close

Вход

Забыли?

вход по аккаунту

?

Passive microwave snow mapping in Quebec

код для вставкиСкачать
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter free, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
«
PASSIVE MICROWAVE SNOW MAPPING IN QUEBEC
by
XIAO, RENMENG
GEOGRAPHY DEPARTMENT
McGILL UNIVERSITY, MONTREAL, QUEBEC
A Thesis
Submitted to
the Faculty of Graduate Studies and Research
McGill University
In Partial Fulfillment of the Requirements for the Degree
of
Master of Science
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1+1
National Library
of Canada
Bibiiotheque nationale
du Canada
Acquisitions and
Bibliographic Services
Acquisitions et
services bibliographiques
395 Wellington Street
Ottawa ON K1A0N4
Canada
395. rue Wellington
Ottawa ON K1A0N4
Canada
Your me Voire reference
Our Sle Notre reference
The author has granted a non­
exclusive licence allowing the
National Library of Canada to
reproduce, loan, distribute or sell
copies of this thesis in microform,
paper or electronic formats.
L’auteur a accorde une licence non
exclusive pennettant a la
Bibliotheque nationale du Canada de
reproduire, preter, distribuer ou
vendre des copies de cette these sous
la forme de microfiche/film, de
reproduction sur papier ou sur format
electronique.
The author retains ownership of the
copyright in this thesis. Neither the
thesis nor substantial extracts from it
may be printed or otherwise
reproduced without the author’s
permission.
L’auteur conserve la propriete du
droit d’auteur qui protege cette these.
Ni la these ni des extraits substantiels
de celle-ci ne doivent etre imprimes
ou autrement reproduits sans son
autorisation.
0 612 29810-8
-
-
Canada
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
Abstract
The objective of this research is to map snow cover in
the
Quebec
area
using
passive
microwave
and
other
remote
sensing data. The areal snow extent and snow water equivalent
are determined and a twelve year snow
produced
for
the
purpose
of
water equivalent map is
analyzing
interannual
snow
variability.
The presence of vegetation cover will affect the data
obtained with passive systems.
For heavily vegetated areas
such as Quebec, the vegetation effect should be predetermined
and classified to reduce the error on snow water equivalence
calculation.
In dry snow conditions, forest coverage and snow density
are the two major error parameters in passive microwave snow
mapping.
The error on snow water equivalence estimation
is
directly proportional to the error in estimated snow density
and forest coverage. For Quebec, ignoring the fraction of the
forest
cover
may
cause
up
to
49%
snow
depth
or
water
equivalence underestimation.
The ground measured snow depth and snow density data are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
necessary for calibrating satellite derived snow depth and
(
mean snow density within forest covered regions.
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(
Resume
Le but de la recherche est de cartographier 1 1enneigement
au
Quebec
satellites.
par
1'utilisation
L'etendue
de
de
donnees
1 ’enneigement
obtenues
et
par
1 ’equivalence
eau/neige sont determines et une carte d 1equivalence eau/neige
sur une periode de 12 ans est etablie dans le but d'analyser
les variations interannuelles d'enneigement.
La Presence d'un convert vegetal affectera les donnees
obtenues
par satellite.
Pour
les regions
vegetal important, tel le Quebec,
devrait
etre
predetermine
et
ayant
un
convert
1'impact de la vegetation
classifie
pour
reduire
les
erreurs de calcul sur 1 1equivalence eau/neige.
La converture forestiere et la densite de la neige sont
les deux principales sources d'erreurs dans la cartographie de
1'enneigement par
satellite.
L'erreur
sur
1'estimation
de
1'equivalence eau/neige est directement proportionnelle aux
erreurs
d 1estimation
de
la
densite
de
la
neige
de
la
couverture forestiere. Pour le Quebec, ignorer la part de la
foret peut etrainer une sous estimation allant jusqu'a 49% de
1'enneigement ou de 1'equivalent eau/neige.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Les mesures au sol de la profondeur et de la densite de
la neige sont necessaires pour calibrer 1'evaluation de ces
parametres par satellite dans les regions forestieres.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
My deepest thanks gratitude to Dr. John Lewis, my thesis
supervisor
financial
who
provided
support
to
equipment,
help
me
laboratory
complete
this
space,
and
project.
His
guidance, helpful criticism, and concern were truly valuable
and very important for me. This research experience provided
me with the opportunity to learn a lot of things about science
and scientists.
Thank also to Dr. A. T. C. Chang, and Dr. K. Steffen for
many useful information and suggestions about this project.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
Contents
1. Introduction
1
2. Background and Literature Review
7
2.1
Theoretic background
7
2.2
SMMR and SSM/I data
10
2.2.1
Frequency and channel
15
2.2.2
Polarization
16
2.2.3
Snow extent and
2.2.4
Snow water equivalent
21
2.2.5
The effect of vegetation cover
22
2.2.6
Passive microwave experiments
26
2.3
depth
NOAA-AVHRR data
3. Snow Mapping
Procedures
17
28
35
3.1
Study area
36
3.2
Geographic Information System
40
3.3
Data base
42
3.3.1
Base layer
43
3.3.2
SMMR data layer
44
3.3.3
AVHRR data layer
45
3.3.4
Ground truth data layer
49
3.3.5
Vegetation coverage data layer
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.4
Calculating snow extent and water equivalence
53
3.4.1
Snow depth monitoring using SMMR and SSM/I
53
3.4.2
Snow extent mapping
54
3.4.3
Calculating snow water equivalence
55
3.4.4
Calculating total water equivalence
57
3.5
Verification and analyses
58
3.5.1
Snow extent verification
58
3.5.2
Snow depth verification
59
3.5.3
Snow water equivalence verification
62
4. Results
65
4.1
Annual snow depth changes
65
4.2
Annual snow water equivalence profile
67
4.3
Twelve years snow water equivalence profile
67
4.4
Snow depth distribution profile
70
5. Conclusions
References
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
€
Tables
2.2-1
:SMMR performance characteristics
12
2.2-2
:Regression parameters for SMMR versus SSM/I
2.2.6
:Summary of experiments on the remote
sensing of snow with passive microwaves
27
2.3
:AVHRR band width
29
3.3.3
:NOAA-9 central wave numbers
for AVHRR IR channels
48
3.3.5
:Winter time forest coverage of Quebec
52
3.4.3
:Ground observed snow density
56
14
3.5.3
: Errors for different snow density
64
4.3
: Statistics of snow covers in 1979-1990
68-3
4.3-1
:Statistics of
snow cover of 1979
71-13
4.3-2
:Statistics of
snow cover of 1980
71-14
4.3-3
:Statistics of
snow cover of 1981
71-15
4.3-4
:Statistics of
snow cover of 1982
71-16
4.3-5
:Statistics of
snow cover of 1983
71-17
4.3-6
:Statistics of
snow cover of 1984
71-18
4.3-7
:Statistics of
snow cover of 1985
71-19
4.3-8
:Statistics of
snow cover of 1986
71-20
4.3-9
:Statistics of
snow cover of 1989
71-21
4.3-10:
Statistics of snow cover of 1988
71-22
4.3-11:
Statistics of snow cover of 1989
71-23
4.3-12:
Statistics of snow cover of 1990
71-24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Figures
3.1-1
:Study area and ground data locations
36-1
3.1-2
:Study area in SMMR data layer
43-1
3.3
:Vegetation coverage zones
51-1
3.5.1
:Overlap of AVHRR and SMMR
58-1
3.5.2
:Snow depth verification
60-1
3.5.3
: Snow water equivalence verification
62-1
4.1-1 : Snow depth change between
Dec. 1987 and Jan. 1988
66-1
4.1-2 : Snow depth change between
Dec. 1988 and Jan. 1989
66-2
4.1-3
: Snow depth change between
Jan. 1988 and Feb. 1988
66-3
: Snow depth change between
Jan. 1989 and Feb. 1989
66-4
4.1-5 : Snow depth change between
Feb. 1988 and Mar. 1988
66-5
4.1-6 : Snow depth change between
Feb. 1989 and Mar. 1989
66-6
4.1-4
4.1-7
4.1-8
: Snow depth change between
Mar. 1988 and Apr. 1988
66-7
: Snow depth change between
Mar. 1989 and Apr. 1989
66-8
4.2
:Annual snow water equivalence profile
67-1
4.3
:Twelve years snow extent and
water equivalence profile
68-1
4.4
:Snow distribution profile
70-1
4.3-1
:Water equivalence of Feb. 1979
71-1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3-2
:Water equivalence of Feb.
1980
71-2
4.3-3
:Water equivalence of Feb.
1981
71-3
4.3-4
:Water equivalence of Feb.
1982
71-4
4.3-5
:Water equivalence of Feb.
1983
71-5
4.3-6
:Water equivalence of Feb.
1984
71-6
4.3-7
:Water equivalence of Feb.
1985
71-7
4.3-8
:Water equivalence of Feb.
1986
71-8
4.3-9
:Water equivalence of Feb.
1987
71-9
4.3-8
:Water equivalence of Feb.
1988
71-10
4.3-11: Water equivalence of Feb. 1989
71-11
4.3-12: Water equivalence of Feb. 1990
71-12
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1. Introduction
The distribution and amount of snow cover over Northern
Quebec are extremely important for hydrological applications
as snow is one of the main local sources of water supply and
hydroelectric generation. In Quebec, ground snow surveys are
the conventional data source for estimating the distribution
and amount of snow water equivalent on the ground throughout
the winter. Snow courses have in fact formed the basis of the
Canadian snow survey network for over 50 years.
Recent advances in remote sensing technologies offer the
potential for significant changes in snow survey procedures.
A
combination
of
ground-based,
airborne,
and
satellite
information, and ultimately remote sensing methods alone, may
be the most efficient means of snow survey data collection to
meet a variety of user needs.
A major challenge in the remote sensing of snow cover is
the
development
extent,
of
snow depth,
algorithms
that
determine
snow
areal
and snow water equivalent using sensors
which have all weather capabilities. Microwave sensors are
uniquely suited for the remote sensing of snowpacks. Although
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
visible and infrared techniques have been used for providing
snowpack information,
very little potential is foreseen,
terms of sensing snow depth,
in
because these wavelengths are
limited to sensing surface conditions. Unlike the visible or
infrared
part
of
the
spectrum,
the
relatively
longer
wavelength of microwave energy can penetrate the snowpack and
interact with or obtain information from not only the snowpack
but also the underlying soil surface.
Because of this,
how
microwave energy is scattered back the snow can be utilized to
provide
snow
addition,
clouds
depth
or
water
equivalent
information.
In
microwave energy has the capability to penetrate
therefore
providing
snowpack
information
under
all
weather conditions.
There
are
two
kinds
of
microwave
microwave and passive microwave.
measures
the
energy
radiation
that
of
1982],
1980,
1991]
naturally
using
Microwave Radiometer
upwelling
snowpack
data
1987,
from
(SMMR)
1990],
the
surface
and underlying
[Goodison et al, 1986],
[Chang et al,
active
Passive microwave sensing
is emitted by both
ground. Investigations
1976,
the
sensing:
and
Scanning
[Kunzi et al,
[Foster et al,
Multichannel
on NIMBUS-7 have described the
potential for using passive microwave sensors to monitor snow
cover.
C
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Active microwave sensing uses a radar system that sends
microwave pulses
energy
towards
reflected back
to
the ground and then measures
the
sensor.
The
total
amount
the
of
energy received by the sensor is the sum of backscattering at
the
surface
snowpack,
of
the
snowpack,
and backscattering
at
backscattering
within
the
the surface of the ground
under the snowpack. Because the active microwave sensing uses
a radar system to send microwave pulses, the active microwave
data are very expensive and, especially for global and large
regional surveys of snow data, more difficult to justify.
The
passive
purpose of
microwave
this
data
study
for
is to
snow
evaluate
mapping
and
the
snow
use of
water
equivalent calculation in for the Province of Quebec and the
Labrador area of Newfoundland.
The specific objectives of the study are to:
1) evaluate passive microwave methods for mapping snow
extent and compare the results with NOAA-AVHRR satellite data
to verify snow extent for selected years.
2) evaluate passive microwave methods for monitoring snow
water equivalent and compare the results with ground measured
snow d a t a .
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
3)
correct for the effects of forest types on snow water
equivalent calculation.
4)
provide a climatological
indication of
interannual
variability of snow cover characteristics from 1979 to 1991.
This thesis contains the following information and new
findings:
1) quantitatively analyzed and verified snow extent and
snow
water
equivalence
maps
derived
from
SMMR
and
SSM/I
passive microwave data for Quebec.
2) that the passive microwave derived snow extent map is
very accurate.
3) that when snow depth is more than 92 cm, the SMMR 18
GHz and 37 GHz responses are out of the linear portion, hence
the maximum snow depth which can be monitored with SMMR data
is s 92 cm.
4) that the snow density varies in important ways and the
errors
on
snow
density
and
forest
cover
have
a
directly
proportional impact on the error in calculated for snow water
equivalence; hence, the accuracy of estimated snow density and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
forest
coverage
is
critical
in
calculating
snow
water
equivalence.
5)
that
there
is
a
linear
relationship
between
extent and total snow water equivalence,
hence
could
total
be
used
for
a
rough
estimate
of
snow
snow extent
snow
water
equivalence.
6) that the annual snow water equivalence profile is a
good indicator of runoff from snowmelt.
This thesis consists of five chapters:
Chapter 1 serves as an introduction to the paper.
Chapter
2 is a literature
review which describes
the
theoretical background on passive microwave remote sensing for
snow cover monitoring.
Important
investigations
of passive
microwave remote sensing on snow extent, water equivalent, and
effects of forest canopy are reviewed.
Chapter 3, mapping areal snow water equivalent,
main part of the paper.
is the
The procedure of snow mapping
for
Quebec is described step by step in this section. The snow
extent and snow water equivalent of Quebec are mapped using
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the formula introduced by Chang and others (1987). The effects
of
forest
are
considered
when
equivalent. The snow extent
calculating
and
snow water
the
snow water
equivalent
are
verified by incorporating high resolution NOAA AVHRR data and
ground measured snow data.
Chapter 4 describes
and discusses
the
results
of
the
procedure which includes a twelve year period of snow water
equivalent
maps,
annual
snow
accumulation
and
snow
depth
change maps, snow frequency distribution, and interannual snow
variability profiles.
Chapter 5 discusses the conclusion of the research and
also indicates future topics for research.
C
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
Chapter 2 - Background and Literature Review
The technology of microwave remote sensing dates from the
mid-1970s.
The
estimation
of
snow
prediction of snow melt runoff,
extent
and
depth,
and
have advanced rapidly since
the launching in June 1987 of a satellite with capabilities
especially designed for this purpose. Significant experiments
have been carried out in northern regions (Canada, Greenland,
Finland) and mountainous regions (the Rockies and Cascades).
My
Quebec
research
relies
heavily
on
the
methods
of
calculation by A. T. C. Chang [Chang et al, 1976, 1979, 1987,
1990,
2.1
].
