Effects of spectral response function on surface reflectance and NDVI measured with moderate resolution satellite sensors Extension to AVHRR NOAA-17, 18 and METOP-A


Alexander, P.Trishchenko

Remote Sensing of Environment 113(2): 335-341

2009


This work extends the previous study of Trishchenko et al. [Trishchenko, A. P., Cihlar, J., & Li, Z. (2002). Effects of spectral response function on surface reflectance and NDVI measured with moderate resolution satellite sensors. Remote Sensing of Environment 81 (1), 1–18] that analyzed the spectral response function (SRF) effect for the Advanced Very High Resolution Radiometer (AVHRR) onboard the NOAA satellites NOAA-6 to NOAA-16 as well as the Moderate Resolution Imaging Spectroradiometer (MODIS), the VEGETATION sensor (VGT) and the Global Imager (GLI). The developed approach is now applied to cover three new AVHRR sensors launched in recent years on NOAA-17, 18, and METOP-A platforms. As in the previous study, the results are provided relative to the reference sensor AVHRR NOAA-9. The differences in reflectance among these three radiometers relative to the AVHRR NOAA-9 are similar to each other and range from − 0.015 to 0.015 (− 20% to + 2% relative) for visible (red) channel, and from − 0.03 to 0.02 (− 5% to 5%) for the near infrared (NIR) channel. The absolute change in the Normalized Difference Vegetation Index (NDVI) ranged from − 0.03 to + 0.06. Due to systematic biases of the visible channels toward smaller values and the NIR channels toward slightly larger values, the overall systematic biases for NDVI are positive. The polynomial approximations are provided for the bulk spectral correction with respect to the AVHRR NOAA-9 for consistency with previous study. Analysis was also conducted for the SRF effect only among the AVHRR-3 type of radiometer on NOAA-15, 16, 17, 18 and METOP-A using AVHRR NOAA-18 as a reference. The results show more consistency between sensors with typical correction being under 5% (or 0.01 in absolute values). The AVHRR METOP-A reveals the most different behavior among the AVHRR-3 group with generally positive bias for visible channel (up to + 5%, relative), slightly negative bias for the NIR channel (1%–2% relative), and negative NDVI bias (− 0.02 to + 0.005). Polynomial corrections are also suggested for normalization of AVHRR on NOAA-15, 16, 17 and METOP-A to AVHRR NOAA-18.

Remote
Sensing
of
Environment
113
(2009)
335-341
ELSEVIER
Contents
lists
available
at
ScienceDirect
Remote
Sensing
of
Environment
journal
homepage:
www.elsevier.com/locate/rse
RcinoteScnsing
Envinihrry..
-rat
-
0
Effects
of
spectral
response
function
on
surface
reflectance
and
NDVI
measured
with
moderate
resolution
satellite
sensors:
Extension
to
AVHRR
NOAA-17,
18
and
METOP-A
Alexander
P.
Trishchenko
*
Canada
Centre
for
Remote
Sensing
Natural
Resources
Canada,
Ottawa,
Ontario,
Canada
KM
0Y7
ARTICLE INFO
ABSTRACT
Article
history:
Received
22
May
2008
Received
in
revised
form
29
September
2008
Accepted
4
October
2008
Keywords:
Satellite
Surface
reflectance
AVHRR
NOAA
NDVI
Spectral
response
function
Spectral
correction
Global
change
detection
This
work
extends
the
previous
study
of
Trishchenko
et
al.
[Trishchenko,
A.
P.,
Cihlar,
J.,
&
Li,
Z.
(2002).
Effects
of
spectral
response
function
on
surface
reflectance
and
NDVI
measured
with
moderate
resolution
satellite
sensors.
Remote
Sensing
of
Environment
81
(1),1-18]
that
analyzed
the
spectral
response
function
(SRF)
effect
for
the
Advanced
Very
High
Resolution
Radiometer
(AVHRR)
onboard
the
NOAA
satellites
NOAA-6
to
NOAA-
16
as
well
as
the
Moderate
Resolution
Imaging
Spectroradiometer
(MODIS),
the
VEGETATION
sensor
(VGT)
and
the
Global
Imager
(GLI).
The
developed
approach
is
now
applied
to
cover
three new
AVHRR
sensors
launched
in
recent
years
on
NOAA-17,
18,
and
METOP-A
platforms.
As
in
the
previous
study,
the
results
are
provided
relative
to
the
reference
sensor
AVHRR
NOAA-9.
The
differences
in
reflectance
among
these
three
radiometers
relative
to
the
AVHRR
NOAA-9
are
similar
to
each
other
and
range
from
-0.015
to
0.015
(-20%
to
+2%
relative)
for
visible
(red)
channel,
and
from
-0.03
to
0.02
(-5%
to
5%)
for
the
near
infrared
(NIR)
channel.
The
absolute change
in
the
Normalized
Difference
Vegetation
Index
(NDVI)
ranged
from
-0.03
to
+0.06.
Due
to
systematic
biases
of
the
visible
channels
toward
smaller
values
and
the
NIR
channels
toward
slightly
larger
values,
the
overall
systematic
biases
for
NDVI
are
positive.
The
polynomial
approximations
are
provided
for
the
bulk
spectral
correction
with
respect
to
the
AVHRR
NOAA-9
for
consistency
with
previous
study.
Analysis
was
also
conducted
for
the
SRF
effect
only
among
the
AVHRR-3
type
of
radiometer
on
NOAA-15,
16,
17,
18
and
METOP-A
using
AVHRR
NOAA-18
as
a
reference.
The
results
show
more
consistency
between
sensors
with
typical
correction
being
under
5%
(or
0.01
in
absolute
values).
The
AVHRR
METOP-A
reveals
the
most
different
behavior
among
the
AVHRR-3
group
with
generally
positive
bias
for
visible
channel
(up
to
+5%,
relative),
slightly
negative
bias
for
the
NIR
channel
(1%-2%
relative),
and
negative
NDVI
bias
(-0.02
to
+0.005).
Polynomial
corrections
are
also
suggested
for
normalization
of
AVHRR
on
NOAA-15,
16,
17
and
METOP-A
to
AVHRR
NOAA-18.
Crown
Copyright
©
2008
Published
by
Elsevier
Inc.
All
rights
reserved.
I.
Introduction
Trishchenko
et
al.
(2002)
provided
an
analysis
and
recommenda-
tions
regarding
the
effect
of
spectral
response
function
(SRF)
on
reflectances
and
normalized
difference
vegetation
index
(NDVI)
for
various
moderate
resolution
sensors.
The
SRF
effect
was
analyzed
for
the
Advanced
Very
High
Resolution
Radiometer
(AVHRR)
onboard
the
NOAA
satellites
NOAA-6
to
NOAA-16
as
well
as
the
Moderate
Resolution
Imaging
Spectroradiometer
(MODIS),
the
VEGETATION
sensor
(VGT)
and
the
Global
Imager
(GLI).
