Evaluation of ClearSky2 for PM2.5 Concentrations and Plume Rise

ClearSky2 (CS2), a system simulating PM2.5 concentrations from agricultural field burning smoke, is applied to and evaluated using observations of PM2.5 and plume rise, for the RARE 2013 campaign. CS2 is an air quality simulation system that uses WRF (Weather Research and Forecasting Model) meteorology to predict PM2.5 concentrations from agriculture burning for use in the management of such burning to protect air quality and human health. CS2 models fires as point sources, with burn parameters, such as acres to burn and fuel load in tons per acre, expressed in point source Sparse Matrix Optimized Kernel for Emissions (SMOKE) input files. SMOKE calculates estimates of fire emissions varying over time and space, including plume rise. SMOKE results include multiple emitted species, but CS2 models just PM2.5, treating it as an inert tracer, using an inert tracer version of the Community Model for Air Quality (CMAQ v4.7.1) to simulate dispersion, resulting in hourly gridded concentration maps.

In 2013, an agricultural burning emissions study called RARE observed PM2.5 and other VOC concentrations from field burns on two farms, near Nezperce, ID and near Walla Walla, WA. RARE measured these concentrations with both airborne and ground-based instruments. CS2 was used to model the RARE field burning, using both 4-km and 1.333-km WRF forecast meteorology. CS2-simulated RARE PM2.5 and plume heights are compared to observed Environmental Beta Attenuation Monitor (EBAM) and Light Detection and Ranging (LIDAR) observations. Although CS2 typically utilizes the 4-km WRF meteorology, here we also report a trial using the 1.33-km WRF, so ClearSky2 (CS2) is renamed to ClearSky3 (CS3). Comparison of CS2 and CS3 results against EBAMS and LIDAR observations show: 1) CS2 under-estimates PM2.5 but gets plume rise reasonably well, and 2) CS3 (only for Walla Walla) does better in PM2.5 timing and concentration and gets plume rise reasonably well.

While these RARE results are sparse, and more replications are needed, they suggest greater confidence in CS3 over CS2 for two reasons. First, CS3 has a greater ability of to spatially discriminate EBAMs stationed around each agriculture burn. Second, CS3 seems able to model PM2.5 concentrations more accurately than CS2, in both in magnitude and in timing. Further research should include: 1) wind rose plots to display microscale wind shifts to support analysis of errors in PM2.5 concentrations, and 2) HYSPLIT trajectories for comparing the 4-km and 1.333-km plume paths.