11th ICRS Abstract book - Nova Southeastern University

11th ICRS Abstract book - Nova Southeastern University 11th ICRS Abstract book - Nova Southeastern University

24.12.2012 Views

Oral Mini-Symposium 17: Emerging Techniques in Remote Sensing and Geospatial Analysis 17-9 Mapping The Habitats And Biodiversity Of Ningaloo Reef, Western Australia Using Hyperspectral Imagery Halina KOBRYN* 1 , Nicole PINNEL 1 , Thomas HEEGE 2 , Lynnath BECKLEY 1 , Matt HARVEY 1 , Suzanne LONG 3 1 Environmental Science, Murdoch University, Murdoch, WA 6150, Australia, 2 EOMAP GmbH&Co KG, D-82205 Gilching, Germany, 3 Department of Conservation and Environment, Kensington, Australia The largest hyperspectral survey of a coral reef (3400 km 2 ) was undertaken in April 2006 and forms the core data set for mapping habitat components and biodiversity of the Ningaloo Marine Park, Western Australia. Optically deep waters of this region are ideally suited for remote sensing techniques and airborne data were collected by HyVista. The data are at 3.5 m spatial resolution for a 1km wide terrestrial coastal strip and out to 20m depth over lagoon and reef areas and covers wavelengths from visible to near infrared at 15nm intervals. Hyperspectral data were corrected for atmospheric, air-water interface and water column effects using the physics-based Modular Inversion & Processing System. This approach allows for quantitative and automated steps as well as the removal of subjectivity from the classification process, allowing improved transferability to additional sampling locations, field spectral datasets and extension of the monitoring to other seasons. Underwater field spectra were collected using an OceanOptics spectrometer as well as underwater photographs, to allow for accurate interpretation and validation. Results of this mapping can be compared to the transect data collected by divers from other studies and also to earlier habitat maps prepared by expert interpretation of aerial photography. Comparisons of classification results for Coral Bay area show promising results in the discrimination of branching, tabulate and massive corals as well as macro-algal assemblages. Remote sensing offers unique tools which are non-invasive, quantitative and enable mapping of large areas into a seamless data set which can be integrated with human use data, oceanographic circulation models and other spatial data sets. The hyperspectral data are being used to develop a high-resolution characterisation of the entire reef, shallow water habitats and terrestrial landforms of the coastal strip in order to support sound conservation and management of the Ningaloo Marine Park. 17-10 Development Of A Field Test Environment For The Validation Of Coastal Remote Sensing Algorithms: Enrique Reef, Puerto Rico James GOODMAN* 1 , Miguel VELEZ-REYES 2 , Samuel ROSARIO 2 , Shawn HUNT 2 , Fernando GILBES 2 1 University of Puerto Rico at Mayaguez, Miami, FL, 2 University of Puerto Rico at Mayaguez, Mayaguez, Puerto Rico Remote sensing is increasingly being used as a tool to quantitatively assess the location, distribution and relative health of coral reefs and other shallow aquatic ecosystems. As the use of this technology continues to grow and the analysis products become more sophisticated, there is an increasing need for comprehensive ground truth data as a means to assess the algorithms being developed. The University of Puerto Rico at Mayagüez, one of the core partners in the NSF sponsored Bernard M. Gordon Center for Subsurface Sensing and Imaging Systems, is addressing this need through the development of a fully-characterized field test environment on Enrique Reef in southwestern Puerto Rico. This reef area contains a mixture of benthic habitats, including areas of seagrass, sand, algae and coral, and a range of water depths, from a shallow reef flat to a steeply sloping forereef. The objective behind the test environment is to collect multiple levels of image and field data with which to validate physical models, inversion algorithms, feature extraction tools and classification methods for subsurface aquatic sensing. Data collected from Enrique Reef currently includes airborne, satellite and field-level hyperspectral and multispectral images, in situ spectral signatures, water bio-optical properties and information on habitat composition and benthic cover. We present a summary of the latest results from Enrique Reef, discuss our concept of an open testbed for the remote sensing community and solicit other users to utilize the data and participate in ongoing system development. 