Exposure to ambient black carbon derived from a unique inventory and high-resolution model

Abstract

Black carbon (BC) is increasingly recognized as a significant air pollutant with harmful effects on human health, either in its own right or as a carrier of other chemicals. The adverse impact is of particular concern in those developing regions with high emissions and a growing population density. The results of recent studies indicate that BC emissions could be underestimated by a factor of 2-3 and this is particularly true for the hot-spot Asian region. Here we present a unique inventory at 10-km resolution based on a recently published global fuel consumption data product and updated emission factor measurements. The unique inventory is coupled to an Asia-nested (∼50 km) atmospheric model and used to calculate the global population exposure to BC with fully quantified uncertainty. Evaluating the modeled surface BC concentrations against observations reveals great improvement. The bias is reduced from -88% to -35% in Asia when the unique inventory and higher-resolution model replace a previous inventory combined with a coarse-resolution model. The bias can be further reduced to -12% by downscaling to 10 km using emission as a proxy. Our estimated global population-weighted BC exposure concentration constrained by observations is 2.14 μg⋅m(-3); 130% higher than that obtained using less detailed inventories and low-resolution models.

Keywords: air pollution; climate change; emission inventory; model resolution.

Fig. 1.

Global distribution of BC emissions in 2007. (A) The 0.1° × 0.1° distribution of BC emissions by PKU-BC-2007. (B) Difference between the PKU-BC-2007 (downscaled to 0.5° × 0.5°) and MACCity (2007, 0.5° × 0.5°) inventories.

Fig. 2.

Comparison between the modeled and observed surface BC concentrations for the four inventory/model combinations. (A) E_MAC/M_IN. (B) E_PKU/M_IN. (C) E_MAC/M_INz. (D) E_PKU/M_INz. The sites in Asia (blue), Africa (black), Europe (green), and North America (red) are marked with different colors. Error bars (vertical lines) are derived by using the first and the third quartiles of PKU-BC-2007 as inputs. NMB is given for each region. The dashed lines show the range of a factor of two deviation of the modeled concentration from the observation and percentages of sites with deviations less than a factor of 2 (F₂) are listed.

Fig. 3.

Spatial distribution of population exposure to BC concentrations for the (A) world, (B) East Asia, and (C) South Asia. The concentrations are derived by downscaling the results from E_PKU/M_INz in Asia and from E_PKU/M_IN in the other regions to 0.1° × 0.1° grids using emission as a proxy. Contours in B and C denote population density.

Fig. 4.

Cumulative frequency distributions of BC exposure in different regions. The BC concentrations are derived by downscaling the results given by E_PKU/M_INz in Asia and by E_PKU/M_IN in the other regions to 0.1° × 0.1° grids using emission as a proxy.

Fig. 5.

Comparison of the calculated surface-air BC concentrations at observation sites and population exposure concentrations in four regions of South Asia, East Asia, Europe, and North America among different methods. The methods compared are four inventory/model combinations (E_MAC/M_IN, E_PKU/M_IN, E_MAC/M_INz, and E_PKU/M_INz) and the two with downscaling [(E_PKU/M_INz)_d for E_PKU/M_INz and (E_PKU/M_IN)_d for E_PKU/M_IN]. The model-calculated average surface-air concentrations at all observation sites are shown as bars, which are compared with the observed mean concentrations (dashed lines) by the marked percentages (modeled concentrations divided by observed ones). Error bars for individual bars are derived by using the first and the third quartiles of PKU-BC-2007 as inputs, which represent the uncertainty range of modeled surface BC concentrations resulting from the uncertainty in emissions. The population-weighted BC exposure concentrations at these sites are shown as diamonds.