Transcript Spectral stray light algorithm

Stray light
Marcel Dobber (KNMI)
OMI ST meeting 11
20-22 June 2006
Stray light algorithms in the GDPS

LSA / USA stray light algorithm.

Spectral stray light algorithm.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 2
LSA / USA stray light algorithm (current)

Currently (GDPS 0.9.15, OPF 31):
 Off for UV1, on for UV2 and VIS.
 Average all pertinent columns in LSA, same in USA.
 Average LSA and USA.
 Apply to all rows and columns.
 In case of asymmetric illumination profile (e.g. north and south
pole):
 Undercorrection for illuminated parts of CCD.
 Overcorrection for not illuminated parts of CCD.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 3
LSA / USA stray light algorithm (future)

Future (GDPS 0.9.16, OPF ?):
 On / Off startus for UV1, UV2 and VIS is TBD.
 Average all pertinent columns in LSA, same in USA.
 Multiply with predefined shapes as f(column).
 Separate shapes for earth, sun, WLS, all LSA and USA.
 Interpolate linearly between LSA and USA.
 Apply to all columns and rows.
 In case of asymmetric illumination profile (e.g. north and south
pole) the correction should work much better than the previous
one.
 Partitioning with spectral stray light algorithm remains TBD.
 More details by Robert Voors in calibration working group.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 4
Spectral stray light algorithm (current)

Currently (GDPS 0.9.15, OPF 31):
 Windows in OPF define source regions.
 In target subchannel stray light described by polynomials of
relatively low order (e.g. max 2).
 Average over all rows in source region.
 Apply equally to all rows in target region.
 Can not deal properly with inhomogeneous illumination profiles
on CCD.
 Currently some polynomials in OPF diverge as a result of
extrapolation in ground measurement results (filters).
 e.g. towards overlap UV2-VIS.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 5
Spectral stray light algorithm (future)

Future (GDPS 0.9.16, OPF ?):
 Conceptual definition source windows and polynomials in target
subchannel will remain unchanged.
 Polynomials are not row-dependent, because we could never
determine such OPF keydata.
 New (row) window definitions in source area to determine how
many rows shall be averaged in source area.
 New other (row) window definitions in target area to determine
how many rows shall be averaged / applied in target area.
 OPF keydata will need to be determined:
 based on existing keydata, largely from on-ground
measurements with filters.
 Tuning very likely required.
 Partitioning with LSA / USA stray light algorithm remains TBD.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 6
GDPS status

v0.9.16 (7 June 2006, TBC):
 New algorithm for USA / LSA stray light correction replaces old
one.
 See TN-OMIE-KNMI-792, issue 2, 30 March 2006.
 See TN-OMIE-KNMI-814, issue 1, 30 May 2006.
 New algorithm for spectral stray light correction replaces old one.
 See TN-OMIE-KNMI-802, issue 1, 27 March 2006.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 7
Plan for OPF for GDPS 0.9.16



Optimise LSA / USA and spectral stray light parameters.
Adapt radiometric calibration parameters.
GDPS 0.9.16 + OPF in ps9999 (8 weeks Sept-Oct. 2006):


L1 and L2 to provide feedback.
After this GDPS 0.9.16 + OPF are ready for reprocessing.
OMI ST Meeting 11, 20-22 June 2006
Marcel Dobber, KNMI
Slide 8