An Innovative Algorithm for Unmanned Validation of Automatic Snow Depth Measurements

Ultrasonic snow sensors provide high temporal resolution snow depth and snowfall data used for hydrological applications and avalanche risk assessment. Nevertheless automatic snow depth measurements can be affected by errors and uncertainties due to the wind transport or interferences below the sensor. For this reason the data acquired by the automatic network of Arpa Piemonte the regional environmental agency in Piemonte, Italy have been usually validated by snow operators. This work presents an innovative algorithm developed by Arpa Piemonte in collaboration with the Earth Sciences Department (University of Torino) for the automatic validation of ultrasonic-based snow depth measurements. This automatic procedure is able to find suspect data i.e. isolated peaks, outliers and errors due to the sensor malfunctioning. Several tests on the air temperature have been implemented and coupled with a snow-melting model to verify the measurement reliability. This technique has been validated comparing the output of the algorithm to the historical snow depth series manually checked. In the winter season 2012 this procedure has become operational and it is currently applied to half-hourly data. The automatically validated data are further daily checked by snow operators who approve or reject the algorithm corrections. This new process of automatic validation has improved the data quality removing anomalous spurious snow depth values and reducing the subjectivity of the manual validation.