basis for the classification in section 4.
For the forest classification study, data for 196 forest
stands in the two test sites were collected from previ-
ous ERS-1 studies (van der Linden, 1995). These
include descriptions of (main) species, height, density,
age and area. In this study a part of these stands
could not be used because of the incomplete overlap
of the JERS-1 images (differing somewhat between
dates) with the ERS-1 images. A database was created
for all stands present in these images. The database
includes data on stand averaged oc? and the within
stand standard deviation of o°.
For the extraction and calibration of the JERS-1 back-
scatter, parameters and algorithms were used descri-
bed by Kleijweg & Groot (1995). The ERS-1 images
were calibrated according to a procedure described by
Laur (1992).
3. TEMPORAL RADAR BACKSCATTER SIGNATURES
3.1 Introduction
The ERS-1 and JERS-1 recordings of the growing
season of 1993 provided a large number of microwave
measurements in time. In the 1993 growing season 20
ERS-1 recordings with an interval of about 10 to 16
days are available. For the JERS-1 only four recordings
during the growing season with an interval of 30 to 70
days are available. It is therefore important to consider
the measurement frequency in relation to the shape of
the time series curves for each crop with ERS-1 first.
Subsequently, the performance of JERS-1 can be
studied with the conclusions of the ERS-1 in mind. Of
course a different backscatter behaviour can be expec-
ted from the various agricultural crops with L-band HH
of JERS-1 compared to C-band VV of ERS-1.
The main crops at the Flevoland site have been selec-
ted and ordered into 4 groups:
1. crops with abundant leaves and biomass, like sugar
beet and potato;
2. crops with less biomass like grass, lucerne and
rapeseed;
3. vertically oriented crops like wheat, barley and
maize;
4. miscellaneous crops like bush and fruit trees.
Results for group | and lll only will be presented since
these are the most important crops in the region.
3.2 Average backscatter ERS-1
The average curves in group | show a similar backscat-
ter behaviour (e.g. figure 1 for sugar beet). The peak
at day 108 can be noticed for potato as well as for
sugar beet. On that day, it was raining probably during
the ERS-1 observation. After the peak and subsequent
dip, a clear increase in backscatter can be noticed.
This increase is most likely due to a combined effect
of soil moisture and biomass, until complete cover has
been reached. Then the backscatter signal saturates
and remains on a more or less stable level.
t€:
9 T + + —
3 so 100 150 200 250 300 Daynr 159
ne ERS-1
|
ad
| JERS-1
as 4
[
20 +
|
|
28 +
gamma (dB)
Figure 1. ERS-1 and JERS-1 backscatter signatures of
sugar beet for the 1993 growing season.
a - Te
3 30 100 150 200 250 250
: Daynr
T AEs
JERS-1
ad
|
|
i
-25 +
gamma (dB)
Figure 2. ERS-1 and JERS-1 backscatter signatures of
winter wheat for the 1993 growing season.
The crops in the third group show a similar behaviour
as well. The vertically oriented crops like winter wheat
(figure 2) and barley have a clear dip in backscatter in
their curves when the stems are elongating (down-
wards) and the ears are growing and filling (upwards).
It seems that the growth started at the same time
(when average temperature is rising) for the three
crops, because in the beginning of the growing season
the curves look rather similar. The wheat is sown
before the winter in 1992 and the small amount of
biomass is apparently not affecting the curve com-
pared to the bare soil of the other two crops. A shift
between the crops can be noticed during the upward
move of backscatter when the crop is maturing, which
offers quite good distinction between the crops. From
day 130 till 200 about 2-3 dB difference can be noti-
ced between the crops in this group.
3.3 Average backscatter JERS-1
In figures 1 and 2 also the JERS-1 average signatures
148
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
