Arctic clouds containing both liquid droplets and ice crystals are difficult to simulate. We tested how ice-related changes in a weather model, designed for temperate regions, perform in the Arctic. Ice crystal shape is the dominant factor: making crystals spherical substantially reduces cloud ice and shifts it to snow. These modifications are moderate in temperate regions but stronger in the Arctic. As this rests on a single case, they should be validated across diverse cases and climates.

Due to the limitations of past data-based models and the high cost of numerical weather prediction computing, accurately forecasting precipitation proximity remains challenging. A dual encoder-decoder framework is proposed to enhance short-term forecasting and reduce underestimation in extreme precipitation by using spatio-temporal information conversion equations and adaptive weighted gradient loss. Demonstrates better accuracy than existing deep learning methods in precipitation datasets.

Reduced-complexity models are an important tool in climate science, helping us understand and estimate future climate change. We present the experimental protocol for the next phase of the reduced-complexity model intercomparison project, which aims to compare results from many such models to better understand their behaviour. This knowledge will guide how these models are developed and used in the future, including in the upcoming Intergovernmental Panel on Climate Change (IPCC) assessment report (AR7).