IASI Channel Selection in Cloudy Conditions
Pauline Martinet, Lydie Lavanant, Nadia Fourrié, Florence Rabier, François Faijan
Contact: pauline.martinet@meteo.fr
2) Method
Introduction
The atmospheric parameters for which information is expected are:
The InfraRed Atmospheric Sounding Interferometer IASI onboard Metop satellite provides temperature and humidity profiles at a high spectral resolution. With 8461 channels, it covers a range
from 645cm-1 to 2760cm-1.
Temperature (T)
Humidity (q)
Liquid Water Content (ql)
Ice Water Content ( qi)
As it is not feasible to assimilate all the 8461 channels in real time, the European Centre for
Medium Range Weather Forecast (ECMWF) monitors a subset of 366 channels operationally. This
selection called CM2009 (Collard and McNally 2009 1) hereafter was carried out through clear
atmospheric profiles.
The cloud profiles were extracted from the convective scale model AROME W MED over the
Mediterranean Sea in the context of the HyMeX campaign.
The cloud profiles were selected within IASI footprints homogeneously covered by clouds by a
screening procedure based on the AVHRR imager
(Martinet et al 20113). 20 profiles were chosen:
All NWP centres intend to take advantage of cloud-affected radiances to improve the analysis in
meteorologically sensitive areas. With the view to add the cloud variables (liquid water content,
ice water content) in the state vector of the assimilation system, this poster describes a
methodology to evaluate the quality of the CM2009 selection on cloudy profiles and the
relevance of a new channel selection.
5 high opaque clouds
5 semi-transparent ice clouds
5 liquid clouds
5 mixed phase clouds
The methodology is based on the informationcontent theory of Rodgers (Rodgers 2000 2)
using Degrees of Freedom for Signal (DFS) as
the figure of merit of the selection.
3) Results
(ql,qi)
300
SELC4 is chosen:
Higher DFS values
Information on T,q for clear profiles
Computation time
290
280
270
260
250
240
300
270
270
260
260
250
250
240
240
230
230
220
500
1000
1500
EXP
300
290
2000
(T,q,ql,qi)
Temperature
CM2009
280
270
4.11
(68%)
4.24
(71%)
4.41
(73%)
2000
6.00
SELC2
260
250
SELC4
240
230
220
500
1000
2500
1500
All channels
3000
2500
BT
~66% of channels. shared between a
constant selection and each cloud
dedicated channel selection
220
500
1000
1500
Liquid Water
Content
Humidity
2.79
(69%)
2.93
(73%)
3.08
(76%)
3000
4.04
2000
Ice Water
Content
1.45
(76%)
1.59
(84%)
1.54
(81%)
1.90
0.17
(59%)
0.20
(66%)
0.19
(63%)
0.30
2500
Loss of information negligible when
applying a cloud-type dedicated channel
to a different one.
3000
Total
8.52
(70%)
8.95
(73%)
9.22
(75%)
12.24
Humidity
σa liq 8461
Pressure
500
σa liq 566
σ liq 8461
200
200
σ
b
300
300
400
400
500
a
σ
σb
500
700
700
700
800
800
800
800
900
900
900
900
1000
0
Temperature
1
2
σ ice 366
100
200
σa ice 566
200
300
σ
a
σ ice 8461
a
b
Pressure
400
500
6
7
1000
0
8
−4
x 10
Humidity
0
100
3
4
5
Standard Deviation
0.2
σ ice 8461
a
−4
x 10
σ ice 8461
a
σ
b
300
300
400
400
400
500
700
700
700
700
800
800
800
800
900
900
900
900
2
1000
0
1
2
3
4
5
Standard Deviation
6
7
8
−4
x 10
1000
0
0.2
0.4
0.6
0.8
Standard Deviation
1
1.2
−4
x 10
3
4
Standard Deviation
5
6
7
−5
x 10
IWC
1000
1200
800
1000
1200
300
2000
2200
1600
1800
2000
2200
1400
1600
Mixed phase
1800
2000
2200
1800
2000
2200
1400
Opaque
250
300
250
1000
0
Boudala Aggregates
1400
1600
280
260
260
240
240
220
220
800 1000 1200 1400 1600 1800 2000 2200
σ ice 8461
a
b
300
2
3
4
Standard Deviation
5
6
7
−5
x 10
Figure 2: Analysis-Error Standard Deviations for each cloud type: liquid clouds in black, ice clouds in blue, mixed phase clouds in red. Solide lines represent the performance on the new selection of 566 channels, dotted lines the best performance using all the channels and dash-dot lines the performance of the CM2009 selection of 366 channels.