The
results
were
derived
relative
to
the
reference
sensor
AVHRR
NOAA-9
for
consistency
with
the
approach
of
the
International
Satellite
Cloud
Climatology
Project
(ISCCP)
(Rossow
&
Schiffer,
1999).
The
study
of
Trishchenko
et
al.
(2002)
demonstrated
that
the
differences
in
SRF
are
significant
and
must
be
taken
into
account,
in
*
588
Booth
Street,
Ottawa,
Ontario,
Canada
KIA
0Y7.
TeL:
+1
613
995
57
87;
fax:
+1
613
947
14
06.
E-mail
address:
trichtch@ccrs.nrcan.gc.ca
.
particular
for
studies
concerning
the
inter-annual
variations
in
satellite
time
series.
The
effect
is
comparable
in
magnitude
to
the
uncertainties
caused
by
sensor
calibration,
atmospheric
and
angular
correction
and
can
lead
to
systematic
biases
if
neglected.
Even
among
"the
same
type"
instruments
such
as
AVHRR,
the
effect
of
the
varying
spectral
response
function
on
surface
and
top
of
the
atmosphere
(TOA)
spectral
reflectances
and
the
Normalized
Difference
Vegetation
Index
(NDVI)
is
sufficiently
large
to
require
correction.
Relative
to
the
AVHRR/NOAA-9,
differences
between
various
AVHRR
sensors
were
found
to
vary
from
-25%
to
+12%
for
visible
(red)
reflectance,
and
from
-2%
to
+4%
for
the
near
infrared
(NIR)
reflectance.
The
absolute
differences
in
NDVI
among
various
AVHRRs
ranged
from
-0.02
to
+
0.06.
The
most
consistent
with
AVHRR/NOAA-9
results
were
obtained
for
AVHRR/NOAA-11
and
-12.
The
corrections
must
be
implemented
for
other
AVHRRs
and
especially
for
the
AVHRR/3
on
NOAA-15
and
-16.
Reflectances
and
NDVI
from
MODIS
differ
from
AVHRR/NOAA-9
by
as
much
as
30-40%.
Likewise,
VGT
and
GI1
also
exhibit
considerable
differences
relative
to
AVHRR
observations
and
should
be
always
corrected
in
comparing
long-term
tome
series.
0034-42571$
-
see
front
matter.
Crown
Copyright
C
2008
Published
by
Elsevier
Inc.
All
rights
reserved.
doi:10.1016/j.rse.2008.10.002
Spectral
response
functions
0
4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
Wavelenghth
[juL]
Fig.
1.
The
spectral
response
functions
of
visible
(red)
and
NIR
channels
for
the
AVHRR
NOAA-17
(a),
NOAA-18
(b)
and
METOP-A
(c)
shown
as
solid
lines
relative
to
the
AVHRR
NOAA-9
(dashed
line).
Typical
spectral
reflectance
curve
for
green
vegetation
is
shown
on
each
panel
as
dash-dot
line.
100
80
60
c*"
40
20
O
fi
100
n
80
a)
u)
60
Q
40
rn
,(1)
20
is
46
80
co
60
40
20
-
a
-N-17
Ch.1
Ch.2
green
vegetation
I
-
b)
--,
-
N-18
-
Ch.1
Ch.2
-
r
I
-
c)
-
METOP-A
Ch.1
Ch.2
336
A.P.
Trishchenko
/
Remote
Sensing
of
Environment
113
(2009) 335-341
The
method
was
found
quite
efficient
and
was
applied
in
many
studies.
Swinnen
and
Veroustraete
(2008)
used
it
for
extending
the
SPOT-VEGETATION
NDVI
time
series
back
in
time
with
AVHRR
data
for
Southern
Africa
and
found
that
accounting
for
the
SRF
effect
considerably
improved
the
consistency
between
sensors.
Stow
et
al.
(2004)
employed
it
for
remote
sensing
studies
of
vegetation
and
land-
cover
change
in
Arctic
tundra
ecosystems.
Li
et
al.
(2007)
considered
these
results
in
assessing
the
cloud
discrimination
capabilities
of
current
and
future
sensors.
Chuvieco
et
al.
(2008)
used
proposed
SRF
corrections
in
generation
of
long-time
series
of
burn
areas
in
the
boreal
forest
using
data
for
different
AVHRR
sensors
processed
at
the
Canada
Centre
for
Remote
Sensing
(CCRS)
(Latifovic
et
al.,
2005).
Venturini
et
al.
(2004)
employed
these
results
for
analysis
of
evaporative
fractions
estimated
from
AVHRR
and
MODIS
over
South
Florida.
The
study
of
Teillet
et
al.
(2007)
emphasized
the
importance
of
SRF
effects
on
sensor
radiometric
cross-calibration.
An
alternative
approach
for
correction
of
SRF
effect
was
proposed
in
Trishchenko
et
al.
(2008)
based
on
empirical
regression
between
observations
for
overlapping
periods.
This
method,
however,
requires
significant
preprocessing
efforts
to
derive
first
the
parameters
of
bi-directional
reflectance
distribution
function.
The
AVHRR
continues
to
be
widely
used
operational
sensor
for
large
number
of
environmental
and
climate
applications.
After
the
study
of
Trishchenko
et
al.
(2002)
three
new
AVHRR
radiometers
were
launched
on
NOAA-17,
18
and
Meteorological
Operational
satellite
(METOP-A)
platforms.
This
paper
provides
recommendations
regard-
ing
the
SRF
effect
for
these
new
sensors
to
achieve
consistency
in
time
series
of
clear-sky
products
derived
from
historical
AVHRR
observa-
tions.
As
in
the
previous
study,
the
results
are
provided
relative
to
the
reference
sensor
AVHRR
NOAA-9.
In
addition,
similar
analysis
was
conducted
only
among
all
five
AVHRR-3
type
of
sensors
flown
on
NOAA-15,
16,
17,
18
and
METOP-A.
The
differences
were
assessed
relative
to
AVHRR
NOAA-18.
The
results
for
the
AVHRR-3
group
may
be
of
interest
if
only
the
last
decade
is
analyzed.
As
in
the
previous
study
of
Trishchenko
et
al.
(2002)
the
recom-
mendations
are
provided
for
reflectances
in
spectral
band
1
(red)
and
band
2
(near
infrared
-
NIR)
p
NIR
at
the
surface
and
the
TOA
levels,
as
well
as
for
the
vegetation
index
NDVI
defined
as
NIR
red
NDVI
-
p
NIR
+
pred
The
6S
radiative
transfer
code
(Vermote
et
al.,
1997)
was
employed
for
model
simulation
of
the
signal
at
the
TOA
level
under
various
atmospheric
conditions
and
observational
geometries
similar
to
Trishchenko
et
al.
(2002).