Theoretical background
Microwave remote sensing can be accomplished either by
measuring emitted radiation with a radiometer (passive) or by
measuring the intensity of the return of a microwave signal as
radar (active)
the
natural
[Hall et al, 1984]. For passive microwave data
microwave
emission
is
recorded
as
brightness
temperature (Tb) , which is expressed in units of temperature
(Kelvin) because for wavelengths in the microwave range the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
radiation
€
emitted
from
a
perfect
emitter
proportional to its physical temperature
(black
(T)
body)
is
[Chang et al,
1987] .
Most real objects emit only a part of the radiation that
a black body would emit at its physical temperature. The ratio
of the emission from a body to that of a black body at the
same temperature is the emissivity,
(e), which is defined as:
e = Tb/ T .
This
radiometry
object
is
a
fundamental
[Zwally et
depends
very
al,
much
equation
of
1977] . As
on
its
passive
the
microwave
emissivity
composition
and
of
an
physical
structure, measurement of emissivity provides information on
the physical properties and conditions of the emitting medium.
The microwave emission from a layer of snow over a ground
medium consists of two contributions: 1) emission by the snow
volume and 2) emission by the underlying ground [Stiles et al,
1979] . Both of
the
two
contributions
are
governed
by
the
transmission and reflection properties of the air-snow and
snow-ground interfaces and by the absorption or emission and
scattering properties of the snow layer [Stiles et al, 1981].
The
electrical
properties
dielectric constant
of
snow
are
described
by
the
(e) . Liquid water in a snowpack changes
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the dielectric constant of snow because liquid water has very
high dielectric losses. For wet snow, the dielectric losses
become large and scattering is unimportant. Therefore, a wet
snowpack
radiates
like
a
black
body
at
the
physical
temperature of the snow layer and is indistinguishable from a
snow free soil surface [Kunzi et al, 1982].
For dry snow, the dielectric losses are very small and
the extinction coefficient is dominated by scattering. As an
electromagnetic wave emitted from the underlying earth surface
propagates
through
the
snowpack,
it
is
scattered
by
the
randomly spaced snow particles in all directions. The dry snow
itself absorbs very little microwave energy
from the wave
propagation and contributes very little to radiative emission.
When
the
snowpack
grows
deeper,
more
microwaves
from the
ground will be back scattered. This causes the scene to have
a
lower
brightness
experiment
temperature.
For
Ka-band
(26-40
GHz),
shows that the penetration depth is upto 87 cm
[Lytle et al, 1994].
The effects of scattering of microwave radiation by snow
crystals
has
been
discussed extensively
in the
literature
[Chang et al, 1976, Kong et al, 1979, Rango et al, 1976, Fung,
1981, Tiuri et al, 1981, Kunzi et al, 1982, and Rott et al,
1989] . The scattering occurs as a result of the dielectric
differences
between the air and
ice crystals.
For a given
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
wavelength,
large snow grains cause more scattering,
which
means scattering efficiency will increase with snow crystal
size and frequency [Ulaby et al, 1980, Richter et al, 1991]
Longer wavelengths
( >15cm
.
) are insensitive to snow depth
variations and the snowpack will appear relatively invisible
to the sensor1s wavelengths.
To summarize, microwaves have the ability to penetrate
deep into dry snow cover and offer additional information for
calculating
obtained
the
snow
water
from optical or
equivalent,
infrared
which
sensors
can
not
[Stiles
be
et al,
1981]. In addition, only microwaves can be used under cloudy
conditions which makes successive observations of microwave
data available without concern for weather conditions.
2.2
SMMR and SSM/I data
From November 1978, the Scanning Multichannel Microwave
Radiometer (SMMR) on the Nimbus-7 satellite has been acquiring
passive microwave data that can be used to measure snow extent
and calculate snow water equivalent on an areal basis.
The SMMR is a 10 channel instrument which receives both
horizontal
0.81,1.4,
and
vertical
polarizations
at
wavelengths
of
1.7, 2.8, and 4.6 cm [Gloersen et al, 1977, 1984].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The band width at each channel
is
250 MHz.
There are six
separate radiometers on SMMR, one for each wavelength but for
the 0.81 cm channel there are two radiometers, one for each
polarization.
Horizontal
and
vertical
polarizations
are
observed on alternate scans for the four longest wavelengths
and on each scan at the 0.81 cm wavelength [Gloersen et al,
1977,
1984],
The raw SMMR data are calibrated and remapped into evenly
spaced cells across the swath and stored on 'CELL' tapes. Four
different
cell
sizes
are used:
156*156
km
(all channels),
97.5*97.5 km (all but 4.6 cm channels), 60*60 km (0.81, 1.4,
and 1.7 cm channels),
and 30*30 km
(0.81 cm channels).
The
purpose of the remapping is to permit proper multispectral
retrievals of the geophysical parameters and to reduce the
radiometric noise inherent in the instrument [Gloersen et al,
1984] .
The calibration of the SMMR data consisted of two major
procedures:
first,
radiometric
data
digitized
signals
conversion
into
of
normalized
received
(Cr)
radiometer gain variations by
the
raw
counts
have
N
been
SMMR
in
digital
which
the
adjusted
for
interpolation between counts
obtained when observing the warm Cw and cold Cc references:
N = (Cr - Cw) / (Cc - Cw) .
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Second,
conversion of normalized counts
into absolute
radiances by means of the relation:
Tb(K) = A + B * N
where A and B are functions of the instrument temperatures and
were
calculated
from
laboratory
measurements
of
the
SMMR
sensors prior to launch.
Table 2.2-1
SMMR Performance Characteristics
PARAMETER
Wavelength cm
Frequency GHz
Bandwidth MHz
MHz
Freq.
K
Dynamic
K
Accuracy
Temp. R a s . K
dB
Noise
Beam Effic . %
Footprint Km
In
June
ch. 1
ch. 2
ch. 3
ch. 4
ch. 5
4 .54
6 .60
250
10-110
10-330
< 2.0
0.9
5.0
87.0
148
2 .80
10.69
250
10-110
10-330
< 2.0
0.9
5.0
87.0
91
1.66
18.00
250
10-110
10-330
< 2.0
1.2
5.0
87.0
55
1.36
21.00
250
10-110
10-330
< 2.0
1.5
5.0
87.0
46
0.81
37.00
250
10-110
10-330
< 2.0
1.5
5.0
87.0
27
1987,
the
U.S.
Air
Defense
Meteorological
Satellite Program launched its "F " satellite which carries
the first spaceborne passive microwave imager,
the Special
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sensor
Microwave
/
Imager,
or
SSM/I. The
SSM/I
passive
microwave data is mainly used for geophysical retrieval of
land
surface
characteristics.
SSM/I
measurements begun in 1978 by SMMR.
continues
the
snow
In contrast with SMMR,
SSM/I provides near global coverage every day. SSM/I operates
at four frequencies:
horizontal
frequency
and
19.35,
vertical
except
22
22.24,
37.0, and 85.5 GHz, with
polarizations
GHz,
which
has
measured
only
a
at
each
vertical
polarization channel [Cavalieri, 1988].
In order to provide SMMR or SSM/I data to the user, NASA
distributes the passive microwave satellite data on CD-ROM.
The first CD-ROM product produced from passive microwave data
was the SMMR gridded brightness temperatures
Polar Region, October 28,
1978 - January 31,
for the North
1980.
The data
were distributed beginning in June 1987.
The
SSM/I
Brightness
Temperature
Grids
for
the
Polar
Regions on CD-ROM, July 9, 1987 - July 31, 1989 is the second
CD-ROM product distributed.
The SSM/I CD-ROMs
are mastered
using the ISO 9660 industry standard. The disks are compatible
with the MS-DOS operating system. Software to access the data
under the MS-DOS is distributed with the SSM/I CD-ROM.
The pixel size of the CD-ROM SMMR or CD-ROM SSM/I map
should not be confused with the resolution of SMMR or SMM/I
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
data. Although the SMMR's 18 GHz (or SMM/I's 19 GHz) channel
and 37 GHz channel have different resolutions
(55 km and 27
km),
channels
on
the
CD-ROM
map,
resampled
to
therefore,
on the CD-ROM,
the
latitude
V40
data
by
V40
the 18 GHz
in
both
longitude
grid
are
cells;
(or 19 GHz for SMM/I)
SMMR map and 37 GHz SMMR map have the same pixel size, or the
same resolution.
The digital number (DN) of the SMMR and SSM/I data on the
CD-ROM provided
by
the
National
Snow and
Ice Data
Centre
(NSIDC) are calibrated brightness temperatures in tenth of a
degree Kelvin. For example a DN of 1000 represents a Tb of 100
K [Chang et al, 1990].
Table 2.2-2
Regression parameters for SMMR versus SMM/I
Slope
Intercept (K)
18h / 19h
0.940
2. 62
37h / 37h
0. 954
2. 85
18v / 19v
0.870
21. 9
37v / 37v
0.861
30 .2
Channel Pair
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
If both SMMR and SSM/I data are used, the correlation for
SMMR data versus SSM/I data should be established. Based on
the experiment by Steffen and others (1993) the correlations
are R > 0.99 with the regression data shown in Table 2.2-2:
Some of
conclusions
about
the passive
microwave
snow
mapping using the SMMR and/or SSM/I data are cited as follows:
2.2.1
Frequency and channel
1)
Boundaries
between
dry snow areas
and
snow
free
areas can be defined by the 37 GHz or 18 GHz data because of
the sharp decrease in brightness temperature when going from
snow free land to a snow surface [Foster et a l , 1991].
2)
against
channel.
Areas with very wet soil cannot be discriminated
snow
With
covered
two
areas
by
frequencies
using
it
is
data
from
possible
to
only
one
separate
clearly wet soil from dry snow [Kunzi et al, 1982].
3)
brightness
For
all
channels,
temperature
as
wet
snow
may
have
snow
free
land.
the
Thus,
same
clear
separation of the wet snow covered areas from snow free areas
is not possible with SMMR data at a single time period [Kunzi
et al, 1982] .
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4)The
^
strongest
information
on
snow depth or
snow
water equivalent is obtained from the 37 GHz data, but there
is also significant information in the 18 GHz data [Kunzi et
al, 1982] . An evaluation of the various algorithms that have
been derived, shows that only algorithms including the 37 GHz
channel provide adequate agreement with the manually measured
snow depth and snow water equivalent values. The use of both
18 GHz and 37 GHz channel often gives better results than the
37 GHz channel alone, because the 18 GHz channel data helps to
reduce the effects of the snow and ground temperatures and the
atmospheric water vapour on changes in Tb [Chang et al, 1979].
2.2.2
Polarization
The
polarization
describes
the orientation
in the xy
plane of the electric field vectors of a wave travelling in
that
direction.
Radiation
emanating
from
an
object
is
restricted to a vertical or horizontal direction of vibration.
At a zero degree view angle, there is no distinction between
the horizontal and vertical components. Any observations at
the nadir are considered to be non-polarized.
The
horizontal
features
of
the snowpack will
tend to
absorb more horizontal components of the microwave electric
field than
the vertical
components.
Vertical polarizations
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
will always
produce higher brightness
while horizontally polarized data,
parallel
to
the
surface,
are
more
temperature readings
with its electric field
sensitive
to
differing
surface conditions [Matzler et al, 1984]. Both vertically and
horizontally
polarized
data
from
37
GHz
or
18
GHz
SMMR
channels are suitable for defining the dry snow boundaries
[Foster et al, 1984].
A problem in assessing snow cover with passive microwave
data
is
the
effect
of vegetation
cover.
For example,
the
polarization difference at the 37 GHz channel deceases rapidly
with increasing vegetation and approaches zero for vegetation
with water
content
greater than
1 kg/m2 [Choudhury et
al,
1987] .
This problem will be discussed in more detail in a later
section.
2.2.3
Snow extent and depth
Several microwave remote sensing methods are available to
retrieve snow cover and snow depth information for specific
regions
and
seasonal
conditions. These
methods
have
been
derived from research using microwave remote sensing from both
satellite and aircraft or field studies.
A straight-forward
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
method is to examine the difference between the brightness
€
temperature observed for snow covered ground and that for snow
free ground. The general form of microwave snow cover method
is:
At =
ts
- Tg
where A t is change in brightness temperature, Ts is observed
brightness
temperature
for
snow
cover
ground,
Tg is
observed brightness temperature for snow free ground.
may
be
either
frequency,
or
the
brightness
the
brightness
temperature
at
temperature
a
the
The T
single
at
several
frequencies or polarizations [Hallikainen et al, 1984, 1992].
Once the A t is known it can be used as a threshold to
determine the snow boundaries. Usually, the 3 7 GHz SMMR data
are
used
because
of
the
sharp
decrease
in
brightness
temperature when going from snow free land to snow surface
[Foster et al, 1991] .
As
multichannel
information
than
a
data
single
can
provide
channel,
gradient between the 18 GHz and 37 GHz
polarization
was
introduced
[Kunzi
an
more
snow
algorithm
cover
using
a
channels in horizontal
et
al,
1982] .
The
algorithm's physical basis is derived from the fact that the
apparent dry snow brightness temperature is a strong function
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of
frequency
and
is
relatively
insensitive
to
physical
temperature changes in snowpack. The algorithm is described as
follows
[Kunzi et al, 1982]:
GT = ( Tb(fx) - Tb(f2) ) / ( fi - f2 )
K/GHz
where GT is the brightness temperature gradient, f3 and f2 are
the higher and lower frequencies of the sensor.
To determine the presence or absence of dry snow cover,
a threshold (D) is selected such that
If GT <= D,
dry snow is present,
If GT
no dry snow.
> D,
The value of D is determined empirically and adjusted for
the best correspondence between SMMR data and ground d a t a .
Cavalieri
brightness
negative
(1984)
also used the combination of the two
temperatures
gradient
to
ratio,
give
a
(NGR),
single
which
parameter,
is
actually
the
a
modification of Kunzi's brightness temperature gradient and is
defined as:
NGR = “1000 * ( Tbl8 - Tb37 ) / ( Tbl8 + Tb37 )
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
where
Tbl8 and Tb37 are
the
brightness
temperatures
at
the
respective frequencies.
This
formulation
removes
the
effects
of
the
physical
temperature which allows changes in brightness temperature to
be more evident.
Because using T18h-T37h often gives
better results than
using other channel combinations [Chang et al, 1990], a method
of differences, between the SMMR 37 GHz and 18 GHz horizontal
channels,
is
temperature
used
to
derive
relationship
for
a
a
snow
uniform
depth-brightness
snow
field.
The
equation developed i s :
SD = 1.59 * ( T 18h - T37h ) = 1.59 * A T b
where SD is snow depth in cm, h is horizontal polarization,
and 1.59 is a constant derived by using the linear portion of
the 37 GHz and 18 GHz responses to obtain a linear fit of the
difference
between
the
brightness temperature
two
is
frequencies.
less than
the
If
the
37 GHz
18
GHz
brightness
temperature, the snow depth is zero, no snow cover is assumed
[Chang et al, 1990] .
C
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
c
2.2.4
Snow water equivalent
The snow water equivalence can be calculated a s :
SWE = d * depth
where d is the snow density in grams per cubic centimetre.
