North Atlantic forcing of climate and its uncertainty from a multi‐model experiment

• Published in: Quarterly Journal of the Royal Meteorological Society
• Main Authors: Rodwell, M. J., Drévillon, M., Frankignoul, C., Hurrell, J. W., Pohlmann, H., Stendel, M., Sutton, R. T.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2004
• Subjects: North Atlantic; North Atlantic oscillation
• Online Access: http://dx.doi.org/10.1256/qj.03.207; https://api.wiley.com/onlinelibrary/tdm/v1/articles/10.1256%2Fqj.03.207; https://rmets.onlinelibrary.wiley.com/doi/pdf/10.1256/qj.03.207

Abstract To understand recent climate change in the North Atlantic region and to produce better climate forecasts with uncertainty estimates it is important to determine the atmospheric ‘response’ to Atlantic sea‐surface temperature (SST) forcing. There have been conflicting results regarding the strength, character and tropical‐versus‐extratropical origin of this response. For model‐based studies, this may indicate differing sensitivities to Atlantic SST, but the comparison is complicated by changes in experimental design. Here, a highly controlled experiment with five atmospheric models is undertaken. The influence of realistic (if reasonably strong) and optimally chosen North Atlantic (equator to 70 ° N) SST anomalies is isolated. Unexpected global agreement between the models is found (e.g. the North Atlantic Oscillation (NAO), Eurasian temperatures, rainfall over the Americas and Africa, and the Asian monsoon). The extratropical North Atlantic region response appears to be associated with remote Caribbean and tropical Atlantic SST anomalies, and with local forcing. Some features such as the European winter‐temperature response would be stronger than atmospheric ‘noise’ if the prescribed SST anomalies persisted for just two years. More generally, Atlantic air–sea interaction appears to be important for climate variability on the 30‐year timescale and, thus, to be important in the climate‐change context. The multi‐model mean response patterns are in reasonable agreement with observational estimates, although the model response magnitudes may be too weak. The similarity between their responses helps to reconcile models. Inter‐model differences do still exist and these are discussed and quantified. © Crown copyright, 2004.