Use of MODIS products to simplify and evaluate a forest fire plume dispersion model for PM10 exposure assessment

Plume dispersion models may improve assessment of the health effects associated with forest fire smoke, but they require considerable expertise in atmospheric and fire sciences to initialize and evaluate. Products from MODIS (Moderate Resolution Imaging Spectroradiometer) sensors can simplify the process by providing (1) estimates of fire location, size and emission rates, and (2) data useful for assessing model output. By grouping individual MODIS fire pixels into discrete events we simulated the growth and decay of large fires and estimated their total burned area. Radiative power measurements for each fire pixel were multiplied against a fuel-specific coefficient to estimate particle emission rates. Using the CALMET/CALPUFF package we modeled the dispersion of these particles throughout a 325,000 km 2 area with complex terrain. Moderate agreement (mean r = 0.61) between estimated and measured PM10 concentrations was observed at five of six sites. Because surface measurements are only made at a limited number of locations, we used aerosol optical thickness (AOT) and color imagery product from MODIS for further evaluation. Strong trend association was observed between surface concentrations, model estimates and the AOT measurements. When CALPUFF plume contours were compared to smoke outlines traced from MODIS images we found an average overlap of 50% with better performance under high wind conditions. We conclude that this relatively simple and globally applicable approach can provide a strong foundation for enhanced smoke exposure modeling and public health risk assessment. (c) 2008 Elsevier Ltd. All Fights reserved.