Investigating the ability to discriminate, identify, and map coral reef benthos and substrates using remote sensing techniques

Leiper, Ian (2011). Investigating the ability to discriminate, identify, and map coral reef benthos and substrates using remote sensing techniques PhD Thesis, School of Geography Planning and Environmental Management, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
s40114060_phd_abstract.pdf Final thesis abstract application/pdf 31.98KB 1
s40114060_phd_finalthesis.pdf Final thesis application/pdf 2.21MB 14
Author Leiper, Ian
Thesis Title Investigating the ability to discriminate, identify, and map coral reef benthos and substrates using remote sensing techniques
School, Centre or Institute School of Geography Planning and Environmental Management
Institution The University of Queensland
Publication date 2011-05
Thesis type PhD Thesis
Supervisor Stuart Phinn
Arnold Dekker
Total pages 120
Total colour pages 26
Total black and white pages 94
Language eng
Subjects 04 Earth Sciences
Abstract/Summary Coral reefs are the rainforests of the sea; they are one of the most diverse and rich ecosystems on earth. Unfortunately, the health of corals reefs has been deteriorating over recent decades, largely linked to the pressures placed on them by humans – both directly and indirectly. Remote sensing can provide data collected at appropriate spatial and temporal scales on these (often) spatially expansive and remotely located ecosystems, which will assist implementation of monitoring and mitigation strategies to sustain and conserve these vital ecosystems. The ability to map coral reef health, through mapping the composition of coral reefs, using remote sensing techniques has been identified as one of the major goals of coral reef remote sensing. Research undertaken during this thesis contributes significant knowledge to achieving this goal, by investigating the ability to discriminate, identify, and map coral reef benthos and substrates using remote sensing techniques; in particular, the potential to map live coral and macro-algae, which can be used to assess coral reef health and status. Three studies are presented in this thesis; each focused on a specific objective, and collectively address the research aim and provide a significant contribution to coral reef remote sensing and the future (operational) use of satellite sensors to map and monitor coral reef health. Research was conducted at a single site: Heron Reef, Great Barrier Reef, Australia. The first study investigated the spectral reflectance of coral reef benthos and substrates at a spatial scale directly linked to the pixel size of high spatial resolution imaging systems, by incorporating multiple benthos and substrate types into a spectrometer field-of-view in-situ. 334 spectral reflectance signatures were measured of 19 assemblages of coral reef benthos and substrate types. Using first derivative values, a discrimination decision tree was developed to discriminate and identify the assemblages; it was possible to identify 15 assemblages with a mean overall classification accuracy of 62.6%. The field survey component of this study showed that much of Heron Reef was dominated (greater than 65% cover) by a single benthos or substrate type at the pixel scale of high spatial resolution remote sensors (pixel size less than 5 m). An observed similarity in spectral reflectance shape between pure endmember spectra and mixed endmember spectra dominated by a relative benthos or substrate type, indicated that much of Heron Reef would be successfully mapped using pure endmember spectra as reference data to supervise image classification. The image data would need to have high spectral, spatial, and radiometric resolution; and the mapping algorithm should be based on spectral shape rather than magnitude of reflectance. The second study investigated the ability to map benthos, substrates and bathymetry at Heron Reef using high spatial resolution hyper-spectral airborne image data and a reef-up mapping approach that was based on spectral shape (Spectral Angle Mapper). Compact Airborne Spectrographic Imager (CASI) data was acquired on Heron Reef in July 2002 and processed to convert the at-sensor radiance values to subsurface remote sensing reflectance. The spectral reflectance of coral reef benthos and substrate types were measured in-situ, and using the Hydrolight® 4.2 radiative transfer model a spectral reflectance library of subsurface remote sensing reflectance was simulated using water column depths from 0.5 to 10.0 m at 0.5 m intervals. Sediment, benthic micro-algae, turf algae, crustose coralline algae, macro-algae, and live coral were accurately mapped to a depth of around 8.0 m; in waters deeper than 8.0 m the match between the classification image and field validation data was poor. A bathymetric map was produced for water column depths of 0.5 to 10.0 m, at 0.5 m intervals, and showed high correspondence with in-situ sonar data (R2 value of 0.93). This study highlighted the importance of sensor spectral bandwidth and placement, in order to capture spectral reflectance features which characterise coral reef benthos and substrates, and a classification approach that can utilise such data. The third study assessed the ability to discriminate benthos and substrates using high spatial resolution multi-spectral satellite image data – a data source that is more likely to be used by coral reef managers and stakeholders due to its cost and availability. Specifically, this study assessed the ability of the Worldview-2 sensor, with a focus on its unique set of spectral bands, to discriminate coral reef benthos and substrates. In-situ spectral reflectance measurements of coral reef benthos and substrates were modelled to units of remote sensing reflectance, and integrated across the spectral response curves of the Worldview-2, QuickBird, CASI-2, and Hyperion sensors. Worldview-2 showed an improved ability compared to QuickBird, another high spatial resolution multi-spectral sensor, to capture the spectral profile of coral reef benthos and substrates; yet, it appeared to only adequately discriminate two benthos and substrate types (sediment and benthic micro-algae). It did not have sufficient spectral resolution to discriminate live coral and macro-algae, which is required to determine coral reef status and health. The two hyper-spectral sensors (CASI and Hyperion) were able to capture the characteristic absorption features of all benthos and substrate types due to their spectral band placement and width. These results indicated that a reef-up approach to mapping coral reef benthos and substrates (including live coral and macro-algae), using in-situ spectral reflectance measurements to supervise image classification, is not plausible using Worldview-2 due to the number, placement, and width of the sensors spectral bands. Studies undertaken during this thesis provide a significant contribution to coral reef remote sensing by establishing the link between in-situ spectral reflectance studies and coral reef mapping. Key findings from this research were that a reef-up approach using Spectral Angle Mapper and high spatial resolution hyper-spectral image data, can accurately map coral reef benthos and substrates at Heron Reef, including live coral and macro-algae, to a water depth of 8.0 m. Multi-spectral image data did not have sufficient spectral resolution to discriminate and identify coral reef benthos and substrates, which are characterised by spectral absorption features across a narrow wavelength range in a limited portion of the visible spectrum. Because current and planned hyper-spectral satellite sensors have a pixel size greater than the spatial scale of coral reef benthos and substrates on most reefs, research into spectral mixing, and unmixing, is a research priority. This is a difficult area of research, as shown by the results of the first study, which could only separate with moderate accuracy the mixed endmember assemblages of coral reef benthos and substrates measured in-situ; yet, spectral unmixing techniques must be developed to quantify the cover of benthos and substrate types such as live coral and macro-algae from moderate spatial resolution pixels of hyper-spectral satellite sensors, to determine coral reef health and status. Because the endmembers used throughout this study have spectral reflectance signatures which are consistent globally, the results presented are likely to be transferable to coral reefs other than Heron Reef.
Keyword Remote sensing
coral reef
spectral reflectance
CASI-2
Worldview-2
hyper-spectral
multi-spectral
Additional Notes Please note these numbers refer to the page number printed on the thesis document, not the number of the page of the pdf file. 17, 19-20, 24, 35, 38, 40, 46, 49, 60, 67, 69-71, 73-74, 79, 87-88, 90, 94-96, 99-100, 102.

 
Citation counts: Google Scholar Search Google Scholar
Created: Fri, 18 Nov 2011, 01:40:48 EST by Mr Ian Leiper on behalf of Library - Information Access Service