'e€ water,
effect of
ses were
. Median
sification
tion was
the best
ots pine.
all below
ılliainen,
equency:
3°), Jers
HH, 23°
he dates
images.
ring that
um prior
to 14 ın
‚5.4.1994
ave been
eraged to
e taken
the table
> table 1)
).11.1997
;RS- and
acted due
e images
| used in
ch image
d Survey
; Finnish
e used as
once data
ixels and
me under
sting), 2.
agricultural field (101149102), 3. urban area (7284-3641),
4. water (69844194), 5. forest, stem volume under 100
m°/ha (83442504), 6. forest, stem volume 100 - 200 m°/ha
(2124213) and forest, stem volume over 200 m/ha
(697-1394).
3. SEASONAL VARIATION OF BACKSCATTER
The first six ERS-1 images were taken during summer
and autumn conditions (snow-free and unfrozen soil).
Temporal variations of backscattering coefficient in these
images are caused by the changes in the vegetation
cover or changes in the vegetation and soil moisture.
Ground freezing happened in the end of November, so in
the December and early January frozen ground was
covered by thin layer of dry snow. In the mid-January,
there was a short warm period (temperature over 0 C^,
raining water) and snow falls in the late January. There
were no significant snow falls after that.
One JERS image was taken during winter (image 15)
and another during late spring (image 16). There was a
few days cold period (temperature below 0 C?) and a
little snowfall prior taking winter JERS-image, so it can
be assumed that ground was frozen and covered with
very thin layer of snow. There was about three weeks
warm period with rain, fall of snow or wet snow prior
taking late spring JERS-image. So, it can be assumed
that snow was melting increasing snow water equivalent
in spring image, and it is possible that ground was
atleast partly snow-free.
Radarsar images (images 17 to 20 in the table 1) were
taken within 2.5 month period, the first one in mid-
November and the latest end of January. Ground was
snow-free and unfrozen in the image 17. There was
snowfall when the image 18 was taken, but ground was
unfrozen. In the early December there were cold period
with snowfall, so ground freezing happened during that
Classes mire, agncultural field, urban area and water
T T T
Backscattenng coefficient (dB)
i 1 i
Number of image
Figure 1: Backscattering coefficients of classes
mire (solid line), agricultural field (dotted line),
urban area (dash-dot line) and water (dashed
line).
time. Warmed period (temperature over zero) followed in
the mid-December. During January, there were few
snowfalls and some warm days, but weather was getting
colder with small snowfall just prior taking the image
20.
Backscattering coefficients of different classes were
computed from ERS-1 and Jers-images. The seasonal
variations of backscattering coefficients of different
classes are represented in figures 1 and 2. Figure 1
represents classes mire (solid line), agricultural field
(dotted line), urban area (dash-dot line) and water
(dashed line), and figure 2 represents classes forest stem
volume under 100 m/ha (solid line), forest stem volume
101 - 200 m/ha (dotted line), forest stem volume over 201
m/ha (dash-dot line), and water (dashed line) is included
for comparison. Generally the mean backscatter of
classes is higher during summer than winter, except
water. Also, the temporal variations of the mean
backscatter are higher during summer and autumn than
during winter. Variations of class water are largest
because, during summer the backscatter of ERS-1
images is sensitive to wind conditions, and during winter
it is sensitive to the aging of snow and the snow
moisture. Backscatter from forest classes is lowest when
ground is frozen and covered with thin layer of snow.
Accumulation of new snow increases backscatter and
aging of snow decreases it. Variations of forest classes
increase according to the increase in stem volume. Also
differences in backscatter between summer and winter
increase with the stem volume. Deviations of forest
classes decrease as stem volume increases. Differences
in backscatter of mire and forest classes are largest
during summer, otherwise backscatter behaves quite
similar way. The backscatter from urban area does not
vary much, it is almost undependent from season.
Largest difference between backscatter from winter Jers
and ERS-1 images is in class water, backscatter of water
from Jers is much lower than from ERS-1.
Forest classes and water
T T
Backscattering coefficient (dB)
| i
8
Number of image
Figure 2: Backscattering coefficients of forest
classes: vol. « 100 m?/ha (solid), 101 - 200
m?/ha (dotted), » 201 m?/ha (dash-dot) and
water (dashed).
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 567
