Prescribed burning (PB) is an effective and economical way of maintaining and improving the ecosystem, and reducing wildfire risk. However, pollutants emitted from the burns may be transported long distances, mix with emissions from other sources, and contribute to air quality problems in downwind urban areas. In the future, compliance with ambient air quality standards may require tougher restrictions on PB emissions. Since the alternatives of PB are costly, it is important for land managers to be able to control the emissions from their PB operations, and to minimize their air quality impacts.

A modeling system is being developed to predict the impacts of prescribed burns on regional air quality. The objectives of this system are to better characterize the emissions from the burns, to accurately simulate the dispersion of the smoke plumes, and to increase the resolution of regional-scale models, so that the impacts of PB emissions can be discerned from other pollution sources in a given region. The system consists of the MM5 meteorology and CMAQ chemistry/transport models, both equipped with adaptive grid refinement capability, and the Daysmoke plume model for sub-grid scale treatment of the PB plumes.

Forest fire plumes are not well resolved in current regional-scale air quality models due to insufficient grid resolution. Our approach to this problem is to utilize a dynamic, solution-adaptive grid algorithm. We developed an adaptive grid version of CMAQ that has all the necessary functions for tracking the formation of ozone and secondary particulate matter. Increased grid resolution downwind from targeted sources improves the simulation of secondary pollutant formation and helps discern the impacts from other sources in the region.

The adaptive grid version produces more accurate solutions than the original, static grid CMAQ which uses comparable or even larger computational resources. It resolves emissions and chemical transformations much better and reduces numerical diffusion significantly (Figure 1). This will be demonstrated in applications to PB related smoke events.

(a)  (b)

Figure 1.  Comparison of PM_2.5 concentrations (mg m^-3) at Fort Benning, GA during a prescribed burn on April 9, 2008: (a) Standard CMAQ with 1.33 km grid resolution, and (b) Adaptive CMAQ with dynamically adapting mesh. Boxes shown are cutouts of the Fort Benning area from the model domain.

Biomass burning plumes are typically parameterized as power plant plumes in CMAQ. While this may be a good assumption for some cases, it is an oversimplification in many others, especially in the case of PB plumes. Daysmoke is a plume model specifically designed for PB plumes with the potential for use in other types of biomass burnings. We coupled Daysmoke with CMAQ for sub-grid scale resolution of PB plumes.

In general, sub-grid scale models produce accurate solutions but this accuracy is lost upon transfer to the grid model. In our approach, the sub-grid solution is expanded into Fourier series. The transfer occurs only when the frequencies supported by the grid can represent the solution reasonably well, assuring maximum benefits from sub-grid scale modeling. This will be demonstrated with applications to prescribed burns and comparisons to measurements at Fort Benning, Georgia.

In January and April 2009, field studies were conducted at Fort Benning to collect data for model evaluation. Ground-based mobile units tried to cover three 60 degree arc zones along the plume direction, respectively 1-3 km, 3-5 km, and 5-7 km downwind from the burn plots. Each unit measured PM_2.5 and CO in real time while moving around in its assigned zone according to wind shifts, measurement levels, and road availability. In addition, ceilometer measurements downwind of the burns and digital photogrammetry from nearby fire towers were conducted to determine the plume height, plume direction, and the number of updraft cores.

A preliminary evaluation of the modeling system with collected data is presented here. Air quality and related data from nearby regional monitors are also used in the evaluation. Lessons learned from this evaluation will help design the field campaign for the 2010 burning season. There will be another cycle of this data collection and model evaluation process after which the models will be refined and finalized. The resulting modeling system is expected to help land managers minimize the regional air quality impacts of their PB operations.