The Assimilation of
IASI ozone channels
Wei Han1 and Tony McNally2
1. Chinese Academy of Meteorological Sciences, CMA
2. ECMWF
Thanks to :
Rossana, Peter, Elias, Marco, Alan, …
CAMS/CMA
Outline
z Background of the study
 It’s a time to start the work
z Model bias or Observation bias?
 dTb~ H(dxa) Æ <O-B> are mainly due to model bias
z Experiments
 Baseline
 Baseline+SBUV+OMI
 Baseline+IASI ozone channels
z Verification against independent observations
 MLS and sonde
z Summary and Discussions
CAMS/CMA
Background
z Ozone analysis
 Importance
Î Assimilation of radiances (IR channels)
Î Stratosphere(T,O3,wind), medium-range weather forecast
Î Reanalysis and Climate study
 Difficulty: Theare is NO unbiased obs. as backbone
Î Observation:Realtime:(SBUV,OMI),Delay: (Sonde,MLS)
Î Model: simple photochemistry model(bias)
z IASI observation information
 321 oznone channels at 9.6um
 Global and All day coverage, esp. for polar night
 Improved RT model
CAMS/CMA
 Cloud detection (QC for IR)
The key issues
z Model bias
 simple photochemistry model(bias)
z Observation bias
 Cal/Val
 Non perfect Cloud Detection
 RT model bias
z No anchoring
 NO realtime unbiased Ozone Obs. AS BACKBONE
z Anchoring VARBC against model biases
 Prior constraint on the Anchoring channel
 VARBC to remove the interchannel bias
CAMS/CMA
Ozone data coverage in 12h
N.H. Summer
SBUV(Black)
and OMI(grey)
IASI over Sea
IASI over Sea
and sea ice and
LAND
CAMS/CMA
Use of the IR spectrum with IASI
321 OZONE CHANNELS
153 Long-wave
temperature
CAMS/CMA
Water vapour
Short-wave temperature
Jacobians and Active Channels
Color Dased Line:
321 Channels at 9.6um O3 band
Black Thick Line:
16 Channels selected
CAMS/CMA
BIAS analysis: OBS or BACKGROUND?
z Operational against sonde
 3 months
z IASI O3 channel FG departures
 In Spectral space
 Geophysical and Seasonal Variation
 Compare with other Channels
z Offline RTTOV
 H: dO3 Æ dTb
 1DVAR: dTbÆdO3
CAMS/CMA
It’s mainly the
model bias
IASI CH 1578,1585,1786 Mean FG Depart.(no bias cor.)
Active Data(f5li:2009020112-2009022812)
IASI CH 1578,1585,1786 Mean FG Depart.(no bias cor.)
Active Data(f5li:2008080712-2008090712)
200808
135
-135
2
-90
-45
200902
45
90
0
135
2
Ch1578
0
0
30
30
60
60
60
90
30
45
0
0
60
-45
30
-90
0
-135
1
-60
-90
-45
0
45
90
-60
-60
-30
Strong
O3
-60
-135
-30
-30
-30
1
-135
135
-90
-45
0
45
90
135
0
0
90
-135
135
135
Ch1585
0
-30
-60
Stronger
O3
-60
60
90
-30
30
45
0
0
0
30
30
-30
-45
60
60
-60
-90
60
45
30
0
0
-45
-30
-90
-60
-135
-1
Exp: f5li
-2
-2
-135
-90
-45
0
45
90
-135
135
-90
-45
0
45
90
The Assimilation system is FROZEN in this peroid
90
-135
135
30
0
-30
45
90
135
-60
Window
-60
-60
-60
0
135
-30
-30
-45
90
0
0
-90
45
30
30
Channel
-135
0
Ch1786
-4
CAMS/CMA
-45
60
60
-3
-90
60
45
30
0
0
-45
135
-30
-90
60
-135
-1
-135
-90
-45
0
45
90
135
-3
-4
2008JA
2009JF
N.H
TR
S.H
CAMS/CMA
Marco(2009) Monitoring Results cycle [CY33R1]
z N.H
9.6 um band of O3
z TR.
z S.H
CAMS/CMA
2008JF,0-30N, Sonde and Analysis
analysis
sonde
CAMS/CMA
analysis-sonde
Bias(Xa-Sonde)ÆBias(Tb):
H(O-X)~dTb=Yo-H(xb)
Biases at IASI O3 band
are consistent with biases of analysis
dTb(dO3)
Bias(N.H)
CAMS/CMA
d=Yo-Hx=(Yt+b)-H(Xt+Eb)
<d>=<b>-H(Eb)
<b>=<d>+ H(Eb)
MLSÆ Model Bias
and Obs. Bias
<O-B>,CH1585
20090615~20090630
1671
−135
f9h2_d
−f9h2_Eb
f9h2_b
0
45
90
135
60
30
0
1643
30
2
60
1652
−45
0
1658
−90
1639
−60
Model Bias
−60
−135
1585
−90
−45
0
45
90
<Xa-MLS>,30-50hPa
1579
Estimated Obs.
