Tải bản đầy đủ - 0 (trang)
An Integrated Approach to Benthic Habitat Mapping Using Remote Sensing and GIS: An Example from the Hawaiian Islands

An Integrated Approach to Benthic Habitat Mapping Using Remote Sensing and GIS: An Example from the Hawaiian Islands

Tải bản đầy đủ - 0trang

212



A.E. Gibbs and S.A. Cochran



recreational opportunities. Many of the world’s reefs have been severely damaged

over the past few decades due to a combination of factors including habitat destruction, land-based pollution, sedimentation, overfishing, vessel groundings, coastal

development, disease, and climate change.

In contrast to many coral reefs around the world, where nearly 70% are threatened or destroyed, Hawai`i’s coral reefs are generally in good condition, with most

degradation occurring near urban areas and at popular tourist destinations in response to land-based sources of pollution, overfishing, recreational overuse, and

invasive species (Wilkinson 2004, Friedlander et al. 2005).

With escalating population and development pressures, Hawai`i’s coral reefs will

increasingly become threatened. Baseline habitat maps and monitoring programs are

an essential step toward evaluating reef health and assisting in management of these

important resources. Prior to the 1990s, however, very few maps of the modern,

shallow-water coral reefs of Hawai`i existed.



9.1.1 History of Coral Reef Mapping in the Hawaiian Islands

Early works were merely descriptive in their nature (for example, MacCaughey

1918, Pollock 1928). Beginning around 1960, the coral reefs in Kane`ohe Bay, O`ahu

became some of the most heavily studied in the islands due to tremendous degradation, as well as their proximity to researchers at the University of Hawai`i marine

lab on Coconut (Moku O Lo`e) Island. While the body of work in Kane`ohe Bay is

extensive, most of these coral reef surveys, and others conducted around the state

during the late 20th century, used a line-transect method to assess coral health and

coverage for various other studies (for example, Banner 1968, Fitzhardinge 1985,

Alifio 1986), and as such, provided no maps or method of quantifying the spatial

extent of Hawaiian reefs.

In 1984, the University of Hawai`i Marine Options Program undertook a coral

reef mapping effort on the island of Moloka`i for the U.S. Army Corps of Engineers (Manoa Mapworks 1984). Qualitative field data were collected over a twoweek period using scuba and snorkel, and maps were plotted using 1:6 K and 1:24 K

black-and-white aerial photography from 1975 as a base layer. These maps provide a

useful background to the Moloka`i reef ecosystem, however, the aerial photographs

were not georeferenced, and thus no accurate measurements of scale and distribution can be made from them.

In 1998, the U.S. Geological Survey (USGS) recognized the need for accurate

maps of Hawai`i’s coral reefs to provide a baseline for future change assessments

and that these maps should include the geometry and distribution of coral cover

(Field and Reid 1998). In order to be useful for management decisions, the accuracy of the maps would need to be verified using ground-truth methods. Concurrently, in response to the mandate set forth by Executive Order 13089, the National

Oceanic and Atmospheric Administration (NOAA) National Ocean Service (NOS)

implemented a program to provide digital maps of coral reefs within U.S. waters,

including territories, for use in a Geographic Information System (GIS). A digital



9



An Integrated Approach to Benthic Habitat Mapping



213



Fig. 9.1 Map of the main eight Hawaiian islands. The dark black lines show the locations of recent

coral reef mapping efforts by the USGS



atlas of the benthic habitats of the main Hawaiian Islands was completed by NOAA’s

Biogeography team in 2003 (Coyne et al. 2003).

Coral reef mapping efforts by the USGS in the Hawaiian islands include major efforts on the south coast of Moloka`i (Cochran-Marquez 2005) and within or

adjacent to three National Park lands along the Kona coast of Hawai`i: Pu`ukohola

Heiau National Historic Site (PUHE), Kaloko-Honokohau National Historical Park

(KAHO), and Pu`uhonua O Honaunau National Historical Park (PUHO) (Gibbs

et al. 2007, Cochran et al. 2007a,b). Limited reconnaissance mapping was also conducted off the south shore of O`ahu and the west coast of Maui (Gibbs et al. 2005)

(Fig. 9.1). A combination of aerial photography, high-resolution lidar bathymetry,

and in situ observations were utilized in each of these mapping efforts. The methodologies employed and general results from one of the National Park studies, KalokoHonokohau, are presented here.



