» ERS-1
ameter
g new
cal and
the C-
etation
hat the
band in
JERS-1
iuse of
d. The
When
JERS-1
r crops
3 were
d more
multi-
. visual
is well
g new
) other
nes an
micro-
wave-
s offe-
arame-
ly, we
5-1 for
e ERS-
cation,
1 data
rowing
tance,
/ember
optical
scenes of the Flevoland and Veluwe region (The Ne-
therlands) were recorded by the JERS-1 and made
available by NASDA to this study. ESA-ESTEC made
as many as possible ERS-1 scenes of Flevoland availa-
ble to the study.
To compare the applicability of JERS-1 L-band SAR
with ERS-1 C-band SAR the following activities were
performed:
(1) Comparison of time series of JERS-1 and ERS-1
radar data: for agricultural areas, radar backscatter
data were extracted from the available images for
selected polygons. The obtained time series for L- and
C-band data were compared with each other.
(2) Agricultural classification experiment: for the gro-
wing season of 1993, 3 JERS-1 L-band images were
used in a multitemporal object-based classification.
The classification results were compared with a com-
posed digital reference map of the research area.
Finally, the obtained classification accuracies for the
different land cover types were evaluated in compari-
son with results based on 3 ERS-1 images.
(3) Forest classification experiment: because of the
larger wavelength and larger tree crown penetration,
the JERS-1 L-band was expected to yield other infor-
mation about forests than C-band ERS-1. Therefore,
the utility of JERS-1 and ERS-1 SAR data for forest
inventory was compared.
The current paper presents detailed results of the
above studies.
2. MATERIAL AND METHODS
2.1 Test Sites
The test sites selected for the ERS-1 and JERS-1
campaign were the Flevoland agricultural test site, the
Horsterwold site (forested) and the Speulderbos site
(forested) in the Netherlands.
The Flevoland site is situated in Southern Flevoland, a
polder reclaimed in 1966 from lake lJssel, and it is in
cultivation by individual farmers and a state owned
farm. The fields of the state farm are relatively large
(approximately 1500 x 500 m) compared to the fields
cultivated by private farmers (approximately 500 x
200 m). The general altitude is 3 metres below sea
level.
The Horsterwold site is also located in Southern Flevo-
land. The forest is managed by the "Directie Flevo-
land". Planting started in 1973. When completed the
forest will be 4000 ha large and, herewith, will be the
largest deciduous forest in the country. The site com-
prises large numbers of poplar stands and stands of
other deciduous species like: willow, elm, beech, oak,
alder, maple and ash.
The Speulderbos site, located at the Veluwe which is a
part of the province of Gelderland and features the
largest forested region in the Netherlands, is managed
by the State Forest Service. The area measures 2390
ha and contains many species in many age classes.
Scots pine, Douglas fir, Japanese larch, beech and oak
stands prevail. Corsican pine, European larch, Norway
spruce and grand fir stands occur less frequently.
These two forest test sites comprise a fair range of
species, age classes, soils and other environmental
conditions. Stand areas typically range from one to
several hectares, which is quite small but normal for
large parts of Europe.
For ERS-1 and JERS-1 application research (classifica-
tion, parameter estimation, monitoring), actual field
data (ground truth) had to be collected during ERS-1
and JERS-1 overflights. From November 1992 until
November 1993, microwave and optical scenes of the
Flevoland test site were obtained from the ERS-1 and
JERS-1 on a regular basis. As a result, ground truth
data also were acquired on a regular basis for that
period. A complete overview of the ground truth can
be found in Vissers (1994).
2.2 Satellite Data
The following ERS-1 SAR images covering Flevoland in
1993 were available for this project: 12/01, 16/02,
23/03, 11/04, 18/04, 27/04, 16/05, 23/05, 01/06,
20/06, .27/06,.06/07,.25/07,.01/08, .10/08,.29/08.
05/09, 03/10, 10/10, 19/10.
The following JERS-1 SAR images were available for
1993: 12/02, 28/03, 12705, 07/08, 20/09.
For comparison of JERS-1 and ERS-1 classification
results ERS-1 images close to the acquisition dates of
the JERS-1 images were selected. This results into the
images given in table 1.
Table 1. Selected SAR images from the total set for
classification purposes.
JERS-1 ERS-1
12-02-1993" 16-02-1993"
28-03-1993" 23-03-1993"
12-05-1993 16-05-1993
07-08-1993 10-08-1993
20-09-1993? 05-09-1993?
' not used for classification of agricultural crops
2) not used for forest classification
2.3 Reference Data
A SPOT-1 optical image of the 2nd July of 1993 was
used to identify 530 agricultural fields of 24 different
crops which were indicated on the "Vegetation inven-
tory map Zuidelijk Flevoland (NL) 1993" (Vissers,
1994). By digitizing the lot boundaries on the screen of
the image processing system a vector map was con-
structed. The vegetation map thus obtained is the
147
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
