Simulacijski model promjene zemljišnog pokrova u Požeško-slavonskoj županiji

master's thesis

• Title (english): The simulation model of land cover change in the Požega-Slavonia County
• Author: Tomislav Jogun
• Mentor: Aleksandar Lukić (mentor)
• Committee member: Aleksandar Lukić (predsjednik povjerenstva)
• Committee member: Ivan Zupanc (član povjerenstva)
• Committee member: Marin Cvitanović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Geography); Zagreb
• Defense date and country: 2016-02-22, Croatia
• Scientific / art field, discipline and subdiscipline: INTERDISCIPLINARY AREAS OF KNOWLEDGE; Geography

Abstract

Predmet ovog rada su stohastični simulacijski modeli za promjene zemljišnog pokrova. Prostorno-vremenski okvir istraživanja je Požeško-slavonska županija 1985. – 2027. godine. Zemljišni pokrov i njegove promjene 1985. – 2013. godine utvrđene su hibridnom nadziranom klasifikacijom Landsat snimki, uz visoku ostvarenu točnost. Glavni trend u tom razdoblju bila je ekstenzifikacija zemljišnog pokrova. Tematske karte zemljišnog pokrova i prostorne varijable bile su ulazni podaci za simulacijski model koji je bio implementiran u dodatku MOLUSCE za QGIS. Pri kalibraciji simulacijskih modela korištene su umjetne neuronske mreže (ANN) i logistička regresija (LR). Predviđanje je načinjeno metodom ćelijskih automata (CA). Validacija testnih modela pokazala je da su uspješnije predviđali količinu od lokacija promjena zemljišnog pokrova. Također, ANN su bile uspješnije od LR, a fuzija pojedinačnih rezultata poboljšala je slaganje simulacija s referentnim podacima. Konačni modeli predviđaju nastavak ekstenzifikacije zemljišnog pokrova 2013. – 2027. godine. Prošle i buduće promjene zemljišnog pokrova odražavaju duboke razvojne probleme Požeško-slavonske županije.

Abstract (english)

The paper focuses on the stochastic simulation models for land cover change. Spatio-temporal framework of the research is the Požega-Slavonia County from 1985 to 2027. The land cover and its changes from 1985 to 2013 were determined by the hybrid supervised classification of Landsat imagery, with high accuracy achieved. The main trend in this period was the extensification of the land cover. Thematic maps of the land cover and spatial variables were input data for the simulation model which was implemented in the MOLUSCE plugin for QGIS. Artificial neural networks (ANNs) and logistic regression (LR) were used for the calibration of simulation models. The prediction was performed using cellular automata (CA). Validation of the test models showed that they have more successfully predicted quantity of the land cover change than its location. Also, ANNs were more successful than LR, and the fusion of individual results improved agreement between the simulation and reference data. The final models predict a continuation of the land cover extensification from 2013 to 2027. The past and future land cover changes reflect profound development problems of the Požega-Slavonia County.

• Keywords: ćelijski automati; logistička regresija; MOLUSCE; Požeško-slavonska županija; promjene zemljišnog pokrova; simulacijski model; umjetne neuronske mreže
• Keywords (english): cellular automata; logistic regression; MOLUSCE; Požega-Slavonia County; land cover change; simulation model; artificial neural networks
• Language: croatian
• URN:NBN: urn:nbn:hr:217:556650
• Study programme: Title: Geography; specializations in: Physical Geography with Geoecology, Spatial Planning and Regional Development, Heritage and Tourism, Geographic Information Systems; Course: Geographic Information Systems; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra geografije (magistar/magistra geografije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-03 15:33:44