Shallow-water, aggregated cover maps were produced by combining as many as four or more detailed habitat types into general cover categories. The original detailed habitat maps were produced by rule-based, semi- automated image analysis of high-resolution satellite imagery for nine locations in the Northwestern Hawaiian Islands. This project is a cooperative effort between the National Oceanic and Atmospheric Administration, State of Hawaii Department of Land and Natural Resources, and the U.S. Fish and Wildlife Service to produce benthic habitat maps and georeferenced imagery for the Northwestern Hawaiian Islands. This project was conducted in support of the U.S. Coral Reef Task Force.
The National Ocean Service is conducting research to digitally map biotic resources and coordinate a long-term monitoring program that can detect and predict change in U.S. coral reefs, and their associated habitats and biological communities. This project was conducted in support of the U.S. Coral Reef Task Force (http://coralreef.gov).
ground condition
These maps are NOT to be used for navigation. These aggregated cover maps of the shallow-water benthic habitats of the Northwestern Hawaiian Islands have been reviewed by numerous decision makers and scientists familiar with the Northwestern Hawaiian Islands region. Although they represent the best available characterizations of the shallow-water coral reef ecosystems of this region the maps should be considered DRAFT versions.
1305 East West Highway, N/SP
NOAA, University of Hawaii, U.S. Fish and Wildlife Service
Statistical analysis to determine accuracy was performed using a subset of the 1,130 site specific benthic characterization data (376 points were used in the accuracy assessment) available for the NWHI. The overall accuracy of the aggregated cover maps was 72 percent (Kappa Statistic 0.59, Tau Coefficient 0.62), and ranged from 94 percent accuracy for unconsolidated sediment with little or no algae or seagrass to 9% for unconsolidated sediment with seagrass present. Correctly classifying benthic habitats that include algae and live coral was challenging. For example, it is difficult to separate Unconsolidated Rubble with dense (>50 percent cover) algae from Pavement with dense (>50 percent cover) algae. Also, it is difficult to separate Unconsolidated Rubble and Sand and Rubble from Pavement habitats. These challenges also were encountered when differentiating between the various habitats with live coral cover. The aggregated cover category Hardbottom with indeterminate cover was not included in the accuracy assessment. The aggregated Hardbottom with indeterminate cover includes those detailed habitat categories where a specific cover was not identified on the map, and no comparison to the field-based benthic characterization information could be performed. An assessment of the overall accuracy of the hardbottom and unconsolidated was performed. That analysis included habitat categories with indeterminate cover and their associated field-based benthic characterization data. That analysis revealed that these two habitats were correctly identified 82% and 90% of the time respectively.
Multispectral satellite imagery was positioned using National Geodetic Survey ground control data for Kure Atoll, Midway Atoll, Pearl and Hermes Atoll, Lisianski Island, Laysan Island, French Frigate Shoals, and Nihoa Island. Satellite imagery of Necker Island was positioned using satellite ephemeris data. Satellite imagery of Maro Reef was positioned using a combination of satellite ephemeris data and supplemental bathymetry transect data. Once maps were positioned, validation was performed by comparison to repositioned vector shoreline data depicted on recently-revised NOS nautical charts. The draft classified benthic habitat maps were reviewed by scientists in Hawaii with first-hand local knowledge of the Northwestern Hawaiian Islands. Comments were received and incorporated into draft maps. Draft maps also underwent a statistics-based accuracy assessment.
High-resolution satellite imagery was analyzed using rule-based, semi-automated, digital image analysis techniques to produce benthic habitat maps of the Northwestern Hawaiian Island. In total, 1,130 site specific examples of the spectral and spatial characteristics of differing habitats were available and incorporated into the analysis process for the NWHI. The minimum mapping unit (MMU) for identifying habitats or features was 100 sq. m. Generally, features to a depth of approximately 30 m can be analyzed from the satellite imagery. Due to the depth of water surrounding Gardner Pinnacles, very little characterization of the benthic habitats of Gardner Pinnacles was possible. No satellite imagery was available for Gambia shoals.
Kure Atoll, Midway Atoll, Pearl and Hermes Atoll, Lisianski Island, Laysan Island, French Frigate Shoals, and Nihoa Island are positioned to within five (5) meters at sea level. Necker Island is positioned to within 15 meters at sea level. Maro Reef is positioned to within eight (8) meters at sea level. Spatial data meets or exceeds national map accuracy standards for fixed features.
none reported
High resolution satellite imagery was analyzed to produce benthic habitat maps of the Northwestern Hawaiian Island.
