D. Heimann, Institut für Physik der Atmosphäre, DLR Oberpfaffenhofen, 82230 Wessling, Germany
One day before the MAP-meeting at Bad Tölz, a workshop on regional climate and climate regionalization in the Alpine region was held at the same place. About 25 scientists from Switzerland and Bavaria gathered and twelve talks were presented.
The workshop aimed at
Several research projects are currently undertaken to investigate potential climate changes due to increased greenhouse gas concentrations and their impact in the Al pine region. In Switzerland climate-related projects are funded by the national research programme NFP31 and the Swiss Priority Programme Environment (SPPU). Bavarian projects are partly coordinated by the Bavarian Climate Research Programme (Bay FORKLIM) and are funded by Bavarian state ministries and other sources (European Commission, German Federal Ministry for Education, Science, Research, and Technology etc.).
The presented projects divide into two general types: Some projects establish high-resolution climatologies or evaluate long-term trends from regional observations of the last years, decades, or even centuries. Other projects take advantage from large-scale numerical analyses or GCM climate simulations and employ different downscaling methods to regionalize these informations to the Alpine topography.
Wanner 1 reported on synoptic downscaling. His project searches for statistical correlations between the northern Atlantic oscillation and regional climate patterns in the Alps. A 200-year time sequence of European "Grosswetterlagen" was derived from pressure observations since 1780. F. Neidhöfer 1 demonstrated a modified ordinary kriging method which was used to generate twodimensional fields of precipitation sums ac cording to different "Grosswetterlagen" on the base of a 20-year set of daily precipitation reports of 149 stations in Switzerland. Major uncertainties occured near steep mountains because the method does not consider dynamic interactions of the large scale flow with mesoscale topographical features.
Climate-change scenarios and impact studies for Alpine forests and grasslands were presented by D. Gyalistras 4 . High-resolution fields of observed monthly temperature means and precipitation sums in Switzerland (50 years of measurements at 40 sta tions) were used to downscale ECHAM1 climate change scenarios ( 2 x CO 2 , IPCC-A) by means of canonical correlation analysis to the regional scale. Uncertainties of the downscaling procedure were particularly large for summer precipitation. The downs caled scenarios were combined with a spatial interpolation procedure and a stochastic weather generator to derive scenarios at a very high spatial (1~km) and temporal (1~h) resolution. Additional climate simulations with different GCMs are at present evaluated to provide a measure of uncertainty. Changes in Swiss annual mean temperature and precipitation downscaled from four GCM-experiments showed strong north-south con trasts and ranged from +1 to +3 K and -15 to +35 %, respectively.
Three talks dealt with thunderstorms in the Alpine area. H. Huntrieser 3 introduced new stability indices that were found to improve thunderstorm forecasts in Switzerland on the base of the Payerne radiosounding. The indices distinguish whether or not thunderstorms are likely to appear and whether they will be isolated or wide-spread. S. Willemse 3 evaluated damage claims provided by the Swiss hail insurance company and worked out a 40-year climatology of hail events in Switzerland. She found a high year-to-year variability but no significant trend in hail occurrence. H.Höller 7 derived statistics of the southern German thunderstorm activity from six years of radar observations at Oberpfaffenhofen (near München). He demonstrated how polarimetric radar informations allow to discriminate different types of thunderstorm cells and types of precipitation. Frequent hail events were observed particularly in some parts of the northern Alpine foothills. The statistics of radar observations were confirmed by a two year set of lightning location data as were shown by U. Finke 7 . He found bimodal tem poral distributions of the lightning activity in the lapse of day and year as well. Daily maxima appear between 15 and 17 and around 20 UTC. Yearly maxima were observed in May and July.
Regional climate modeling was central to the following talks. Most groups use the dynamical downscaling method, i.e. a mesoscale models is nested into a large-scale grid on which either numerically analyzed or prognostically simulated meteorological fields provide appropriate boundary values.
Keuler 5 spoke about general problems in regional climate modeling. He used a nested regional model (McGaP, 20 km resolution) that was permanently forced at the lateral and upper boundaries by large-scale coarse-grid analyses. He showed how the resulting monthly means depend on the spatial resolution of the regional model and what problems arise when results are compared with observations (e.g. correction of elevation, spatial representativeness).
M. Beniston and M. Rotach 2 applied the dynamical downscaling method to the output of large- scale climate simulations (ECHAM3, T106-resolution) representing the present-day climate and a doubled CO 2 scenario, respectively. Five consecutive months (January and July, respectively) were simulated using the RegCM-2/MM4 regional model. The strong warming and decrease in precipitation that was simulated in a doubled CO 2 July climate by the ECHAM3 model appeared weaker in the regional model runs. More reliable results are expected from future ECHAM4 simulations.
D. Lüthi 3 also presented results of dynamical downscaling. He applied the limited area model of the Swiss Meteorological Institute (at 54 and 14 km horizontal resolution) for winter and summer months. At the lateral boundaries the model was driven by EC MWF analysis data. Comparisons with observations and different test runs revealed a remarkable sensitivity of the simulated amount of precipitation to the prescribed soil water content specifically in summer. The temperature (daily amplitude) was strongly affected by this parameter, too.
Some of the abilities of the Penn State/NCAR non-hydrostatic mesoscale model (MM5) were demonstrated by G. Grell 6 . He introduced a hierarchy of nested grids from 45 km down to 1.7 km horizontal resolution. Initialized by radiosounding data the develop ment of deep convection over southern Germany was simulated at different resolu tions. Convective processes are parameterized except for the 1.7 km grid where the non-hydrostatic model version explicitely resolves convective circulations. The model is currently used to do climate simulations on local non-hydrostatic scales over 1-3 month periods and to produce an assimilated (Four Dimensional Data Assimilation) consistent meteorological data set over a one month summer period (down to 1.7 km resolution).
An alternative regionalization method was introduced by U. Fuentes 7 . The so-called statistical-dynamical downscaling combines statistics of large-scale weather types and regional model simulations. Based on a 10-year sequence of daily ECMWF-analyses climatological fields were derived for the entire Alpine area at 20 km resolution separately for the winter and summer half year. At present, the main emphasis is directed to a verification at 30 SYNOP- stations in and around the Alps. V. Sept 7 applied the statistical-dynamical downscaling procedure to assess the regional frequency of convective storms in southern Bavaria. He showed that stability indices which were derived from the regional model output are a useful indicator for deep convection.
The workshop revealed a broad variety of research approaches towards a better under standing of the Alpine climate and the implementation of tools and methods to region alize global climate change assessments. Conventional surface measurements, radio soundings, damage claims, and remote-sensing techniques (radar, lightning detection) were utilized in order to establish highly resolved climatological fields at least in certain subareas of the Alps. Three different downscaling procedures are currently used to transmit large-scale low-resolution information ("Grosswetterlagen", numerical analyses, GCM simulations) to mesoscale high-resolution fields of climatologically relevant parameters or their long-term changes:
Nevertheless, many problems were encountered by the reseach groups, some of which are not yet solved. The following open questions were identified to be common to most, if not all participants:
The participants agreed with a following workshop which is provisionally
scheduled in February 1996. This workshop should provide a forum
of intense discussions about se lected problems.
MAP Data Centre - April '05 - MAP WebMaster