Abstract
For many ecological processes, near-surface temperature is one of the main controlling parameters. Since meteorological observations in high mountains are sparse, climate model data are frequently used. State of the art climate reanalysis products or general circulation models have a horizontal spatial resolution on the order of 50 � 50 km, i.e. far too coarse to represent the spatial variability of near-surface temperatures in mountain regions. Based on the assumption that elevation is the main factor determining the distribution of temperatures, we present a SAGA-GIS based approach for elevation and bias correction of climate model output data. Modeled temperature and geopotential height at different pressure levels were used to derive local temperature profiles by means of a polynomial regression approach. Atmospheric temperatures at surface level can then be derived from the regression equation. Compared with a simple elevation adjustment of ERA-Interim near-surface temperatures using an invariant lapse rate of 0.65�C per 100 m, the method showed good results, particularly in complex terrain. The method was used to generate high-resolution daily temperature fields for a target area in Central Asia for the period from 1989 onwards. Gridded trends of different temperature indices were calculated. For winter and spring, the highest temperature trends were detected in the high mountain regions. In summer, the calculated trend magnitudes were generally smaller. During the post-monsoon season, trends were more pronounced at low elevations. For the southern slopes of the Himalayan Arc and the valleys of the Tibetan Plateau, a significant decrease in frost days and an increase in growing degree days were detected.
