Regional Data Impact Studies at NCAR And The JCSDA WMO Observation Impact Meeting, Geneva, Switzerland, March 27th 2008 Dale Barker, T. Auligne, M. Demirtas, H. C. Lin, Z. Liu, S. Rizvi, H. Shao, Q. Xiao, and X. Zhang National Center for Atmospheric Research, Boulder, Colorado, USA
39
Embed
Regional Data Impact Studies at NCAR And The JCSDA
Regional Data Impact Studies at NCAR And The JCSDA. WMO Observation Impact Meeting, Geneva, Switzerland, March 27th 2008 Dale Barker, T. Auligne, M. Demirtas, H. C. Lin, Z. Liu, S. Rizvi, H. Shao, Q. Xiao, and X. Zhang National Center for Atmospheric Research, Boulder, Colorado, USA. - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Regional Data Impact Studies at NCAR And The JCSDA
WMO Observation Impact Meeting,
Geneva, Switzerland,
March 27th 2008
Dale Barker, T. Auligne, M. Demirtas, H. C. Lin, Z. Liu, S. Rizvi, H. Shao,
Q. Xiao, and X. Zhang
National Center for Atmospheric Research, Boulder, Colorado, USA
WRF DA Research To Operations (NCAR/AFWA)
• NCAR/AFWA DA Program initiated in August 2006.• NCAR responsible for WRF-Var development and initial testing.• JCSDA provides Community Radiative Transfer Model (CRTM), etc.
• WRF Community contributions include radar, radiance (RTTOVS), GPS, etc.
• Data Assimilation Testbed Center (DATC) performs rigorous testing prior to ops.
AFWA: Pre-operational testing, implementation
NCAR (DTC, DATC): Extended period testing
NCAR/MMM (WRF-Var, ARW)
JCSDA(CRTM, GSI)
WRFCommunityR&D
Testbeds
Operations
1) WRF-Var Overview
2) Antarctica: COSMIC/AMSU/AIRS/MODIS Impact.
3) AFWA: AMSU Impacts
4) Korea: Radar Impacts.
5) Summary
Outline of Talk
WRF-Var Data Assimilation Overview
• Goal: Community WRF DA system for regional/global, research/operations, and deterministic/probabilistic applications.
• Techniques: 3D-Var, 4D-Var (regional), Hybrid Variational/Ensemble DA.
NOGPS 6 hourly cycling run without assimilating COSMIC data
WGPS 6 hourly cycling run with assimilating COSMIC data
1
WGPS_damp3 Same as WGPS , except uses damp_opt=3 in WRF
50mb
3
WGPS_250mb Same as WGPS , except assimilates COSMIC data below 250mb .
NMC method using forecasts for May 2004
250mb
50mb 31
NODA_10mb 12 hourly cold -start run of WRF
NOGPS_10mb Same as NOGPS , except for model top at 10mb
WGPS_10mb Same as WGPS , except for model top at 10mb and assimilation of
COSMIC up to 10mb
NMC method using forecasts
from NODA_10mb for Oct 2006
10mb
10mb
57
1
Sensitivity Study of Stratospheric COSMIC Data Assimilation
NOGPS WGPS WGPS_250mb WGPS_damp3 WGPS_10mb
WGPS_250mb vs WGPS & WGPS_250mb vs NOGPS:Assimilation of COSMIC data only in troposphere sustains positive impacts in troposphere and decreases the RMSE of T forecasts in stratosphere as shown in WGPS. WGPS_damp3 vs WGPS:The enhanced damping at the model top only marginally changes the RMSE of T(U) forecasts.
WGPS_10mb vs WGPS:Moving the model top to 10mb decreases the RMSE of U and T forecasts in the stratosphere.
