Partitioning indoor-generated and outdoor-generated PM2.5 from real-time residential measurements in urban and peri-urban Beijing

Limited number of projects have attempted to partition and quantify indoor- and outdoor-generated PM 2.5 ( PM 2.5 ig and PM 2.5 og ) where strong indoor sources (e.g., solid fuel, tobacco smoke, or kerosene) exist. This study aimed to apply and refine a previous recursive model used to derive infiltration efficiency (F inf ) to additionally partition pollution concentrations into indoor and outdoor origins within residences challenged by elevated ambient and indoor combustion-related sources. During the winter of 2016 and summer of 2017 we collected residential measurements in 72 homes in urban and peri-urban Beijing, 12 of which had additional paired residential outdoor measurements during the summer season. Local ambient measurements were collected throughout. We then compared the calculated PM 2.5 ig and using (i) outdoor and (ii) ambient measurements as model inputs. The results from outdoor and ambient measurements were not significantly different, which suggests that ambient measurements can be used as a model input for pollution origin partitioning when paired outdoor measurements are not available. From the results calculated using ambient measurements, the mean percentage contribution of indoor-generated PM 2.5 was 19 % (σ = 22 %), and 7 % (11 %) of the total indoor PM 2.5 for peri-urban and urban homes respectively during the winter; and 18 % (18 %) and 6 % (10 %) of the total indoor PM 2.5 during the summer. Partitioning pollution into PM 2.5 ig and PM 2.5 og is important to allow investigation of distinct associations between health outcomes and particulate mixes, often with different physiochemical composition and toxicity. It will also inform targeted interventions that impact indoor and outdoor sources of pollution (e.g., domestic fuel switching vs. power generation), which are typically radically different in design and implementation. • Measured indoor PM 2.5 was partitioned by sources with a method adapted for LMICs. • Ambient measurements can be used as an input for indoor pollution partition. • Peri-urban has significant higher infiltration efficiency & indoor-generated PM 2.5 . • Higher indoor-generated PM 2.5 was found in peri-urban Beijing than anywhere studied. • We suggest using partitioned pollution in health studies to reduce measurement error.