Tuesday, 5 June 2018: 2:15 PM
Colorado A (Grand Hyatt Denver)
Following the successes of the research, development and operational implementation of the GSI based hybrid ensemble-variational (EnVar) data assimilation (DA) for the global and convection allowing hurricane (HWRF) modeling systems (Wang et al. 2013; Wang and Lei 2014; Kleist and Ide 2015ab; Lu, Wang et al. 2016, 2017), the GSI-based EnVar hybrid systems are further developed to 1) use ensemble produced by regional model’s own cycled EnKF in place of readily available GFS ensemble member to provide flow-dependent error covariance, and to 2) enable the convective-scale radar data assimilation capability including the direct assimilation of both the radial velocity and reflectivity observations (Wang and Wang 2017). The new direct reflectivity assimilation replaces the current operational diabatic digital filter and cloud analysis. These new developments have been tested in the operational HRRR and NAMRR contexts. A single domain covering the entire COUNS region is employed with a 3km resolution. Conventional data (i.e., surface station, wind profiler, rawinsonde soundings, etc.) are assimilated hourly; while radar DA can be conducted hourly or sub-hourly.
Ten retrospective forecasts during 2015 and 2016 are used. The examined configurations include localization scales respective for conventional and radar DA, the Relaxation to Prior Spread (RTPS) inflation factor, and the radar DA frequency and length. Results show that, 1) experiments with relatively small horizontal and vertical localization scales in conventional DA perform much better than large localization scales, especially for the weakly forced cases; for radar DA, experiments with large horizontal and vertical localization scales produce more accurate storms; 2) a higher RTPS factor performs better than a smaller one; 3) as increasing radar DA length and frequency, the system improves the convective-scale predictions. Experiments also show the new direct radar reflectivity approach outperforms the cloud analysis. The impact of the static covariance will also be presented in the conference.
The new system was tested in real time during the 2017 HWT spring experiment. Various verifications and diagnostics were performed. The system is also planned to be tested in real time during the 2018 HWT spring experiments. Results will be shared during the conference.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner