|
Environmental Case Study: Coeur D'Alene River Basin Project In October of 1998, a hyperspectral airborne survey of selected areas in the Coeur díAlene River Basin was undertaken for the U.S. Environmental Protection Agency. The objective of the study was to show "proof of concept" for the use of hyperspectral imagery to provide mineral maps of the mine wastes and tailings in selected areas of the Coeur díAlene project area. Figure 1 shows a general map of the Coeur d'Alene River Basin, Idaho study area with flight lines overlain on the geography. |
||
The pilot study concentrated on the Lateral Lakes area, from Rose Lake to the Old Mission State Park, and in the Canyons ñ parts of Canyon Creek Canyon and Ninemile Creek Canyon, both located north and northeast from Wallace, Idaho. This overview will concentrate on those areas and discuss the preliminary results of the study. Two hyperspectral sensors were used to fly the area. These are the CASI (Compact Airborne Spectral Imager) for the visible and near-infrared wavelength regions and the SFSI (SWIR Full Spectrum Imager) for the short-wave infrared region. These instruments were flown by Borstad Associates, Sidney, B.C. The NASA, AVIRIS (Airborne Visible InfraRed Imaging Spectrometer) sensor had also flown most of the area in 1993. The sensor itself is shown below at the AVIRIS facility in the Jet Propulsion Laboratory, Pasadena, California. It occupies the second seat in the ER-2 aircraft. |
||
 
As a comparison the next picture shows the next generation hyperspectral SWIR range sensor, the SFSI, installed in a survey aircraft, ready for flight. |
||
|
The companion sensor to the SFSI, CASI, is used for the Visible and Near InfraRed regions, and is smaller then the SFSI. SFSI and CASI are flown at approximately 10-12,000 ft to attain a ground pixel size of 4 meters. SFSI has 240 bands with a 10 nm spectral resolution. CASI has 15 selectable bands with varying resolutions from 2-4 nm as a function of wavelength region. AVIRIS has approximately 220 bands, flies at approximately 65,000 ft and has a ground pixel size of 17-20 m and a spectral resolution of from 12-15 nm. AVIRIS data are relatively easy and fast to process. They were used to determine the general distribution of the inorganic materials in the target areas. AVIRIS data are used in this report to show the results of the processing and discuss the project. At this time, July of 1999, the Environmental Protection Agency has not publicly released the mineral maps which were constructed from the CASI and SFSI image data. A true-color CASI image is shown later in this paper as an example of the resolution of the two hyperspectral sensors. However, it shows only the terrain, much like a digital photograph, and does not contain any mineral distribution information. AVIRIS is an excellent sensor. It has been the pioneer for hyperspectral data. However, it does not have the spatial and, possibly, the spectral detail required for a study of this detail. Some of the mine dumps are less then 20 meters in size or close to 20 meters, and therefore reliable discrimination of the contents is difficult. However, AVIRIS can be used to illustrate the greater detail provided by hyperspectral sensors as compared to the Landsat Thematic Mapper satellite sensor. It also demonstrates the type of ground coverage available from hyperspectral sensors as compared to field mapping and ground sample collection. This type of a survey requires initial ground truth to assist the sensors in the discrimination of the ground materials. The Coeur díAlene River Basin area is complicated by extensive vegetation cover and similarity of tailings with country rock constituents. This makes it difficult to discriminate the tailings. The sources of the Canyon mineralized materials were easier to determine. Ground sample sites were chosen throughout the study area based on previous information of what they contained. These included concentrations of tailings along the river, rock outcrops of local geologic formations, numerous dumps and tailings associated with mines in the two canyons and profiles through the streambed in Canyon Creek. For each location, Visible, NIR, and SWIR measurements were collected along with X-ray Fluorescence (XRF) data. Extensive ground information already exists through the efforts of several EPA contractors including CH2M-Hill and URS Greiner. Some of these data were integrated with the infrared and visible spectroscopy, chemistry from XRF, and X-ray diffraction data collected from the specific calibration sample sites. Those sample sites, for the Canyons and the Lateral Lakes areas, are shown in Figures 2A and 2B. |
||
 
|
||
 |
||
|
