RTFDDA-3DVAR Hybrid System


RTFDDA (Real-Time Four-Dimensional Data Assimilation and forecasting system) is a mesoscale numerical weather modeling technology that has been developed for applications that desire rapid-update, high-resolution, high-accuracy, and customized weather information for specific regions. An important feature of RTFDDA is that it allows for smooth and uninterrupted assimilation of diverse weather observations and produces physically consistent and dynamically balanced 4D weather analyses and cloud/precipitation “spun-up” predictions.

RTFDDA integrates different community WRF data assimilation tools and activates these tools to formulate hybrid data assimilation that could provide the best data assimilation strategies for a given application scenario. For data sparse regions, an RTFDDA and 3DVAR (WRFDA and GSI) hybrid data assimilation approach has been developed, with RTFDDA taking a greater roles on fine-grid simulation of meso-beta and meso-gamma scale processes and 3DVAR on coarse grids for synoptic scale simulation. 3DVAR is able to assimilate satellite radiance measurements that provide important weather information which is valuable in regions where conventional observations are sparse. In addition, 3DVAR possesses a better ability to assimilate Doppler radar observations of radial velocities than RTFDDA, whereas RTFDDA with HLHN (hydrometer and latent heat nudging) is good at assimilate radar reflectivity. Thus, a RTFDDA-3DVAR-HLHN hybrid data assimilation scheme is ideal for effective radar and lightning data assimilation for convective weather prediction. 


One of the challenges for numerical weather prediction in the regions where conventional weather data are excessively sparse, such as over oceans, the Middle East, Africa etc., is the limitation of the observations that can be used for properly initializing mesoscale models. An RTFDDA-GSI hybrid data assimilation scheme (Fig. 1) has been developed in order to incorporate unconventional observations, especially remote sensing measurements such as satellite radiance, for model initialization. RTFDDA-GSI hybrid data assimilation approach is also employed to assimilate Doppler radar radial velocity observations in RTFDDA. A major benefit of the RTFDDA-GSI hybrid data assimilation technology over the typical GSI-only model initialization scheme is that it keeps the advantage of the RTFDDA in generating 4-D dynamically consistent and physically spun-up analysis and forecasts

Figure. 1 Schematics of the RTFDDA, GSI and HLHN hybrid for radar data assimilation.
Figure. 1 Schematics of the RTFDDA, GSI and HLHN hybrid for radar data assimilation. 

Another important utilization of GSI/WRFDA 3DVAR is to assimilate Doppler radar radial velocity. In this situation, Doppler radar reflectivity measurements are assimilated with a hydrometeor-latent-heat-nudging (HLHN) scheme where radar reflectivity is used to retrieve precipitation particles (snow, rain drops and graupel) which are then nudged into WRF along with adjustments of latent heat releases. Figure 2 shows a diagram of the RTFDDA-GSI-HLHN radar data assimilation (RDA) capability. Sensitivity study was conducted with the assimilation of radar reflectivity measurements with HLHN to study the impact of data frequencies between six minutes and one hour, relaxation strength.

Incorporating radar data into the real-time operational framework of RTFDDA has been one of the major undertakings of the next–generation RTFDDA development. To assimilate radar radial winds and reflectivity into the WRF-based RTFDDA system, a hybrid approach that couples RTFDDA and GSI with a hydrometeor and latent heat nudging (HLHN) technique has been developed. RTFDDA-GSI/HLHN hybrid radar data assimilation has been tested with retrospective case studies for convective systems in the Colorado Front Range, the Army Aberdeen Test Command (ATC) range, Redstone Test Command (RTC) range and Shenzhen metropolitan regions in China. A prototype RTFDDA-GSI-HLHN hybrid system was also run in real-time for the ATC domain. 

Plans for FY2019

Research on RTFDDA-GSI-HLHN hybrid data assimilation will be focused on enhancement of radar data assimilation (RDA) and lightning data assimilation (LDA) in 2019. GSI will be assessed and tuned for assimilating Doppler radar radial velocities. Strategies for nudging hydrometeors (rain, snow, and graupel mixing ratios) and the corresponding latent heat derivation from radar reflectivity and lightning observations will be studied. The RTFDDA-GSI-HLHN technology has been proposed for developing a real-time operational weather forecasting system for Indonesia, with a cloud-resolvable 3-km grid covering major islands and water bodies of the country and two 1-km grid model domains for concentrated oil and gas fields.