Initialization and Prediction of Hurricane Ike via Assimilation of Airborne Doppler Radar Radial Velocity Observations using the ARPS 3DVAR

In this study, the impact of assimilating airborne Doppler radar (ADR) retrieved 3-D wind and radial velocity observations on the initialization and prediction of Hurricane Ike (2008) is examined using a three-dimensional variational (3DVAR) system in the Advanced Regional Prediction System (ARPS). Comparing the results of assimilating retrieved 3-D wind and radial velocity respectively with of no assimilation of any observation data (CNTL initialized by GFS 0.5 degree analysis data), it shows that both of assimilation experiments can improve the prediction of track and intensity for Hurricane Ike, but the experiment (EXPVr) which assimilates ADR radial velocity directly has more improvement than that (EXPret) assimilates retrieved 3-D wind. The forecast of rain pattern and wind fields show that, over the forecasting period, the vortex structure of EXPVr is closer to and tighter than that of EXPret and CNTL.

Considering the partial coverage of ADR observations around the hurricane vortex, and the typically small scale structures radar data are supposed to observe, standard background error correlation scales typically used by 3DVAR system are often inappropriate, a set of experiments are conducted, which choose different horizontal influence radius ranging from 10 km to 140 km. The results show that both the intensity forecast as measured by the minimum central sea level pressure and the track prediction of all the above experiments are closer to observations than the experiment without ADR data. When increasing the horizontal influence radius, at the beginning, the mean error of intensity and track are both decreasing, and then both are enlarging. The turning point of track error taking advantage of horizontal influence radius, 40 km, is prior to that of intensity error having horizontal influence radius, 120 km. While the horizontal influence radius is larger than 40 km, although, for the Hurricane Ike intensity, the mean error is continuously reducing, the mean error of the track raises quickly. Comparing the intensity error lapse rate with the track error increasing rate as the horizontal influence radius is between 40 km and 120 km, the former is much smaller than the latter. So, when the observations only cover a part of hurricane vortex field, there exists a best horizontal influence radius. The intensity and track prediction can be improved through choosing the best value. Based on the above experiments, another batch of experiments are conducted which examine that the impact of different assimilation cycles from 1 to 4 times on the vortex structure of initial condition. The more assimilation cycles, the stronger three-dimensional winds around the hurricane vortex at the initial time.