Validation of OMPS-LP Radiances
P. K. Bhartia, Leslie Moy, Zhong Chen, Steve Taylor
NASA Goddard Space Flight CenterGreenbelt, Maryland, USA
Motivation
• Find causes of large radiance residuals from the L2 algorithm
• Improve altitude registration methods• Isolate systematic errors in measured and
calculated radiances• Evaluate accuracy of MLS and NCEP data• Better understand information content of
measurement
Methodology
• Radiance Simulation– Bass & Paur cross-sections– Atlas/SUSIM solar irradiance– MLS O3, temp and GPH profiles– OMPS-NP reflectivity
• limb RTM is scalar but nadir is vector – NO2 climatology, No aerosols
• Measured data– Solar irradiances – Ungridded UV (290-350 nm) radiances from two “high
gain” images
MLS GPH uncertainties
64 km48 km32 km16 km
Z*
500 m error in GPH will produce ~8% error in calculated radiances Error in T that causes GPH errors will produce additional error in radiances
Scalar radiance error at TH = 40 km, R = 0.3Error is shown for λ
= 325, 345, 385, 400, 449, 521 nm (solid lines) and 602, 676, 756, 869, 1020 nm (dashed line)
Same scalar radiance error pattern prevails for all wavelengths
Amplitude at 345 nm is reduced (-3.5% to +5.5%), due to larger R
% change in 350 nm radiance due to aerosols
% change shown for TH = 20, 25, 30, 35, 40 km
Surface reflectivity = 0
λ = 350 nm
LP Focal Plan Schematic
Designed for sequencing HG Long/LG long/HG Short/LG short: 1: 4.5: 7: 4.5 Total dynamic range gain: x140
Two interleaved exposures in 1:31 ratio
Ratio: 1:4.5
Low gain
High gain
High Gain Image in UV
No HG data
HG long
There are systematic differences between HG & LG images so our plan is not to use LG image in the UV. This will free up some bytes for other use.
HG short
No data
Optical distortions in HG Image Variation of wavelength with TH
• variation is smaller than instrument bandpass, but still needs to be corrected.• variation is 4 times worse for LG image.
Fixed column no
Optical distortions in HG Image Variation of TH with wavelength
Fixed row no
Wavelength Under-sampling
Without under-sampling corrn interpolation error can be as large as 3%
Radiances convolved with OMPS bandpass
Solar Irradiance (SI) Comparison
SUSIM smoothed with OMPS bandpass
OMPS in FWHM/10 steps
No adj
+0.6 nm shift
error decreases
SI comparison results
• We have ~0.6 nm error in in the 290-320 nm band
• Error decreases with increase in so it is not pixel shift type error
• Bias remaining after +0.6 nm shift is partly due to error and partly due to radiometric calibration errors
Explanation of large radiance residuals in L2
• Partly due to error• Partly due to error in SI assumed in calculating
the radiances in L2– L2 SI doesn’t agree with Atlas/SUSIM
Radiance Comparison Example (49.5 km, 70S)
Black: Measured
Red: Calculated
Radiance bias
Irradiance bias
There may be spectral bias between radiance and irradiance.
Alt Registration using 305 nm
No O3 abs
strong O3 abs
Strong sens to TH
weak sens to TH
305 nm not affected by reflectivity
An example: 70S April 2, 2012
600 m error
No TH error
Variation in est. TH error with alt is probably due to error in MLS GPH
Conclusions
• Release 1 data has 0.6 nm error in the UV.– This amounts to 10% error in O3 x-section. – Limb UV ozone profiles are therefore not reliable.– Error in VIS profile is TBD.
• Release 1 altitude registration seems quite accurate (±300 m)– To improve accuracy we will need to rely on multiple
approaches. • MLS temperature probably has larger bias in the
mesosphere than MLS has estimated
Conclusions (cont’d)
• Given uncertainties and low vert resolution of of NCEP temperature data it may not be possible to produce accurate/high precision MR profiles from LP above ~40 km– Density profiles are not affected by this problem– It may be possible to improve NCEP temp profiles
above 40 km using LP, but this needs further investigation.