Introduction
Local and regional climate changes have been observed in the last century (e.g., retreat of Alp glaciers) and are the aim of recent research. The causes of climate changes are difficult to detect, because the climate system is extremely complex due to the nonlinear interactions of its components.
Helpful tools in climate research are global climate models, which allow the implementation of experiments to explore the nature of the climate system. They are an important resource in understanding the issues involved in detecting the causes of climate changes, but, because of their coarse resolution in space and time, they have difficulties in an adequate precise reproduction of local and regional climate conditions.
In order to gain information about the regional climate, a variety of downscaling methods can be used. Dynamic downscaling is one of them and imbeds a regional model with a higher resolution within a global model, whereby the results of the global model are only used to set up the initial and boundary conditions of the regional model. The horizontal grid space in regional models usually varies from 15 to 50 km, whereas the grid space of global models is in the range of 100 to 400 km (see Figure 1).
Usual grid spaces of a general circulation model (~ 3.75° left) and a regional climate model for the Arctic (~ 0.5° right). At each case only every second grid line is shown.
The climatology of a regional model is determined by the succession of weather events simulated from the contribution of lateral boundary conditions and internal model physics. In a regional model major ridges, plains, coastal and other finer-scale features are resolved. Regional climate model studies have shown that regional models consistently improve the simulation of regional spatial patterns of precipitation, temperature or cloud formation compared to the driving global model, essentially as a result of better representation of topography, nonlinear energy transfer, hydrodynamic instability processes and other high-resolution forcings.
HIRHAM flow-chart
Failure analyses of global climate models have also revealed problems in a realistic simulation of typical Arctic phenomena, such as Arctic haze, sea ice or stratiform clouds. Considering the importance of this region with regard to the radiation balance and the energy budget of the earth, it seems to be desirable to apply a regional climate model to the Arctic. Particularly since greenhouse gas scenarios of global climate models as well as recent observations have revealed larger climate changes in polar regions than in other regions of the earth.
Following that, our research group uses the regional atmospheric climate model HIRHAM for the whole Arctic region. At the lateral and lower boundaries HIRHAM is one-way driven by different global data sets: observational data analyses or global climate model output (see Figure 2). Because of the continuous boundary forcing in the regional model, the large-scale circulation of the global model is preserved, if the downscaled region has an optimized size.
However, the predictability of the climate with a regional climate model still depends on the choosen resolution of the model. In case a model is imbedded in a global model, the error propagation is partly limited by the boundary conditions. Finally, the accuracy of a regional model depends on the physical parameterizations and boundary conditions used.