Bias
45
90
135
1.5
1.0
30
0.5
0
1521
0.0
1509
−30
−30
1513
−0.5
−60
−1.0
−2
CAMS/CMA
−1
0
dTb[Unit:K]
1
Southern Hemisphere
2
−135
−90
−45
0
45
90
135
−1.5
mmr(mg/Kg)
0
60
−45
60
O-B
1536
−90
30
1574
20090615~20090630
−135
0
1578
1479
0
−3
−4
−6
135
−60
Channel
−30
−30
−2
1626
1587
0
OSE Experiments
z BASELINE: f9h2—No Ozone Observations
 2 months :20090501-20090702
z EXP1: f9w4---Baseline+SBUV+OMI
 SBUV: NOAA-17 and NOAA-18,6 Layers
 OMI: Total Column
z EXP2:f9h1----Baseline+IASI Ozone Channels
 Activate 16 O3 channels at 9.6um band
 Zero bias correction for ONE channel(CH1585) to ANCHOR
 VarBC applied to other 15 channels
z EXP3:f9hc---Use IASI Ozone Channels over land
z VERIFY against independent data
 Sonde,MLS
CAMS/CMA
 2 weeks20090615-20090630
Experiments period and verification period
VERIFICATION
Against independent
data
CAMS/CMA
Impact on analysis
IASI -- BASELINE
z Largest Impact
Jacobian
N.H.
Background Error
Averaged in N.H
B convolved
With Jacobian
Impact on analysis
CAMS/CMA
Mean analysis difference
Winter
IASI-Baseline
UV-Baseline
CAMS/CMA
Summer
IASI O-B
CH:1579
CH:1585
CH:1786
CAMS/CMA
BASELINE
BASELINE+IASI
Verify against MLS(2 Weeks,20090615-20090630)
BIAS: <AN-MLS>
BASELINE
No O3 OBS
BASELINE
+SBUV+OMI
BASELINE
+IASI 16 O3
Channels
CAMS/CMA
std: <AN-MLS>
Bias and std: verify against MLS
z Black: BASELINE
z RED: BASELINE+IASI
S.H
CAMS/CMA
TR.
N.H
Zonal Mean Bias
Winter
BASELINE
IASI
UV
CAMS/CMA
Summer
Std. difference for summer period
IASI
Std. of Baseline
2.0
1.5
10
1.0
100
0.5
1000
90
60
30
0
Latitude
−30
−60
−90
mmr(mg/Kg)
Pressure, hPa
1
IASI including
Sea ice and Land
0.0
The standard deviation of baseline differences has been subtracted such that
negative value indicate the degree of improvement relative to the baseline.
CAMS/CMA
Verify against MLS
BASELINE
No O3 OBS
BASELINE
+SBUV+OMI
BASELINE
+IASI 16 O3
Channels
BASELINE
+IASI 16 O3
Channels+Land
CAMS/CMA
31.6hPa
100hPa
Positive bias in the
Southern polar
vortex
Polar Night
d=<Yo-H(xb)>
NegativeÆPositive
Reduce Ozone
Positive bias
becomes smaller
1.5
1.5
0
0
0
2
1.0
−45
1.0
45
90
0.0
31.6hPa
0.5
mmr(mg/Kg)
−90
−90
−90
90
−0.5
−135
135
−1.0
45
0
0.5
−90
−45
0.0
−90
−90
90
CH1585
−0.5
−135
135
−1.0
−2
−4
−135
135
−6
180
−1.5
BASELINE
June 15
TCO
June 19
June 23
CAMS/CMA
June 27
June 17
180
IASI O3 Over antarctic
−1.5
K
45
mmr(mg/Kg)
−45
180
O3 -- BASELINE
Verify against sonde data in June 2009
CAMS/CMA
Improve the fit to other IASI channel: CH290
CAMS/CMA
Summary and Discussions
z BIAS analysis: OBS or BACKGROUND?
 the Bias is maily from BACKGROUND: 75% VS 25%
Î Activate the IR O3 channel in assimilation
Î Anchoring VarBC against model bias
z Experiment Results
 VERIFY against independent obs.
Î SONDE: only about 10 stations, but better at UTLS
Î MLS: better in S.H and N.H. There is problem in Tropics
 Impact on other sounding channels
Î IASI 290: improved the fit
 Polar night
z Future work
CAMS/CMA
 Bias Correction, Anchoring, UV+IR, Wind Coupling
THANKS FOR YOUR ATTENTION
THANKS FOR
EUMETSAT NWP SAF VS project
CAMS/CMA
Backup slides
CAMS/CMA
CAMS/CMA
Analysis Bias: verify against MLS
BASELINE: No O3 OBS
31.6hPa
100hPa
CAMS/CMA
BASELINE+IASI O3 Channels
How to anchor VarBC : next step
CH:1579
CH:1585
Anchoring:no
bias correction
CH:1786
CAMS/CMA
CAMS/CMA
Fiona Hilton,ITSC16
Interactive O3-Wind
z Interactive ozone-wind
coupling
ur
∂δ O3 ur b
b
+ V ∇δ O3 + δ V ∇O3 = 0
∂t
z WHY turn it off?(Elias Holm)
 Bias of the model
 Thin layers observation
CAMS/CMA
TL model: