New top-down estimation of daily mass and number column density of black carbon driven by OMI and AERONET observations

Abstract

This work uses a mixture of observations from surface remote sensing (AERONET) and satellite remote sensing (OMI) to uniquely compute the atmospheric column loading of black carbon (BC) mass concentration density (MCD) and number concentration density (NCD) on a grid-by-grid, day-by-day basis at 0.25°x0.25° over rapidly developing and biomass burning (BB) impacted regions in South, Southeast, and East Asia. This mixture of observations is uniformly analyzed based on OMI NO₂ retrievals, OMI Near ultraviolet band absorption aerosol optical depth and single scattering albedo (SSA), and AERONET visible and near-infrared band SSA observations, in connection with an inversely applied MIE mixing model approach. This method uniquely solves for the unbiased spatial and temporal domains based on variance maximization of daily NO₂. These locations in space and time are then used to quantify the distribution of all possible individual particle core and refractory shell sizes as constrained by all band-by-band observations of SSA from AERONET. Finally, the range of NCD and MCD are computed from the constrained range of per-particle core and refractory shell size on a grid-by-grid and day-by-day basis. The maps of MCD and NCD are consistent in space and time with known urban, industrial, and BB sources. The statistical distributions are found to be non-normal, with the region-wide mean, 25th, 50th, and 75th percentile MCD [mg/m²] of 90.3, 56.1, 81.1, and 111 respectively, and NCD [x10¹² particles/m²] of 8.76, 4.63, 7.39, and 11.3 respectively. On a grid-by-grid basis, a significant amount of variation is found, particularly over Myanmar, Laos, northern Thailand, and Vietnam, with this subregional mean, 25th, 50th, and 75th MCD [mg/m²] of 90.7, 56.1, 81.3, and 112 respectively and NCD [x10¹² particles/m²] of 9.66, 5.49, 8.33, and 12.3 respectively. On a day-to-day basis, events are determined 121 days in 2016, during which the computed statistics of MCD and NCD have mean and uncertainty ranges which scale with each other. However, there are 11 days where the uncertainty ratio of NCD values is larger than 1 while the uncertainty ratio of MCD is small, and 5 days where the reverse is observed, indicating that the particle size is strongly atypical on these days, consistent with mixed aerosol sources, a substantial change in the aerosol aging, or other such factors including a substantial region of overlap between BB and urban sources. The high values observed from March to May lead to an extended BB season as compared to previous work focusing on fire radiative power, NO₂, and models, which show a shorter season (usually ending in early April). The results are consistent with BC being able to transport significant distances. The new approach is anticipated to provide support for improving radiative forcing calculations, estimating emissions inventories, and providing a basis by which models can compare against observations.