Estimation of winter precipitation in a high-altitude catchment of the Eastern Italian Alps: validation by means of glacier mass balance observations

Abstract

This work analyses the estimation of winter season (October 1 to May 31) precipitation over a high altitude catchment (Val di Peio) of the Eastern Italian Alps. The extrapolation of precipitation over ungauged areas is problematic in this basin due to the combined effect of measurement errors and orographic effects. The study is based on the availability of long term series of weather data, snow observations and glacier mass balance measurements. The error in precipitation measurement at the uppermost weather station was assessed by comparison with snow water equivalent data. The error estimates were consistent with the outcomes of a precipitation correction model, with the aerodynamic correction as its main component. This correction procedure was used to compute correction factors for the entire precipitation gauge network. Both corrected and uncorrected precipitation data were used to estimate the spatial distribution of precipitation over the study area by means of a technique that accounts for the precipitation-altitude relationship. Winter balance observations over Careser glacier (the longest mass balance series in Italy), which is within the study area, were used to assess the improvement given by the precipitation correction procedure, showing a reduction of bias from -38% to –2.5% and a reduction of the RMSE from 410 to 171 mm water equivalent. The use of corrected precipitation data led to a 31.7% (182 mm) increase of the winter season basin-averaged precipitation, whereas the winter season average vertical gradient increased from 4.0% km^–1 to 21.9% km^–1. Overall, these results highlight the need for precipitation correction in precipitation analyses over snow-dominated mountain areas. This paper also provides evidence of a considerable interannual variability in the correction factors for snow at the uppermost weather station. Further improvements in precipitation estimations would require an analysis of the dominant processes controlling this variability.