Sensitivity of Alpine snow cover to European temperature

The number of days with snow cover at 268 Alpine climate stations in the winters of 1961-2000 has been investigated with respect to the mean winter temperature over Europe. The corresponding description, originally developed for Austria and recently applied to Switzerland, consists in fitting a logistic curve to the observed data. The slope of this curve, originally the hyperbolic tangent function, is interpreted as the sensitivity of the snow duration-temperature relationship. Here we first demonstrate with a physical-statistical model that the proper logistic curve is not the hyperbolic tangent, but the error function, generated through the pdf of the fluctuating temperature; the slope of this curve is inversely proportional to the standard deviation of temperature. Since the station temperature used for this local model is on a scale much too small for global climate models, we simulate, secondly, the temperature with the concept of the Alpine temperature: It is the spatial Taylor expansion of the seasonal European temperature in vertical and horizontal directions. This improved model yields, for the same Austrian and Swiss data, both a better fit and a slightly smaller sensitivity of the snow-temperature curve than the original hyperbolic model. Thirdly we apply our improved model to a considerably larger Alpine data set comprising also data from France, Germany, Italy and Slovenia and find a sensitivity of about -0.33 (+/- 0.03) per degree warming. It is representative for the entire Alpine region and corresponds to a maximum reduction of the snow cover of 30 days in winter at a height of 700 in for 1 degrees European warming. The implication is that the relation between the natural fluctuations of winter snow duration and European temperature may be an estimate for a trend of snow duration in case of a future European temperature trend. Copyright (C) 2007 Royal Meteorological Society.