The Analysis of the Thunderstorm Forecasts on the Friuli-Venezia Giulia Plain

Ristret

This work looks at four years of objective and subjective thunderstorm forecasts on the Friuli Venezia Giulia plain. This analysis is based on the usual attributes derived by the contingency table, that is the Probability Of Detection (POD), the False Alarm Rate (FAR) and the Hit Rate (HR), as well as more sophisticated skill indexes such as the Heidke skill Score (HSS). The categorical approach with the Brier Skill Score (BS) is adopted only for subjective forecasts. Results show that objective thunderstorm forecasts have an higher POD than subjective forecasts, but unfortunately even an higher FAR. This produces comparable HRs for both forecasts, although subjective forecasts are globally better than objective ones. Based on their behavior, to obtain a better product it is worth using a combination of both of them in the operative weather forecasts routinely issued by the Regional Meteorological Observatory of ARPA-FVG. The BS computed for the subjective forecasts highlights the positive feedback obtained from this analysis, carried out yearly by the authors. In particular, the subjective forecasts over the last year are much more calibrated.