Evaluation of the stability indices for the thunderstorm forecasting in the region of Belgrade, Serbia

•We evaluate the ability of the stability indices to forecast the thunderstorms.•We set up the optimal thresholds for the various stability indices.•The best rank sum score had Lifted, K, Showalter, Boyden and Total totals index.•Lifted index had the highest skill for thunderstorm forecasting.•The Boyden index had the poorest skill scores.

The pre-convective atmosphere over Serbia during the ten-year period (2001–2010) was investigated using the radiosonde data from one meteorological station and the thunderstorm observations from thirteen SYNOP meteorological stations. In order to verify their ability to forecast a thunderstorm, several stability indices were examined. Rank sum scores (RSSs) were used to segregate indices and parameters which can differentiate between a thunderstorm and no-thunderstorm event. The following indices had the best RSS values: Lifted index (LI), K index (KI), Showalter index (SI), Boyden index (BI), Total totals (TT), dew-point temperature and mixing ratio. The threshold value test was used in order to determine the appropriate threshold values for these variables. The threshold with the best skill scores was chosen as the optimal. The thresholds were validated in two ways: through the control data set, and comparing the calculated indices thresholds with the values of indices for a randomly chosen day with an observed thunderstorm. The index with the highest skill for thunderstorm forecasting was LI, and then SI, KI and TT. The BI had the poorest skill scores.