A EURASIAN CASE STUDY ON CLIMATE CHANGE AND AGRICULTURAL LAND SUITABILITY USING INTERPRETABLE MACHINE LEARNING

Abstract

As a result of climate change's impact on the appropriateness of agricultural land, this research seeks to solve the pressing worldwide problem of food security. Predicting the risks of land suitability deterioration and changes in irrigation patterns, which directly effect food security, is the fundamental focus of our study. The fight against hunger and malnutrition is one of the United Nations' sustainable development objectives, and this study fits in with that. Our research focusses on Central Eurasia because it is a prime example of an area dealing with distinct social and economic issues; this makes it ideal for studying how climate change affects food security. We use interpretable machine learning methods to assess how various carbon emission scenarios would affect the usability of agricultural land in the face of climate change. The generated model performs well in a multi-class land suitability classification test, with an accuracy of 86% and a mean average precision of 72%. Our study gives policymakers important insights into the most susceptible locations in Northern Asia and Eastern Europe. Insights like these are crucial for humanitarian crisis prevention strategy, which includes allocating vital resources like water and fertilisers. The findings prove that machine learning is a potent instrument for foreseeing and controlling the effects of climate change on food security.