![[logo] US EPA](http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Imaging Spectroscopy for Detecting Fugitive Environmental Contaminants
RESEARCH SUMMARY
Investigators: E. Terrence Slonecker and David J. Williams
U.S. Environmental Protection Agency, Office of Research and Development, Reston, Virginia 20192.
The intentional or accidental release of contaminants into the environment is an inevitable consequence of anthropogenic activity. Industrial, mining and even natural processes can cause the release of substances into the air, land and water that may be harmful to environmental quality. Reducing the risk to human and ecological ecosystems from waste-related problems is one of the highest priority areas for research in the EPA (EPA 1999). Hazardous and other wastes are regularly released into the environment as a result of normal industrial processes and accidental spills and releases. Under the Resource Conservation and Recovery Act (RCRA), a total of 400,000 facilities have reported generating RCRA-regulated hazardous wastes in the United States, and generate over 200 million tons of hazardous waste each year (OSW 1993a, OSW 1993b). Approximately 2000 sites have released hazardous wastes into the environment at levels of concern that require corrective action (OSW 1993c).
Releases from abandoned hazardous waste facilities are significant and are regulated under the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA), also know as ‘Superfund’. Over 40,000 sites have been identified as potential candidates for Superfund remedial action and approximately 1300 sites are listed on the Superfund National Priorities List (NPL) as the highest priority for cleanup action. Further, 4300 Superfund emergency actions have been initiated to eliminate immediate risks to human health from abandoned hazardous waste facilities. These emergency actions have at least temporarily displaced over 15,000 residents and provided alternative drinking water supplies to over 350,000 people (OERR 1996a, OERR 1996b).
Additionally, significant health hazards result from the release of other substances such as oil and other petroleum products. Accidental releases of oil and related products into the environment threaten public health and safety through contamination of drinking water, fire and explosions, diminished air and water quality, destruction of recreational areas, and compromised agriculture. Annually, over 20,000 oils spills are reported each year releasing between 10 and 25 million gallons into the environment (OERR 1996c).
Imaging Spectroscopy also known as hyper-spectral remote sensing (HRS), is an imaging technique that is capable of identifying materials and objects in the air, land and water on the basis of the unique spectral reflectance patterns that result from the interaction of solar energy with the molecular structure of the material. It is a relatively new scientific capability within the field of remote sensing that is currently making the transition from purely research and development into commercial applications. Improvements in sensor design and other technical advances have permitted the development of hand-held, airborne and satellite spectral sensing instruments that can measure the near-continuous response of reflected and emitted electromagnetic energy. In many cases, the measurement of this energy into recorded spectra, gives unique information related to the structural makeup of the material or substance.
Imaging spectroscopy is an emerging research and development area in the geographic discipline of remote sensing. As a result of technical advances in sensors and electronic design, HRS instruments are able to collect reflected and emitted electromagnetic energy in hundreds of discrete spectral channels. This allows a near-continuous recording of hundreds of very narrow spectral bands across the reflected and emitted parts of the electromagnetic spectrum. Classic multi-spectral remote sensing instruments, such as the Landsat Thematic Mapper, collect data in a few selectively-placed, broad portions of spectrum. HRS instruments collect a discrete image for hundreds of narrow (~10 nm) bands resulting in an image with hundreds of separate bands. A further development, now in the R&D stage, is Ultra-spectral Remote Sensing (URS), which, in general, collects data in thousands of very narrow (~1-2 nm) bands. Figure 1 shows basic bandwidth and differences between multi-, hyper-and ultra- spectral remote sensing instruments.
Most materials on the earth's surface contain characteristic absorption and reflectance features, based on their molecular structure, that can be identified through the unique spectral trace, or spectra, of that particular substance. This unique interaction between matter and energy is recorded in each pixel of a hyper-spectral image. By comparing the plot of spectral responses through the hundreds of spectral channels with a library of known spectral reflectance, the imaging scientist can now identify specific substances such as minerals, chlorophyll, dissolved organics, atmospheric constituents, and specific environmental contaminants. Other applications include water and air quality analysis, vegetative species identification, and fate and transport modeling.
Several airborne hyperspectral instruments, such as the AVIRIS (NASA) and the TRWIS-3 (TRW), and several others, are already operational, and many more are planned for the near future, including satellite-based systems. Table 1 shows examples of current and future hyperspectral systems. The emergence of hyperspectral imaging moves general remote sensing applications from the area of basic landscape classification into the realm of full spectral quantification and analysis. The same type of spectroscopy applications, that have been utilized for decades by chemists and astronomers, are now possible through overhead imaging applications.