High-resolution satellite imagery was analyzed using rule-based, semi-automated, digital image analysis techniques to produce benthic habitat maps of the Northwestern Hawaiian Island. Shallow-water, aggregated habitat cover maps were produced by combining (aggregating) as many as four or more detailed habitat categories into generalized cover habitat types. The list of these aggregated cover types and those detailed habitat categories included in each aggregation are presented below. Following careful evaluation of the digital satellite imagery, selected sites were visited in the field for typological validation. This validation included: (1) areas in the satellite imagery with confusing or difficult to interpret spectral signatures, (2) transects across many representative habitat types occurring in different depths and water conditions, (3) a survey of the Zones, and (4) confirmation of preliminary habitat delineations after a first draft habitat map was produced. Navigating to field sites was accomplished in a variety of ways including uploading position coordinates from the mosaic into an onboard GPS and navigating to those waypoints, using an onboard PC connected to GPS allowing navigation using digital nautical charts or the mosaic, and visual navigation using landmarks visible in the satellite imagery. On most occasions, field activities were conducted with the guidance of local experts. Satellite imagery was used in the field to facilitate comparison of signatures in the imagery to actual habitats at each site. Individual sites were visually evaluated by snorkeling and free diving or directly from the boat in shallow, clear water. Habitat transitions were evaluated by swimming transects across habitat types to further guide placement of habitat boundaries. Habitat type(s), zone, approximate depth, position (GPS), image number, and other descriptive information were recorded at each site. Field data for each site were then compiled into a text table with GPS position fields to allow overlay of the field information on the mosaic and habitat maps. Where depth and water clarity permitted, the satellite images were used to navigate across multiple bottom features allowing continuous confirmation of habitat types and transitions between each site.
Benthic features were assigned one of thirty habitat types ranging from sand to coral. A complete description of these classification categories can be found in the NWHI_class_scheme.pdf (http://ccmaserver.nos.noaa.gov/rsd/products/pdf/NWHI_class_scheme.pdf). RESULTS AGGREGATED COVER TYPE: Habitat_name, Numeric_value, Red_color_value, Green_color_value, Blue_color_value RESULTS DETAILED HABITAT TYPE: Habitat_name, Habitat_classification_code
NOS - Biogeography Team
The name of the benthic habitat identified and depicted.
NOS - Biogeography Program
A hierarchical, 4-digit code assigned to each habitat type.
A two- or three-digit numeric value assigned to each habitat type.
A numeric value ranging from 0-255 assigned to the red channel of the RGB color palette.
A numeric value ranging from 0-255 assigned to the green channel of the RGB color palette.
A numeric value ranging from 0-255 assigned to the blue channel of the RGB color palette
Included all classes in the hardbottom substrate category that had more than 10 percent live coral cover. Hardbottom that was not classified with respect to structure (class 2020), but which had live coral, was also included in this category. Bottom habitat consisting of mixtures of live coral, crustose coralline algae and macroalgae also dell into this category as long as there was enough live coral to meet the 10 percent threshold.
Aggregated Cover Type
Detailed Habitat Type
Included all classes in the hardbottom substrate category that had more than 10 percent live crustose coralline algal cover, but not more than ten percent live coral cover. Bottom habitats with a mixture of crustose coralline algae, live coral (not more than 10 percent) and macroalgae fell in this category if there was enough crustose coralline algae to meet the 10 percent threshold.
Included all classes in the hardbottom substrate category that had less than 10 percent live coral cover, crustose coralline algal cover or macroalgal cover.
NOS - Biogeography
Included all classes in the hardbottom substrate category that had more than 10 percent macroalgae, but not more than 10 percent live coral cover or 10 percent crustose coralline algal cover. Typically these habitats also had a high percentage of turf algal cover.
NOS- Biogeography Program
Included all classes in the hardbottom substrate category for which a specific bottom cover was not identified. Included were area classified only to the first or second level of the detailed classification scheme.
Biogeography Program
NOS Biogeography Program
Included all classes in the unconsolidated substrate category with not more than 10 percent cover of live submerged vegetation (macroalgae/seagrass).
Included all classes in the unconsolidated substrate category with more than 10 percent cover of live submerged vegetation (macroalgae/seagrass). Seagrass was very uncommon in NWHI (only a small area was found at Midway during the field survey) and was not identified on any of the habitat maps.
Aggregated Cover Sites
Areas where the bottom cannot be detected (identified) in the imagery. In most cases, this occurs in water more than 30 m deep.
Areas determined to be above the water line in the imagery at the time the imagery was acquired.
Areas where the bottom type is unclassified because of turbidity in the water, surface glint, or other types of interference. This category also includes areas where the seabed cannot be classified due to a lack of supplemental field-based information of for other reasons.
Referes to areas within the geographic bounds of the habitat map that lie outside the bounds of the aquired imagery. This is not a habitat category in the classification scheme.
Not to be used for navigation.
