Three-dimensional Analyses of Several Thunderstorm Cases Observed using Hunan's Regional Radar Network

The assimilation of radar data into storm-scale weather prediction models represents a significant scientific and technological challenge. Several data assimilation methods are candidates for use, including ensemble Kalman, filter methods and variational (3DVAR, or 4DVAR) assimilation methods. Much of the radar data assimilation research using the ensemble Kalman filter and 4DVAR is found to be computationally very expensive and it will be challenging to implement these two approaches in the next several years. In contrast, the 3DVAR approach is computationally very efficient and also produces reasonable results. In this study, we will present results from assimilating radar observations of reflectivity and radial velocity for several thunderstorms observed by Human regional Doppler radar network using the ARPS 3DVAR system formulated in an incremental form. For every selected case, the 3DVAR analyses will begin with the first 30 dBZ radar echo associated with the eventual thunderstorm of interest and continue until the end of the slected thunderstorm. Three-dimensional storm-scale analyses will be produced every 5-10 minutes with high spatial resolution using a cycling approach where the Advanced Regional Prediction System (ARPS) model is used to advance the 3DVAR analysis forward in time between data insertions. Preliminary tests indicate that these Chinese storms usually produce heavy rains, similar to Mesoscale Convection System (MCS) frequently observed within US, but not well-organized as US MCSs. The circulations within these severe storms are generally weaker than US severe storms. The benefits and limitations of the 3dvar approach and the usefulness of the 3DVAR system will be assessed and discussed in the conference.