Deep Learning in Rice Yield Prediction G. S. V. M. Ishan 2020/

Abstract

ABSTRACT Rice, being the staple food across much of Asia, holds paramount importance in countries like Sri Lanka. The yield of paddy rice is profoundly affected by climate variations, making accurate forecasting crucial for ensuring food security. However, predicting rice yield entails navigating through intricate non-linear relationships between climate factors and agricultural output. In this study, I leverage Convolutional Neural Networks (CNNs) to forecast rice yield using extensive historical weather and yield data from 25 key rice-producing cities in Sri Lanka. Our analysis encompasses a comprehensive array of climate variables, including temperature, radiation, precipitation, wind speed, and more, spanning the years 2004 to 2023. By employing CNNs, I demonstrate the efficacy of this advanced machine-learning technique in unraveling the complexities of rice yield prediction. This approach here not only provides valuable insights into the interplay between climate dynamics and rice cultivation but also offers a powerful tool for policymakers in formulating effective agricultural policies in Sri Lanka. This study underscores the significance of CNNs in enhancing rice yield prediction accuracy, thereby contributing to the sustainable management of food resources in Sri Lanka and beyond. Keywords: Rice yield prediction, Convolutional Neural Networks, Weather data, Climate, Sri Lanka, Agricultural forecasting, Deep Learning, Paddy Yield