Ensemble climate projections are a vital tool for understanding historical and future changes, informing regional and local modeling efforts, as well as for conducting impact assessments. However, using all available models from a multi-model ensemble such as CMIP6 is often not feasible and does not necessarily provide the best possible representation of climate, its changes, and its uncertainties.
Therefore, many methods have been developed to aggregate, weight, filter, constrain, and sub-select models in recent years. These methods are based on internal model consistency, comparing models with observations, or an assessment of model inter-dependencies among other things. They cover a range of spatial scales from global to local and temporal scales from near-term predictions to long-term projections.
This session welcomes contributions focusing on, but not limited to, aggregating, weighting, filtering, constraining, and sub-selecting multi-model ensembles. This includes:
- application-oriented work including model constraining and sub-selection for impacts and regional applications
- work on observational and emergent constraints, weighting and filtering approaches, as well as qualitative and quantitative model sub-selection
- work on methods to aggregate multi-model ensembles and quantify uncertainties using classical statistics as well as machine learning
- verification of model constraining and sub-selection methods using, for example, out-of-sample historical observations, pseudo-observations, or explainable artificial intelligence (XAI) methods