The first AVIRIS image presented here shows the Lateral Lakes Study Area from Rose Lake to Cataldo. This study area was chosen by the client. It was processed as a pseudo-true color image, to display, in blue, the 2200 nm region for the Al-OH clays present in the tailings. The corrections applied for atmospheric affects apply only to the river floodplain. Therefore, blue areas on the hillsides and above the floodplain are from shadows and country rock. Interstate 90 (I-90) was constructed on tailings. The tailings along the river bed are well displayed, except for vegetation interference. The SFSI gives finer detail of the tailings distribution. This has been field checked. |
||
The preceding image provided a general idea of the distribution of the target materials. The objective then was to determine the actual composition of the tailings. The PIMA™ infrared spectrometer was used to ground truth the image data. PIMA™ provided a spectral library of mineral data from the Coeur díAlene study areas which essentially told the image processors what spectral signatures were diagnostic. A portable XRF unit provided chemical information on the metals present in the samples. |
||
The dominant infrared-active minerals in the Coeur díAlene Pilot Study areas are (from top to bottom on the below figure) muscovite [A], illite [B], illite/siderite [C], smectite [D], silica [E], and sphalerite [F]. Most sulfides do not have absorption features in the SWIR. Field work and discussions with U.S. Geological Survey personnel, established that the infrared-active minerals in the tailings, in the Lateral Lakes study area, were dominantly mica/degraded mica/illite and siderite with silica and that much of it is found 10î below the surface. The spectral signature is shown in the following plot of PIMA™ spectra from several localities in the Lateral Lakes region. The picture below is from Pit A, dug into the Dredge Tailings Piles, north of I-90 near the Old Mission State Park and shows a soil profile into the tailings. The initial layers are smectitic and grade through a mixture of tailings and soil to tailings. This is one of the reference sites for the tailings signature spectrum of mica/illite plus siderite. |
||
Using the mica/siderite spectra, processing of the AVIRIS data was started. The first image produced used a low cutoff. Best matches are red then yellow. Blue is background. |
||
A higher cutoff was applied resulting in the Rule image below where red is the best match, then yellow, blue and magenta. |
||
This image shows only the highest concentrations of tailings and does not target mixtures of tailings with uncontaminated soil very well. There are two problems: spatial resolution and vegetation. The SFSI images show more of the tailings as the 4 meter spatial resolution can see between clumps of grass and bushes. The following picture illustrates this. PIC00023 is a picture of the area at the east side of the study area between the freeway and the frontage road, where dark orange brown exposed tailings can be clearly seen. On the AVIRIS image, these tailings are not discriminated; on the SFSI image, they are. This has been field checked. |
||
This further demonstrates that the higher spatial resolution of the SFSI and CASI systems is an advantage over AVIRIS because of the greater impact of vegetation cover (on the tailings) upon discrimination by AVIRIS. |
||
This picture of a sandbar on the river at the east edge of the study area, also demonstrates that vegetation in AVIRIS does mask the tailings. The SFSI, with the 4 meter pixels, has the ability to see the smaller pieces of exposed ground. |
||
|
||
|
When the Crosta processing procedure for discriminating iron and clay was run on the AVIRIS data, the image shown below was produced. The oranges and red indicate the presence of iron. This is a very broad category of processing. It was done on this data set as the original hypothesis linked iron minerals and the tailings. This was substantiated with the PIMA™ data showing the iron carbonate. Additionally, the association is visually striking. The tailings have a distinctive reddish brown, sandy appearance. Please note that the highest concentrations of the orange color, which is roughly equivalent with the tailings, are found in the river bends, marshes close to the river where floods would have flushed tailings from the river bed and banks, in the sand bars within the river, and in the dredge tailings pile near Old Mission State Park. |
||
This analytical technique is complicated by the fact that there is pyrite in the Pritchard Formation, a local country rock, and it oxidizes in the ground water to form iron-oxide coatings on many fracture and bedding plane surfaces as shown below. |
