Transcript Document

OTHER SATELLITE DATASETS
1. Ocean Biology
2. Vegetation Cover
3. Fire Counts
OCEAN BIOLOGY FROM SPACE
Links to Atmospheric Composition:
1. Source of organics
2. Related to nutrient deposition?
Chlorophyll is a marker of ocean
productivity because
phytoplankton need chlorophyll
(and sunlight) to convert
nutrients into plant material
Plankton includes microscopic
organisms as well as large
organisms such as jellyfish
OCEAN COLOUR
= light intensity at visible wavelengths
True Colour
Pure Water: deep-blue/black
Productive Water: blue-green
 chlorophyll absorbs red light, but reflects blue & yellow light = GREEN
OCEAN COLOUR SENSORS
CZCS: Coastal Zone Color Scanner
Nimbus 7(1978 - 1986)
SeaWifs: Sea-viewing Wide Field-ofview Sensor
SeaCare (1997-
MODIS: Moderate Resolution Imaging
Spectrometer
Terra (1999-)
Aqua (2002-)
RETRIEVAL OF CHLOROPHYLL
1. “Atmospheric Correction”: remove atmospheric component to backscattered
radiance (including Rayleigh and multiple-scattering from aerosol)
2. Convert water-leaving radiance to chlorophyll concentrations.
Water-leaving reflectance ratio @
443 nm to 550 nm
* note correction for white caps can be complicated
SeaWiFS
uses 490nm
& 555 nm
“Universal” relationship
CZSC
wavelengths
[Gordon et al., 1997]
SEAWIFS CHLOROPHYLL-A
Annual mean (1997-2004)
The plankton populations are dependent on a variety of factors, including ocean
currents, temperature, availability of nutrients, amount of sunlight, and ocean depth
SPRINGTIME CHLOROPHYLL (MODIS & SEAWIFS)
Bloom in the Northern oceans is the result of cold winter waters which encourage
mixing from the deep (and hence the presence of nutrients at the surface) + the
springtime sunlight.
VEGETATION COVER
Links to Atmospheric Composition:
1. Natural emission source
2. Biomass burning fuel load
3. Surface roughness (deposition)
Instrument requirements include: spectral coverage (visible) and small footprint
Examples:
AVHRR
MODIS
SPOT
NORMALIZED DIFFERENCE VEGETATION INDEX (NDVI)
Plants absorb in the visible (PAR) and scatter in the
NIR (where energy isn’t useful for organic synthesis)
NIR  Vis
NDVI 
NIR  Vis
HEALTHY
SPARSE
NDVI is a QUALITATIVE measure of vegetation density
www.earthobservatory.nasa.gov
NDVI FROM DIFFERENT SENSORS
653 nm vs 865 nm
628 nm vs. 850 nm (broader bands)
http://geog.utm.utoronto.ca/yuhonghe/Research.html
MAPS OF NDVI…
AVHRR
NDVI can also
be used as an
indicator for
drought
AUG
93
www.earthobservatory.nasa.gov
LEAF AREA INDEX (LAI)
LAI=2.25
LAI=4.75
leaf area
LAI 
land area
Soybean
The relationship
between LAI & NDVI
is not unique
(depends on
veg/land type)
LAI is related to
primary
production (PP)
LAI is a QUANTITATIVE measure of vegetation density
LAI RETRIEVALS
Retrieval (like ocean biology) involves removing an atmospheric component and then
looking at the spectrally resolved light radiance using a LUT based on a canopy
radiation model. In this case, biomes are assumed to constrain architecture of individual
trees (transmission and reflectance of light through the canopy depends on this)
[Myeni et al., 2002]
LAI FROM AVHRR (1981-1991)
March 1991
ANIMATION:
http://earthobservatory.nasa.gov/Features/LAI/Images/lai_640by480_30fps.mov
CANADIAN VEGETATION COVER FROM SPOT-4
http://ess.nrcan.gc.ca/ercc-rrcc/proj3/theme6/index_e.php
FIRE COUNTS
Links to Atmospheric Composition:
1. LARGE emission source (often in remote regions)
Fires smoulder (500K) or flame (1200K) with very strong IR emissions
CONTEXTUAL ALGORITHMS FOR FIRE DETECTION
Compare pixel to its nearest neighbours (based on brightness temperatures in
Channels 3 and 4 – 3.7 m and 10.8 m):
As developed for AVHRR:
1. Identify pixels that MIGHT be a fire: T3 > 311 K, T3-T4 > 8 K
2. Remove pixels with high reflectance: 2<20%
3. Compare pixel in question to its neighbours:
Neighbours define a background (b)
Fire identified when: T3-[T3b+2T3b] > 3 K
T34 > T34b+2T34b
[Flasse and Ceccato, 1997]
Advantage: adaptive algorithm that doesn’t depend on the environment.
Earlier threshold techniques would require (for example) different threshold of
response over a cool forest vs. a dry savannah.
AVHRR FIRE DETECTION
[Stroppiana et al., 2000; Dwyer et al., 2000]
MODIS FIRE DETECTION
Based on similar principles used with AVHRR and uses similar wavelengths (4
m and 11 m).
There are two 4 m channels with different
saturation T (331K and 500K).
Multiple additional channels are used for cloud
masking, rejection of false alarms
Absolute thresholds are also applied T4>360K
(320 K at night)
September 4, 2001
[Giglio et al., 2003]
[Justice et al., 2002]
ANIMATION:
http://earthobservatory.nasa.gov/GlobalMaps/view.php?d1=MOD14A1_M_FIRE#
MODIS BURNED AREA PRODUCT
Burned area is not necessarily directly proportional to fire counts. We would need
to account for where the fire is burning and with what intensity.
 Need to link the change in vegetation with fire activity
Ai   fi
=effective fire area per fire pixel
  f (NDVIi )
  f (% tree coveri )
  f (fire sizei , % tree coveri , % herbaceous coveri )
High uncertainties on these products, tough to validate.
[Giglio et al., 2006]
FIRE RADIATIVE POWER
FRP=The instantaneous rate of energy released from combustion (megawatts)
Estimated from empirical relationship [Kaufman et al., 1998]:
FRP  (4.34 10-19 )(T48  T48 )A pix
T4=4 m brightness T
Apix=total area (km2) of the pixel
Integral over lifetime of fire (Fire Radiative Energy) should be proportional to
total mass of fuel burned
An example from geostationary SEVERI
Over Angola in 2003
Feb 7, 2004
http://www.eumetsat.int/Home/Main/Access_to_Data/Meteosat_Meteorological_Products/Product_List/SP_1145433873287?l=en
MODIS RAPID RESPONSE PROVIDES GLOBAL NRT
FIRE PRODUCTS
04/01/10 – 04/10/10
http://rapidfire.sci.gsfc.nasa.gov/firemaps/
Rapid Response Imagery is 10 day composite of MODIS Terra & Aqua detected fire
locations (daily data is available).