Sensitivity of the CALMET-CALPUFF model system on estimating PM10 concentrations at a mining site in northern Colombia

Abstract

Pollutant dispersion models are essential tools to assess the influence of emission sources in the atmosphere. We performed PM10 simulations using the CALMET-CALPUFF system at a mining site in northern. We set the dynamic parameters (TERRAD, RMAX1, RMAX2, R1, R2, and Calmet-Bias) and established different horizontal grid spacing (0.5 km, 1 km, 2 km, 3 km, and 5 km). Also, we adjusted the surface roughness length (0.1 m–0.5 m) values to determine the sensitivity of the dispersion model. Blocking effects due to terrain, represented by TERRAD, and horizontal resolution in the simulations significantly influence PM10 estimates. TERRAD values below 6 km reduced the values of MSE (0.75 μg/m3 - 4 μg/m3), BIAS (0.5 μg/m3–6 μg/m3), and FB (up to 0.32) at all receptors. The modeling's horizontal resolution of 0.5 km allowed us to identify high PM10 concentrations areas. However, it increases the MSE values (6.42 μg/m3–12.29 μg/m3) compared to the EPA-Federal-Land-Manager-recommended parameter configuration settings of the modeling system. Increasing surface roughness length values slightly increased the MSE values (<2 μg/m3) in the urban area, whereas the MSE values were slightly reduced in the rural receptors (<3 μg/m3). Recognizing the parameters of the CALMET-CALPUFF model system to which solutions are most sensitive allows faster calibration and representative simulations that are input to public health policies.

Keywords

• Horizontal resolution
• Atmospheric modeling
• Parameterization
• Sensitivity analysis