Analysis of Canadian Tropospheric Ozone Measurements from Geostationary Orbit and An Assessment of Non-Coincident Limb-Nadir Matching for Measuring Tropospheric Nitrogen Dioxide

This thesis work attempts to improve the quality of surface-level pollutant concentrations retrieved from satellite-borne optical instruments. In the first part of the present work, an analysis is performed to determine potential benefits of implementing a different radiative transfer model than the one planned for retrieving Canadian tropospheric ozone concentrations with future measurements from the Tropospheric Emissions: Monitoring of Pollution (TEMPO) optical instrument, planned to be launched in 2022 into geostationary orbit to measure tropospheric pollutants over the majority of North America. The plane-parallel Earth-atmosphere geometry assumption for multiple-scattered electromagnetic radiation in the planned radiative transfer model for the TEMPO ozone retrieval algorithm has minimal effect for heritage instruments that look at angles close to straight down and measure at local times where the Sun is far above the horizon. However, it is demonstrated in the present work for simulated TEMPO measurements over the Canadian Oil Sands that the retrieval error for a radiative transfer model with a plane-parallel geometry can reach approximately 15% at 13:00 local time, 25% in March or September near local sunrise, 50% in June near local sunrise, and 80% in December near local sunrise, while a radiative transfer model with a spherical geometry results in error up to an order of magnitude smaller in each case. Further work is required to assess the effects of the geometry assumptions on different orders of scattering and of measurement noise. In the second part of the present work, a novel method of estimating tropospheric NO2 pollution using non-coincident limb- and nadir-viewing instrument measurements is further assessed with a reanalysis using new datasets produced by the Ozone Monitoring Instrument (OMI), the Optical Spectrograph and Infrared Imager System (OSIRIS), and a photochemical box model, and an analysis using OSIRIS and the TROPOspheric Monitoring Instrument (TROPOMI). A bias is demonstrated in the current publicly available OSIRIS NO2 density profile data, leading to the development of an updated dataset that is shown to agree with a previously validated dataset within retrieval error bounds above the tropopause. The OSIRIS-OMI reanalysis demonstrates biases of up to 0.5*10^15 molecules/cm^2 due to the different photochemical box model input parameters and up to 0.2*10^15 molecules/cm^2 due to the use of the latest OMI NO2 dataset. The OSIRIS-TROPOMI analysis demonstrates a positive average bias of approximately 0.5*10^15 molecules/cm^2 in the limb-nadir matching with TROPOMI relative to that with OMI due to TROPOMI-OMI tropospheric and stratospheric NO2 column density biases. Error range estimates of photochemical box model input parameters and of different versions of OMI datasets, further analysis of local and yearly dependencies of OSIRIS-OMI limb-nadir matching biases, and further studies on latitudinal and seasonal dependencies of TROPOMI-OMI dataset biases are recommended for future work