George Pouliot* and Thomas Pierce*

1
Developing Emission Inventories for Biomass
Burning for Real-time and Retrospective Modeling

• George Pouliot and Thomas Pierce
• Atmospheric Sciences Modeling Division/ARL/NOAA
• Tom Pace
• Office of Air Quality Planning and Standards/EPA
• 5th Annual CMAS Conference
• October 17, 2006
• In partnership with the USEPA, National Exposure
  Research Laboratory

2
Wildland Fire Emissions Selected
Acknowledgements

• NOAA/ARL-RTP (R. Mathur)
• NOAA/ARL-SS (R. Draxler)
• NOAA/NESDIS (S. Kondragunta, X. Zhang, M.
  Ruminski)
• NOAA/NWS (P. Davidson, J. McQueen)
• NOAA/Research (S. Fine)
• UNC-CEP (U. Shankar, J. Vukovich)

3
Biomass Burning Emissions

• GOAL (1) Develop a method to estimate
  real-time biomass burning emissions for the
  EPA/NOAA Air Quality forecast system (PM2.5 and
  Ozone)
• Key Issues
• Fire locations
• Fire size, fuel loading, and emission factors
• Forecasts of biomass burning
• Agricultural burning

4
Real-time Biomass Burning Emissions Questions

• What accuracy in space and time is needed?
• Is persistence sufficient? (We currently assume
  that current fires continue to burn for the next
  48 h.)
• Do we need a fire behavior model?
• Can we use a simple one-size-fits-all model?
• Can we combine different data sets and approaches
  to make a better forecast?

5
Real-time Biomass Burning Emissions

• Existing data sources and methods
• BlueSky (U.S. Forest Service)
• HySplit smoke plume estimates (NOAA/ARL)
• Limited chemistry and physics
• HMS simple approach (NOAA/EPA-AMD)
• One-size-fits-all
• Modified WF-Abba approach (NOAA-NESDIS)
• Satellite-derived fire detects and fuel loading
• Preliminary results for the last two will be
  discussed

6
One Size Fits All approach

• Adapt older HySplit assumption that all fire
  detects use the same emission factor
• Use SMOKE updates from the BlueSky-EM tool to
  create emissions for ETA/CMAQ system
• Use daily/near-real-time Hazard Mapping System
  (HMS) product from NOAA/NESDIS for fire detects
• Use the raw version of HMS product and not the
  HySplit version

7
One Size Fits All approach

• Other Criteria Pollutants
• Used 1996-2002 emission inventory for wildland
  fires
• Created average ratios of other criteria
  pollutants to PM2.5
• Derived emission factors for other pollutants

8
One Size Fits All approach Wildland Fire
Emission Factors (kg/ha)
Pollutant 7-year (1996-2002) Inventory derived
CO 2250
NH3 6.75
NOx 45
SO2 6.75
VOC 112.5
PM2.5 225
PM10 250
9
One Size Fits All approach

• Burn area for each fire set to 22.3 ha (based
  on analysis of EPA 2001 NEI dataset w/ total
  annual burn area and total number of fires). For
  operational forecasts, assume 16.7 ha/day (75 of
  total burn area).
• Heat output (used for plume rise) set to 725 x
  106 BTUs/day (based on an aggregated set of
  BlueSky simulations).
• Diurnal profile for emissions from the WRAP.
• PM2.5 emission factor set to 225 kg/ha (review of
  existing emission factors, which exhibit a wide
  range 20-800 kg/ha).
• 7 year average NEI factor range 187-279 kg/ha

10
ETA/CMAQ Test CaseJune 18-July 5, 2005
One Size Fits All approach

• 5x ETA/CMAQ system with PM version of CMAQ
• Cold start used for some species, as only the
  PM-3x version of system avbl for June 2005
• 12 km grid/national domain
• Major wildland fires in southwestern U.S.
• Cave Creek Complex Fire Second worst wildfire in
  the state of Arizona
• Fire started in June 21, 2005 by a lightning
  strike
• 243,950 acres burned

11
Cave Creek Complex Fire
News Video of Fire on 22 June 2005 North of
Carefree, AZ
Source http//www.abc15.com/gallery
12
Satellite View of Cave Creek Fire
The Moderate Resolution Imaging Spectroradiometer
(MODIS) on NASA's Terra satellite captured this
image of the fire on June 23, 2005, at 1150
a.m., local time Source http//www.nasa.gov/visio
n/earth/lookingatearth/Arizona_Wildifire06.23.05.h
tml
13
Fire Locations June 18-July 5, 2005 HMS and 209
Ground Reports
Legend RedHMS and Ground Report Match Green
Ground Report Only BlueHMS only
Cave Creek Fire
14
(No Transcript)
15
Difference in Max 8-hr Ozone for June 25
Forecast HMS Fires Case No Fire Case
16
Difference in 24-hr Average PM2.5 June 24
Forecast HMS Fires Case No Fire Case
17
Scatter Plots of Max 1-hr Ozone, Max 8-hr Ozone
and daily Mean PM2.5 for entire episode
Note only model-obs. pairs selected where we
detected a fire impact O3 (Fire-base)gt4ppb PM25
(Fire-base)gt 2ug/m3
18
Summary ofone size fits all approach

