Operational RTFDDA


Real-Time Four-Dimensional Data Assimilation (RTFDDA) and forecasting technology has been developed to meet the need for high-resolution, accurate, and rapidly updated weather information for weather-critical applications.  directed at national defense and security, energy, emergency response and health.  It has been deployed for real-time operational weather support in over 50 applications areas (e.g., national defense and security, energy, emergency response, and health) by US government agencies, international organizations, and commercial entities. This section reviews the following operational RTFDDA NWP projects:  

  1. US Army Test and Evaluation Commands (ATEC) test ranges
  2. Advanced NWP for State Grid Corporation of China (SGCC) 
  3. MAGEN for the Government of Israel
  4. WRF-RTFDDA for wind power prediction of Xcel Energy
  5. CONUS-scale RTFDDA operation for Panasonic Weather Solutions
  6. RTFDDA high-resolution reanalysis and nowcasting for Shenzhen, China


US Army Test and Evaluation Commands (ATEC)

RTFDDA system serves eight Army test ranges located in the US and also supports on-demand test missions of ATEC in other regions aroundthe globe. Both an ensemble-RTFDDA system and a deterministic RTFDDA-LES system have been set up for operational forecasts at the Army Dugway Proving Ground, Utah. More detail on specific advances made in ATEC modeling systems can be found in the 4DWX section of this report.

Advanced NWP for the State Grid Corporation of China (SGCC)

Weather affects every aspect of electric power production from power generation, grid integration and transmission through dispatch and load.  With recent tremendous growth in renewable energy and the deployment of ultra-high-voltage, large capacity electric power transmission systems, the State Grid Corporation of China (SGCC) has critical needs for better weather information. RAL is working in collaboration with the SGCC’s Chinese Electric Power Research Institute (CEPRI) to meet these needs by applying and improving the WRF-RTFDDA technology to support weather modeling capabilities and to develop new advanced weather/climate tools.  Fig. 1 summarizes major NWP systems developed for real-time operational use in China.

MAGEN for Israeli Government

MAGEN (Model for Advanced GENeration of 4D Weather) employs RTFDDA and WRFDA-3DVAR hybrid data assimilation technologies to provide high-resolution weather guidance over the eastern Mediterranean region. The MAGEN system is enhanced with sand and dust prediction capabilities for real-time operational forecasts of dust and visibility. RTFDDA data assimilation is coupled with WRF-Chem online dust modeling for real time forecasting. A large single-domain RTFDDA-Dust model at a grid size of 9km was set up to cover the main dust emission and transport areas over the Mid-East and northern Africa that affect Israel. In addition, a new 3.3km grid high-resolution MAGEN forecasting is also developed to optimally take the advantage of the high-resolution ECMWF global model forecasts working as the initial and lateral boundary conditions. Case studies have been conducted and the results demonstrated the superior performance on the dust storm prediction and lower stratus simulation of the new modeling systems. The MAGEN-dust and the new ECMWF data driven 3km grid MAGEN modeling systems have been installed at Israel for real-time operation.

WRF-RTFDDA for SPIC Wind-Power Prediction

Figure 2. Streamlines (black lines with arrows) of the SPIC RTFDDA 6h wind prediction on Domain 2 (at grid intervals of 3 km), valid at 00:00UTC, Nov. 14, 2017. Terrain is plotted in shaded colors. Solid black dots mark the wind farms. Red vectors are the wind vectors from surface weather stations.
Figure 2. Streamlines (black lines with arrows) of the SPIC RTFDDA 6h wind prediction on Domain 2 (at grid intervals of 3 km), valid at 00:00UTC, Nov. 14, 2017. Terrain is plotted in shaded colors. Solid black dots mark the wind farms. Red vectors are the wind vectors from surface weather stations.    

