Mission | People | Organization | Research | Education | Project | Data | Tools | Links
Oil Spill Detection on Sea Water Using AVIRIS Data
Submerged Aquatic Vegetation Remote Sensing
Oil Spill Detection on Sea Water Using AVIRIS Data
Project Leader: Dr. M. Kafatos (L) and Ms. Foudan Salem, GMUBackground:
Our applied research will assess the detection of oil spills on freshwater and seawater using hyperspectral image analysis, AVIRIS data, and information available for spills in other areas, such as in Santa Barbara, California. In earlier research, the Patuxent River in Maryland, was selected as a prototype for oil spill detection in the relatively calm waters of Chesapeake Bay.
Technical Approach:
Now we plan to use AVIRIS data to study ocean spills as evidenced in the Santa Barbara coastline. The reason for selecting this western coastal zone is that features on the surface make the study more realistic and useful to other costal area applications in Virginia. The detection of an oil slick affected by heavy seas and breaking waves will be different than oil on flowing river waters. The Santa Barbara coastal area will be used for creating oil mitigation simulations and scenarios.AVIRIS DATA SETS
The AVIRIS (Airborne Visible/Infrared Imaging Spectrometer) is a unique optical sensor that delivers calibrated images of the spectral radiance in 224 contiguous spectral channels (bands) with wavelengths from 400 to 2500 nanometers. The main objective of using the AVIRIS data is to identify, measure, and monitor constituents of oil spills based on molecular absorption and particle scattering signatures.The measurements can be converted to ground reflectance data which can then be used for quantitative characterization of surface features. Compared with the AISA sensor used before, the classification processes in the first part of this study was done using 25 bands, which were not useful in spectral matching with different spectral libraries, because of it’s wavelength range from 400 to 800 nanometers. The AVIRIS sensor will allow more specific spectral analysis with wavelengths from 400 to 2500 nanometers and 224 bands.
The AVIRIS scene of the Santa Barbara coast will be obtained from Jet Propulsion Laboratory, Caltech, (JPL/ NASA).
METHODOLOGY
From our own experience and the literature on the spectral properties of various types of oil, we knew that the HSI would see sheens and oil slicks, but we could find no reports in the literature of the spectral properties of oil dispersed with chemical dispersants. If a distinctive spectral signature could be found for the dispersant itself and the dispersed oil, then algorithms could be developed which would optimize the identification and tracking of both. Also, we can obtain operational experience and assess the utility of airborne hyperspectral remote sensing with AVIRIS over an oil spill in preparation for other disasters elsewhere.This activity includes the following tasks
- Compare ”pure” oil spectra from the data sets (AISA and AVIRIS)
- Collect oil spectral information for different oil types and dispersed oil on the sea water
- Determine the optimal spectral bands for detection of this type of oil and dispersant
- Perform accuracy assessments for multi spectral and hyperspectral data sets
- Compare other polluted water spectra (fresh water versus sea water)
- Run current Spectral Angle Mapper (SAM) classifications for the Santa Barbara scene
- Apply the Mixture Tuned Matched Filtering (MTMF) method to AVIRIS data and compare the results with existing methods for spectral analysis
- Use hyperspectral data over land in an attempt to map the effects of soiling of coastal zones and sandy beaches and providing baseline data for long term monitoring of the impact of oil pollutants.
Milestones:
3rd Quarter Assessment of oil detection in seawater analyses
Submerged Aquatic Vegetation Remote Sensing
Project Leader: Dr. Richard Gomez, GMU
BackgroundIt is well established that submerged aquatic vegetation (SAV) is an important part of the food chain in Chesapeake Bay, providing shelter and nursery areas for shellfish and finfish and food for a variety of waterfowl, fish, and invertebrates. There has been a series of massive declines observed in SAV covered areas in the Chesapeake Bay, which are of serious concern to scientists and resource managers even with an increase in restoration efforts, as well as increased limited area coverage that has occurred in recent years. Remote sensing of the Chesapeake Bay SAV ecosystem could be very valuable in establishing trends in areal coverage and could potentially be used to identify the species of SAV growing in particular areas. The tidal Potomac River and Estuary provide an ideal ecosystem for applied research leading to the restoration of SAV baywide.
