The approach in these case studies has been to use ensemble forecast methodologies, depending upon the event being examined. For the smallest scale events, such as a watershed area, computing resources have permitted only a single forecast approach.
We also have examined the sensitivity of precipitation totals and patterns to initial conditions, using an adjoint approach, for a number of the events. The aims were to determine if any statements could be made about which variables are significant and also to see if it was possible to estimate forecast skill a priori.
In the presentation we will provide examples using the various approaches and summarise their usefulness as well as suggest future directions for QPF.