The Use of COMS Clear Sky Radiance in NWP Model; CSR Product in Global
Forecast Model and Clear Pixel Radiance for Local Forecast Model
Jung-rim Lee, Yoonjae Kim, Eunhee Lee, Dong-jun Kim and In-cheol Shin
jrleeng@korea.kr
Korea Meteorological Administration
Impact of New Bias Coefficients
WV CSR in Global Forecast Model



Operationally assimilated since June 27th, 2013.
Use fixed bias coefficients
WV CSR product
- Central wavelength : 6.75 ㎛
Mean
brightness
temperature
of
surrounding
(spatial resolution is about 30 km)
- Only 100 % clear data is used in the 4DVAR
- July, 2013
7x7
clear
- December, 2013
(hPa)
pixels
Monitoring the Bias

Mean RMSE Difference (Exp-Ctr), Negative: improvement

(m)
(m/s)
 In the northern hemisphere, RMS error reduced in December by applying new bias
coefficients, but shows neutral impact in July.
Timeseries of C-B
upgrade model
Percentage Difference of Geopotential Height RMSE, Positive: improvement

- Own analysis
- Sonde
- ECMWF
GPH/Verfi. against ECMWF
GPH/Verfi. against ECMWF
2
1.5
1
0.5
Global 0
NH -0.5
-1
TR
-1.5
SH
-2
100hPa +5D
100hPa +4D
100hPa +3D
100hPa +2D
100hPa +1D
200hPa +5D
200hPa +4D
200hPa +3D
200hPa +2D
200hPa +1D
300hPa +5D
300hPa +4D
300hPa +3D
300hPa +2D
300hPa +1D
500hPa +5D
500hPa +4D
500hPa +3D
500hPa +2D
500hPa +1D
850hPa +5D
850hPa +4D
850hPa +3D
850hPa +2D
SH
850hPa +1D
100hPa +5D
100hPa +4D
100hPa +3D
100hPa +2D
100hPa +1D
200hPa +5D
200hPa +4D
200hPa +3D
200hPa +2D
200hPa +1D
300hPa +5D
300hPa +4D
300hPa +3D
300hPa +2D
300hPa +1D
500hPa +5D
500hPa +4D
500hPa +3D
500hPa +2D
500hPa +1D
850hPa +5D
850hPa +4D
850hPa +3D
850hPa +2D
850hPa +1D
1000hPa +5D
1000hPa +4D
1000hPa +3D
Experiment Using New Bias Coefficients
1000hPa +2D
1000hPa +1D
-2
TR
1000hPa +5D
 C-B is getting larger. Needs for recalculating new bias.
NH
1000hPa +4D
-1
Global
1000hPa +3D
0
1000hPa +2D
1
1000hPa +1D
2
 Overall, applying new bias coefficients made positive impact on forecasting
performance except some part of tropics. At least in the northern hemisphere
including Asia region, obvious improvements are shown.
 Couldn’t find any specific trend with the verification results against sonde.



Periods: ① July (1~30, 30 days), ② December (9~25, 17 days)
Compared with operational model
Verification against own analysis, sonde, ECMWF

Changes in mean analysis (Exp-Ctr), red circles indicate COMS observation domain
Mean RMSE Difference of Geopotential Height, blue: improvement

- Geopotential Height (m), red: higher
- Relative Humidity (%), red: moisture
 In the area of COMS observation domain shows positive impact (East Asia and
Australia). The positive impacts enhanced over the land.
 After changing the bias coefficients, GPH become higher as reaching to low level, and,
RH changed to be more humid in middle and low levels.
Clear Pixel Radiance in Local Forecast Model




Summary
- Temperature (K)
Currently working on the evaluation
Calculated bias coefficients using Local Forecast Model
Bias
exp-op
6
6
6
Local Forecast Model
- Spatially 1.5 km, 70 vertical levels (~39km), 3DVAR
5
5
5
4
4
4
3
3
3
WV Clear Pixel Radiance
- Central wavelength : 6.75 ㎛
- Spatial resolution : about 4 km
- 100 % clear, standard deviation surrounding 5x5 pixels < 0.5 K
2
2
2
1
1
1
-1.5
-1
-0.5
0
-0.1
-0.05
0
0.05
 C-B of COMS CSR in KMA’s global forecast model
is getting slightly larger since its operation.
% difference
0.1
-15
-10
-5
 New bias coefficients are calculated, and tested on
the model for 1 month.
0
 The model analysis got more moisture throughout
all vertical levels after changing bias coefficients.
5
- Relative humidity (%)
 Experiment Method
- Period : July, 2013
- Compared with operation (without COMS CSR)
 Verification against Sonde
-3
-2
6
6
6
5
5
5
4
4
4
3
3
3
2
2
2
1
1
1
-1
0
1
2
-0.2
-0.1
0
0.1
0.2
-10
-5
 Forecasting performance seems improved very
much for all forecast time especially in the long term
forecast.
0
 In the Northern Hemisphere including Asia, and
Australia which is in the COMS observation domain,
the forecast performance improved a lot.
5
- Geopotential height (m)
 Overall, the impact of CSR is quite neutral except in the 6-hour
forecast of temperature and relative humidity.
 In the verification of geopotential height, it made the bias lower,
and led positive impact in low levels.
 In the future, optimization works will be carried out.
-30
-20
-10
6
6
6
5
5
5
4
4
4
3
3
3
18h
2
2
2
12h
1
1
1
0
10
-5
0
5
10
15
-4
-2
36h
30h
24h
6h
0
2
ITSC-19 in Jeju Island, South Korea
 The use of pixel radiance in the local forecast model
is not yet optimized, and more works will be done in
KMA.