A 2XCO2 climate change scenario over Europe generated using a limited area model nested in a general circulation model: 1. Present‐day seasonal climate simulation

Abstract

In this and the companion paper by Giorgi et al. (this issue) we present a regional climate change scenario for Europe and the western Mediterranean basin induced by doubling of carbon dioxide concentration as produced with a limited area model (LAM) nested in a general circulation model (GCM). In this paper we analyze the seasonal climatology of the present‐day climate simulation which is used as a control run in the generation of the scenario. The GCM and the LAM employed for this study are versions of the community climate model (CCM) of the National Center for Atmospheric Research (NCAR) and the NCAR/Pennsylvania State University mesoscale model (MM4). The CCM simulation generally reproduces the basic seasonal migration patterns of the storm tracks which affect the European region but also shows significant biases: (1) a general weakening and southward shift of the cold season North Atlantic jet; (2) cold sea surface temperature bias of up to 4°–6°C in the North Atlantic and the Mediterranean; (3) cold tropospheric temperature bias of several degrees; (4) low tropospheric relative humidity bias; and (5) underprediction of July precipitation, resulting in low soil moisture conditions and exceedingly high surface temperatures. Overall, the nested MM4 produces large‐scale circulations similar to those of the driving CCM and, because of its finer representation of topography and coastlines, substantially improves the spatial distribution of precipitation and surface air temperature not only at high resolution but also at resolutions close to the CCM's. Among the main biases in the nested MM4 simulation are (1) cold bias over land of 1°–5°C in January, April, and October and warm bias in July; (2) underprediction of precipitation over most of continental Europe, maximum in July (simulated precipitation is 60–80% lower than observed) and less pronounced in January, April, and October (simulated precipitation is 10–40% lower than observed). A strong contribution to these biases is given by errors in the driving CCM tropospheric fields. Overall, temperature and precipitation are better simulated in the colder seasons than in summertime and the seasonal cycle is captured better over the Mediterranean regions than over the continental interior.