9.1.2 The Kaloko-Honokohau Study Area

Kaloko-Honokohau National Historical Park is one of three National Park lands located along the western coast of the island of Hawai`i and the only one to include

submerged lands and marine resources within its official boundaries. The park was

established in 1978 and is 1,160 acres in size, including 596 acres of marine area.

KAHO is located adjacent to a moderately well-developed area of the Kona coast.

The park is bordered on the south by the Honokohau small boat harbor and on the

north by a luxury residential/resort and golf course development near Wawahiwa`a



214



A.E. Gibbs and S.A. Cochran



Point (Fig. 9.2). Future development slated for lands adjacent to the southern boundary of the park include a 300% expansion of the small boat harbor along with construction of hotels, condominiums, and a light industrial park.

Marine resources located within KAHO include coral reef and habitat for many

marine animals such as the green sea turtle and a variety of fish and invertebrates. In

addition, many archeological, cultural, and recreational resources are located within

the marine realm of the park, including ancient fishponds and popular scuba diving destinations. Potential threats and stressors to the modern marine environment

include groundwater and surface-water contamination, invasive plants and algae,

fishing pressure, use of monofilament gill nets (which can ensnare marine life or



Fig. 9.2 Location map showing the boundaries of Kaloko-Honokohau National Historical Park

and the area mapped as part of this study



9



An Integrated Approach to Benthic Habitat Mapping



215



become tangled on reefs and left behind as fishing debris), and visitor use impacts,

such as scuba diving and snorkeling. Illegal dumping, oil releases, boat groundings, and other physical damage to reef resources are potential threats from users

of the nearby harbor. A specific issue of concern for the park includes establishing

baseline conditions of the offshore resources prior to the development of adjacent

coastal lands.

In 2003, the U.S. Geological Survey (USGS) Coastal and Marine Geology Program, in cooperation with the National Park Service (NPS), was tasked with developing a detailed benthic-habitat classification map for the marine lands within and

adjacent to the park. The intent of this project was to provide baseline maps, a GIS

database, and a report summarizing the biological and geological resources of these

marine lands in order to facilitate the management, interpretation, and understanding of park resources. The report (Gibbs et al. 2007) and data generated are available

online at: http://pubs.usgs.gov/sir/2006/5256/(last access on 11 March 2008).



9.2 Data and Methods

9.2.1 Classification Standards

A standard for characterization of coral-reef environments was first implemented by

NOAA for mapping the Florida Keys (Rohman and Monaco 2005) and Puerto Rico

and the Virgin Islands (Kendall et al. 2001). This standard for mapping coral reefs

in the United States and its territories characterizes benthic habitats on the basis

of their sea-floor geomorphology, geographic zones, and biological cover using a

minimum mapping unit of one acre. Typically, only color aerial photography or

satellite imagery is used to define habitat boundaries and field reference checks are

conducted using shipboard video or scuba transects.

In the study presented here, benthic-habitat classification maps were created using the standards established by NOAA but at a larger scale (minimum mapping

unit of 100 m2 versus 1 acre) and with additional data sources, including existing

color aerial photography, Scanning Hydrographic Operational Airborne Lidar Survey (SHOALS) bathymetric data, and georeferenced underwater video and still photography. Maps were generated using both ArcView and ArcMap GIS software by

ESRI (http://www.esri.com; last access on 11 March 2008), and a statistical analysis

of accuracy of the resultant maps was performed.



9.2.2 Base Imagery and Data

9.2.2.1 Aerial Photography

High-resolution aerial photomosaics offer a relatively inexpensive and easily acquired foundation for mapping shallow-water structures and features of coral reefs.

As a passive form of remote sensing, aerial photography can provide an excellent



216



A.E. Gibbs and S.A. Cochran



overview of reef habitats due to the typically clear, shallow-water environments in

which reefs are found. Two of the primary limitations in the use of aerial photography for mapping coral reefs are: (1) the absorption of light by the water column precludes interpretation of bottom habitats in water depths greater than approximately

20 m, and (2) the remotely sensed data must be integrated with actual in-the-water

field (groundtruth) observations in order to determine actual sediment type and live

coral abundance, type, and distribution.

In this study, color aerial photographs were used as the base layers for mapping.

