Monitoring Surface Reflectance and Fires using MODIS

1
Monitoring Surface Reflectance and Fires using
MODIS

• Vermote et al.
• University of Maryland/ Dept of Geography
• and
• NASA/GSFC Code 923

2
Solar Energy Paths
3
Different Types of Reflectors
diffuse reflector (lambertian)
Specular reflector (mirror)
Nearly Specular reflector (water)
nearly diffuse reflector
4
Sun glint as seen by MODIS
Gray level temperature image
5
MODIS data illustrating the hot-spot over dense
vegetation
6
Observation Geometry
7
STRE for non absorbing atmosphere and lambertian
ground
Ground reflectance ( albedo for Lambertian)
Atmospheric reflectance
Atmospheric Transmissions
Apparent reflectance at satellite level
Atmosphere spherical albedo
8
The composition of the atmosphere
9
Gaseous Absorption (H2O)
10
Modified SRTE to account for absorption
In case of a pure molecular atmosphere (no
aerosol) we can write
m is the air mass 1/cos(?s)1/cos(?v) Ugas is
the gas concentration
11
Final SRTE approximation
12
Water vapor effect for different sensors in the
near infrared
13
Atmospheric effect Vegetation 1/3
14
Atmospheric effect Vegetation 3/3
No absorption, Continental aerosol
15
Atmospheric effect Vegetation 3/3
Absorption tropical atmosphere, Continental
aerosol
16
Atmospheric effect Ocean 1/2
17
Atmospheric effect Ocean 2/2
18
MODIS aerosol optical depth validation
t470nm(MODIS)
Comparison of aerosol optical thickness retrieved
by MODIS blue channel with AERONET
19
(No Transcript)
20
Prototype correction details over China (monthly
composite)
No aerosol correction
Aerosol correction
21
MODIS 1km aerosol product
Uncorrected for aerosol
Corrected for aerosol
Aerosol optical thickness (1km resolution)
22
MODIS 1km aerosol product
Aerosol optical thickness (1km resolution)
Thermal anomaly RGB (4mic,1.6mic,2.1mic)
23
MODIS Granule over South Africa (Sept,13,2001,
845 to 850 GMT)
The surface reflectance algorithm uses internal
1km aerosol optical depth since collection 3
processing.
Corresponding aerosol optical thickness at 670nm
(0 black, 1.0 and above red) linear rainbow
scale. Clouds are in magenta, water bodies are
outlined in white.
RGB surface reflectance (corrected for aerosol)
RGB no correction for aerosol effect
24
Aerosol 1km retrieval validation is on-going
Comparison of 1km operational aerosol optical
thickness retrieved by MODIS blue channel (120
matches) with AERONET sun photometer measurements
during the March, April, May 2001 period
25
MODIS Surface reflectance validation (1/2)
Comparison of the surface reflectance derived
from ETM (using AERONET data) with the
operational MODIS surface reflectance product.
26
MODIS Surface reflectance validation (2/2)
Harvested corn
Yellow grass
Bare soil
27
Analysis will be extended/automated to the
following possible cases (92 possible)
28
Fires monitoring

• Two algorithms Contextual method based on AVHRR
  heritage, new method called ANO (anomaly) both
  relying on the signal in the middle infrared
  around 4.0mic.

Fire of 700K filling 0.1 of the pixel at
300K Apparent temperature at 4micron 302.3K App
arent temperature at 11micron 300.09K
29
MODIS Active Fire Validation Strategy

• Objective to determine product accuracy
• Initial product intercomparison with
• GOES, TRMM VIRS, DMSP, AVHRR
• Compare fires identified with MODIS to fires
  identified simultaneously in high-resolution
  ASTER imagery global sampling of fire regimes
• Field-based product validation by fire scientists
  and management agencies organized with
  international partners through GOFC/GOLD-Fire
• Mexico
• Russia
• Brazil
• Southern Africa (SAFNET)
• Southeast Asia (SEARIN)
• Australia

