Upload
benedict-wright
View
217
Download
0
Tags:
Embed Size (px)
Citation preview
Progress Update of Numerical Simulation for OSSE Project
Yongzuo Li
11/18/2008
ECMWF T799 nature run
Simulated squall line is focus of regional-scale nature run
Regional-scale Nature Run with MM5
• 9-km domain 521 × 553 × 30
• 3-km domain 511 × 661 × 30
• One-way nesting from 9- to 3-km grid
• Initial and boundary conditions from ECMWF T799 NR
• Integration period: 00 UTC 02 May to 00 UTC 04 May
• Parameterizations:– simple ice microphysics
– Blackadar PBL
– Kain-Fritsch cumulus (9-km domain)
– no cumulus scheme (3-km domain)
MM5 /WRF3 km
MM5 9 km
Simulation Results of MM5 9-km Grid Run
10 M wind Precipitation
Simulation Results of MM5 3-km Grid Run
10 M wind Precipitation
Sensitivity experiments with WRF (cold start)
• 3-km domain 511 × 661 × 30
• Initial and boundary conditions from MM5 9-km [ regional-scale NR ]
• One-way nesting from MM5 9-km grid
• Integration period: 00 UTC 02 May to 00 UTC 04 May
• Parameterizations:– no cumulus scheme
– simple ice microphysics
– YSU PBL
MM5 /WRF3 km
MM5 9 km
• Synthetic observations assimilated using WRF-VAR
WRF 3DVAR Background Error
1) Use NMC method to calculate WRF 3DVAR BE. Starting at 00Z and 12Z respectively, 24 hour forecast using WRF model are performed from 12 April to 14 May 2005. 12 hrs fcst
00Z12 12Z12 00Z13 perturbation 1 at 00Z13
24 hrs fcst 12 hrs fcst 12Z12 00Z13 12Z13 perturbation 2 at 12Z13 24 hrs fcst
WRF 3DVAR Cycling and Simulation
Synthetic observations are extracted from MM5 3-km grid simulation (section 4). 36 hours 00Z02 03Z02 06Z02 09Z03 12Z02 00Z 04 WRF 3DVAR cycling WRF simulation
From 00Z 02 to 12Z 02 WRF 3DVAR cycling is made to assimilate “true data” of synthetic observation from MM5 3-km.
Data assimilation Synthetic observations are taken from MM5 3-km grid.
02 May 02 May 02 May 02 May 02 May
00 UTC 03 UTC 06 UTC 09 UTC 12 UTC SFC SFC SFC SFC SFC UA SAT UA
WRF 3DVAR cycling every 3 hrs
Sites of synthetic UA soundings
5-minute ATMS / CrIS swath
Sites of synthetic surface obs
Synthetic satellite coverage
Current work: Satellite data assimilation Synthetic satellite observations of T and Td are taken from MM5 3-km grid.
02 May 02 May 02 May 02 May 02 May
00 UTC 03 UTC 06 UTC 09 UTC 12 UTC SFC SFC SFC SFC SFC UA SAT UA
WRF 3DVAR cycling every 3 hrs 5-minute ATMS / CrIS swath
Synthetic satellite coverage
Procedure:
Synthetic ATMS / CrIS data available within 3-km grid area from 0730 UTC to 1000 UTC on 02 May
Synthetic ATMS / CrIS data are thinned to closely match MM5 3-km grid points
MM5 T and Td are matched to ATMS / CrIS swath points in horizontal space and vertical space
New synthetic satellite observation dataset is assimilated at 0900 UTC in WRF 3DVAR, with appropriate errors added:
T (± 1K) Td (± 2K)
Comparison of MM5 and WRF Simulations
Comparison of simulation results from MM5, WRF cold start, and WRF data assimilation is made in this section. The results are very close to each other though very subtle difference is seen between MM5 and WRF results. Precipitation, sea level pressure, and 2 meter dew point temperature are compared in following slides.
At beginning, 00Z 02, sea level pressure fields of MM5 and WRF are almost identical as they are interpolated from the same MM5 9-km fields.
MM5 WRF
At beginning, 00Z 02, 2 meter dew point temperature fields of MM5 and WRF are also very clso as they are interpolated from the same MM5 9-km fields.
MM5 WRF
At 18Z 02, locations of WRF and MM5 rainfall are close. But obvious difference of rainfall patterns are found between WRF and MM5.
MM5 WRF 3DVAR
WRF
At 21Z 02, locations of WRF and MM5 rainfall are close. But obvious difference of rainfall patterns are found between WRF and MM5.
MM5 WRF 3DVAR
WRF
At 00Z 03, mean sea level pressure fields of MM5 and WRF are very similar whereas MSL pressure fields from WRF cycling run are closer to WRF cold start run than to MM5 run. It means that model similarity affects results at this moment more than data assimilation during first 12 hours does.
MM5 WRF 3DVAR
WRF
At 00Z 03, precipitation fields of MM5 and WRF are close. Rainfall band of MM5 moves a little faster than that of WRF. Rainfall pattern of WRF cycling is closer to that of WRF cold start than to that of MM5.
MM5 WRF 3DVAR
WRF
At 03Z 03, precipitation fields of MM5 and WRF are close. Rainfall band of MM5 still moves a little faster than that of WRF. Rainfall pattern of WRF cycling is closer to that of WRF cold start than to that of MM5.
MM5 WRF 3DVAR
WRF