17-11 Remote Sensing Of Coral Reef Biogeochemistry Based On Optical Absorptance And Light-Use Efficiency Eric HOCHBERG* 1 , Marlin ATKINSON 1 1 Hawaii Institute of Marine Biology, University of Hawaii, Honolulu, HI We have developed a remote sensing technique for measuring coral reef benthic primary productivity based on light-use efficiency and optical absorptance. The model is GPP = EdAε, where GPP is gross primary production, Ed is seafloor-incident irradiance, A is seafloor absorptance, and ε is the benthic community’s light-use efficiency. Both Ed and A are derivable from various remote sensing data sources. Given appropriate values for ε, it is therefore possible to use remote sensing to determine GPP across spatial scales (meters to many kilometers) and in different reef environments (e.g., fore reef, reef flat). We demonstrate the utility of our approach for measuring the range and distribution of GPP across a reef system. Because reef calcification is inherently linked to photosynthesis, it is possible to define a separate light-use efficiency (εcalc) and thus remotely sense that rate, as well. 17-12 An Investigation Into The Effects Of Coral Cover, Colony Size-Frequency Distribution And Clustering On The Classification Accuracy Of Simulated Reef Images. Alan LIM* 1 , Peter MUMBY 2 , John HEDLEY 2 , Chris ROELFSEMA 3 , Ellsworth LEDREW 1 1 University of Waterloo, Waterloo, ON, Canada, 2 University of Exeter, Exeter, United Kingdom, 3 University of Queensland, Brisbane, Australia Numerous studies have been conducted to compare the classification accuracy of coral reefs maps produced from satellite and aerial imagery at different spatial or spectral resolutions or from images processed to different levels. So far no work has been done that specifically looks at the differing spatial elements of the coral reef ecosystem and their effects on classification accuracy. In this study, we will examine how accuracy is affected by spatial elements of the reefscape by investigating the effects of colony size-frequency distribution, spatial aggregation of the coral colonies and the proportion of live coral cover at different spatial resolutions. One of the main difficulties of such a study would be the acquisition of images that could be used to represent these different reef-scape scenarios. Additional difficulties in using actual imagery would include ensuring that improved accuracy in one instance was not the result of advantageous environmental conditions or on how well the accuracy assessment was carried out. Thus, in order to investigate these issues, we created simulated spectral reef images which could be manipulated to reflect the desired reefscapes. With simulated images, only those characteristics of the reef that were of interest would be allowed to vary, allowing us complete confidence in the results of the accuracy assessments. The spatial elements determine the proportion of pixels in each class which are spectrally ‘purer’ due to lower amount of inter-class mixing. These factors will influence the number of ‘purer’ pixels such that images with higher coral cover, colony clustering and skew will have a greater proportion of ‘purer’ pixels. However, an increase in this proportion is not always related to increases in levels of accuracy. An analysis of the interactions between the various spatial elements provides an insight into how these spatial elements influence classification accuracy. 143

