Evaluation of High-resolution GEOS-Chem Nested Grid Simulations over Africa Using a Novel Surface Aerosol and Gas Dataset and Satellite Observations

DANIEL WESTERVELT, Benjamin Yang, Columbia University

     Abstract Number: 505
     Working Group: Aerosols Spanning Spatial Scales: Measurement Networks to Models and Satellites

Abstract
Air pollution is a growing problem in Africa, with around 1 million premature deaths attributed to poor air quality in 2019 on the African continent. Although measurements of PM2.5 and gaseous air pollutants on the continent are very sparse, new efforts by both African and international organizations have increased public data availability in recent years. New networks of PM2.5 rely on both reference monitoring and consumer-grade low cost sensors, the latter of which, if well calibrated, have been proposed as a potential means of closing this data gap in Africa. These emerging datasets provide a unique opportunity to evaluate regional and global chemical transport models over Africa so that projections and predictions from such modeling efforts can be more reliable. We will present an overview of these new datasets that span the African continent in at least 15 countries and cover total PM2.5, NOx, and O3. We also present simulations of atmospheric chemistry, emissions, transport, deposition, and meteorology using the GEOS-Chem chemical transport model. We employ a 25 by 25-kilometer horizontal spatial resolution regional nested simulation over equatorial Africa and simulate meteorological years 2021 and 2022, allowing us to take advantage of novel surface observations established in only the last few years. Emissions in Africa are from the DICE-Africa inventory. We evaluate daily, monthly, and hourly surface PM2.5 in several cities, including: Accra, Ghana; Lomé, Togo; Kinshasa, DRC; Brazzaville, ROC; Kampala, Uganda; Nairobi, Kenya; Addis Ababa, Ethiopia; and more. Initial results show that this high-resolution model reproduces surface PM2.5 reasonably well. In Accra, March – October 2021 modeled surface 24-hour mean PM2.5 was moderately well correlated with surface observations (r2 = 0.57). Overall, the model underpredicted PM2.5 in March-October 2021, with an average of 23.6 µg m^-3 observed PM2.5 versus 18.8 predicted by GEOS-Chem. Finally, we leverage satellite observations of NO2 from TROPOMI and AOD from MODIS to compare to both the GEOS-Chem model output and the ground-based observations. These findings will help inform and ultimately improve air quality modeling efforts over Africa.