Erschienen in:
Geoscientific Model Development, 13 (2020) 6, Seite 2723-2742
Sprache:
Englisch
DOI:
10.5194/gmd-13-2723-2020
ISSN:
1991-9603
Entstehung:
Anmerkungen:
Beschreibung:
Abstract. Climate simulations require very complex numerical models.Unfortunately, they typically present biases due to parameterizations,choices of numerical schemes, and the complexity of many physical processes.Beyond improving the models themselves, a way to improve the performance ofthe modeled climate is to consider multi-model combinations. In the presentstudy, we propose a method to select the models that yield a multi-modelensemble combination that efficiently reproduces target features of theobservations. We used a neural classifier (self-organizing maps), associated with a multi-correspondence analysis to identify the models that bestrepresent some target climate property. We can thereby determine anefficient multi-model ensemble. We illustrated the methodology with resultsfocusing on the mean sea surface temperature seasonal cycle in the Senegalo-Mauritanian region. We compared 47 CMIP5 model configurations toavailable observations. The method allows us to identify a subset of CMIP5models able to form an efficient multi-model ensemble. The future decreasein the Senegalo-Mauritanian upwelling proposed in recent studies is then revisited using this multi-model selection.