Prescribed burning (PB) is an effective and economical land management tool for maintaining fire-adapted ecosystems, reducing wildfire risk, and improving training realism at Department of Defense (DoD) facilities. Pollutants emitted from prescribed fires, however, may be transported downwind, mix with emissions from other sources, form other pollutants, and contribute to poor air quality of urban areas in the region. As a result, compliance with ambient air quality standards in urban areas proximal to DoD prescribed burning activities may require tougher restrictions on DoD’s air emissions in the future. Because the alternatives to PB are costly and may have undesired ecological impacts, it is important for DoD to be able to control the emissions from its PB operations and to minimize their air quality impacts.
This project aimed to improve the characterization of PB emissions by collecting field data most relevant to the modeling of PB plumes and developing a simulation framework that can accurately predict the impacts of prescribed burns on regional air quality. The objectives were to (1) better characterize the fuel types and loads for the sites to be studied; (2) estimate accurately the emissions from the burns to be simulated; (3) develop an air quality modeling system with sufficient resolution to characterize PB plumes, track their regional transport and chemistry, and discern their impacts from other pollution sources; (4) identify the data needs for model evaluation, collect new data, and evaluate the models; (5) consider the effects of inherent uncertainties in model inputs on model results; and (6) simulate the regional air quality impacts of DoD PB operations in the Southeast and assess alternative burning strategies.
This project applied existing U.S. Forest Service tools to the characterization of fuel beds on DoD lands. PB emission factors derived from prior field measurements of particulate matter (PM), carbon monoxide (CO), and volatile organic compounds (VOCs) in the Southeast were compared to laboratory and in situ emissions measurements. “Daysmoke,” an empirical plume model designed specifically for PB, and the Community Multiscale Air Quality (CMAQ) model were coupled using the improved grid resolution provided by the Adaptive Grid (AG) technology. This enabled adequate representation of plume dynamics and chemistry at local scales as well as accurate prediction of impacts over regional scales. The coupled models were first evaluated using existing data. Additional data needs were identified and fulfilled by monitoring campaigns at Fort Benning and Eglin Air Force Base (AFB). The models were evaluated by comparing model predictions to measurements. The sources of uncertainty in model predictions were investigated through sensitivity analysis. Finally, various scenarios were simulated to quantify the air quality impacts of alternative PB scenarios, such as changing the burning times, frequencies, and methods.
The modeling system developed for predicting the air quality impacts of prescribed burns incorporates several new elements into emission, dispersion, and transport process components, all with significant potential for improving the accuracy of the predictions. Initial evaluations of the components individually and together using field data from burns monitored at Fort Benning and observations from regional networks led to significant accuracy improvements. Daysmoke predicted plume structures and smoke concentrations generally in agreement with observations, whereas AG-CMAQ reduced the artificial diffusion inherent to air quality models and produced better defined plumes compared to the standard CMAQ.
The data collected at Fort Benning consisted of measurements of plume height with a lidar ceilometer and short-range smoke concentrations with ground-based mobile platforms. A more comprehensive field study was conducted at Eglin AFB to collect data for final evaluation of the modeling system. As part of this study, the fuels were measured both before and after the burn. Emissions of carbon dioxide (CO2) and PM2.5 were measured with a platform mounted on a tethered aerostat. Comparison to measurements revealed that although both fuel loads and consumptions were overestimated, by 20% and 10% respectively, total emissions were underestimated by 15%. These results showed that the uncertainties in emission estimates are relatively small. Winds were also measured at Eglin with pilot balloon soundings before the burns and airborne (mounted on the aerostat) and ground-based sonic anemometers during the burns. Low wind speeds (WS) and varying wind directions (WD), particularly during transition from land to sea breeze, were difficult to predict with weather prediction models. For sustained and steady wind periods, typical WS errors were 10 to 30% and WD errors were 10 to 20⁰ at altitudes most relevant to plume transport.
Predicted plume heights were in agreement with the ceilometers’ measurements during initial flaming stages of the burns but modeled plumes collapsed faster. Ground-level PM2.5 also was measured at Eglin, both with fixed and mobile platforms. The agreement of predicted ground-level concentrations with observations varied; typically better agreement was obtained near the burn plot and the differences grew with downwind distance. Simulation of a historic smoke incident with the modeling system resulted in significant under-prediction of PM2.5 concentrations observed at monitors about 80 km downwind. The reasons were investigated through a sensitivity analysis. Emission related parameters such as emission strength, timing, and vertical distribution (plume fraction penetrating into the free troposphere) proved to have little effect on the underestimation. WS and WD emerged as the most prominent factors. Sensitivity to WD was very large as ±10⁰ rotation of the wind field would make the plume totally miss the monitors. Changing WD by 5⁰ increased modeled peak PM2.5 concentrations at the monitors by 20% of the under-prediction amount. More important, reducing WS by 30% compensated for 70% of the under-prediction. Both of these adjustments are well within the uncertainty bounds determined for wind fields and justified by the differences between meteorological model predictions and data from the most proximate sounding.
This research integrated improved emissions data, important burn-front information, and advanced plume modeling techniques in regional air quality models, which enabled more accurate prediction of the air quality impacts of prescribed burns. Using the developed simulation framework, DoD land managers can plan their operations to minimize the impacts to regional air quality. The benefit to the scientific community is access to improved PB emissions and accurate models for the dispersion, transport, and chemistry of these emissions.