Transcript No Slide Title

TAKING COASTAL MAPPING TO A NEW LEVEL
Assessing Habitat Composition and Water Properties of Shallow Coastal Ecosystems, along the
Coast of Puerto Rico using Hyperspectral Remote Sensing
Miguel Goenaga-Jimenez Ph.D Student
e-mail: [email protected]
Dr. James Goodman
e-mail: [email protected]
Dr. Miguel Velez-Reyes - Advisor
e-mail: [email protected]
Laboratory of Applied Remote Sensing and Image Processing.
Bernard M. Gordon Center for Subsurface Sensing and Imaging Systems
University of Puerto Rico at Mayagüez Campus
CCRI
ABSTRACT
RELEVANCE TO THE GOALS OF CCRI
DESCRIPTION OF MAJOR TASKS
The main purpose of the planned project is to provide a more
comprehensive assessment and mapping of the shallow coastal
resources for selected areas along the coast of Puerto Rico.
Specifically, we will extract information on habitat composition and
water properties using the enhanced capabilities of hyperspectral
remote sensing. Output from this research will exceed the existing
map resources for this area by providing managers with a spatially
explicit indication of the distribution of existing coastal resources .
(see Fig. 1 for an example of the proposed resource assessment and
mapping).
The proposed research products directly address the CCRI
research priority for the basic assessment of resources. The
resulting maps provide a quantified assessment of benthic cover
over a broad geographic scale, as well as supplemental information
on bathymetry and water optical properties. This will provide
research scientists and resource managers a means to assess the
spatial relationships between habitat distribution and environmental
stressors, and ultimately a quantifiable baseline for evaluating future
changes to the ecosystem.
The analysis procedure uses a sequence of image processing
steps to resolve the complex interaction of atmospheric conditions,
bathymetry, sea surface state, water optical properties and bottom
composition (summarized in Figure 4 and 5). After preprocessing,
which includes atmospheric correction and sunglint removal, a semianalytical optimization model is employed to retrieve bathymetry and
water properties throughout the study area. Then, using field spectra
data representing the dominant benthic components (i.e., spectral
endmembers for sand, coral, algae and seagrass), a constrained nonlinear spectra unmixing model is utilized to classify the benthic
substrate as a function of the fractional contribution from each
endmember. The final step is to utilize field observations to assess the
accuracy of the resulting image products.
MOTIVATION
Remote sensing is increasingly being employed as a significant
component in the evaluation and management of coral ecosystems.
Advantages of this technology include both the qualitative benefits
derived from a visual overview, and more importantly, the
quantitative abilities for systematic assessment and monitoring.
Advancements in instrument capabilities and analysis methods,
particularly with respect to hyperspectral remote sensing, are
continuing to expand the accuracy and level of effectiveness of the
resulting data products. Not only do hyperspectral instruments offer
the spatial and temporal capabilities of traditional remote sensing,
but also the spectral detail necessary to extract multiple layers of
information from the optically complex environment associated with
coral reefs and other shallow coastal subsurface environments.
(a)
Water Depth
0 m
10
Work for this project will be performed at the University of
Puerto Rico at Mayaguez (UPRM). It will leverage the expertise and
available resources of the Center for Subsurface Sensing and
Imaging Systems (CenSSIS). The CenSSIS mission is “to
revolutionize the existing technology for detecting and imaging
biomedical and environmental-civil objects or conditions that are
underground, underwater, or embedded in the human body.”
CenSSIS employs a unified, multidisciplinary approach to explore
the limits of subsurface sensing and imaging using diverse expertise
in wave physics, sensor engineering, image processing, inverse
scattering and software development. In addition to UPRM, other
core academic partners include Northeastern University (lead
institution), Boston University and Rensselaer Polytechnic Institute.
