Historical evidence of snowfall variability and trends in the Central and Southern Apennine Mountains

In recent years, a great deal of attention has been devoted to the study of past snowfall variability worldwide, mainly in mountain regions.The snow, in fact, is a pivotal component of the hydrological cycle and has a relevant impact on the energy budget, controlling the land surface albedo. In addition, the snow strongly affects the complex ecosystems of mountain areas, as well as the biogeochemical cycles. Therefore, considering also the recent climate changes, that are posing serious threats on the cryosphere and mountain regions, it is crucial to recover and analyse historical long-term time series of snowfall to assess its variability and tendencies.For several reasons, many mountain areas remain under-researched. In the Mediterranean, an example in this sense is represented by the Apennine region (Italy). A considerable lack, in fact, exists in the knowledge of the past snowfall variability for this area, although it has a good heritage of past in situ observations.This work presents an analysis of historical snow precipitation data collected in the period 1951-2001 in Central and Southern Apennines. To pursue this aim, we used the monthly observations of the snow cover duration, number of days with snow and total height of new snow collected at 129 stations located between 288 and 1750 m a.s.l.. Such data have been manually digitized from the Hydrological Yearbooks of the Italian National Hydrological and Mareographic Service. The available dataset has been primarily analyzed to build a reference climatology (related to 1971-2000 period) for the considered Apennine region. More specifically, using a methodology based on Principal Component Analysis and k-means clustering, we have identified different modes of spatial variability, mainly depending on the elevation, which reflect different climatic zones. Subsequently, focusing on the number of days with snow and snow cover duration on the ground, we have carried out a linear trend analysis. An overall negative tendency has been found for both variables. For clusters including only stations above 1000 m a.s.l., a significant (at 95% confidence level) decreasing trend has been found in the winter season (i.e. from December to February): 3.2 6.0 to 0.0 days/10 years for snow cover duration and 1.6 2.5 to 0.6 days/10 years for number of days with snow. Moreover, in all considered seasons, a clear direct relationship between trend magnitude and elevation has emerged. In addition, using a cross wavelet analysis, we found a close in-phase linkage on decadal time scale between the investigated snow indicators and the Eastern Mediterranean Pattern. For both snow cover duration and number of days with snow, such connection appears to be more relevant in full (i.e. from November to April) and in late (i.e. from February to April) seasons.