Sauter, T; Weitzenkamp, B; Schneider, C (2010). Spatio-temporal prediction of snow cover in the Black Forest mountain range using remote sensing and a recurrent neural network. INTERNATIONAL JOURNAL OF CLIMATOLOGY, 30(15), 2330-2341.
Winter tourism is the main economic factor for many different regions in the German Mountain Range. Owing to warming trends experienced in the past and predicted for the future, precise knowledge about the development of snow cover and snow duration in the future is becoming more and more important. On the basis of the International Panel on Climate Change (IPCC) A1B scenario, this paper investigates the possible regional development of snow cover and snow duration in the Black Forest in southwest Germany until 2050. For this purpose, we developed a new method that combines Non-linear AutoRegressive networks with eXogenous inputs (NARX) with Remote Sensing and Geographic Information System (GIS). With this non-parametric approach, we try to define with preferably high accuracy a simple transferable model. Besides the general problem of developing a robust statistical model, our main focus is on the enhancement of the spatial resolution of snow patterns by incorporating complex structures of the underlying terrain using Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data. The results suggest a possible decrease in the number of snow days (snow cover >= 10 cm) in the decade 2041-2050 by 10 to 44% at altitudes higher than 1200 m, by 17 to 57% at 1000-1200 m and 25 to 66% at 500-1000 m. This results in a dramatic shortening of the snow season mainly caused by earlier snow melt initiation rather than by later first snow precipitation in autumn. These considerable changes in the snow season would cause enormous losses in the skiing and tourism industries. In this context, the obtained high-resolution snow data and maps can provide a useful tool and decision-making aid for the economy and politics. Copyright (C) 2009 Royal Meteorological Society