Assimilation of Aircraft Reconnaissance Observations with HWRF Hybrid Data Assimilation System

The initialization of the HWRF model has been largely relied on an advanced vortex initialization scheme with no inner core data assimilation. With the upgrade of the GFS data assimilation system to a hybrid EnKF-3DVAR system, an economic one-way hybrid data assimilation system has been developed and implemented for operational HWRF starting with the 2013 hurricane season. The forecast ensemble generated from the global hybrid system is used in GSI to provide flow-dependent background error covariance.

Although initial efforts of assimilating NOAA P3 Tail Doppler Radar (TDR) data into HWRF using static 3DVAR data assimilation scheme didn't show positive impact, the assimilation of TDR data from 19 P3 missions with the hybrid data assimilation scheme significantly improved intensity forecasts, as seen in our HFIP real-time forecasts during the 2012 hurricane season.

Encouraged by those results and in support of the HFIP program, we further extended our capability of assimilating aircraft observations including inner core dropsonde, flight level and SFMR (Stepped Frequency Microwave Radiometer) 10m wind speed retrievals into our data assimilation system. Retrospective experiments have been conducted using aircraft observations collected from 2008 to 2012 to evaluate the impact of TDR data versus the impact of flight level and SFMR data. The average intensity forecast error with the assimilation of TDR data from 84 P3 missions is found reduced by about 10%. The assimilation of flight level and SFMR data can slightly improve track forecasts and the intensity forecasts after 30 hours forecast lead time. More details of these results and potential future efforts will be presented at the conference.

The assimilation of TDR data into HWRF became operational in 2013 hurricane season. The first cycle with TDR data assimilated for tropical storm Karen changed the forecast scenario of this storm and had positive impact on the forecast guidance of NHC. In preparing for the 2014 HWRF implementation, we try to hence data assimilation by assimilating more observations, improving background error covariance and using hourly FGAT. The impact of TDR data in the 2014 version of HWRF will also be presented at the conference.