In collaboration with the Renewable Energy Branch of the State Power Investment of China, the RTFDDA and ensemble RTFDDA technologies are being applied for wind prediction at three large wind farms in the middle China. The wind farms are located in two regions featured by complex terrain, including steep mountains and river corridors. Up to three days forecasts of general wind evolution and rapid-updated 0 – 6 h forecasts of wind ramps are required for all wind farms. To meet these goals, a high-resolution (with 1-km grids) RTFDDA system was designed to run hourly forecasting cycles for predicting wind ramps and a 3-km grid 30-member ensemble RTFDDA system is developed to produce 0 – 72 h forecasts at three-hour forecast cycles. Scientific research foci of this project include a) performing assimilation that maximizes the impact of very dense surface automatic weather station (AWS) observations, hub-height wind measurements of wind turbines at wind farms and wind farm met-tower weather observations for wind prediction. A next-generation RTFDDA data assimilation technology, known as Four-Dimensional Relaxation Ensemble Kalman Filter (4D-REKF), will be applied in these modeling systems; b) developing machine-learning post-processing tools based on the NCAR analog ensemble (ANEN) bias correction technology and quantile-regression (QR)-based ensemble forecast calibration technology to enhance the model prediction of wind turbine hub-height wind speed; and c) studying the WRF model dynamics and physics to improve its boundary layer flow simulation over complex terrain. Fig.2 shows a sample comparison of the model surface wind forecasts with the observations of the automatic weather stations.

PWS CONUS-scale RTFDDA Operation

Figure 3. Same as last year, no change. Update of land uses for the PWS-NCAR CONUS 4-km grid RTFDDA land use using the new land cover-based datasets. Many differences can be seen, especially the improved representation of the urbanization in the recent decades (red areas).
Figure 3. Same as last year, no change. Update of land uses for the PWS-NCAR CONUS 4-km grid RTFDDA land use using the new land cover-based datasets. Many differences can be seen, especially the improved representation of the urbanization in the recent decades (red areas).

NCAR and PWS (Panasonic Weather Solutions; formerly AirDat LLC) have been long-term partners in developing RTFDDA technology for TAMDAR (Tropospheric Airborne Meteorological Data Reporting) data quality-control, optimization of TAMDAR impact in regional NWP, and in developing operational RTFDDA forecasting systems. A CONUS-scale operational WRF-based RTFDDA data assimilation and forecasting system at 12/4-km resolution was deployed at PWS in 2009, and has been continuously running since then. A major task now underway focuses on enhancing the PWS-NCAR 12/4km CONUS RTFDDA system with radar data assimilation (RDA). More recent land cover data are used to better specify the RTFDDA model land use types, which significantly improves the representation of the current land surface properties, especially urbanization over the last 30 years (Fig. 3). The RAL hydrometeor-latent-heat-nudging (HLHN) radar data assimilation (RDA) scheme has been assessed with case studies of convective events with a 2-km grid CONUS-scale RTFDDA system. In case studies for two convective storms the impact of the radar data assimilation has been assessed. The system will be implemented on a super-computing cluster, newly purchased by PWS for real-time operational forecasting. Future work includes optimization of radar data assimilation in different regions with complex terrain.

RTFDDA High-resolution Reanalysis and Nowcasting for Shenzhen, China

Shenzhen is a major city located in the Pearl River Delta in southern China; the municipality, which includes both urban and rural areas, spans 2,050 square kilometers. NCAR has been collaborating with Shenzhen Meteorological Bureau (SZMB) to develop urban-scale precision climate reanalysis, real-time weather analysis and short-term weather prediction based on RTFDDA. The R&D goal is to effectively integrate a high-density observation network with advanced remote sensing instruments, including ultra-dense surface Automatic Weather Stations (AWS), wind profilers, radiometers, met-towers, Doppler radars, the Global Positioning System (GPS), lightning, and other platforms, into the RTFDDA system to provide continuous weather analysis and forecasts, and generate a five-year microclimatology for the Shenzhen area. The modeling system was configured with four nested domains with horizontal grid sizes at 27km, 9km, 3km and 1 km, respectively. The 1-km domain covers Shenzhen municipality, Hong Kong, and the neighboring area.  

The main accomplishments in FY2016 include: 1) developing a real-time operational rapidly updated urban-scale precision prediction system (RUPPS) at the HPC center of the Shenzhen Meteorological Bureau. The system runs hourly analysis and forecast cycles, producing 24 h forecasts at 1km grid spacing every hour; 2) deploying RTFDDA radar data assimilation (RDA) and lightning data assimilation (LDA) capabilities for the region and evaluating their impact on nowcasting of severe convection with case studies on 10 convective weather cases occurred during 2017; and 3) working with the SZMB scientists and engineers to generate climate service products based on the last 5.5 year of climate-FDDA reanalysis generated last year.


