My research will focus on the impact of human activity on the vitality of coral reefs through the connectivity of marine and terrestrial components of island ecosystems and will relay on the creation of a coastal terrain model as a primary diagnostic tool. It is important to understand interactions of the system beyond the scale of the coastal interface to mesoscale level analysis of material and energy exchange in order to provide support for informed decision making to mitigate the depletion of resources and also so that human populations may be prepared to face events such as hurricanes or tsunamis. Space and sea based remote sensing technologies have enabled the collection of data to produce detailed coastal terrain models that include benthic substrate and habitat classification, ground cover characteristics and terrain detailed enough to run oceanographic, hydrographic and atmospheric simulations. Issues with analytical accuracy remain to be solved for space and aerial remote sensing technology to live up to its potential for bathymetric and reef morphology determination, benthic habitat classification with in-situ ground truth data remaining essential, but progress is being made. Geographic Information Systems (GIS) are intimately involved with finding solutions to some of these remote sensing problems and are essential to the construction of an effective terrain model.
Annotated Journal Articles
Li R., Liu J., and Felus
Y. (2001). Spatial modeling
and analysis for shoreline change detection and coastal erosion monitoring, Marine Geodesy, 24:1-12
A
coastal terrain model of the south central
Clarke S. and Burnett K. (2003). Comparison of digital elevation models for
aquatic data development, Photogrammetric
Engineering and Remote Sensing, 69(12):1367-1375
The Coastal Landscape Analysis and Modeling Study assessed the utility of 3 types of digital elevation model (DEM) datasets for accuracy in the modeling of terrestrial aquatic systems in the Coast Range Province of Oregon. Terrain models for the assessment of hydrologic units constructed using 10x10 meter resolution “drainage enforced” DEM data outclassed those made with the latest 30x30 meter USGS DEM data sets by performing as better predictors of stream placement and hierarchy and as superior classifiers of hydrologic units, slope and aspect. Drainage enforcement was defined as the addition of “breaklines” along stream courses where there are abrupt changes in terrain orientation. Readily available data sources and contemporary ESRI (ArcInfo version 7.2) software were used and compare 10 meter and 30 meter DEMs made from thirteen 7.5’ quadrangles to demonstrate advantages of finer resolution, drainage enforced data. This study provided a good description of creating a geographic information system (GIS) to determine precise terrain characteristics and classifications in completion of this analysis while also defining and demonstrating the utility of drainage enforcement in hydrologic unit definition.
Miller S.N., et al. (2007). The automated geospatial watershed
assessment tool, Environmental Modeling
and Software, 22(3):365-377
Scientists from
the
Gahegan M. and Ehlers M. (2000). A framework for the modeling of uncertainty
between remote sensing and geographic information systems, ISPRS Journal of Photogrammetry and Remote
Sensing, 55(3):176-188
Sources of uncertainty inherent to the collection and analysis of remotely sensed data and its incorporation into a GIS result from the conversion of spectral data from the sensor into an image, classification of this image data into geographic information, the manipulation of this information between different data models and projections, and its subsequent use with other data sets in geographical analyses. These sources of error build upon themselves in a manner that is difficult to quantify causing problems for geographic information scientists because good science requires the documentation of accuracy issues in order to be quantitative rather than qualitative. Specific sources of uncertainty are identified and quantified and a flow model for tracking its accumulation is proposed. Uncertainty is defined as quantifiable error with the addition of more intangible doubts resulting from analytical methods. They do a very succinct job of identifying Points in geographic conversion and classification process that leads to specific measures of uncertainty are identified and framework to track its accumulation during geoprocessing is proposed. Attention to this paper allow preparation for thoughtful and thorough selection of data sources, documentation of their lineage and processing methods as data are added to a coastal terrain model to assure data quality and increase the credibility through error/uncertainty analysis.
Kenyon J.C., et al. (2006). Towed-diver surveys, a method for mesoscale
spatial assessment of benthic reef habitat: a case study at Midway Atoll in the
Hawaiian Archipelago, Coastal Management,
34(3):339-349
The Coral Reef Ecosystem Division (CRED) of the National Oceanic and Atmospheric Administration, Pacific Islands Fisheries Science Center has built upon the experience of established reef survey methods and developed a towed SCUBA diver methodology that has been proven safe (with extensive training) and accurate in describing Meso-scale reef habitat changes. The speed at which the small boat tows the divers, 2.5 -3.5 km/hr, allows for 2 – 3 kilometers to be covered during the typical 50 minute survey while the divers control the instrument sled and record substrate data on a five minute interval while a camera photographs the substrate every 5 seconds. This data is geo-coded by a layback model as it is integrated into GISystem allowing for site specific analysis of digital images data collected during tows. Water tight descriptions of the “manta tow” survey method, the geo-rectification of photographic position and the image classification procedure used by the study are provided as well as an accuracy assessment that points out both strengths and weaknesses. These in-situ data should useful both in an analysis of changes in species diversity and distribution, as a source for ground truth data for initiating spectral classification of remotely sensed imagery, and as a final mesoscale comparison to classifications of the benthos.
