University of the Aegean
Τηλεπισκόπηση, τεχνητή νοημοσύνη και συστήματα γεωγραφικών πληροφοριών στην εκτίμηση κινδύνου δασικών πυρκαγιών
Abstract
dc:descriptionFire danger rating systems have been adopted by many developed countries dealing with wildfire prevention and presuppression planning, so that civil protection agencies are able to define areas with high probabilities of fire ignition and resort to necessary actions. This dissertation focuses on the development of a fire ignition risk scheme that can be an integral component of a quantitative Fire Danger Rating System. The methodology used, estimates the geo-spatial fire risk regardless of fire causes or expected burned area, while it has the ability of forecasting based on meteorological data. The main output of the scheme is the Fire Ignition Index (FII) which is based on three other indices: the Fire Weather Index (FWI); the Fire Hazard Index (FHI); and the Fire Risk Index (FRI). These indices are not just a relative probability of fire occurrence, but a rather quantitative assessment of fire danger in a systematic way. Remote sensing data have been utilized in order to retrieve part of the input parameters to the scheme, while Remote Automatic Weather Stations and the SKIRON/Eta weather forecasting system provided real-time and forecasted meteorological data, respectively. The relationship between wildfire occurrence and the input parameters has been investigated by neural networks, whose training was based on historical data. The FWI function was more easily approached, while the FRI had better classification percentages regarding the validation fires. According to the operational validation under realistic conditions, the FII identified the high risk areas where most of the fire ignited. The dissertation also presents the results of sensitivity analysis performed in a back-propagation neural network in order to distinguish the influence of each variable in the proposed fire ignition risk scheme. To evaluate the three indices developed within the above scheme, 5 different methods were used. Results showed that the presence of rainfall, the 10-h fuel moisture content and the month of the year parameter are the most significant variables of the FWI, FHI and FRI, respectively. On the contrary, relative humidity, elevation and day of the week have a small contribution to fire ignitions in the study area. The results retrieved by sensitivity analysis can be used by forest managers and other decision-makers dealing with wildfires in order to take the appropriate preventive measures, emphasizing on the important factors of fire occurrence.
Degree
thesis:*- Grantor dc:publisher
- University of the Aegean
- Year dc:date
- 2007
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Vasilakos, Christos
- Βασιλάκος, Χρήστος
Subjects
dc:subject × 20- Συστήματα γεωγραφικών πληροφοριών
- Τηλεπισκόπηση
- Τεχνητή νοημοσύνη
- Νευρωνικά δίκτυα
- Πυρκαγιές, Δασικές
- Εκτίμηση κινδύνου
- Γεωπληροφορική
- Geographical Information Systems (GIS)
- Remote sensing
- Artificial intelligence
- Neural networks
- Wildfires
- Risk assessment
- Geoinformatics
- Φυσικές Επιστήμες
- Γεωεπιστήμες και Επιστήμες Περιβάλλοντος
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Earth and Related Environmental Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/18686