Assuming a
snow
mean snow density of 0.3 g/cm3 for a uniform
field, the water equivalence of the snow cover can
be expressed as
[Chang et al, 1990]:
SWE = 0.3 * 1.59 * ( T 18h - T37h )= 0.48 *
The
then
gradient algorithm was
extended
to
A T b cm.
determine the
depth ofthe snowpack [Kunzi et al, 1984]:
SWE = 0.3 * ( GT - 0.085 ) / 0.036 cm.
A generalized snow density can be calculated with the
formula [Bilello, 1969]:
d =
where T
0.152 - 0.0031T + 0.019W
is average seasonal air temperature C, W
seasonal wind speed m/sec2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
isaverage
Snow density can also be determined in situ using hand­
held electromagnetic sensors [Denoth, 1989, Jones et al, 1983,
Kendra et al, 1994] .
2.2.5
The effect of vegetation cover
In the 37 GHz and 18 GHz channels, vegetation is a strong
absorber of microwave radiation and dominates the upwelling
microwave radiation [Wang, 1985, Rignot et al, 1994]. It will
influence both passive and active microwave readings.
If the field of view of the sensor includes a mixture of
two or more objects,
and
their
physical
the respective emissivities are e^ . .e„
temperature
are Tx. . .Tn, the
brightness
temperature Tb is approximately a linear combination of these
individual brightness temperatures
[Patil et al, 1981]:
Tb = fi*Tbl + f2*Tb2 + .. . + fn*Tbn
where Tbn=en*Tn and fj_. . .fD are the ground cover percentage of
object l...n within the view field of the sensor.
If
pixel,
there
are
more
than
two
materials
within
a mixed
multichannel signals should be used to determine the
fractional ground cover (f)
[Hallikainenn et al, 1984].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C
In order to remove the forest effects from snow cover
data, the percentage of the scene covered by forest,
be
determined
first.
For
an
individual
mixed
f, must
pixel
the
brightness temperature of the pixel i s :
Tb = f *Tbf + (l-f)*T bs
where Tbf is the brightness temperature of a pure forest pixel
and the Tbs is the brightness temperature of a pure snow pixel,
the coverage of forest in the pixel i s :
f = ( 1b
1bs ) / ( Tbf - Tbs ) .
Because of the large footprint of the SMMR data
(30*30
km) it is difficult to find a pure snow pixel or a pure forest
pixel for a scene. Therefore,
in this research the Advanced
Very High Resolution Radiometer (AVHRR) data will be used to
determine the areal
forest
coverage,
f.
The
resolution
of
AVHRR data is 1.1 km and the infrared AVHRR data is suitable
for separating snow and vegetation [Baglio et al, 1989]. The
equation for calculating the forest coverage within a AVHRR
pixel then becomes:
f' = ( DN - DNS ) / ( DNf - DNS )
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
where DN is the digital number of a AVHRR scene pixel. As the
forest coverage of each individual pixel is known, the areal
forest coverage f can be easily calculated:
f = ( Ef'
) / n
where n is the number of pixels included in the scene.
Once the forest coverage is known the following equation
can be used:
Tbr = Tb - ( f * Tbf )
where Tbr is residual brightness temperature , Tbr= (1- f )*Tbs, and
Tbf is the temperature brightness of the forest which can be
calculated a s :
Tbf = e f * T
where ef is the emissivity of the forest, approximately 0.90,
and the T is the average of the maximum air temperature.
This algorithm was successfully used to remove forest
effects from SMMR data, using the 37 GHz channel [Hall et al,
1984] .
The
shortcoming
of
this
algorithm
is
that
the
emissivity of the forest and the average of the maximum air
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
temperature are used as parameters. Usually, the desired air
«
temperature data are not available.
Assuming the emissivities of forest for both 18 GHz and
37 GHz are the same, the brightness temperature difference for
a SMMR pixel will be:
= (1-f) *
A T br
ATb
where f is the forest coverage within the pixel,
remotely
sensed
residual
brightness
temperature
A T br
is the
difference
from which the effects of forest cover have been removed, and
the
ATb
is the brightness temperature difference of a pure
snow cover pixel.
smaller
than
A T b,
If f is non zero,
the
SWE
derived
the
with
will always be
A T br
A T br
then
will
be
smaller than the actual SWE. To remove the effects of forest
and get the actual SWE the equation
SWE = 0.48 *
A T br
/ (l~f) cm3
should be used [Foster et al, 1991] , where the average forest
coverage
f can
be
derived
with
the
AVHRR
data
or
ground
measured data.
One of the advantages of this method is, with the ground
measured snow cover parameters as training data,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the actual
winter-time
€
forest coverage can be derived
from the above
equation [Foster et al, 1991]:
f = 1 - ( 0.48 * A T br / SWE )
where f is the forest coverage within a SMMR pixel, 0.48*ATbr
is the remotely sensed snow water equivalence, and SWE is the
ground measured snow water equivalence.
This approach is very useful because of the difficulty of
getting winter time vegetation coverage
information over a
remote a r e a .
2.2.6
Passive microwave experiments
A number of different kinds of passive microwave remote
sensing experiments
have been performed.
These experiments
include laboratory experiments, small scale field experiments
which
use
hand-held
using
airborne
or truck mounted
sensors,
spaceborne satellite.
and
sensors,
experiment
using
experiments
data
from
Truck and aircraft data usually have
been used to verify radiative transfer models. Satellite data,
on the other hand,
are especially useful
for assessing the
venture of large area remote sensing of snowpacks [Foster et
al, 1991].
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Table
c
2.2.6
lists
the
numerous
passive
microwave
experiments conducted through 1995.
Table 2.2.6
Summary of Experiments on the Remote Sensing of
Snow with Passive Microwaves
1962
1971
1973
1976
1979
1979
1980
1980
1981
1981
1982
1982
1982
1982
1986
1987
1987
1988
1990
1991
1992
1995
Platform
Wavelength
Mt. Rainer
laboratory
Western USA
N. Hemisphere
Colorado
Canada
US, Canada
Colorado
US
Finland
Greenland
Finland
N . Hemisphere
Michigan
Saskatchewan
N . Hemisphere
US
Finland
US
US
Finland
Ross Sea
air-borne 1.55
0.81, 2.2
air-borne 1. 55
Nimbus-5 0.95, 1.35
air-borne 1.7, 2.1
Nimbus-6 0 .81
Nimbus-6 1.55, 0.81
truck
0.3,0.81,2.8
Nimbus-7 0 .8,1.66,2.8
Nimbus-7 0.8,1.4,1.66
Nimbus-6 0.94, 1.34
tower
0.81,2.5,6.0
Nimbus-7
0 .8,1 .66,2.8
0 .8,1 .66,2.8
Nimbus-7
Nimbus-7
Nimbus-7
0 .8,1.66,2.8
Nimbus-7
0 .8,1.66,2.8
Nimbus-7 0 .8,1.66,2.8
0 .8,1.66,2.8
Nimbus-7
Nimbus-7 0 .8,1.66,2.8
0 .8,1.66,2.8
Nimbus-7
Nimbus-7
0 .8,1 .66,2.8
to
00
Meier,
Meier et al,
Schmugge,
Kunzi et al,
Chang et al,
Rango et al,
Foster et al,
Ulaby et al,
Buark et al,
Hallikainen,
Rotman et a l ,
Tiuri,
Kunzi et al,
Hall et al,
Goodison,
Chang et al,
Choudhury,
Hallikainen,
Chang et al,
Foster et al,
Hallikainen,
Zibordiet al
Location
o
00
Reference
Some of the more important and widely cited experiments
from Table 2.2.6 a r e :
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1)
The experiment conducted by Kunzi and others (1982) .
The goal of the study was to show that how the three snow
cover parameters - extent, water equivalent, and onset of snow
melt can be determined using SMMR data.
2)
The
experiment done
by Chang
and others
(1987),
which introduced the multi-channel algorithm and the equation:
SW = 0.3 * 1.59 * (T18h - T 37h) .
3)
The research of Foster and others (1991), examined
how trees affect the radiation reaching the satellite,
and
introduced an algorithm to estimate snow water equivalent in
the boreal forest regions.
2.3
NOAA AVHRR data
An AVHRR sensor can be used to verify the extent of snow
and the accuracy of the snow boundary. This is how the AVHRR
data
will
be
used
in
this
study.
The
following
summary
indicates the limitations of AVHRR sensors and why they cannot
provide information about snow depth.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The
sensor
Advanced
aboard
Very
the
High
NOAA
Resolution
polar
Radiometer
orbiting
(AVHRR)
satellites
is
a
radiometer designed to measure the upwelling radiance from the
earth and atmosphere in visible,
infrared regions
sensor
is
a
5
of
the
channel
near infrared,
spectrum
[Rango,
instrument.
The
and thermal
1986].
The AVHRR
visible
and
near
infrared channels (channels 1 , 2 ) use 0.1 square inch silicon
detectors to measure incident radiation. The infrared channels
(channels
3,
4,
5)
use
cooled
detectors
made
of
indium-
antimonide (channel 3) and mercury-cadmium-telluride (channel
4, 5) .
Table 2.3
AVHRR Band Width
Band Width mm
Channel #
1
2
3
4
5
0.58
0.72
3.55
10.3
11.5
-
0.68
1.10
3.93
11.3
12.5
The instantaneous field of view of all AVHRR channels is
1.4
(
milliradians,
resulting
in
a
ground
resolutions
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of
approximately 1.1 km at nadir for an altitude of 833 km. The
^
acquisition repeat-cycle of AVHRR is 12-hours,
and one night for each calendar day
i.e. one day
[Moreno et al,
1993],
[Mouginis et al, 1994]. The AVHRR data is shipped in 10-bit
precision, uncalibrated format on either 1600 BPI CCTs or 6250
BPI CCTs.
Two major disadvantages of the AVHRR sensor, that limit
the use of AVHRR as primary sensor
for operational remote
sensing of seasonal snow cover, are:
1 ) neither
visible
and
infrared
wavelengths
can
penetrate into the snow cover, therefore the AVHRR
sensor can not derive any information about snow
depth. This means only snow extent information can
be monitored with AVHRR sensor.
In this study the
AVHRR data are used to verify the accuracy of the
snow boundary derived with the passive microwave
MSSR dat a .
2) due to the sensor characteristics of the AVHRR
(infrared
clouds)
radiation
can
not
penetrate
through
many of the AVHRR scenes are obscured by
clouds. In fact, for the Quebec area it is hard to
find
successive
cloud
free
AVHRR
data
continuous snow cover monitoring purposes.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for
The
AVHRR
sensor
has
some
advantages
in
monitoring
seasonal snow cover. The ground resolution of 1.1 km at nadir
is available globally. The Landsat Thematic Mapper (TM) and
Multispectral
Sensor
System
(MSS)
instruments
offer
much
smaller resolutions (30m - 80m) than AVHRR, but with a repeat
cycle of 16 days, they are too infrequent for use as a primary
sensor
for operational
remote
sensing
channel wavelengths of the AVHRR,
channels,
offer
greater
of
snow
cover.
The
being similar to the TM
capabilities
for
multispectral
analyses than does the MSS which has only 4 channels,
all
limited to the visible and near infrared regions.
Experiments with AVHRR data on distinguishing between
clouds and ice or snow covered surfaces have been conducted
[Allen et al, 1990],
et al, 1989],
al,
1986],
1987] . An
[Ebert, 1989],
[Key, 1990],
[Saunders
example
[Saint et al, 1981],
et al,
of
[Gesell, 1989],
cloud
1988],
and
[Harrison
[Saunders et
[Yamanouch et
contamination
procedures
al,
was
conducted by Allen et al (1990) through repetitive compositing
of
several
images
and
selection
of
pixels
with
minimum
brightness. Analyzing bivariate histograms of AVHRR channels
1 vs 4 and 2 vs 3 enabled Saint et al (1981) to discriminate
between vegetation, clouds, mist, and snow.
After the cloud is removed,
Allen et al
(1990)
use a
change detection classification method to monitor the snow
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cover. The fundamental assumption of this method is that the
significant increases in measured reflections in a study area
from summer to winter,
resulted
primarily
snowpack.
This
from
excluding
the
any
cloud
reflection
technique requires
that
contamination,
off
the
a summer
seasonal
(no snow)
image be acquired and that the summer and all winter images be
normalized
to
the
same
solar
illumination
conditions.
The
summer image data then are subtracted from the winter image
data to produce the snow map.
A
multispetral
snow/cloud
distinguishing
procedure
introduced by Baglio and others (1989). They a use 2-dimension
image classification approach to distinguish cloud from snow
and ice.
The AVHRR channel 1 is used as one dimension and
channel 3 minus channel 4 is used as the second dimension.
This technique enables the assignment of each pixel on the
AVHRR
scene
a
three
partition:
snow,
land,
and
cloud
or
mixture. The pixels with low values on both dimension 1 and
dimension 2 will be assigned as land, pixels with low values
on dimension 1 and high values on dimension 2 will be assigned
as snow, and pixels with high values on dimension 1 will then
be assigned as cloud.
The theoretical background of this method is as follows:
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1)
In the visible
snow
and
cloud
and
are
near
bright
infrared bands,
while
other
both
surface
features are darker.
2) In the middle infrared band,
snow becomes dark
while cloud remain bright but land can be dark.
These features with the use of the two AVHRR bands (one
visible and the other middle infrared), are sufficient for
most snow discrimination exercises.
The AVHRR band 3, however, is not a pure middle infrared
band
but
thermal
is on
the
infrared.
border
In the
between
daytime,
reflected
infrared
both thermal
and
energy and
reflected solar radiation are added to the signal,
so AVHRR
band 4, which is a thermal infrared band, must be employed.
AVHRR band 4 is subtracted from band 3, so that the residual
approximate the reflected middle infrared component.
Sakellariou et al
(1988) used different dimensions for
the purpose of multispectral classification. AVHRR channel 2
is used as dimension 1 and AVHRR channel 2 minus channel 1 is
dimension 2. The cloudy pixels are distinguished from clear
pixels
by assigning
all
clear
pixels
to one of
the three
different surface types. In this scheme, a snow pixel has high
volume on dimension 1 and low volume on dimension 2 , a land
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
pixel has low value on dimension 1 and high value on dimension
2 , an open water pixel has low values on both dimensions 1 and
2,
and
the
cloudy
pixels
have
moderate
values
on
both
dimensions 1 and 2 .
This technique was modified by Li et al (1991) to deduce
the
nature
of
the
surface
under
a
thin
cloud.
The
same
dimensions with AVHRR channels 1 and 2 are used and the cloudy
pixels are further classified to land, ocean, and snow or ice.
For example, a cloudy pixel with a high value on dimension 2
should be assigned as land, a cloudy pixel with a low value on
dimension 2 will be assigned as snow, and a cloudy pixel with
moderate values on dimension 2 will be assigned as ocean.
A considerable amount of material on snow cover mapping
has been published which utilizes AVHRR data; however, there
is a restricted selection of imagery due to cloud free dates
[Sutherland et al, 1986, Zibordi et al, 1995] .
