Monitoring and Forecasting wet-snowfall and snow sleeves accretion on overhead power lines in the Italian western Alps

Heavy wet snowfall events are responsible for several and harmful winter blackouts on Italian power networks due to the formation of cylindrical snow sleeves on overhead line conductors and shield wires. Despite field observations and modelling studies in the last decades, accurately forecasting these events remains challenging because of the peculiar local meteorological conditions. Although the total number of snowfall events decreased over the alpine region during the last 20 years, last winter season (2023-2024) witnessed several significant wet snowfall events, characterized by mixed convective-advective precipitation, causing several power disruptions, and underscoring the persistent threat posed by this atmospheric phenomenon.

The current research activity carried out in RSE involves monitoring systems of sleeve formations at the Wet-snow Ice Laboratory Detection (WILD) station in the south-western Alps, a historical reconstruction of wet snow loads through the MEteorological Reanalysis Italian Dataset (MERIDA), and an operational forecast system Wet-snow Overload aLert and Forecasting (WOLF) for the identification of weather conditions favorable to the formation of snow sleeves on the overhead conductors.

WOLF consists of: a) a version of the Weather Research and Forecasting (WRF) model at 4 km spatial resolution properly configured to optimize the description of temperature and precipitation, variables primarily involved in the identification of wet snow and in the modelling of snow sleeves accretion b) an identification method for wet snow conditions, either considering a “thermal window” or the “snow ratio” of precipitation, and c) a sleeve accretion model (Makkonen model for wet-snow, ISO 12494:2017).

This study presents some exemplary case studies from the 2023-2024 winter season to highlight current challenges of wet snowfall prediction. High resolution simulations (1 km) were carried out for an accurate description of the mountainous area of south-western Alps where the WILD station is located. Different model drivers (IFS and GFS), model resolutions and wet-snow identification methods were evaluated against meteorological and snow mass measurements collected at the WILD station.

Preliminary results showed that the snow load forecast accuracy is primarily related to the choice of the model driver and, secondarily, to the wet snow identification method. The analyzed case studies are particularly challenging because of mixed convective-advective characteristic and because of complex orography. Additional case studies should be investigated to further assess the relative skill of meteorological drivers and of wet-snow identification methods for the snow load forecast in the WOLF system and in high resolution simulation focusing on the WILD station.