Μελέτη διαχείρισης και εκτίμησης κινδύνου πυρκαγιάς με εφαρμογή πολυκριτηριακής ανάλυσης και αλγορίθμων μηχανικής εκμάθησης σε περιβάλλον G.I.S.

Χατζηγεωργιάδης, Μηνάς

Master Thesis

Author

Χατζηγεωργιάδης, Μηνάς

Date

2022

Abstract

In the next few years, climate change is expected to lead to a significant increase of fire danger in Greece, provoking detrimental consequences in the environment and human beings. The effects of climate change have already been observable, especially during the summer period. For this reason, the term “Climate Crisis” is often used instead of “Climate Change”, so as to point out that the impacts of climate change will not be faced in the distant future; on the contrary, they are currently visible in the 21st century. The Geographical Information Systems (GIS) enable the conduction of in-depth studies and management models on fire risk, as well as the design of plans on proactive measures. In the present study, a flexible GIS-based workflow is suggested, with the aim to present the factors responsible for fire risk and their combination for fire hazard assessment, as well as monitoring shelter management in order to eliminate the possible dangers. The National Park Forest of Dadia-Lefkimi-Soufli, including the wider area of Evros, was chosen for the implementation of the workflow and the intergraded models in the study area, as it is one of the most important protected areas in Greece. The study was based on freely available and reliable research data, derived from national and international sources. These data were analysed with the aid of the GIS platform, in order to better illustrate the factors contributing to fire risk. Throughout the study, basic GIS tools and previous applications were used, such as machine learning algorithms (Support Vector Machine, Unsupported Machine Learning), map algebra and geostatistics. Afterwards, the factors were combined using multi-criteria decision analysis (MCDA) and analytic hierarchy process (AHP), in order to create a fire risk assessment model. The model was applied in Evros county (as mentioned previously) with the aim to measure the fire risk in the National Park Forest of Dadia-Lefkimi-Soufli. The results of this fire assessment study showed 3 major fire risk areas. The first one is located in the northern part on the national park, near the settlement of Giannouli. The second and widest area is located in the centre of the national park, around the invisible straight line that links the settlement of Kotronia and the settlement of Lagina. The third area is located on the south and more specifically in the northern part of the settlement Lefkimi. It should be pointed out that the average risk in the area is very high, fact that stresses the importance of monitoring the national park, especially during the summer period. The results of fire risk for the National Park Forest of Dadia-Lefkimi-Soufli, the data for the existing monitoring shelters of the fire department in Evros, along with the protected areas constituted the basis for investigating the monitoring capabilities of the fire brigade and creating a model which detects the ideal places for fire lookout tower construction. In order to evaluate the potential of the existing infrastructure, a basic visibility model was created. According to the results, the existing fire monitoring infrastructure network, in full function, covers the biggest part of the county. More specifically, the 88% of the county is monitored by at least one fire lookout and the biggest part of these districts includes forests and protected areas. The data provided from the spots of the existing fire lookout towers were processed using the Random Forest Regression algorithm and were combined with location suitability factors (altitude, road distance, slope, ridge). Also, Multicriteria analysis was used, in order to classify Evros county according to the possibility of constructing fire lookout towers. The suitable areas, according to the above mentioned criteria, cover a very limited district comparing to the total area of Evros county. Particularly, the suitable areas cover the 14% and the highly suitable areas only 1%. Consequently, the creation of a model capable of detecting these areas appears to be of great importance. Out of the highly suitable areas, 4 new spots were selected, which are expected to expand surveillance in the protected areas of the county. The selection of these spots was succeeded by the combination of the results of the visibility model (offered from the existing monitoring infrastructure of the fire brigade), protection policies in force in each area, land cover and fire risk models. Each one of the 4 spots can improve the visibility of important forestal areas, which are included in Natura network 2000. Apart from the suitability model, alternative forest monitoring methods are suggested, with the placement of monitoring devices into wind parks. In the following years, fire risk is expected to increase significantly in many areas of Greece. As a result, more and more public and private stakeholders will be obliged to face the dangers of wildfires in areas under their responsibility. In this context, the models which were created and applied in GIS platform, are presented on a general basis in the present thesis and can be used in additional areas of Greece apart from the Evros county. In addition to the model flexibility, the freely available data that were used, can give the opportunity to apply the workflow developed by any stakeholder.

Postgraduate Studies Programme

Κλιματική Κρίση και Τεχνολογίες Πληροφορικής και Επικοινωνιών / MSc in Climate Crisis and Information and Communication Technologies

Department

Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών Συστημάτων

Number of pages

196

Language

Greek

Description

Κάποιες εικόνες έχουν καλυφθεί καθώς περιέχουν εμπιστευτικά δεδομένα για τα πυροφυλάκια και τις περιπόλους της πυροσβεστικής στο Ν. Έβρου

URI

https://dione.lib.unipi.gr/xmlui/handle/unipi/14540
http://dx.doi.org/10.26267/unipi_dione/1963

Collections

Τμήμα Ψηφιακών Συστημάτων

Show full item record

Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα