Feasibility of machine learning-based rice yield prediction in India at the district level using climate reanalysis and remote sensing data

CONTEXT: Yield forecasting, the science of predicting agricultural productivity before the crop harvest occurs, helps a wide range of stakeholders make better decisions around agricultural planning. OBJECTIVE: This study aims to investigate whether machine learning-based yield prediction models can capably predict Kharif season rice yields at the district level in India several months before the rice harvest takes place. METHODOLOGY: The methodology involved training 19 machine learning models such as CatBoost, LightGBM, Orthogonal Matching Pursuit, and Extremely Randomized Trees on 20 years of climate, satellite, and rice yield data across 247 of India's rice-producing districts. In addition to model-building, a dynamic dashboard was built understand how the reliability of rice yield predictions varies across district. RESULTS AND CONCLUSIONS: The results of the proof-of-concept machine learning pipeline demonstrated that rice yields can be predicted with a reasonable degree of accuracy, with out-of-sample R2, MAE, and MAPE performance of up to 0.82, 0.29, and 0.16 respectively. This performance outperformed test set performance reported in related literature on rice yield modelling in other contexts and countries. In addition, SHAP value analysis was conducted to infer both the importance and directional impact of the climate and remote sensing variables included in the model. Important features driving rice yields included temperature, soil water volume, and leaf area index. In particular, higher temperatures in August correlate with increased rice yields, particularly when the leaf area index in August is also high. Building on the results, a proof-of-concept dashboard was developed to allow users to easily explore which districts may experience a rise or fall in yield relative to the previous year. The dashboard show that the model may perform better in some regions than in others. For instance, the absolute percentage error for predicted versus actual yields ranged from an average of 7.1 % in districts in Uttarakhand to an average of 14.7 % in Uttar Pradesh. SIGNIFICANCE: This study underscores the potential for policymakers to consider scaling and operationalizing machine learning approaches to rice yield prediction in the context of agricultural early warning systems to deliver timely crop yield forecasts on a rolling basis throughout the season, thereby equipping agricultural decision-makers with the ability to make informed choices on irrigation scheduling, fertilizer application, and harvest planning to optimize crop output and resource use.