||
The question remained as to what iron minerals were present, aside form the siderite, in the tailings. Subsequent processing for hematite, goethite and jarosite showed an absence of these minerals. X-ray diffraction analyses could identify only very small amounts of these minerals, if any. The assumption was then made that the iron phases present were mostly amorphous iron oxides and hydroxides. Several images were created using visible-range iron information. In this first image below (72_287), the reds indicate high concentrations of iron while yellow and green are mixtures of materials where iron dominates the visible spectrum. This is a useful image and although it does highlight much of the tailings, it also picks up outcrops and other concentrations of iron. |
||
This image has been warped and rectified and can be displayed against a grid. It is a rule image where only one category of Information is displayed with degrees of confidence indicated by the colors from red-yellow-green-blue-black. To fit these images on a vertical page format, they are slightly distorted in scale. The image below is another Rule Image (Fe23orul). It has much tighter processing statistics, has been refined by adding clay mineral spectra to the processing, and more specifically targets the tailings. |
||
 
The following image (W23FeRul) is another Rule Image. It incorporates a more specific tailings signature with the iron. The degrees of confidence are arranged from white-red-yellow-green-blue with open water shown in black. It appears to be a very representative image of the area. |
||
|
||
Figure 2B is reproduced here for reference as it shows the Canyon Creek and Ninemile Canyon study areas. |
||
AVIRIS did not fly exactly the same area as the SFSI/CASI present survey and it flew in 1993; however, the ground locations are reasonably close. The AVIRIS image here ( CanCKiarr) was processed to show the areas of concentration of clays in the canyons. Blue in the image highlights clays seen in the stream beds and around the mine dumps and other disturbed areas. |
||
This is a CASI true-color image of the Canyons study area. Note the forest fire which appears as blue smoke over the northern and eastern parts of the image. The town of Wallace is at the lower right end of the image. The Star Tailings Repository lies along the Canyon Creek drainage at Woodland Park, which is the lower canyon in the image. The ground spatial resolution is 4 m. A larger-scale view of this image will show the individual tamarack trees which turn yellow in the Fall. CASI imagery is very sharp and provides an excellent base for other data. |
||
Image (C_67_ctg) shows the same Canyon areas, as in CanCkiarr, but this time processed to emphasize the materials on the dumps and in the streams that are associated with the mining activity. This is another Rule Image and is graded from red-> yellow->blue-> white for degrees of confidence of fit. The areas of highest concentration of mines show up as red and yellow. |
||
This is an expanded view of Canyon Creek and Ninemile Creek. Areas of mining activity are in red and yellow. It is hypothesized that blue is probably indirectly associated with alteration. This expanded-scale image shows how pixilated the 20-m data becomes with enlargement. However, it still is possible to discriminate many of the mines. |
||
One of the major problems in the Canyons is the precipitation of metals from mine adits draining into the watershed. The picture below shows this occurring at Gem. Although this coating appears to be from acid water, it is not. It contains a neutral silica with amorphous iron oxide phases, mixed with lead, zinc, and manganese. |
||
The following is the AVIRIS iron image for Canyon Creek . It was processed with the assumption that the iron on the rocks is amorphous. It covers from west of Burke to Gem. It demonstrates that AVIRIS can show the high concentrations of iron in the stream beds. The large pixel size and poorer spectral resolution do not allow for fine spatial resolution and discrimination of the iron, however, it does show its general distribution and helps target Regions of Interest (ROI). |
||
This AVIRIS Rule Image shows both canyons and is striking in how it highlights the mines, the roads, the disturbed areas and concentrations of mineralized materials in the stream beds. Degrees of Confidence are red-yellow-green. The rectified version of this image can be used to locate the various dumps and adits with some certainty. X-ray fluorescence data show that the iron coatings on the rocks consist of amorphous iron, silica, zinc, variable lead, and manganese, with selected local concentrations of other elements. The lead actually occurs in higher concentrations in the soil of the stream banks than in the sediments and fines within the stream bed. Zinc, however, is concentrated more in the stream sediment. |