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(
Chapter 3 - Snow Mapping Procedures
This
chapter
mapping areal
three
major
describes
snow extent
steps:
(1 )
my
procedures,
using
GIS
for
and water equivalent. There are
preparation
of
data
layers,
(2 )
calculation of snow extent and snow water equivalent, and (3)
verification of the results. After describing the procedures
for steps 1 and 2 , closer attention is given to the exercise
in verification because it indicates how procedures must be
adjusted
in order to make
SMMR and SSM/I
a practical
and
reliable tool for predicting snow melt runoff.
The three major steps in the procedure are:
1 ) prepare the data layers:
- merge SMMR images,
- interpolate, and calibrate AVHRR images,
- register ground truth data and calculate the mean
snow density,
- calculate vegetation coverage.
2 ) calculate snow extent and snow water equivalent:
- the calculation is based on the formula introduced by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
<
1 :20
1 mm
.000
20 km
3
Fig. 3.1-1
Study area and ground data locations
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chang, A. T. and others (1987).
- the effects of forest are considered when calculating
the snow water equivalent.
- the calculation is performed with a GIS map overlay
function.
3) verify the results:
- overlay snow extent maps derived from SMMR with AVHRR
data to verify the SMMR snow extent map,
- verify snow depth maps using ground truth data,
- verify water equivalence maps using ground truth data.
Chapter 3 .3 will describe the data layers. Chapter 3.4
describes the snow mapping procedures. Chapter 3.5 describes
the verification of the snow maps.
3.1
Study area
The study area encompasses the Quebec Province and the
Labrador area of Newfoundland Province, i.e., the land between
45°N and 63°N latitude, and 52°W and 80°W longitude. The total
area is about 1,550,000 km2. Figure 3.1-1 shows the study area
at a 1:20,000,000 scale on a Mercator projected map. Figure
3.1-2 shows the study area using a stereographic projected
SMMR m a p .
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
1
: 20,000,000
Fig. 3.1-2
Study area in SMMR data layer
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Quebec
ranks
as
Canada's
largest
province,
occupying
roughly 15 percent of the country1s landmass. The study area
spans nearly 1500 km from east to west and 2000 km from north
to south.
This
area
offers
climates:
from
the
a
striking
often-steep
rang
of
landscapes
shores
of
the
and
mighty
St
Lawrence to the cultivated terraces of the Appalachian valleys
to the wide-open spaces of the northern tundra.
The St Lawrence Lowlands are lodged between the Canadian
Shield,
to the north, and the Appalachian Mountains,
to the
southeast. This region is the northeastern extension of a much
larger
farther
platform
into
the
of
unfolded
sedimentary
continent's
interior.
rocks
The
extending
lowland
rise
gradually to the northeast so that the area around Quebec City
has
a
higher
elevation
(100m
above
sea
level)
than
the
Montreal plain, which rarely surpasses the 70m mark.
The Appalachian Mountains cross into Quebec from Vermont
and New Hampshire and run to the northeast along the boundary
with the US and New Brunswick.
The highest summits
heights
most of the hilly
of
1100m to
1200m,
but
rarely rises above 500m.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
attain
terrain
The Canadian Shield covers over 80 percent of the study
area. The granitic and gneissic rocks of this immense craton
are
the
roots
of
ancestral
mountain
ranges
that
were
repeatedly uplifted and eroded over billions of years. Today
this
expanse
is
generally
higher than
600m above
Pleistocene
glaciations
flat
sea
and
level.
created
monotonous,
Glacial
a
rising
abrasion
grainy,
no
during
striated
land
surface interspersed by rocky knobs. Glacial meltwaters filled
valley bottoms with sand and gravel accumulations impounding
millions
of
lakes.
Today
16
percent
of
the
area
is
still
covered by inland lakes. The landforms of the Shield consist
of extensive plateaus interrupted by a few mountain massifs.
Only near the rim of the Shield is the land deeply incised by
rivers flowing towards the surrounding lowlands.
Owing to its high latitude and location at the eastern
margin
of the continent,
fluctuations
the
in temperature.
study area
Cold winters
undergoes
and
extreme
surprisingly
warm periods during the summer months are the hallmarks of a
continental climate. With increasing northern latitude summers
become cooler and winters turn frigid, but the sizeable gap in
seasonal temperatures remains. For example, Montreal's average
summer temperature is 22°C, and the winter averages -9°C. For
KUUJJUARAPIK,
Hudson
Bay,
an
the
Inuit
settlement on the eastern
equivalent
figures
are
11°C
shore of
and
-23°C.
Precipitation is abundant and locally reinforced by the nearby
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
open seas of the Atlantic Ocean and Hudson Bay. Annual totals
average between 35cm and 110cm. The amount of annual rainfall
generally decrease inland and northward and is fairly evenly
spread between summer rains and winter snows.
Latitude is an important factor in the distribution of
natural
vegetation
cover
since
it
largely
determines
the
length and average temperatures of the growing season. In the
study area,
the growing season for the northern portion is
less than 40 days long and for the southern portion is more
than 180 days.
Most
common
in the southern portion of the area,
the
mainly broad-leaf hardwood or deciduous forest is dominated by
maple species mixed with beech,
oak.
This
areas.
becomes
forest persists
With
increasing
interspersed
hickory,
in mostly
latitude
with
ash,
and
hilly and mountainous
and
balsam
basswood,
altitude
fir
and
it gradually
yellow
or
white
the
dense
birch.
In
boreal
the
central
forest
portion
of
the
study
area,
is dominated by straight-trunked,
coniferous
trees. Extending broadly in homogeneous stands, these needleleaf
softwood
growing season.
forests
are
better
adapted
to
the
shorter
Common associations include fir stands with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
white birch or black spruce, lichen-spruce woodlands as well
as Jack pine and birch-aspen stands.
Farther north at the fringes of the hight-latitude boreal
forest
begins
mainly
black
stunted
the
taiga.
spruce,
Well-spaced
white
in their growth,
birch,
clusters
or
of
tamarack
decrease in height to
-
trees
-
already
shrub-like
forms and give way to ground cover such as lichens and arctic
mosses. The tree line marks the northern limit of the taiga.
Two
factors,
low
average
summer
temperatures
and
lack
of
available water, limit the growth and reproduction of trees.
The northernmost vegetational zone, the tundra, has been
called a cold desert.
Interspersed by bedrock outcrops and
field of shattered rocks,
a thin carpet of grasses,
mosses,
lichens, and flowering herbs clings tenuously to the ground.
Widely
scattered low shrubs
of willow and birch manage to
survive only in sheltered pockets. In the tundra, year-round
moisture
is
scarce
and
summers
are too
short
and
cold to
software
program
support the growth of trees.
3.2
Geographic Information System
The
study
is
performed
using
a GIS
entitled GRASS. The techniques needed for snow mapping and
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(
water equivalent calculation include: 1 ) placing the data into
a
geographic
framed
reference;
2)
building
the
spatial
interrelations between data layers; 3) performing snow extent
and water equivalent calculations and verifications;
and 4)
producing
easily
a
high
quality
map
output.
A
GIS
can
integrate and facilitate these tasks.
GRASS
version
4.0
is
distributed
Construction Engineering Research Lab,
runs under UNIX environment.
by
the
Champaign,
US
Army
IL. GRASS
A SUN work station is employed
for running GRASS.
GRASS is a raster based
functions
that
can
program which has many build-in
perform a variety
of tasks. Functions
frequently used in this study are as follows:
"v.digit",
a
function for
labelling,
and
converting
digitizing,
vector
maps.
editing,
In
this
study it was used to input the boundaries of the
studying area.
"i.points",
a
function that enables the
mark control points
user
to
for creation of a coordinate
transformation matrix which is needed for rectify
an image.
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
"i .rectify", a function that rectifies an image by
computing
a
coordinate
transformation
for
each
pixel in the image using the transformation matrix
created by the "i.points"
was
applied
to
every
function.
digital
This function
image
so
that
the
images can be overlaid.
"r.combine",
a function which perform map overlay
procedure for several map layers.
"r.mapcalc",
performs
a
full
set
of
arithmetic
operations on the map layer d a t a .
"r.report",
for the map
a function that lists area statistics
layers. For each category
for a map
layer, the total area is given in units of pixels,
hectares, sq. k m s , sq. miles, or % coverage.
"p.map",
a function to produce a colour hardcopy
from raster or vector map layers.
3. 3
Data base
Five data layers are used in the snow mapping analysis
These layers are: 1) base map layer,
C
2) SMMR data layer,
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
AVHRR
c
data
layer,
4)
ground truth map,
and
5)
vegetation
coverage layer.
3.3.1
Base layer
A 1975 northern hemisphere stereographic map is used as
the
basic
layer.
The
map
projection characteristics.
is
selected
because
of
its
The stereographic projection is
officially used by American Geographical Society for Arctic
and Antarctic maps,
and is widely used for large continent
sized areas and polar aspects.
consideration
is
this
However,
projection
is
the most important
employed
by
NASA
in
representing SMMR and SSM/I satellite data.
Based
northern
on
these
hemisphere
stereographic
map,
properties,
the boundary
of
Quebec
from
the
including
longitude and latitude are digitized into GRASS.
Figure
3.1-2
shows
the
digitized
Quebec boundary
coordinates.
€
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
and
€
3.3.2
SMMR and SSM/I data layer
Twelve years (1979-1990) of winter time (December-April)
SMMR or SSM/I passive microwave data are used for mapping snow
extent and calculating snow water equivalent. All the data are
from the National Snow and Ice Centre in Boulder, CO. The SMMR
and SSM/I are purposely designed for research in the polar
regions.
For
Quebec,
the
combination
of
4
images
from
4
successive days is needed to produce a fully covered Quebec
scene (one data layer). Therefore 240 SMMR and SSM/I images
(12 years * 5 months * 4 days) are combined to produce 60 data
layers for the 60 months.
calibrated brightness
Each layer contains SMMR or SSM/I
temperatures
from 18 GHz
and 37 GHz
horizontal channels.
As both SMMR and SSM/I data are used in this research,
the calibration for SMMR data versus SSM/I data is performed
based on the experiment by Steffen and others (1993).
The
SMMR
and
SSM/I
CD-ROM
data
are
resampled
to
V*0
latitude by V40 longitude grid cells. In this research the 53°N
latitude is used to determine the SMMR and SSM/I pixel size
over the whole research area. Thus, a SMMR or SSM/I pixel is
standardize
for a unit area of
V4°
at
53°N latitude by l
A°
longitude, or for an areal extent of 455 km2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The total SMMR and SSM/I data layer (one image) contains
c
3379 pixels - equal to 1,539,445 km2.
Because the SMMR and SSM/I images and the base layer have
the same map projection, there is no mis-registration problem
in overlaying the passive microwave data layer on the base map
using GRASS.
3.3.3
AVHRR data layer
The AVHRR data layer is created using a multispectral
method with AVHRR visible channel 1 and infrared channels 3
and 4. The pixel resolution of the AVHRR layer is 1.1 km at
nadir.
The high-resolution AVHRR data layer is used to verify
the snow extent determined with passive microwave data.
The
AVHRR data have to be pre-processed before they can be used as
a
data
layer.
interpolation
The
which
stereographic system,
raw
data
to
pre-processing
re-registers
includes
the
raw
(1)
data
image
to
the
(2 ) data calibration which converts the
temperature
values
or
albedo, and
removal.
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
(3)
cloud
In the process of image rectification and interpolation,
internal errors can be produced. Usually,
it is difficult to
find enough control points and get their accurate positions
from an image. Fortunately,
for the AVHRR images,
2048
line,
pixels
of
each
scan
the
earth
among the
location
data
(latitude and longitude) are sampled every 40 pixels starting
at
pixel
25
(25,
65,
105,...,
1985,
2025).
There
are
51
possible earth location values for each scan line. This AVHRR
data
characteristic
makes
the
creation
of
a
coordinate
transformation matrix easier and is essentially error free.
In this study, 30 evenly distributed pixels are selected
to
form an AVHRR scene as part of
the development of
the
transformation matrix. The result of the interpolation shows
that the maximum error, which occurred along the east and west
edges of the image, is 12 pixels (13 km) and within the study
area, the overlapping errors are less than 5 pixels or 6 km.
In
using
AVHRR
bands
3,
4,
and
1
for multispectral
classification of snow mapping, AVHRR thermal data values for
channels 3, 4 need to be converted to temperature values, and
the
AVHRR
converted
visible
to
data
albedo.
values
According
for
channel
to
Kidewell
1 need
to
(1986),
calibration procedures are as follows.
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
be
the
With the AVHRR images, the calibration coefficients are
appended
to
each
scan
line.
The
calibration
coefficients
consist of slope and intercept values for each of the five
channels of the scan line. Once the calibration coefficients
have been extracted,
they must be scaled.
The slope values
must be divided by 230 and the intercept values by 222 . The
scaled slopes and intercepts can then be used for calibration.
For thermal channels 3 and 4 the scaled channel values as
well
as
the
intercept,
are
in
units
of
milliwatts/m2-
steradian-cm'1 per count.
The energy measured by the sensor is computed as a linear
function of the input data values
:
E = S * C + I
where E is the energy value in milliwatts/m2-steradian-cm'1, C
is the digital value of a AVHRR image pixel (ranging from 0 to
1023 counts), and S and I are the scaled slope and intercept
values, respectively. The conversion from energy to brightness
temperature
is
performed
using
the
inverse
of
radiation equation:
T(E) = C2 * v / ( I n ( 1 + C1 * v 3 / E ) )
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Plank's
where T is the brightness
f
temperature
(°K)
for the energy
value E, v is the central wave number of the channel filter
(cm'1),
and
C: and
C2 are
constants
(Cx =
1.1910659*10'5
milliwatts/m2-steradian-cm'4, C2 = 1.438833 cm-°K) .
The central wave numbers v for thermal channels 3, 4, and
5 of NOAA-9 are shown as a function of temperature in table
3.3.3:
Table 3.3.3
NOAA-9 Central Wave Numbers for AVHRR IR Channels
Temperature (0K )
180 - 225
226 - 275
276 - 320
Ch. 3 (cm"1)
2670.93
2674.81
2678.11
Ch. 4 (cm'1)
928.50
929.02
929.46
ch. 5 (cm'1)
844.41
844.80
845.19
For the visible channels, the scaled slope values are in
units of percent albedo/count for the slope and in percent
albedo for the intercept.
The percent albedo measured by the sensor is computed as
a linear function of the input data value:
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A = S * C + I
«
where A is the percent albedo,
C is the digital value of a
AVHRR
I
image
pixel,
intercept values.
and
S and
are
The AVHRR data
the
scaled
slope
of visible channel
and
1 was
calibrated using the above equation.
Cloud cover creates a problem in using AVHRR data,
as
AVHRR scenes are frequently obscured with clouds. A number
experiments on cloud removal have been done,
but
for this
study the problem is solved in a straightforward manner: one
cloud-free
scene
was
selected
from
twenty-two
available
scenes. This scene is used for the snow boundary verification.
3.3.4
Ground truth data layer
The ground measured snow data is issued by Direction des
Reseaux
Atmospheriques,
Gonvernement
du
Quebec.
Ministere
The
data
de
is
1 1Environnement,
collected
from
3 60
stations over Quebec.
The snow data contain two kinds of
location of
station
position
the station and the
information
in
includes
latitude
and
the
information
- the
snow characteristics. The
name of
longitude,
the station,
its
altitude,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
its
its
declination,
and the name of the person who performed the
measurement.
