Tangent height verification algorithm Chris Sioris, Kelly Chance, and Thomas Kurosu Smithsonian Astrophysical Observatory
Algorithm described in Sioris et al., JGR, in press. - find the longest median wavelength for the set of pixel that has the same knee TH on the measured TH grid such that the median is < 305 nm -compare this knee TH (e.g. ~45 km) to one produced by a limb radiative transfer model at the median wavelength for the same conditions (SZA, d , T[z], p[z], O 3 [z]) -algorithm takes 3 sec per limb scan (at 800 MHz) with McLinden et al. model. -this approach has been compared to one where a quadratic in is fitted to observed and simulated knee THs (and the difference between the ‘obs’ and ‘sim’ is the TH offset)
Median used for 305 nm technique
Results Orbit 3422: nm gives wrong TH offset due to wrong mesospheric O 3 in the model
# of limb scans=325, # of orbits = 17, mostly from Aug. and Sep. ’02
Analysis of five consecutive SCIAMACHY orbits
Paired t-test: every 2nd orbit has same latitude sampling scheme, Applied to NH extratropics where algorithm is less sensitive to assumed atm orbit pairs lat range (°) p (that drift is insignificant) drift/orbit (km) 2995 vs % vs % vs % -0.24
Offset statistics (km, obs - model TH) GLOBAL Average: 0.06 stdev: 1.42 maximum: 6.03 minimum: -4.38
Expected error budget for spectral knee at ~305 nm 1. theoretical limit to due spectral resolution and sampling and shot noise ~40 metres 2. completely wrong surface albedo, trop. cloud info: <50 3. completely wrong temperature profile: < % bias in O 3 column (variability for worst-case 200 latitude, for most months above 45 km) 5. interannual variability, (July=worst-case, [Keating <200 et al., 1990]) 6. Neglect of diurnal variation below 53 km ([Keating et al.]) since LT at poles is not LT at equator <50 7. Radiative transfer accuracy planetary and gravity wave activity affecting p at 46 km [Barnett and Labitzke, 1990]: worst-case: hi lat winter 1280 best-case: equator 20-40
Worst-case total error (added in quadrature): ~1300 m at the equator: ~330 m
Application to Level 2: NO 2 profile intercomparison with SAGE III SAGE III pointing for solar occultation is accurate to ~100 m. Vertical column (21-37 km) agreement improves marginally from 6.5% to 3.8% but profile agreement improves from 21% to 11% as sharp features are captured only after TH correction of 2.6 km