ult in an
season,
tinental
Will be
»solution
mosaics
)0 meter
osaicked
> canvas
8 million
ics.
n with
nts on all
terest to
] largest
> Congo
changes
n basin
/er basin
t border
into an
re. The
> part of
her may
toring of
thus not
sily to hit
the case
d than it
sensitive
oral and
n that is
e Congo
and low
the time
Annan nn
rn Africa
of great interest in central Africa is also a possible
comparison with an available ERS-1 mosaic covering the
Congo basin. The data were acquired during the summer
of 1994 by using a transportable ground receiving station
located in Gabon at the time and the mosaic was
generated by the Joint Research Centre of the E.C. within
the framework of the TREES project (Tropical Ecosystem
Environment Observations by Satellite). It covers the
major part of an area between South 6° - North 10° and
East 8°-28° (some 400 ERS-1 scenes) and provides a first
opportunity to compare data from C- and L-band SAR at
such a large scale.
3.2 Coverage and data acquisition schedule
The JERS-1 data acquisitions over central and western
Africa can be divided into three main areas: eastern
Africa between South 6° - North 5° / East 35°-40°; central
Africa/Congo basin between South 8° - North 8° / East
10°-40°: and a 600 km wide strip along the coast of
western Africa, approx. North 4°-10° / West 14°-East 10°.
This covers a continuous stretch from the Indian Ocean
coast of Kenya and Tanzania in the east to Sierra Leone
on the Atlantic coast on the west, amounting to about 8
million km2, or some 2000 scenes. Data acquisitions over
this entire strip were accomplished in one uninterrupted
effort between early January and early April, 1996,
coinciding with the overall low water mark of the Congo
river. As the JERS-1 satellite acquires data at adjacent
swaths with one day intervals (going from east to west),
this results in a temporally very homogeneous data set.
The study of the temporal and spatial extent of flooded
environments applies to the Congo basin only (South 8? -
North 8° / East 10°-40°), thus excluding the eastern and
western parts during the second data take. The
acquisitions are planned to concur with the peak of high
water in the Congo river in October and November, 1996,
adding another 5 million km2 and some 1200 scenes to
the low water figures above.
3.3 Data processing
3.3.1 Raw data processing: The major part of the
data acquired during the low water period in early 1996
were down-linked at NASDA Earth Observation Center,
where the raw data processing will be performed. NASDA
has during the spring of 1996 up-graded its SAR
processor to remove radiometric noise which was present
in the earlier version of the processor. This included
removal of a vertical striping noise and intensity
differences between near and far range. The output data
Is standard NASDA high resolution scenes, i.e. ground
range, amplitude, 16 bits, 12.5 meter pixels.
$3.2 Generation of low resolution data and
Image mosaics: The high resolution scenes will be
shipped to JRC (Institute for Remote Sensing
Applications) in Ispra, Italy, for further processing, that
is, downsampling to low resolution scenes at 100 meter
pixel size and generation of texture images. JRC will also
generate 100 meter mosaics by utilizing mosaicking
Software developed at JRC within the TREES project.
597
Also here will the 100 meter mosaics be generated in a
number of more facile regional mosaics. 400 meter semi-
continental mosaics will be generated by either JRC or
NASDA.
3.3.3 Land cover classification: In cooperation
with JRC, and sponsored by NASA, the Jet Propulsion
Laboratory are considering a classification effort over
certain areas of Africa by utilizing the same classification
algorithm as is being used at JPL for the Amazonian data.
Although no simultaneous field measurements were made
during the satellite acquisitions, some ground truth data
can be obtained from local scientists in the area.
4. SOUTH-EAST ASIA AND PAPUA NEW
GUINEA
4.1 Mapping of a large distributed target
South-East Asia and Papua New Guinea (referred to
below as SEA & PNG) comprises a significantly different
environment than in both South America and Central
Africa. The region consists of a large number of islands
which more often than not are rugged with topographic
features and this brings about a new challenge to our
mapping project. In an operational point of view, the well
distributed archipelago results in that the data takes will
have to be divided in to a number of smaller sub regions
rather than acquiring the data in one large sweep. The
region is currently projected to be acquired one time,
yielding a total number of some 3000 scenes.
Acquisitions are primarily planned by utilizing the on-
board tape recorder, with the two ground receiving
stations in Bangkok, Thailand, and Pare pare, Indonesia,
as back-up.
Although certain parts of SEA & PNG during the years
have been monitored and mapped quite extensively,
others, such as e.g Papua New Guinea/lrian Jaya and the
inner parts of Borneo/Kalimantan, still remain virtually
unknown. This project will thus provide a unique
opportunity to get an up-to-date radar "snap shot" over
these areas, adding objective data to speculations on
controversial issues such as e.g. deforestation.
Acquisitions over well documented areas in turn, such as
e.g. parts of Thailand and the Philippines, will result in a
large scale homogeneous data set which may help to
increase the over all understanding and usefulness of
SAR data in general.
4.2 Coverage and data acquisition schedule
The data acquisitions over the SEA & PNG region have
been tentatively scheduled as follows:
Papua New Guinea/lrian Jaya: March - April 1996
Malay Peninsular and Sumatra: August 1996
Borneo/Kalimantan, Java and Sulawesi: Sept.- Oct. 1996
The Philippines: December, 1996
Burma, Thailand, Cambodia , Laos and Vietnam:
January - February, 1997.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