The CenSSIS research group at UPRM is located in the
Laboratory for Applied Remote Sensing and Image Processing in
the UPRM Research & Development Center, with participation of
faculty and students from the Colleges of Engineering and Arts and
Sciences. UPRM is the lead institution in the research efforts for the
development of algorithms to extract subsurface information using
imaging spectroscopy, also called hyperspectral imaging, in
translucent media such as coastal environments and biomedical
applications. UPRM also leads the SeaBED testbed effort in
collaboration with Woods Hole Oceanographic Institution to develop
an algorithm validation platform that will lead to important
contributions in the use of satellite optical sensors and underwater
vehicles for the monitoring of shallow and deep coral reef systems.
Figure 4. Hyperspectral image processing scheme.
METHODS AND APROACH
20
(b)
(c)
AVIRIS: Hyperspectral
2004 Puerto Rico
RGB (650 nm, 450 nm)
Study Area: Year-1
Efforts for image analysis in the initial phase of this project will be
performed in the coastal area around La Parguera in southwestern
Puerto Rico. An example of the AVIRIS imagery for this area is
illustrated in Figure 2. Note that the imagery is 99% cloud-free and
does not exhibit any significant effects from sunglint (specular
reflection from the water surface). Therefore, this imagery represents
an outstanding basis for application of subsurface image processing
algorithms. Figure 2 also provides an example image of Enrique Reef
acquired using the IKONOS satellite. Although the IKONOS
instrument collects multispectral data (low spectral resolution), its
spatial resolution (4m pixels) is used here to demonstrate the level of
detail possible (3-4m pixels) from the proposed data collection using
a hyperspectral instrument from either HyVista Corporation or
Spectra Vista Corporation. Such imagery would have similar spectral
resolution to the AVIRIS data, and thus be appropriate for the
proposed image analysis techniques, but it also provides the added
benefit of higher spatial resolution.
Figure 1. Example of (a) proposed image analysis product, (b) existing habitat
map 2004, and (c) hyperspectral imagery from a previous study in Enrique
Reef, Puerto Rico.
Puerto Rico
We propose to utilize recent advances in the field of hyperspectral
remote sensing image analysis (Goodman 2004; Goodman et al. 2005
in review) to augment the existing NOAA habitat maps in selected
areas of Puerto Rico and provide an additional, more detailed, level of
habitat information previously unavailable (see Fig. 1a for an example
of proposed map product). Initial analysis will be performed using
imagery from NASA’s Airborne Visible Infrared Imaging Spectrometer
(AVIRIS) acquired in August 2004 (spatial resolution of 20m pixels) and
a second more detailed level of analysis will be achieved through the
collection of higher spatial imagery (3-4m pixels depending on the
instrument selected, either through HyVista Corp. or SpectraVista
Corp.).
Data produced for each pixel in an image includes water depth,
water optical properties, and a quantitative assessment of the percent
habitat cover of coral, algae, seagrass and sand. Efforts in the first year
of the project will center on the area around La Parguera in
Southwestern Puerto Rico, while the focus of the second year is open
to CCRI input for defining specific campaign areas (note that the
AVIRIS data is already available for much of coastal Puerto Rico). The
resulting maps will be geo-registered to the existing NOAA maps and
made available as both jpegs and raster data layers. Specific
deliverables for the proposed sections of coastal imagery in each year
will include (note that indicated depth ranges are for clear water and will
be reduced in turbid conditions).
DELIVERABLE PRODUCTS
As mentioned in the introduction of this proposal, specific
deliverables for the proposed sections of coastal imagery in each
year will include (note again that the depth range provided is for clear
water and will be reduced in more turbid conditions):
•• Map of the per-pixel benthic distribution of coral, algae, seagrass
and sand in water depths down to 5-10m;
•• Map of per-pixel optical water properties in water depths down to
20-30m;
•• Per-pixel bathymetric estimates for water depths down to 20-30m;
•• Spatially coincident color jpeg images of the coastal areas.
These raster mapping products will be delivered in an ENVI
compatible format (for use in other remote sensing analysis) as well
as in a jpeg format for use in GIS mapping software (e.g., ArcView).
