DEEP LEARNING-BASED ANALYSIS OF THE RELATIONSHIPS BETWEEN CLIMATE CHANGE AND CROP YIELD IN CHINA
Keywords: Climate Change, Grain Yield, Rice Yield, Food Security, Deep Learning
Abstract. Climate change is an important factor in vegetation growth, and it is very significant to understand the relationship between climate change and rice yield. China is a food-importing country whose grain consumption is higher than grain production, and which relies on imports of rice, soybean, wheat and other grains. Therefore, in order to secure food security for 1.6 billion people in China, it is necessary to grasp the relationship between climate change and rice yield. In this study, 16 administrative districts in China were selected and designated as study area. This study used annual rice production from the USDA (United States Department of Agriculture) for each of China’s major administrative regions from 1979 to 2009, as well as average climate data from July to August, which were meteorological observations collected from the CRU (Climate Research Unit). Using this data, the rice crop was increased in 10 administrative regions in China and the reduction in rice harvest in 6 administrative areas was confirmed. The relationship between selected rice production and climate change was nonlinear and modelled using a deep neural network, and the validation statistics showed that the performance of DNN was 32-33% better than that of MLR (multiple linear regression). Therefore, a more quantitative analysis of the relationship between climate change and rice yield changes has been made possible through our prediction model. This study is expected to contribute to better food self-sufficiency in China and forecast future grain yields.