1. An overview of the KIAPS numerical weather prediction systems
Song-You Hong and KIAPS staffs
(Korea Institute of Atmospheric Prediction Systems: KIAPS)

2. Three-stage development plan of KIAPS Global Model
step3
step2 .
■ Finalize the KIAPS operational version 1.0 and run on semi-operational basis
■ Stabilize the model system by further diagnostics and verifications
■ Release the KIAPS system to international users
Finalize the
operational system
step1
Release beta version
■ Complete developing major model components based on KIAPS own research
■ Release the KIAPS beta version model
■ conduct semi realtime experiments with KIAPS beta version and evaluate it skills to the KMA operational model
Foundation and basic
research
■ Establish the foundation of KIAPS research and development environment
■ Efforts on laying out future model development – model dynamics, physics and data assimilation

3. Spectral element method
KIM Configuration (2016.8)
scheme
References
Numerical model
-Dynamical
core
-Physics
Equation
Nonhydrostatic
Choi et al. 2014
Choi and Hong 2016
Horizontal discretization
Spectral element method
Radiation
RRTMG
Iacono et al. 2008
Land surface
Noah V3.4.1
Ek et al. 2003
Koo et al. 2016
Ocean surface layer
Kim and Hong
Kim and Hong 2010
Vertical diffusion
YSU
ShinHong
Hong et al. 2006
Shin and Hong 2015
Lee et al. 2016
Gravity wave drag
O: Kim-Arakawa
Hong et al. 2008
Choi and Hong 2015
C: Chun and Baik
Chun and Baik 1998
Deep conv.
SAS
Han and Pan 2011
Lim et al. 2014
Han et al. 2016
Shallow conv.
GRIMs
Hong et al. 2013
Microphysics
WSM5
Hong et al. 2004
Bae et al. 2016
Cloudiness
Prognostic
Park et al. 2016
Assimilation
OPS
KPOP
Var
3DVar
Song and Kwon 2015
LETKF
Framework
Semi-real run
Cylc

4. Dynamical core

5. Atmospheric Model test
Development of Non-hydrostatic Dyn.
WRF governing equations are adapted
- Cubed sphere, horizontal: Spectral element, vertical: FD, terrain-following
- Flux-type compressible governing equations in perturbation form with prognostic variables
- Time-split time integration: Slow mode  third-order Runge-Kutta
Horizontal sound wave and gravity wave  Forward-Backward
vertical sound wave and buoyancy  implicit
2D Shallow Water Test
2D (x-z) Slice
Dry Dynamical Core Tests 3D
Dry Dynamical Core Tests 3D
Aqua-Planet Experiments (APE) 3D
Atmospheric Model test
1. Advection of cosine bell
2. Steady state
3. Translating low
4. Zonal flow over an isolated mountain
5. Rossby-Haurwitz wave
1. Linear hydrostatic mountain wave
2. Simulations The Schar-type mtn.
3. Density current
4. Inertial-gravity wave
5. Rising thermal bubble
(Choi et al., 2014)
1. Steady-sate
2. Baroclinic wave
3. 3D Rossby-Haurwaitz wave
4. Mountain-induced Rossby wave
5. Pure gravity wave
6. DCMIP pure advection test
7. DCMIP Mountain-wave over a Schar-type
8. Held-Suarez
1. Control APE
2. 3KW1 APE
1. the 5-case for a severe weather
2015
(Choi and Hong, 2016)

6. Increase the order of the diffusion operator
The fourth-order diffusion scheme  sixth-order diffusion
Outer loop time-split horizontal diffusion (5 times per a time-march)
Increased dt and enhanced energy spectra are achieved.
ne120 dt
Diffusion order
Diffusion Coeff.
KIM2.2(CTL) 15 2
1.8E13
KIM2.3 60 3
3.0D21

7. New profiles of A and B for hybrid-sigma coordinate
Lower level associated with pure pressure coor. (15  33 hPa)
To get rid of discontinuity for layer depth (dB/dη)
TMPA
OLD (50lev)
KIM2.4
KIM2.5
NEW (50lev)

8. Variable resolution using Schmidt transform
[Example ne6np4 grid with Magnification factor S = 2]
[Baroclinic Instability test at 9-day]
[DCMIP51 – Idealized Tropical Cyclone experiment w/ simple-physics]

9. Physics modules

10. RADIATION V2.4
REDU-MCICA: Increase the efficiency of MCICA with keeping accuracy (reduced 30% computational time)
SNOW ALBEDO: Improve
the cold bias at the lower levels
RRTMG O3 input data
Improvement of upper level T forecast skills
CTL
GMAO
TROP
TROP

