Back to search

Aristotle University Of Thessaloniki (AUTH)

Χαρτογράφηση της δασικής καύσιμης ύλης με μεθόδους τηλεπισκόπησης και τα γεωγραφικά συστήματα πληροφοριών

Abstract

dc:description

Judicial wildland fire prevention and management require precise information on fuel characteristics and spatial distribution of the various vegetation types present in an area, since fuel is the only factor related to forest fire behaviour that humans can control, manage and modify. A few years ago, the above information for pre-fire planning was relied on ground measurements, orthophotomaps and vegetation maps, resulting to the lack of up-to-date data for fuel mapping. However the recent advances in remotely sensed satellite data in conjunction with G.I.S. and special software offer great capabilities in fuel mapping. The main purpose of this thesis is to investigate the possibility of developing new methods to classify satellite imagery for more accurate and useful fuel type mapping. Three study areas were selected: the prefecture of Chalkidiki, the prefecture of Attica and the University forest of Taxiarhis in the Prefecture of Chalkidiki. In the first area, a classification method was developed; in the second one the transferability of the method was evaluated. In the third area another method of classification was developed and the results of fuel classification of the two methods were compared. A SPOT5 imagery were acquired, since they have high spectral and spatial resolution as well. An object based image analysis was employed for the first method of fuel classification. Originally, an object based model including multi-scale image segmentation and classification was developed and applied in Chalkidiki. The model created was based on a nearest neighbour classifier. The overall accuracy of the fuel classification was very satisfactory. Following, evaluation of the transferability of the developed model took place by transferring and applying the model rule set, after minor adjustments, in Attica. The overall accuracy in the second area was also very satisfactory. Extensive field work was carried out in both test sites for model development, training and evaluation. In the present thesis, a new method of fuel type mapping was developed, based on the use of neural networks. A software for classifying satellite data was developed, since the existing commercial software, relating to remotely sensed satellite data, do not include neural network classification tools or their tools lack of variety of options (training algorithms, training parameters etc.). The neural network classification was applied to the third study area. In order to increase the accuracy, some additional parameters were embodied to the initial data. Finally, a comparison of the results of the object based model and the neural network classification showed that the two methods were satisfactory accurate and the object based model classification a bit better.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2010

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Στεργιόπουλος, Ιωάννης

Subjects

dc:subject × 21

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/28267

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Στεργιόπουλος, Ιωάννης. Χαρτογράφηση της δασικής καύσιμης ύλης με μεθόδους τηλεπισκόπησης και τα γεωγραφικά συστήματα πληροφοριών. Aristotle University Of Thessaloniki (AUTH), 2010. http://hdl.handle.net/10442/hedi/28267