Research and development are our continuous effort in deploying and improving the RTFDDA technologies to support existing and new weather-critical applications. We strive to produce high-fidelity, high-resolution 4-D weather information, including microclimatology, current weather and short-term forecasting for these applications. The work will include advancing the core model sciences and technologies as well as enhancing the capabilities of on-going and new operational systems. Plans for several other on-going RTFDDA projects are as follows.

RTFDDA for SPIC Wind-Power Prediction

The work of this project during FY2018 will be focused on RTFDDA and ensemble-RTFDDA system refinements to achieve accurate boundary layer wind forecasts for the three wind farms. In addition, collecting and processing the wind farm data, including met-tower observations and wind-turbine nacelle wind measurements, and performing strict data quality control are critical tasks to warrant successful use of these data for RTFDDA data assimilation to improve the model forecasts and for machine-learning model-output post-processing.  The NCAR analog ensemble (ANEN) bias correction technology and quantile-regression (QR) ensemble forecast calibration algorithm will be employed for the model forecast post-processing. Objective verification and graphic display of the results will be accomplished to support the system refinement research and uses of the forecast products by end-users.

Israel MAGEN Systems

New research proposals will be initiated to significantly augment the capabilities of the existing MAGEN hybrid system to assimilate data from dust models to improve the dust forecasting capability. Plans for further enhancements to the MAGEN models also include increasing model resolution, cloud analysis and assimilation, ensemble data assimilation and probabilistic prediction, adaptive observation, and improvement of integration with the European Centre for Middle-range Weather Forecasts (ECMWF) model output

Joint Research Program with the China Electric Power Research Institute

RAL is working with CEPRI, a major electric power research institute under the State Grid Corporation of China, to establish a CEPRI-NCAR joint program for electrical power meteorological research and ensure a long-term, stable collaborative research program with multi-discipline electric power and meteorological sciences. CEPRI and NCAR will jointly develop advanced numerical weather prediction and data assimilation systems with the technologies including the Weather Research and Forecasting (WRF) model, Four-Dimensional Data Assimilation (FDDA), Ensemble-Kalman-Filter FDDA, Gridpoint Statistical Interpolation (GSI), Climate-FDDA, data quality control, advanced model output statistical bias correction, and ensemble calibration technologies.

Because of the diverse weather regimes across the country, the joint research program will conduct weather forecasting experiments to advance electric power weather simulation and forecasting capabilities. High-resolution weather reanalyses for generating a multi-year electric power weather/climate reanalysis historical database will be performed, and fine-scale weather modeling and forecasting systems toward renewable energy resource assessment and power forecasting applications and effectively support large-scale integration of renewable energy on to the power grids will be carried out. Electric-power meteorological disasters are the key research foci.  NCAR and CEPRI will jointly research and develop analysis and forecasting technologies and develop electrical power grids weather safety forecasting and early warning systems.

Panasonic Weather Solutions

Work will be continued to develop and optimize the NCAR-PWS CONUS-scale 2.5-km operational RTFDDA system, with assimilation of radar reflectivity. Case study and statistical verification of real-time operation will be conducted.  

Shenzhen Meteorological Bureau

RAL will continue to optimize the WRF-RTFDDA setting for SZMB area to reduce the simulated wind bias, especially over the ocean.  A new collaborative project has been established to implement and improve radar data assimilation and lightning data assimilation for the severe convection nowcasting and short-term forecasting at the Shenzhen area. SZMB has recently upgraded its weather radar system to a modern dual-polarimetric technology. It is important to take advantage of the hydrometeor-recognition capability in RTFDDA-RDA. In addition, several other tasks will be taken during FY2018 and beyond, including 1) refining RTFDDA observation data quality control to leverage the general use of the data besides the RTFDDA data assimilation; 2) developing observation sensitivity study (OSS), observation system experiment (OSE) and observation system simulation experiment (OSSE) technologies to provide guidance on planning future observation systems for the region; and 3) optimal use of QPE and QPF products to aid flooding warning for sub-districts and reservoirs in Shenzhen.