Development and evaluation of portable reduced-complexity air quality models for policy assessment in Africa

REACH: an accessible and globally portable air quality model for policy and health impact assessmentAtmospheric models serve an essential role in guiding air pollution management by assessing health impacts from various sources. In the US and other developed nations, chemical transport models (CTMs) are used to predict pollutant concentrations and to evaluate compliance with air quality standards. CTMs are state-of-the-science models but are resource-intensive requiring skilled technical personnel, limiting their use in many policy applications, particularly in developing countries. Reduced-complexity models (RCMs) were developed to address those limitations; they apply simpler chemistry, are easier to use, require fewer and simpler inputs, are faster than CTMs, and can assess more policy scenarios. However, most RCMs to date are limited in scope to the United States, but there are many countries with higher pollutant concentrations and limited resources to manage air pollution. To address this need, we present the Rapid Estimation of Air Concentrations for Health (REACH), a location-agnostic Gaussian plume model with chemistry that predicts PM2.5 concentrations and is easily portable to any region of the globe. Here we report the development of REACH and its evaluation in the United States against ambient measurements and other RCMs. One important output of REACH are estimates of marginal social costs, which are the total mortalities downwind from a source location caused by emitting an additional tonne of a PM2.5 precursor. For the US, REACH estimates average marginal social costs for 2005 ground PM2.5, SO2, NOx, NH3, and VOCs as 142, 73, 10, 73, and 11 $/ktonne, respectively. We demonstrate that the REACH marginal social costs are generally comparable to AP2, EASIUR and InMAP RCMs with R2 ranging from 0.38-0.77 depending on the PM2.5 species. REACH is open-source and aims to aid in air quality decision-making in the US and internationally and to serve as a point of comparison with other RCMs. Development and Evaluation of the REACH Air Quality Model for Policy Assessment in Southern AfricaThe growing health burdens from air pollution in Africa prompt immediate action and policy development to improve public health. There are limitations in applying the most robust state-of-the-art chemical transport models (CTMs) in Africa because of the requirements of high technical expertise and computational costs. Reduced-complexity models (RCMs) are often more suitable for policy assessment than CTMs as they are relatively easier to use and have less resource constraints making them more feasible. Given these benefits, it is advantageous to apply them in resource-scarce regions to predict health impacts from changing emissions and pollutant levels. To help address the needs for RCMs in Africa, here, we apply the REACH RCM to Southern Africa (REACH-SA) with the primary aim of predicting marginal social costs. As part of the evaluation process, the model annual-average PM2.5 was compared to observations in South Africa and performs reasonably well (MFB: 13 %, MFE: 31 %). Given the limited speciated PM2.5 measurements in South Africa, the availability of NOx and SO2 observations allowed for a comparison of REACH-US and REACH-SA calibrations. The ratio of REACH-SA to REACH-US calibrations for NOx and SO2 is 1.2 and 0.7, respectively, indicating that the REACH-US calibrations are transferrable to another region. Using REACH, we also evaluate how well a generic global emissions inventory (EDGAR) explains observed PM2.5 concentrations compared to a region-specific inventory (DACCIWA). Surprisingly, EDGAR performs marginally better than DACCIWA. On average in the region, REACH-SA marginal social costs with EDGAR and DACCIWA are comparable, suggesting that the marginal social costs are not largely dependent on the choice of emissions inventory. This has important implications particularly for primary PM2.5 where marginal social costs are insensitive to the emissions inventory and is useful for policy applications where primary PM2. emissions are dominant. PMCAMx simulations of ambient PM2.5 in East AfricaIn Africa, there are limited tools to manage air pollution. Reduced-complexity models (RCMs) serve as valuable resources in predicting marginal social costs as a health impact metric for policy assessment. They are meant to be complementary to chemical transport models (CTMs), which have more detailed chemical mechanisms and processes, because RCMs require less time and technical expertise to run. Despite the benefits of RCMs with their ease-of-use and feasibility for policy analysis, there are very few RCMs developed for Africa. To help democratise tools for air quality management, we aim to develop the EASIUR RCM for East Africa. As EASIUR is a regression model derived from CTM output, this paper focuses on first configuring the PMCAMx CTM for East Africa. Here, we run the PMCAMx model at 30 km to simulate ambient PM2.5 concentrations. The monthly PM2.5 evaluations in 2023 against AirQo low-cost sensors show reasonable agreement in Kampala with 12% differences in the annual-average PM2.5. Because of the large contribution of primary PM2.5 emissions in the region, the model predictions at 10 km spatial resolution show significant differences. The difference in model PM2.5 predictions with the coarser and higher resolutions (over 40 µg/m3 difference in Kampala) suggest that the model is highly sensitive to horizontal spatial resolution. Longer simulations are needed to evaluate the 10 km model against observations and to conclude the impact of spatial resolution on the PM2.5 output.