• Daily HMS product appears to reasonably capture
  large fires.
• A preliminary system using HMS fire detects has
  been tested successfully. A final report has
  been delivered by UNC-CEP.
• The fire emissions algorithm resulted in higher
  ozone and PM2.5 concentrations for a major fire
  event, which appear to agree more closely with
  observations than the base model.
• Additional tests and comparisons with NESDIS
  (post-WF-ABBA-approach) are underway.
• The preliminary system is ready for experimental
  real-time tests.
• Other work could include improved
  characterization of fire size, fuel loading,
  emission factors, and temporal profiles.

19
NOAA/NESDIS Satellite derived Emission estimates

• Developed by X. Zhang, S. Kondragunta, F. Kogan,
  J Tarpley, and W. Guo
• NOAA/NESDIS has developed a new algorithm to
  derive biomass burning emissions of PM2.5 from
  remotely-sensed fire products in near-real-time
  for regional and global air quality modeling
  applications

20
NOAA/NESDIS Satellite- derived Emission estimates

• Fuel loading
• Fraction of fuel consumed
• Emission factor
• Fire locations
• Fire size

21
Fuel Loading Database

• Uses maximum monthly MODIS Leaf Area Index (LAI)
  and allometric models that relate leaf foliage
  biomass with other biomass components in forests,
  shrubs, and grasses

Emission Factors

• Determined from a fuel moisture category using
  AVHRR Normalized Vegetation Index (NDVI) product

22
NOAA/NESDIS Satellite derived Emission estimates

• Work in progress
• Still refining method to estimate fire size from
  satellite
• Initial work so far is to look at fire locations
  and compare with ground reports
• Plan to test ETA/CMAQ for same 2005 episode

23
Fire Locations June 18-July 9, 2005 NESDIS and
209 Ground Reports
Legend RedNESDIS and Ground Report
Match GreenNESDIS only BlueGround only
24
Fire Locations June 18-July 9, 2005 NESDIS and
209 Ground Reports
Legend RedNESDIS and Ground Report
Match GreenNESDIS only BlueGround only
Cave Creek Fire
25
Biomass Burning Emissions Retrospective Modeling

• GOAL (2) Develop a method to estimate biomass
  burning emissions for the National Emissions
  Inventory for years gt 2002 (e.g. 2005)
• The 2002 NEI for biomass burning is the most
  comprehensive to date
• Many spent by RPOs to create this inventory
• Develop a method for biomass emission inventory
  that is better than pre-2002 methods and not as
  costly as 2002

26
Biomass Burning Emissions Retrospective Modeling

• What sources of data do we have for retrospective
  modeling?
• 209 Reports
• GOES satellite products
• MODIS satellite products

27
Biomass Burning Emissions Retrospective Modeling
Questions

• How much detail do we need in the inventory?
• Do we try to characterize only the largest fires
  accurately?
• What satellite data should we use?
• MODIS and/or GOES

28
Biomass Burning Emissions Retrospective Modeling
Questions

• How accurate are MODIS, GOES and 209 reports?
• See A. Soja et. al.
• "How well do satellite data quantify fire and
  enhance biomass burning emissions estimates?"

29
Biomass Burning Emissions Retrospective Modeling
Proposed Approach

• Combine MODIS fire detects and 209 reports to get
  fire locations
• Use land cover data to distinguish between ag
  burning, wildfires and Rx burning
• Get fuel loading from a national database (FCCS
  or NFDRS)
• Develop inventory for 2005
• Test methodology by redoing inventory for 2002
  and compare to 2002 NEI

30
Biomass Burning Emissions

• For forecasting air quality, we recommend a
  combination of available methods and data but,
  fire behavior modeling will be needed for true
  forecasts
• For retrospective air quality modeling,
  satellite-derived data offer much promise for
  capturing the temporal and spatial variation of
  large fires

31
Biomass Burning Emissions
Disclaimer The research presented here was
performed under the Memorandum of Understanding
between the U.S. Environmental Protection Agency
(EPA) and the U.S. Department of Commerces
National Oceanic and Atmospheric Administration
(NOAA) and under agreement number DW13921548.
Although it has been reviewed by EPA and NOAA and
approved for publication, it does not necessarily
reflect their views or policies.