Keynote Talk: Progress, Problems, and Prospects for Radar Data Assimilation in Stormscale Models

It is fifteen years since the first attempt at radar data assimilation and numerical prediction of an observed thunderstorm. Similar to the initial attempt at synoptic-scale prediction by L.F. Richardson, the first storm prediction quickly went awry. However, in the last fifteen years significant progress has been made in both radar-data assimilation and short-term prediction of moist convection. In this talk I will describe some of that progress, as well as some of the challenges ahead.

The field of numerical storm prediction can largely be decomposed into two main problems; convective-scale initialization and the effect of the large-scale environment on the convective scale. For the first problem, I will describe some of the progress that has been made in radar data assimilation with a number of different methods such as 3D and 4D variational data assimilation and the Ensemble Kalman Filter. It will also be shown that storm-scale NWP has the ability to improve over simpler techniques based on spatial extrapolation of existing radar echoes. However, these short term numerical predictions can be extremely sensitive to the specification of the large-scale environment.

The second problem, that of the interaction of large-scale and convective-scale motions, remains a significant challenge. I will conclude the talk by examining the possibility of using convective-scale data to project back on to the larger scale.