Mumby P.J., et al. (2004). Remote sensing of coral reefs and their
physical environment, Marine Pollution
Bulletin, 48(3-4):219-228
Recent improvements in instrumentation and analysis have expanded scientific ability to enhance in situ reef observations with remote sensing technologies and it is useful to conduct periodic summation of a field’s progress and potential. The authors succinctly yet thoroughly answer the two questions “What ecological properties of reefs can we measure using remote sensing?” and “What environmental properties of reefs can we measure using remote sensing?” while enunciating the advantages and disadvantages of remote sensing in general and between satellite systems in particular. Advances and challenges are discussed in the realms of the use of high spatial and spectral resolution imagery in the analysis of coral reefs and other coastal systems, the combined use of SAR (synthetic aperture radar) and optical data for the classification of coastal habitats, the spatial and temporal analysis of changes in bio-optical properties of the oceans, and the use of acoustic ground discrimination systems (multibeam and sidescan sonar) for seafloor studies. The study areas of coastal zone monitoring, biological integrity and function of biotope and habitat classes, optical properties and radiative transfer modeling of aquatic systems, further use of high resolution imagery and the use of synergies of multiple remote sensing platforms are indicated as important foci for future research. The assessment of satellite data for deriving bathymetry and classifying substrate categories and the acknowledgment of the accuracy and utility of multi-beam bathymetry support the use of these technologies in further research supports their use in coastal terrain modeling.
Bello-Pineda J., Liceaga-Correa M.A.,
Hernandez-Nunez H. and Ponce-Hernandez R. (2005). Using aerial video to train the supervised
classification of Landsat TM imagery for coral reef
habitats mapping, Environmental
Monitoring and Assessment, (105):145-164
A challenge to mitigating global threats to coral reef health is the lack of information concerning the distribution, use and status of reef resources. This lack of information is particularly acute in developing countries making objective, scientifically based natural resource management decision-making difficult. Remote sensing technology has increasingly proved to be a useful tool for resource mapping, change detection, bathymetric mapping and fisheries management/stock assessment and may potentially fill the gap. This paper focuses on the use of a GIS that combines Landsat TM imagery, aerial photography, aerial video and a bathymetric model to enhance training sight selection for image classification and then provides a comparison of the outcome with maps resulting from field surveys. Landsat imagery was georeferenced with ground control points collected with a GPS using a first order polynomial transformation and then atmospherically corrected with dark pixel subtraction. Aerial photographs and videos were georeferenced to the Landsat image as a base using an affine projection. The bathymetric model was derived using methods developed by Green et al. (2000) selecting pixels of uniform substrate to define linear relationships between depth and radiance for TM spectral bands 1, 2 and 3 and applying the resulting transforms to each band. The overlay of these four data sources enabled the identification of habitats while considering different spatial and spectral resolutions as well as depth. Training sites were selected for 12 different substrate categories and the imagery was classified using a maximum likelihood algorithm. Though similar spectral characteristics caused some misclassification, a limited amount of common sense reclassification resulted in a 77% overall accuracy. Broader categories would increase accuracy, but the 77% figure falls within accepted standards and the 12 classes are more useful for resource mangers.
