Convective-scale Data Assimilation and Numerical Weather Prediction: Past, Present and Future

Explicit prediction of convective storms dates back to the vision of Douglas Lilly and his colleagues, as put forth in his easy "Numerical prediction of thunderstorms - Has its time come?", published almost twenty years ago (Lilly 1990, QJRMS). Advanced numerical techniques and physics development enabling accurate non-hydrostatic prediction models, promising variational data assimilation techniques for radar data assimilation, the deployment of national Doppler weather radar network and the advent of massively parallel supercomputers, coupled with better understanding on thunderstorm dynamics and predictability formed the foundation for convective-scale numerical weather prediction (NWP). In fact, the conditions were considered ripe that the National Science Foundation funded the Center for Analysis and Prediction of Storms as one of its first Science and Technology Centers in 1989 to address exactly this problem.

Significant progresses have occurred in the past twenty years on all fronts of convective-scale NWP. Advanced community nonhydrostatic models are now readily available while their data assimilation systems are also maturing. Realtime convection-resolving predictions are routinely made and even ensemble forecasting at convective-scale has become a reality. Significant progresses have also been in convective-scale data assimilation, in particular the assimilation of Doppler radar data.

In this talk, the short history of storm-scale data assimilation and NWP will be briefly reviewed first. Recent progresses in the areas of data assimilation, simulation and realtime forecast experiments will be highlighted. State of the science research that promises future improvements to storm-scale NWP as well as remaining challenges will be discussed. A personal vision for the future will be offered. The talk will try to draw upon materials from the research in the speaker's own research group as well as from the community at large.