In most meteorological or hydrological models, the distinction between snow and rain is based only on a given air temperature. However, other factors such as air moisture can be used to better distinguish between the two phases. In this study, a number of models using different combinations of meteorological variables are tested to determine their pertinence for the discrimination of precipitation phases. Spatial robustness is also evaluated. Thirty years (1981-2010) of Swiss meteorological data are used, consisting of radio soundings from Payerne as well as present weather observations and surface measurements (mean hourly surface air temperature, mean hourly relative humidity, and hourly precipitation) from 14 stations, including Payerne. It appeared that, unlike surface variables, variables derived from the atmospheric profiles (e.g., the vertical temperature gradient) hardly improve the discrimination of precipitation phase at ground level. Among all tested variables, surface air temperature and relative humidity show the greatest explanatory power. The statistical model using these two variables and calibrated for the case study region provides good spatial robustness over the region. Its parameters appear to confirm those defined in the model presented by Koistinen and Saltikoff.