The snow information includes the date of the
measurement,
the
snow
depth,
snow
density,
and
water
equivalent.
For this research, the following dates of ground measured
snow data are used - January 1988 to April 1988 and January
1989 to April 1989.
vegetation
coverage
These data are suitable for calculating
and
verifying
the
passive
microwave
derived snow depth or water equivalence. The locations of the
ground stations are displayed on Figure 3.1-1.
3.3.5
Vegetation coverage data layer
The preparation of the vegetation coverage data layer
includes two steps:
1 )vegetation coverage classification and
2 )fractional forest coverage calculation.
The vegetation coverage classification data are included
in the data base. The average fractional forest cover within
each
SMMR pixel
is
then
estimated
using
a combination
of
ground based measurements and microwave techniques.
The vegetation coverage classification data is taken from
maps published in The National Atlas of Canada (1975), and The
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Vegetation
Chart,
published by
(1986) . Based on these two maps,
the
U.
S.
Army
Air
Forces
the study area is divided
into 3 zones as showed in Fig. 3.3.
Zone 1 is the area above 55° N. This is a tundra zone.
The vegetation within this zone are lichen heath and patches
of needleleaf trees. A fractional forest cover of10% is used.
Zone 2 is the
woodland zone.
area between 55° N And 52° N.
It is a open
The vegetation within this zone consists of
scattered or shrubby needleleaf trees with lichen heath. The
forest coverage of this zone is 20%-40%.
Zone 3 is the area between 52° N and St. Laurent
This is the boreal forest zone.
River.
The vegetation within this
zone consists of needleleaf trees such as spruce, black pine,
balsam fir, and tamarack. The coverage is estimated for this
zone at 30%-60%.
The vegetation coverage for Zones 2 and 3 is calculated
using the equation introduced by Foster and others (1991):
f = 1 - ( 1.59 * A T br / SD )
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3.5
1
:
20,000,000
Fig. 3.3
Vegetation coverage zones
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
where f is the forest coverage within a SMMR pixel, 1.59*ATbr
is the remotely sensed snow depth,
and the SD is the ground
measured snow depth.
Twenty control points with ground measured snow depth are
used
to
calculate
the
winter
time
forest
coverage
using
equation 3.3.5, the results are displayed in Table 3.3.5.
The final vegetation coverage data layer contains three
zones. The average fractional forest coverage within each SMMR
pixel of Zone 1, Zone 2, and Zone 3 are 10%, 31%, and 49%.
Table 3.3.5
Winter Time Forest Coverage of Quebec
Zone 3
Zone 2
SD cm 1 59*ATbr cm
72.6
78 .0
61.5
59.2
79.5
62.0
63.2
66.8
43 .0
57 .0
45 .0
45.0
57.0
51.0
38.0
36.0
f %
41
27
27
24
28
18
40
46
Average f : 31
SD cm
16.9
30.1
31.5
32.5
21.8
22.6
22.8
35.8
35.1
34 .0
15.9
23 .8
1.59*ATbr cm
10 .0
17 .0
14 .0
14 .0
10 .0
14 .0
11.0
19 .0
14 .0
14 .0
10 .0
13 .0
f %
41
44
56
57
54
38
52
47
60
59
37
45
Average f : 49
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.4
Calculating snow extent and water equivalent
3.4.1
Snow depth monitoring using SMMR and SSM/I
€
The
method
used
for
monitoring
the
snow
depth
of
a
uniform snow field is obtained by applying the equation 3.4.1
SD — 1.59 * ( T 18h - T37h )
3.4.1
to the SMMR data layer. SD is snow depth in cm, Tlah and T37h
are
brightness
temperatures
of
SMMR
18
GHz
and
37
GHz
horizontal channels. The value 1.59 is a constant derived by
using the linear portion of the two channels'
responses to
obtain a linear fit of the difference.
Once
the
forest
coverage
is
known
and
assuming
the
emissivities of forest for 18 GHz and 37 GHz are the same, the
brightness temperature difference for a pixel is:
A T br = (1-f) * A T j
where f is the forest coverage within a pixel,
remotely sensed brightness
mixed
pixel,
and
the
A T bs
temperature difference
is
difference of a pure snow pixel.
the
brightness
from the
temperature
If f is non zero, the A T br
will always be smaller than A T bs, therefore,
€
A T br is the
the SD derived
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
with A T br then will be smaller than the actual SD. To remove
«
the effects of forest and obtain the actual SD the equation
SD = 1.59 * (T18h - T37h) / (1-f)
is utilized.
In this research,
the study region was divided into 3
vegetation zones and coverages of 10%, 31%, and 49% which are
the percentages assigned to zones 1 to 3 respectively.
Snow
depth of each zone is calculated separately incorporating the
constant value for each vegetation zone.
3.4.2
Snow extent mapping
The snow extent can be derived from the snow depth maps
where
snow depths are calculated by applying the equation
3.4.1
SD = 1.59 * (T18h - T37h) = 1.59 * A T b
to each pixel of the SMMR data layer.
It is obvious that there is snow when the snow depth (SD)
> 0 and is no snow when SD s 0.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
If
snow
extent
is
the
only
value
of
interest,
the
equation can then be simplified as:
s
-
Tl8h
~
^37h
snow presence
if S > 0
no snow
if S 5 0.
The SD ^ 0 pixels on the passive microwave images define
the boundary of the snow line and thus snow extent.
3.4.3
Calculating snow water equivalence
Once the snow depth is known, the water equivalence can
be calculated a s :
SW = d * SD
(g/cm3)
3.4.3
where SD is snow depth and d is the snow density in gram per
cubic centimetre.
Assuming a mean snow density for a uniform snow field,
the water equivalence of the snow cover is calculated using a
uniform snow density value.
The problem is then how to find
the correct snow density which minimizes the error for the
water equivalence estimation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Some experiments use values of 0.25 to 0.3 as the mean
snow density
[Rango et al,
1979] . The ground measured snow
data shows that 0.3 is too high for the snow cover in Quebec.
Table
3.4.3
is
the
statistical average
result
from ground
measured snow density data.
Table 3.4.3
Ground Observed Snow Density
1986
1987
1988
1989
167
142
69
146
mean
0.228
0.227
0.231
0.225
0.23
o
0.035
0.033
0 .034
0 .037
0 .035
max.
0.34
0.35
0.34
0 .37
0 .35
min.
0.16
0.12
0.16
0.15
0.15
n
Based
density
on
of
Table
0.23
was
3.4.3,
used
a
four year
average
for calculating
the
average
mean
snow
snow water
equivalence for the entire geographical region. The equation
used i s :
SW = 0.366*(T18h-T37h)/(l-f) .
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which is applied to the snow depth data layers to build the
«
snow water equivalence maps. The results will be discussed in
the next chapter.
3.4.4
Calculating total snow water equivalence
The total snow water equivalence for the research area is
the
summation
(pixel)
of
of
the
the water
data
layer.
equivalence
The
data
in
are
each
unit
resampled
area
to V40
latitude by V40 longitude grid cells, and each pixel is counted
as an unit area of 455 km2.
The
equation
for
calculating
the
total
snow
water
equivalence (TSW) in (10‘2 MT) is:
TSW = 455 * 0.366
*
( 2 [T18h(i, j )-T37h(i, j )]/[l-f (zj ] iez1,jez1
+
2 [T18h(i, j )-T37h(i, j )]/[l-f (z2)] iez2,jez2
+
2 [Tlsh(i, j )-T3vh(i, j )]/[l-f (z3)] iez3,jez3 }
where (i,j) is the position of a pixel in SMMR data layer; and
zx, z2, z3 represent zone 1, zone 2, and zone 3 of vegetation
coverage.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Verification and analyses
3.5
<
Verification of results is the key step in confirming and
adjusting procedures.
For technical reasons,
certain scenes
have been selected for evaluating the different aspects.
evaluate estimates of snow extent,
1988
scene
conditions
was
chosen,
permitted
because
for example,
the
comparison
of
a February-
favourable
the
SMMR
To
clear-sky
data
with
a
favorable AVHRR image. For evaluating the snow depth, a scene
of
February
1989
was
selected,
because
of
the
quanity
of
is
to
ground truth data for that particular month.
3.5.1
Snow extent verification
A
cloud-free
AVHRR
high
resolution
image
used
verify the SMMR derived snow extent. Both AVHRR and SMMR data
have the same date in February 1988. AVHRR band 3, band 4, and
band 1 are used to perform the multispectral classification
snow mapping. The AVHRR derived snow map and SMMR derived snow
map are overlaid to display the places where the SMMR derived
snow boundary and AVHRR derived snow boundary are mismatched.
The overlay of AVHRR derived snow map and SMMR derived
snow map are shown as figure 3.5.1.
In the figure, the pink
pixels
the
indicate
the
places
where
SMMR
derived
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
snow
1 : 20,000,000
Fig. 3.5.1
Overlap of AVHRR and SMMR
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
boundary and AVHRR derived snow boundary are mismatched. That
^
is, a SMMR snow pixel is overlain by one or more AVHRR no-snow
pixels (over estimate) or a SMMR no snow pixel is overlain by
one or more AVHRR snow pixels (under estimate).
Two conclusions on AVHRR based SMMR derived snow extent
verification can be drawn from Figure 3.5.1:
1) the result of snow extent derived with SMMR data is
quite good, as most of the SMMR derived snow boundary pixels
are
identical with the
snow boundary pixels on AVHRR data
layer.
2) the error on SMMR derived snow boundary is less than
1 pixel, or 1/4 degree of latitude and longitude, because the
maximum mismatch along the snow boundary is one SMMR pixel in
width.
3.5.2
Snow depth verification and analyses
The SSM/I derived snow depth map of February 1989 was
verified with ground measured snow data. All available ground
snow data
in February
1989
(total of ninety seven control
points) are used. The SMMR derived snow depth on each control
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
point was compared with the ground measured data. The result
€
is shown in Figure 3.5.2.
On
the
regression
figure,
line
and
the
difference
between
the
solid
line
1:1
the
dashed
indicates
the
difference between the mean SMMR derived snow depth and the
mean ground snow observations. The errors are about 3 cm for
a thin snowpack and 4 cm for a thick snowpack. The regression
(0.972) indicates that the equation used for calculating the
snow depth is stable.
The standard errors and t-test for the total of ninetyseven data points are calculated. The a is ±2.372 cm for the
ground measured snow depth and a = ±2.281 cm for the SMMR
derived snow depth.
The standard error of the SMMR derived
snow depth data is less than that of the ground measured snow
depth data, because 1) the snow depth variation information
has
been
sensor,
smoothed
due
to
the
low
resolution
of
the
SMMR
and 2) using equation 3.4.1, the maximum snow depth
which SMMR or SSM/I can monitor is limited to about 90 cm.
In Figure 3.5.2,
for example, we can see that near the
end of the regression line there are five data points at which
the ground measured snow depth is between 97 cm and 107 cm,
but the calculated snow depth is fixed on 90 cm. The reason is
after propagating through 90 cm of snow, most of the microwave
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Y = 3 .9 8 + 1 .0 1 X
R—0.97241
1:1
100
80
Meosured
60
40
Ground
Snow
Oepth
(cm )
120
20
••
0
0
20
40
60
80
100
SMMR Oerived Snow Oepth (cm )
F i g . 3.5.2
Snow depth verification
<
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
emission from the underlying surface is scattered back by the
snow. Therefore, the signal received by SMMR is weak and the
difference
between
the
brightness
temperature
of
18
GHz
channel and 37 GHz channel is no longer sufficient to reflect
the snow depth.
(An experiment [Lytle et al, 1994] shows the
penetration depth for 37 GHz is 87 c m ) . In other words, when
snow depth is greater than 90 cm, for the 18 GHz and 37 GHz,
the
calibration
equation
3.4.1
exceeds
its
domain.
For
calculating snow depth greater than 90 cm, the constant (1.59)
in equation 3.4.1 should be modified to a coefficient which
can
account
for
the
change
between
18
GHz
and
37
GHz
brightness temperatures and then one would obtain a new linear
fit for the difference between the two frequencies.
The result of the t-test on comparison of two populations
is t=1.4096, which indicates the means of the samples are not
different at the 0.05 level.
The
Figure
3.5.2
also
indicates
that
for
most
data
points, the SMMR derived snow depth is slightly less than the
ground observed snow depth, and the difference is consistent,
since the regression line and the 1:1 line are parallel. The
consistent underestimate of snow depth may be caused by using
an incorrect constant in the equation (1.59 has been used) or,
most probably, by an underestimated forest coverage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Forest coverage,
f,
is an important parameter in snow
depth estimating. According to equations 3.4.1 and 3.4.3, an
underestimation of forest cover by ten percent will cause ten
percent
underestimation
on
snow
depth,
thus
ten
percent
underestimation on water equivalence.
Using more
calculating
ground observed
the
forest coverage
increase the accuracy of
3.5.3
data as
control points
with equation
3.3.5
in
may
forest coverage estimation.
Snow water equivalence verification and analyses
Figure
3.5.3
is
the
result
of
water
equivalence
verification. The same ninety seven control points which were
used to construct
Figure 3.5.2 were used
to verify snow water
equivalence. As a
result of using a mean
snow density of0.30
g/cm2 in water
water
equivalence
equivalence
regression
additional
of
error
calculating,
verification
snow
is
depth
is
the
reduced
the
from
of
0.972
(the
to
0.925.
The
variations
of the
snow
verification)
caused by
regression
density ( shown in table 3.4.3 and 3.5.3).
The standard errors for the snow water equivalence data
are the same as the standard errors for the snow depth data
since the equation is only being changed by a constant. They
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ground Meosured
Woter Equivalence
(g/sq. cm)
«
30 i
25 -
20
Y*=0.59+0.96X
R=0.92507
-
15 -
10
-
5 -
0 H----- 1------- 1------ 1------ 1------ 1------ 1
0
5
10
15
20
25
30
SMMR Oerived Water Equivalence (g /s q . cm )
Fig. 3.5.3
Snow water equivalence verification
<
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
are a = ±2.372 cm for ground measured snow water equivalence
data and o = ±2.281 cm for SMMR derived snow water equivalence
da t a .
The
3.5.3
major difference between
is
that
parallel
in Figure
to the
1:1
3.5.3
line.
Figure
3.5.2
the regression
The water
and
Figure
line
is not
equivalence
is
under
estimated by 2% when the water equivalence is higher than 20
g/cm2, and over estimated by 5% when the water equivalence is
lower than 5 g/cm2.
This means that within the research area
the snow density is not uniform but rather variable,
would
expect
for
this
large
geographical
area.
as one
The
snow
density for a thin, loose snow is less than that for a heavily
packed snow. The snow density of 0.23 is adequate as a mean
value for the snowpack where the water equivalence is between
15 and 20 g/cm2, or the depth is between
However,
a snow density
less
than
0.23
65 cm and 85 cm.
should be used to
calculate the water equivalence for a snowpack thinner than 65
cm.