30
Active Fire Validation Collocating ASTER and
MODIS data
Aug 17 2001 0908 UTC 18.8S 19.9 E (NE Namibia)
White squares MODIS fire pixels
Burn scar
Fire fronts
R 2.16 µm G 1.65 µm B 0.56 µm
Smoke
31
MODIS GLOBAL NEAR REAL TIME FIRE MONITORING
http//rapidfire.sci.gsfc.nasa.gov
32
MODIS Fire Detections Quebec, Canada 07/06/02
33
Fire Animation July 2001-2002
34
Investigating Global Fire Patterns
35
TISI Temperature Independent Spectral Indices
(Becker and Li, 1990) Method adapted to MODIS
(Petitcolin and Vermote, RSE, 2002)
36

• Jornada, New Mexico (2001-132)

37
Jornada Site Land Surface Temperature, ASTER
(USDA-ARS method) versus MODIS
38
Jornada Site Emissivities ASTER (USDA-UARS
method) versus MODIS
39
Figure 1b RGB image of MODIS surface reflectance
over South Africa acquired on August 25,2000 at
855 GMT
Figure 1a RGB image of MODIS corrected
reflectance over South Africa acquired on August
25,2000 at 855 GMT
Red3.75mic Green1.6mic Blue2.1mic
3.75mic reflectance
40
Figure 2 Gray scale image of the surface
reflectance computed in band 20 (3.75mic) scaled
between 0.0 and 0.25 reflectance unit)
corresponding to Figure 1. The brightest feature
on the images correspond to fire (red boxes) and
clouds (white boxes). The green box is used to
establish a correlation between the surface
reflectance at 3.75mic and the middle infrared in
the subsequent slide.
41
The reflectance at 3.75mic (band 20) is highly
correlated with the reflectance in the shortwave
infrared (2.13mic) since the same processes (e.g
liquid water absorption) govern the signal at
both wavelengths for pixels located in the green
box of the previous slide. Clouds or fire pixel
will not follow this relation and therefore
appear as anomaly
42
Gray scale image of the reflectance anomaly
(scaled between 0.0 and 0.25 re- flectance unit)
observed in band 20 after correcting for surface
reflectance estimate based on band 7 (using the
previously defined relationship) and filtering
clouds. The bright correspond to active fires.
43
Fire detection on corrected reflectance
background. Fire are flagged using a simple
threshold on the reflectance anomaly (0.1
reflectance unit)
44
3.75mic reflectance
T31 (295K -310K)
AOT (0.0-1.0)
Analysis of MODIS aerosol product with the
thermal infrared show evidence of the cooling
effect of large smoke concentration (Robock, A.,
Science, 1988).
45
RGB composition of Landsat/ETM
data (R1.6mic,G0.87mic,Blue0.67mic) with the
overlay of the present MODIS fire detection
algorithm results (crosses). The white, red,
yellow green and blue crosses correspond
respectively to anomaly greater than 0.5, 0.4,
0.3, 0.2 and 0.1.
46
Fire detection validation with ASTER data
(2000-328-0920)
47
Regional active fire validation results - 7 ASTER
scenes, Africa 2000
ANO20
Algorithm Detection Rate
Number of 1km MODIS pixels containing one or more
30m ASTER fire pixels
48
Example Fire/Scars/Smoke monitoring (1/3) using
3.75 µm reflectance product (post-launch)
RGB surface reflectance
RGB corrected reflectance
Optical depth at 0.67mic (0-0.7)
Optical depth at 0.67mic (0-0.7)
RGB corrected reflectance
RGB surface reflectance
August 31, 2000
RGB 3.75mic,1.6mic,2.1mic
RGB 3.75mic,1.6mic,2.1mic
49
Example Fire/Scars/Smoke monitoring (2/3)
Optical depth at 0.67mic (0-0.7)
Optical depth at 0.67mic (0-0.7)
RGB corrected reflectance
RGB surface reflectance
RGB corrected reflectance
RGB surface reflectance
Sept 3, 2000
RGB 3.75mic,1.6mic,2.1mic
RGB 3.75mic,1.6mic,2.1mic
50
Example Fire/Scars/Smoke monitoring (3/3)
Optical depth at 0.67mic (0-0.7)
Optical depth at 0.67mic (0-0.7)
RGB corrected reflectance
RGB surface reflectance
RGB corrected reflectance
RGB surface reflectance
Sept 7, 2000
RGB 3.75mic,1.6mic,2.1mic
RGB 3.75mic,1.6mic,2.1mic
51
MODIS burned area product description