11. GWDO V2.4
Resolve the problem of too strong continental high pressure systems
(Choi et al., in preparation) hr
1021
1019
UM(Anal)
H: 1021
L1: 980
L2: 970
UM(Fcst)
H: 1019
L1: 980
L2: 969
1030
KIM v2.1
(ORG:
Alpert et al., 1988)
H: 1030
L1: 961
L2: 955
1022
KIM v2.1
(updated:
Choi-Hong, 2015)
H: 1022
L1: 972
L2: 956
More GWD at low level→ reduce wind sfc wind speed → reduce mass convergence → weaker high

12. CPS V2.4|
Entrainment: Increasing the dependency of environment moisture
 increase entrainment rate in drier environment
CPS trigger considering environment moisture
 suppress triggering with the condition of dry low levels
Improvement of precipitation forecast a
SAS_MAY2016 a
SAS_MAY2016
ETS against AWS
(Korea)
Bias against CPC (Global)
Before
After

13. PROGC+SCV V2.5|
Modification of cloud formation for the supersaturated cloud in the prognostic cloud scheme
Change of vertical diffusion profile for temperature and specific humidity
Improvement of downward SW flux at surface
CERES
CTL
MOD
178.6 W/m2
157.7 W/m2
166.6 W/m2
Improvement of temperature at lower level against sonde
East Asia
Europe
East Asia
Europe

14. GWDC V2.5
Stationary convective GWD
in an uniform atmosphere
(Jeon et al. 2010/Chun-Baik 1998)
Spectral convective GWD
(Song-Chun 2005; Choi-Chun 2010)
Rayleigh friction or
uppermost diffusion
Spectral frontal GWD
(Charron-Manzini 2002; Richter et al. 2010)
Improvement of U wind against FNL analysis data and radiosonde
JJA 2013
CTL
MOD
Feb. 2014
CTL
MOD
Contour: RMSE
Shading: Bias
RMSE = 2.713
PC=0.977
RMSE = 2.613
PC=0.980

15. KIAPS Package for Observation
Processing (KPOP) and
Data Assimilation Systems
(3DVAR and LETKF)  4D EnVar

16. KIAPS Package for Observation Processing (KPOP)
2013
2014
2015
2016
Conven-tional
Data
Sonde
Surface
Aircraft
Satellite
AMSU-A
ATMS
MHS
IASI
CrIS
GPS-RO
AMV
ASCAT
SSMIS
Frameworks
ODB (Obs. Data base)
Comparison
with KMA-op.
Monitoring system
Primitive
Improved
Semi-operational
current

17. 3DVAR Song and Kwon 2015 (MWR)
3DVAR system built on KIM (cubed sphere grid using Real-observations)
For radiance assimilation, the Jacobian matrix gained from the KPOP system is used for operator (brightness temperature)
The observation data assimilated so far
Sonde, Surface, Aircraft, AMSU-A, IASI, GPO-RO, AMV, ATMS, CrIS, MHS,
CSR, ScatWind
Aircraft
data assimilation
RMSE
250hPa
Results of 3DVAR system
AMSU-A
data assimilation

18. Development of hybrid DA Hybrid-4DEnVar
4D extension of 3DVAR (seven time bin for an analysis window)
Error of the Day of error from LETKF
Add alpha control variable for Ensemble background error
Temperature and moisture improvement for all level
↑ RMSD of U analysis
against UM analysis

19. Verification/Analysis
KIM 2.4 (Jul 2016)

20. Verification/Analysis Application (Physics)
NE60, 240-hr
July 2013
AC of 500 hPa GPH (NH)
Bias of 24-h Acc. Precip. (3-d avg)
V2.3 : 0.832
V2.4 : 0.837
RMSE of 500 hPa GPH (NH)
ETS of 24-h Acc. Precip. (3-d avg)
37.040
37.083
V2.3 : 38.48
V2.4 : 37.92
V2.3
V2.4
[against FNL analysis]
[against TMPA observation (Global)]

21. Verification/Analysis Application (Physics)
Temperature bias (shade) and RMSE (contour) against sonde (+240 h forecast)
KIM2.3
KIM2.4
Jul. 2013
(Tropics)
Feb. 2014
(Asia)

22. Verification/Analysis KIM skill statistics
July 2013 case NH
500 hPa GPH AC NH
500 hPa GPH RMSE
UM: N512 GDPS
KIM: NE60 with GFS analysis

23. Verification results of semi-real run (2016. 7)

24. Verification for rain against CPC (July 2016)
Tropics
GLOB
KIM-COLD
KIM (~25km) with UM initial condition
+240h forecast at 00, 12 UTC
KIM-WARM
KIM (~25km) with 3DVAR (~50km)
+240h forecast at 00, 12 UTC
+6h forecasts at 06, 18 UTC

25. Monthly verification statics
KIM v2.1
v2.2
v2.3
v2.4 NH
500 hPa GPH AC NH
500 hPa GPH RMSE