Accuracy Assessment

1

• Estimating Biomass Burning Fire Radiative Energy
  Using MODIS

Evan A. Ellicott1, Eric F. Vermote1,2, Louis
Giglio3, Gareth Roberts4 1Department of
Geography, University of Maryland,
USA 2NASA/GSFC 3Science Systems and
Applications, Inc., Greenbelt, USA 4Kings
College, London, England
Estimating the MODIS FRP Diurnal Cycle
FRE Results
We used Europes Meteosat geostationary satellite
sensor, SEVIRI, and NASAs Tropical Rainfall
Monitoring Mission (TRMM) satellite sensor, VIRS,
to obtain greater FRP retrieval frequency in
order to characterize the fire diurnal cycle.
These sensors offer higher temporal resolution
than MODIS (e.g. SEVIRI sensor provides 15-minute
temporal resolution), but with limited spatial
coverage.
MODIS and Fire Radiative Energy
The goal of this project was to estimate the fire
radiative energy emitted from biomass burning
using NASAs MODIS sensor. Onboard the
sun-synchronous polar-orbiting satellites Terra
and Aqua, MODIS acquires four observations of
nearly the entire Earth daily at 1030 and 2230
(Terra) and 0130 and 1330 (Aqua), equatorial
local time. Although global in spatial scale,
the MODIS sensors do not provide adequate
temporal detail to characterize the fire diurnal
cycle the key component to quantifying the fire
radiative energy (FRE) released. Therefore we
developed a methodology to parameterize the fire
cycle from the MODIS fire observations using a
function of the variation in fire intensity
between the Terra-MODIS and Aqua-MODIS retrievals
and curve fitting of higher temporal fire
observations made by other satellite sensors.

• In addition, high latitude MODIS retrievals were
  also incorporated because of the large number of
  overpasses made a function of its polar orbit.
• Regions were chosen to characterize the fire
  diurnal cycle (Fig. 2) .

Fig 2 Regions to investigate the fire diurnal
cycle
Fig. 8 Estimated mean MODIS FRE for 2001 - 2007.
Spatial resolution is 0.5. Temporal resolution
is based on total FRE values per 30-day increment.

• A Gaussian function (Eq. 2) was used to fit the
  SEVIRI, TRMM, and MODIS regional FRP retrievals
  (Fig. 3).

Accuracy Assessment
Eq. 2
What is Fire Radiative Energy?
Evaluation of our FRE estimate shows a slight
overestimation by MODIS (Fig. 9). When SEVIRI is
adjusted for under-detection (relative to MODIS)
and cloud obscuration Roberts Wooster, 2008
the relationship is still good, but MODIS FRE is
now underestimated (Fig. 10).
Where t is time, b is the lower boundary of FRP
retrievals, peak is the maximum FRP value, h is
the hour corresponding with peak, and ? is the
standard deviation.

• Fire Radiative Energy (FRE) is the integral of
  the rate of energy released from combustion.
• Units are in megajoules (MJ)
• The instantaneous rate of energy is the Fire
  Radiative Power (FRP) measured in megawatts (MW,
  or MJ-sec)

Fig 3 Normalized FRP from SEVIRI observations,
binned in hourly increments, and fitted using a
Gaussian function. Observations made within a
10?x10? African site centered at 5?N 15?E
(circled in Fig. 2).
Ratios between Terra and Aqua (T/A) monthly MODIS
FRP observations (Fig. 4) were used to
parameterize the Gaussian function. Our
contention is that the variation in the T/A ratio
can serve as a proxy for the diurnal cycle
trajectory. The relationship between T/A ratios
and Gaussian function parameters (h, ?, b) from
SEVIRI, TRMM, and MODIS data was fitted for each
of the above regions (Figs. 5-7). Modeled
parameters were then used to estimate the diurnal
cycle of MODIS FRP. MODIS FRP values for each
site were then integrated to calculate the FRE
(Fig. 8).
Applications of FRE
Fig. 9 Comparison of SEVIRI and estimated MODIS
FRE for southern Africa, July 2004. Observations
made at 2.5? resolution (n 49)

• The approach to quantify the rate of fire
  radiative energy emitted during biomass
  combustion was first developed by Kaufman et al.
  1996 and later refined by Wooster et al.
  2005.
• Integrating FRP produces FRE which can be used
  to estimate the total mass of fuel combusted
  (Fig. 1)
• With an emission coefficient (Ec), the total
  fire emissions can be calculated (Eq. 1).

Fig. 10 Monthly comparison of adjusted SEVIRI
and MODIS FRE for Africa (2004). MODIS is only
adjusted for transmission.
Conclusions
We have demonstrated an approach to estimate
discrete MODIS FRP retrievals and calculate FRE.
The methodology described is a significant
contribution to satellite based fire science. To
date, the calculation of FRE from MODIS FRP has
not been achieved and therefore this estimate is
a first of its kind. Accuracy assessment
indicates that our approach produces realistic
results.
Fig. 4 Terra Aqua FRP for 60 months from the
region used in Figure 3.
Fig. 5 b parameter vs. T/A ratio
REFERENCES Kaufman, Y. J. et al., 1996.
Relationship between remotely sensed fire
intensity and rate of emission of smoke SCAR-C
experiment, Global Biomass Burning,
685696. Roberts, G. J., and M. J. Wooster
(2008), Fire detection and fire characterization
over Africa using Meteosat SEVIRI, Ieee
Transactions on Geoscience and Remote Sensing,
46(4), 1200-1218Doi 10.1109/Tgrs.2008.915751.Woost
er, Wooster, M. J., G. Roberts, G. L. W. Perry,
and Y. J. Kaufman (2005), Retrieval of biomass
combustion rates and totals from fire radiative
power observations FRP derivation and
calibration relationships between biomass
consumption and fire radiative energy release,
Journal of Geophysical Research-Atmospheres,
110(D24), D24311, doi 10.1029/2005jd006318.
Fig. 1 Biomass combustion rate
Eq. 1
Fig. 6 c parameter vs. T/A ratio.
Fig. 7 d parameter vs. T/A ratio.