||
|
||
Blue indicates Canyon Creek; maroon indicates NineMile Canyon; yellow is Lateral Lakes |
||
|
||
|
(The lower detection limit for the portable XRF data was 100 ppm. 50 ppm was assumed for the samples which returned values below the detection limit. Maximum value cutoffs were: Fe = 100,000 ppm, Pb = 15,000 ppm, and Zn = 10,000 ppm.) After the minerals are identified, their associations must be established. This was determined from the PIMA™ and XRF data. From the infrared perspective, and what can be detected from the images, the mineralized rock contains mica and illite with sulfides. This rock is degraded in the dumps and fines are created from mechanical weathering. The overall alteration is degraded mica and illite. It occurs as haloes around and within the dumps. |
||
This AVIRIS image of the two canyons, shows a very general distribution of mixed mineralized (white) and altered materials (red). It is noisy because an attempt has been made to separate sub-pixel-scale dominant phases. |
||
|
The area of Canyon Creek inclusive of Gem to Tamarack #7 contains mine adits, dumps, and waste piles and provides a good example. The next image, which is also AVIRIS, shows in one classification image the mineralized rock (color purple), the fines created from that rock (color cyan), and background alteration (color white) similar to that just seen in the preceding image. A relatively broad cutoff value was used for the matching statistics, which means that the areas shown for each classification may also contain other materials than what is specific to a particular classification. This is a very compelling image and illustrates what can be done with the airborne hyperspectral data. Comparing against topographic maps, mine location maps, and orthophotographs, it can be seen that the purple and cyan areas on the image match mines, and their ìdownstreamî effects, on the reference documents. The 20-m pixels are coarse. However, they can be used to locate the mineralized material in general. |
||
Orthophoto of the area from Gem to Burke in Canyon Creek. This figure is an enlarged AVIRIS image showing mineralized rock (purple), mineralized fines (cyan), and altered areas (white) in the area from just east of Gem to Burke in Canyon Creek. This should be matched against the orthophoto to show correlations with mines and dumps. The purple classification corresponds to mineralized rock containing sulfides, +carbonates and quartz. As this source rock is weathered and mechanically degraded (to progressively finer grain sizes), it is carried away from the point sources (i.e., adits, veins, dumps). This transported material becomes concentrated in the drainages and stream beds; it is represented by the cyan color classification. Several areas are visible where the cyan has moved beyond the purple classification, either downslope from a migrating dump, or downstream from an adit located near a stream. These transport effects become more immediately visible when the classification image is compared to the orthophotograph of the same area in which topographic and drainage patterns are evident. It should be emphasized that the techniques discussed remain at a preliminary stage of development. Additional work needs to be done to verify the classifications. Although at least half a dozen sample sites occur within this image and mineralogical and chemical data are known for each site, further field checking must be done to validate the image classifications and the techniques that created them. |
||
|
||
|
This report is a summary compilation and is provided without references and in an informal format wit the objective of presenting only a brief overview of a major study. Many terms are not defined. Please contact the authors for more detailed information and explanations. The image data included in this report are AVIRIS from the NASA-owned sensor and is public domain information. Processing was done by William Peppin of Advanced Software Applications exclusively for Spectral International. Information discussed relative to mineral and chemical composition of tailings and mine dumps is available from the literature on the area and also through personal communications with United States Geological Survey, United States Bureau of Land Management, United States Fish and Wildlife and United States Forest Service personnel. Proprietary information is not included in this document. |
||
|
||
|
SPECTRAL INTERNATIONAL Peters Geosciences Borstad Associates Ltd. |
||
 |
 |
 |
 |
 |
 |
 |
 |
|