Using MODIS fire count data as an interim solution for estimating biomass burning emission of aerosols and trace gases

1
Using MODIS fire count data as an interim
solution for estimating biomass burning emission
of aerosols and trace gases

• Mian Chin, Tom Kucsera, Louis Giglio, Thomas
  Diehl
• NASA Goddard Space Flight Center, U.S.A.

2
Emission, emission, emission

• Emission is one of the most important factors
  that determines the amount of aerosols and trace
  gases in the atmosphere
• The quality of global model simulations
  critically depends on the accuracy of emissions
  used in the model

3
Emissions in the GOCART model for aerosol
simulations (1)

• Fossil fuel/biofuel consumptions
• Emit SO2, BC, OC
• We currently use the IPCC 2000 emissions, based
  on energy use, population density, and technology
• We assume these emissions are relatively constant
  with some seasonal variations

• Volcanic/biogenic emissions
• Volcanic emission of SO2 based on the global
  volcanism database and TOMS SO2 index
• Ocean emission of DMS from ocean using empirical
  relationship between the winds and DMS seawater
  concentrations
• Biogenic emission of OC based on global inventory

4
Emissions in the GOCART model for aerosol
simulations (2)

• Dust and sea-salt emissions
• We use empirical relationships between emission
  and meteorological conditions
• Dust emission is a function of surface type,
  surface wetness, and wind speed
• Sea-salt emissions is a function of wind speed

• Biomass burning emissions
• We currently use the monthly averaged emission
  data based estimated based on the TRMM and ATSR
  fire data, MODIS burned area estimates, and dry
  mass burned (van der Werf et al., 2003, 2005)
• No daily variations is given

5
Challenges in estimating biomass burning emissions

• Biomass burning emission is highly variable with
  space and time
• It is difficult to use a climatology to model
  the biomass burning emission for a particular
  region at a certain time
• Only satellite data can provide global coverage
  of fire monitoring at real time, but converting
  the satellite fire data to biomass burning
  emission takes considerable efforts, making near
  real time simulation impossible
• These products are only available for monthly
  average which are not adequate for fires that
  last just a fraction of a month

6
Using MODIS fire counts for daily
near-real-time fire emission

• Here we explore the possibility of using MODIS
  fire counts (at 1-km2 pixel resolution) to model
  daily biomass burning emissions of aerosols and
  trace gases as an interim solution
• This methods can be used in aerosol forecast for
  mission support, in which the near real time fire
  counts can be incorporated into the model

7
Emission of aerosols and trace gases from fire

• Mass of tracer i (Mi) emitted from fire
• Mi A B C Ei
• A Area burned
• B Biomass density (or fuel load)
• C Completeness of burning (or burning
  efficiency)
• Ei Emission factor of tracer i

Dry mass burned
8
Area burned (A)

• This is probably the most difficult quantity to
  determine on daily bases
• Currently we assume that each 1-km2 MODIS fire
  pixel is filled with fire, such that the burned
  area within a model gridbox (1.25ºlong x 1ºlat or
  2.5ºx2º) total number of 1-km2 fire pixel
  within the box

Terra-MODIS fire counts 20040701
9
Biomass density (B) and Completeness of burning
(C)
Based on Hoelzemann et al., JGR 2004
10
Emission factors for tracers (Ei)
Emission factors (g tracer / kg dry matter) for
selected tracers from Andreae and Merlet, GBC
2001

• Ecosystem-dependent
• Burning stage-dependent
• Also depending on temperature, moisture, etc.
• Large uncertainties

Savana/ Grassland Tropical Forest Extratropical forest Biofuel Agriculture Residual
BC 0.480.18 0.660.31 0.560.19 0.590.37 0.690.13
OC 3.41.4 5.21.5 8.6 9.7 4.01.2 3.3
SO2 0.350.16 0.570.23 1.0 0.270.3 0.4
CO 6520 10420 10737 7831 9284
CO2 161395 158090 1569131 155095 1515177
11
Example BC biomass burning emission used in the
GOCART model
(MBC ABCEBC )
BC biomass burning emission July 1 2004
12
GOCART model simulation of aerosols
Example Total aerosol optical thickness at
550 nm, July 1 2004 (including biomass burning,
anthropogenic, dust, and sea-salt emissions)
Comprehensive evaluation with satellite and other
data are in progress
13
GOCART model simulation of aerosols
Total aerosol optical thickness at 550 nm, July
2004 (including biomass burning, anthropogenic,
dust, and sea-salt emissions)
GOCART
MODIS
Comprehensive evaluation with satellite and other
data are in progress
14
Comparison with AEORNET AOT over North America
during INTEX-A
AERONET
Total
Sulfate
Dust
OC
BC
Sea-salt
15
Future plan

• Using MODIS fire counts for daily emissions
• Better estimates of area burned
• Using the relationship between Terra-MODIS fire
  counts and area burned at different regions
  (Giglio et al., 2005)
• Using combined Terra- and Aqua-MODIS fire counts
• Better estimates of seasonal variations of dry
  mass burned
• Linking MODIS fire counts to the monthly averaged
  dry mass burned estimates (van der Werf et al.,
  2005)
• Using aerosol emission derived from MODIS fire
  radiative energy and aerosol optical depth
  (Ichoku and Kaufman)

16
Acknowledgment

• MODIS fire team for fire counts data
• MODIS aerosol team for providing aerosol data
  (special thanks to Rob Levy)
• AERONET team
• Funding from NASA EOS

17
GOCART model simulation of aerosols (Mi
ABCEi)
Example MODIS fire counts and BC biomass burning
emission, July 1 2004
BC biomass burning emission July 2004
MODIS (Terra) fire counts July 2004