SongYou Hong

1
Development of a new global forecast system in KMA

• Song-You Hong
• Yonsei University

2
Background (1) Performance of KMA GDAPS
500 mb GH RMS Error in 2001
3
Background (2) Status of the KMA GDAPS
JMA GSM as of 1989 - authority
KMA GDAPS
Flexibility -not compatible in workstation
Collaboration - Inaccessible from Universities
Need for a new model system
4
Strategy plan
Model
Data assimilation
Pre-evaluation
Design
1st yr (2002)
1st step
Pre-evaluation
Development
2nd yr (2003)
Setup the new model
Development
3rd yr (2004)
Parallel evaluation
2nd step
4th (2005)
Operational implementation
5th (2006)
5
Development team
International collaboration (NCEP, ECMWF, Scripps)
KMA (Operational evaluation)
Yonsei Univ. - Hong (PI)
Yonsei Univ -Hong (Physics)
KMA (Data assimilation)
Pukyung Univ -Chung (Dynamics)
6
Current status
-CVS-GSM on multi-platform machines with open-MP
and MPI Server set up at Yonsei University
(134.75.155.160) - Dry primitive global DFS
system setup with multiple levels Chung et
al. (2003) - YOnsei University Research
version modeling System (YOURS) (?? Not yet
developed, ? Not yet implemented ) Cumulus
parameterization scheme (Pan and Wu
1995, Hong and Pan 1998, Byun and Hong 2004 ??)
Vertical diffusion and PBL (Noh et al. 2003,
Hong et al. 2004) Shallow convection scheme
(Kim et al. 2004 ??) Explicit cloud scheme
(Hong et al. 2004) Gravity wave drag (Kim and
Arakawa 1995, Chun and Baik 1998) Radiation
(Chou et al, Byun and Hong 2004 ??) Land
surface (Pan and Mahrt, Chen and Dudhia, J. Hong
2004 ??)
7
Implementation
Resolution WRF, MM5, RSM and GSM (grid size
from 250 m to T62 ) Tests Case studies,
NWP, Seasonal, regional climate
8
Comparison of the performance of MRF and CCM3
Development of new physics parameterization
Future plans
9
(No Transcript)
10
(No Transcript)
11
Precipitation (Snowfall for 3-7 January 2001)
84hr fcst (MRF)
84hr fcst (CCM3)
GPCP Precip. (1.7.00UTC)
12
Precipitation (Heavy rainfall for 14-15 July 2001)
36hr fcst (MRF)
36hr fcst (CCM3)
GPCP Precip. (7.14.00UTC)
13
Precipitation (Heavy rainfall for 14-15 July 2001)
60hr fcst (MRF)
60hr fcst (CCM3)
GPCP Precip. (7.14.00UTC)
84hr fcst (MRF)
84hr fcst (CCM3)
14
Evaluation statistics
Region GL, NH, TR, SH Fcst hour 36, 60, 84,
108 hr
T126 5-day fcst
T213 cases snow heavy rainfall
15
2001? 7? bias
16
2001? 7? bias
17
24-h accumulated precipitation valid at 0000 UTC
15 July 2001
OBS
CCM3
SAS
18
Physics development

• PBL scheme
• Cloud scheme

19
The MRFPBL
Known problems and analysis of Stevens
(2000) Based on the Troen and Mahrt (1986)
Explicit representation of the entrainment
process Based on Noh et al. (2003, BLM)
Improvement of the K-profile model for the
PLANETARY BOUNDARY LAYER based on LARGE EDDY
SIMULATION DATA Y. Noh, W.G. Cheon and S.Y.
Hong S. Raasch
Too much mixing when wind is strong Too early
development of PBL Too deep and dry moisture in
PBL Too high PBL height
YSUPBL
20
Implementation of N2002
5
4
4
3
3
2
2
Theta-prime due to thermal Excess is less than 1
C
1
1
flux
21

• CASES 97 (Kansas)
• 24 hr simulation from 00Z 29 April 1997
• 10 km grid
• Comparison of YSU and MRF PBL heights

22
PBL height at 18Z
YSU
MRF
23
Cold front (10-11 Nov 2002)
Not in notes

• 4 km grid (cloud-resolving)
• YSU PBL compared to MRF PBL
• Showing how different pre-frontal soundings
  affect frontal convection

