Comprehensive Particulate Matter Modeling: A One Atmosphere Approach

1
Forecasting the Impacts of Wildland Fires
Yongtao Hu1, William Jackson2, M. Talat Odman1
and Armistead G. Russell1 1School of Civil and
Environmental Engineering, Georgia Institute of
Technology, Atlanta, Georgia 2USDA Forest
Service, Asheville, North Carolina
Presented at the 6th Annual CMAS Conference,
October 2nd, 2007
2
Wild land Fires and Air Quality
3
Hi-Res Air Quality Forecasting SystemServing
Metro-Atlanta Area since 2006
4
Hi-Res Modeling Domains
4-km
12-km
36-km
5
Hi-Res Cycle
What could be done?

• Hi-Res cycle allows sufficient time for extra
  efforts.

• For prescribed burning, the air quality forecast
  ahead of the actual igniting would help plan and
  conduct burns.
• For existing/ongoing wildfires, the air quality
  forecast would warn people to avoid unhealthy air
  exposures at the following days.

6
Estimate Emissions of Potential Fires

• Using models the Fire Emission Production
  Simulator (FEPS) and the Consume 3.0
  (http//www.fs.fed.us/pnw/fera/research/smoke/cons
  ume/index.shtml )
• Prescribed fire collect pre-burning information
  from the burning plans prepared in advance.
• acreage of planned burning area, approximate
  locations, fuel load descriptions, igniting
  method and operation schedules
• Existing/ongoing wild fire determine the most
  likely fire locations on the following days
  according to the analysis of forecast
  meteorological conditions combined with the
  information on previous days burning locations.
• Then collect and estimate other fire information
  approximate acreage of burning area, fuel
  consumption and expected fire temperatures
• Allocate estimated potential fire emissions to
  the corresponding Hi-Res grid cells according to
  the geographical information.

7
Wild-Land-Impacts on Air Quality

• One way to calculate the air quality impacts of a
  fire is to run two simulations
• Run (1) typical emissions default in Hi-Res
• Run (2) estimated potential fire emissions
  added in
• and to take their difference
• Impact Air Quality (2) Air Quality (1).

• A more efficient way is to estimate the
  contribution of the fires by calculating
  emission sensitivities with the Decoupled Direct
  Method (DDM) provided by the Hi-Res system.
• Requires a single model run with potential fire
  emissions added in.

8
Application to prescribed fire forecast and
hindcast the February 28th, 2007 episode

• Forecast to test the predictive capability of
  this system.
• Forecast meteorology
• Emissions estimated from pre-burning information
• Hindcast to identify key weaknesses in the
  system.
• Re-analysis data (through FDDA) to predict the
  meteorology
• Post-burning information to estimate emissions

9
Smoke Detected by Geostationary
Satellite(115-145 pm EST on February 28th,
2007 )
10
Ambient Monitoring and Prescribed Burning Sites
11
Hourly PM2.5 Mass
12
Hourly Ozone Concentrations
13
PM Impact of the Oconee NF and Piedmont NWR Fires
14
Observed and Predicted Max. Concentration and
Predicted Max. Impact from the Fires within
Atlanta Urban Area

• Sensitivity analysis has shown that observed
  ozone peaks can only be reached at 5 times
  typical biogenic VOC emissions from burning
  areas. Bursts of fire-induced isoprenoid
  (isoprene and monoterpenes) emissions are
  reported in the literature.

15
Organic Matter to PM2.5 Ratios

• Increased VOC emissions also make up part of
  missing secondary organic aerosol (SOA).
  Evaporation and re-condensation of leaf surface
  wax may be another source of SOA as suggested by
  GC/MS analysis. Also background primary OM from
  other fires missing in the typical inventory.

16
1-hr Exposures to Ambient PM2.5
17
Forecast, Hindcast and Observed Plumes
18
Application to wild fires the May 18th-23th,
2007 episode of Georgia-Florida wildfires
MODIS aerosol optical depth (AOD) on May 21st (L)
and 22nd (R), 2007
Observed Hourly PM2.5
G-F fires plume reached Atlanta after long-range
transport through Alabama under easterly winds on
21st that turned to westerly on 22nd.
19
Wildfire Impacts on Hourly PM25
Simulated period May 18 23, 2007 Preliminary
Results on May 22

• Possible reasons could be the absence of the
  surface thermal changes induced by the fire from
  the meteorological model and the coarse vertical
  resolution in CMAQ above 1-km from the ground.

20
Summary

• We have developed and tested a modeling system to
  forecast wildland-fire-impacts on air quality in
  Atlanta, Georgia.
• The application to forecast the prescribed fires
  on February 28, 2007 was successful and indicates
  that the fires could be reasonably well estimated
  using the system. The forecast predictions are
  in good agreement with observations, though the
  hindcast improves significantly on timing and
  location.
• More efforts are needed to improve the capability
  of the system to simulate wildfires.