Table 1: Example Hyperspectral Instruments
| Sensor | Name | Manufaturer | Bands | Spectral Range (nm) |
|---|---|---|---|---|
| ASAS | Advanced Solid-State Array Spectrometer | NASA (Goddard) | 62 | 400-1200 |
| AIS-1 | Airborne Imaging System 1 | NASA (JPL) | 128 | 800-1600/1200-2400 |
| AISA | Airborne Imaging Spectrometer for Applications | SPECIM, Ltd | 286 | 400-1250 |
| CIS | Chinese Imaging Spectrometer | Shanghai Institute of Techical Physics | 91 | 00 |
| Dais 7915 | Digital Airborne Imaging Spectrometer | GER Corp | 79 | 400-12000 |
| IRIS | Infrared Imaging Spectroradiometer | ERIM | 256 | 2000-15000 |
| MIVIS | Mutispectral Infrared And Visible Imaging Spectrometer | Daedalus | 102 | 433-12700 |
| VIMS-V | Visible Infrared Mapping Spectrometer | ASI | 512 | 300-1050 |
| AVIRIS | Advanced Visible and Infrared Imaging Spectrometer | NASA (JPL) | 224 | 400-2500 |
| HYDICE | Hyperspectral Digital Imagery Collection Experiment | NRL | 210 | 400-2500 |
| HYMAP | Airborne Hyperspectral Scanner | Integrated Spectronics | 200 | 400-12000 |
| MAS | MODIS AIRBORNE SIMULATOR | Daedalus | 50 | 530-14500 |
| MODIS | Moderate Resolution Infrared Spectrometer | NASA | 36 | 400-14400 |
| TRWIS III | TRW Imaging Spectrometer | TRW | 384 | 300-2500 |
| NEMO | Navy Earth Map | U.S. Navy | 210 | 400-2500 |
| Warfighter1 | Observer Warfighter | U.S. Air Force | 280 | 400-5000 |
While much work has been accomplished on the spectral identification of relatively pure substances (chemicals and contaminants) in the laboratory, this type of spectral data is rarely useful in a complex matrix of natural resources such as soil, wood, chlorophyll, metals, and water, any combination of which might be represented in the classic ‘mixed pixel’ remote sensing problem. In general, any given remote sensing scene, photograph or image is often likely to be dominated by vegetation, even in urban areas. Chlorophyll and other vegetative pigments are highly reflective to the visible and near-infrared parts of the solar reflective spectrum. Further, it is likely that vegetation is often in some way affected by a contaminant released into the environment. Through uptake, sorption or other processes, many contaminants can affect vegetation reflectance. The ability to detect and map areas of contaminant release based on changes in spectral reflectance of vegetation would be a significant advance in environmental monitoring.
Reflectance Spectroscopy
Hyperspectral imaging and analysis techniques are related to a special class of spectroscopic analysis called reflectance spectroscopy. Reflectance spectroscopy is the study of energy as a function of wavelength that has been reflected from or scattered by a solid, liquid or gas (Clark 1995). It differs from other spectroscopic techniques in two primary ways. First, energy is reflected or scattered back to the sensing device. Most other spectroscopic techniques pass some form of energy through a substance suspended in a gaseous or liquid medium, or measure some secondary response to energy. The differences in the physical processes of reflectance versus transmittance are important. Second, reflectance spectroscopy, especially as applied to overhead remote sensing, is limited to the .4 - 2.5 µm range of the electromagnetic spectrum where solar radiation is passed through atmosphere. Laboratory techniques often have the advantage of much wider portions of the EM spectrum.
However, many substances still have unique spectral reflectance signatures in the .4 - 2.5 µm range caused mainly by absorption of photons by specific electronic, vibrational or scattering processes that often occur with common earth materials in this part of the spectrum. The reflectance characteristics in the range of .4 - 1.0 µm are typically influenced by the presence of transition metals, such as iron, and are extremely diagnostic for minerals. Electronic transitions which cause absorption in the crystal field of minerals also cause unique absorptions patterns in the area of .9 µm. The .4 - 1.0 µm region is also extremely diagnostic for vegetation. Chlorophyll absorption features occur around .48 µm and .68 µm and are a function of electronic transitions in the carotenoid pigments associated with the photosynthetic process (Vane and Goetz 1988). Figure 2 shows the general concept of imaging spectroscopy and example spectra.
Vegetation reflectance also has diagnostic characteristics in the EM spectrum past 1.0 µm. The classic bimodal distribution of green vegetation is caused by reflectance peaks generally centered at .8 and 1.3 um and is called the infrared plateau. High reflectance in this area is related to structure of leaf tissue and contains diagnostic information based on the cellular arrangement of the tissue as well as in the hydration state of the leaf. The steep rise in the spectral reflectance curve for vegetation at .8 µm is called the red edge of the chlorophyl band and is one of the fundamental characteristics of vegetation analysis by imaging spectroscopy methods. Several researches have documented diagnostic ‘blue shifts’ in the location of the red edge, (the movement of the location of red edge towards lower wavelength values around .7 µm), as a result of chemical stress (Chang and Collins 1983, Rock et al.1988a, Vane and Goetz 1988). Several researchers have demonstrated that spectral reflectance can detect stress in vegetation due to morphological and/or pigmentation changes in the leaf tissue.
