Retrieval and validation of global tropospheric nitrogen dioxide (NO2) vertical profiles obtained via cloud-slicing TROPOMI partial columns

Observations of the vertical distribution of nitrogen oxides (NO_x ≡ NO + NO₂) in the troposphere are severely limited, despite its influence on ozone formation. Here, we derive vertical profiles of the NO_x component NO₂ by applying cloud-slicing to partial columns of NO₂ from the space-based TROPOMI instrument. This yields seasonal means of NO₂ volume mixing ratios at ~100 km resolution for multiple years (March 2018 to February 2022) on a global scale in the upper troposphere (180-320 hPa and 320-450 hPa), the middle troposphere (450-600 hPa and 600-800 hPa) and the boundary layer (800 hPa to the Earth’s surface). We evaluate our product against in situ NO₂ measurements from NASA DC-8 aircraft campaigns over Canada (ARCTAS, ATom, INTEX-A), the Eastern US (ATom, SEAC4RS, INTEX-A), the North and South Atlantic (ATom), and the Central and South Pacific (ATom) and use our validated dataset to assess state-of-knowledge of global tropospheric NO_x as simulated by GEOS-Chem.  In the middle troposphere, cloud-sliced NO₂ has a mean value of 20-40 pptv and deviates by < 5 pptv where NO₂ from aircraft observations exceeds the instrument detection limit. The consistency between cloud-slicing results and aircraft observations here is due to high sampling frequency and ideal conditions for cloud-slicing. Differences with aircraft observations are larger (up to 120 pptv) in the upper troposphere between 320-180 hPa where aircraft observations may be susceptible to biases and where cloud-sliced NO₂ data are relatively sparse. In the boundary layer, retrievals consistent with the aircraft observations are only possible over marine environments where NO₂ concentrations differ by < 35 pptv compared to > 450 pptv over terrestrial regions. This is because large land-based NO_x sources cause steep vertical NO₂ gradients that are problematic for cloud-slicing which assumes NO₂ is well mixed throughout the troposphere. We find that NO₂ concentrations above the Eastern US differ by < 20 pptv when comparing cloud-sliced tropospheric vertical profiles to simulated vertical profiles from the GEOS-Chem chemical transport model. However, GEOS-Chem consistently underestimates concentrations of NO₂ in the remote troposphere, simulating concentrations that are 50% less than the mean cloud-sliced NO₂ observations. This is a result of the limited number of current NO₂ observations used to validate models like GEOS-Chem which are limited in both time and space. By deriving tropospheric vertical profiles from cloud-slicing satellite observations there is an opportunity to obtain routine NO₂ observations which can then be compared to aircraft measurements and simulations from the GEOS-Chem model. From this, we can determine the environmental factors that impact tropospheric NO_x on a global scale and address long-standing uncertainties in our understanding of NO_x in the troposphere.