A Decision Support System for Regional Avalanche Forecasting

Abstract

Besides the text-based and Swisswide national avalanche forecasts regional avalanche forecasts have been introduced in Switzerland in 1996. Each of the regional forecasts covers an area of about 3.000 km2 and shows the avalanche hazard in graphical form. The regional bulletins are published at 8 a.m. This paper shows a computer based system that contains • a statistical model (NXD-REG) that forecasts the avalanche hazard for Switzerland. The model is based on the nearest-neighbor approach which is applied for 60 manual snow stations. The results of each station are then compiled into a map of the current and future avalanche hazard. The model has been validated on a historic database of 10 years by cross-validation. In 47% of the situations the nearest-neighbor model was in agreement with the conventional hazard estimations of the SLF avalanche warning service. The model's results are an additional information for constructing avalanche hazard estimations. • a GIS-based editor (BULLED) for constructing maps of the past or current avalanche hazard for Switzerland. The maps can be edited, copied, printed, exported in all important graphics formats and saved in a relational database. The editor also manages the division of Switzerland into more than 100 sub-areas. These sub-areas are needed for defining all geographical terms used in the text-based national bulletin and for quick map construction. The results of NXD-REG are fully integrated with BULLED. • a text-based editor (REGBUL) for writing the forecasts for six different regions . The data of automatic weather stations and the official weather forecast of Switzerland are included automatically. Furthermore a set of structured sentences allows the avalanche forecaster to quickly compile a few sentences about the current situation. The structure of the sentences allows us to translate fully automatically into French. The hazard estimation created with BULLED is used for the region edited. These three complementary approaches are the basis of Computer Aided Regional Avalanche Forecasts. The data of automatic stations and model calculations of SNOWPACK are further important prerequisites but are not presented here. The system has been fully implemented and is in operational use at SLF since February 2000.