30.01.2013 Views

ALPMON FINAL REPORT - ARC systems research

ALPMON FINAL REPORT - ARC systems research

ALPMON FINAL REPORT - ARC systems research

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Contract ENV4-CT96-0359 <strong>ALPMON</strong><br />

The concept of object-wise data management is common in GIS <strong>systems</strong> like ArcView or the newest<br />

ArcInfo 8.0 release. Clicking on the object you will open the data base with the attributes assigned to this<br />

object. Looking for areas with specific properties, you will succeed starting a data base search following<br />

the rules of .sql language of relational data base <strong>systems</strong>.<br />

The result of object oriented image analysis is a data base with information on each object distinguished<br />

during the segmentation and classification process. This allows to identify objects with statistics changed<br />

between the observations by data base .sql queries and more, to quantify the changes. An image analysis<br />

system like eCognition, which additionally is able to use existing administrative or thematic boundaries<br />

from other investigations, has major advantages for specific purposes, like they occur for example in the<br />

forest administration, which is managing the forest on stand level, for biotope monitoring, a.s.o.<br />

Post-processing<br />

There is still a need for advanced post-classification techniques that can at least partially compete with<br />

human interpretation skills. While the spectral classification and the recognition of land cover types can<br />

be highly automated, the interpretation of spatial context and the derivation of land use patterns is still<br />

a complex matter. Post-classification as applied in <strong>ALPMON</strong> is a first step towards rule based<br />

generalisation but is limited to simple spatial parameters. Recent developments, such as object based<br />

classification schemes or graph based representation, offer new answers to these questions, which<br />

become even more evident when the very high spatial resolution data will be used in the near future.<br />

Concerning landscape monitoring not only satellite image classification could be an automated way.<br />

Another possibility to get cheaper results in landscape monitoring is the automatic recognition of<br />

objects. This would have to be tested in future.<br />

Table 23 finally illustrates by way of example the improvements in algorithms and software which are<br />

expected to increase the accuracy and performance of the erosion risk assessment. Part of these<br />

parameters do also apply to other alpine applications.<br />

Table 23: Expected methodological improvements useful to erosion risk assessment<br />

Methodological<br />

Improvements<br />

Elimination of the overcorrections<br />

in topographic<br />

normalisation<br />

Unsupervised classification<br />

algorithm taking into account<br />

the Mahalanobis distance<br />

instead of the Euclidean<br />

distance (like ISODATA).<br />

Database containing the<br />

phenological calendars of the<br />

most common species in the<br />

study area, better if updated to<br />

the actual meteorological trend.<br />

Benefits<br />

� Improvement of the land cover accuracy. Reduction of<br />

manpower in revision of the classification.<br />

� Increase in discrimination of vegetation classes.<br />

� Helps in selecting the best acquisition dates for the satellite<br />

images.<br />

� Useful in interpretation of spectral signatures.<br />

Rock-falls risk sub-model. � Accounts for the mass movements over slopes deepest than<br />

45°. The sub-model should be integrated to the existing FSTAB.<br />

Model tuning support system � Should help the geologist not very expert in fuzzy set logic in the<br />

definition of the inference rules and the tuning of the<br />

membership functions.<br />

3.4.3 Needs from Users Point of View<br />

The inventory of actual vegetation coverage is considered as the starting point of a monitoring program to<br />

supervise local or regional changes due to global change effects. Applying the methods developed under<br />

<strong>ALPMON</strong> it will be possible to draw up an inventory of forest habitats, tree species, tree ages, stocking<br />

JR, RSDE, ALU, LMU, Seibersdorf, WSL 89

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!