Weight assignment in regional climate models

Abstract

An important new development within the European ENSEMBLES project has been to explore performance-based weighting of regional climate models (RCMs). Until now, although no weighting has been applied in multi-RCM analyses, one could claim that an assumption of 'equal weight' was implicitly adopted. At the same time, different RCMs generate different results, e.g. for various types of extremes, and these results need to be combined when using the full RCM ensemble. The process of constructing, assigning and combining metrics of model performance is not straightforward. Rather, there is a considerable degree of subjectivity both in the choice of metrics and on how these may be combined into weights. We explore the applicability of combining a set of 6 specifically designed RCM performance metrics to produce one aggregated model weight with the purpose of combining climate change information from the range of RCMs used within ENSEMBLES. These metrics capture aspects of model performance in reproducing large-scale circulation patterns, meso-scale signals, daily temperature and precipitation distributions and extremes, trends and the annual cycle. We examine different aggregation procedures that generate different inter-model spreads of weights. The use of model weights is sensitive to the aggregation procedure and shows different sensitivities to the selected metrics. Generally, however, we do not find compelling evidence of an improved description of mean climate states using performance-based weights in comparison to the use of equal weights. We suggest that model weighting adds another level of uncertainty to the generation of ensemble-based climate projections, which should be suitably explored, although our results indicate that this uncertainty remains relatively small for the weighting procedures examined.

Exploring metrics of extreme daily precipitation in a large ensemble of regional climate model simulations

“…The final weights differed by a factor of 1.3 between the best and worst model. These weights have been used in the final ENSEMBLES weighting scheme, which is discussed in Christensen et al (2010).…”

“…The intended use of these weights is to quantitatively weight model results according to their performance, and thus obtain more precise climate change predictions. How this can be accomplished is not a topic of this paper, but is discussed by Christensen et al (2010) elsewhere in this CR Special issue.…”

“…In the final ENSEMBLES weighting system (Christensen et al 2010), the metric is used to compute weighted average results. For this purpose, the final weights of each model are normalized such that the sum of weights of the RCMs is 1.…”

Projected Change—Models and Methodology

“…Samuelsson et al 2011) or for a large number of models (e.g. Christensen et al 2010). However, studies on RCM performance focusing on the Baltic Sea region remain few (e.g.…”

“…10.2) for the Baltic Sea region with respect to the daily gridded observational data set based on European Climate Assessment & Dataset information (E-OBS) (Haylock et al 2008). Nine RCMs were used: C4IRCA3, KNMI-RACMO2, DMI-HIRHAM5, ETHZ-CLM, HadRM3Q0, HadRM3Q16, MPI-REMO, Had-RM3Q3, and SMHIRCA (for documentation on the individual models, see Christensen et al 2010; data are available from http://ensemblesrt3.dmi.dk/). The maps show gridpoint-wise model performance, and as an estimate of the spread, the nine sets of results for each grid point are sorted resulting in an approximate 5th percentile corresponding to the lowest value, a median, and an approximate 95th percentile corresponding to the largest value.…”