Predicting patterns of near-surface air temperature using empirical data

The signal of recent global warming has been detected in meteorological rec ords, borehole temperatures and by several indirect climate indicators. Ant hropogenic warming continues to evolve, and various methods are used to stu dy and predict the changes of the global and regional climate. Results derived from GCMs, palaeoclimate reconstructions, and regional clim ate models differ in detail. An empirical model could be used to predict th e spatial pattern of the near-surface air temperature and to narrow the ran ge of regional uncertainties. The idea behind this approach is to study the correlations between regional and global temperature using century-scale m eteorological records, and to evaluate the regional pattern of the future c limate using regression analysis and the global-mean air temperature as a p redictor. This empirical model, however, is only applicable to those parts of the world where regional near-surface air temperature reacts linearly to changes of the global thermal regime. This method and data from a set of approximately 2000 weather stations with continuous century-scale records of the monthly air temperature was applie d to develop the empirical map of the regional climate sensitivity. Data an alysis indicated that an empirical model could be applied to several large regions of the World, where correlations between local and global air tempe rature are statistically significant. These regions are the western United States, southern Canada, Alaska, Siberia, south-eastern Asia, southern Afri ca and Australia, where the correlation coefficient is typically above 0.9. The map of regional climate sensitivity has been constructed using calcula ted coefficients of linear regression between the global-mean and regional annual air temperature. As long as the correlations between the local and g lobal air temperature are close to those in the last several decades, this map provides an effective tool to scale down the projection of the global a ir temperature to regional level. According to the results of this study, m aximum warming at the beginning of the 21st century will take place in the continental parts of North America and Eurasia. The empirical regional clim ate sensitivity defined here as the response of the mean-annual regional te mperature to 1 degreesC global warming was found to be 5-6 degreesC in sout hern Alaska, central Canada, and over the continental Siberia, 3-4 degreesC on the North Slope of Alaska and western coast of the U.S.A., and 1-2 degr eesC in most of the central and eastern U.S.A. and eastern Canada. Regions with negative sensitivity are located in the southeastern U.S.A., north-wes tern Europe and Scandinavia. The local tendency towards cooling, although s tatistically confirmed by modern data, could, however, change in the near f uture.