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 />

forest. Furthermore, streams could not be classified due to their representation in mixed pixels<br />

together with bank vegetation. On the other hand the auxiliary data supported classification itself by<br />

establishing rules for selected classes (e.g. sealed surfaces do not occur on steep slopes).<br />

Partly, additional features derived from the high resolution satellite data sets (see WP8) were<br />

integrated in the classification process. Texture features were used for the delineation of sealed<br />

surfaces in the Dachstein and Cordevole test sites. The forest / non-forest separation in the Dachstein<br />

test site was based on a fused Landsat TM/Spot pan image.<br />

Some main difficulties crystallised in most classifications:<br />

� Forest age in many test sites could not be classified due to the natural appearance of the forest<br />

with a mixture of all tree age categories. Only the Dachstein test site is dominated by managed<br />

forest with a common age class within one stand.<br />

� Forest type in some test sites is mainly mixed, with only a few pure coniferous or broad-leaved<br />

stands.<br />

� Tree species composition is an important information for two applications (national park<br />

management and avalanche risk). However, some tree species could not be separated (spruce,<br />

pine, and stone pine). Dwarf mountain pine could not be separated sufficiently with the satellite<br />

data available for most test sites. This is on one hand due to spectral similarity with other<br />

coniferous trees and, on the other hand, due to mixture with other tree species in one stand.<br />

However, in the Dachstein and the Mangfall test site it was possible to classify dwarf mountain pine<br />

and green alder by means of winter respectively spring images, where the low growing trees are<br />

covered with snow whereas the high growing trees are snow free.<br />

� Wet lands were well recognised in the Mangfall test site by using the thermal infrared band of<br />

Landsat TM (TM6). However, this approach is not applicable e.g. to the Dachstein test site, as wet<br />

land areas here are relatively small and beyond the resolution of TM6 with 120m.<br />

� Sealed surfaces cannot be separated from rock, gravel, and soil based only on the spectral<br />

characteristics of these classes, as the spectral signatures are too similar. Additional information<br />

has to be introduced to classify this category with higher accuracy, or the information must be<br />

taken from GIS data.<br />

On the other hand, most categories required by the customers could be classified with sufficient<br />

accuracy. In general no problems occurred with the classification of the forest type in those areas,<br />

where pure coniferous or broad-leaved stands are common. Forest age could be classified at least in<br />

some test sites, where pure age stands occur. In the Dachstein test site it was possible to separate<br />

pure larch stands from mixed spruce/larch stands and pure spruce stands (possibly mixed with pine<br />

and /or stone pine). In the Dachstein and Mangfall test sites even dwarf mountain pine and green alder<br />

could be classified by means of winter satellite data. Furthermore, some classes could be divided into<br />

subclasses (e.g. pasture and meadow, carbonate and silicate rocks) giving more detailed information.<br />

Problems with the classification of water bodies or sealed surfaces could be overcome by introducing<br />

auxiliary data into the classification process. In general, it can be stated that the outcome of<br />

classification was satisfactory with respect to the local as well as the European wide applications. This<br />

was confirmed by the customers in the first <strong>ALPMON</strong> customer work shop held from February 4-5,<br />

1999 at Innsbruck, as well as in the final project evaluation.<br />

2.2.7 Verification of Classification Results (WP10)<br />

The verification phase was intended to give evidence on the quality of the classification results<br />

(Congalton, 1991). The results of the verification are confusion matrices for each classification giving<br />

true error rate estimates for each class and for the total classification. The accuracy of the<br />

classification results was determined for all classified parameters based on the ground information<br />

collected at each sample plot. The statistical analyses of accuracy included Kappa statistics, confusion<br />

matrices, regression analyses of observed versus estimated values, etc.<br />

All analysis of classification results was based on the use of an error matrix or contingency table. It is a<br />

very effective way to represent accuracy, in that the accuracy of each category is plainly described<br />

along with both the errors of inclusion (commission errors) and errors of exclusion (omission errors)<br />

present in the classification. Error matrix can be used for a series of descriptive and analytical<br />

statistical techniques.<br />

JR, RSDE, ALU, LMU, Seibersdorf, WSL 23

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

Saved successfully!

Ooh no, something went wrong!