According to equation 3.4.3,
the error for snow water
equivalence estimation is directly proportional to the error
of snow density. Table 3.4.3 shows that the ground observed
average
snow density values vary
therefore,
the
from 0.15
snow water equivalence
to
0.35
g/cm3,
calculated with the
maximum snow density will be 2.3 times (0.35/0.15 = 2.33) as
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
much as the snow water equivalence calculated with the minimum
snow density.
Table 3.5.3 shows the error generated by an inaccurate
snow density:
Table 3.5.3
Errors for different Snow Density
d
error
0.23
0.15
0 .30
0.35
0
-35 %
30 %
52 %
Using more precise variables to evaluate snow density for
equation 3.4.3 can significantly increase the accuracy of snow
water equivalence calculation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t
Chapter 4 - Results
Some of the observations for the 12 years of record will
permit us to make simpler, more economical, and more practical
measures.
optimal
We
can
depths
identify
(paragraph
optimal
4.2)
months
for
(paragraph
estimation
or
4.1),
special
attention. We shall see that the spatial pattern of snow cover
is rather consistent from year to year (paragraph 4.3),
and
that the seasonal pattern is consistent, too (paragraph 4.2).
4.1
Annual snow depth changes
The maps shown on figures 4.1-1 to 4.1-8 indicate the
snow depth changes
images
the
colors
during
the
from yellow
indicated winters. On
to
dark
green
these
represent
an
increase of snow depth from 5 to 50 cm; the colors from dark
blue to red represent a decrease of snow depth from 5 to 50
cm.
The first two figures show the snow depth changes between
December and January. The large dark green area represents a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
substantial amount
(36-50cm)
of snow depth increase during
this period.
Figures
4.1-3 and
4.1-4
show the
snow
depth
changes
between January and February. The yellow and light green areas
stand for a small snow depth increase during this period.
Figures
4.1-5 and
4.1-6
show the
snow
depth
changes
between February and March. The dark blue area shows that in
some
of the Quebec
area
the
snow depth begin
to
decrease
depth
changes
during this period.
Figures
4.1-7 and
4.1-8
show the
snow
between March and April. The large pink and red areas show
that in most of the Quebec area the snow depth has a sharp
decrease during this period.
The snow depth change images like Figures 4.1-1 to 4.1-8
could be used to monitor snow melt and predict runoff caused
by melting
snow.
By observing
the snow depth change
on a
regular base during early spring, runoff is expected when red
areas appear on the snow depth change image.
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
: 20,000,000
Fig. 4.1-1
Snow depth change between
Dec. 1987 and Jan. 1988
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
1
: 20,000,000
Fig. 4.1-2
Snow depth change between
Dec. 1988 and Jan. 1989
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 : 20,000,000
Fig. 4.1-3
Snow depth change between
Jan. 1988 and Feb. 1988
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
+ 36-50cm
+ 21-35cm
+ 5-20c m
- 5-20cm
- 21-35cm
- 36-50cm
1
: 20,000,000
Fig. 4.1-4
Snow depth change between
Jan. 1989 and Feb. 1989
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
+ 36—50cm
+ 21-35cm
+ 5-20cm
- 5-20cm
- 21-35cm
- 36—50cm
1
:
20 ,000,000
70W
Fig. 4.1-5
60W
Snow depth change between
Feb. 1988 and Mar. 1988
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
€
+ 36-50cm
+ 21-35cm
+ 5-20cm
- 5-20cm
- 21—35cm
- 36-50cm
1
:
20,000,000
Fig. 4.1-6
€
Snow depth change between
Feb. 1989 and Mar. 198 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
+ 36-50cm
+ 21-35cm
+ 5—20cm
- 5-20cm
- 21-35cm
- 36-50cm
1
: 20,000,000
Fig. 4.1-7
Snow depth change between
Mar. 1988 and Apr. 1988
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
+ 36-50cm
+ 21-35cm
+ 5-20cm
- 5-20cm
- 21-35cm
- 36-50cm
1
: 20,000,000
Fig. 4.1-8
Snow depth change between
Mar. 198 9 and Apr. 1989
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
4.2
Annual snow water equivalence profile
The same conclusion on annual snow accumulation changes
can also be derived from the annual snow water equivalence
profile of Figure 4.2.
The figure shows that the total water equivalence of the
research area increases sharply between December and January;
the water equivalence reaches the maximum value in February;
and it decreases quickly between March and April.
4.3
Twelve years snow water equivalence profile
Figure
4.3
is the twelve years
equivalence profile,
snow extent and water
calculaed with February SMMR or SSM/I
data.
The snow extent curve and the water equivalence curve on
Figure 4.3 are very much alike, although the snow extent curve
is more smooth. The maximum snow water equivalence occurred in
1988 and the minimum water equivalence is in 1980.
Table
4.3
shows
the
quantity
snow
extent
and
water
equivalence data of the twelve years period. During the twelve
years,
for Quebec,
the maximum snow extent was 1,539,445 km2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
«
140 n
120
c
I—
c
1988
1989
1990
-
o
100 -
o
80 4>
O
C
4>
o
>
'5
CT
UJ
60 -
40 -
20
-
Dec
Jon
Feb
Mor
Apr
Month
Fig. 4.2
Annual snow water equivalence profile
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1989), and the minimum was 1,102,010 km2 (1983). The maximum
snow water equivalence amount was 105,232 billion kg (1988),
and the minimum was 51,456 billion kg (1980).
Based on the data in Table 4.3, the calculated regression
between
the
total
snow
extent
and
the
total
snow
water
equivalence regressions is 0.9230.
The high correlation between the snow extent curve and
the snow water equivalence curve in Figure 4.3 means that the
total snow extent is a good indicator of the total amount of
snow water equivalence. In other words, snow extent could be
used for a rough estimate of total snow water equivalence. For
example, the numerical ranking (highest to lowest) of the six
odd years' snow extent i s :
1989, 1987, 1981, 1985, 1979, 1983
which is very close to the numerical ranking of total snow
water equivalence amount:
1989, 1987, 1981, 1979, 1985, 1983.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
160 -i
c
mean
o
-
•
o
c
•
5
*5
ar
UI
i-
mean
80 -
-
0.8
-
0.4
3a
&
40 -
■79 *80 ‘81 '82 '83 *84 *85 '86 '87 '88 '89 ‘90
YEAR
Fig. 4.3
Twelve years snow extent and
water equivalence profile
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Snow Extont (million aq. km)
120
f
100
-
1.6
-
1.4
-
1.2
-
3
80
60
-
-
r
40
1978
Fig. 4.3
1980
1982
1984
1986
1988
1990
Twelve years snow extent and
snow water equivalence
f
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
1.0
Snow Extent (million sq. km)
Water Equivalence (billon Ton)
120 -i
€
Table 4.3
Statistics of the Snow Cover in 1979-1990
year
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
total
snow
pixels
2, 968
2, 548
2, 995
2, 989
2,422
2,971
2, 971
2, 958
3, 015
3, 204
3, 219
3, 201
total snow
cover area
( km2 )
1,350,440
1,159,340
1,362,725
1,359,995
1,102,010
1,351,805
1,351,805
1,345,890
1,371,825
1,457,820
1,464,645
1,460,550
total water
equivalence
( billion kg )
84,557.20
51,456.05
95,991.35
81,495.05
53,457.95
88,242.20
83,315.05
72,067.45
101,979.15
105,232.40
104,731.90
97,770.40
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figures 4.3-1 to 4.3-12 are maps depicting twelve years
«
of snow water equivalence values in February. The colour codes
for the digital images are:
yellow:
0.1-3 g/cm2
light green:
4-8
g/cm2
dark green:
9-11 g/cm2
light blue:
12-15
g/cm2
dark blue:
16-19 g/cm2
pink:
>20
g/cm2
Figure 4.3-2 is the snow map of February 198 0,
the year
which had the minimum total snow water equivalence during the
twelve years. On this map, most of the Quebec area is covered
with thin snow (yellow or green), the entire Labrador area is
uncoloured, which means no snow.
Figure 4.3-10 is the snow map of February 1988, the year
which had the maximum snow water equivalence for the twelve
years. On this map, the entire geographical area of Quebec is
covered with snow, there is no uncoloured areas. 1988 is also
one of the few years in which the snow water equivalence was
>15 g/cm2 (dark blue area). This also coourred in 1987. Unlike
the other maps where Northern Quebec, i.e. the area between
Hudson Bay and Ungava Bay, was covered with thin snow (appears
as yellow, for water equivalence 3g/cm2 or snow depth 13cm),
the
1988
map
is
the
only
map where
the
entire
region
of
Northern Quebec is covered with heavy snow (appears as dark
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
green
€
and
blue,
for
water
equivalence
>15g/cm2 or
depth
>65cm).
Another interesting situation is that all of the twelve
images show that in Quebec, the area where the maximum amount
of snow accumulation occurs (blue and dark blue area) is the
geographical
centre
of Quebec
(about
70W,55N), not the
northern part of Quebec.
While
images
equivalence
provide
4.3-1
to
distribution,
detailed
snow
4.3-12
tables
extent
depict
4.3
and
and
snow
the
snow
4.3-1
water
-
water
4.3-12
equivalence
information for each of the twelve years. The quantity of snow
water equivalence data was calculated with the digital images
using the equations described in 3.4.3 and 3.4.4.
4.4
Snow depth distribution profile
Figure 4.4 shows six years of snow depth distribution for
February.
Each curve represents a relationship between snow
having a specific depth and the area occupied by the snow
cover. The total snow extent, therefore, is equal to the total
area under the curve. The figure shows that the shapes of the
curves are similar, which means the snow depth distributions
are approximately the same for these six years. It also shows
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
600
1980
1982
1984
1986
1988
1990
500
No. of Pixels
400
300
A
200
'/A
it V /
S-•*:
y \
100
v ./
100
20
40
60
Snow Depth (c m )
Fig. 4.4
Snow distribution profile
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
that areas where the snow depth is greater than 20 cm (water
c
equivalence of 4.5 g/cm2) accounts for the maximum area.
Although the 20 cm depth snow accounts for the largest
percent of total snow covered area,
total snow water equivalence.
it does not reflect the
In other words,
there
is no
immediate relationship between the area of 20 cm snow and the
total snow water equivalence.
A larger 20 cm snow coverage
does not mean a greater amount of total snow water equivalence
for a given month or years.
On the other hand, the 60 cm snow depth area does reflect
the total snow water equivalence. A larger 60 cm snow depth
(dark blue) area will lead to a greater amount of total snow
water equivalence. This is because a larger heavy snow area
normally means a larger total snow extent and the 60 cm snow
area,
itself,
equivalence.
contributes
a
large
amount
of
snow
water
As heavy snow tends to be concentrated in the
centre of Quebec,
this area may be used as an index for the
province-wide snow conditions.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t
20-22g/cm2
16-19g/cmr
12-15g/cmr
/
8-llg/cm2
4-7g/cm2
\
0.1-3g/cm2
1
: 20,000.000
Fig. 4.3-1
Water equivalence of Feb. 1979
t
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20-22g/cm2
lS-lOg/cm2
12-15g/cm2
8-llg/cm2
4—7g/cm2
x
0.1-3g/cm2
1
: 20,000,000
60N
50N
70W
Fig. 4.3-2
60W
Water equivalence of Feb.
1980
t
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
20-22g/cm 2
16-19g/cirf
12-15g/cm?
/
8 -llg /c m 2 ^
4—7 g / cm?
0.1-3g/cm2
\
1
: 20,000,000
Fig. 4.3-3
Water equivalence of Feb. 1981
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
20-22g/cm 2
16-19g/cmr
12-15g/cm2
B-llg/cm2
4—'7 g/cm2 \
O.l-Bg/cm2
1
:
20 ,000,000
Fig. 4.3-4
Water equivalence of Feb. 1982
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20-22g/cm 2
16-19g/cm2
12-15g/cm2
B -llg/cm 2
4—7 g/cirf
0.1-3g/cm2
1
:
20 ,000,000
■W A »A9PI
Mff»
Fig. 4.3-5
Water equivalence of Feb. 1983
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20-22g/cm2
16-19g/cm?
12-15g/cmr
/
8-llg/cm2
4-7g/cm2
0.1-3g/cm2
1
:
20,000,000
Fig. 4.3-6
Water equivalence of Feb. 1984
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
20-22g/cm 2
16-19g/cmr
12-15g/cm2
8 -llg /c m 2
4—7g/cirf
0.1-3g/cm 2
1
:
20,000,000
Fig. 4.3-7
Water equivalence of Feb.
1985
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
20-22g/cm 2
16-19g/cirf
12-15g/cin2
8 -llg /cm 2
4—7g/cirf
0.1-3g/cm2
1
(
: 20,000,000
Fig. 4.3-8
Water equivalence of Feb. 1986
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20-22g/cm 2
16-19g/cm2
12-lSg/cm 2 /
B -llg/cm 2
4-7g/cm 2 \
0.1-3g/cm2
1
:
20,000,000
Fig. 4.3-9
Water equivalence of Feb. 1987
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20-22g/cm 2
16—lOg/cm2
].2-l5g/cwr /
8-llg/cm 2
4-7g/cm 2 X\
0.1-3g/cm2
1
:
20,000,000
Fig. 4.3-10
Water equivalence of Feb. 1988
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
12-15g/cm?
