Speaker
Description
Nitrogen dioxide (NO2) has absorption lines in UV and VIS spectral windows and can interfere with the ozone retrievals derived from Ozone Mapping and Profiling Suite Limb Profiler (OMPS-LP). To correct for this effect, we constructed NO2 climatology using simulated NO2 profiles with the GMI GEOS CCM model by averaging model data over 6 years. This climatology provides vertical NO2 profiles as a function of latitude, season, and local solar time. In this study, we evaluate the quality of the GMI climatology and compare it with the available measurements of NO2 profiles from SAGE III and Odin OSIRIS. We also evaluate sensitivity of the ozone retrieval algorithm to variations in stratospheric NO2.
OMPS-LP makes hyperspectral observations covering a spectral range from 295-1000 nm, but the coarse spectral resolution of LP hinders retrievals of NO2 profiles with traditional inverse models. Here we explore a new approach based on a machine learning (ML) algorithm to derive stratospheric NO2 vertical profiles from OMPS-LP. We train the ML algorithm using radiances calculated with the Gauss-Seidel limb scattering radiative transfer model and NO2 profiles derived from the GMI model. We evaluate NO2 profiles predicted by the ML algorithm and compare them with available measurements.