Verification and analysis of a climate simulation of temperature and pressure fields over Norway and Svalbard

The monthly mean 2 m temperature and sea-level pressure (SLP) fields from t he most recent integration (GSDIO) with the Max Planck Institute's global c oupled climate model ECHAM4/OPYC3 are compared to historical data over Norw ay including Svalbard. For temperature, observations from selected stations are directly compared to values at grid points nearby. For SLP, modelled a nd observed gridded fields over the area 20 degrees W-40 degreesE, 50-85 de greesN are compared by means of empirical orthogonal function (EOF) analysi s. Finally, the connections between SLP and temperature over Norway are ded uced for both historical data and results from the GSDIO integration and th en compared to each other. The GSDIO 'control climate' grid point temperatu res over Norway are in most cases found to be realistic, whenever it is pos sible to find stations with similar altitude and distance from the coast. T he GSDIO 'future climate' indicates an annual mean warming of 0.2 to 0.5 de greesC decade(-1) on the Norwegian mainland, and 0.8 degreesC decade(-1) at the Svalbard grid point up to 2050. The strongest warming is simulated in winter, in the inland, and at high latitudes. The GSDIO 'control climate' S LP gradients imply on average westerly winds over Norway that are too weak. The GSDIO 'future climate', however, indicates an increase in the westerly wind component. Observations after 1960 show an increase in the westerly f ield of the same magnitude as in the GSDIO results during the same period. The observed connections between atmospheric circulation and temperatures i n Norway are satisfactorily reproduced in the GSDIO integration, especially in winter. The winter warming in the GSDIO integration may partly be expla ined by the increase in the westerly wind component. On the Norwegian mainl and, a linear regression model based on atmospheric circulation indices acc ounts for 1/3 to 2/3 of the 'scenario' warming in January. In July, the lin ear regression model does not account for any warming at all, The warming w hich is not accounted for by the linear regression model may be caused by n on-linear processes (e.g. air-sea-ice interactions) or directly connected t o changes in the climate forcing.