Using Combustion Tracers to Estimate Surface Black Carbon Distributions in WRF-Chem

Abstract:

Black Carbon (BC) emissions significantly affect the global and regional climate, air quality, and human health. However, BC observations are currently limited in space and time; leading to considerable uncertainties in the estimates of BC distribution from regional and global models. Here, we investigate the usefulness of carbon monoxide (CO) in quantifying BC across continental United States (CONUS). We use high resolution EPA AQS observations of CO and IMPROVE BC to estimate BC/CO ratios. We model the BC and CO distribution using the community Weather Research and Forecasting model with Chemistry (WRF-Chem). We configured WRF-Chem using MOZART chemistry, NEI 2005, MEGAN, and FINNv1.5 for anthropogenic, biogenic and fire emissions, respectively. In this work, we address the following three key questions: 1) What are the discrepancies in the estimates of BC and CO distributions across CONUS during summer and winter periods?, 2) How do BC/CO ratios change for different spatial and temporal regimes?, 3) Can we get better estimates of BC from WRF-Chem if we use BC/CO ratios along with optimizing CO concentrations? We compare ratios derived from the model and observations and develop characteristic ratios for several geographical and temporal regimes. We use an independent set of measurements of BC and CO to evaluate these ratios. Finally, we use a Bayesian synthesis inversion to optimize CO from WRF-Chem using regionally tagged CO tracers. We multiply the characteristic ratios we derived with the optimized CO to obtain BC distributions. Our initial results suggest that the maximum ratios of BC versus CO occur in the western US during the summer (average: 4 ng/m3/ppbv) and in the southeast during the winter (average: 5 ng/m3/ppbv). However, we find that these relationships vary in space and time and are highly dependent on fuel usage and meteorology. We find that optimizing CO using EPA-AQS provides improvements in BC but only over areas where BC/CO ratios are close to observed values.Black Carbon (BC) emissions significantly affect the global and regional climate, air quality, and human health. However, BC observations are currently limited in space and time; leading to considerable uncertainties in the estimates of BC distribution from regional and global models. Here, we investigate the usefulness of carbon monoxide (CO) in quantifying BC across continental United States (CONUS). We use high resolution EPA AQS observations of CO and IMPROVE BC to estimate BC/CO ratios. We model the BC and CO distribution using the community Weather Research and Forecasting model with Chemistry (WRF-Chem). We configured WRF-Chem using MOZART chemistry, NEI 2005, MEGAN, and FINNv1.5 for anthropogenic, biogenic and fire emissions, respectively. In this work, we address the following three key questions: 1) What are the discrepancies in the estimates of BC and CO distributions across CONUS during summer and winter periods?, 2) How do BC/CO ratios change for different spatial and temporal regimes?, 3) Can we get better estimates of BC from WRF-Chem if we use BC/CO ratios along with optimizing CO concentrations? We compare ratios derived from the model and observations and develop characteristic ratios for several geographical and temporal regimes. We use an independent set of measurements of BC and CO to evaluate these ratios. Finally, we use a Bayesian synthesis inversion to optimize CO from WRF-Chem using regionally tagged CO tracers. We multiply the characteristic ratios we derived with the optimized CO to obtain BC distributions. Our initial results suggest that the maximum ratios of BC versus CO occur in the western US during the summer (average: 4 ng/m3/ppbv) and in the southeast during the winter (average: 5 ng/m3/ppbv). However, we find that these relationships vary in space and time and are highly dependent on fuel usage and meteorology. We find that optimizing CO using EPA-AQS provides improvements in BC but only over areas where BC/CO ratios are close to observed values.