ALPMON FINAL REPORT - ARC systems research

Contract ENV4-CT96-0359 ALPMON
the data costs. The cost estimations for the European customer requirements are calculated for one
entire satellite image (Table 16) as well as for the whole Alps (Table 17). The cost respectively time
effort estimated for the scale of 1:50.000 to 1:100.000 varied between 1,78 EURO/0.05 hours per km 2
and 2.04 EURO/0.03 hours per km 2 .
All cost estimates are performed under the assumption that the processing procedure is already
developed. A further assumption was that the aerial photos (for reference area investigation) and the
DEM are provided by the customer.
In general, the more satellite scenes are treated for one application, the cheaper the interpretation per
km 2 becomes. However, the area of interest always has to be taken into consideration when
calculating the costs for a specific application. The costs for the interpretation of one satellite scene
remain more or less the same, whether the area of interest is the entire image or only one km 2 . This
explains the differences in costs per km 2 given in Table 16 and Table 17 for classification of one entire
Landsat TM scene (32.400 km 2 ), all scenes covering the Alps (360.000 km 2 ), and the area restricted to
the Alps (180.000 km 2 ).
Conventional methods could only be evaluated as far as they were applicable to the specific
requirements and as far as they were previously used by the customers. For most applications this was
not the case, as earth observation data provided additional information to the customers, which was
not available in this kind before. However, two figures on more or less conventional methods can be
given for comparison with the ALPMON remote sensing approach.
The first concerns CORINE land cover, which is investigated by visual interpretation of Landsat TM
false colour composites. The costs for the interpretation of CORINE land cover were estimated by the
Federal Environment Agency of Austria with 9.50 EURO/km 2 . Compared to the cots estimated for
(semi)automatic classification with approx. 2 EURO/ km 2 (Table 16) the costs of the conventional
visual interpretation are nearly five times higher. However, it has to be kept in mind, that the results of
both approaches are not directly comparable.
For most of the national applications the only alternative to satellite image classification would be aerial
photo interpretation or field work. It has to be emphasised, that the different methods are not directly
comparable, as they deal with different scales and in most cases also different nomenclatures
respectively parameters and number of classes to be investigated. Only for forestry related aerial photo
interpretation figures are available from an area which was investigated in Carinthia/Austria. The
figures given in Table 18 apply to a very detailed interpretation of aerial photos for forest parameters
using an analytic stereo plotter, including vectorisation of the interpretation results. The costs are given
only for the forest covered areas, which in Austria can be assumed with approx. 50% of the area
covered by the aerial photos. They were estimated with 1325 EURO per km 2 . It can be expected that
these costs may be further reduced by a maximum factor of 5 when less parameters are interpreted or
less detail is required. This would lead to a figure of approx. 130 EURO/km 2 for a land cover
interpretation down to 80 EURO/ km 2 for the investigation of only a few land cover parameters.
Concluding it has to be stated that also here a direct comparison of the conventional approach with the
satellite classification is not possible, as the scale and detail of interpretation in both cases is quite
different. However, a tendency can be clearly read from these figures, indicating that satellite image
classification for the described applications and required parameters is much cheaper than the
conventional methods available, and that for this reason satellite image classification in many cases
made an investigation of the required parameters possible for the first time.
JR, RSDE, ALU, LMU, Seibersdorf, WSL 66