Iterative maximum a posteriori (IMAP)-DOAS for retrieval of strongly absorbing trace gases: Model studies for CH4 and CO2 retrieval from near infrared spectra of SCIAMACHY onboard ENVISAT
[摘要] In the past, differential optical absorption spectroscopy (DOAS) hasmostly been employed for atmospheric trace gas retrieval in theUV/Vis spectral region. New spectrometers such as SCIAMACHY onboardENVISAT also provide near infrared channels and thus allow for the detection of greenhouse gases like CH
4, CO
2, or N
2O. However, modifications of the classical DOAS algorithm are necessary to account for theidiosyncrasies of this spectral region, i.e. the temperature andpressure dependence of the high resolution absorption lines.Furthermore, understanding the sensitivity of the measurement ofthese high resolution, strong absorption lines by means of anon-ideal device, i.e. having finite spectral resolution, is ofspecial importance. This applies not only in the NIR, but can alsoprove to be an issue for the UV/Vis spectral region.
This paper presents a modified iterative maximum a posteriori-DOAS(IMAP-DOAS) algorithm based on optimal estimation theoryintroduced to the remote sensing community by rodgers76.This method directly iterates the vertical column densities of theabsorbers of interest until the modeled total optical density fitsthe measurement. Although the discussion in this paper lays emphasison satellite retrieval, the basic principles of the algorithm alsohold for arbitrary measurement geometries.This new approach is applied to modeled spectra based on acomprehensive set of atmospheric temperature and pressure profiles.This analysis reveals that the sensitivity of measurement stronglydepends on the prevailing pressure-height. The IMAP-DOAS algorithmproperly accounts for the sensitivity of measurement on pressure dueto pressure broadening of the absorption lines. Thus, biases in theretrieved vertical columns that would arise in classical algorithms,are obviated. Here, we analyse and quantify these systematic biasesas well as errors due to variations in the temperature and pressureprofiles, which is indispensable for the understanding ofmeasurement precision and accuracy in the near infrared as well asfor future intercomparisons of retrieval algorithms.
