2) Water Quality
Water quality assessment requires the detection and measurement
of pollution sources and the mixing of pollutants. By remote
sensing this is generally possible for visible pollutants such
as sewage and bulk oil.
Non-visible pollution of a chemical nature can only be
detected and traced by an indicator such as a dye-tracer or
by temperature difference. Both types of indicators have been
used in experiments in the Wilhelmshaven area. With Rhodamin-
dye a chemical plant discharge was investigated and thermal
contrasts served to trace the discharge of a paper plant and
a thermal power plant.
The power plant investigations were combined with a test
on the accuracy of water temperaturg measurement by
remote sensing. Using the thermal M S-Scanner data
directly, without corrections yielded RMS errors of + 1,5%
at check points. An atmospheric model, taking into account
meteorological measurements of air temperature and humidity
during the data acquisition period resulted in RMS errors of
* 0,4 C. If, finally, water temperature measurements were used
at few control points RMS errors od + 0,2 C were obtained at
check points, after application of an interpolation procedure
according to a low order polynomial.
3) Tidal flat classification
Both non-supervised and supervised classification attempts
have been made for tidal land, trying to separate the most impor-
tant tidal land classes by the usual multivariate procedures.
As usual in classification, difficulties were encountered in the
definition of classes by radiometric properties. On the other
hand it was difficult to find suitably pure sample areas for con-
ducting supervised classification. Particularly for low resolution
Landsat-images in 4 channels (pixel size 80 m) and for medium reso-
lution M S- images in 11 channels (pixel 4 m) the classification
results were poorer than for 3-color separated aerial color photo-
graphs (pixel size o,4 m), despite the fact that for the color
photos the spectral channels were very wide and overlapping.
4. Sonar Imaging
Into the uses of remote sensing in coastal areas, there
should also be included such non-classical RS devices as the
underwater sonar.
Its geometry and its imaging problems are similar to those
of SLAR. Geometry is the lesser problem. It can be taken care
of if the sensor movements are accurately known.
Th
th
Fo.
or
Bu
vi
re.
ti!
SO!
|
th:
ae:
2)
re:
3)
at
4)
int
5)
anc
Suc
hax
dev