The
set
of
surface
spectral
reflectances
was
employed
in
the
same
manner
as
in
Trishchenko
et
al.
(2002).
Namely,
17
representative
spectra
were
selected
for
12
surface
classes:
1)
coniferous
forest,
2)
deciduous
broadleaf
forest,
3)
closed
shrubland,
4)
open
shrubland,
5)
drygrass/savanna,
6)
grassland,
7)
cropland,
8)
crop/natural
vegetation
mosaic,
9)
barren/desert,
10)
water
bodies,
11)
fresh
snow,12)
coarse
granular
snow.
The
sources
of
surface
spectral
information
are
described
in
Trishchenko
et
al.
(2002).
The
solar
zenith
angle
(SZA)
varied
from
to
85°.
The
viewing
zenith
angle
(VZA)
varied
from
to
65°,
and
the
relative
azimuth
angle
(RAA)
varied
from
to
180°.
The
total
water
vapor
amount
in
the
atmosphere
varied
from
0.05
cm
to
5
cm,
ozone
columnar
amount
varied
from
150
DU
to
450
DU.
Aerosol
optical
depth
varied
from
0.01
to
0.6.
AVWRR
NOAA-17
relative
to
NOAA-9
15
Surface
TOA
:
.1.1.1.1
Ch.1
Surf
_
Ch.1
TOA
10
0.01
0
5
o
o
°
gr-
_5
E_
0.00
0
0)
-K
-10
.
-
-0.01
.
4
-15
<
-20
8
-0.02
0
-25
a)
os
-
b)
Z
-30
3-
„,
30
0.03
0.03
Ch.2
Surf
Ch.2
TOA
25
<
20
0
relative
difference
-
0.02
A
absolute
difference
0
15
10
-
0.01
5
0
. .
0.00
-5
-10
c)
-
d)
-0.01
15
14
0.07
0
NDVI
Surf
s
,
NDVI
TOA
U
12
C
10
a)
0
06
0.05
a)
6
0
0
0
9
°
0
0
0
.
0
e
0.04
0.03
4
0.02
2
0.01
0
0.00
e)
-2
iT)
lY
4
-0.2
-0.01
0
0.02
0.0
0.2
0.4
0.6 0.8
1
0
-0.2
0.0
0.2
0.4
0.6
0.8
1
NDVI
surf
N
DV
'IDA
Fig.
2.
The
absolute
(solid
triangles)
and
relative
(open
circles)
differences
in
surface
and
TOA
reflectances
for
channels
1,
2
and
NDVI
between
the
AVHRR
NOAA-17
and
the
AVHRR
NOAA-9.
All
data
points
are
plotted
versus
NDVI
of
particular
sensor.
Quadratic
best
fits
for
absolute
(solid)
and
relative
(dashed)
differences
are
also
shown.
Parameters
of
fitting
curves
are
given
in
Table
1.
Abso
lu
te
difference
NO
AA-
17
-
NO
AA-
9
Ch.2
Surf
0
relative
difference
A
absolute
differense
-
....
0
.0
t
k3
0
%rig
c)
0.02
0.01
0.00
-0.01
-0.02
0.03
0.03
0.02
0.01
0.00
-0.01
0.0
0.2
0.4
0.6
0.8
1
0
-0.2
0.0
0.2
0.4
0.6
0.8
NDVI
surf
NDVITOA
Surface
I
a)
Ch.1
Surf
:
a
s
r;B
TOA
Ch.1
TOA
b)
0
G
NDVI
Surf.
0
m.
a
e)
Ch.2
TOA
yL
d)
0
NDVI
T•A
-
t
r
)
,
a
0
G
00
a)
'15
8
6
4
2
0
-2
a)
-4
-6
-0.2
0.02
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
10
20
15
10
0
5
0
0
-a
-5
0)
-10
<
-15
<
-20
O
-25
Z
30
Er;
30
..
„1.
25
_;"e
20
O
15
Z
10
as
5
r
0
<
l
c
_
5
<
-10
O
-15
Z
-20
N
14
12
10
6
Ch.2
Surf
A
tr
r
,
ROC
'
relative
difference
absolute
difference
:"
'
F
A
C)
Ch.1
TOA
b)
Ch.
TqA
tip
:
d)
NDVI
TOA
0.2
0.4
0.6
0.8
NDVI
-
roA
002
0
01
0
00
-0.01
-0.02
0.03
0.02
0.01
0.00
-0.01
-0.02
0.03
0.10
0.08
0-06
0.04
0.02
000
-0.02
-0.04
10
0.2
0.4
0.6
0.8
1
0
-0.2
0.0
NDVI
surf
NDVI
Surf
0
00
e)
..1.1
,
1.1
Ch.1
Surf
0
a
0
00
I I I I
o
O
_25
26
15
b
-
-
9
-
10
0
5
0
a
-5
°?
-10
<
-15
<
-20
--...
-30
37
20
15
10
0
5
z
0
-5
-10
-15
-20
-25
-30
'-'
20
(ll
0
15
10
5
3.
Corrections
for
AVHRR
NOAA-17,18
and
METOP
relative
to
NOAA-9
The
spectral
response
function
effect
in
this
section
is
determined
relative
to
AVHRR
NOAA-9
selected
as
a
reference
sensor
in
Trishchenko
et
al.
(2002).
This
radiometer
is
often
considered
as
a
reference
instrument
follow
to
Rossow
and
Schiffer
(1999).
Previous
study
of
Trishchenko
et
al.
(2002)
analyzed
the
SRF
effect
for
the
AVHRR
radiometer
onboard
the
NOAA
satellites
NOAA-6
to
NOAA-16
as
well
as
MODIS,
VGT
and
GU
sensors.
With
addition
of
three
new
sensors,
this
collection
of
AVHRR
sensors
includes
now
13
instru-
ments.
Three
new
radiometers
continue
the
series
of
AVHRR-3
type
instruments
that
include
also
AVHRR
NOAA-15
and
16.
Unlike
all
previous
AVHRR
instruments
that
were
launched
on
the
NOAA
platforms,
the
most
recent
AVHRR
was
launched
in
2006
onboard
METOP-A
satellite
in
the
framework
of
the
joint
program
established
by
the
European
Space
Agency
(ESA)
and
the
European
Organization
for
the
Exploitation
of
Meteorological
Satellites
(EUMETSAT).
The
AVHRR
contributed
by
NOAA
is
also
scheduled
to
be
launched
on
follow-up
satellites
METOP-B,
and
METOP-C.
Figs.
2-4
show
the
absolute
and
relative
differences
in
reflectance
or
NDVI
between
specific
sensor
and
the
AVHRR
NOAA-9.
The
reflectance
is
computed
according
to
Eq.
(1).
Abso
lu
te
differ
ence
NO
AA-
18
-
NO
AA-
9
Trishchenko
/
Remote
Sensing
of
Environment
113
(2009) 335-341
337
AVHRR
NOAA-18
relative
to
NOAA-9
Fig.
