Advances in Computational Modeling for Climate Change Prediction Based on Machine Learning

Abstract

Climate change prediction is an important tool for addressing global environmental challenges and making scientific decisions. Traditional physical models such as general circulation models (GCMs) have been widely used for climate prediction, but they have significant limitations in high temporal and spatial resolution prediction and simulation of extreme weather events. In recent years, machine learning methods have provided new solutions for climate prediction with their powerful nonlinear feature modelling capabilities and data-driven properties. In this paper, we systematically review the research progress of climate change prediction based on machine learning, focusing on analysing the applicability, advantages and disadvantages, and key technologies of traditional machine learning methods, such as decision trees, support vector machines, and random forests, deep learning methods, such as convolutional neural networks, recurrent neural networks, and long and short-term memory networks, as well as cross-domain fusion methods, such as the combination of machine learning and physical models, and reinforcement learning and transfer learning. Meanwhile, key challenges in the field of climate prediction are explored, including data scarcity, technical barriers to multimodal data fusion, insufficient model interpretability, and generalization capability issues for cross-region and cross-time prediction. Combined with application scenarios such as agriculture, energy and disaster warning, this paper summarizes the practical application effects of different methods and proposes future optimization directions, such as high-quality data acquisition and preprocessing, model lightweight design, interpretability enhancement and the potential of interdisciplinary cooperation. This paper aims to provide comprehensive technical references and development suggestions for climate.