Assessing The Accuracy Of Snow-Rain Phase Separation Models at Meteorological Stations of the Mountainous region of Zagros, Iran

In this study, the possibility of snow estimation using the snow-rain phase separation models that usually integrated into the snowmelt models or in the numerical weather prediction (NWP) models are evaluated in the Zagros region, Iran. For this purpose, daily precipitation (rain and snow) and temperature from 36 weather stations distributed over Zagros and having data records over the 1951-2015 period were used. The performances of the snow-rain phase separation models in accurately predicting the phase of precipitation at the studied stations were evaluated through constructing contingency tables between the predictions and observations and eventually evaluating the result using a set of statistical skill scores. A statistically significant relationship was found between the phase of precipitation predicted by the models and the snow occurrences observed at all selected stations. The USACE, Pipes and Quick, Hyperbolic tangent function and Kienzle were found as the best models while the Mccabe and Wolock model and the Brown model based on the maximum and minimum of temperature resulted in the weaker predictions. The snow-rain separation temperature threshold at the selected stations varies from -1.7 to 5 degree Celsius, nonetheless, it it is between 0 and 2 degree Celsius in more than 75% of stations. It was also found that the temperature range within which both snow and rain can occur simultaneously varies in all considered stations, but in most of the stations it is between 9 and 13 degrees Celsius.