Original paper

Statistically downscaled projections of local scale temperature in the topographically complex terrain of Austria up to the end of the 21st century

Landgraf, Martin; Matulla, Christoph; Haimberger, Leopold

Meteorologische Zeitschrift Vol. 24 No. 4 (2015), p. 425 - 440

59 references

published: Jul 21, 2015
published online: Apr 13, 2015
manuscript accepted: Feb 11, 2015
manuscript revision received: Jan 29, 2015
manuscript revision requested: Jul 30, 2014
manuscript received: May 30, 2014

DOI: 10.1127/metz/2015/0620

BibTeX file

O

Open Access (paper can be downloaded for free)

Download paper for free

Abstract

This paper provides local scale temperature scenarios for Austria from the middle up to the end of the 21st century. Climate simuations based on the IPCC emission pathways A2 and A1B have been carried out with the global climate models ECHAM5 (three A1B and two A2 realizations) and HadGEM2 (three A1B realizations). The corresponding large scale projections of sea level pressure and 850 hPa temperature fields are statistically downscaled to stations spread across Austria using a perfect prognosis (PP) approach and Multiple Regression Models. The downscaling performance is assessed by a split sample test. Simulated time series are compared to actual measurements by means of the simulated variance, the root mean square and the mean error. Performances are highest during the cold season and sites located in valleys exhibit somewhat lower values. In summer performances show about a 10 percent lower skill than in winter. Downscaled local scale scenarios differ between seasons, scenarios, GCMs and regions in Austria. A1B estimates derived from ECHAM5 indicate a winter-temperature increase of approximately 3 °C at the end of the 21st century compared to present conditions, which is about one degree above the HadGEM2 based projections. This situation is reversed in summer: the HadGEM2 based projections show a warming of about 4 °C while those derived from ECHAM5 indicate a 0.5 °C lower warming. Statistically downscaled winter warming rates at stations < 1400m$<1400\,\text{m}$ NN can be roughly split into three regions. The stations exhibiting the largest warming rates are located in the north-eastern parts of Austria, whilst stations showing the lowest increases are located south of the Alpine ridge. The rest of the stations are found north of the Alpine crest in the north-western parts of Austria. In summer spatially rather uniform temperature increases are detected. Mountain sites above 1400 m NN exhibit an average warming of around 0.5 °C higher than the remaining stations.

Keywords

Climate changeStatistical DownscalingHISTALP databaseLocal scale seasonal temperature projections