The product consists of identified convective cells and their tracks from previous positions as well as nowcasts of storm strike probabilities. In addition, each identified cell is assigned a severity class.
Convective cells are identified from radar mosaics and classified into one of several danger categories using a climatology of storm events and also fault messages. Power cut incidents, which are caused by severe weather, are used to train classification software to detect thunderstorms, which potentially cause different categories of power failures. At European scale, the OPERA domain was divided into six subdomains, and severity class boundaries were determined separately for each region using CDFs of 500000 individual storms.
There are large uncertainties in forecasts of convective cells because their size and shape changes in time. Therefore, identified convective cells are extrapolated to give a probabilistic forecast that also captures the uncertainty in the future location and severity class. The extrapolation is based on velocities that are estimated from previously observed cells.