ecific
which always
upwards and
NOAA /VHRR
size of
ion, repe-
ities could
a high
gether with
3.
accuracy and
er the vital
e of the
cted inter-
to be taken
apping. From
ed (Fig.
l| data. This
ut numerical
tric as well
le.
es of the
27.
VI Supporting activities
Interpretation of satellite data will always have to be accompanied by
supporting activities, which can be divided in the following major groups:
- ground observations and measurements
- spectral properties of snow
- underflights
Ground truth is necessary for a better understanding of the study area and
the influence caused by the various snowparameters on the registration of
the snowcover by different sensor systems. For digital processing, applying
supervised technology, the optimal selection of the test samples is of
particular importance. Finally the verification of the classification results
has to be secured.
Underflights by aeroplanes or helicopters may serve similar purposes such as
theverification of the results and selection of the test samples.
The spectral characteristics of snow vary considerably, due to illumination,
season, climatic region, surface conditions etc. Besides snow is always
covering an underlying stratum which may influence the spectral reflectance,
or is partly covered by vegetation, especially trees. These problems will not
be treated in this paper. Sofar it may be concluded that for efficient snow
studies data in the following three channels are absolutely essential:
- one band in the visible spectrum (Landsat-5) to determine the areal extent;
- one band in the near IR (Landsat-7) to separate dry and wet snow;
- one band in the 1.55-1.75 pm (to be included in Landsat-D) to separate
snow from clouds.
The ultimate goal will be to reduce ground measurements and observations to
a minimum, e.g. by building up data banks on the spectral behaviour of snow
etc.
VII Conclusions
Methods for snowcover monitoring today may be regarded as quasi operational.
For further projects it is especially important that the selected method is
appropriate regarding the size, structure and climatic conditions of the
study area, sensor system, accuracy needed to fulfill the specific task
within a very short time lapse. The most promising approach offer interactive
interpretation systems,where the knowledge of an experienced interpreter can
be combined with fast digital classification systems.