3.
The
same
as
Fig.
2,
but
for
the
AVHRR
NOAA-18.
Parameters
of
fitting
curves
are
given
in
Table
2.
The
paper
is
organized
as
follows.
Section
2
describes
the
special
features
of
instrument
spectral
response
functions
for
three
AVHRR
sensors
on
NOAA-17,
18
and
METOP-A.
Section
3
presents
results
and
derived
corrections
relative
to
AVHRR
NOAA-9
sensor.
Section
4
provides
results
of
analysis
and
recommendations
for
corrections
of
AVHRR-3
radiometers
onboard
NOAA-15,
16,
17,
and
METOP-A
relative
to
AVHRR
NOAA-18.
Section
5
summarizes
the
study.
2.
Spectral
response
functions
for
AVHRR
NOAA-17,18
and
METOP-A
The
spectral
response
functions
for
AVHRR
bands
1
and
2
of
NOAA-
17,
18
and
METOP-A
are
shown
in
Fig.
la-c
together
with
curves
for
AVHRR
NOAA-9
plotted
for
the
reference.
The
data
for
spectral
response
functions
were
taken
from
the
NOAA
Center
for
Application
and
Research
(STAR)
web-site
http://www.orbit.nesdis.noaa.gov/
smcd/spb/fwu/solar_cal/spec_resp_func/index.html.
The
typical
spec-
tral
dependence
of
green
vegetation
is
also
shown
in
Fig.
1.
The
major
observed
differences
are
seen
as
more
narrow
shape
of
band
1
for
the
AVHRR
NOAA-17,
18
and
METOP-A
relative
to
AVHRR
NOAA-9,
and
as
a
shift
of
SRF
for
band
2
of
the
AVHRR
NOAA-17,
18
and
METOP-A
away
from
the
red
edge
region
around
0.7
pm
toward
longer
wavelengths.
The
above
differences
lead
to
the
following
systematic
differences
in
general:
a)
slightly
darker
band
1
reflectances,
b)
slightly
brighter
band
2
reflectances,
c)
slightly
larger
NDVI.
All
these
features
were
observed
in
a
previous
study
Trishchenko
et
al.
(2002)
for
AVHRR-3
type
of
sensors.
They
are
manifested
in
a
similar
way
for
new
sensors
although
there
are
some
minor
differences
related
to
a
specific
SRF
shape
for
each
spectral
band.
An
interesting
feature
of
Fig.
lc
is
a
small
spectral
leak
for
band
1
of
the
AVHHR
METOP-A
in
the
blue
region
which
leads
to
some
noticeable
differences
for
the
results
at
the
TOA
level
due
to
intensive
Rayleigh
and
aerosol
scattering
for
short
wavelengths.
S
T
(X)F(X)dX
x
P(N)
„„„
max
f
(X)F(X)dX
x„„„
(1)
where
ST
(X),
Si
(X)
are
the
upward
or
downward
solar
spectral
radiances
and
F
(X)
is
the
instrument
spectral
response
function.
The
AVHRR
METOP-A
relative
to
NOAA-9
Surface
TOA
Fig.
4.
The
same
as
Fig.
2,
but
for
the
AVHRR
METOP-A.
Parameters
of
fitting
curves
are
given
in
Table
3.
Abso
lu
te
difference
MET
OP-
A
-
N
OAA-
9
a)
7
-
Ch.1
Ch.2
--
NDVI
...
f
METOP-A
-
Ch.1
Ch.2
-
-
-
-•-
NDVI-
.
.
.
-0.04-0.02
0.00
0.02
0.04
0.06
b)
N17
-
Ch.2
NDVI:
-20
-10
0
10
20
30
Cl)
4
E
.
0
N18_
-
Ch
.1
-
Ch
.2
.
NDV1
-0.04
-0.02
0.00
0.02
0.04
0.06
d
N18
-
Ch.1
-
Ch
.2
NDVI
-20
-10
0
10
20
30
0
0
!VI
EToP-A.
-
Ch.1
-
Ch.2
-
NDVI
-
r-
-0.04
-0.02
0.00
0.02
0.04
0.06
Absolute
difference
-20
-10
0
10
20
30
Relative
difference
[%]
co
0
C
z
0
O
C
0
338
A.P.
Trishchenko
/
Remote
Sensing
of
Environment
113
(2009) 335-341
Absolute
difference
Relative
difference
[%]
Fig.
5.
Absolute
(left
panels)
and
relative
(right
panels)
differences
in
TOA
reflectances
and
NDVI
for
the
AVHRRs
on
NOAA-17
(top),
NOAA-18
(center)
and
METOP-A
(bottom)
relative
to
the
AVHRR
NOAA-9.
spectral
radiances
are
computed
at
the
TOA
and
surface
levels.
The
surface
reflectance
is
assumed
to
be
Iambertian.
The
results
for
each
sensor:
AVHRR
NOAA-17,
18
and
METOP-A
are
plotted
in
Figs.
2-4.
The
statistical
distributions
of
differences
at
the
TOA
are
displayed
in
Fig.
5.
The
distributions
are
not
shown
at
the
surface
level
because
they
can
be
determined
relatively
easy
from
Figs.
2-4
for
each
sensor.
The
results
in
Figs.
2-4
are
plotted
against
NDVI
computed
at
the
surface
or
TOA
level
depending
on
the
analysis.
Although
the
magnitude
of
SRF
effect
is
influenced
by
many
factors,
such
as
surface
reflectance
spectrum,
atmospheric
state,
geometry
of
observations
that
may
alter
the
spectral
com-
position
of
the
radiance
spectra
within
the
instrument
spectral
re-
sponse,
we
found
that
NDVI
serves
as
a
good
indicator
of
the
effect
(Trishchenko
et
al.,
2002).
This
occurs
because
the
effect
of
spectral
overlap
or
spectral
separation
between
visible
and
NIR
channels
over
the
"red"
edge
(0.7
pm)
in
the
pixel
surface
reflectance
spectrum
is
related
to
NDVI.
In
addition,
NDVI
generally
characterizes
the
shape
of
surface
spectra,
and
as
such
it
also
influences
the
magnitude
of
SRF
effect.
The
quadratic
fits
are
also
shown
in
Figs.
2-4
for
all
panels
except
panel
f)
that
shows
the
relative
NDVI
differences.
This
is
because
relative
difference
may
be
very
large
due
to
division
by
small
NDVI
value
when
it
is
close
to
zero.
Except
the
limited
num-
ber
of
points
for
the
extreme
geometry
conditions
(large
viewing
and
solar
zenith
angles)
the
dependence
of
differences
on
NDVI
is
captured
reasonably
well
by
quadratic
fit.
