The classification scheme describes four broad classes, eleven specific categories,
and a number of detailed sub-categories, for a total of twenty-seven distinct
habitats (Table 2.1). The hierarchical scheme allows users to expand or collapse
the level of thematic detail as needed. The classification scheme also describes
eleven zones, which refer to a habitat’s location within the coral reef ecosystem.
These zones correspond to typical reef geomorphology terms found in current
scientific literature (e.g., reef flat, fore reef, back reef, reef crest), but are not
indicators of the substrate or biological cover type.
To incorporate the hierarchical classification scheme into a GIS, the authors
utilized the Habitat Digitizer Extension, a tool created for ArcView/ArcGIS by
NOAA in order to facilitate heads-up (on-screen) digitizing of benthic habitat
polygons (public domain software; search the ArcScripts database at
www.esri.com to download for free). The extension allows users to assign attributes to the benthic habitat polygons based on a predetermined and customizable
classification scheme using a point-and-click window dialog. The extension also
allows the user to set a minimum mapping unit (MMU) restriction, if desired, so
that polygons smaller than the MMU will not be digitized.
Habitat boundaries (areas with similar specific colors or patterns) are handdigitized and assigned attributes based on the classification scheme. A 1-acre
minimum mapping unit (MMU) was chosen for the NOAA benthic habitat mapping projects based on the scale of the aerial photography and the objectives of the
mapping program (Kendall et al. 2001). Occasionally the brightness, contrast, and
color of the digital image were adjusted in order to enhance subtle features and
assist with interpretation. In addition, the original 1:24,000 scale prints and diapositives were available for viewing as needed, as was external information such
as nautical charts and supplemental anecdotal evidence.
After the first draft of a habitat map was completed, scientists and local experts
visited sites in order to field check and validate information contained in the map.
The primary purposes of these visits included visually documenting those areas in
the aerial photographs that were difficult to interpret for one reason or another, as
well as verifying assigned polygon attributes. After navigating by GPS to a particular field site, scientists made observations either by snorkel from the surface,
free diving, or directly from the boat using a viewing box when conditions (depth
and water clarity) permitted. Field check observations were then used to revise and
correct the draft maps as needed. Local scientific and management experts
reviewed the final drafts, and their recommendations, especially for those polygons
labeled as ‘‘unknown’’, were incorporated into the final maps (Fig. 2.4).
The validity, or usefulness, of any interpretation or classification map may be
determined with an accuracy assessment that compares the map with what is
actually found in the field. In addition to determining the overall accuracy of a
map, it is helpful to determine the accuracy of a map from both the producer and
user points of view. Producer accuracy denotes which points on a map are classified correctly during a field assessment; user accuracy signifies the probability
that a point in a given class is actually represented by that class in the field
(Lillesand and Kiefer 1994).
38
S. A. Cochran
and a number of detailed sub-categories, for a total of twenty-seven distinct
habitats (Table 2.1). The hierarchical scheme allows users to expand or collapse
the level of thematic detail as needed. The classification scheme also describes
eleven zones, which refer to a habitat’s location within the coral reef ecosystem.
These zones correspond to typical reef geomorphology terms found in current
scientific literature (e.g., reef flat, fore reef, back reef, reef crest), but are not
indicators of the substrate or biological cover type.
To incorporate the hierarchical classification scheme into a GIS, the authors
utilized the Habitat Digitizer Extension, a tool created for ArcView/ArcGIS by
NOAA in order to facilitate heads-up (on-screen) digitizing of benthic habitat
polygons (public domain software; search the ArcScripts database at
www.esri.com to download for free). The extension allows users to assign attributes to the benthic habitat polygons based on a predetermined and customizable
classification scheme using a point-and-click window dialog. The extension also
allows the user to set a minimum mapping unit (MMU) restriction, if desired, so
that polygons smaller than the MMU will not be digitized.
Habitat boundaries (areas with similar specific colors or patterns) are handdigitized and assigned attributes based on the classification scheme. A 1-acre
minimum mapping unit (MMU) was chosen for the NOAA benthic habitat mapping projects based on the scale of the aerial photography and the objectives of the
mapping program (Kendall et al. 2001). Occasionally the brightness, contrast, and
color of the digital image were adjusted in order to enhance subtle features and
assist with interpretation. In addition, the original 1:24,000 scale prints and diapositives were available for viewing as needed, as was external information such
as nautical charts and supplemental anecdotal evidence.
After the first draft of a habitat map was completed, scientists and local experts
visited sites in order to field check and validate information contained in the map.
The primary purposes of these visits included visually documenting those areas in
the aerial photographs that were difficult to interpret for one reason or another, as
well as verifying assigned polygon attributes. After navigating by GPS to a particular field site, scientists made observations either by snorkel from the surface,
free diving, or directly from the boat using a viewing box when conditions (depth
and water clarity) permitted. Field check observations were then used to revise and
correct the draft maps as needed. Local scientific and management experts
reviewed the final drafts, and their recommendations, especially for those polygons
labeled as ‘‘unknown’’, were incorporated into the final maps (Fig. 2.4).
The validity, or usefulness, of any interpretation or classification map may be
determined with an accuracy assessment that compares the map with what is
actually found in the field. In addition to determining the overall accuracy of a
map, it is helpful to determine the accuracy of a map from both the producer and
user points of view. Producer accuracy denotes which points on a map are classified correctly during a field assessment; user accuracy signifies the probability
that a point in a given class is actually represented by that class in the field
(Lillesand and Kiefer 1994).
38
S. A. Cochran