Nowcasting and Very Short-range Forecasts of the Convective System: The Korean Perspective Dong-Eon Chang, Y. H. Lee, J.-C. Ha, H. C. Lee, Y.-H Kim Forecast Research Lab National Institute of Meteorological Research Pre-CAS TECO, 16-17 Nov 2009, Incheon, Korea
27
Embed
Nowcasting and Very Short-range Forecasts of the Convective System: The Korean Perspective
Pre-CAS TECO, 16-17 Nov 2009, Incheon, Korea. Nowcasting and Very Short-range Forecasts of the Convective System: The Korean Perspective. Dong-Eon Chang, Y. H. Lee, J.-C. Ha, H. C. Lee, Y.-H Kim Forecast Research Lab National Institute of Meteorological Research. Background. - PowerPoint PPT Presentation
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
Nowcasting and Very Short-range Forecasts
of the Convective System: The Korean
Perspective
Nowcasting and Very Short-range Forecasts
of the Convective System: The Korean
Perspective
Dong-Eon Chang, Y. H. Lee, J.-C. Ha, H. C. Lee, Y.-H Kim
Forecast Research Lab
National Institute of Meteorological Research
Dong-Eon Chang, Y. H. Lee, J.-C. Ha, H. C. Lee, Y.-H Kim
Forecast Research Lab
National Institute of Meteorological Research
Pre-CAS TECO, 16-17 Nov 2009, Incheon, KoreaPre-CAS TECO, 16-17 Nov 2009, Incheon, Korea
BackgroundBackground
Isolated thunderstorm
Cloud clusterConvection band
Squall line
Convection band
Cloud cluster
Squall line
Isolated thunderstorm
Not defined
Total
No. of events 31 53 8 13 8 113
Ratio(%) 27.4 46.9 7.1 11.5 7.1 100
Heavy rainfall events (2000-2006)
Lee and Kim (2007)
4 Typical heavy rainfall types
• In Korea 45 % of Casualties by natural disaster is caused by Heavy rainfall events (NEMA, 2006)
Forecast SkillForecast Skill
Forecast Length
Extrapolation
NWP
Fore
cast
Skill
Be
st
3-8 h
Explicit model
By J. Wilson (NCAR)
Position, Intensity
Initiation, growth
Extrapolation
○ ⅹ
NWP ⅹ ○
Nowcasting : 0~2hrNowcasting : 0~2hr
Very short-range forecast : Very short-range forecast : ~12h~12h
By WMO Tech Note No. 1024 By WMO Tech Note No. 1024
KLAPS : Korea Local Analysis and Prediction System
KLAPS Data IngestKLAPS Data Ingest
KLAPS Data Ingest : LightningKLAPS Data Ingest : Lightning
CTL
LGT
If Lightning(grid) ±30min-> cloud base = LCL-> fill the cloud cover 0.9-> cloud ω * 2
CTL
LGT
If Lightning(grid) ±30min-> cloud base = LCL-> fill the cloud cover 0.9-> cloud ω * 2
Lightning NetworkLightning Network
IMPACT (IMProved Accuracy from Combined Technology) - Sensor : IMPACT ESP, LDAR II - Method : MDF + TOA and TOA, Detect CG and CC - Period : Since March 2001
IMPACT (IMProved Accuracy from Combined Technology) - Sensor : IMPACT ESP, LDAR II - Method : MDF + TOA and TOA, Detect CG and CC - Period : Since March 2001
Build deep convective cloudBuild deep convective cloud
KLAPS Data Ingest : Radar KLAPS Data Ingest : Radar reflectivityreflectivity
uf_to_nc.exe
Remapping
(remap_polar_netcdf.exe)
Mosaic
mosaic_radar.x
……
Raw data (Polar coordinate)
Composite site (nearest site)
……
Remapping to Cartesian grid (each site)
Raw data (UF)
Polar netcdf file
3-D LAPS GRID(vxx)
2-D LAPS GRID(vrc)
3-D LAPS GRID(vrz)
Elev 0.0° Elev 7.03°
uf_to_nc.exe
Remapping
(remap_polar_netcdf.exe)
Mosaic
mosaic_radar.x
…………
Raw data (Polar coordinate)
Composite site (nearest site)
……
Remapping to Cartesian grid (each site)
Raw data (UF)
Polar netcdf file
3-D LAPS GRID(vxx)
2-D LAPS GRID(vrc)
3-D LAPS GRID(vrz)
Elev 0.0° Elev 7.03°
uf_to_nc.exe
Remapping
(remap_polar_netcdf.exe)
Mosaic
mosaic_radar.x
……
Raw data (Polar coordinate)