• Monthly geolocated product
• 500m pixel dimension
• defined in the Level 3 MODIS land tile grid
• 24002400 500m pixels, 1010o area
• approximately 11Mb per tile
• Pixel values
• n approximate Julian day of burning
• 0 not burned
• 999 could not perform mapping due to missing
  data or persistent cloud
• 991-998 QA codes

D. Roy
52
Time series of observed (open triangles) and
predicted (filled triangles) MODIS band 5
(1.240mm) land surface reflectance data and
corresponding Z-score values Single pixel
located near the Angolan-Namibian border, fire on
day 275
D. Roy
53
MODIS 500m burned area results, Angolan-Namibian
border August 21 September 18, 2000
54
MODIS 1km day night active fire results,
Angolan-Namibian border August 21 September
18, 2000
55
Emissions Model Outline
Area burned
FuelLoading
Combustion Completeness
Emission Factors
x
x
x
Emissions (g)
(ha)
(kg/ha)
(g/kgdry fuel)
()
Definition of terms PLITTER(litter/total
fuel) PGREEN(green grass/total grass) MCE
CO2/(COCO2) Total Fuel LitterGrass
(live,dead)twigs
Data Inputs/Source
Analyses

• MODIS Burned Area (1km, 500m), D.Roy
• JRC-GBA 2000 (1km)

Data prep. for model integration

• UVA fuel load(1km)
• S. Princes NPP?? (8km)

Data prep., PGREEN, PLITTER, Total Fuel
Emissions from grassland savannas
Model Output
Cgf(PGREEN) ZIBBEE
1km, 15-day

Combustion Completeness
Regional pyrogenic emissions
Cwf(PGREEN) ZIBBEE
UMD Tree Cover 10
Emissions from woodland savannas
CO2, CO, CH4, NMHC, PM2.5
Egf(MCEg)f(PGREEN)
Emission Factors
ZIBBEE
Unpub.
Ewf(MCEw)f(PLITTER)
Korontzi UMd
UMD Tree Cover 10
ZIBBEE, S92
Unpub.
56
Preliminary Results MODIS-derived regional
emissions
57
Preliminary Comparison of Emissions Estimates
58
Emission estimate formulas

• Emission Aburned x C x Eeff x Fload
• Aburned is the area burned (retrieved)
• C is the combustion completeness (guessed)
• Eeff Emission Efficiency (guessed)
• Fload is the fuel load (computed with Biomass
  model)
• Emission Efactor x

59
Martin J. Woosters results (in press)
60
Bi-spectral Infrared Detection (BIRD) a
technology demonstration

• Small satellite test of new infrared push broom
  sensor
• Sampling of fire events
• Payload
• a two-channel infrared Hot Spot Recognition
  Sensor system (HSRS)
• MIR3.4-4.2µm, TIR 8.5-9.3µm
• 370m IFOV, 190km
• a Wide-Angle Optoelectronic Stereo Scanner
  (WAOSS-B)
• VIS 600-670nm, NIR 840-900nm
• 185m IFOV, Swath 533km
• D. Oertel, DLR, Germany - www.dlr.de/bird

61
FIRE RADIATIVE ENERGY Comparisons of derivation
method developed for BIRD and for MODIS. Applied
to BIRD- Sydney Fires Jan 2002M. Wooster, (
Dept. Geography, Kings College, London)
62
MODIS
BIRD
Fire Images Middle Infrared
63
MODIS Fire detection
BIRD Fire detection
64
MODIS BIRD Fire Radiative Energy Comparison
MODIS BIRD

