Articles | Volume 10, issue 3
Research article
21 Mar 2017
Research article |  | 21 Mar 2017

Improved OSIRIS NO2 retrieval algorithm: description and validation

Christopher E. Sioris, Landon A. Rieger, Nicholas D. Lloyd, Adam E. Bourassa, Chris Z. Roth, Douglas A. Degenstein, Claude Camy-Peyret, Klaus Pfeilsticker, Gwenaël Berthet, Valéry Catoire, Florence Goutail, Jean-Pierre Pommereau, and Chris A. McLinden

Abstract. A new retrieval algorithm for OSIRIS (Optical Spectrograph and Infrared Imager System) nitrogen dioxide (NO2) profiles is described and validated. The algorithm relies on spectral fitting to obtain slant column densities of NO2, followed by inversion using an algebraic reconstruction technique and the SaskTran spherical radiative transfer model (RTM) to obtain vertical profiles of local number density. The validation covers different latitudes (tropical to polar), years (2002–2012), all seasons (winter, spring, summer, and autumn), different concentrations of nitrogen dioxide (from denoxified polar vortex to polar summer), a range of solar zenith angles (68.6–90.5°), and altitudes between 10.5 and 39 km, thereby covering the full retrieval range of a typical OSIRIS NO2 profile. The use of a larger spectral fitting window than used in previous retrievals reduces retrieval uncertainties and the scatter in the retrieved profiles due to noisy radiances. Improvements are also demonstrated through the validation in terms of bias reduction at 15–17 km relative to the OSIRIS operational v3.0 algorithm. The diurnal variation of NO2 along the line of sight is included in a fully spherical multiple scattering RTM for the first time. Using this forward model with built-in photochemistry, the scatter of the differences relative to the correlative balloon NO2 profile data is reduced.

Short summary
A new OSIRIS NO2 retrieval algorithm is described and validated using > 40 balloon-based profile measurements. The validation results indicate a slight improvement relative to the existing operational algorithm in terms of the bias versus the balloon data, particularly in the lower stratosphere. The implication is that this new algorithm should replace the operational one. The motivation was to combine spectral fitting and the SaskTRAN radiative transfer model to achieve an improved product.