1. Development of Wildland Smoke Marker Emissions Maps for the Conterminous United States Leigh Patterson 06/15/09 M.S. Defense

2. Acknowledgements Advisor: Dr. Jeff Collett Committee: Drs. Kreidenweis, Schichtel, and Rocca FLAME Partners: Cyle Wold, Dr. Wei Min Hao, Dr. Bill Malm Sample Analysis: Amy Sullivan, Mandy Holden Funding: Joint Fire Science Program, National Park Service, American Meteorological Society Friends and family

3. Outline Introduction & motivation Fire Lab at Missoula Experiment (FLAME) Relationship between smoke marker emissions and vegetation type Fuel Characteristic Classification System Smoke marker emissions maps Biomass burning carbon apportionment

4. Fire Impacts Radiative budget ▫OC – reflective ▫EC – absorptive Visibility ▫Regulated by Clean Air Act  Wildfires – natural  Prescribed fires – manmade Health effects ▫Ultra-fine particles provoke alveolar inflammation (Seaton 1995) ▫During a fire episode in California, 117 hospital admissions for smoke reactions (Shusterman et. al., 1993)

5. Total TC – IMPROVE measurements TC enhancement Effects of fires on air pollution Debell et. al., 2006Park et. al, 2007 TC, µ g m -3

6. What are smoke markers? Chemical compounds used to fingerprint smoke ▫K +  Produced from loss of potassium in plant matter (Arianoutsou and Margaris, 1981) ▫Anhydrosugars  Includes levoglucosan, mannosan, and galactosan  Produced from combustion of cellulose and hemicellulose Marker criteria: unique, constant, inert, and measurable (Khalil and Rasmussen, 2003)

7. Why do we need accurate source profiles? MM-CMB models use profiles to apportion wildfire smoke ▫CMB models attempt to apportion 100% of the PM to various sources ▫If one source is incorrectly apportioned, the apportionment of other sources will be misestimated Correct geographic profiles are most important to determine wildfire smoke contribution (Sheesley et. al., 2007)

8. Why do we need accurate source profiles? MM-CMB models use profiles to apportion wildfire smoke ▫CMB models attempt to apportion 100% of the PM to various sources ▫If one source is incorrectly apportioned, the apportionment of other sources will be misestimated Correct geographic profiles are most important to determine wildfire smoke contribution (Sheesley et. al., 2007)

9. FLAME Study Burned over 33 fuels in over 100 burns in two campaigns in a burn chamber in Missoula, MT Mostly single component burns Measured physical, optical and chemical properties Picture courtesy of Gavin McMeeking

10. Vegetation Relationship Levoglucosan/OCCellulose dry mass Hoch, 2007 Sullivan et. al., 2008

11. Vegetation composition & anhydrosugar relationship Levoglucosan & Cellulose Mannosan/Galactosan & Hemicellulose

14. Vegetation source profiles FLAME groupsNon-FLAME groups Separate vegetation groups Source profile = median profile of each group ▫Averages have outlier problems Problem: The FLAME study does not sample all different types of vegetation in the U.S. Identify source profiles in lit Take median of each study Average the medians to calculate final source profile

15. Fuel Characteristic Classification System Fully descriptive fuelbed model Defines 113 fuelbeds across U.S. Assigns characteristics for six strata in each fuelbed Maps fuelbeds across conterminous U.S. with 1 km resolution Ottmar et. al., 2007

16. Emissions Algorithm B j = fuel loading CE j = combustion efficiency e ij = emissions factor (source profile) Emission i = Emissions i = Canopy (3 stories): Hardwood branches Softwood branches Hardwood leaves Softwood needles Shrubs Shrub branches Shrub leaves Non-woody vegetation Grasses Litter Duff

17. Source Profiles Litter Type Assigned by FCCSEmissions Type Assigned NeedlesSoftwood Needles Broadleaf DeciduousHardwood Leaves Broadleaf EvergreenShrub Leaves Palm FrondSaw Palmetto Leaves Grasses Litter: 5 different categories are multiplied with weightings Duffs are assigned same source profile as litters

18. Duff: Calculated vs. Measured Calculated levoglucosan yields match measured Mannosan and galactosan are underestimated K+ is grossly overestimated ▫Burn conditions ▫Correction factor of 2.65 is applied

20. Spatial Distribution of fuelbeds

21. Levoglucosan/OC Map

23. Mannosan/OC Map

25. Galactosan/OC Map

27. K+/OC Map

29. Can a national source profile apply? Levoglucosan/OCMannosan/OC

30. Source Apportionment Samples from IMPROVE site in Rocky Mountain National Park ▫Weekly: 06/28/05 – 08/16/05 Attempts to apportion carbon resulted in overestimation of biomass burning carbon concentrations Carbon concentrations and biomass burning carbon concentration courtesy of Amanda Holden

31. Simple Fire Model Fires identified by MODIS thermal anomalies Seven 48 hour HYSPLIT back trajectories were calculated Source profiles of fires within 2 degrees latitude and 2 degrees longitude of a trajectory were averaged No accounting for fire size, distance to sampler, or dispersion

32. New Source Apportionment Estimates using levoglucosan source profile improved K+ and galactosan source profiles yield reasonable results Mannosan too high Uncertainty can be assessed

33. New Source Apportionment Estimates using levoglucosan source profile improved K+ and galactosan source profiles yield reasonable results Mannosan too high Uncertainty can be assessed

34. In summary Vegetation composition & anhydrosugar relationship Differences in means of different vegetation types Source profiles ▫Vegetation ▫Fuelbed Fuelbed profile maps Source apportionment

35. Any Q uestions ?

36. Kuo et. al. 2008 Figure

37. Grouping vegetation types Sullivan et. al. – 6 groups ▫Grasses ▫Leaves ▫Needles ▫Branches ▫Straws ▫Duffs Shrub Leaves Hardwood Leaves Shrub Branches Softwood Branches

38. Physical Fuel Loadings Canopy: ▫Split into hardwoods and softwoods  Hardwoods: 84% wood, 16% leaves (Wiedenmyer et. al., 2006)  Softwoods: 79% wood, 21% needles (Wiedenmyer et. al., 2006) Shrub: ▫39% wood, 61% leaves (Wiedenmyer et. al., 2006)

39. Smoke Contribution Canopy: ▫Combustion efficiencies: 30% for wood, 90% for leaves/needles (Wiedenmyer et. al., 2006)  Hardwoods: 64% wood, 36% leaves  Softwoods: 56% wood, 44% needles Shrub: ▫Combustion efficiency: 30% for wood, e (-.013*TCP) for leaves (Wiedenmyer et. al., 2006)  For 50% shrub coverage: 27% wood, 73% leaves

40. Combustion efficiencies Canopy: Total trees available to fire depends on fire characteristics. Combustion efficiency: 0.9 for leaves, 0.3 for wood (Wiedenmyer et. al. 2006) Shrub: 0.3 for wood, CE = exp(-0.013*TCP) for leaves (Wiedenmyer et. al. 2006) Non-woody vegetation: 0.98 (Wiedenmyer et. al.,2006) Litter-lichen-moss: 1 (Reinhardt et. al. 2003) Ground fuels: CE = (26.1 – 0.225 * DM + 0.0417 * DEPTH)/DEPTH (Brown et. al. 1985)

41. Source Profiles Sheesley et. al. 2007