Probing Vegetation, Climatic Data, and Machine Learning for Agricultural Planning and Climate Action: A Case Study from North India

This research investigates relationships among vegetation indices (VIs), climatic variables (CVs), and crop productivity and applies machine learning models vis-a-vis linear regression. Mathura, a district in Uttar Pradesh, North India, is an agricultural hotspot and heavily populated; it is imperative to probe the interplay of various factors affecting crop yield. We extracted normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) from the MODIS (MOD13Q1) dataset, nine CVs from the CRU climate data, and wheat data from India’s Ministry of Agricultural and Farmers Welfare. We found that NDVI consistently reflects higher values than EVI during the Rabi season. Both VIs exhibit increasing trends, including seasonal and monthly fluctuations, meaning a continuous increase in vegetation over 20 years. However, a strong negative correlation exists between VIs and diurnal temperature range, indicating the difference between day and night temperatures is associated adversely with the vegetation. On the contrary, a strong positive correlation is observed between VIs and precipitation. Additionally, the study compared the effectiveness of a multiple linear regression (MLR) model and a random forest (RF) machine learning model in predicting yield variability. The RF model, considering years, all nine CVs, and two VIs, demonstrated higher predictive efficiency than the MLR model. The findings highlight the importance of integrating climatic and vegetation data in agricultural planning and climate action. The research lays a robust foundation for more detailed future studies to build knowledge and capacity to address climate change using various machine learning models and integrating VIs and CVs.