SNOW HYDROLOGY USING MULTI-CHANNEL, MULTI-POLARIZED, MULTI-TEMPORAL SYNTHETIC
APERTURE RADAR
Bajzák, Dénes. Professor of Engineering, Faculty of Engineering and Applied Science, Memorial University of
Newfoundland, St. John's, Nfld., Canada. A1B 3X5.
Roberts, Bruce A. Research Scientist, Natural Resources Canada, Canadian Forest Service, Atlantic Forestry Centre,
Sir Wilfred Grenfell College, P.O. Box 960, Corner Brook, Nfld, Canada. A2H 6J3.
Deering, Keith W., Forest Ecologist, Natural Resources Canada, Canadian Forest Service, Atlantic Forestry Centre, Sir
Wilfred Grenfell College, P.O. Box 960, Corner Brook, Nfld, Canada. A2H 6J3.
Commission VII, Woking Group
Key Words: Snow hydrology, snow-water-equivalent (SWE), multi-channel multi-temporal SAR.
ABSTRACT
The estimation of spring run-off is a very important aspect of the water management of reservoirs. One of the inputs to
models used for this purpose is the measurement of snow accumulation on the ground during the winter months. The
Churchill Falls (L) Co. uses ground measurements in several snow courses to predict the amount of snow melt water
through the determination of snow-water-equivalent (SWE) for management of water in the reservoirs of the hydro-electric
power plant at Churchill Falls, Labrador.
Evaluation of research literature suggested that the SWE could be determined by the simultaneous use of multi-channel
(X, C, and L bands) synthetic aperture RADAR (SAR), (NASA 1981). This project was designed to test this theory under
field conditions. Multi-temporal (frozen ground with no snow, and ground with maximum snow cover) X and C band SAR
data were obtained over several test sites. A detailed ground survey of snow conditions immediately followed the
‘maximum snow cover SAR mission. Half of the test sites were located at 30° and the other half at 60° incidence angles.
Like (VV and HH) and cross (HV and VH) polarizations were used. Data analysis to date included the summarization of
snow characteristics and of distribution, and the association of C and X band SAR data with ground observations of SWE.
INTRODUCTION
The hydro-electric power plant at Churchill Falls is one of
the largest power generating facilities in the word with a
drainage area of 69 267 km containing five reservoirs.
Effective water management in the reservoirs requires the
estimation of snow melt at spring time. This prediction is
based on field observations in several snow courses during
the winter months. The collection of field data is very
expensive as most of the courses can be reached only by
helicopter. Climate modellers maintain that in order to
estimate how long snow cover will affect the energy and
moisture exchange at the land surface, estimates of snow
cover extent are not enough, rather they also need to know
average snow water equivalent (SWE) (Rango, 1996).
Since these snow course data do not provide accurate
estimate of run-off, they could be supplemented by SAR
observations provided by satellites.
Researchers suggested that snow properties and SWE can
theoretically be related to multi-channel RADAR returns
from snow fields in open areas (NASA 1981 and Shi et.al.
1990, 1995). In 1989 the Canada Centre of Remote
Sensing (CCRS) had introduced an airborne RADAR
development program as the fore runner for the application
of a new Canadian RADARSAT satellite which was
launched in late 1995. This made it possible for us to
execute an experiment under field conditions to test the
above theory.
Bernier and Fortin (1991, 1998) evaluated the potential of
C- and X- band SAR data to monitor dry and shallow snow
cover. Another application of C band SAR for mapping
melting snow (Donald et. al. 1993) and mapping of
discontinuous permafrost terrain (Granberg 1994) were
reported in the Canadian Journal of Remote Sensing.
Rango (1994) recognized the need for future research to
correct for variations in vegetation cover and snow grain
size for use in a snowmelt-runoff model (SRM).
Our experiment was designed such that the airborne SAR
data acquisition and ground data on snow could be
collected nearly simultaneously in pre-determined sites at
two time intervals (with and without snow cover).
Statistical analysis of the relationship between ground and
RADAR data should provide an estimate of SWE that could
be built into a hydrological model to predict spring run-off.
EXPERIMENTAL DESIGN
The 'Earth Observation System' approach (NASA 1981)
recommends the use of L, C, and X band RADAR
simultaneously over the snow field to find the SWE from the
SAR data without any ground observation. Data of these
bands provide three simultaneous equations from which
the SWE can be obtained. Since the CCRS SAR had only
two bands (X and C) we required two coverages of the
same area: one when the ground was frozen without snow
and another one with maximum snow depth accumulation.
This would replace the use of L band.
The incidence angle is one of the most determining factors
in RADAR signal return. To simplify our analysis we used
only two incidence angles: 30? and 60? for data acquisition.
As the ground data collection had to immediately follow the
662 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
SAR
test
to 1
obte
with
The
onc
corr
spe
SAF
one
to 3
On
ana
wer
tem
Airl
The
afte
fligl
poli
wer
only
of I
mo
Gr
Sin
afte
sta
ext
wel
anc
or.
foll
ten
gra
doc
ph
tes
pet
pre
line
the
dis
