IASI bias correction at KIAPS observation processing system
Hyoung-Wook Chuna, Yoonjae Kimb, Peter Westonc, and Sujin Haa
aKorea
Institute of Atmospheric Prediction System, Korea, (hw.chun@kiaps.org)
bKorea Meteorological Administration, Korea, cMet Office, Exeter, UK.
Introduction
1
Scan biases
4
(a) KIAPS (120 scan positions)
IASI
 The processing system of Infrared Atmospheric Sounding
Interferometer (IASI) radiances developed at Korea Institute of
Atmospheric Prediction Systems (KIAPS) is introduced and
its preliminary results are described here.
IASI scan position
Flight direction
Brightness temperature (TB) observed for one month in
November, 2012 is extracted from IASI Level1d BUFR data
using ECMWF BUFR decoder. The background TB is
simulated by RTTOV version 10.2 using UM 6-hour forecast
data.
Flight
direction
IASI 1d data
(MetOp-A, B )
CrIS
developing
Preliminary
version
(C) Met Office (30 scan positions)
Fig. 3. IASI scan position
- IASI: 120 scan position = (4 x 30)
- AMSU-A: 30 scan position
In level 1d data, 1 pixel in 4 is chosen by using AVHRR data.
The selected pixel is the most warmest and homogeneous
field of view.
Air-mass coefficients
5
Done
Fig. 4. IASI scan bias of KIAPS (a, b) and Met Office (c). In KIAPS, 120 scan
position is default. To compare with Met Office, the scan biases are calculated
for 30 scan position. The scan biases are totally different between KIAPS and
Met Office. The variations along with scan positions are larger in KIAPS for
temperature channels, on the other hands the variations are larger in Met
Office for water vapor channels.
KIAPS
Blacklist
- Bad channels
(Channel selection)
- Bad regions
Blacklisting
8 5 12 9
IASI
Bkg.
u,v,t,q,p,sfc
Sanity Check
Obs. TBs
(O)
Under
4 1
Scan direction
IASI processing in KIAPS
Future
plan
7 6 11 10
AMSU
 And also we remove bias of observed TB for each IASI
channel and scan position. We assumed that the bias of
background TB is negligible compared to that of observed TB.
Lat, Lon, Time
3 2
(b) KIAPS (30 scan positions)
AMSU scan position
 We adopt NWP SAF cloud detecting algorithm (McNally and
Watts, 2003) to remove cloud contamination in observed TB
for each IASI channel.
2
AMSU scan position
AMSU-A
Met Office
Long-term
check
RTM
(RTTOV)
Offline code
Static BC
Bias Corr.
Obs. TBs
(C)
Criteria
for outlier
removal
Bkg. TBs
(B)
Cloud screening
Obs.
Error
Coeff.
for
Static BC
Monitoring
system
Ana.
FSO
…
Statistical analysis
Outlier removal
(C-B check)
Monthly or bi-weekly data:
O, B, C, QCFlags, Predictors
Qualified and
corrected
Obs. TBs (C)
3
Factor
for
monitoring
Thinning
Obs. data
for DA
system
DA
system
IASI bias correction
Obs. TBs
(O)
Bkg. TBs
(B)
Bias scan = O − B
QCFlag
CldFlag
OutFlag
Qulalified
O-B
Scan correction
scan
Mean (O-B) each
scan position
−O −B
Fig. 5. The coefficients for air-mass correction of KIAPS and Met Office.
Slope (Thick 850-300)
• Similar spectral variation of slopes at high-peaking temperature channels
• The slopes of Met Office are larger at low-peaking temperature channels
Slope (Thick 200-50)
• Similar spectral variation of slope for all temperature channels
Constant
• The spectral variations are totally different in water vapor channels
scan
Bias scan
6
O-B - Biasscan
Scan corrected
O-B
Results of bias correction
After BC @ ch401
Before BC @ ch401
After BC @ ch167
Before BC @ ch167
Predictors
Multiple
regression
Slopes and constant
Predictor1:
Thickness(850-300 hPa) – Thickness(850-300 hPa)
Multiple regression between
scan corrected O-B and predictors
Biasairmass = S1 P1 + S 2 P2 + Const
Predictor2:
Thickness(200-50 hPa) – Thickness(200-50 hPa)
-5
Fig. 1. The predictors for air-mass correction.
Fig. 2. IASI channels passed to the
OPS and VAR in UM (Weston, 2011,
Fig.2)
Original IASI: total 8461 channels
KIAPS: 314 channels are used
(level1d data from KMA)
- 300 channels (Collarad, 2007)
- 14 channels (CNES monitoring)
* Channel selection
is our future work
UM
- OPS: 183 channels
- VAR: 138 channels
-3
-1
1
O – B (K)
3
5
-5
-3
-1
1
C – B (K)
3
5
-5
Fig. 6. O-B distribution before and after bias correction at ch 167 and ch 401.
- channel 167 (686.5 cm-1, peak @ ~ 70 hPa, 18.8 km, Index: 49)
- channel 401 (745.0 cm-1, peak @ ~640 hPa, 4.0 km, Index: 113)
- The irregular biases in each latitude bins are reduced after bias correction
7
Summary
 The framework for IASI processing in KIAPS is established.
-3
-1
1
O – B (K)
3
5
-5
-3
-1
1
C – B (K)
3
KIAPS
Met Office
Background
UM ver 7.9
physics PS28,
SURF 18.5
UM ver 8.2
Data processing
KOPS
OPS
RTTOV version
UM: ver. 7
KOPS: ver. 10.2
OPS: ver. 9.3
Day/Night
Day & Night
Day & Night
Region
All surface type
(Land, Ocean, SeaIce)
All surface type
(Land, Ocean, SeaIce)
Time window
6 hour (+3h, -3h)
6 hour (+3h, -3h)
Thinning
Not thinned data
Not thinned data
5
The bias correction coefficients are calculated for each IASI channel and scan position during Nov. 2012. The bias correction
coefficients of KIAPS are compared with Met Office values. The air-mass correction slopes are similar in two organization, but scan
biases and constants are some different. The differences of bias coefficients between KIAPS and Met Office may be due to the
RTTOV version and UM version used in each organization.
The calculated coefficients are apply to the same time period and then the irregular biases in each latitude bins are reduced .
International TOVS Study Conferences 19| Jeju Island, South Korea, 26 March – 1 April 2014