The
parameters
of
quad-
ratic
fit
for
each
sensor,
spectral
band
and
TOA
or
surface
level
are
given
in
Table
1
for
the
AVHRR
NOAA-17,
Table
2
for
the
AVHRR
NOAA-18,
Table
3
for
the
AVHRR
METOP-A.
The
tables
also
contain
r
2
and
a
(coefficient
of
determination
and standard
deviation
of
the
fit)
for
each
fitting
curve.
The
r
2
-values
vary
between
0.75
and
0.98
except
for
the
NDVI
relative
difference
and
the
NIR
TOA
reflectance
for
the
AVHRR
METOP-A
where
they
can
be
as
low
as
03-0.5.
The
a-values
of
quadratic
fit
for
absolute
differences
are
typically
better
than
0.005.
The
relative
a-values
are
close
to
1%
or
smaller
for
most
cases,
but
they
can
be
slightly
higher
than
2%
for
the
relative
NDVI
fit.
They
are
close
to
1%
for
relative
differences
for
visible
band.
The
overall
distribution
of
differences
describing
the
SRF
effect
for
new
AVHRR
sensors
relative
to
AVHRR
NOAA-9
is
shown
in
Fig.
5.
It
presents
the
histograms
of
relative
and
absolute
differences
for
visible,
NIR
reflectance
and
NDVI
at
the
TOA
level
for
each
sensor.
The
differences
in
reflectance
among
three
radiometers
relative
to
AVHRR
NOAA-9
are
similar
and
range
from
-0.015
to
0.015
(-20%
to
+2%
relative)
for
visible
channel,
and
from
-0.03
to
0.02
(-5%
to
+5%)
for
NIR
channel.
Absolute
change
in
NDVI
ranges
from
-0.03
to
+
0.06.
Due
to
systematic
biases
of
visible
channels
toward
smaller
values
and
NIR
Table
1
Parameters
of
quadratic
best
fit
to
absolute
spectral
correction
Ap=p-p
N
OAA_9
and
relative
spectral
correction
Ap
=
Pg'"°":
(%)
for
visible
(red)
channel
(Ch.1)
reflectance,
NIR
channel
(Ch.2)
reflectance
and
NDVI
for
the
AVHRR
NOAA-17
relative
to
the
AVHRR
NOAA-9.
X
denotes
NDVI
observed
by
the
AVHRR
NOAA-17
Parameter
Absolute
correction
r
2
a
Relative
correction
(%)
r
2
a(%)
Chi
Surf
Chi
TOA
Ch.2
Surf
Ch.2
TOA
NDVI
Surf
NDVI
TOA
0.00026-0.0224
X+
0.0121
X
2
0.00089-0.0221
X+
0.0131
X
2
-
0.00191+0.0174
X-0.0029
X
2
-
0.00218+0.0147
X-0.0007
X
2
-
0.00077
+0.0897
X-
0.0340
X
2
-
0.00141
+0.0752
X-0.0144
X
2
0.81
0.76
0.84
0.80
0.97
0.98
0.0021
0.0024
0.0022
0.0020
0.0035
0.0027
-
0.110+2.774
X-35.725
X
2
0.111-6.455
X-16.418
X
2
-
0.160+4.650
X-1.027
X
2
-
0.537
+5.452
X-1.629
X
2
8302
-1.687
X-0.507
X
2
0.98
0.96
0.95
0.94
030
130
0.95
0.28
032
1.10
p
denotes
a
channel
reflectance
or
NDVI
at
the
surface
(Surf)
or
TOA
level.
Table
2
The
same
as
Table
1
but
for
the
AVHRR
NOAA-18
Parameter
Absolute
correction
r
2
a
Relative
correction
(%)
r
2
a
(%)
Chi
Surf
Chi
TOA
Ch.2
Surf
Ch.2
TOA
NDVI
Surf
NDVI
TOA
0.00011-0.0191
X+
0.0075
X
2
0.00178-0.0254
X+
0.0141
X
2
-
0.00308+0.0265
X-0.0085
X
2
-
0.00520+0.0233
X-0.0070
X
2
-
0.00162+0.0947
X-0.0338
X
2
-
0.00661
+0.0875
X-0.0132
X
2
0.80
0.75
0.81
0.68
0.97
0.97
0.0022
0.0030
0.0032
0.0038
0.0039
0.0037
-
0.059+4.801
X-39.601
X
2
0305-6.849
X-18.754
X
2
-
0.292+6.724
X-2.257
X
2
-
1.443
+7.278
X-
2.486
X
2
11.030-12.849
X+
9.878
X
2
0.98
0.96
0.93
0.88
0.41
0
132
0.99
0.44
0.62
2.07
AVHRR
NOAA-17
relative
to
NOAA-18
Surface
TOA
8
8'
9,
6
0
O
4
2
00
D
.c
k
-2
<
-4
O
_6
-.....
-8
co
8
t-
<C
0
0
z
a)
10
5
81)
'13
a)
-10
T.)
1
--0.2
0.0
0.010
6
4
2
0
-2
-4
-6
-8
15
0
-5
0.2
0,4
0.6 0.8
1
0
-0.2
0.0
0.2
0.4
0.6 0.8
NDVI
surf
NDVIT0A
1
0
0
0.
3
0.005
0.000
-0.005
0.010
0.005
0.000
-0.005
0.03
0.02
0.01
0.00
-0.01
0.010
0.010
0.02
Abso
lu
te
differen
ce
NO
AA-
17
-
NOAA-
18
NDVI
TOA
ati
R.
b)
Ch.1
TOA
.1,
d)
f
t
Ch.
TOA
a
Q.
a
Ch.1
Surf
c•cP
a
2
.a
Ch.2
Surf
2
P
2
2
0
2
eon
a
a!
o
relative
difference
.
absolute
difference
e)
NDVI
Surf
s`
0
0
0
6
4
2
Co
0
-2
-4
0-6
?I
C
!
0
O
A.P.
Trishchenlco
/
Remote
Sensing
of
Environment
113
(2009) 335-341
339
Table
3
The
same
as
Table
1
but
for
the
AVHRR
METOP-A
Parameter
Absolute
correction
r
2
CY
Relative
correction
(%)
r
2
(%)
Ch.1
Surf
0.00016-0.0238
X+
0.0152
X
2
0.79
0.0020
-0.146+0307
X-30.282
X
2
0.98
132
Ch.1
TOA
0.00189-0.0208
X+
0.0122
X
2
0.76
0.0023
0.617-5.942
X-13.625
X
2
0.93
1.08
Ch.2
Surf
-0.00406+0.0308
X-
0.0150
X
2
0.75
0.0036
-0393+6.881
X-3.069
X
2
0.89
0.53
Ch.2
TOA
-0.00752+0.0266
X-0.0131
X
2
0.53
0.0054
-2.170+7.187
X-2.858
X
2
0.74
0.95
NDVI
Surf
-0.00150+0.1055
X-0.0571
X
2
0.98
0.0031
15.195-25.667
X+
18.001
X
2
0.76
2.17
NDVI
TOA
-0.01216+0.0784
X-0.0173
X
2
0.94
0.0049
channels
toward
slightly
larger
values,
overall
systematic
biases
for
NDVI
are
mostly
positive.
