Back to search

Institutes outside Greece

Mediterranean forest species mapping using hyperspectral imagery

Abstract

dc:description

Mediterranean forests, although a small fraction of the world's forest cover, need to be understood, described and monitored in order to be preserved. Spatial information about their location, their extent, their structure and their biodiversity, among others, is necessary and traditionally gathered in every available way. In particular, forest species mapping is an objective and a prerequisite in various applications that span in the fields of ecology, biology, forestry and agriculture, such as resource inventories, biodiversity assessment, fire hazard assessment, conservation planning, assessment of per-species carbon sequestration etc. Adding to that, forest species information is also essential for the national reporting obligations towards national and international policies, such as the UN Framework Convention on Climate Change (UNFCCC) and its Kyoto Protocol, the UN Convention on Biological Diversity (UNCBD), the UN Collaborative Programme on Reducing Emissions from Deforestation and Forest Degradation in Developing Countries (UN-REDD), the Ministerial Conference on the Protection of Forests in Europe (MCPFE), and the Streamlining European 2010 Biodiversity Indicators (SEBI2010). Advances in hyperspectral technology provide researchers with an opportunity to explore challenges that were either difficult or impossible to approach using multispectral data, including forest species mapping. This study examines the mapping of Mediterranean forest species, based on EO-1 Hyperion satellite spectral imagery (30m, 196 channels). Two pixel-level analysis methodologies were evaluated, based on Spectral Angle Mapper (SAM) and Support Vector Machines (SVM), as well as an object-based analysis methodology (GEOBIA). These were applied in two study areas with different species composition spatial patterns, namely the island of Thasos and Taxiarchis in Chalkidiki. Extensive fieldwork provided the reference data for estimating the accuracy of the maps. The preprocessing phase included correction steps and reduction of the hyperspectral dimension. In the case of Thasos, where there were two species of pine, the methodologies with SAM, SVM and GEOBIA, achieved overall accuracy of 94%, 89% and 85.3% respectively. In the case of Taxiarchis, where more species were present, the overall accuracy achieved was 80%, 82.6% and 74.1%. All three developed methodologies achieved very accurate results, in some cases comparable to forest inventory maps. The first (SAM) was the simplest to implement while the second required both a series of configurations and custom software. The lowest accuracy of the latest methodology (GEOBIA) can be attributed to the way accuracy is estimated, since in the case of object-based analysis, alternative methods of accurate accuracy estimation are still being investigated. The result of the proposed methodologies can meet the current needs for spatial vegetation data, both regionally and nationally. They also demonstrate the value of satellite hyperspectral imagery in mapping Mediterranean forest species.

Degree

thesis:*
Grantor dc:publisher
Institutes outside Greece
Year dc:date
2013

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Galidaki, Georgia
  • Γαλιδάκη, Γεωργία

Subjects

dc:subject × 18

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/50335

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

Galidaki, Georgia; Γαλιδάκη, Γεωργία. Mediterranean forest species mapping using hyperspectral imagery. Institutes outside Greece, 2013. http://hdl.handle.net/10442/hedi/50335