Oxford MIPAS meeting#22
07 Jan 03

• mwmake/ updated (v29DEC02) to include
  • In-flight NESR values
  • SO2 variability reduced to 10% of climatological variability (since climatological variability allows for volcanic activity)
  • 'ails' uncertainty error replaced by 2% 'spread' in AILS width (see below)
  • 'gain' error increased from 1% to 2%

• spectra/
  • Jacobian and error spectra now moved to crun directory
  • namfil.for in /mwmake and /rec directories updated for new location
• OM/ missing-band occupation matrices replaced by those which include CPU considerations - these should more closely resemble the nominal OMs and cause less disturbance to the retrieval characteristics when used instead (not yet delivered to ESA, though)

• (CP) cdproc now modified (but not yet delivered) to process L2 ftp data in the same way as CD data, ie to extract residuals and generate profile maps
• (AD) L2 data collected by ftp from ESRIN and Kiruna up until Jan 5th, but waiting for CP to return to process it.

• Joint retrieval of O3 and H2O from 15-68~km using 10 x 1cm-1 microwindows (no masks) and ESA pT profiles
• Forward model includes: O3, H2O, CO2, N2O, CH4, NO2, HNO3, NH3, CH3Cl, HCN, C2H2
• 5 profiles selected from orbit 2081 for AMIL2DA project.

Source	Fit	Source	Fit	Source	Fit
o3	0.01	ch4	-1.01	so2	-0.02
nonlte	0.08	hno3	-0.77	hcn	-7.19
h2o_st	-0.11	nh3	-0.12	ccl4	0.75
no2	-0.03	n2o5	1.54	hno4	1.01
pre	2.71	cof2	2.46	co2mix	2.55
clono2	1.30	co2	2.13	c2h6	0.11

• Small O3 signature, as desired, but larger stratospheric H2O (-ve sign indicates that retrieved H2O is too high)
• 2.71 pressure error corresponds to (ESA) underestimating pressure by 5.4%.
• Large +ve residual signatures (as expected) from gases not modelled in RAL retrieval (ClONO2, N2O5, COF2, CCl4, HNO4)

• Currently modelled AILS uncertainty (`ails' in error plots) is equivalent to 5% width uncertainty in A band but closer to 1% in AB,B and C bands
• Observed fit of 2nd derivative to monthly averaged residual signatures shows no clear bias, but suggests a uniform fit coefficient of +/-0.04 (equivalent to 2% AILS width) is more representative of observed uncertainty in AILS shape.

1. SO2 variability - decreased to 10% of original size since no volcanic activity has occured
2. Radiometric Gain - increased from 1% to 2%
3. Instrument Line Shape uncertainty - changed to represent 2% uncertainty in AILS width (`spread') given observed 2nd derivative signature in residuals (see above), replaces original `ails' perturbed line shapes
4. NESR - increased to use in-flight average for orbit 2081 rather than pre-flight measurements based on a low detector temperature
5. High altitude variability added (`hialt') - assuming an uncertainty in the retrieved species which increases to climatological values within 10km of the top retrieval level

• Clearer +ve H2O residuals in the non-H2O microwindows
• Low H2O residuals in the H2O microwindows - suggesting that the ESA H2O retrieval is working correctly with the microwindows it is using
• Conclusion: eliminate retrieval/convergence problems as a possible cause of the bias

• H2O is well-fitted across entire H2O microwindow, not just unmasked points, so bias unlikely to be caused by sensitivity of spectral masks to AILS shape
• Revised error analysis shows that although re-evaluated AILS uncertainty (`spread') is twice as large as previous estimate (`ails'), it is still not significant.
• Conclusion: eliminate AILS shape/masks as a possible cause

• Monthly results are generally consistent in showing apparent 2-3 ppmv overestimate of H2O above >25km
• Also show a large underestimate at 12 km.
• Note consistency between these results and apparent MIPAS bias relative to other data sources from the Validation workshop (x-axis scaling is a bit arbitrary). This may just be coincidence since there is no reason to expect the H2O spectroscopy to be more accurate in the non-H2O microwindows than in the H2O microwindows.