Abstract. This study assesses the impact of revised volatile organic compound (VOC) and organic aerosol (OA) emissions estimates in the GEM-MACH (Global Environmental Multiscale‒Modelling Air Quality and CHemistry) chemical transport model, driven with two different emissions input datasets, using observations from the 2013 Joint Oil Sands Monitoring (JOSM) intensive field study. The first emissions dataset (base-case run) makes use of regulatory reported VOC and particulate matter emissions data for the large oil sands mining facilities in northeastern Alberta, Canada, while the second emissions dataset (sensitivity run) uses emissions estimates based on box-flight aircraft observations around specific facilities (Li et al., 2017, Zhang et al., 2017) and a mass-balance analysis (Gordon et al., 2015) to derive total facility emission rates. The preparation of model-ready emissions files for the base-case and sensitivity run is described in an accompanying paper by Zhang et al. (2017). The large increases in VOC and OA emissions in the revised emissions data set for four large oil sands mining facilities were found to improve the modeled VOC and OA concentration maxima in plumes from these facilities, as shown with the 99th percentile statistic and illustrated by case studies. The results show that the VOC emission speciation profile from each oil sand facility is unique and different from standard petrochemical-refinery emission speciation profiles used for other regions in North America. A feedback between larger long-chain alkane emissions and higher secondary organic aerosol (SOA) concentrations was found to be significant for some facilities and improved OA predictions for those plumes. The use of the revised emissions data resulted in a large improvement of the model OA bias; however, the decrease in OA correlation coefficient suggests the need for further improvements to model organic aerosol emissions and formation processes. Including intermediate volatile organic compound (IVOC) emissions as precursors to SOA and spatially allocating more PM1 POA emissions (primary organic aerosol of 1.0 μm or less in diameter) to mine-face locations are both recommended to improve OA bias and correlation further. A systematic bias in the background OA was also predicted on most flights, likely due to under-predictions in biogenic SOA formation. Overall, the weight of evidence suggests that the new aircraft-observation-derived organic emissions help to constrain better the fugitive organic emissions, which are a challenge to estimate in the creation of bottom up emission inventories. This work shows that the use of facility-specific emissions, based on direct observations, rather than generic emission factors and speciation profiles can result in improvements to model predictions of VOCs and OA. Emissions estimation techniques, such as those used to construct the inventories in our study, may therefore have beneficial impacts when applied to other regions with large sources of VOCs and OA.
An economic analysis of oil sands mining options