Oral Mini-Symposium 17: Emerging Techniques in Remote Sensing and Geospatial Analysis 17-13 Comparison Of in Situ Temperature Data From The Southern Seychelles With Sst Data: Can Satellite Data Alone Be Used To Predict Coral Bleaching Events? Ben STOBART* 1 , Nigel DOWNING 2 , Raymond BUCKLEY 3 , Kristian TELEKI 4 1 Marine Reserves, Spanish Institute of Oceanography, Palma de Mallorca, Spain, 2 Cambridge University Coastal Research Unit, Cambridge University, Cambridge, United Kingdom, 3 College of Ocean and Fishery Sciences, University of Washington, Seattle, WA, 4 International Coral Reef Action Network (ICRAN), Cambridge, United Kingdom Degree-heating-weeks data derived from satellite sea surface temperature (SST) readings are increasingly being used to predict where bleaching is likely to occur, though predictions have not always been calibrated and corroborated by field observations. While SST can provide a good indication of water temperature, local oceanographic conditions will determine the depth to which satellite SST readings are representative. In 2003 the Aldabra Marine Programme initiated a temperature monitoring network, currently involving 40 temperature data loggers, in the southern Seychelles at Aldabra, Assomption, Astove and St. Pierre The annual temperature cycle in the region involves a shorter period of winter lows (min 23 ºC) between June and October, and extended high summer temperatures (max 30 ºC) between December and April. We compare SST temperature data from the four locations with in situ temperature measurements at 6m, 10m and 20m depth. In situ data is most similar to satellite SST during the winter period (typically not more than 1 ºC difference), and differs most during the summer period (up to 4 ºC difference). This seasonal difference is due to water column stratification during the summer, which is typified by calm weather with weak winds (though remaining a period of occasional cyclone activity). While during the winter rough weather fuelled by strong southeasterly winds reduces stratification. During the period of stratification in situ water temperatures are most similar to satellite data at 6m, followed less so by 10m and 20m depth. A combination of greater stratification during the summer period, along with periods of cool water upwelling, may in some cases reduce the reliability of satellite derived SST data for predicting bleaching events. We propose that during bleaching events exposure of coral communities at Astove, St Pierre and select sites at Aldabra to thermal stress may be reduced by local oceanographic conditions. 17-14 New Ecological Insights From A 21-Year Coral Reef Temperature Anomaly Database Kenneth CASEY* 1 , Elizabeth SELIG 2 , John BRUNO 3 1 National Oceanographic Data Center, NOAA, Silver Spring, MD, 2 Curriculum in Ecology, University of North Carolina - Chapel Hill, Chapel Hill, NC, 3 Department of Marine Science, University of North Carolina - Chapel Hill, Chapel Hill, NC A wide range of new ecological insights has been enabled by advances in satellite remote sensing of the physical characteristics of the ocean. Dramatically improved algorithms coupled with advances in computational capabilities have resulted in new products with finer resolution, longer temporal coverage, greater accuracy, and better consistency. The Coral Reef Temperature Anomaly Database (CoRTAD), based on 21 years of AVHRR Pathfinder sea surface temperatures, is one such product that is yielding insights into the spatial and temporal characteristics of thermal stress and its influence on coral bleaching and disease. The development of the CoRTAD will be presented along with selected research highlights from its application to understanding thermal stress patterns, coral disease, and marine protected area design and effectiveness. In addition, information on how the extensive collection of information in the CoRTAD can be accessed and applied to global, regional, and local coral studies will be provided. 17-15 A Methodology For Using Satellite-Based Temperature And Light Measurements For Predicting Coral Bleaching Severity And Mortality William SKIRVING* 1 , Roberto IGLESIAS-PREITO 2 , Susana ENRIQUEZ 2 , Tyler CHRISTENSEN 1 , John HEDLEY 3 , Mark EAKIN 1 , Ove HOEGH-GULDBERG 4 , Sophie DOVE 4 , Scott HERON 1 , Peter MUMBY 3 , Alan STRONG 1 , Gang LIU 1 , Jessica MORGAN 1 , Dwight GLEDHILL 1 1 NOAA Coral Reef Watch, Silver Spring, MD, 2 Universidad Nacional Autonoma de, Puerto Morelos, Mexico, 3 University of Exeter, Exeter, United Kingdom, 4 University of Queensland, St Lucia, Australia The current NOAA Coral Reef Watch (CRW) suite of satellite products is designed to help coral reef managers monitor heat stress to better understand and predict mass coral bleaching. Although these products perform well when used to describe the onset of coral bleaching, they are not as accurate in describing the severity and mortality associated with mass coral bleaching events. The coral bleaching HotSpot and Degree Heating Week products are based purely on sea surface temperature (SST), yet coral bleaching is a physiological response that results from a combination of temperature and light. Here, we describe a potential major evolution of the NOAA CRW satellite products. A new methodology under development combines satellitederived SST data with a new satellite-derived solar radiation product to better predict the severity and mortality of mass coral bleaching events. This new methodology is novel in that it goes beyond just examining the thermal stress, but actually combines thermal stress measurements of the existing CRW suite with light measurements from the Geostationary Environmental Satellites to provide a measure of the total photo-thermal damage. 17-16 Producing A Satellite Sst Climatology – How Long Is A Piece Of String? Scott F. HERON* 1 , William J. SKIRVING 1 , Gang LIU 2 , Tyler R.L. CHRISTENSEN 2 , C. Mark EAKIN 1 , Jessica A. MORGAN 2 , Dwight K. GLEDHILL 2 , Alan E. STRONG 1 1 NOAA Coral Reef Watch, Silver Spring, MD, 2 IMSG at NOAA Coral Reef Watch, Silver Spring, MD Coral reef ecosystem stress often occurs in response to abnormal environmental conditions (e.g., temperature, salinity, and light) rather than the absolute level of these. For example, corals in the Persian Gulf are accustomed to warmer summer conditions than corals off the coast of Brazil; as such, ocean temperatures of 30°C would be “comfortable” for the former but stressful for the latter. Identifying anomalous conditions requires good knowledge of the baseline of usual (“normal”) conditions. Here we discuss whether the existing 22-year satellite sea surface temperature (SST) record is of sufficient length to calculate a long-term average (climatology) that can sensibly be used as a baseline for monitoring the health of corals. We also discuss issues related to global warming in determining this baseline and the relevance of adaptation by corals. At present, NOAA Coral Reef Watch uses near-real-time satellite temperatures to determine regions that experience thermal anomalies that have been linked to coral bleaching events. Inherent within the present operational process is a SST climatology that was defined using satellite SST data and which underpins the satellite-monitoring success. 144

