Primjena metoda daljinskih istaraživanja i GIS-a u kartiranju ležišta gipsa na Kosovom polju

master's thesis

• Title (english): Application of remote sensing and GIS methods for mapping gypsum occurances in Kosovo Polje
• Author: Dominik Mandić
• Mentor: Borna Lužar-Oberiter (mentor)
• Committee member: Borna Lužar-Oberiter (predsjednik povjerenstva)
• Committee member: Nenad Tomašić (član povjerenstva)
• Committee member: Kristina Pikelj (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Geology); Zagreb
• Defense date and country: 2023-09-20, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Geology

Abstract

U ovom radu istražene su mogućnosti kartiranja ležišta gipsa na Kosovom polju primjenom metoda daljinskih istraživanja i GIS-a. Postojeće geološke karte istraživanog područja su digitalizirane i vektorizirane pomoću programa QGIS. Prikupljene satelitske snimke Sentinel-2A senzora obrađene su programom SNAP, te je izrađena K-means obrada podataka u nekoliko verzija sa različitim brojem klasa. Svaka klasa opisuje dominantnu karakteristiku površine terena (stijene, vegetacija itd.). Terenski podaci su prikupljeni na ukupno 50 točaka gdje su zabilježene koordinate i opisane karakteristike površine terena. Svi podaci su uvršteni u jedinstveni GIS projekt te analizirani. K-means klasifikacija je bila u mnogo slučajeva uspješna pogotovo u verziji s većim brojem klasa, međutim radi ograničenja u prostornoj rezoluciji dobivene podatke nužno je dodatno evaluirati. Uz postojeće ulazne podatke nije bilo uvijek moguće jednoznačno razlikovati izdanke različitih vrsta stijena (npr. gipsa, vapnenca i breča) kao zasebne klase, nego su svi bili svrstani u klase karakterizirane stijenskim površinama. U ovome radu je korišten tek osnovni alat strojnog učenja, međutim rezultati nisu zanemarivi, već ih je moguće koristiti kao preliminarne u početnoj fazi geološkog kartiranja mineralnih sirovina.

Abstract (english)

In this thesis, the possibilities of mapping gypsum deposits in Kosovo polje were investigated using remote sensing methods and GIS. Existing geological maps of the research area were digitized and vectorized using the QGIS software. Collected satellite images of the Sentinel-2A sensor were processed with SNAP software, and K-means data processing was performed in several versions using different numbers of classes. Each class describes the dominant characteristic of the terrain surface (rocks, vegetation, etc.). Field data was collected at a total of 50 points where coordinates were recorded, and the characteristics of the terrain surface were described. All data were compiled in a single GIS project and analyzed. K-means classification was successful in many cases, especially in the version with a larger number of classes, however, due to limitations in spatial resolution, the obtained data must be additionally evaluated. With the existing input data, it was not always possible to unambiguously distinguish the outcrops of different rock types (eg. gypsum, limestone and breccia) as separate classes, but they were all classified into classes characterized by rock surfaces. In the presented work only a basic machine learning tool was used, however the results are not negligible. Rather, they can be used as preliminary in the initial phase of geological mapping of mineral resources.

• Keywords: evaporiti; Kosovo polje; metode daljinskih istraživanja; GIS; K-means
• Keywords (english): evaporites; Kosovo polje; remote sensing methods; GIS; K-means
• Language: croatian
• URN:NBN: urn:nbn:hr:217:137145
• Study programme: Title: Geology; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni magistar/magistra geologije (sveučilišni magistar/magistra geologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2025-01-30 08:43:02