Technical Approach
The general approach to this research will be relatively simple. While working within the EPA/EPIC mission to provide remote sensing support to Regional and Program Offices for waste site operations, the research will identify sites of preliminary interest based on their contamination profile, accessibility, and relative environmental risk. Working with EPA On-Scene Coordinators (OSCs) EPIC personnel will travel to individual sites to record field spectra with EPIC’s field Spectroradiometer. Spectra will be collected at areas where contamination is known to exist and where recent in situ samples have been collected and verified by laboratory analysis methods. Spectra will also be collected in nearby areas of the same vegetation/soil profile where no specific contamination exists. Comparison and preliminary analysis of these spectral profiles will determine the feasibility of spectral identification of the contaminant profile and the value in acquiring hyperspectral remote sensing data of the area. If preliminary results show that spectral identification and analysis of the contamination profile is feasible, the collection of hyperspectral data will be attempted through a variety of intergovernmental and/or commercial mechanisms. Analysis of hyperspectral data will be coordinated with OSCs and site-specific operations. Although the spectral analysis will be driven by site-specific parameters, the initial focus of this research will concentrate on vegetation effects from fugitive contaminants and larger areal units, such as watersheds, which will be imaged and analyzed to determine the extent and distribution of contaminants. In this way, imaging spectroscopy will be able to contribute to the assessment of relative risk to human and ecological health and will provide data important to the effective management of that risk.
The overall objective of this research is to determine the capability of imaging spectroscopy in the solar reflected region to identify and monitor fugitive contaminants in the environment. Under the Resource Conservation and Recovery Act (RCRA), the Clean Water Act (CWA), and the Comprehensive Environmental Response and Compensation and Liability Act (CERCLA, aka ‘Superfund’), EPA monitors known, suspected or potential chemical releases into the environment at industrial facilities, waste sites and abandoned mining areas. Current techniques for identifying, quantifying and mapping the extent of contamination from a wide range of chemical releases, involves time-consuming and expensive in situ sampling techniques and laboratory analyses. With the development of hyperspectral remote sensing platforms, the ability to directly or indirectly identify chemical contaminants in the environment may be possible through aircraft and satellite remote sensing systems. Because the landscape in many parts of world is dominated by vegetation, this research effort will concentrate on the ability of imaging spectroscopy to identify contamination effects through changes or anomalies in vegetation and chlorophyll reflectance patterns.
This research will utilize field and overhead remote spectral data collection at a wide variety of EPA sites where known contamination profiles exist to: 1) identify those chemicals and contamination scenarios that can be identified through solar reflected hyperspectral remote sensing techniques, 2) develop a library of those unique spectral signatures, 3) identify secondary, or indirect spectral measures of the effects of fugitive contaminants, such as vegetation stress, or soil reflectance anomalies. Initially, this research effort will focus on vegetation signatures that are the result of chemical effects.
NASA: http://aviris.jpl.nasa.gov/
USGS: http://speclab.cr.usgs.gov/
References:
Chang, S.H. and W. Collins. 1983. Confirmation of the Airborne Biogeophysical Mineral Regional Water Quality Exploration Technique Using Laboratory Methods. Economic Geology 78:723-736.
Clark, R.N. 1995. Reflectance Spectra. AGU Handbook of Physical Constraints 178-188. requirements. U.S.
Green R.O., M.L.Eastwood, C.M.. Sarture, T.G. Chrien, M. Aronsson, B.J. Chippendale, J.A. (EPIC) by Lockheed Faust, B.E. Pavri, C.J. Chovit, M. Solis, M.R. Olah and O. Williams. 1998. Imaging Spectroscopy and the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). Remote Sensing of the Environment 65(3): 227-248.
Multispectral Users Guide. 1995. Spectral Information Technology Applications Center, Faifax, Virginia. operations.
Office of Emergency and Remedial Response (OERR). 1996a. Comprehensive Environmental Environmental Response Compensation and Liability Information System (CERCLIS). Washington, D.C: U.S. Environmental Protection Agency.
Office of Emergency and Remedial Response (OERR). 1996b. National Environmental Response Compensation and Indicators Data Compilation. Washington D.C.: U.S. Environmental Protection Agency. Office of Emergency and Remedial Response (OERR). 1996c. Emergency Response Notification System (ERNS) Database. Washington D.C: U.S. Environmental Protection Agency.
Office of Solid Waste (OSW). (1993a). RCRA Data Summary. Washington D.C. U.S. Environmental Protection Agency.
Office of Solid Waste (OSW). (1993b). The Biennial RCRA Hazardous Waste Report. Washington D.C. U.S. Environmental Protection Agency.
Office of Solid Waste (OSW). (1993c). Draft Regulatory Impact Analysis for Final Rulemaking on Corrective Action for Solid Waste Management Units. Washington D.C. U.S. Environmental Protection Agency.
Rock, B.N., T. Hoshizake, and J.R. Miller. 1988a. Comparison of In Situ and Airborne Spectral Measurments of the Blue Shift with Forest Decline. Remote Sensing of the Environment 24(1):
U.S. Environmental Protection Agency (USEPA). 1999. Waste Research Strategy. Office of Research and Development. EPA/600/R-98/154.
Vane G. and A. Goetz. 1988. Terrestrial Imaging Spectroscopy. Remote Sensing of the Development. Environment 24(1): 1-29.