800 1000 1200 1400 1600 1800 2000 2200
Wyser Aggregates
Wyser Hexagonal
300
280
280
260
260
240
240
220
220
800 1000 1200 1400 1600 1800 2000 2200
1
Boudala Hexagonal
300
280
σa ice 366
500
600
1.5
2
σ
600
1
Standard Deviation
1
200
600
0.5
300
σa ice 566
600
1000
0
800
1800
Crystal Shape
100
300
500
b
0
σa ice 566
200
σb
1000
0
1.2
σa ice 366
100
σa ice 566
1
LWC
0
σa ice 366
0.4
0.6
0.8
Standard Deviation
Pressure
2
1200
1400
1600
Semi−transparent
800 1000 1200 1400 1600 1800 2000 2200
Figure 4: Location of the 200 selected channels when considering different ice cloud parameterizations and
different ice crystal shapes.
Future prospects
References:
1. Collard A.D and McNally A.P, 2009: The assimilation of Infrared Atmospheric Sounding Interferometer radiances at ECMWF.
Quart.J.Roy.Meteor.Soc, 135,1044-1058.
2. Rodgers CD 2000: Inverse methods for atmospheric soundings: theory and practice. World Scientific: Singapore.
3. Martinet P, Fourrié N, Guidard V, Rabier F, Montmerle T, Brunel P: Towards the use of microphysical variables for the assimilation of cloud-affected
infrared radiances. Quart.J.Roy.Meteo.Soc, submitted.
Use of the new channel selection in 1D-Var experiments with simulated
observations.
Comparison with a new channel selection on different air mass types
(midlatitude, tropical, polar).
Sensitivity of the retrieval to different cloudy B matrices.
Evolution of the increments in a one-dimensional version of AROME.
parameterization
700
1.5
a
500
600
1
Standard Deviation
1000
2200
250
400
600
0.5
σ liq 8461
300
600
0
Pressure
100
600
1000
0
Ice
σ liq 566
σa liq 366
a
a
Pressure
Pressure
Liq
400
σa liq 366
100
σ liq 8461
200
σb
300
σa liq 566
IWC
0
BT
a
200
σa liq 366
100
σ liq 566
LWC
0
Pressure
100
800
2000
In RTTOV, possibility to choose the scheme to use to parameterize the
effective diameter and the ice shape (hexagonal ice crystals or ice aggregates).
Figure 4 compares the new channel selection for two parameterizations and
two ice shapes.
Robustness of the selection with respect to the cloud optical
parameterization.
Selection more sensitive to the ice crystal shape.
BT
a
Ice Water content
Pressure
σ liq 366
Liquid Water Content
Humidity
0
1200
1800
6) Dependence to the ice cloud parameterization
Small loss of information compared to the 8461 channels
Useful improvement of the analysis over the background for liquid clouds
Little information extracted from mixed phase clouds (not shown)
Temperature
1000
1600
Figure 3: Location of the selected channels for each cloud-type
dedicated channel selection (SELC4 method is used).
5) Reduction of Background Errors
0
800
1400
Liquid
250
300
Robustness of the selection with
respect to the cloud type
Table 1: DFS values for each variable and each channel selection.
Temperature
1200
BT
280
1000
BT
280
800
300
BT
290
BT
290
All profiles
250
(T,q,ql,qi)
SELC4
BT
300
New Analysis
Error Covariance
Matrix A
4) Dependence to the cloud type
220 20 profiles. The CM2009 selection of ECMWF is repreFigure 1: Location of the selected channels averaged over all the
500
1000
1500
2000
2500
3000
sented with black points, the 200 new channels in red points.
(ql,qi)
SELC2
Update of the
background
error covariance
matrix B
300
230
BT
200 channels added to the CM2009 selection
SELC2: ql, qi in the state vector
SELC4: T, q, ql, qi in the state vector
60 channels shared by the two selections
Choose the channel j
that most improves
the DFS
DFS=tr(I-AB-1)