Nowcasting the precipitation phase combining weather radar data, surface observations, and NWP model forecasts

Abstract

Heavy snowfall events can cause substantial transport disruption and exert a negative socioeconomic impact, particularly in low-altitude and midlatitude regions, where it seldom snows. Such problems may be exacerbated if there are rapid transitions between different precipitation phases within the same event. Previous studies have addressed this issue using precipitation-phase nowcasting techniques, often focusing on critical infrastructures such as airports. Very short-range forecasts are usually based on trends of observations and numerical weather prediction models. Nowcasting schemes considering the precipitation phase generally merge extrapolated surface observations, modelled vertical temperature profiles, and extrapolated weather radar precipitation fields. In this study, a precipitation-phase nowcasting scheme was developed and evaluated, initially using eight different algorithms to classify precipitation into rain, sleet or snow, together with a probabilistic weather radar data extrapolation technique. In addition, three combinations of the previous algorithms were also evaluated. The nowcasting scheme was applied to a midlatitude region in the Northwestern Mediterranean to assess its performance during eight snowfall events. Single and combined algorithms were compared to determine their suitability in conditions close to freezing point, when there is increased uncertainty about the precipitation phase. The results indicate that, although single and combined algorithms perform similarly, the latter can provide valuable information during event monitoring. Precipitation phase transitions were also analysed, finding that on average they can be forecast correctly with a lead time of 120 min. The proposed methodology can be readily applied to other regions where ground-based observations, weather radar data, and model forecasts are available