Downscaling local extreme temperature changes in south-eastern Australia from the CSIRO Mark2 GCM

作者：Schubert S^*

来源：International Journal of Climatology, 1998, 18(13): 1419-1438.

DOI：10.1002/(SICI)1097-0088(19981115)18:13<1419::AID-JOC314>3.3.CO;2-Q

• regional climate change
• statistical downscaling
• daily temperature extremes
• synoptic scale atmospheric circulation
• GCM validation
• south-eastern Australia

Climate impact studies crucially rely on climate change information at high spatial and temporal resolutions. Since the most developed tools for estimating future climate change-the general circulation models (GCMs)-still operate on rather coarse spatial scales, their output has to be downscaled in order to provide the needed high resolution input for climate impact models.
In this study, a perfect prognosis approach is employed to downscale daily local temperature extremes at several stations in south-eastern Australia from synoptic scale atmospheric circulation fields. The statistical model combines principal component analysis and linear multiple regression and is suitable to explain a considerable part of both short and long frequency variations of local temperature extremes. Using simulated and observed daily data the regional to local performance of the Mark2 GCM, developed by the Commonwealth Scientific and Industrial Research Organisation (CSIRO), was validated over the Australian region. While the regional temperature extremes simulated under present-day climate conditions are in general agreement with the observed climate, there are highly significant differences on the local scale. The observed daily synoptic scale atmospheric circulation, however, is well reproduced by the GCM. This supports the idea of using these reliably simulated climatic parameters to estimate the changes in local temperature extremes under altered global climate conditions.
The downscaling model was applied to synoptic scale atmospheric circulation fields generated by the CSIRO Mark2 GCM under 1 x CO2 and 2 x CO2 conditions. Compared to the extreme temperature changes simulated by the GCM directly, the downscaled variations are much weaker. Several sources of uncertainty might be causing these differences. Firstly, the statistical model is stationary. Therefore, it is not capable of including changes in the relationships between circulation and local temperature which are likely to occur under 2 x CO2 climate changes. It also only captures circulation-related temperature variations. An analysis of the circulation changes under 2 x CO2, however, reveals almost no significant differences to the present day conditions. That leads to the conclusion, that the quite dramatic changes of local temperature extremes under doubled CO2 are not driven by circulation changes and might be: almost solely caused by changed radiative properties of the atmosphere.

• 出版日期1998-11-15