Towards Probabilistic Methods for Forecasting Snow Load on Power Lines in Italy

Wet snowfall events often cause significant and damaging winter blackouts in Italy's power networks due to the formation of cylindrical snow sleeves on overhead conductors.  

In the last decade, observations at the WILD (Wet-snow Ice Laboratory Detection) monitoring station in the Italian western Alps allowed to thoroughly investigate wet-snow and to improve snow sleeve accretion models. These models were employed to develop a historical reconstruction of snow load over Italy based on MERIDA (MEteorological Reanalysis Italian DAtaset). Moreover, RSE developed WOLF (Wet-snow Overload aLert and Forecast), an operational forecast system specific for wet-snowfall events and snow sleeves formation on overhead lines. WOLF is based on the WRF model, which provides precipitation, wind and temperature fields, and on the Makkonen’s snow sleeve accretion model, which, depending on meteorological information, calculates the growth of predicted snow mass on reference conductors of the transmission and distribution power network in each domain cell.  

Past case studies showed that accuracy limitations were primarily due to the intrinsic uncertainty in modeled meteorological fields produced by a deterministic NWP forecast. Sensitivity tests with different model configurations and GCM drivers showed variable performances, without establishing an optimal configuration for the heterogenous set of snowfall events analyzed.  

In this work, we assess the benefits of a probabilistic multi-model approach, combining the resulting snow load predictions obtained from different model runs and exploring other post-processing methods suitable for wet-snow forecasting. We employed freely available output runs of NWP models from different providers such as the Italian Mistral open data hub (https://meteohub.mistralportal.it/app/datasets) and we compared results with observations of recent snowfall events over the western alpine area. 

Probabilistic forecasts for this specific application will be further investigated and refined over more case studies to reduce forecast uncertainty of the WOLF system during the next winter seasons.