Oral Mini-Symposium 17: Emerging Techniques in Remote Sensing and Geospatial Analysis<br />

17-9<br />

Mapping The Habitats And Biodiversity Of Ningaloo Reef, Western Australia Using<br />

Hyperspectral Imagery<br />

Halina KOBRYN* 1 , Nicole PINNEL 1 , Thomas HEEGE 2 , Lynnath BECKLEY 1 , Matt<br />

HARVEY 1 , Suzanne LONG 3<br />

1 Environmental Science, Murdoch <strong>University</strong>, Murdoch, WA 6150, Australia, 2 EOMAP<br />

GmbH&Co KG, D-82205 Gilching, Germany, 3 Department of Conservation and<br />

Environment, Kensington, Australia<br />

The largest hyperspectral survey of a coral reef (3400 km 2 ) was undertaken in April 2006<br />

and forms the core data set for mapping habitat components and biodiversity of the<br />

Ningaloo Marine Park, Western Australia. Optically deep waters of this region are ideally<br />

suited for remote sensing techniques and airborne data were collected by HyVista. The<br />

data are at 3.5 m spatial resolution for a 1km wide terrestrial coastal strip and out to 20m<br />

depth over lagoon and reef areas and covers wavelengths from visible to near infrared at<br />

15nm intervals. Hyperspectral data were corrected for atmospheric, air-water interface<br />

and water column effects using the physics-based Modular Inversion & Processing<br />

System. This approach allows for quantitative and automated steps as well as the removal<br />

of subjectivity from the classification process, allowing improved transferability to<br />

additional sampling locations, field spectral datasets and extension of the monitoring to<br />

other seasons. Underwater field spectra were collected using an OceanOptics<br />

spectrometer as well as underwater photographs, to allow for accurate interpretation and<br />

validation. Results of this mapping can be compared to the transect data collected by<br />

divers from other studies and also to earlier habitat maps prepared by expert<br />

interpretation of aerial photography. Comparisons of classification results for Coral Bay<br />

area show promising results in the discrimination of branching, tabulate and massive<br />

corals as well as macro-algal assemblages. Remote sensing offers unique tools which are<br />

non-invasive, quantitative and enable mapping of large areas into a seamless data set<br />

which can be integrated with human use data, oceanographic circulation models and<br />

other spatial data sets. The hyperspectral data are being used to develop a high-resolution<br />

characterisation of the entire reef, shallow water habitats and terrestrial landforms of the<br />

coastal strip in order to support sound conservation and management of the Ningaloo<br />

Marine Park.<br />

17-10<br />

Development Of A Field Test Environment For The Validation Of Coastal Remote<br />

Sensing Algorithms: Enrique Reef, Puerto Rico<br />

James GOODMAN* 1 , Miguel VELEZ-REYES 2 , Samuel ROSARIO 2 , Shawn HUNT 2 ,<br />

Fernando GILBES 2<br />

1 <strong>University</strong> of Puerto Rico at Mayaguez, Miami, FL, 2 <strong>University</strong> of Puerto Rico at<br />

Mayaguez, Mayaguez, Puerto Rico<br />

Remote sensing is increasingly being used as a tool to quantitatively assess the location,<br />

distribution and relative health of coral reefs and other shallow aquatic ecosystems. As<br />

the use of this technology continues to grow and the analysis products become more<br />

sophisticated, there is an increasing need for comprehensive ground truth data as a means<br />

to assess the algorithms being developed. The <strong>University</strong> of Puerto Rico at Mayagüez,<br />

one of the core partners in the NSF sponsored Bernard M. Gordon Center for Subsurface<br />