M. Wooster, ( Dept. Geography, Kings College,
London)
65
AQUA surface reflectance product 250m (V.3.1.9
in production since 7/12/02)
66
TRMM VIRS-Derived Diurnal Burning Cycle (July)
Giglio/Kendall/Justice Jan. 2001
67
TERRA Morning thermal anomaly
68
AQUA Afternoon, thermal anomaly
69
Heavy aerosol typing/detectionDust could be
easily confused with clouds at high optical
thickness middle/shortwave infrared reflectances
show different signature for dust versus clouds
which enable to recover those situations and
detect strong dust events.
Experimental dust storm mask, the area detected
as strong dust concentration are colored in
red-orange.
False RGB image (2.1mic Blue, 1.6mic Green,
3.75mic Red), the low clouds appear whiter than
the dust in this false RGB
RGB image (no aerosol correction) showing a dust
storm (yellow-white) over the Mediterranean sea
70
Heavy/Peculiar aerosol typingMt Etna eruption
data acquired by MODIS offers unique opportunity
to develop volcanic ash detection technique
middle/shortwave infrared reflectances show a
very specific signature of the volcanic ash plume.
RGB image of the Mt Etna volcanic ash plume. The
plume appears gray.
False RGB image of the Mt Etna volcanic ash
plume this time using 2.1mic (Blue) ,1.6mic
(Green) and 3.75mic reflectance (red)
71
CONCLUSIONS

• MODIS narrow land bands located outside strong
  gaseous absorption features and an operational
  aerosol retrieval method make it possible to
  produce surface reflectance at a global scale on
  a daily basis. This enables the inversion of
  directional surface properties and computation of
  climate model parameters (albedo).
• The wide spectral range of MODIS offers unique
  opportunity to study surface and aerosol
  properties (reflectance from 0.412mic to 4.0mic,
  emissivity at 8.55mic,11mic and 12mic).
• Validation of the reflectance product is on-going
  to produce a realistic estimate of error bars
  under representative atmospheric and geometric
  conditions.
• Fire detection validation/comparison of both
  algorithm is on-going.
• Need to demonstrate how the MODIS Burned Area
  contributes to National/Regional Emissions
  Estimates
• Fire Energy computation is an interesting
  development for consolidating/improving emissions
  estimates and facilitate validation by
  introducing a quantitative measurement.

72
References

• Vermote E.F., El Saleous N, Justice C, 2002,
  Atmospheric correction of the MODIS data in the
  visible to middle infrared First results, Remote
  Sensing Of Environment, (In press).
• Vermote, E. F. and Roy, D.P., Land Surface
  Hot-Spot observed by MODIS over Central Africa,
  International Journal of Remote Sensing, Cover
  Letter, (In Press)
• Petitcollin F. and Vermote E. F., Land Surface
  Reflectance, Emisivity and Temperature from MODIS
  Middle and Thermal Infrared data, Remote Sensing
  Of Environment, (In press).
• Roger J.C and Vermote E. F., 1998, A method to
  Retrieve the Reflectivity Signature at 3.75mm
  from AVHRR data, Remote Sensing of the
  Environment, 64103-114.
• Vermote E.F., El Saleous N.Z., Justice C.O.,
  Kaufman Y.J., Privette J., Remer L., Roger J.C.,
  Tanré D.,1997, Atmospheric correction of visible
  to middle infrared EOS-MODIS data over land
  surface, background, operational algorithm and
  validation, Journal of Geophysical Research, Vol.
  102,. d14, 17,131-17,141.
• Kaufman Y.J., Tanré D., Remer L., Vermote E.F.,
  Chu A., Holben B.N., 1997, Operational Remote
  Sensing of Tropospheric Aerosol Over the Land
  from EOS-MODIS, Journal of Geophysical Research,
  Vol.102, D14, 17,051-17,068.
• Vermote E.F., Tanré D., Deuzé J.L., Herman M.,
  Morcrette J.J., 1997, Second Simulation of the
  Satellite Signal in the Solar Spectrum an
  overview, IEEE Transactions on Geoscience and
  Remote Sensing, 35,3,675-686.