The images were scanned, orthorectified, and a digital mosaic with a resolution of

0.16 m per pixel was produced. The horizontal accuracy of this photography was

better than 2 m and most seafloor features were recognizable to a water depth between 15 and 25 m.

9.2.2.2 High-Resolution Bathymetry

High-resolution bathymetry was a second integral base data type used for delineating the habitat and morphological environments in this study. Scanning Hydrographic Operational Airborne Lidar Survey (SHOALS) bathymetric data were collected over the Hawaiian islands during 1999 and 2000 by the U.S. Army Corps

of Engineers. Lidar is an active remote sensing technology that utilizes laser energy to detect distance between source and receiver. The SHOALS technology (see

also Lillycrop et al. 1996, Guenther et al. 2000, Irish et al. 2000) determines water

depth by comparing the time difference between a pulse of laser energy reflected

off the surface of the water and one reflected off the sea floor. This time difference

is difficult to resolve in shallow water (<∼ 1 m) or where waves are breaking. The

maximum water depth of data collection is limited by the combined effects of the

incident sun angle and intensity, the reflectance or radiance of bottom material, and

water clarity–including the type and quantity of particles in the water column. The

SHOALS system is typically capable of sensing bottom depths equal to two or three

times the Secchi depth–the depth to which an 8-inch (20-cm) disk with alternating

black and white quadrants can be seen from the surface (Tyler 1968). In the KAHO

study area, the maximum water penetration was 42 m. The bathymetric data have a

nominal horizontal point spacing of 4 m (±3 m) and a vertical resolution of ±15 cm.

For further details regarding SHOALS data, see http://shoals.sam.usace.army.mil

(last access on 11 March 2008).

Continuous bathymetric data were obtained for nearly the entire park area, with

the exception of the shallow coastal waters in Honokohau and Kaloko Bays and

a swath of missing data in the central part of Honokohau Bay. Bathymetry in this

central area was obtained from the historical National Ocean Service (NOS) survey H09336 of 1968 (http://www.ngdc.noaa.gov/mgg/bathymetry/hydro.html; last

access on 11 March 2008). From the combined SHOALS/NOS data, a triangulated

irregular network (TIN) of the point data was generated, from which gridded surfaces were created. Isobaths, hillshades, and slope maps were derived from these

grids using standard ArcMap functions and then used to assist in the interpretation

of seafloor morphology and benthic habitat distribution.



9



An Integrated Approach to Benthic Habitat Mapping



217



9.2.3 Groundtruth Data

9.2.3.1 Underwater Video and Still Photography

Several types of camera systems and collection methods were used to collect the

groundtruth data. In water depths shallower than a boat could navigate, and along

scuba transects, video and still photography were collected by snorkelers and scuba

divers using hand-held video recorders and digital still cameras. A Global Positioning System (GPS) unit mounted on a surface float tethered to the snorkeler or diver

provided positional information for the imagery.

In deeper water, video imagery was obtained by either towing a camera behind a

moving vessel or dropping it over the side of the vessel while remaining on a fixed

station or drifting slowly. Camera tows were designed to rapidly collect video imagery over large geographical areas. To avoid collisions with the seafloor, however,

the camera had to be towed several meters above the bottom. This limited the observable detail of the seafloor in the video and these images were thus most useful

for providing information regarding spatial transitions in coral cover and habitat. In

contrast, during collection of video using the on-station drop or drift configuration,

the camera could be lowered to within centimeters of the seafloor, which provided

exceptionally detailed information on substrate type, benthic cover, and habitat

diversity.

The camera system used for shipboard operations was a watertight video camera

illuminated with a light-emitting diode (LED) light ring designed by SeaViewer

Underwater Video Systems (http://www.seaviewer.com; last access on 11 March

2008). When rigged for towing, the camera was mounted in a small aluminum frame

with a rear-mounted plastic fin (Fig. 9.3A). When rigged for on-station dropping,

the camera and light were integrated with two battery-powered lasers and a Seabird

CTD (Conductivity, Temperature, Depth) instrument in a steel frame (Fig. 9.3B).

Live video from both systems was viewed in a shipboard laboratory on a monitor

and recorded directly to miniDV tape (Fig. 9.3C). Time, date, location, and ship

speed were overlaid on the video using Sea-Trak GPS Video Overlay, also developed

by the SeaViewer Company.

