Satellite Wind Data Assimilation in HRRRAK-like Model

Location: National Center for Atmospheric Research, Boulder, Foothills Lab Building 2, Room 1001
Speaker:
Jiang Zhu, University of Alaska
Description:

High-Resolution Rapid Refresh for Alaska (HRRR-AK) model is one of the NOAA operational convective scale forecast system. Satellite wind data are not assimilated in the HRRR-AK model. Thanks to its high latitude, Alaska benefits from many polar-orbiter passes each day. Wind data derived from satellite observations have a good potential to improve the HRRR-Alaska short-term forecast. The purpose of the research is to investigate if the assimilation of satellite wind data can improve accuracy of the HRRR-AK model forecast. Experimental environment was setup in AWS cloud system, as well as the local machine in University of Alaska. The experimental model is HRRR-like model. It uses similar configuration, parameters, and initial fields as HRRR-AK model. In the experiment, the model run in three modes. Control run does not assimilate any observation data. Two experiment runs, one assimilates wind data only, and one assimilates wind data plus conventional observation, respectively. For simplicity, GSI 3D-Var analysis is used in the data assimilation experiments. The case study shows that valid VIIRS wind data is coarse in the domain; the impact of data assimilation of VIIRS wind data is very limited. The WRF model configured like HRRR-AK (smaller domain size and grid resolution) was used to do 24-hour forecasts 4 times daily. A month of forecasts are analyzed in terms of RMSEs. The preliminary conclusion shows that the VIIRS wind data does not improve the HRRR-AK-like model short-term forecast due to very coarse data distribution in the model domain. Further study includes evaluation of assimilation of all available satellite-derived wind data into the model.