Large scale validation of SCIAMACHY nadir reflectance

1
Large scale validation of SCIAMACHY nadir
reflectance

• Gijs van Soest, L. Gijs Tilstra, Juan Acarreta,
  Piet Stammes

2
Outline

• Reflectance validation
• Known issues
• Method and data
• Large data set validation
• New key data an improvement?

3
Reflectance and irradiance validation

• Errors in reflectance cause errors in geophysical
  parameters retrieved from it
• Spot and characterize possible problems in
  SCIAMACHY level 1 data
• Instrument, calibration data, algorithms
• Approach compare measured and calibrated UV
  spectra with radiative transfer calculations
• Single states careful, detailed characterization
• Large data sets spot drifts, possible
  geographical seasonal trends

4
UV reflectance spectrum
5
Known issues of UV reflectance

• Radiometric calibration reflectance is 20 too
  small
• Also in other channels
• Showstopper for non-DOAS retrievals (O3 profile)
• Spectral feature associated with polarization
• Near 350 nm BrO and CHOH retrieval affected
• Improved in recent (V5.0x) data
• Spectral stray light, memory effect

6
Large scale validation method data

• Comparison of SCIA nadir reflectance spectra with
  RTM LIDORT
• Data from Sep Dec 2003, Apr Aug 2004,
  decontaminations excluded
• 4 substates per state East, East-C, West-C,
  West
• Wavelength range 250 400 nm or 270 340 nm
• NRT SCIA L1B data (SCI_NL__1PN)
• Processor version 4.0x in 2003, 5.0x in 2004
• L1C using SciaL1C tool, all calibrations on

7
Large scale validation method data

• Inputs for RTM
• Cloud mask based on FRESCO
• GOME ground albedo data base
• O3 profiles
• IFE limb retrieval (FTP server)
• TOMS v8 climatology using TOSOMI O3 columns
• Temperature and geopotential height from ECMWF
• 1100 states between 09/2003 and 12/2003
• 9400 states between 04/2004 and 08/2004

8
Histograms of SCIA/RTM-1 (2004)
east
Ozone TOMS v8 climatology
9
Map of ltSCIA/RTM-1gt 320-340 nm
370-390
10
Map of ltSCIA/RTM-1gt 270-290 nm
11
w/ IFE limb O3 profiles 270-290 nm
12
Difference map 2003 270-290 nm
CHANNEL 1 DATA FROM ESRIN NRT ORBITS
(SCI_NL__1PNPDE) PRIOR TO MARCH 2004 SHOULD NOT
BE USED!
13
Effect of decontamination 270-290 nm
decontamination orbits 12031-12208
Signal for ? lt 300 nm is so small that reliable
calibration is a real problem
14
New radiometric key data set

• Based on measurements from the PI period, IFE
  have calculated new radiometric key data
• Two sets Spectralon and NASA sphere
• Irradiance now more or less OK (w/in 5 from
  reference in UV/Vis/NIR)
• Assess quality of reflectance with new KD

15
Reflectance comparison with MERIS
old
Spectral averaging of SCIA, spatial averaging of
MERIS. Orbits 2509/2510
several versions of new
16
Reflectance comparison with RTM DAK
Sahara state from orbit 2509 KNMI polarization
correction Rest as in other SCIA/RTM study
17
Regular key data (SCIA/5.00) NASA sphere
key data
? strong unphysical spectral features
18
Regular key data (SCIA/5.00) Spectralon
key data
? offset smaller, but unphysical spectral
features remain
19
Channel 3
2509 Sahara state Channel 3 Spectralon NASA
sphere V5.00
20
Conclusion (1)

• Large scale comparison between SCIA and RTM
  suited for finding geographical/slowly varying
  features in data
• Flexible tool, possible to include several data
  sources
• No temporal trends found technique may be used
  for degradation monitoring
• Ch 1 extremely sensitive to calibration (low
  signal)
• Dont use ESRIN NRT Ch1 data before 03/2004!!!

21
Conclusion (2)

• New IFE Key Data investigated
• Irradiance looks reasonable
• Reflectance offset improved
• Many spectral features introduced
• Reflectance is sometimes very noisy
• Alternative use current key data and correct the
  reflectance with a smooth function based on
  validation result