Exceedances of critical loads were calculated for terrestrial and aquatic ecosystems and were shown to be highly dependent on the assumed level of base cation emissions. Our estimates of the deposition of S, N and base cations suggest that base cation neutralization occurs in the immediate vicinity of the oil sands, but exceedances of critical loads for both S and N occur further downwind. The area of exceedance depends on the ecosystem examined, varying from 1,800 km2 with respect to S for forest ecosystems, to over 720,000 km2 (larger than the area of France) with respect S+N for aquatic ecosystems. Model-estimated wet deposition of sulphur and nitrogen was in very good agreement with downwind observations of ions within precipitation, though biased high or low close to the oil sands with respect to deposition within snowpack or aircraft observations, respectively.
Ammonia in the Boreal Forest:
Initial model estimates of vertically integrated ammonia were biased significantly low relative to aircraft and satellite observations. These biases were greatly reduced through the adoption of a bi-directional flux algorithm for ammonia, suggesting that re-emission of ammonia from temporary reservoirs in vegetation may be the controlling factor for ammonia concentrations in boreal forest regions. Forest fire emissions of ammonia were found to be the largest direct source of ammonia emissions in the boreal forests.
Secondary organic aerosols:
SOA from oil sands sources originate in chemical pathways which may be unique to the region. Lagrangian box-modelling simulations of the emissions and fate of intermediate volatility organic compounds suggest that they may play a dominating role in organic aerosol formation from oil sands sources. Laboratory box modelling along with 3D GEM-MACH simulations suggest that acid-catalyzed SOA formation contribute to the total SOA loading downwind of the oil sands.
Plume rise from oil sands sources:
Two studies of plume rise from industrial sources were carried out; (a) using plume rise algorithms driven by meteorological observation data from tall towers, a WindRASS, and aircraft, and (b) using the high resolution GEM-MACH model as a test-bed for different plume rise approaches. The standard Briggs (1984) algorithms uniformly under-estimated plume height – this may be due to the observed and complex nature of atmospheric stability in the oil sands region. Improvements to the plume rise algorithms have centered on the adoption of approaches which take into account the local changes in the vertical profile of potential temperature. The potential for double-counting of mixing through the use of algorithms which describe plume rise to equilibrium, and the vertical diffusion process in 3D models, will be discussed.
Meteorological impacts of oil sands emissions:
The feedback version of GEM-MACH (which simulates the particle direct and indirect effects on radiative transfer) was used to examine the effects of model-generated aerosols on simulated meteorology. The use of the feedback scheme was found to result in a significant increase in the average cloud liquid water content over the oil sands region, in a 29-day simulations during summer conditions. The adoption of feedbacks was found to result in improvements of air-quality component predictions relative to surface observations.
Forest fire simulations:
The effects of resolution and emissions algorithms on model simulations of plumes from forest fires in the summer of 2017 was examined in three parallel 10-km to 2.5-km experimental forecast simulations. Comparisons between 10-km and 2.5-km resolution forest fires within the high resolution domain shows the expected increase in near-fire concentrations, and the impact of different approaches for plume rise and emissions estimates on predictions of O3, PM2.5, NO2 and other pollutants will be discussed.
Atmospheric Hg simulations:
Simulations of multiple-years with a research version of GEM-MACH that includes the chemistry, reactions and transport for gaseous elemental mercury, particle bound mercury, and gaseous oxidized mercury were carried out for both 10km and 2.5km model resolutions. The dominant source of Hg in the region by mass was found to be forest fires, though local increases of Hg in the vicinity of the oil sands were observed in snowpack data, and were well simulated by the high resolution version of the model.
Monitoring network analysis:
Associative analyses of monitoring network data collected using both passive and continuous sampling methodologies were carried out, using hierarchical clustering with respect to the metrics of 1-R (R=Pearson Correlation Coefficient) and the Euclidean distance. These allowed a relative ranking of station records according to their degree of similarity, as an aid towards determining potential station location redundancies. GEM-MACH output at station locations was shown to be a suitable surrogate for observation data in hierarchical clustering for at least some chemical species.
Oil sands emissions data collection and processing:
The above studies were supported by a dedicated emissions improvement effort. A new inventory was constructed from multiple sources of emissions data to provide better input data for modelling. Satellite-derived data for land use classification were used to update spatial allocations. Continuous emissions monitoring data of emitted mass, volume flow rates, and temperatures were collected and replaced derived estimates from annual emissions. Aircraft-observation-based estimates of some emissions from oil sands sources differed significantly from previous estimates; the new estimates were incorporated into revised inventories and their impact was studied using GEM-MACH.
We close with a brief discussion of modelling plans for a second phase of the Joint Oil Sands Monitoring Plan, which includes field forecasts for a second aircraft monitoring campaign to take place in the spring and summer of 2018.