11th ICRS Abstract book - Nova Southeastern University

Poster Mini-Symposium 17: Emerging Techniques in Remote Sensing and Geospatial Analysis
17.573
Determination Of Water Depth From Ikonos And Quickbird Satellite Imagery
Donald FIELD* 1 , Amit MALHOTRA 2
1 NOAA Center for Coastal Fisheries and Habitat Research, Beaufort, NC, 2 NOAA Center
for Coastal Fisheries and Habitiat Research, Beaufort, NC
IKONOS and QuickBird multispectral images were used to create bathymetric maps for
an area of approximately 100 sq. km. in the Dry Tortugas. A previously described
algorithm utilizing a ratio of the blue and green bands for deeper areas and the blue and
red bands for shallow areas was applied to both image sources. A georeferenced
database of nearly 25,000 soundings was used to develop the algorithm and assess the
accuracy of the final bathymetric maps. Initial application of the algorithm to the
QuickBird imagery yielded much lower accuracies than with IKONOS. Closer
examination of the QuickBird imagery revealed two issues: 1) radiometric miscalibration
in the green band; and 2) an irregularity referred to as the “circuit board pattern”, that was
most obvious over areas of deep water. While the initial algorithm was based on a linear
function, an additional ratio algorithm was developed for the QuickBird imagery based
on a second order polynomial regression curve. The final bathymetric maps for both
image sources were able to compensate for variable bottom types and albedo (sand,
seagrass, colonized hardbottom). Both image sources were also able to predict depths up
to 15 m, but past that depth, error between the bathymetry estimates and known depths
increased rapidly. While the overall accuracies for both image sources were similar, the
more variable, noisy nature of the QuickBird data may be inadequate for some uses.
17.574
Mapping Coral Reef Benthic Communities With High Spatial Resolution Image
Data
Ian LEIPER* 1 , Stuart PHINN 1 , Chris ROELFSEMA 1
1 University of Queensland, Brisbane, Australia
An new approach using image pixel scale in-situ spectral reflectance libraries to
supervise classification of high spatial resolution satellite image data was tested on Heron
Reef, Great Barrier Reef, Australia. Spectral reflectance signatures of homogeneous
benthic features (~2.0 cm sample area diameter) on coral reefs have previously been
shown to be distinguishable from one another. However, the occurrence of such pure
endmembers is rare at the spatial scale of high spatial resolution airborne and satellite
sensors (1.0 - 5.0 m pixel size). We investigated the separability of in-situ spectral
signatures collected at a lower spatial resolution (~1m sample area diameter) that is more
appropriate to the types of reef communities being mapped with currently available high
spatial resolution remote sensors. 3500 geo-located photos were collected during a field
survey of Heron reef, and used to create a classification scheme based on Bray-Curtis
Similarity Analysis. For each class, in-situ spectra were then collected on Heron Reef
using an Ocean Optics VIS-NIR spectrometer across a range of 400-900nm, with a 1.0 m
sample area diameter. Multivariate techniques were used to explore these spectral
reflectance signatures and identify features that could be used to discriminate between the
classes. A Quickbird multi-spectral satellite image was classified using the in-situ
spectral library to supervise classification, and a standard error assessment conducted
using the field survey data. This research presents an approach to mapping coral reef
benthic communities based on benthic community spectral properties.
17.575
Pre-Processing Of 2005 Aviris Data For Coral Reef Analysis
brad LOBITZ* 1 , liane GUILD 2 , marcos MONTES 3 , roy ARMSTRONG 4 , James
GOODMAN 5
1 California State University, Monterey Bay, moffett field, CA, 2 NASA Ames Research Center,
moffett field, CA, 3 Naval Research Laboratory, Washington, DC, 4 Dept. of Marine Sciences,
Univ. of Puerto Rico, Mayaguez, Mayaguez, Puerto Rico, 5 Center for Subsurface Sensing and
Imaging Systems, Univ. of Puerto Rico, Mayaguez, Mayaguez, Puerto Rico
Abstract – In December 2005, the Airborne Visible Infrared Imaging Spectrometer (AVIRIS)
was flown on the Twin Otter over Puerto Rico and the US Virgin Islands for assessment of the
2005 Caribbean coral reef bleaching event. A hand-held spectroradiometer was also operated in
the field to collect concurrent field data for dominant bottom types and coral. Benthic-type
classifications of the AVIRIS imagery will be generated with these spectral data, after a number
of image processing steps. An overview of the pre-processing phases, including stray light and
sun glint suppression and atmospheric correction with Tafkaa will be presented.
Keywords: coral reefs, airborne remote sensing, AVIRIS, imaging spectrometry, hyperspectral,
Tafkaa.
17.576
Use Of Aerial Photographs And Acoustic Techniques For Mapping Benthic Marine
Communities Of Martinique
Helene LEGRAND* 1 , Ian SOTHERAN 2 , Robert FOSTER-SMITH 2 , Cécile PERES 3 , Yann
ROUSSEAU 3 , Philippe LENFANT 4 , René GALZIN 4 , Jean-Philippe MARECHAL 3
1 Observatoire du Milieu Marin Martiniquais, FORT DE FRANCE, Martinique, 2 Envision,
Newcastle, United Kingdom, 3 Observatoire du Milieu Marin Martiniquais, Fort de France,
Martinique, 4 EPHE - Université Perpignan, Perpignan, France
A remote sensing survey of the marine benthic communities was conducted along the whole
coastal zone of Martinique (French West Indies). A Geographic Information System (GIS) was
developed to locate and assess the health status of the marine habitats.
Several techniques (visual and acoustic) were combined to produce the final maps. Benthic
communities in shallow waters (0-7 m) were identified visually from aerial photographs (IGN
2004 campaign) and ground-trusted. Habitat health status was assessed based on
hypersedimentation, macroalgae and coral necrosis levels.
The sublittoral zone (7-30m) was surveyed using Acoustic Ground Discrimination System
(RoxAnn) and interferometric swath bathymetric system. The relatively high resolution of the
swath system provided information on bedform features and major seabed sediment categories
as well as a detailed bathymetry. A drop-down video camera was used to check the biology and
physical characteristics of the seafloor.
Six community types were identified: coral community, seagrass, mixed community
(assemblage of coral, sponges and gorgonians with seagrass beds), soft bottom community,
sponge and gorgonian community, and algal community.
The final maps show significant differences in the distribution of habitats and communities
between the Caribbean and the Atlantic sides. The Atlantic coast has an outer spur and groove
reef and inner lagoon system which extend about 4kms away from the land. The Caribbean
coast has sporadic fringing reefs amongst sandy sediment habitats. The largest reef is located in
the south.
Community health appears largely degraded all around the island, especially in large bays and
downstream of river mouths, where anthropogenic impacts are more important.
The amount of numeric data collected and the marine GIS developed are valuable resources for
managers and policy makers for the coastal zone preservation of Martinique.
407