Revisiting the high tropospheric ozone over southern Africa: role of biomass burning and anthropogenic emissions

Abstract. Tropospheric ozone over southern Africa is particularly high and causes tremendous health risks and crop yield losses. It has been previously attributed to the influence by biomass burning (BB), with a neglected contribution from anthropogenic emissions. However, due to the lack of measurements for ozone and its precursors, the modeled impacts of BB and anthropogenic emissions on tropospheric ozone levels in southern Africa were not well evaluated. In this study, we combined the nested GEOS-Chem simulation with a horizontal resolution of 0.5°×0.625° with available multiple observations at the surface and from space to quantify tropospheric ozone and its main drivers in southern Africa. Firstly, BB emissions from current different inventories exhibit similar peaks in the summer season but also have large uncertainties in southern Africa (e.g., uncertainty of a factor of 2–3 in emitted NOx). The model–satellite comparison in the fire season (July–August) in 2019 shows that using the widely used Global Fire Emissions Database version 4.1 (GFED4.1) inventory, the model tends to overestimate by 87 % compared to OMI NO2, while the Quick Fire Emissions Dataset (QFED2) inventory can greatly reduce this model bias to only 34 %. Consequently, the modeled tropospheric column ozone (TCO) bias was reduced from 14 % by GFED4.1 to 2.3 % by QFED2. In addition, the QFED2 also has a much better spatial representativeness than GFED4.1. The simulated surface daily maximum 8 h mean (MDA8) ozone was decreased from 74 ppb by GFED4.1 to only 56 ppb by QFED2. This suggests a highly overestimated role of BB emissions in surface ozone if GFED4.1 is adopted. The model–observation comparison at the surface shows that the global Community Emissions Data System (CEDSv2) anthropogenic inventory tends to underestimate anthropogenic NOx emissions in typical southern African cities and likely misrepresented anthropogenic sources in some areas. That means that urban ozone and PM2.5 concentrations in southern Africa may be strongly underestimated. For example, a 10-fold increase in anthropogenic NOx emissions can change the ozone chemistry regime and increase PM2.5 by up to 50 µg m−3 at the city of Luanda. Furthermore, we also find that TROPOMI can already capture the urban NO2 column hotspots over low-emission regions like southern Africa, while this is unavailable from the OMI instrument, highlighting the critical role of high-resolution measurements in understanding atmospheric chemistry issues over southern Africa. Our study presents a deeper understanding of the key emission sources and their impacts over southern Africa that will be helpful not only to formulate targeted pollution controls, but also to enhance the capability to predict future air quality and climate change, which would be beneficial for achieving a healthy, climate-friendly, and resilient development in Africa.