24
09Z 10 Nov 2002 CAPE
MRF
YSU
X
X
25
09Z 10 Nov 2002 Sounding
MRF
YSU
26
12Z 10 Nov 2002 Reflectivity
MRF
YSU
27
Cold Front case summary

• Shallower PBL with YSU PBL
• More CAPE with YSU PBL
• More convection at south end of front as observed

28
Experimental Design
Experiments
Ensemble runs

• boreal summer (JJA)
• 10 member ensemble run with NCEP Reanalysis II
  data
• LAF 5.110 0000 UTCs
• for three years with different SST conditions
• 1996 normal
• 1997 El Nino
• 1999 La Nina

29
Experimental Design
(mm mon-1)
30
Result
Precipitation of 1996 summer
SAS
RAS
MRF
YSU
Y-M
(mm mon-1)
31
Result
Pattern correlation (50S-60N)
SAS
RAS
32
Result
El Nino
Precipitation anomaly (1997-1996)
Pattern correlation
OBS
(mm mon-1)
MRF
YSU
SAS
RAS
33
Result
La Nina
Precipitation anomaly (1999-1996)
Pattern correlation
(mm mon-1)
34
Summary
New PBL SAS convection
Eastern Pacific
Western Pacific
Decrease of tropical rainfall Less mixing due
to New PBL -gt Less CAPE (Less convective
activities) Higher SLP -gt Intensifying
Walker circulation
Increase of tropical rainfall Larger
low-level moisture advection due to the
intensified Walker circulation -gt More CAPE
(More convective activities) Lower SLP -gt
Intensifying Walker circulation
35
WSMMP (WRF-Single-Moment- MicroPhysics)Hong,
Dudhia and Chen (2004)

old
new
23 25 June 1997 Heavy Rainfall Case (Vertically
integrated cloud ice)
36
Case description

• Case1 14-15 July 2001

• Case2 23-25 June 1997

37
Model setup
38
Experiment design

• 5 species (qv, qc, qi, qr, qs)
• WSM5 experiment excuted by applying new
  microphysics developed by Hong et al. to mixed
  phase scheme in MM5

39
MM5 Results Case1
WSM5-MSM5
WSM5-NOVI
Sedimentation of cloud ice effect
New microphysics effect
40
MM5 Results Case2
WSM5-MSM5
WSM5-NOVI
Sedimentation of cloud ice effect
New microphysics effect
41
MM5 Results
Case2
Case1
42
MM5 Regional Climate Run
NOVI
WSM5
MSM5
Accumulated precipitation from 0000 UTC 01 Jul
2002 to 0000 UTC 31 Aug 2002
43
MM5 Regional Climate Run
WSM5-MSM5
WSM5-NOVI
WSM5
300 mb Temperature bias from analysis averaged
from 0000 UTC 01 Jul 2002 to 0000 UTC 31 Aug 2002
44
Bulk parameterization

• RH83, WSM5

• D89, WSM3

- 5 species of hydrometeors
- 3 species of hydrometeors
45
Results Korea WSM6
24hr acc. prcp.
PLIN
OBS
WSM6
rain graupel snow cloud ice cloud water
0?
46
Results United States WSM6
24hr accumulated precipitation valid at 1200 UTC
16 June 2002
WSM6
PLIN
OBS
47
Results United States WSM6
Domain averaged precipitation Time series
WSM6
PLIN
? Weisman et al., 2003 Preliminary results from
Explicit convective forcasts using WRF model
48
Dynamical core

• Terrain following coordinate system  
• Fully compressible fluid system  
• Numerical method
• - horizontal plane Double Fourier
  Spectral method
•     - vertical direction linear finite
  element method
•     - time-discretization semi-implicit
  w-velocity solver

49
Dynamical core
50
DFS model (I)Barotropic Instability and Vortex
Amalgamation
51
Strategy
Dynamics (DFS) Physics (YOURS) gt
Next-generation KMA GSM
Time schedule
2004.12 Implement DFS and YOURS A meta
version 2005.12 Construct the code A final
version
Implementation
Resolution WRF, RSM and GSM (grid size from 250
m to T62 ) Tests Case studies, NWP,
Seasonal, regional climate
52
The End ?????.