EXISTING MAPS AND PROPOSED PRODUCTS
Until recently, existing maps of coastal resources were generally
limited both in the type of information included and the level of detail
provided. Fortunately, funding was allocated to NOAA’s Biogeography
Program to address this shortcoming, particularly in and around coral
reef ecosystems, and to produce a series of maps describing the
dominant benthic habitats (e.g., Coyne et al. 2003; Kendall et al. 2001).
The maps were produced using a hierarchical classification scheme
based primarily on manual interpretation of aerial photographs. Results
of this effort represent a significant improvement over previously
available large-scale map resources. However, because of the project’s
immense scope, the maps still have limitations with respect to smallscale resource assessment. For instance, individual reefs are typically
described according to general categories (i.e., the patch reefs in Fig.
1b) and spatially explicit quantitative habitat information is limited.
Figure 5. Multi-level array of optical measurements.
LITERATURE CITED
Figure 2. Proposed study area for Year-1 of project, illustrating remote sensing
imagery already available for analysis: AVIRIS data of the La Parguera region
(bottom); and IKONOS data of Enrique Reef (upper right).
Study Area: Year-2
Efforts in Year-2 of the project will focus on an area to be
determined with direct input from CCRI. Available areas already
covered by AVIRIS in August 2004 are depicted in Figure 3. It should
be noted, however, that portions of some of these areas are
obscured by clouds and thus not available for analysis. Nevertheless,
like the imagery illustrated in Figure 3, much of the AVIRIS data is
cloud-free and of excellent quality for application of the image
analysis techniques. Additionally, as with the study area in Year-1,
the proposed higher spatial resolution imagery can be collected at
any number of locations throughout Puerto Rico.
Coyne, M.S., Battista, T.A., Anderson, M., Waddell, J., Smith, W., Jokiel, P.,
Kendall, M.S. and Monaco, M.E. (2003). Benthic Habitats of the Main Hawaiian
Islands. (CD-ROM), National Oceanic and Atmospheric Administration, Silver
Spring, Maryland.
Goodman, J.A. (2005). Hyperspectral remote sensing of shallow coral reef
ecosystems using AVIRIS and HYPERION. 8th International Conference on
Remote Sensing for Marine and Coastal Environments, Halifax, Nova Scotia,
Canada, 17-19 May.
Goodman, J.A. (2004). Hyperspectral remote sensing of coral reefs: deriving
bathymetry, aquatic optical properties and a benthic spectral unmixing
classification using AVIRIS data in the Hawaiian Islands. PhD Dissertation,
University of California, Davis.
Goodman, J.A., Lee, Z.P. and Ustin, S.L. (2005, in review). Application of a
semi-analytical model to derive bottom depth and water properties from AVIRIS
data in the Hawaiian Islands. Submitted to Remote Sensing of Environment.
Goodman, J.A. and Ustin, S.L. (2003). Airborne hyperspectral analysis of coral
ecosystems in the Hawaiian Islands. In Proceedings 30th International
Symposium on Remote Sensing of Environment.
Kendall, M.S., Monaco, M.E., Buja, K.R., Christensen, J.D., Kruer, C.R.,
Finkbeiner, M. and Warner, R.A. (2001). Benthic Habitats of Puerto Rico and the
U.S. Virgin Islands. (CD-ROM), National Oceanic and Atmospheric
Administration, Silver Spring, Maryland.
Lee, Z.P., Carder, K., Mobley, C.D., Steward, R., and Patch, J. (1998).
Hyperspectral remote sensing for shallow waters. 1. a semi-analytical model.
Applied Optics, 37, 6329-6338.
Lee, Z.P., Carder, K., Mobley, C.D., Steward, R., and Patch, J. (1999).
Hyperspectral remote sensing for shallow waters: 2. deriving bottom depths and
water properties by optimization. Applied Optics, 38, 3831-3843.
Lee, Z.P., Carder, K., Peacock, T.G., Davis, C.O., and Mueller, J.L. (1996).
Method to derive ocean absorption coefficients from remote-sensing reflectance.
Applied Optics, 35, 453-462.
SPONSOR
Figure 3. Areas covered by the AVIRIS mission in August 2004.