Channel selection for IASI in clearsky conditions

1
Channel selection for IASI in clear-sky
conditions

• Florence Rabier and Nadia Fourrié
• Météo-France
• ITSC-XII February 2002

2
Rationale and objectives

• Advanced IR sounders
• large volume of data, prohibitive in operational
  NWP
• Selection of individual channels
• Which channel selection method leads to the best
  analysis accuracy?
• In practice, how can this method be applied
  robustly to a large set of atmospheric
  conditions?
• IASI Infrared Atmospheric Sounding
  Interferometer developed by CNES-EUMETSAT

3
Linear estimation theory

• Retrieval xa xb K(y-Hxb)
• Covariance matrix A-1B-1HTR-1H
• Gain matrix K A HTR-1
• Data Resolution Matrix DRMHK
• Model Resolution Matrix MRMKH
• Metric based Jacobian matrix HR-1/2HB1/2
• Degree of freedom for signal DFS Tr (I-AB-1)
• Shannon entropy reduction ER -1/2 log2 AB-1

4
Linear estimation theory

• Resolution matrices
• xa- xb K(y-Hxb) KH (x- xb) MRM (x- xb)
• ya- yb H (xa- xb)HK (y- yb) DRM (y- yb)
• Link the analysis and the signal from the data
• Diagnostics of retrieval accuracy
• Standard-deviations of analysis errors
• ?a(i)
• Vertical resolution (Purser and Huang)
• Resol(i) dz (i) /MRM(i,i)

5
Channel selection methods

• Methods based on the DRM (Menke, Prunet)
• Equation ya- yb DRM (y- yb)
• Select the most useful data in the analysis
• Method based on Jacobians (Goldberg, Aires)
• Characteristics of HR-1/2HB1/2
• For each parameter to be retrieved, select the
  most useful channel
• Iterative method (Rodgers)
• Measures of improvement ER or DFS (AB-1)
• Iteratively, pick up the most useful channel to
  improve on the current analysis. Update the
  analysis errors.

6
Methods based on the DRM (Menke)

• Data resolution matrix DRMHK
• From ya-ybDRM (y-yb), the diagonal elements of
  DRM indicate how much weight a datum has in its
  own analysis
• These diagonal elements measure the
   importance  of the various channels
• The method needs the computation of A

7
Methods based on the DRM (Prunet)

• SVD of H, with metrics B and R
• G R-1/2HB1/2 U?VT
• Truncation in ?2 such that eigenvalues of GTG
  B1/2HT R-1HB1/2 , equivalent to sb2/ so2
  represent 10 of contribution of the observations
  to the analysis
• G R-1/2HB1/2 gtUp?pVpT
• DRM VpVpT . Its diagonal elements are used as
  channel  importance 

8
Method based on the Jacobians (Goldberg, Aires)

• Is it based on the shape of the weighting
  functions
• Normalisation of H R-1/2HB1/2
• For each retrieved parameter, at each level in
  the vertical, one selects channels
• Among those peaking next to the level
• With the largest ratio
• Amplitude of the peak/Width of the weighting
  function

9
Iterative Method (Rodgers)

• This method is a step by step selection scheme.
  At each step, BiAi-1 is updated by using the
  most informative channel among those which have
  been previously selected.
• After normalisation of the Jacobian by R
• Ai-1Bi-1hTh
• Where B0B and h is a line of H
• The selection criterion is either DFS or ER
• DFS(h)iTr(I-ABi-1)hTBih/(1 hTBih )
• ER (h)i-1/2 log2det(ABi-1)1/2 log2(1 hTBih )

10
Experimental context

• 500 atmospheric situations
• Profiles (T,Q), various sites and dates
• IASI data simulated with RTIASI
• (Matricardi and Saunders)
• 8461 radiances (645 cm-1 2760 cm-1)
• B based on a 60-level ECMWF matrix
• O from CNES, F0.2K
• Removal of bands sensitive to trace gases
• (700-720, 1000-1080, 1267-1312, 2092-2355 cm-1)

11
Results on mid-lat profiles

• 24 atmospheric situations
• Profiles (T,Q), one site at various dates
• 4 channel selection methods tested
• For each profile, optimal selection performed
• Results averaged over all profiles

Retrieval i Profile i Selection i
Selection i
Profile i
12
Results on mid-lat profiles
13
First Channels for T Iterative method
14
First Channels for T Iterative method
15
Non-optimal set of channelsIterative method

• For a set of profiles, optimal selection
  performed
•  Constant  selection obtained by averaging the
  ranks of the channels Cst selection Ave
  (Selection i)
• Non-optimal retrievals
• Would allow to pre-compute a constant selection
  off-line, and to apply it to new profiles in real
  time

Selection i
Profile i
Retrieval j Profile j Cst Selection
Profile j
16
 Constant  selection (300 channels) Iterative
method (492 profiles)
17
Results for analysis errorsIterative and
Jacobian methods (300 channels, 492 profiles)
18
Results for analysis vert resolutionIterative
and Jacobian methods (300 channels, 492 profiles)
19
Influence of number of channels Results for
analysis errorsIterative method (24 profiles)
20
Influence of number of channels Results for
analysis vert resolutionIterative method (24
profiles)
21
Conclusions

• Iterative method
• Among 4 channel selection methods tested, the
  iterative method is giving the best results
• Main strength
• Update the error covariance matrix each time a
  channel is selected
• Constant selection gives promising results
• Pre-selection based on a set of profiles, then
  applied to all profiles
• Robustness selection performed for 62 profiles
  out of 492 gave 84 of channels in common with
  the one computed on all 492 profiles

22
Perspectives

• Method can be applied to other sounders
• Thépaut and Fourrié
• Study to be extended
• Inclusion of different scan angles, surface
  types, cloud conditions
• Possible operational channel selection
• Pre-selection based on monitoring statistics
• Use several sets of channels for various
  configurations of scan angles, surface types,
  cloud conditions and also air-mass