Variational Doppler Radar Assimilation System (VDRAS) for Ramp Prediction Juanzhen (Jenny) Sun MMM/RAL, NCAR May 12, 2010
Outline Challenge of ramp forecast VDRAS - background and technique Real-time low-level frequent wind analysis 0-6 hour convective forecast initialized by high-resolution data Summary May 12, 2010
Challenge of ramp prediction Sudden wind change associated with thunderstorms Current operational NWP has difficulty in forecasting the pattern and timing of storms Require high resolution data (radar, lidar, mesonet, ) and advanced data assimilation techniques May 12, 2010
Current Skill in Rainfall Prediction Critical Success Index (CSI) 1.0.8.6.4.2 Accuracy of Rainfall Nowcasts NWP >1 mm/h GRID MESH 20 km Jun-Oct 2002 Courtesy of Shingo Yamada JMA 0 1 2 3 4 5 6 Forecast Length, hours
General description of VDRAS VDRAS is an advanced data assimilation system for high-resolution (1-3 km) and rapid updated (12 min) wind analysis The main sources of data are radar radial velocity, reflectivity, and highfrequency surface obs. The core is a 4-dimensional data assimilation scheme based on a cloudscale model Nowcasting can be produced by the cloud model or providing initial conditions for WRF VDRAS has been installed at more than 20 sites for various applications May 12, 2010
History of VDRAS Development milestones 1991: First version of VDRAS developed and successfully applied to simulated radar data (Sun et al 1991) 1997: Extended to a full troposphere cloud model (Sun and Crook 1997,1998) 2001: Applied to lidar data for convective boundary layer analysis (VLAS) 2005: Added the capability to cover multiple radars (Sun and Ying 2007) 2007: Coupling with mesoscale models (mm5 or WRF) 2008: Began to explore how to use VDRAS analysis to initialize WRF May 12, 2010
History of VDRAS cont Real-time installations 1998: Implemented at Sterling, NWS 2000: Installed at Sydney, Australia for the Olympics 2000-2005: Field Demonstration for FAA aviation weather program Beijing 2008 Olympics 2003-now: Run at various mission agencies ATEC Dugway Utah 2006-2008: Real-time demonstration for Beijing Olympics 2008 Currently: NWS at Melbourne, Florida NWS at Dallas, Texas ATEC at Dugway, Utah Beijing, China Taipei, Taiwan May 12, 2010
Major components of VDRAS Data Ingest Rawinsondes Mesoscale model Profilers Mesonet Doppler radars/lidars Data Preprocessing Quality control Interpolation Background analysis First Guess Output &Display (CIDD) Plots and images Animations Diagnostics and statistics 4DVAR Assimilation Cloud-scale model Adjoint model Cost function Weighting specification Minimization May 12, 2010
How VDRAS analysis is produced with time 0 min 4DVar 12 min 18 min 6-min Forward Integration 4DVar 30 min time KVNX KDDC KICT KTLX Cold start Mesoscale analysis as first guess 6-min Forecast as first guess; Mesoscale analysis Output of u,v,w,t,qv,qc,qr Output of u,v,w,t,qv,qc,qr Model data Sounding VAD profile Surface obs. Model data Sounding VAD profile Surface obs.
VDRAS analysis by assimilating 8 NEXRADs Radar radial velocities over IHOP domain Analyzed temperature Red contour: 25 dbz ref.
High-resolution data assimilation reveals how low-level wind evolves with cold pools 0611 2046 UTC - 0612 1250 UTC; every 24 min Pert. Temp. (color) Wind vector at 0.1875km (black arrow) Red contour (25 dbz observed reflectivity) QuickTime and a BMP decompressor are needed to see this picture. 4DVar analyses with radar data assimilation via VDRAS
VDRAS wind analysis over complex terrain in Taiwan Red contours: observed 25 & 35 dbz reflectivity black contour: 100 meter terrain line QuickTime and a BMP decompressor are needed to see this picture. SPol
VDRAS Verification from previous studies Dual-Doppler verification ACARS (Sun and Crook 2000) Cpol Dual-Doppler (Crook and Sun 2004) Kurn ell Research aircraft (Sun and Crook 1998) Profiler, AWS (Sun et al. 2010) rms(u dual u vdras ) = 1.4 m/s rms(v dual v vdras ) = 0.8 m/s May 12, 2010
VDRAS verification for Olympics 2008 FDP VDRAS cold pool compared with AWS
0-6 hour forecast is very sensitive to initial conditions No radar Physics Initial conditions With radar
2-hour forecast of a gust front using VDRAS cloud model Advantage: Balanced initial Conditions QuickTime and a Cinepak decompressor are needed to see this picture. Disadvantage: The model is Relatively simple
Inserting VDRAS analysis into WRF inner domain Interpolated fields of VDRAS analysis (MESO) to RTFDDA_d02 U-wind at the 1 st level Without (left) and with (right) weighting on RTFDDA 19 UTC 15 June 2002 VDRAS RTFDDA_d02
2-hour WRF forecast No VDRAS OBS With VDRAS
5-hour WRF forecast No VDRAS OBS With VDRAS
WRF Radar data assimilation Front Range Radar data assimilation testbed through NCAR s Short Term Explicit Prediction (STEP) program WRFDA 3DVAR + DDFI WRFDA 3DVAR + RTFDDA WRFDA EnKF HRRR with second-pass DDFI Obs. No radar With radar
Summary Accurate nowcasting of thunderstorms and the associated sudden wind change is critical for ramp prediction VDRAS produces robust and accurate high temporal and spatial resolution analysis of wind by combining radar, surface, and mesoscale model data and has a good potential for application of ramp prediction High-resolution data assimilation systems that aim at the improvement of 0-6 wind and precipitation forecasts are being actively developed and tested at NCAR May 12, 2010