Very short range forecasting of precipitation: Comparing NWP and extrapolation techniques

Presently, the state-of-the-science technique for very short range (0-3 hour) forecasts of convection involve spatial extrapolation of existing radar echoes (sometimes called Nowcasting). These techniques are computationally inexpensive, however, their forecast skill rapidly decrease due to the fact that system evolution is typically not accounted for. An alternative technique is to assimilate stormscale data into an Numerical Weather Prediction model. In contrast to Nowcasting techniques, stormscale assimilation is computationally expensive and often difficult to code. However, since NWP models have the ability to forecast system evolution, the technique has the potential to improve over Nowcasting methods.

At the National Center for Atmospheric Research we have been examining, over the last decade, the ability of NWP models to explicitly forecast convective storms in the 0-3 hour timescale. The technique we have been using is four-dimensional variational assimilation of radar data into a cloudscale model. We are performing a systematic comparison of stormscale NWP and Nowcasting methods for a number of convective cases from the midwest of the USA as part of the RCWF program supported by the FAA. At the Conference, the case studies will be briefly described verification results presented and then the pros and cons of very short range NWP discussed.