Measuring and modeling of residential black carbon concentrations in two megacities, China

Abstract

Indoor exposure to black carbon (BC) may differ greatly from ambient concentrations measured at stationary monitoring stations. This study characterized the variation between measured indoor and outdoor levels of BC and developed models to predict BC exposures within residences. Indoor and outdoor PM2.5 filter samples were collected for five consecutive days from 66 residences in Beijing (BJ) and Nanjing (NJ), China, across two seasons. City-specific indoor BC models were developed using liner mixed effect models and assessed by R-2 and root mean square error (RMSE). The mean (SD) of indoor BC concentrations in BJ and NJ were 3.0 mu g/m(3) (0.6 mu g/m(3)) and 3.2 mu g/m(3) (0.5 mu g/m(3)), respectively. Both the indoor and outdoor BC concentrations were significantly higher during heating season than non-heating season. In general, indoor levels of BC were strongly associated with outdoor measurements (BJ: r(s) = 0.74, p < 0.001; NJ: r(s) = 0.76, p < 0.001), but were often lower. The critical determinants influencing indoor BC exposures varied by city, including outdoor BC concentration, window opening time, presence of indoor smoking, and the volume of the room where the indoor monitor was placed. The final models accounted for 66.4%-86.5 % (RMSE: 0.070-0.106) of the variance in residential-indoor BC exposures. By incorporating the key BC exposure determinants identified here, this modelling approach can improve the estimates of BC exposure and better link BC exposure to health risk assessment.