4.
The
SRF
effect
among
AVHRR-3
sensors
relative
to
AVHRRNOAA-18
The
previous
study
by
Trishchenko
et
al.
(2002)
showed
that
the
most
significant
SRF
effect
is
observed
between
AVHRR
instruments
from
different
groups.
This
is
especially
true
for
the
differences
between
AVHRR-3
group
that
includes
AVHRR
NOAA-15,
16,
17,
18
and
METOP-A
and
the
rest.
To
understand
better
how
significant
are
these
differences
among
the
most
recent
AVHRRs,
we
conducted
a
similar
analysis
within
this
group
using
the
AVHRR
NOAA-18
as
a
reference
sensor.
It
is
expected
that
these
results
will
be
useful
for
analysis
of
consistency
in
the
AVHRR
time
series
for
the
last
decade
after
the
launch
of
NOAA-15
in
1998.
Because
the
spectral
response
functions
for
these
AVHRR
radiometers
are
quite
similar,
one
can
expect
that
the
magnitude
of
the
SRF
effect
will
be
smaller
than
the
one
determined
relative
to
the
AVHRR
NOAA-9.
Calculations
confirmed
that
this
is
indeed
the
case.
The
differences
for
the
AVHRR
NOAA-15,
16
and
17
were
quite
close.
The
SRF
effect
for
the
AVHRR
METOP-A
is
somewhat
larger
due
to
more
distinct
shape
of
the
spectral
response
curves
of
visible
and
NIR
bands.
The
results
of
analysis
are
shown
in
Figs.
6-8.
Figs.
6
and
7
show
similar
results
as
in
Figs.
2-4,
but
computed
for
the
AVHRR
NOAA-17
and
the
AVHRR
METOP-A
relative
to
the
AVHRR
NOAA-18.
We
do
not
show
figures
for
the
AVHRR
NOAA-15
and
16
because
results
look
similar.
The
parameters
of
quadratic
regression
for
each
sensor
relative
to
the
AVHRR
NOAA-18
are
provided
in
Tables
4-7
for
the
AVHRR
NOAA-15,
16,
17
and
METOP-A
correspondingly.
Fig.
8
shows
the
distribution
of
differences
for
all
sensors
including
the
AVHRR
NOAA-
15
and
16.
Results
in
Fig.
8
are
shown
at
the
TOA
level,
similar
to
Fig.
5.
The
absolute
and
relative
errors
are
shown
on
the
panels
on
the
left
and
right
side
correspondingly.
The
magnitude
of
the
SRF
effect
within
the
AVHRR-3
group
computed
relative
to
the
AVHRR
NOAA-18
is
noticeably
smaller
than
the
magnitude
of
the
effect
computed
relative
to
the
AVHRR
NOAA-9.
The
differences
are
smaller
by
about
factor
of
3.
They
normally
vary
within
±
0.01
for
absolute
values
(except
the
METOP-A
TOA
NDVI
AVHRR
METOP-A
relative
to
NOAA-18
Surface
TOA
--.._
-8
0
.7
) .
8
0.010
0.005
0.000
-0.005
0.010
0.010
Abso
lu
te
difference
METOP-
A
-
N
OAA-
18
Ch.1
Surf
0
64
:
-
4.
2
a
:a)
I
1
Ch.1
TOA
A
b)
6
_
Q
Ch.2
Surf
relative
difference
Ch.
TOA
4
-•
absolute
difference-
0.005
0
2
Z
0
0.000
v
2
-2
CL
-4
1
-0.005
0
I-
-6
U.1
_
8
c)
0.010
2
15
0.03
.......
NDVI
Surf
NDVI
TOA
V
10
2
pi
p
0.02
0)
5
°
go
2
0.01
'
07.)
0
A
0.00
"o
o
cco0
D
a)
-0.01
-10
e)
fi
-D_02
tp
-15
0
103
-0.2
0.0
0.2
0.4
0.6 0.8
1
0
-0.2
0.0
0.2
0.4
0.6 0.8
1
Fig.
6.
The
absolute
(solid
triangles)
and
relative
(open
circles)
differences
in
surface
and
TOA
reflectances
for
channels
1
and
2
and
NDVI
between
the
AVHRR
NOAA-17
and
the
AVHRR
NOAA-18.
All
data
points
are
plotted
versus
NDVI
of
particular
sensor.
Parameters
of
fitting
curves
are
given
in
Table
6.
NDVI
surf
NDVIT0A
Fig.
7.
The
same
as
Fig.
6,
but
for
the
AVHRR
METOP-A.
Parameters
of
fitting
curves
are
given
in
Table
7.
N15
-
Ch.1
Ch
2
NOVI
-
a
-
c)
N16
Ch1
Ch.2
NDVI.
d)
N16
Ch.1
Ch.2
NOV
I
-
cn
0
U
z
0
C.D
-0.02
-0.01
0.00
0.01
0.02
-15
-10
-5
0
5
10
15
e
N17
Ch.1
Ch
2
-
-
NDVI
.
f)
N17
Ch.1
Ch.2
NDVI'
p
0
U
-0.02
-0.01
0.00
0.01
0.02
-15
-10
-5
5
10
15
g)
METOP-A.
-
Ch
1
Ch.2
h)
METOP-A
Ch
-1
Ch.2
NDVI
-
E
z
0
U
C
p
0
-0.02
-0.01
0.00
a
01
0.02
-15
-10
-5
0
5
10
15
Absolute
difference
Relative
difference
[%]
•••
-0.02
-0.01
0.00
0.01
0.02
"
N15
Ch.1
Ch.2
NDVI
10
15
0
0
0
0
b)
340
A.P.
Trishchenko
/
Remote
Sensing
of
Environment
113
(2009) 335-341
Absolute
difference
Relative
difference
[%]
Fig.
8.
The
absolute
(left
panels)
and
relative
(right
panels)
differences
in
TOA
reflectances
and
NDVI
for
AVHRRs
among
AVHRR-3
radiometers
on
NOAA-15
(a-b),
NOAA-16
(c-d),
NOAA-17
(e-f)
and
METOP-A
(g-h)
relative
to
the
AVHRR
NOAA-18.
where
the
difference
can
be
as
low
as
-
0.02).
The
relative
differences
are
usually
within
±5%.
Although
the
coefficients
of
determination
r
2
presented
in
Tables
4-7
for
quadratic
fit
are
smaller,
indicating
the
lesser
quality
of
regression,
the
overall
standard
deviations
a
are
typically
less
than
0.5%
and
only
in
two
instances
(relative
NDVI
differences
for
the
AVHRR
NOAA-15
and
16)
are
close
or
exceed
1%.
