be related to the large-scale fields of the type predicted by General Circulation Models (GCMs).

Accordingly, an analysis of the relationship between groundnut yield data (from the International Crops Research Institute for the Semi-Arid Tropics in India) and monthly rainfall data (from the Indian Institute for Tropical Meteorology) is presented. The yield data are on the district level, and upscaled subdivisional values are compared with the subdivisional rainfall data. This highlights the regional variability of the yield response to rainfall.

The spatial structure of the relationship between rainfall and groundnut yield has been explored using empirical orthogonal function (EOF) analysis. A coherent, large scale pattern emerges for both rainfall and yield. On the subdivisional scale, the first Principal Component of rainfall is well correlated with the first PC of yield (r^2=0.53, p<10^-4), demonstrating that the large scale patterns picked out by the EOFs are correlated. Further, district-level EOFs of the yield data demonstrate the validity of upscaling these data to the subdivisional scale. Similar patterns are produced using data on both these scales, and the first PCs are very highly correlated (r^2=0.96). Hence a common spatial scale ($\sim 300$km) has been identified, typical of that used in seasonal weather forecasting, which can form the basis of future crop modelling work for the case of groundnut production in India.