8 -llg /cm 2
4-7g/cm 2
0.1-3g/cm2
1
:
20,000,000
Fig. 4.3-11
Water equivalence of feb. 1989
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
V /
20-22g/cm 2
16-19g/cm2
12—15g/cirf
8 -llg /cm 2
4-7g/cm 2 \
0.1—3g/cirf
I : 20,000,000
Fig. 4.3-12
Water equivalence of Feb. 1990
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
<
Table 4.3-1
Statistics of Snow Cover of 1979
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
20
26
28
29
32
37
total
total
number of
pixels
171
254
238
384
263
295
414
242
223
151
80
126
82
26
7
3
1
1
1
1
1
1
1
2
2,968
total snow
cover area
( km2 )
77,805
115,570
108,290
174,720
119,665
134,225
188,370
110,110
101,465
68,705
36,400
57,330
37,310
11,830
3,185
1, 365
455
455
455
455
455
455
455
910
1, 350,440
total water
equivalence
( billion kg )
778 .05
2,311.40
3,248.70
6,988.80
5,983.25
8,053.50
13,185.90
8,808.80
9,131.85
6,870.50
4,004.00
6,879.60
4,850.30
1,656.20
477.75
218.40
77.35
81.90
91.00
118.30
127.40
131.95
145.60
336.70
84,557.20
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
Table 4.3-2
Statistics of Snow Cover of 1980
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
total
total
number of
pixels
241
391
331
428
330
281
296
120
92
31
7
2,548
total snow
cover area
( km2 )
109,655
177,905
150,605
194,740
150,150
127,855
134,680
54,600
41,860
14,105
3,185
1,159,340
total water
equivalence
( billion kg )
1,096.55
3,558.10
4,518.15
7,789.60
7,507.50
7,671.30
9,427.60
4, 368 .00
3,767.40
1,410.50
350.35
51,465.05
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-3
Statistics of Snow Cover of 1981
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
20
24
total
total
number
of pixels
156
222
171
317
220
230
363
290
355
193
86
108
101
86
74
20
1
1
1
2,995
total snow
cover area
( km2 )
70,980
101,010
77,805
144,235
100,100
104,650
165,165
131,950
161,525
87,815
39,130
49,140
45,955
39,130
33,670
9,100
455
455
455
1,362,725
total water
equivalence
( billion kg )
709.80
2,020.20
2,334.15
5,769.40
5,005.00
6,279 .00
11,561.55
10,556.00
14,537.25
8,781.50
4,304.30
5,896.80
5,974.15
5,478.20
5,050.50
1,456.00
77. 35
91.00
109.20
95,991.35
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-4
Statistics of Snow Cover of 1982
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
32
33
35
36
38
total
total
number of
pixels
151
253
280
463
305
278
389
269
221
130
81
64
58
30
11
1
2
1
1
1
2,989
total snow
cover area
( km2 )
68,705
115,115
127,400
210,665
138,775
126,490
176,995
122,395
100,555
59,150
36,855
29,120
26,390
13,650
5, 005
455
910
455
455
455
1,359,995
total water
equivalence
( billion kg
)
687.05
2,302.30
3,822.00
8,426.60
6,938.75
7,589.40
12,389.65
9,791.60
9,049.95
5,915.00
4,054.05
3,494.40
3,430.70
1,911.00
750.75
145.60
300.30
159.25
163.80
172.90
81,495.05
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
Table 4.3-5
Statistics of Snow Cover of 1983
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
total
total
number of
pixels
298
329
282
355
255
224
238
139
112
82
35
40
25
8
2,422
total snow
cover area
( km2 )
135,590
149,695
128,310
161,525
116,025
101,920
108,290
63,245
50,960
37,310
15,925
18,200
11,375
3,640
1,102,010
total water
equivalence
( billion kg
)
1,355.90
2,993.90
3,849.30
6,461.00
5,801.25
6,115.20
7,580.30
5,059.60
4,586.40
3,731.00
1,751.75
2,184.00
1,478.75
509.60
53,457.95
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-6
Statistics of Snow Cover of 1984
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
total
total
number of
pixels
185
293
253
373
217
234
340
183
208
181
126
135
123
62
45
12
1
2, 971
total snow
cover area
( km2 )
84,175
133,315
115,115
169,715
98,735
106,470
154,700
83,265
94,640
82,355
57,330
61,425
55,965
28,210
20,475
5,460
455
1,351,805
total water
equivalence
( billion kg )
841.75
2,666.30
3,453.45
6,788.60
4,936.75
6,388.20
10,829.00
6,661.20
8,517.60
8,235.50
6,306.30
7,371.00
7,275.45
3,949.40
3,071.25
873 .60
77.35
88,242.70
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-7
Statistics of Snow Cover of 1985
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
total
total
number of
pixels
168
283
230
444
297
325
324
182
222
136
65
110
89
58
23
11
3
1
2,971
total snow
cover area
( km2 )
76,440
128,765
104,650
202,020
135,135
147,875
147,420
82,810
101,010
61,880
29,575
50,050
40,495
26,390
10,465
5,005
1, 365
455
1,351,805
total water
equivalence
( billion kg )
764.40
2,575.30
3,139.50
8,080.80
6,756.75
8,872.50
10,319.40
6,624.80
9,090.90
6,188.00
3,253.25
6,006.00
5,264.35
3,694.60
1,569.75
800.80
232.05
81.90
83,315.05
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-8
Statistics of Snow Cover of 1986
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
total
total
number of
pixels
184
377
287
486
310
329
340
231
162
99
45
53
21
21
10
3
2,958
total snow
cover area
( km2 )
83,720
171,535
.130,585
221,130
141,050
149,695
154,700
105,105
73,710
45,045
20,475
24,115
9,555
9,555
4,550
1, 365
1,345,890
total water
equivalence
( billion kg )
837.20
3,430.70
3,917.55
8,845.20
7,052.50
8,981.70
10,829.00
8,408.40
6,633 .90
4,504.50
2,252.25
2,893.80
1,242.15
1,337.70
682.50
218.40
72,067.45
C
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-9
Statistics of Snow Cover of 1987
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
total
total
number of
pixels
97
208
161
389
284
255
368
238
189
173
90
137
105
95
76
67
45
28
10
3, 015
total snow
cover area
( km2 )
44,135
94,640
73,255
176,995
129,220
116,025
167,440
108,290
85,995
78,715
40,950
62,335
47,775
43,225
34,580
30,485
20,475
12,740
4,550
1,371,825
total water
equivalence
( billion kg )
441.35
1,892.80
2,197.65
7,079.80
6,461.00
6,961.50
11,720.80
8,663.20
7,739.55
7,871.50
4,504.50
7,480.20
6,210.75
6,051.50
5,187.00
4,877.60
3,480.75
2,293.20
864.50
101,979.15
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-10
Statistics of Snow Cover of 1988
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
total
total
number of
pixels
67
229
215
392
437
329
267
306
172
97
87
97
151
201
71
41
36
6
3
3,204
total snow
cover area
( km2 )
30,485
104,195
97,825
178,360
198,835
149,695
121,485
139,230
78,260
44,135
39,585
44,135
68,705
91,455
32,305
18,655
16,380
2, 730
1, 365
1,457,820
total water
equivalence
( billion kg
)
304.85
2,083.90
2,934.75
7,134.40
9,941.75
8,981.70
8,503.95
11,138.40
7,043.40
4,413.50
4,354.35
5,296.20
8,931.65
12,803.70
4,845.75
2,984.80
2,784.60
491.40
259.35
105,232.40
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Table 4.3-11
Statistics of Snow Cover of 1989
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
total
total
number of
pixels
41
104
134
401
564
433
260
310
252
129
166
138
107
119
34
21
6
3,219
total snow
cover area
( km2 )
18,655
47,320
60,970
182,455
256,620
197,015
118,300
141,050
114,660
58,695
75,530
62,790
48,685
54,145
15,470
9,555
2,730
1,464,645
total water
equivalence
( billion kg )
186.55
946.40
1,829.10
7,298 .20
12,831.00
11,820.90
8,281.00
11,284.00
10,319.40
5,869 .50
8,308.30
7,534.80
6,329 .05
7,580.30
2,320.50
1,528.80
464.10
104,731.90
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4.3-12
Statistics of Snow Cover of 1990
water
equivalence
( g/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
total
total
number of
pixels
84
226
197
452
541
290
236
264
239
141
175
128
71
105
47
14
3,210
total snow
cover area
( km2 )
38,220
102,830
89,635
205,660
246,155
131,950
107,380
120,120
108,745
64,155
79,625
58,240
32,305
47,775
21,385
6, 370
1,460,550
total water
equivalence
( billion kg
)
382.20
2,056.60
2,689.05
8,226.40
12,307 .75
7,917.00
7,516.60
9,609.60
9,787.05
6,415.50
8,758.75
6,988.80
4,199.65
6,688.50
3,207.75
1,019.20
97,770.40
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
€
Chapter 5 - Conclusion
The following conclusions can be derived from chapter 3
and 4:
a)
the
passive
microwave
techniques
are
adequate
in
mapping snow extent and estimating snow volume on a regional
basis. Actually, microwave remote sensing is the only way to
monitor on a regular basis the snow depth from satellites.
b) snow extent and snow depth maps can be derived using
the equation
SD = 1.59 * (T18h - T37h) / (1-f)
where f is forest coverage. Determining the forest coverage is
an error generating procedure. By this it is meant that the
error in determining forest coverage is directly proportional
to the error in calculated snow depth.
c) using the above equation the maximum snow depth which
can be monitored with SMMR or SSM/I data is 92 cm.
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
d)
the
forest
coverage
can
be
derived
from
ground
observed snow depth data, with the equation
(1-f) = 1.59 * (Tlah - T37h) / SD
where SD is ground measured snow depth.
e) snow water equivalence is calculated as snow depth
multiplied by snow density.
determining
procedure.
the
To
snow
Snow density is a variable,
density
minimize
the
is
another
errors,
error
ground
and
generating
observed
snow
density should be used.
f) the annual snow water equivalence variation profile
can be used for predicting snow melt runoff. The profile also
provides useful information to climate modelers for validating
how
their
models
handle
temperature
and
precipitation
in
different areas and at different times of the winter.
g) there is a linear relationship between snow extent and
total snow water equivalence. Snow extent could be used for a
rough estimate of total snow water equivalence.
h) for estimating runoff,
it is more useful to evaluate
the maximum extent of snow coverage and the extent of deep
snow (>60cm), rather than the modal or mean depth. Evaluations
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in central Quebec may be indicative of the conditions across
the province.
Future
work
on
passive
microwave
snow
mapping
might
include:
- Incorporate
seasonal
and
geographical
variations
of
density value, which would describe more of the spatial
variation
in
snow
density,
and
thus
provide
a
more
scientific basis for making rational adjustment to the
snow water estimate. Instead of using one snow density as
a constant in calculating snow water equivalence,
use
more control points to build a interpolated snow density
layer which assign unique snow density to each pixel.
- Develop a better ground truth method or an independent
remote sensing measure for forest cover. Instead of using
one forest coverage as a constant in calculating snow
depth for each vegetation coverage zone, use more control
points
to build
a
interpolated
forest
coverage
which assign unique forest coverage to each pixel.
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
layer
c
REFERENCES
Allen, R. C ., Durkee, P. A., and Wash, C. H., 1990, Snow/Cloud
Discrimination with Multispectral Satellite Measurements,
Journal of Applied Meteorology, V29, p994-1004
Baglio,
J.
V. ,
and
Holroyd,
E.
W. ,
1989,
Methods
for
Operational Snow Cover Area Mapping Using the Advanced
Very
High
Resolution
Radiometer,
Research
Technical
Report, 8Ip
Bernier,
P. Y.,
1987,
Microwave Remote Sensing of Snowpack
Properties: Potential and Limitations, Nordic Hydrology,
V18, pl-20
Bilello,
M.
A.,
Regional
1969,
Relationships
Variations
in
Snow
Between
Cover
Climate
Density
and
in
North
C.,
1981,
America, CRREL Research Report, V267, 15p
Burke,
H.
K.,
Bowley,
C.
J.,
and
Barnes,
J.
Comparison of Theoretical and Actual Satellite Microwave
Brightness Temperatures to Determine Snowpack Properties,
Environment Research and Technology, A653-F, pl24-137
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Cavalieri, D. J., 1988, NASA Ice and Snow Validation Program,
NASA Technical Memorandum 100706, 108p
Cavalieri, D. J., and Gloersen, P., 1984, Determination of Sea
Ice
Parameters
With
the
Geophysical Research,
Derived
Global
SMMR,
Journal
of
V89 No. D 4 , p53 55-5369
Chang, A. T. C . , Foster, J.
7
NIMBUS-7
L . , and Hall, D. K., 1987, Nimbus-
Snow Cover
Parameters,
Annals
of
Glaciology, V9, P39-44
Chang,
A.
T.
C.,
Forter,
J.
L., and
Hall,
D.
K.,
1990,
Satellite Sensor Estimates of Northern Hemisphere Snow
Volume, International Journal of Remote Sensing, Vll No.
1, pl67-172
Chang, A. T. C., Foster, J. L., Hall, D. K., Powel, H. W. , and
Chien,
Y.
L.,
1990,
Nimbus-7 SMMR Derived Global Snow
Cover and Snow Depth Data Set,
Documentation,
product
Description, and User's Guide, NASA/Goddard Space Flight
Center,
38p
Chang, A. T. C., Foster, J. L., and Hall, D. K. , 1990, Effect
of Vegetation Cover on Microwave Snow Water Equivalent
Estimates, Proceedings of the International Symposium on
Remote
Sensing
Association
of
and
Water
Resources,
Hydrogeologists
and
International
The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Netherlands
Society for Remote Sensing, p!37-145
c
Chang, A. T.
C., Gloersen,
P., Schmugge, T., Wilheit,T., and
Zwally,
H. J., 1976, Microwave EmissionFrom Snow
Glacier
Ice, Journal of Glaciology, V16, p23-39
Chang, A. T.
Shiue,
C., Hall,
J. C.,
D. K. , Foster, J. L.,
1979,
Characteristics
Rango,
and
A., and
Passive Microwave Sensing of Snow
Over
Land,
Satellite
Hydrology,
5th
Annual W. T. Pecora Memorial Symp. on Remote Sensing, S.
Dakota, p213-217
Choudhury, B. J. , Tucker, C. J. , Golus, R. E., and Newcomb, W.
W., 1987, Monitoring Vegetation Using Nimbus-7 SMMR Data,
International Journal of Remote Sensing, V8, p533-538
Denoth,
A.,
1989,
Snow Dielectric Measurements,
Adv.
Space
Res., V9 No. 1, p23 3 -243
Dozier, J., 1987, Recent Research in Snow Hydrology, Reviews
of Geophysics, V25 No. 2, pl53-161
Ebert,
E.,
1989,
Analysis
Imagery Using Pattern
of
Polar
Clouds
Recognition,
From
Journal
Meteorology, V28, p382-399
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Satellite
of Applied
Foster, J. L. , Chang, A. T. C., Hall, D. K., and Rango, A.,
1991,
Derivation
of
Snow Water
Equivalent
in
Boreal
Forests Using Microwave Radiometry, Journal of the Arctic
Institute of North America, V44 Supp.
1, pl47-152
Foster, J. L. , Hall, D. K., Chang, A. T. C., and Rango, A.,
1984, An Overview of Passive Microwave Snow research and
Results, Reviews of Geophysics and Space Physics, V22 No.
2, pl95-208
Foster,
J.
L.,
Allison,
Rango,
L.
monitoring
Microwave
J.,
A.,
and
Hall,
D.
Diesen,
K.,
B.
Chang,
C.,
A.
1980,
T.
C.,
Snowpack
in North America and Eurasia Using Passive
Satellite Data,
Remote
Sensing Environment,
V10, p285-298
Frank, C., Itten, K. I., and Staenz, K., 1988, Improvement in
NOAA-AVHRR
Snow
Cover
Prediction,
Proceedings
Determination
of
IGARS
'88
for
Runoff
Symposium,
ESA
Publications Division, p433-437
Franklin,
J.,
1986,
Thematic Mapper Analysis of Coniferous
Forest Structure and Composition,
International Journal
of Remote Sensing, V7 No. 10, pl287-1301
Fung,
A. K.,
1981,
A Review of Volume Scatter Theories for
Modelling Applications,
Coherent
and
Incoherent Radar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Scattering From Rough Surfaces and Vegetated Area, ESA
«
SP-166, p83-92
Gesell,
G.,
Using
1989,
An Algorithm for Snow and
AVHRR
Data,
International
Ice Detection
Journal
of
Remote
Sensing, V10, p897-905
Gloersen, P., and Baratn, F. T., 1977, A Scanning Multichannel
Microwave Radiometer for Nimbus and Seaset, IEEE Journal
of Oceanic Engineering, OE-2, pl72-178
Gloersen, P., Cavalteri, D. J., Chang, A. T. C., Wilheit, T.
T., Campsell, W. J., Johannessen, 0. M. , Katsaros, K. B.,
Kunzi, K. F. , Ross, D. B., Staelin, D., Windsor,
L.,
Barath,
Pamseier,
F.
