Crop Yield Analysis using SIF and Climate Variables: A Case Study in Punjab, India

Abstract

Regular and faster crop yield prediction can mitigate the extreme effects of severe weather events, such as drought, heavy rainfall, etc. This work explores a precise, scalable, and automatic way to understand rice yield dynamics and its correlation with satellite-based solar-induced fluorescence (SIF), climate variables, such as temperature and rainfall, as they exhibit a significant correlation with rice yield. A district-wise analysis in Punjab, India, is carried out using Pearson correlation coefficient and different regression techniques, such as linear, ridge, lasso, and elastic net for the Kharif season. A comparative study shows that the elastic net performs better than the other models, with the best coefficient of determination (R2) of 0.792 and root mean square error (RMSE) of 300.5 kg/ha. This study can be extended in multiple dimensions by including a variety of crops, climate factors, and multi-satellite SIF data for any crop yield pattern analysis and prediction. © 2022 IEEE.