Predicting new snow density in the Italian Alps: A variability analysis based on 10 years of measurements

Despite its strong impact on the time evolution of the snowpack, current estimation of new snow density (ρ_hn) is usually accomplished either by using local empirical techniques or by assuming a constant snow density. Faced with the lack of an estimation model of ρ_hn valid for a wide spatial scale and supported by a suitable number of observations, this study aims to develop simple monthly linear regression models at scale of the entire Italian Alpine chain based on 12,112 snowfall observations at 122 stations, using only air temperature as predictor. Moreover, the remaining variance is investigated in both time and space, also considering some qualitative features of the snowfall events. The daily ρ_hn measurements present a mean value of 115 kg m^?3 (105 and 159 kg m^?3 for dry and wet conditions, respectively). The mean air temperature of the 24 hr preceding the snowfall event has been found to be the best predictor of the ρ_hn, within 31% of uncertainty. The analysis of associated residues allows supporting the idea that the adoption of a more local approach than the one analysed here is not able to substantially increase the predictive capabilities of the model. In fact, the main factor explaining the remaining variance over the air temperature is the wind, but in a complex orography, as mountain regions are, supplying realistic local wind fields is particularly challenging. Therefore, we conclude that using only the daily mean temperature as predictor is a good choice for estimating daily new snow density at scale of Italian Alpine chain, as well as at more regional scale.