The
standard
deviations
for
absolute
differences
are
also
several
times
smaller
than
previously
described
in
Section
3.
This
is
explained
by
smaller
magnitude
of
the
SRF
effect
for
sensors
with
similar
SRF.
The
smaller
magnitude
of
the
effect
reduces
the
correlation
while
the
quality
of
regression
fit
on
the
absolute
scale
may
still
be
very
good.
5.
Conclusions
Long-term
monitoring
of
the
Earth's
environment
by
satellite
sensors
requires
consistent
and
comparable
measurements.
One
of
the
longest
satellite
time
series
used
for
terrestrial
monitoring
and
climate
change
studies
come
from
the
NOAA
AVHRR
sensors
spanning
the
period
of
almost
three
decades.
During
this
period
thirteen
AVHRR
sensors
have
been
operated
by
the
NOAA.
Despite
similar
design
and
comparable
spectral
channels,
each
instrument
has
unique
properties
in
terms
of
spectral
response
functions.
This
new
study
provides
an
extension
of
previous
work
by
Trishchenko
et
al.
(2002)
to
recently
launched
AVHRR
sensors
onboard
NOAA-17,
18
and
METOP-A.
The
results
could
be
used
for
improving
consistency
between
AVHRR
sensors
by
apply
corrections
for
the
spectral
response
function
effect
in
the
way
as
it
was
proposed
in
the
previous
study.
The
corrections
are
derived
to
normalize
all
sensors
to
AVHRR
NOAA-9.
They
are
obtained
as
the
second
order
polynomials
of
sensor-
observed
NDVI.
The
NDVI
was
used
as
a
predictor
because
it
is
related
to
the
effect
of
spectral
overlap
or
spectral
separation
between
visible
and
NIR
channels
over
the
"red"
edge
(0.7
um)
in
the
pixel
reflectance
spectrum
(Trishchenko
et
al.,
2002).
In
addition,
NDVI
generally
characterizes
the
shape
of
surface
spectra,
and
as
such
it
also
influences
the
magnitude
of
SRF
effect.
Although
the
magnitude
of
SRF
effect
depends
on
many
factors,
such
as
surface
reflectance
spectrum,
atmospheric
state,
geometry
of
observations,
NDVI
was
found
to
be
the
most
essential
factor
and
was
employed
for
parameterization.
The
magnitude
of
the
SRF
effect
for
new
AVHRR
sensors
onboard
NOAA-17,
18
and
METOP-A
was
found
similar
to
the
values
reported
in
previous
study
for
AVHRR
NOAA-15
and
16.
The
differences
in
Table
4
The
same
as
Table
1
but
for
the
AVHRR
NOAA-15
relative
to
the
AVHRR
NOAA-18
Parameter
Absolute
correction
r
2
CY
Relative
correction
(%)
r
2
(%)
Oil
Surf
Ch.1
TOA
Ch.2
Surf
Ch.2
TOA
NDVI
Surf
NDVI
TOA
0.00012
-0.0029
X+
0.0040
X
2
-
0.00093
+0.0027
X-
0.0003
X
2
0.00233
-0.0151
X+
0.0136
X
2
0.00527-0.0120
X+
0.0122
X
2
0.00159-0.0078
X+
0.0040
X
2
0.00892
-0.0119
X-0.0031
X
2
0.48
0.29
0.47
0.21
0.68
0.59
0.0003
0.0011
0.0017
0.0032
0.0010
0.0032
-
0.030-2.186
X+4314
X
2
-
0.240-0.010
X+2.920
X
2
0.291-2.722
X+
2.519
X
2
1.610-2.063
X+
1.553
X
2
-
5.242+21.885
X-19.512
X
2
0.80
0.65
0.49
0.25
0.73
0.29
037
0.29
0.61
1.45
Table
5
The
same
as
Table
1
but
for
the
AVHRR
NOAA-16
relative
to
the
AVHRR
NOAA-18
Parameter
Absolute
correction
r
2
CY
Relative
correction
(%)
r
2
a
(%)
Chi
Surf
Chi
TOA
Ch.2
Surf
Ch.2
TOA
NDVI
Surf
NDVI
TOA
0.00012
-
0.0026
X+
0.0049
X
2
-
0.00110+0.0044
X-0.0008
X
2
0.00149-0.0099
X+
0.0089
X
2
0.00310-0.0073
X+
0.0072
X
2
0.00121-0.0042
X-
0.0052
X
2
0.00572-0.0113
X-0.0071
X
2
0.62
039
0.50
0.23
0.88
0.73
0.0004
0.0014
0.0011
0.0019
0.0011
0.0026
-
0.032-4.249
X+10341
X
2
-
0.288+0.265
X+5364
X
2
0.190
-1.829
X+
1.686
X
2
0.944-1329
X+1.011
X
2
-
2.576+11.419
X-11334
X
2
0.97
0.82
0.54
031
0.55
031
0.46
0.18
034
1.00
A.P.
Trishchenlco
/
Remote
Sensing
of
Environment
113
(2009) 335-341
341
Table
6
The
same
as
Table
1
but
for
the
AVHRR
NOAA-17
relative
to
the
AVHRR
NOAA-18
Parameter
Absolute
correction
r
2
a
Relative
correction
(%)
r
2
a(%)
Ch.1
Surf
Ch.1
TOA
Ch.2
Surf
Ch.2
TOA
NDVI
Surf
NDVI
TOA
0.00013
-0.0032
X+
0.0045
X
2
-
0.00103
+0.0036
X-0.0009
X
2
0.00119-0.0092
X+
0.0055
X
2
0.00314-0.0087
X+
0.0060
X
2
0.00094-0.0057
X+
0.0003
X
2
0.00567-0.0132
X-
0.0015
X
2
0.49
031
0.71
038
0.81
0.70
0.0004
0.0013
0.0010
0.0020
0.0009
0.0026
-
0.046-2.556
X+
5311
X
2
-
0.241+0.172
X+
3.427
X
2
0.141-2.110
X+
1.286
X
2
0.962-2.032
X+
1.019
X
2
-
2.406+9.862
X-
9.230
X
2
0.86
0.68
0.83
0.55
0.58
031
0.43
0.16
038
0.88
Table
7
The
same
as
Table
1
but
for
the
AVHRR
METOP-A
relative
to
the
AVHRR
NOAA-18
Parameter
Ch.1
Surf
Ch.1
TOA
Ch.2
Surf
Ch.2
TOA
NDVI
Surf
NDVI
TOA
Absolute
correction
r
2
0.00006-0.0050
X+
0.0081
X
2
0.64
0.00027+0.0042
X-0.0011
X
2
0.29
-
0.00096+0.0044
X-0.0068
X
2
0.44
-
0.00245
+0.0037
X-0.0069
X
2
0.07
0.00008
+0.0117
X-0.0248
X
2
0.71
-
0.00609-0.0093
X-0.0048
X
2
034
a
Relative
correction
(%)
r
2
a
(%)
0.0006
-0.108
-6.069
X+
13.160
X
2
0.95
0.43
0.0016
0332
+1.120
X+
6.930
X
2
0.74
0.85
0.0007
-0.099+0.182
X-
0.853
X
2
0.65
0.14
0.0017
-0.780-0.086
X-
0360
X
2
0.04
0.40
0.0023
3.721
-
11.136
X+
6.780
X
2
0.84
0.85
0.0047
reflectance
among
three
radiometers
relative
to
AVHRR
NOAA-9
range
from
-0.015
to
0.015
(-20%
to
+2%
relative)
for
visible
channel
(red),
and
from
-0.03
to
0.02
(-5%
to
5%)
for
NIR
channel.