Simultaneous navigation, recording of ship position, and feature annotation were

conducted in real time using hardware and software developed by Red Hen Systems (RHS; http://www.redhensystems.com; last access on 11 March 2008) on a

PC laptop. Location data were recorded using a hand-held GPS receiver. The RHS

hardware transmitted NMEA-formatted GPS data at two-second intervals to the first

audio channel of the video tape. A database was simultaneously created to crossreference the GPS locations and video time codes. This technique allowed for navigation and video to be viewed in real time and the location of features of interest

and comments (for example, start/end of lines, substrate types) to be added to the

database during data collection. For post-survey analyses, this technique allowed

rapid random access to the original video by selecting locations along the navigation trackline within MediaMapper and GeoVideo (an extension developed by Red



218



A.E. Gibbs and S.A. Cochran



Fig. 9.3 Photograph of the

(A) SEAVIEWER camera

system rigged to collect

towed video, and (B) with

SeaBird CTD acquisition

system rigged to collect drop

or drift video and CTD

information. (C) Photograph

of shipboard laboratory set up

for navigation, recording of

ship’s position, and

annotation of bottom features



Hen Systems for integration with the ESRI ArcMap platform) software packages.

Video could be interactively queried and geographically referenced feature annotations could be added to the database.

Nearly 48 trackline kilometers (22 h) of underwater video footage and more than

500 still images (89 towed lines, 124 on-station drop/drift sites, 5 scuba transects,

and 3 snorkel transects), were collected during three field surveys between December 2003 and August 2004 (Fig. 9.4).



9



An Integrated Approach to Benthic Habitat Mapping



219



Fig. 9.4 Aerial photomosaic of the study area overlain with video trackline locations



9.2.3.2 Video Mosaics

Recent advances in software development have allowed digital video to be converted to georeferenced image strips that can be imported into a GIS. Researchers at

the University of New Hampshire (UNH) are developing software tools for pattern

recognition from one video frame to the next, which results in a continuous image

mosaic made from overlapping video frames (Rzhanov et al. 2004). Collaborators

from the USGS and UNH used the sea-floor video acquired from KAHO to make

georeferenced mosaics of selected tracklines within the study area. Original video

on miniDV tape was converted into Audio Video Interleave (AVI) format using commercial software and then divided into 2-min sections in order to limit file size and

to minimize propagation errors. Using the suite of UNH-developed software, the



220



A.E. Gibbs and S.A. Cochran



Fig. 9.5 Example of an underwater video mosaic overlain on aerial photography. (A) Aerial photomosaic with habitat polygon boundaries, and (B) with video mosaic added



video was then de-sampled, keeping every 15th–20th frame (depending on camera

sled speed), and the outer edges of the AVI were cropped to remove the navigation

and time stamps (Sea-Trak) that were superimposed on the video. With the clean

AVI, both automatic and manual pattern recognition were performed, calculating

the X-Y shift and rotation from one frame to the next. An image mosaic was generated using the video frames and the offset information. Finally, the image mosaics

were georeferenced using a combination of GIS techniques, the navigation information on the original video, and comparison with the aerial photography where

shallow portions of the reef were visible. Once the imagery was properly georeferenced, it was available within the GIS for direct comparison and groundtruthing of

the benthic habitats (Fig. 9.5).



9.2.4 Benthic Habitat Mapping Using GIS

Digital benthic habitat maps were created using ESRI’s ArcMap 8.3 and ArcView 3.2 software with a habitat digitizing extension created by NOAA (see

http://www.ccma.nos.noaa.gov/products/biogeography/digitizer/ to download the

extension; last access on 11 March 2008). The habitat digitizing extension allows

users to delineate habitat areas and assign attributes to the habitat polygons based

on a predetermined classification scheme using a point-and-click menu system.



9



An Integrated Approach to Benthic Habitat Mapping



221



We delineated and classified 1,185 polygons, covering more than 2, 479 km2 . A

minimum mapping unit (MMU) of 100 m2 was used; however, smaller features were

mapped if they carried habitat significance (for example, an individual coral colony

2 m in diameter located in an otherwise uncolonized area). Features were digitized

and interpreted primarily using the orthomosaics. In areas where seafloor features

were too deep to be resolved in the aerial photograph, the morphological characteristics of the sea floor observed in the bathymetry, combined with underwater imagery,

were used to define and classify the habitat polygons.