Composite site (nearest site)
……
Remapping to Cartesian grid (each site)
Raw data (UF)
Polar netcdf file
3-D LAPS GRID(vxx)
2-D LAPS GRID(vrc)
3-D LAPS GRID(vrz)
Elev 0.0° Elev 7.03°
uf_to_nc.exe
Remapping
(remap_polar_netcdf.exe)
Mosaic
mosaic_radar.x
…………
Raw data (Polar coordinate)
Composite site (nearest site)
……
Remapping to Cartesian grid (each site)
Raw data (UF)
Polar netcdf file
3-D LAPS GRID(vxx)
2-D LAPS GRID(vrc)
3-D LAPS GRID(vrz)
Elev 0.0° Elev 7.03°
Operational FeaturesOperational Features
• Forecasts(~12h) guidance ready by 42 min from initial time
• 3D analysis is produced within 10 min each hour
3D Analysis (every 3D Analysis (every hour)hour)
Forecasts (every 3 Forecasts (every 3 hour)hour)
Diabatic InitializationDiabatic Initialization• Diabatic initialization is unique technique of the KLAPS for the
improvement of precipitation forecast in the early integration time.
• Variational adjustment process is applied to produce dynamically balanced wind fields
Effect of Diabatic InitializationEffect of Diabatic Initialization
Optimization of initialization- Seeking optimal cloud updraft- Tuning of radar reflectivity threshold
Ingest of VAD wind Adapting WDM microphysics scheme
Wind Profiler
VAD
Optimization of Cloud updraft Optimization of Cloud updraft velocityvelocity
- W to height ratio Cu types (0.5)
- W to height ratio Sc types (0.05)
- W for St (0.01)
0.45.0 1 x
50.005.0 2 x
05.001.0 3 x
Wmax = depth * / dx for CuWmax = depth * / dx for Sc W = for St
1x2x3x
i
iETS Fitness 50,,2,1 i
Genetic AlgorithmGenetic Algorithm
Start
Initialization
Fitness Evaluation
Selection
Crossover
Mutation
Fitness Evaluation
Terminal condition
End
NO
YES
The Genetic Algorithm (GA) is a global optimization approach based on the Darwinian principles of natural selection.
This method, developed from the concept of Holland [1975], aims to efficiently seek the extrema of complex function .
The PIKAIA seeks to maximize a function f(X)
in a bounded n-dimensional space,
),,,( 21 nxxx 0.1,0.0kx Each generation has 20 chromosomes. The
crossover probability is set to 0.85, implying that 85% of the chromosomes in a generation are allowed to crossover in an average sense. The maximum and minimum mutation probability is set to 0.05 and 0.005, respectively.
Parameter estimationParameter estimation
animation
• GA shows quick convergence. The parameter X1 converged within 5~6th generation.
• Optimal value X1 = 3.95 X2 = 0.22 X3 = 0.035
Optimization ResultsOptimization Results
CTRL Optimized Exp
AWS
RADAR
6h rainfall
50~80 mm50~80 mm
Performance - examplesPerformance - examples
Performance - examplesPerformance - examples
KLAPS vs Regional ModelKLAPS vs Regional Model
• Precipitation verification score (ETS) for Jun – Aug 2009
Summary and ConclusionSummary and Conclusion
MAPLE with KMA operational radar observation provided useful guidance up to 2~3hr.
Diabatic initialization of KLAPS showed promising results in the very short-range precipitation forecasts, and optimization of some parameters using GA was quite successful and efficient.
In the future, blending of MAPLE and KLAPS precipitation forecast will be tested.
Thank youThank you
■ ETS ■ BIAS
Default :10.3 :
0.310.2 :0.210.1 :0.11 :
xxxx
Sensitivity to parameter X1Sensitivity to parameter X1
Verification ScoresVerification Scores
ETSETS BIASBIAS
thresholdthreshold::
1mm/1mm/3hr3hr
threshold:threshold:10mm/3hr10mm/3hr
WSM vs WDMWSM vs WDM
A CASE A CASE (INIT: 2008. 6. 18. (INIT: 2008. 6. 18. 00UTC)00UTC)