Stumpf R.P., Holderied
K., and Sinclair M. (2003). Determination
of water depth with high-resolution satellite imagery over variable bottom
types, Limnology and Oceanography,
48(1-2):547-556
Though methods have been developed for the determination depth from reflectance data, obstacles to accuracy remain in the form of the affects of variable bottom albedo and water clarity on the measurement of variable attenuation rates of blue, green and red spectral ranges of light – the key physical factor allowing the derivation. Due to spectral band similarity the analytical methods discussed apply to imagery from both the Ikonos and Landsat platforms, however, the high (4x4 meter) resolution of Ikonos data makes it superior to the lower resolution (30x30 meter) Landsat data for the derivation of bathymetry and terrain modeling of coral reefs. Previous attempts at deriving depth relied on a linear algorithm (Lyzenga 1978) suspected of allowing variable substrate classes and water conditions to affect depth determination. This research supports the use of a ratio transform algorithm that is less affected by environmental factors because it avoids using fixed variables and relies on fewer parameters so that it requires less “tuning”. The ratio transform, comparing reflectance values of the blue and green spectral bands is potentially more robust because the different albedo of variable bottom types affects both bands in the same manner allowing for the stronger signal of variable attenuation rates through the water column to dominate the signal. The results of both the linear and ratio transforms were compared against depths from LIDAR transects with the depth derivations from the ratio transform proving more accurate, especially in turbid waters and at depth. The limit of the ratio transform’s effectiveness was about 15 meters due to the absolute attenuation of both bands.
Headley J.D., Harborne J.D. and Mumby P.J.
(2004). Simple and robust
removal of sun glint for mapping shallow-water benthos, International Journal of Remote Sensing, 26(10):2107-2112
The problem of sun glint off of sea surface waves rendering many satellite images unusable for the analysis of coral reefs was solved by the methods described in Hochberg et al. (2003), but they leave room for improvement. As presented, the procedures are unduly sensitive to outlier values in the determination of the linear relationship between NIR and visible bands because only two pixels are used, require exhaustive masking of land and clouds due to this sensitivity, and are presented in such a mathematically rigorous manner that many potential users may miss the simplicity of the process. Methods presented in this paper build on work of Hochberg et al. (2003) to improve the robustness of the technique by utilizing a large number of pixels to perform a linear regression in establishing the NIR/visible relationship, thus eliminating the need for masking. The methods in Hochberg et al. (2003) are supported by agreement with the assumptions that shallow water can be considered optically deep to the NIR wavelengths and that the real index of refraction is nearly equal between visible and NIR wavelengths so that the relationship between NIR brightness and visible band sun glint is linear. It is then proposed that samples of multiple pixel groups that exhibit sun glint over deep water should be manually selected for a linear regression analysis of NIR brightness against each visible band. A simple equation is provided for the final correction using the slope and image pixel values and a four step implementation procedure is suggested (which would be straightforward using ENVI).
Hochberg E.J. and Atkinson M.J. (2000). Spectral discrimination of coral reef
benthic communities, Coral Reefs,
19(2):164-171
Understanding the distribution of benthic communities, even in the broad categories of coral, algae and sand, and the ability to track changes in these distributions are fundamental to the study of reef ecology while the ability to distinguish between these communities by their spectral characteristics is essential to the use of remote sensing for this application. Very little information regarding the spectral profiles of the components of reef benthos is available. This research focuses on the establishment of baseline spectral data and its use in the classification of hyperspectral imagery by collecting both ex-situ and in-situ spectral data from three coral species, five algal assemblages and three sand/coral rubble combinations and classifying a AAHIS (advanced airborne hyperspectral imaging system) image. To enhance the spectral data, fourth derivative analyses were performed for each substrate category to produce sharp peaks at the wavelengths of defining features, followed by stepwise selection to eliminate redundant wavelengths and finishing with a linear discriminant function analyses to indicate wavelengths with the best separation. These same steps were applied in classifying the imagery and the outcome was compared to distribution data resulting from SCUBA dive transects demonstrating very similar spatial distributions even though no corrections for water column effects were made. However, the imagery was not georectified nor georeferenced so that statistical comparison was not possible. Nevertheless, the high spatial correlations between the classification and ground truth along with the high spectral separation achieved in digital image processing proves the ability to separate benthic communities by spectral characteristics.
Additional Related Information Sources
(future annotations)
Journal Articles
Lundblad, E., Wright, D.J., Miller,
J., Larkin, E.M., Rinehart, R., Battista, T., Anderson, S.M., Naar, D.F., and Donahue, B.T. (2006). A benthic terrain classification scheme
for
Jiang Y.W., et al. (2004). A geographical information
system for marine management and its application to
Books
Wright, D. J.,
B.T. Donahue, and D.F. Naar (2002). Seafloor Mapping and GIS
Coordination at
Electronic Media
NOAA
Biogeography Program (2003). Benthic Habitat Mapping:
NOAA National
Centers for Coastal Ocean Science (NCCOS) (2005). Shallow-Water Benthic Habitats of
Kyle
Hogrefe mailto:hogrefek@geo.oregonstate.edu
Last updated: March 16, 2007