Present‐day interannual variability of surface climate in CMIP3 models and its relation to future warming

Abstract Interannual variability (IAV) of 2m temperature ( T ), sea level pressure ( SLP ) and precipitation ( P ) in the CMIP3 20th century model simulations are compared with IAV in observational and reanalysis data sets using standard deviation based variability indices. Further, the relation between the representation of T IAV and the amplitude of future warming is investigated. In the Northern Hemisphere (NH) extratropics, T and SLP IAV are (in contrast to P ) in general well represented although a few models perform much worse than others. General problem regions are: (1) sea ice boundary regions, where well‐known biases in the mean states exist; and (2) the Pacific Ocean and Central Africa where SLP IAV is consistently underestimated. T and SLP IAV discrepancies are often found in similar regions and are large in well‐known bias problem regions in the tropics and subtropics and high mountain regions. ‘Bad’ IAV representation also occurs in regions with small biases. T IAV is in general better reproduced over land than over sea and in the extratropics than in the tropics. Among the ‘good’ IAV models there is no robust relation between the tropics (sea only) and the extratropics (land only). The relation between the model's ability to correctly represent T IAV and projected temperature changes is slightly negative (more warming for better IAV representation) but except for the NH summer season not significant when the worst models in terms of IAV representation are omitted. This suggests that aggregated over large regions (with exception of NH summer) no robust relations are found between the model's ability to correctly represent T IAV and the projected temperature change. Copyright © 2010 Royal Meteorological Society