Sensing and Imaging Systems, is addressing this need through the development of a<br />

fully-characterized field test environment on Enrique Reef in southwestern Puerto Rico.<br />

This reef area contains a mixture of benthic habitats, including areas of seagrass, sand,<br />

algae and coral, and a range of water depths, from a shallow reef flat to a steeply sloping<br />

forereef. The objective behind the test environment is to collect multiple levels of image<br />

and field data with which to validate physical models, inversion algorithms, feature<br />

extraction tools and classification methods for subsurface aquatic sensing. Data collected<br />

from Enrique Reef currently includes airborne, satellite and field-level hyperspectral and<br />

multispectral images, in situ spectral signatures, water bio-optical properties and<br />

information on habitat composition and benthic cover. We present a summary of the<br />

latest results from Enrique Reef, discuss our concept of an open testbed for the remote<br />

sensing community and solicit other users to utilize the data and participate in ongoing<br />

system development.<br />

17-11<br />

Remote Sensing Of Coral Reef Biogeochemistry Based On Optical Absorptance And<br />

Light-Use Efficiency<br />

Eric HOCHBERG* 1 , Marlin ATKINSON 1<br />

1 Hawaii Institute of Marine Biology, <strong>University</strong> of Hawaii, Honolulu, HI<br />

We have developed a remote sensing technique for measuring coral reef benthic primary<br />

productivity based on light-use efficiency and optical absorptance. The model is GPP = EdAε,<br />

where GPP is gross primary production, Ed is seafloor-incident irradiance, A is seafloor<br />

absorptance, and ε is the benthic community’s light-use efficiency. Both Ed and A are derivable<br />

from various remote sensing data sources. Given appropriate values for ε, it is therefore<br />

possible to use remote sensing to determine GPP across spatial scales (meters to many<br />

kilometers) and in different reef environments (e.g., fore reef, reef flat). We demonstrate the<br />

utility of our approach for measuring the range and distribution of GPP across a reef system.<br />

Because reef calcification is inherently linked to photosynthesis, it is possible to define a<br />

separate light-use efficiency (εcalc) and thus remotely sense that rate, as well.<br />

17-12<br />

An Investigation Into The Effects Of Coral Cover, Colony Size-Frequency Distribution<br />

And Clustering On The Classification Accuracy Of Simulated Reef Images.<br />

Alan LIM* 1 , Peter MUMBY 2 , John HEDLEY 2 , Chris ROELFSEMA 3 , Ellsworth LEDREW 1<br />

1 <strong>University</strong> of Waterloo, Waterloo, ON, Canada, 2 <strong>University</strong> of Exeter, Exeter, United<br />

Kingdom, 3 <strong>University</strong> of Queensland, Brisbane, Australia<br />

Numerous studies have been conducted to compare the classification accuracy of coral reefs<br />

maps produced from satellite and aerial imagery at different spatial or spectral resolutions or<br />

from images processed to different levels. So far no work has been done that specifically looks<br />

at the differing spatial elements of the coral reef ecosystem and their effects on classification<br />

accuracy. In this study, we will examine how accuracy is affected by spatial elements of the<br />

reefscape by investigating the effects of colony size-frequency distribution, spatial aggregation<br />

of the coral colonies and the proportion of live coral cover at different spatial resolutions.<br />

One of the main difficulties of such a study would be the acquisition of images that could be<br />

used to represent these different reef-scape scenarios. Additional difficulties in using actual<br />

imagery would include ensuring that improved accuracy in one instance was not the result of<br />

advantageous environmental conditions or on how well the accuracy assessment was carried<br />

out. Thus, in order to investigate these issues, we created simulated spectral reef images which<br />

could be manipulated to reflect the desired reefscapes. With simulated images, only those<br />

characteristics of the reef that were of interest would be allowed to vary, allowing us complete<br />

confidence in the results of the accuracy assessments.<br />

The spatial elements determine the proportion of pixels in each class which are spectrally<br />

‘purer’ due to lower amount of inter-class mixing. These factors will influence the number of<br />

‘purer’ pixels such that images with higher coral cover, colony clustering and skew will have a<br />

greater proportion of ‘purer’ pixels. However, an increase in this proportion is not always<br />

related to increases in levels of accuracy. An analysis of the interactions between the various<br />

spatial elements provides an insight into how these spatial elements influence classification<br />

accuracy.<br />

143

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!