Model intercomparison for the present day, the mid-Holocene, and the Last Glacial Maximum over western Europe

Abstract

The focus of this paper is an evaluation of the performance over western Europe of an ensemble of General Circulation Models (GCMs) used to simulate climates at the present-day, the mid-Holocene and the Last Glacial Maximum (LGM). Experiments are considered from the Palaeoclimate Modeling Intercomparison Project (PMIP), and from more recent sensitivity studies for the same time slices. Surface air temperature and precipitation are compared between models, and with observed and proxy data sets, using evaluation measures such as correlation coefficients and root mean squared errors. Hierarchical and k-means cluster analysis are used to quantitatively assess relative model performance. Comparisons are made for western Europe for a gridded domain between 38N13W (south west corner) and 65N30E (north east corner), and for four transects within this domain. Inter-experiment comparisons reveal that the ranges of results from the sensitivity studies are within those of the PMIP models. There is no month-to-month or time slice-to-time slice consistency in the clustering patterns produced, although one clear characteristic is that experiments performed with models from the same institution tend to cluster. Inter-model agreement at past time slices is considerably reduced when we consider differences from the present-day rather than absolute values, and this is particularly the case for mid-Holocene temperature, when the insolation forcing only represents a small departure from modern values. Statistical comparisons of the models with observed and proxy data sets demonstrate a lack of consistency in model performance between months, transects and time slices. Copyright 2004 by the American Geophysical Union.