Search

We have created a performance-based assessment of CMIP6 models for Europe that can be used to inform the sub-selection of models for this region. Our assessment covers criteria indicative of the ability of individual models to capture a range of large-scale processes that are important for the representation of present-day European climate. We use this study to provide examples of how this performance-based assessment may be applied to a multi-model ensemble of CMIP6 models to (a) filter the ensemble for performance against these climatological and processed-based criteria and (b) create a smaller subset of models based on performance that also maintains model diversity and the filtered projection range as far as possible. Filtering by excluding the least-realistic models leads to higher-sensitivity models remaining in the ensemble as an emergent consequence of the assessment. This results in both the 25th percentile and the median of the projected temperature range being shifted towards greater warming for the filtered set of models. We also weight the unfiltered ensemble against global trends. In contrast, this shifts the distribution towards less warming. This highlights a tension for regional model selection in terms of selection based on regional climate processes versus the global mean warming trend. [ABSTRACT FROM AUTHOR]

We have created a performance-based assessment of CMIP6 models for Europe that can be used to inform the subselection of models for this region. Our assessment covers criteria indicative of the ability of individual models to capture range of large-scale processes that are important for the representation of present-day European climate. We use this study to provide examples of how this performance-based assessment may be applied to multi-model ensemble of CMIP6 models to a) filter the ensemble for performance against these climatological/processed-based criteria and, b) create a smaller sub-set of models based on performance, that also maintains model diversity and the filtered projection range as far as possible. Filtering by excluding the least realistic models leads to higher sensitivity models remaining in the ensemble as an emergent consequence of the assessment. This results in both the 25th percentile and the median of the projected temperature range being shifted toward greater warming for the filtered set of models. We also weight the unfiltered ensemble against global trends. In contrast this shifts both the distribution of towards less warming. This highlights a tension for regional model selection in terms of selection based on regional climate processes versus the global mean warming trend. [ABSTRACT FROM AUTHOR]