Towards high-resolution NO2 monitoring from space: evaluation of a discrete-wavelength DOAS NO2 retrieval using OMI and TROPOMI observations

The use of satellite NO₂ data for air quality studies is increasingly revealing the need for observations with higher spatial and temporal resolution. One way to achieve increased spatial resolution is to reduce the spectral information needed for the retrieval. Conventional satellite instruments measure continuous reflectance spectra at high spectral resolution and typically use a few hundred wavelengths to retrieve NO₂.

In this work we propose a variant of the well-established Differential Optical Absorption Spectroscopy (DOAS) technique, in which only 10 discrete spectral channels are used to retrieve NO₂ slant column densities (SCDs). Spectra from Level 1B data measured by the Ozone Monitoring Instrument (OMI) and TROPOspheric Monitoring Instrument (TROPOMI) over clean and polluted regions are discretised by applying Gaussian filters at NO₂-absorbing wavelengths. NO₂ SCDs are retrieved from these discretised spectra and compared with those retrieved in existing Level 2 datasets for these instruments (OMI: QA4ECV, TROPOMI: Operational), which use conventional DOAS algorithms.

We find very good agreement between our algorithm and Level 2 data for both OMI (mean bias: < 5 %, uncertainty: 0.97 x 10¹⁵ molecules cm^-2) and TROPOMI (mean bias: < 11 %, uncertainty: 0.68 x 10¹⁵ molecules cm^-2). Residual discrepancies can be explained by differences in retrieval implementation. The low biases between our retrieval algorithm and conventional DOAS algorithms demonstrate the efficacy of discrete-wavelength retrievals. These allow for simpler, more economic satellite instrument designs for NO₂ monitoring at high spatial and temporal resolution. Constellations of small satellites with such instruments on board would be a valuable complement to current and upcoming high-budget hyperspectral instruments. Future work involves further development of a new instrument concept based on this method.