Using Satellite Data to Investigate the Relationship Between Field Size and Agricultural Productivity in Eastern Uttar Pradesh, India

Abstract

This study investigates the relationship between field size and agricultural productivity in smallholder farming systems, focusing on three districts in Eastern Uttar Pradesh, India. We employ a Mask R-CNN image segmentation model and high-resolution satellite imagery to predict field boundaries, and estimate crop yield using maximum NDVI from Sentinel-2 imagery during the winter wheat growing season. Our findings reveal a slight positive relationship between field size and productivity across 7 of 8 zones in the study area when controlling for village-level effects. This is contrary to the consensus of an inverse productivity-size relationship in smallholder farming systems. This study provides a framework to investigate the relationship between productivity and field size at a large and dense scale, without the use of self-reported yield data. Limitations include the reliance on maximum NDVI as a yield proxy and the spatial resolution of Sentinel-2 imagery. Future research should explore alternative yield metrics and higher-resolution satellite data to refine our understanding of the productivity-size relationship in smallholder farming systems.