The
absolute
change
in
NDVI
ranged
from
-0.03
to
+0.06.
Due
to
systematic
biases
of
visible
channels
toward
smaller
values
and
NIR
channels
toward
slightly
larger
values,
the
overall
systematic
biases
for
NDVI
are
positive.
The
coefficient
of
determination
(r2)
for
quadratic
regression
vary
mostly
between
0.75
and
0.98
except
for
NDVI
relative
difference
and
NIR
TOA
reflectance
for
METOP-A.
The
standard
deviation
(a)
of
quadratic
fit
for
absolute
differences
are
typically
better
than
0.005.
The
relative
a-values
are
close
to
1%
or
smaller
for
most
cases,
except
relative
NDVI
fit.
To
characterize
the
magnitude
of
the
SRF
effect
among
AVHRR-3
type
of
radiometers
onboard
NOAA-15,
16,
17,
18,
and
METOP-A
a
similar
analysis
was
completed
by
selecting
AVHRR
NOAA-18
as
a
reference
instrument.
The
magnitude
of
observed
SRF
effect
among
this
group
was
found
smaller
by
about
factor
of
3
relative
to
numbers
reported
above
with
a
reference
to
AVHRR
NOAA-9.
The
quadratic
polynomial
corrections
as
function
of
NDVI
are
also
provided
for
these
sensors
to
normalize
the
reflectances
and
NDVI
to
AVHRR
NOAA-18
values.
Acknowledgements
This
work
was
conducted
at
the
Canada
Centre
for
Remote
Sensing
(CCRS),
Earth
Sciences
Sector
of
the
Department
of
Natural
Resources
Canada
as
part
of
the
Project
J35
of
the
"Enhancing
Resilience
in
a
Changing
Climate"
Program.
The
study
was
also
supported
by
the
Canadian
Space
Agency
under
the
Government
Related
Initiative
Program
(GRIP).
References
Chuvieco,
E.,
Englefield,
P.,
Trishchenko,
A.
P.,
&
Luo,
Y.
(2008).
Generation
of
long
time
series
of
burn
area
maps
of
the
boreal
forest
from
NOAA-AVHRR
composite
data.
Remote
Sensing
of
Environment,
112(5),
2381-2396.
Latifovic,
R.,
Trishchenko,
A.
P.,
Chen,
J.,
Park,
W.
M.,
Khlopenkov,
K.
V.,
Fernandes,
R.,
Pouliot
D.,
Ungurenu,
C.,
Wang,
S.,
&
Cihlar,
J.
(2005).
Generating
historical
AVHRR
1-km
baseline
satellite
data
records
over
Canada
suitable
for
climate
change
studies.Canadian
Journal
of
Remote
Sensing,
31,
324-346
No.
5.
Li,
Z.,
Li,
J.,
Menzel,
W.
P.,
Schmit,
T.
J.,
&
Ackerman,
S.
A.
(2007).
Comparison
between
current
and
future
environmental
satellite
imagers
on
cloud
classification
using
MODIS.
Remote
Sensing
of
Environment,
108(3),
311-326.
Rossow,
W.
B.,
&
Schiffer,
R.
A.
(1999).
Advances
in
understanding
clouds
from
ISCCP.
Bulletin
of
the
American
Meteorological
Society,
80,
2261-2287.
Stow,
D.
A.,
Hope,
A.,
McGuire,
D.,
Verbyla,
D.,
Gamon,
J.,
Huemmrich,
F.,
Houston,
S.,
Racinef,
C,
Sturmg,
M.,
Tapeh,
K,
Hinzman,
L,
Yoshikawai,
K.,
Tweediej,
C.,
Noylek,
B.,
Silapaswanl,
C,
Douglasm,
D.,
Griffithn,
B,
Jiao,
G.,
Epsteino,
H.,
Walkerp,
D.,
Daeschnera,
S,
Petersena,
A.,
Zhouq,
L,
&
Myneni,
R.
(2004).
Remote
sensing
of
vegetation
and
land-cover
change
in
Arctic
Tundra
Ecosystems.
Remote
Sensing
of
Environment,
89(3),
281-308.
Swinnen,
E.,
&
Veroustraete,
F.
(2008).
Extending
the
SPOT-VEGETATION
NDVI
time
series
(1998-2006)
back
in
time
with
NOAA-AVHRR
data
(1985-1998)
for
Southern
Africa.
IEEE
Transactions
on
Geoscience
and
Remote
Sensing,
46(2),
558-572.
Teillet
P.
M.,
Fedosejevs,
G.,
Thome,
K.
J.,
&
Barker,
J.
L
(2007).
Impacts
of
spectral
band
difference
effects
on
radiometric
cross-calibration
between
satellite
sensors
in
the
solar-reflective
spectral
domain.
Remote
Sensing
of
Environment,
110(3),
393-409.
Trishchenko,
A.
P.,
Cihlar,
J.,
&
Li,
Z.
(2002).
Effects
of
spectral
response
function
on
surface
reflectance
and
NDVI
measured
with
moderate
resolution
satellite
sensors.
Remote
Sensing
of
Environment,
81(1),
1-18.
Trishchenko,
A.
P.,
Luo,
Y.,
Khlopenkov,
K.
V.,
&
Wang,
S.
(2008).
A
method
to
derive
the
multi-spectral
surface
albedo
consistent
with
MODIS
from
historical
AVHRR
and
VGT
satellite
data.
Journal
of
Applied
Meteorology
and
Climatology,
47(4),
1199-1221.
Venturini,
V.,
Bisht,
G.,
Islam,
S.,
&
Jiang,
L
(2004).
Comparison
of
evaporative
fractions
estimated
from
AVHRR
and
MODIS
sensors
over
South
Florida.
Remote
Sensing
of
Environment,
93(1-2),
77-86.
Vermote,
E.,
Tanre,
D.,
Deuze,
J.
L,
Herman,
M.,
&
Moricette,
J. J.
(1997).
Second
simulation
of
the
satellite
signal
in
the
solar
spectrum:
An
overview.
IEEE
Transactions
on
Geoscience
and
Remote
Sensing,
35,
675-686.