9.2.5 Classification Scheme

The classification scheme used was based on a scheme established by NOAA’s biogeography program in 2002 (Coyne et al. 2003) for the main eight Hawaiian islands

and subsequently revised in 2004 (NOAA 2005). Developed with input from coral

Table 9.1 List of individual habitat components in the classification scheme. Numbers represent

the 4-digit identifier (ABCD)

Major structure (A)



Dominant structure (B)



1 Unconsolidated Sediment



1 Mud

2 Sand



2 Reef and Hardbottom



1 Aggregate Reef

2 Spur-and-Groove

3 Individual Patch Reef

4 Aggregated Patch Reef

5 Volcanic Pavement with 10–50% Rocks/Boulders

6 Volcanic Pavement

7 Volcanic Pavement with > 50% Rocks/Boulders

8 Volcanic Pavement with Sand Channels

9 Reef Rubble



3 Other



0 Unknown

1 Land

2 Artificial

3 Artificial/Historical



9 Unknown



0 Unknown



Major biologic cover (C)



Percent cover (D)∗



0 Unknown

1 Uncolonized

2 Macroalgae

3 Seagrass

4 Coralline Algae

5 Coral

6 Turf

7 Emergent Vegetation

8 Mangrove

9 Octocoral



0 Unknown

2 10– < 50%

3 50– < 90%

4 90–100%

∗ Because < 10% coverage is considered to be

uncolonized, percent category (1) < 10% is not used.



222



A.E. Gibbs and S.A. Cochran



reef scientists, managers, local experts, and others, the hierarchal scheme allows

users to expand or collapse the level of thematic detail as necessary. We used the

NOAA definition of benthic habitats and classification scheme as a starting point

to provide continuity with other habitat maps. We then made modifications to the

classification scheme where necessary to improve the characterization of benthic

habitats and geologic substrates found along the Kona coast.

The classification scheme uses four basic attributes to describe each mapped

polygon: (A) the major structure of the substrate, (B) the dominant structure (a subset of the major structure), (C) the major biologic cover found on the substrate,

and (D) the percentage of major biological cover. The structure combined with the

overlying biologic cover is referred to as a “habitat”. Each polygon is coded with

a 4-digit identifier (ABCD) that reflects the combination of the individual habitat

components. See Table 9.1 for a list of habitat components. If a polygon includes

two or more substrate or coverage types, the polygon is identified with the dominant

one. Each polygon is also coded with a fifth attribute, “zone”, which refers to the

location of the habitat community within the coral reef ecosystem. These zones correspond to typical reef geomorphology (for example, reef flat, reef crest, fore reef,

shelf, etc.; Spaulding et al. 2001, Kendall et al. 2001, Coyne et al. 2003). Detailed

descriptions of habitats and zones, including example photographs, may be found

in Gibbs et al. (2007).



9.3 Application of Techniques and Habitat Classification

9.3.1 Habitat and Geology

A thematic map showing the distribution of benthic habitats within the study area

is shown in Fig. 9.6. This 2-dimensional thematic map provides a highly detailed

representation of the aerial extent and distribution of the various benthic habitats.

The ability to directly query the digital data and view the data in 3 dimensions,

however, expands the utility and usefulness of the information, allowing for a more

complete assessment of the environment and the interpretation of potential controls

on the distribution of corals in this environment.

Evaluation of the bathymetric data, including the development of shaded relief

(hillshades) and slope maps, along with profile transects or cross-sections, provided

information on the shape or morphology of the seafloor (Fig. 9.7). Groundtruth data

provided additional detail on the structures at a finer scale than the bathymetric

data allowed. Gibbs et al. (2007) describe the geology of the marine region of the

park and suggest that the underlying geologic framework and morphology of the

submerged volcanic flows provide the primary control on benthic habitats within

the park. The seafloor of KAHO is composed of multiple, smooth to undulating

pahoehoe-type basalt flows that form flat to gently sloping benches, shear vertical walls, and steep flow-front escarpments. In some locations the basalt surface



Tài liệu bạn tìm kiếm đã sẵn sàng tải về

An Integrated Approach to Benthic Habitat Mapping Using Remote Sensing and GIS: An Example from the Hawaiian Islands

Tải bản đầy đủ ngay(0 tr)

×