Catalog Home Page

HyMap airborne hyperspectral imagery and field-based ecological analysis for plant community assessment on Rottnest Island, Western Australia

Mukaromah, Laily (2017) HyMap airborne hyperspectral imagery and field-based ecological analysis for plant community assessment on Rottnest Island, Western Australia. Masters by Research thesis, Murdoch University.

[img]
Preview
PDF - Whole Thesis
Download (8MB) | Preview

Abstract

Accurately classifying and mapping plant communities is essential for natural resource conservation and management planning. In recent decades, the application of hyperspectral remote sensing to biodiversity assessment and vegetation mapping has become an increasingly effective approach in supporting conservation efforts. The key focus of this work was to classify and map the vegetation on Rottnest Island, Western Australia by conducting a vegetation survey and using hyperspectral imagery. The application of HyMap airborne hyperspectral data to map plant communities seeks to capture unique spectral characteristics of plant communities, thereby allowing more accurate mapping and analysis of the complexity and heterogeneity of vegetation.

Rottnest Island, Western Australia, represents a valuable natural habitat for biodiversity conservation due to its unique features, but is highly vulnerable to disturbance, especially from invasive species. The aim of this research was twofold:

1. to classify the distribution patterns of plant communities as well as their floristic composition through a quadrat survey and investigate their relationship to environmental factors

2. to map the communities using HyMap hyperspectral data and compare the remote sensed map with the field data.

The study is divided into two major parts. Firstly, analysis of vegetation was based on a vegetation survey for 210 plots derived from a stratified random sample. TWINSPAN classification and Mutidimensional scaling (MDS) ordination were used to elucidate plant communities and the ecological relationships among them. Secondly, HyMap data was used to map the vegetation communities using the Spectral Angle Mapper (SAM) classification.

Eleven vegetation types were identified from the TWINSPAN analysis comprising a variety of heathy communities, woodlands, coastal scrub, and halophytic lake-shore communities. NMDS revealed distance from the coast and disturbance by fire as potential drivers of community distribution. Heath communities were extensive across the island, and were characterised by dominance of Acanthocarpus preissii. In addition, results underlined that disturbance likely has a considerable impact on the distribution of invasive plants, particularly Trachyandra divaricata. The HyMap image classification demonstrated its utility in mapping the vegetation with evaluation performed using accuracy matrices indicated a high accuracy (86.2%) and reliability (Kappa coefficient 0.84).

This study explored the relationship between field vegetation data and the unique spectral signatures extracted and classified for each community type. The map is an important summary of the vegetation patterns for comprehensive cover of the entire landscape of Rottnest Island and will be useful for management strategies and further research, including, monitoring change and potential disturbance impacts on the island. The high accuracy of the classification based on remotely-sensed data highlights the growing efficiency of such data as surrogates for field-based approaches to landscape analysis and interpretation.

Publication Type: Thesis (Masters by Research)
Murdoch Affiliation: School of Veterinary and Life Sciences
Supervisor: Enright, Neal and Kobryn, Halina
URI: http://researchrepository.murdoch.edu.au/id/eprint/38428
Item Control Page Item Control Page

Downloads

Downloads per month over past year