Climate Forecasting

Climate forecasting aims to predict future weather and climate patterns, improving disaster preparedness and resource management. Current research emphasizes developing advanced machine learning models, including deep learning architectures like UNet++, masked autoregressive models, and multimodal large language models, to enhance forecast accuracy and resolution across various timescales (from daily weather to seasonal and annual climate). These models are increasingly incorporating diverse data sources, such as meteorological raster data and textual event descriptions, to improve prediction of both numerical variables and open-set climate events. The resulting improvements in forecasting accuracy have significant implications for mitigating climate change impacts and informing effective policy decisions.