Insights from the BRAVO study on nesting global models to specify boundary conditions in regional air quality modeling simulations

Abstract

Regional air quality simulations used for source apportionment must consider the role of boundary conditions on estimated species concentrations. This technical note, as part of the Big Bend Regional Aerosol and Visibility and Observational (BRAVO) Study, examines how sulfate concentrations simulated by the Regional Modeling System for Aerosols and Deposition (REMSAD) were influenced by two sets of sulfur boundary conditions: (1) spatially invariant lateral boundary conditions, in which sulfur levels were fixed in space on all four boundaries, and (2) boundary conditions derived from Georgia Tech/Goddard Global Ozone Chemistry Aerosol Radiation Transport Model (GOCART), resulting in sulfur levels that varied spatially. A series of REMSAD simulations were evaluated, and used to apportion sulfate compounds at Big Bend National Park (NP) in southwestern Texas. The spatially invariant boundary conditions specified sulfur dioxide and particulate sulfate mixing ratios near the surface at 200 and 280 ppt, respectively, and dropped to 50 and 70 ppt, respectively, at the highest model layer. The GOCART-derived boundary conditions, however, could be considerably higher or lower, e.g., sulfur dioxide mixing ratios ranging from 2000 ppt along the highly populated northeastern portion of the boundary to less than 10 ppt in the upper model layers. The introduction of the spatially varying GOCART sulfur mixing ratios resulted in a modest overall improvement in the ability of REMSAD to simulate regional sulfate concentrations (an improvement from 48% to 45% and 32% to 21% in normalized error and normalized bias, respectively). However, the sulfate source apportionment at Big Bend NP, a relatively remote area of North America, was significantly influenced, with the estimated contribution of boundary conditions to sulfate levels during the four-month study decreasing from 20% using the original boundary conditions to 7% using the GOCART-derived boundary conditions. This effect was even more pronounced during the 20% clearest days at Big Bend, with 25% of sulfate attributed to the spatially invariant boundary conditions vs. 10% of sulfate attributed to the GOCART-derived boundary conditions. These results do not represent a formal sensitivity study on the role of different boundary condition scenarios as year-specific, temporally varying concentrations were not available during the tightly scheduled BRAVO study. However, the results do indicate the benefits and necessity of incorporating global-scale models to specify boundary conditions in regional-scale air quality model simulations.