In support of this project, GMU is continuing the development of a spectral library database containing selected ground-based and airborne sensor SAV spectra for use in image processing. The long-term goal of the spectral database is to automate the image processing of hyperspectral imagery for potential real-time material identification and mapping of SAV. Further research into the spectral signatures of various SAV species with a focus on bio-chemical differences is contemplated. For this year, the spectral library database for removing potential false positives and mapping SAV by species will be made more effective and computationally efficient. The Spectral Library SAV Information System will also serve as a tutorial capability to train people in the use of this hyperspectral technology for SAV identification and mapping.
George Mason University (GMU) was part of the team that investigated the use of airborne hyperspectral remote sensing imagery for automated mapping of submersed aquatic vegetation in the tidal Potomac River for near to real-time resource assessment and monitoring [1]. In that investigation, field surveys at the tidal Potomac River at the mouth of Nanjemoy Creek at Blossom Point, in southern Charles County Maryland, determined SAV presence, species, and distribution. Airborne hyperspectral imagery, using the HyMap sensor, together with in-situ spectral reflectance measurements using a field spectrometer, were obtained for the pilot sites in spring and early fall of 2000. A spectral library database, which currently resides at GMU containing selected ground-based and airborne sensor spectra, is being developed for use in hyperspectral image processing of SAV. The goal of the spectral database is to automate the image processing of hyperspectral imagery for potential real-time material identification and mapping of SAV. Field based spectra were compared to the airborne imagery using the database to identify and map two species of SAV (Myriophyllum spicatum and Vallisneria mericana). Of the seventeen species of submerged aquatic vegetation that are commonly found in the Chesapeake Bay and its tributaries, these two are among the most abundant. Overall accuracy of the vegetation maps derived from hyperspectral imagery was determined by comparison to a product that combined aerial photography and field based sampling at the end of the SAV growing season. Map accuracy was high and had very low false positive detections. The algorithms and databases developed in this study will be useful with the investigation under this proposed VAccess project, the main objective, which is to prepare the remote sensing and terrestrial ecology communities for exploiting current and forthcoming space-based hyperspectral remote sensing systems.
METHODOLOGY
Imaging spectroscopy, also called hyperspectral imaging (HSI), was originally developed by astronomers to obtain geochemical data from inaccessible planetary surfaces within the solar system [2]. The primary application of this technology has now shifted to the observation of the Earth with airborne and spaceborne hyperspectral remote sensors. AVIRIS and Hyperion are good examples, respectively. The basis of traditional remote sensing theory involves the interaction of electromagnetic radiation with materials on a “macroscopic” level, including reflection, refraction, diffraction, absorption, and scattering effects [2]. The emerging enabling technology known as hyperspectral imagery, places emphasis on the “microscale” monitoring interactions of electromagnetic energy within molecules, crystal lattices, and cell structures [2,3]. This new perspective requires knowledge of quantum mechanics and a “particulate” view of electromagnetic energy [2]. Ecological research, in particular, is likely to benefit rapidly from the increased spectral resolution that hyperspectral sensors can provide. Most natural Earth surface and water materials have diagnostic absorption features in the range of current hyperspectral sensors, i.e., 400 nm – 2500 nm range of the reflected spectrum. With the diverse chemistry of the materials involved, spectral signatures can be complex and sometimes difficult to interpret. However, with the increasing knowledge of the spectral signatures involved, this problem will continue to diminish. Roger N. Clark [4] suggests that the previous disadvantage, for imaging spectroscopy, is turning into a huge advantage, allowing us to probe ever more detail about the chemistry of our natural environment.
