Source Attribution of PM2.5 and Health Effects in Kinshasa, Democratic Republic of Congo

DANIEL WESTERVELT, Paulson Kasereka, Garima Raheja, Jean-Luc Balogije Selenge, Rodriguez Yombo Phaka, V. Faye McNeill, Guillaume Kiyombo Mbela, Marianthi-Anna Kioumourtzoglu, Joel Nkiama Konde, Jean-Pierre Mfuamba Mulumba, Djibi Buenimio, Columbia University

     Abstract Number: 457
     Working Group: Identifying and Addressing Disparate Health and Social Impacts of Exposure to Aerosols and Other Contaminants across Continents, Communities, and Microenvironments

Abstract
Estimates of air pollution mortality in sub-Saharan Africa are limited by a lack of surface observations of fine particulate matter (PM2.5). Despite being large metropolises, Kinshasa, Democratic Republic of the Congo (DRC), population 14.3 million, has had little attention towards air quality monitoring. In 2019, a 5-node PurpleAir network was deployed in the city. Calibrated annual average PM2.5 for 2019 in Kinshasa was estimated at 43.5 µg m^-3, more than 8 times higher than WHO Interim Target 1 of 5 µg m^-3. This initial study motivated additional instrumentation in Kinshasa, Congo as a whole, and neighboring Brazzaville. New deployments included a small Clarity Node-S network, a QuantAQ Modulair, and a reference method PM2.5 MetOne Beta Attenuation Monitor (BAM-1020). In addition, monitoring of gas-phase species, including NO2, O3, CO, CO2 is now underway. Here we present first results from this aggregated, multi-sensor, multi-species network in the Congo. We first conduct a sensor intercomparison, comparing the performance of three different popular sensor brands (PurpleAir, Clarity, and QuantAQ) evaluated against the reference BAM-1020. Initial findings suggest that QuantAQ PM2.5 is most correlated and least biased compared to the reference, followed by PurpleAir and by Clarity. We also use our co-location to develop a simple correction factor using both Multiple Linear Regression and Gaussian Mixture Regression, a probabilistic method that has been shown to perform better than commonly used methods in other African cities. We also leverage on-site gaseous pollutant concentrations, particle size distribution data from an optical particle counter, and anemometer data to draw some initial conclusions about sources of PM2.5 in Kinshasa. In particular, we link factors resolved from a nonnegative matrix factorization method using the gaseous species and particle bin concentrations to particular source profiles (e.g. combustion). We find a 3-factor solution that points to a CO-dominated, larger particle source potentially indicative of secondary particles from local combustion, along with a particle-dominated source indicative of primary particles from combustion, and a regional biomass burning source. We leverage wind speed and direction data as well as diurnal cycles to help discern sources. Finally, we use respiratory hospitalization data from Kinshasa to characterize exposure-resposne curves between PM pollution and respiratory outcomes using outcome-specific Poisson mixed models in an estimation framework that allows for overdispersion and include random intercepts to account for within-zone outcome correlation over time. Our results highlight the need for clean air solutions implementation in the Congo.