3.4 MODERATE SPATIAL RESOLUTION (30 METER) GLOBAL OCEAN
PRODUCTS
For coral reefs, multispectral sensors have been very successful in identifying
geomorphological zones (forereef, reef crest, reef flat, lagoon, etc.) and coral reef
habitat types for moderate complexity mapping (4-8 habitats) with 60-75%
classification accuracy. Classification accuracy declines to a little over 53% for more
than 13 classes (Andréfouët et al., 2003a, b). Accuracy of habitat classifications with
Landsat data is systematically lower than with higher spatial resolution images, such as
Ikonos (4m, Space Imaging, Inc.) data (Andréfouët et al., 2003a; Capolsini et al., 2003).
The proper spectral resolution for accurately distinguishing such important
ecological end-members as coral and algae is not available on any currently orbiting
satellite sensor (Hochberg and Atkinson, 2003). Thus, detailed assessments of the
status of coral reef communities currently requires imaging spectroscopy from airborne
platforms and cannot be effectively conducted over regional and larger scales (Mumby
et al., 1997).
3.4.1 Millennium Coral Reefs - Global reef geomorphology maps from Landsat-7
By using a consistent dataset of moderate-resolution (30 meter) multispectral
Landsat-7 images acquired between 1999 and 2003 (Gasch et al., 2000), a team led by
the Institute for Marine Remote Sensing at the University of South Florida is
developing the first high-resolution global map of shallow coral reef ecosystems. This
effort is summarized at http://eol.jsc.nasa.gov/reefs. In addition to using a consistent
data set processed with consistent methods, the project includes an unprecedented
standardization of the geomorphological description of reefs around the world to build a
globally consistent and rich classification scheme made of several hundred
classes based on geomorphology, exposure, depth and continental/oceanic attributes.
At present, the maps are being constructed scene-by-scene, with mosaics produced
for selected locations (Figure 2).Completion of the mapping is expected in
2004, with product release to follow. More information is available at
http://imars.marine.usf.edu/corals/index.html. The global archive of the underlying
Landsat data was achieved in early 2004 and has been made available to the research
community, see http://seawifs.gsfc.nasa.gov/cgi/landsat.pl.
3.5 HIGH SPATIAL RESOLUTION (≤ 20 METER) DATASETS AVAILABLE FOR
MULTIPLE REGIONS
As studies move from the global to regional and local levels, higher spatial
resolution datasets become important. It also becomes possible to purchase small
amounts of satellite data or aerial remote sensing data. Multispectral satellite sensors
remain key data sources for regional and local scale studies, including Landsat,
ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer), digital
astronaut photography, IKONOS, and Quickbird. For coastal areas and reefs close
enough to major population centers to support aircraft deployment, airborne acquisition
of photographs, multispectral data, hyperspectral data, and LIDAR (light detection and
ranging) provide additional data sources that can be combined to provide extensive
information about coastal environments. The cost effectiveness of studies using aerial
acquisition depends on both the costs of data acquisition and the cost of data
processing. For example, although LIDAR data (see Chapter 7) can provide
293
Data Synthesis for Management
PRODUCTS
For coral reefs, multispectral sensors have been very successful in identifying
geomorphological zones (forereef, reef crest, reef flat, lagoon, etc.) and coral reef
habitat types for moderate complexity mapping (4-8 habitats) with 60-75%
classification accuracy. Classification accuracy declines to a little over 53% for more
than 13 classes (Andréfouët et al., 2003a, b). Accuracy of habitat classifications with
Landsat data is systematically lower than with higher spatial resolution images, such as
Ikonos (4m, Space Imaging, Inc.) data (Andréfouët et al., 2003a; Capolsini et al., 2003).
The proper spectral resolution for accurately distinguishing such important
ecological end-members as coral and algae is not available on any currently orbiting
satellite sensor (Hochberg and Atkinson, 2003). Thus, detailed assessments of the
status of coral reef communities currently requires imaging spectroscopy from airborne
platforms and cannot be effectively conducted over regional and larger scales (Mumby
et al., 1997).
3.4.1 Millennium Coral Reefs - Global reef geomorphology maps from Landsat-7
By using a consistent dataset of moderate-resolution (30 meter) multispectral
Landsat-7 images acquired between 1999 and 2003 (Gasch et al., 2000), a team led by
the Institute for Marine Remote Sensing at the University of South Florida is
developing the first high-resolution global map of shallow coral reef ecosystems. This
effort is summarized at http://eol.jsc.nasa.gov/reefs. In addition to using a consistent
data set processed with consistent methods, the project includes an unprecedented
standardization of the geomorphological description of reefs around the world to build a
globally consistent and rich classification scheme made of several hundred
classes based on geomorphology, exposure, depth and continental/oceanic attributes.
At present, the maps are being constructed scene-by-scene, with mosaics produced
for selected locations (Figure 2).Completion of the mapping is expected in
2004, with product release to follow. More information is available at
http://imars.marine.usf.edu/corals/index.html. The global archive of the underlying
Landsat data was achieved in early 2004 and has been made available to the research
community, see http://seawifs.gsfc.nasa.gov/cgi/landsat.pl.
3.5 HIGH SPATIAL RESOLUTION (≤ 20 METER) DATASETS AVAILABLE FOR
MULTIPLE REGIONS
As studies move from the global to regional and local levels, higher spatial
resolution datasets become important. It also becomes possible to purchase small
amounts of satellite data or aerial remote sensing data. Multispectral satellite sensors
remain key data sources for regional and local scale studies, including Landsat,
ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer), digital
astronaut photography, IKONOS, and Quickbird. For coastal areas and reefs close
enough to major population centers to support aircraft deployment, airborne acquisition
of photographs, multispectral data, hyperspectral data, and LIDAR (light detection and
ranging) provide additional data sources that can be combined to provide extensive
information about coastal environments. The cost effectiveness of studies using aerial
acquisition depends on both the costs of data acquisition and the cost of data
processing. For example, although LIDAR data (see Chapter 7) can provide
293
Data Synthesis for Management