Subseasonal-to-Seasonal and Seasonal-to-Decadal (S2S/S2D) forecasting (from two weeks and a decade ahead) is a rapidly developing area of forecasting, with the potential to provide valuable information for the development of climate services. Although S2S/S2D climate predictions have a comparative lack of skill, over the past decade there has been a substantial research effort to improve prediction skill via novel advanced Statistical, Machine and Deep Learning Methods either in terms of post-processing of the dynamical model output or data-driven models. Additionally, there is a strong interest in understanding predictability on S2S scales using eXplainable Artificial Intelligence (XAI) which could help to improve forecast skill.

 

This session welcomes all aspects of improving forecasting on  S2S/S2D scales including novel statistical, Machine and Deep Learning methods based post-processing (bias correction, multi-model ensemble) of the dynamical model output, and ML/DL models based on teleconnections (empirical/data driven). Abstracts that explore XAI for predictability are also encouraged.