Progresses of BMB project in 2006

1
Progresses of BMB projectin 2006

• Y.-H. Kuo, X.Y. Huang, Y.-R. Guo, and
• Jiqin Zhong
• 28 June 2006

2
BMB B08 operational data assimilation system
NCAR provide Default BES and Software to derive
BES
From NMC
Beijing local data
Conventional data Decoder
First guess WRF Converter
Background Error Statistics (BES) Derivation
GPS TPW AWS Wind-profile Decoder OBS error
?
WRF-Var 3D-Var
cycle
?
WRF Model
Lateral boundaries WRF Converter
6h (3h) forecast
From NMC
Verification package
?
NWP products Other applications
48-h forecast
3
System configuration

• Model domains
• D01 27 km, 15115138
• D02 9 km, 14218438
• D03 3 km, 17219938
• D04 1 km, 21121138 (experimental)
• Analysis scheme
• The latest released version of WRF 3D-Var
• (current OPR MM5 3D-Var)
• D01 and D02 6 hour cycles
• D03 3 hour cycles
• D04 1 hour cycles (experimental)
• Forecast model
• The latest released version of WRF model
• (Current OPR MM5 model)
• Data
• Conventional data and Beijing local data

4
Expected pre-operational and operational
upgrades

• Should follow the table
  Project started March 2005
• Depends on the pre-operational results
  This table is
  version 200512

5
Progresses in 2006Huang, Zhong, and Guo

• BES transfer from CV2 to CV5
• FG transfer from T213 to AVN
• Local OBS data Wind profiler, AWS, and GPS PW
• Obs data processing wind profiler (u,v,w) to
  (ff,dd)
• AWS data SLP unit, observation
  error specifications
• GPS PW observation errors
• WRFVar code bug fixed SFC_assi2 rh_check for
  single OBS test
• Forecast model transfer from MM5 to WRF
• Model settings time-step, physics, interval
  for radiation
• Input files MM5 REGRID to WRF_SI
• Static fields Landuse table from MM5, 27-km
  single domain terrain.
• Verification transfer to WRFVar-based VERIFY
  package
• WRFVar-based VERIFY applied to MM5 forecast.

6
(No Transcript)
7
CV5 Background Error Statistics
8
New BMB BE (cv_option5)
Psi first 5 global eigenvector
Scale_length of control variances
9
Single T obs test (O-B) 1, so 1.0
(b)
(a)
Increment and scale are too small
(d)
(c)
Response in low level is unreasonable
The T/V increments cross-section (West-East
about 2000 km) for Single T OBS tests in BMB
Domain1 (a) T response with old BE (b) T
response with new BE (c) V response with old BE
(d) V response with new BE. The OBS value and
error Is 1.0 located at x76, y 76, z22. The
domain is 151x151x37 with 27-km grid distance.
10
Single T obs test (O-B) 1, so 1.0
Unreasonable U-wind response on d0.983 level
(with old BE)
Reasonable U-wind response on d0.983 level (with
new BE)
11
Single u obs test (O-B) 1, so 1.0
(b)
(a)
(c)
(d)
bad response
(a) U response with old BE (b) U response
with new BE (c) T response with old BE
(ond0.983 level) (d) T response with new BE
(ond0.983 level). The OBS value and error Is 1.0
located at x76, y 76, z22.
12
Conclusions

• CV5 BES better than CV2 BES
• FGAVN12f better than T213, WRF12f equivalent to
  AVN12f
• SFC_assi1 equivalent to SFC_assi2
• Upper air WRF better than MM5,
• Surface MM5 better than WRF
• MM5 landuse table better than WRF landuse table.

13
Experiment design

• Exp.11 WRFVar, CV5 BES, FGavn12f, MM5 model,
• passed4, var_scaling1, len_scaling1,
• check_rh2, sfc_assi1,
• t_err2C, p_err100pa, v_err1.1m/s rh_err10
• Exp.15 Same as Exp.11 but sfc_assi2
• Exp.17 Same as Exp.11 but with WRF model
• Exp.18 Same as Exp.17 but FG wrf12f

14
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 gt exp15
exp17 exp18
exp11 gt exp17
15
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 exp15
exp17 exp18
exp11 gt exp17
16
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 gt exp15
exp17 exp18
exp11 gt exp17
17
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 lt exp15
exp17 lt exp18
exp11 lt exp17
18
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 lt exp15
exp17 exp18
exp11 lt exp17
19
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 exp15
exp17 lt exp18
exp11 lt exp17
20
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 gt exp15
exp17 exp18
exp11 lt exp17
21
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 gt exp15
exp17 exp18
exp11 gt exp17
22
Exp11 - mm5 sfc1 anv12f Exp15 - mm5 sfc2
anv12f Exp17 - wrf sfc1 anv12f Exp18 - wrf
sfc1 wrf12f
?? gt ?????? ?? ?? lt ??????
exp11 exp15
exp17 lt exp18
exp11 lt exp17
23
(No Transcript)
24

• Exp.19 the same to exp17, but radt15,
  time_step150, mp_physics3

25

• Exp.17_newtbl Same as the Exp17 but with
  landuse.tbl file used in MM5 model

26
27km domain TERRAIN DIRRERENCE WRF-MM5_nest
27
27km domain INPUT DIFFERENCE (variable sfcp)
WRF-MM5
28
27km domain TERRAIN DIFFERENCE of MM5 nest -
single
29
27km domain TERRAIN DIRRERENCE WRF-MM5_single
30

• Exp21 MM5 model with TERRAIN_DOMAIN1 of single
  configure

31
Impact of the AWS data assimilation

• According to the verification scores based on
  SYNOP and TEMP observations, AWS data have minor
  negative impact on analyses and neutral impact on
  forecasts.
• According to the verification scores based on AWS
  observations, AWS data have clear positive impact
  on analyses and positive impact on forecasts of
  some variables, for example, mixing ratio.

32
AWS assimilation Verification against the SYNOP
Exp31, WRF MODEL , TEMP SYNOP, Exp32, WRF
MODEL , TEMP SYNOP METAR
33
AWS assimilation
Exp31, WRF MODEL , TEMP SYNOP, Exp32, WRF
MODEL , TEMP SYNOP METAR
34
AWS assimilation
Verification against AWS
Exp31, WRF MODEL , TEMP SYNOP, Exp32, WRF
MODEL , TEMP SYNOP METAR
35

Impact of the GPS PW assimilation

• Exp31, WRF MODEL , TEMP SYNOP,
• Exp33, WRF MODEL , TEMP SYNOP GPSPW with
  default error of 0.2 cm
• WRFVar-based VERIFY can be directly applied to
  non-conventional observations, such as GPS PW.
• No matter RMS or Bias, GPS PW assimilation has
  the positive impact.

36
(No Transcript)
37
The work remained

• Nested run experiments (27/9/3km)
• for mesoscale data assimilation, 27-km single
  domain is not enough
• BES derivation for high resolution, or
  interpolate_stats
• Tuning and multiple outer-loops
• Cycling mode experiments
• Warm-start for operational
• DFI implementation (MM5/WRF)
• Verification package
• WRFVar-based 3D-VERIFY,
• Precipitation-VERIFY
• Case selection
• Convective case in summer season
• Real time GPS data processing and assimilation
• Implementation of ZTD assimilation in WRFVar