RMSE of 36hr Forecasts wrt Sondes
Bias and RMSE of 36hr Forecasts of T wrt Sondes
Assimilation of COSMIC data: • Reduces the bias of T forecasts in the lower-middle troposphere and stratosphere• Decreases the RMSE of T forecasts below 70mb
– First-guess check (innovations < 3o). Error factor tuned from objective method (Desrozier and Ivanov, 2001)
• Imager AIRS/VIS-NIRDay only (cloud coverage within AIRS pixel <5%)
Thinning (120km)345 active data
Thinning (120km)345 active data
Warmest FoV696 active data
Warmest FoV696 active data
• Thinning
AIRS Impact: 36hr Fcst. Score vs. Sondes
Whole Domain High Latitudes (> 60S)
McMurdo
Pegasus North
Black Is.
Western Ross Sea / Ross Is. grids
McMurdo Region & AWS sites
2.2-km
6.6-km, 2.2-km grids
•
Gill
Minna Bluff
Mt. DiscoveryMt. Morning
Transantarctic Mtns.
Impact Of High-Resolution Cycling 2300 UTC 15 May— Hr 23
DA: With MODIS
25 25
ms-1ms-1
L
Von Karman vortex
•
DA: With reduced MODIS
DA - Conventional
CTRL - WRF with GFS ICs
34
Sfc
Win
ds
(ms-1
)
SL
P (
hPa)
Pegasus North Winds
Win
d S
pee
d (
ms-1
)
OBS:WRF:
Hr from 00 UTC 15 May
Win
d S
pee
d (
ms-1
)
Win
d S
pee
d (
ms-1
)W
ind
Sp
eed
(m
s-1)
Hr from 00 UTC 15 May
NoDA
DA - Reduced MODIS Conventional
DA - With MODIS
20
20
35
20
35
35 35
20
35.3
35.3 35.3
35.3 36.6
24.6
31.529.3
Record ends
Record ends Record ends
Record ends
3. AFWA: AMSU Impacts
24hr Forecast Verification Vs. Obs for AFWA Testbed
Conclusions:
1. Regional DA adds significant value (even without radiances).
2. Update-cycling (GFS first guess at 00/12 UTC) superior to full-cycling.
No Data Assimilation
“Update” Cycling
Full-cycling
Meral Demirtas, DATC
East Asia Domain (T46)
Land Use Category
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
• 162*212*42L, 15km• model top: 50mb• Full cycling exp. for a month
• 1 ~ 30 July 2007• GTS+AMSU
• NOAA-15/16, AMSU-A/B from AFWA• AMSU-A: channels 5~9 (T sensitive)• AMSU-B: channels 3~5 (Q sensitive)• Radiance used only over water • thinned to 120km• +-2h time window• Bias Correction (H&K, 2001)
• Compare to GTS exp.• Only use GTS data from AFWA
• 48h forecast, 4 times each day• 00Z, 006, 12Z, 18Z
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
Obs used in assimilation (from AFWA operational datafeed)
Vs. Profiler VSlightly positive impactBeyond 24h
Vs. Profiler USlightly negative impactwithin 24h
Vs. Sound TNeutral
Vs. Sound QNeutral/Slightly negative
Impact Of AMSU Radiances in T46 (Liu et al. 2009)Verification against assimilated obs
Vs. SATEM ThicknessPositive impact
Vs. GPS RefractivityPostive impact
Vs. AIRS retrieval TSlightly positive impact
Vs. AIRS retrieval QSlightly positive impact beyond 24h
Impact decreasesWith forecast range
LBC takes controlFor long range FC
Impact Of AMSU Radiances in T46Verification against unassimilated obs
Atlantic Domain (T8)
Land Use Category
• 361*325*57L, 15km• Quite compute-demanding for WRF forecast
• Select similar data type used by AFWA• No SSM/I retrieval
• GTS+AMSU+MHS (use NCEP BUFR rad.)• NOAA-15/16/18, AMSU-A, ch. 5~10• NOAA-15/16/17, AMSU-B, ch. 3~5• NOAA-18, MHS (similar to AMSU-B)• Radiance used only over water • thinned to 120km• +-2h time window• Bias Correction (H&K, 2001)
• 48h forecast twice each day• 00Z, 12Z
• Might not optimal to use all sensors/satellites at the first try, but I want to test the robustness of the system with all Microwave sensors which can be assimilated in WRF-Var now.