The field of weather forecasting and climate modeling is rapidly advancing with the integration of artificial intelligence (AI) and machine learning (ML) techniques. Recent developments have focused on improving the accuracy and efficiency of forecasting models, with a emphasis on real-time processing and scalability. Researchers are exploring various AI-driven approaches, including convolutional neural networks (CNNs) and transformer-based architectures, to enhance the performance of weather forecasting systems. These innovations have the potential to revolutionize the field, enabling more accurate and reliable predictions, and improving our understanding of complex climate phenomena. Notable papers in this area include those that propose lightweight and efficient models for on-board satellite processing, fully AI-driven global weather forecasting systems, and novel benchmarks for 3D atmospheric structure recovery.

Noteworthy papers in this area are: XiChen, which introduces a fully AI-driven global weather forecasting system that can accomplish data assimilation and medium-range forecasting within 17 seconds. FourCastNet 3, which implements a scalable geometric machine learning approach to probabilistic ensemble forecasting, delivering forecasting accuracy that surpasses leading conventional ensemble models. EPT-2, which presents a foundation AI model for Earth system forecasting that sets a new state of the art in predicting energy-relevant variables and significantly surpasses the ECMWF ENS mean in medium- to long-range forecasting.