R.
T.,
0.,
Gudmandsen,
1984,
P.,
Langham,
A Summary of Results
E. P.
E.,
and
From the
First Nimbus-7 SMMR Observations, Journal of Geophysical
Research, V89 No. D4 , p5335-5344
Goodison,
B.
E.,
Langham,
E.
Equivalent
Radiometry,
Rubinstein,
J.,
on
the
1986,
I.,
Thirkettle,
Determination
Canadian
Prairies
F.
of
W. , and
Snow
Water
Using
Microwave
Proceedings of the IAHS Meeting,
Budapest,
p24-32
(
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Hall,
«
D.
K ., Foster,
Measurements
J.
L.,
and
Chang,
A.
T.
C.,
1982,
and Modelling of Microwave Emission From
Forested Snowfields, Nordic Hydrology, V13, pl29-138
Hall,
D.
K.,
Regional
Seen by
Foster,
J. L.,
Differences
the
Nimbus-7
and
Chang,
A.
T.
in Snowpacks in
SMMR,
C.,
1984,
Northern U.S. as
Proceedings, Eastern
Snow
Conference, 41th Annual Meeting, Washington, D. C., V29,
328p
Hall,
D.
K.,
Forster,
Passive
Sturm,
J.
L.,
Microwave
M. , Benson,
Garbeil,
Remote
C.
S.,
Chang,
H.,
and
Chacho, E.,
and
In Situ
A.
T.
C.,
1991,
Measurements
of
Arctic and Subarctic Snow Covers in Alaska, Remote Sens.
Environ., V38, pl61-172
Hall, D. K., and Martinec, J., 1985, Remote Sensing of Ice and
Snow, Chapman & Hall, 189p
Hallikainen, M. T., 1984, Retrieval of Snow Water Equivalent
From Nimbus-7 SMMR Data: Effect of Land Cover Categories
and
Weather
Conditions,
IEEE
Journal
of
Oceanic
Engineering, OE-9 No. 5, p372-376
Hallikainen, M. T., and Jolma, P. A., 1986, Retrieval of the
Water Equivalent of Snow Cover in Finland by Satellite
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t
Microwave
£
^
Radiometry,
IEEE
Trans,
on
Geoscience
and
Remote Sensing, GE-24 No. 6, p855-862
Hallikainen,
M.
T.,
and Jomla,
P.
A.,
1992,
Comparison of
Algorithms for Retrieval of Snow Water Equivalent From
Nimbus-7 SMMR Data in Finland, IEEE Trans, on Geoscience
and Remote Sensing, V30 No. 1, pl24-131
Hallikainen,
M. T.,
Satellite
Jolma,
Microwave
Types in Finland,
P. A.,
and Hgyppa,
Radiometry
of
J. M . , 1988,
Forest
and
Surface
IEEE Trans, on Geoscience and Remote
Sensing, GRS-26 No. 5, p622-628
Harrison,
A.
R. ,
and
Classification
Lucas,
of
Snow
R.
M. , 1989,
Using
NOAA
Multi-spectral
AVHRR
Imagery,
International Journal of Remote Sensing, V10, p907-916
Jones,
E. B.,
Rango,
A., and Howell,
S. M. , 1983,
Snowpack
Liquid Water Determinations Using Freezing Calorimetry,
Nordic Hydrol., V14, pll3-126
Kendra,
J. R., Ulaby,
F. T.,
and Sarabandi,
K.,
1994,
Snow
Probe for In Situ Determination of Wetness and Density,
IEEE Trans, on Geoscience and Remote Sensing., V32 No. 5,
pll52-1159
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Key, J.,
«
1990, Cloud Cover Analysis with Arctic AVHRR Data,
Journal of Geophysical Research, V95, p7661-7675
Kidwell,
K. B.,
National
1986,
NOAA Polar Orhiter Data Users Guide,
Oceanic
and
Atmospheric
Administration,
Satellite Data Services Division, 166p
Kong,
J. A.,
and Shin,
R.,
1979,
Theory and Experiment for
Passive Microwave Remote Sensing, Journal of Geophysical
Research, V84 No. BIO, p5669-5673
Kunzi, K. F., Fisher, A. d., Staelin, D. H., and Waters J. W . ,
1976,
Snow and
Ice Surfaces Measured by
the Nimbus-5
Microwave Spectrometer, journal of Geophysical Research,
V81, p4965-4980
Kunzi,
K.
F. , Patil,
S.,
and
Pott,
H. , 1982,
Snow
Cover
Parameters Retrieved From Nimbus-7 SMMR Data, IEEE Trans,
on Geoscience and Remote Sensing, GE-20 No. 4, p452-467
Li, Z. Q., and Leighton, H. G., 1991, Scene Identification and
Its
Effect
on
Cloud
Radiative
Forcing,
Journal
of
Geophysical Research, V96 No. D5, p9175-9188
Lillesand, T., Meisner, D., Downs, A., and Deuell, R., 1982,
Use of GOES and TIROS/NOAA Satellite Data for Snow Cover
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
«
Mapping, Photogrammetric Engineering and Remote Sensing,
V48 No. 2, p251-259
Lytle, V. I., and Jezek, K. C., 1994, Dielectric Permittivity
and Scattering Measurements of Greenland Firn at 26.5-40
GHz,
IEEE Trans. on Geoscience and Remote Sensing, V3 2
No. 2, p290-295
Martinec,
J.,
Snowmelt
and
Rango,
A..,
1986,
Runoff Modelling,
Parameter
Journal
Values
of Hydrology,
for
V84,
pl97-219
Matzler, C . , Aebischer, H . , and Schanda, E., 1984, Microwave
Dielectric Properties of Surface Snow,
IEEE Journal of
Oceanic Engineering, OE-9 No. 5, p366-371
Matzler,
C.,
Schanda,
E.,
and Good,
W. , 1982,
Towards the
Definition of Optimum Sensor Specifications for Microwave
Remote
Sensing of
Snow,
IEEE Trans
on
Geoscience
and
Remote Sensing, GE-20 No. 5, p57-66
Meier, M. F.,
1972, Measurement of Snow Cover Using Passive
Microwave Radiation,
International Symposium on Role of
Snow and Ice in Hydrology, Geneva, VI, p73 9-750
Meier, M. F., and Edgerton, A. T., 1971, Microwave Emission
From Snow - A Progress Report,
Proceedings of the 7th
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
International Symp. on Remote Sensing of the Environment,
Michigan, p!155-1163
Moreno,
J.
F.,
and Melia,
J.,
1993,
A Method for Accurate
Geometric Correction of NOAA AVHRR HRPT Data, IEEE Trans.
Geosci. Rem. Sens., V31, p204-226
Mouginis-Mark,
Effects
P. J.,
Garbeil,
of Viewing
H.,
Geometry
on
and Flament,
P.,
1994,
AVHRR Observations
of
Volcanic Thermal Anomalies, Remote Sens. Environ., V48,
p51-60
Patil, S., Kunzi, K. F . , and Rott, H . , 1981, The Global Snow
Cover
Seen
by
the
Nimbus-7
SMMR,
European
Microwave
Conference 11th, p227-232
Rango,
A.,
1986,
Research,
Progress
in Snow Hydrology Remote Sensing
IEEE Trans, on Geoscience and Remote Sensing,
GE-24 No. 1, p47-53
Rango,
A.,
1989, Operational Applications of Satellite Snow
Cover Observations, iVater Resources Bulletin, V16 No. 6,
P1060-1068
Rango,
A.,
Chang,
Utilization
A.
of
T.
C.,
and
Spaceborne
Foster,
Microwave
J.
L.,
1979,
The
Radiometers
for
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Monitoring Snowpack Properties,
Nordic Hydrology,
V10,
p25-40
Rango,
A., and Itten,
Snow
Cover
K. I.,
Monitoring
1976,
and
Satellite Potentials in
Runoff
Prediction,
Nordic
Hydrology, V7, p209-230
Rao, C., 1987, Pre-launch calibration of channels 1 and 2 of
the Advanced Very High Resolution Radiometer, Technical
Report No. 39, NOAA-NESDIS, 45p
Richards, J., Sun, G. Q., and Simonett, D., 1987, L-Band Radar
Backscatter Modelling of Forest Stands, IEEE Trans. on
Geoscience and Remote Sensing, GE-25 No. 4, p487-498
Richter-Menge,
J.,
Colbeck,
S. C.,
and Jezek,
K. C.,
1991,
Recent Progress in Snow and Ice Research, Rev. Geophys.,
Supplement, p218-226
Rignot,
A.,
E. J. M. , Williams, C. L., Way, J., and Viereck, L.
1994,
Mapping
of
Forest
Types
in
Alaskan
Boreal
Forests Using SAR Imagery, IEEE Trans. Geos. Rem. Sens.,
V32 No. 5, pl051-1058
Rosenkranz,
P. W. , 1982,
Inversion of Data From Diffraction
Limited Multiwavelength Remote Sensors,
Radio Science,
V17, p257-267
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Rotman,
£
^
S.
R.,
Inversion
Fisher,
for
A.
Physical
D.,
and
Staelin,
Characteristics
D,
of
H.,
1982,
Snow
Using
Passive Radiometric Observations, journal of Glaciology,
V28, p89-97
Rott, H., and Aschbacher, J., 1989, On the Use of Satellite
Microwave
Radiometers
for
Large
Scale
Hydrology,
Proceedings IAHS 3rd International Assembly on Remote
Sensing and Large Scale Global Processes, Baltimore, p2130
Saint, G., Herbert, P., Leprieur, C., and Deneral, M . , 1981,
Snow Cover Monitoring Using AVHRR Data From TIROS and
NOAA Satellites, Proceedings of the Fifth International
Symposium on Remote Sensing of Environment,
Ann Arbor,
MI, p519-526
Sakellariou, N . , and Leighton, H. G. L., 1988, Identification
of Cloud Free Pixels in Inhomogeneous Surfaces From AVHRR
Radiances, Journal of Geophysical Research, V93, p52875293
Saunders, R. W . , 1986, An Automated Scheme for the Removal of
Cloud Contamination From AVHRR Radiances Over Western
Europe, International Journal of Remote Sensing, V7 No.
7, p867-886
(
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Saunders, R. W., and Kriebel, K. T., 1988, An Improved Method
for Detecting Clear Sky and Cloudy Radiances From AVHRR
Data, International Journal of Remote Sensing, V9, pl23150
Schmugge,
T.
J.,
1980,
Photogrammetric
Microwave
Engineering
Approaches
and
in Hydrology,
Remote
Sensing,
V4 6,
p495-507
Schmugge, T. J., Wilheit, T. T., Gloersen, P., Meier, M. F.,
Frand, D., and Dirmhirm, I., 1973, Microwave Signatures
of
Snow
and
Techniques
Fresh
in
the
Water
Study
Ice,
of
Advanced
Snow
and
Concepts
Air
and
Resources,
National Academy of Science, Washington, D. C., p551-563
Scialdone, J . , and Ulaby, F. T., 1987, Comparison of Northern
Hemisphere Snow Cover Data Sets, Journal of Climate and
Applied Meteorology, V26 No. 1, p53-68
Shiue, J. C., Shin, R., Chang, A. T. C . , Fuchs, J . , Lin, F.,
and Greenan, H . , 1984, Observation of Passive Microwave
Emission
From
Snowpack,
Proceedings,
Eastern
Snow
Conference, 41th Annual Meeting, Washington, D. C., v29,
p75-85
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Steffen,
m
^
K.,
1993,
Climate
Sensitivity
Studies
of
The
Greenland Ice Sheet, Meteorology and Atmospheric Physics,
in press
Stiles, W. H., and Ulaby, F. T., 1979, Monitoring Snow with
Microwaves, Satellite Hydrology, 5th Annual W. T. Pecora
Memorial Symp. on Remote Sensing, S. Dakota, p225-229
Stiles, W. H., Ulaby,
F. T.,
and Rango,
A.,
Measurements of Snowpack Properties,
1981, Microwave
Nordic Hydrology,
V12, pl43-166
Sutherland,
I.,
Monitoring
and
in
Ferner,
Alberta
S.,
Using
1986,
Snowpack
Computer
Depletion
Processed
NOAA
Imagery, Proceedings of the 10th Canadian Symp. on Remote
Sensing, Alberta, p463-471
Tiuri, M. E., 1982,
Microwave
Theoretical and Experimental studies of
Emission
Signature
of
Snow,
IEEE Trans,
on
Geoscience and Remote Sensing, GE-20 No. 1, p51-57
Tiuri,
M. , and
Hallikainen,
M. , 1981,
Microwave
Emission
Characteristics of Snow Covered Earth Surfaces by the
Nimbus-7
Satellite.
Proceedings
of
the
11th
Microwave Conference, p233-238
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
European
Ulaby, F. T., Moore, R. K., and Fung, A. K., 1987, Microwave
Remote
Sensing,
Active
and
Publishing Co., Massachusetts,
Passive,
Addison-Wesley
3 vols.
Ulaby, F. T., and Stiles, W. H., 1980, Microwave Radiometric
Observation of Snowpacks,
Snowpack Properties,
Microwave Remote Sensing of
Proceedings
of
a workshop
NASA
conference, pl87-202
Ulaby, F. T., and Stiles, W. H., 1980, The active and Passive
Microwave
Response
to
Snow
Parameters,
Journal
of
geophysical Research, V85, pl037-1049
Wang,
J.
R.,
1985,
Effect
of
Vegetation
on
Soil
Sensing Observed From Orbiting Microwave
Moisture
Radiometers,
Remote Sensing of Environment, V17, pl41-151
Warren,
S.,
1982,
Optical
Properties
of
Snow,
Reviews
of
Geophysics and Space Physics, V20 No. 1, p67-89
Warren, S., and Wiscombe, W. , 1980, A Model for the Spectral
Albedo of Snow, Journal of the Atmospheric Sciences, V37,
p2734-2745
Woo, M. K., and Stteer, P., 1985, Simulation of Snow Depth in
a
Forest,
Proceedings
of
the
1985
Eastern
Conference, Montreal, p44-54
R eproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Snow
Yamanouchi,T ., Suzuki, K., and Kawaguchi, S., 1987, Detection
of Clouds in Antarctica From Infrared Multispectral Data
of AVHRR, Journal of Meteorology Society, V65, p949-962
Zibordi,
G., Van Woert, M . , Meloni, G. P., and Canossi,
1995,
Intercomparisons
SSM/I
and
AVHRR
Data
of
of
Sea
the
Ice Concentration
Ross
Sea,
Remote
I.,
from
Sens.
Environ., V53, pl45-152
Zwally, J., and Gloersen, P., 1977, Passive Microwave Images
of
the Polar Regions and Research Applications,
Research, V18, p431-44 2
€
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Polar
IMAGE EVALUATION
TEST TARGET (QA-3)
1.0
13-2
“
1.1
____
i«
nni 2 2
1 3.6
= “ IS
KUU.
1.8
1-25 11.4
ui i.6
150mm
IIS/14GE . In c
1653 East Main Street
Rochester. NY 14609 USA
Phone: 716/482-0300
Fax: 716/288-5989
0 1993, Applied Image, Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
4 162 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа