Impact of assimilating airborne Doppler radar velocity data using the ARPS 3DVAR on the analysis and prediction of Hurricane Ike (2008)

Abstract

The ARPS 3DVAR data assimilation system is enhanced and used for the first time to assimilate airborne Doppler radar wind observations. It is applied to Hurricane Ike (2008), where radar observations taken along four flight legs through the hurricane vortex 14 to 18 h before it made landfall are assimilated. An optimal horizontal de‐correlation scale for the background error is determined through sensitivity experiments. A comparison is made between assimilating retrieved winds and assimilating radial velocity data directly. The effect of the number of assimilation cycles, each analyzing data from one flight leg, is also examined. The assimilation of retrieved wind data and of radial velocity data produces similar results. However, direct assimilation of radial velocity data is recommended for both theoretical and practical reasons. In both cases, velocity data assimilation improves the analyzed hurricane structure and intensity as well as leads to better prediction of the intensity. Improvement to the track forecasting is also found. The assimilation of radial velocity observations from all four flight legs through intermittent assimilation cycles produces the best analyses and forecasts. The first analysis in the first cycle tends to produce the largest analysis increment. It is through the mutual adjustments among model variables during the forecast periods that a balanced vortex with lowered central pressure is established. The wind speeds extracted from the assimilated model state agree very well with independent surface wind measurements by the stepped‐frequency microwave radiometer onboard the aircraft, and with independent flight‐level wind speeds detected by the NOAA P‐3 aircraft in‐flight measurements. Twenty‐four hour accumulated precipitation is noticeably improved over the case without radar data assimilation.