The ecological activities of the project will be conducted in phases, and most of the fieldwork will be conducted on the Chesapeake Bay. The principal submerged aquatic vegetation (SAV) species for the Chesapeake Bay are Eurasian watermilfoil (Myriophyllum spicatum) and wildcelery (Vallisneria mericana). Keep in mind that we will use archived data during the first year of this ecology research part of the project (Phase 1). The first phase of the field research focuses on assuring the calibration of the HSI methodology for the two dominant SAV species. At three times during the growing season (late spring, mid summer and early fall), in situ measurements will be made with an Analytical Spectral Devices, Inc (ASD) field portable spectroradiometer provided by VAccess. (See Figure 9). This Field Spectrometer has a 350 to 1050 nanometers spectral range that will be fully used. The in situ measurements over several windows of time will allow for the calibration to identify (i) age-dependent changes in spectral properties and (ii) site-specific affects on spectral properties. The former, which has been reported in an earlier study by one of the principal investigators [1], is attributed to epiphytic colonies of microbes and appears to be minor. The second class of factors is related to covariates associated with the water column (e.g., turbidity, depth). Harvesting of plant samples during each trip and re-testing under controlled laboratory conditions in the Physiological Ecology Laboratory on the George Mason University campus will be used to further investigate these in situ calibration records. The resulting data (in situ and laboratory) will be used to populate the team’s HSI spectral library database.
Technical ApproachThe research will be divided into two phases, a demonstration phase and a validation phase. The use of a demonstration phase will allow for research to begin immediately through the use of archived imagery. A demonstration phase will also allow us to better define all the required steps in the validation phase, thus significantly increasing the chances of a successful research effort. Validation of the demonstration phase results will be performed in phase 2, using new collections that will take place during the demonstration phase with the new imagery being available for the start of the validation phase.
Demonstration phase: In this phase, the ecology application to be addressed will be better defined. We will determine which ecology applications are appropriate for remote sensing technologies. Several applications will be addressed. Data requirements (for both demonstration phase with archived data and validation phase with new imaging collections) will be established. We will identify one site for the demonstration phase using archived data. The 2nd site will be identified for validation phase using new collections (the two sites may possibly be the same). We will collect imagery, ground truth, and other data. We will use HyMap data for this phase and any other data we may find. It is possible that Hyperion data might be available.
Validation Phase: During the validation phase, GPS will be used when collecting spectral truth (we will also collect GPS points at road intersections, building corners, parking lots, etc). Data collected over the pilot sites during the demonstration phase will be used for the validation of the specialized tools for analysis of SAV data. The data collection effort will continue through the second and third years.
MilestonesThe main goal of the proposed project will be to perform analysis of hyperspectral data to better understand the Chesapeake Bay submerged aquatic vegetation (SAV) ecosystem variability and responsiveness to global change.
A spectral library populated with spectral signatures for all SAV species found at the Chesapeake Bay pilot site will be developed and will be made available to the community through the Internet. The data will come from various sources including the laboratory and field measurements done under this project.
References1. D. J. Williams, T. M. O’Brien, N. B. Rybicki, A. V. Lombana, and R. B. Gomez, “Preliminary Investigations of Submerged Aquatic Vegetation Mapping using Hyperspectral Remote Sensing”, to be published in Environmental Monitoring and Assessment, Springer-Vertag, 2002.
2. A.F.H.Goetz (1992), “Imaging Spectrometry for Earth Remote Sensing”, in Imaging
Spectroscopy: Fundamentals and Prospective Applications, F. Toselliand J. Bodechtel, Eds., 1-19, Brussels and Luxembourg.
3. K. Nassau (1983), The Physics and Chemistry of Color – The Fifteen Causes of Color, Wiley Interscience.
4. R.N. Clark (1999), “Spectroscopy and Principles of Spectroscopy Manual of Remote Sensing, A. Rencz, Editor, John Wiley and Sons, Inc.
VAccess-MAGIC © 2001-2002. All rights reserved
