Photochemical grid model performance with varying horizontal grid resolution and sub-grid plume treatment for the Martins Creek near-field SO_2 study

摘要

Near source modeling is needed to assess primary and secondary pollutant impacts from single sources and single source complexes. Source-receptor relationships need to be resolved from tens of meters to tens of kilometers. Dispersion models are typically applied for near-source primary pollutant impacts but lack complex photochemistry. Photochemical models provide a realistic chemical environment but are typically applied using grid cell sizes that may be larger than the distance between sources and receptors. It is important to understand the impacts of grid resolution and sub-grid plume treatments on photochemical modeling of near-source primary pollution gradients. Here, the CAMx photochemical grid model is applied using multiple grid resolutions and sub-grid plume treatment for SO_2 and compared with a receptor mesonet largely impacted by nearby sources approximately 3-17 km away in a complex terrain environment. Measurements are compared with model estimates of SO_2 at 4- and 1-km resolution, both with and without sub-grid plume treatment and inclusion of finer two-way grid nests. Annual average estimated SO_2 mixing ratios are highest nearest the sources and decrease as distance from the sources increase. In general, CAMx estimates of SO_2 do not compare well with the near-source observations when paired in space and time. Given the proximity of these sources and receptors, accuracy in wind vector estimation is critical for applications that pair pollutant predictions and observations in time and space. In typical permit applications, predictions and observations are not paired in time and space and the entire distributions of each are directly compared. Using this approach, model estimates using 1-km grid resolution best match the distribution of observations and are most comparable to similar studies that used dispersion and Lagrangian modeling systems. Model-estimated SO_2 increases as grid cell size decreases from 4 km to 250 m. However, it is notable that the 1-km model estimates using 1-km meteorological model input are higher than the 1-km model simulation that used interpolated 4-km meteorology. The inclusion of sub-grid plume treatment did not improve model skill in predicting SO_2 in time and space and generally acts to keep emitted mass aloft.