The Relative Importance of Assimilating Radial Velocity and Reflectivity Data

The NEXRAD (WSR-88D) Doppler radar network allows meteorologists to track severe weather events and provide better warning information to the public, ultimately saving lives and reducing property damage. However, the assimilation of such data into NWP models to provide physically consistent three-dimensional analyses and short-term forecasts has not been extensively explored. Yet Doppler radar is the only operational instrument capable of providing observations of sufficient spatial and temporal resolution to capture convective-scale phenomena. Therefore, the effective assimilation of Doppler radar data into operational convection-resolving models is of increasing importance in our quest to extend warning lead times. Among the existing data assimilation methods, the 3DVAR system is a very efficient method that can use radar data in real-time mode and in very high resolution both spatially and temporally. In this paper, the impact of assimilating radial velocity and reflectivity data from a WSR-88D network near central Oklahoma is examined using both an idealized case and a real data case. The three-dimensional variational data assimilation system developed for the ARPS model (ARPS 3DVAR), combined with a complex cloud analysis package, is used to produce analyses in high-frequency intermittent assimilation cycles. Our purpose is to examine the relative importance of assimilating radial velocity and reflectivity data on storm-scale data assimilation and forecasting for very strong convective weather events.

For the idealized case, a set of experiments that differ in the type of data used are performed to identify the impact of radial velocity and reflectivity data when using different numbers of NEXRAD radars. It is found that by assimilating radial velocity data only, the model can predict the timing and evolution of a simulated supercell thunderstorm with great accuracy. In contrast, large errors emerge when only reflectivity data are assimilated. These errors are produced during the updating of hydrometer-related variables and the temperature adjustment that occurs in the cloud analysis package. For the observed Greensburg tornadic thunderstorm case of 4-5 May 2007, two preliminary experiments are performed. One uses only radial velocity and the other uses both radial velocity and reflectivity from several nearby radars. It is found that by assimilating only radial velocity data, the model can reconstruct the supercell thunderstorm that produced Greensburg tornado very well, while assimilating both radial velocity and reflectivity does not add much value. These initial results suggest that the assimilation of radial velocity data is essential for the prediction of supercell thunderstorms, likely due to their helical updrafts that play such an important dynamic role in storm development and evolution. Though reflectivity data is fundamental to storm tracking and Quantity Precipitation Estimation (QPE), the assimilation of such data into NWP models may be not as important as radial velocity, because reflectivity is related to more inactive model variables, and a lot of uncertainties in model microphysics further complicates its usage in storm scale NWP. However, for weaker thunderstorms reflectivity data may be very important.