Automated Storm Identification, Tracking and Forecasting: a Radar-Based Method.

The problem at the focus of this research is the estimation of precipitation from storms (generally convective though other types are considered) either for the short -term prediction of severe weather events or for design and planning purposes. The method adopted aims to characterize storms by tracking them as detected by weather radar and determining their behavior in space and time. A 'storm' is defined as a contiguous region exceeding thresholds for radar reflectivity and size. Storms defined in this way are identified at discrete time intervals in the radar data. An optimization scheme is employed to match the storms at one time with those at the following time, with some geometric logic to deal with merging and splitting storms. The short-term forecast of both position and size is based on a weighted linear fit to the storm track history. The nature of the tracked storms was investigated over a 3 year period to produce a climatology of the study region. The accuracy of the short-term forecasts of storm position was determined over that same period and is shown to be comparable with or better than the accuracy of forecasts made by human forecasters in a similar experiment. The possibility of using instantaneous storm properties to improve the size forecast accuracy is investigated. A technique for short-term precipitation forecasting is developed and the results compared to the persistence forecast both for rainfall from convective storms and snowfall from winter storms. It is concluded that the automated technique shows skill both in precipitation and storm position forecasting, that this is an advance over previously available methods and that the method produces data suitable for the stochastic modelling of convective storm precipitation.