NASA EOCAP NAS-13-99004

PROJECT SUMMARY

INTERNATIONAL SMELTER SITE

STOCKTON DISTRICT

BAUER MILL SITE

DRAGON MINE SITE

PROJECT SUMMARY

Hyperspectral Remote Sensing Technology offers a means of characterizing mine and mill wastes, and the surrounding environment, quickly and more efficiently than surface sampling alone, as is presently and routinely done.

The objective of this project is the study of mine wastes and their impact on their environs including the actual properties, associated watersheds, wetlands and other vegetation communities, using a combination of hyperspectral imaging sensors which form a visible and infrared hyperspectral system.

The project team is drawn from nine companies and organizations, representing small business, academia, mining companies, and state and federal government. Expertise and contributions to this project range from many years of hyperspectral and other remote sensing experience, to exploration and environmental applications, and assessments in the mineral industry and government to market definition and development.

Data and interpretations from the commercial sensors used in this project, the SFSI (SWIR Full Spectrum Imager) and the CASI (Compact Airborne Spectrographic Imager), have been compared to AVIRIS data and ground derived information for the same study areas in Utah during this Phase I. Technology deficiencies and operational limitations have been evaluated. Software issues were also investigated.

The applicability of the system to the abandoned mine land (AML) watershed analysis problem is demonstrated fairly effectively in the analytical ground database, software, and image reports compiled and submitted during Phase I.

The project metrics cover the tasks of

  1. Data acquisition and image processing
  2. Image post-processing
  3. Field work
  4. GIS database development
  5. Data analysis and interpretation
  6. Identification of technology deficiencies
  7. Documentation, public outreach, and presentations
  8. Market strategy and market development

NASA benefits from this project through support of the remote sensing community, advancement of data processing and analysis techniques, and through aiding the promotion of hyperspectral sensors into the operational and commercial realm, in line with the goals of the Hyperspectral Initiative within the OES Commercial Strategy.

This project is co-funded with NASA. A graduate thesis at the Colorado School of Mines is funded as part of the project work. Data and interpretations from Phase I have produced eleven formal and informal presentations to environmental and remote sensing conferences and meetings.

PHASE I RESULTS

Hyperspectral Remote Sensing can effectively be applied to mine waste site evaluation projects with some caveats
It is possible to differentiate between acid drainage and neutral drainage sites
It is possible to track toxic metals
It is potentially possible to map smelter plumes
Information previously unknown about the selected sites was discovered through the remote sensing images
In this application, Remote Sensing is not a stand-alone method
Mine Waste Site Evaluations require multiple data sets because the detail required for remediation is greater than for conventional remote sensing products.
Site Specific Spectral Libraries are essential
Until much more efficient identification algorithms exist, it is necessary to know what is on the ground.
The question to be resolved: Is the data good enough for the available algorithms, or do the algorithms require improvement to equal the quality of the data?
The issue here is that it has not been possible to equally evaluate all the hyperspectral data options (low altitude AVIRIS is missing from this study) nor all algorithms (the USGS Tetracorder program still remains elusive). It is more cost-effective to improve the algorithms.
Much of the Hyperspectral Case Study Information in the Literature has not undergone even a cursory Accuracy Assessment and therefore the validity is suspect.
Commercial Viability -
Only if potential clients can be educated.
If better algorithms can be developed.
If the entire acquisition and interpretive procedure can be streamlined.

MAPS SHOWING THE SITES

Sketch-map showing location of Utah study sites: IS - International Smelter; B,S - Bauer Mill & Stockton Mining District; D ­ Dragon Mine (Tintic Mining District).

IMAGE PROCESSING AND ANALYSIS DISCUSSION

An important key to understanding geologically oriented image processing is the understanding of spectral variability that occurs with mineral species. If this is not addressed, then the image processing will produce inaccurate results. We feel this has been happening in many image products being produced by commercially available programs and processors and even with "research facility" algorithms.
Certain applications may require only that the presence or absence of a mineral is verified and its detailed and accurate distribution may not be required to be known for a reconnaissance pass over an area that will be sampled in detail on the ground. In applications where the image data are tied to other information on the ground, such as distribution of toxic materials, the criticality of site-specific spectral libraries and spectral variability is more apparent.
The initial step in production of mineralogical images for any site in this project involves the development of the spectral library or spectral database for that site.
Although the ground data may show the presence of a category of materials, it may not be present in large enough concentrations or areal distribution to be detected from the air. This is a learning process to determine what the optimum parameters are for specific applications. This presents almost insurmountable problems with 20-m pixels (AVIRIS) and even major problems with 4-m pixels.
Vegetation masks the soil signature and anything else with mineral spectral characteristics, as the organic signature is strong. For Bauer, where there is little vegetation on the wastes, this is not a major problem. For the International Smelter area, where revegetation is extensive, the area must be studied in the negative sense (i.e., look for good mineral response as an indication of failure of revegetation).
The vegetation itself can potentially be utilized to evaluate the effectiveness of the re-vegetation over toxic sites. The CASI data, in particular, can be used to create maps of vegetation types (based on pigmentation).
AVIRIS and the newer SFSI hyperspectral imagery have the potential to provide spectacular classification results in real cases, which can lead to important scientific and/or commercial decisions with respect to mineral exploration, mine waste mitigation, or resource assessment. Common practice has failed to lead to a methodology in which anyone but <highly-trained experts can use this information properly. Standardization of methods, technology transfer, and training become important considerations.
The only two ways to obtain confirmation of image processing results are: 1) check that the targets classified have spectral characteristics similar to the reference minerals, and 2) field check on the ground with a spectrometer. Because standardized accuracy assessment protocols do not exist, often neither 1) nor 2) is done.
We have found that making the correction using an internal average relative reflectance (IARR) calculation gives good results provided the bands being analyzed do not span more than about 750 nm.

REFERENCES

 

Braxton, L.P., and B.W. Buck, 1989, Reclamation of the Carr Fork property, Tooele, Utah, in G.E. Cordy, ed., Geology and hydrology of hazardous waste, mining waste, waste water and repository sites in Utah: Utah Geol. Assoc., Publication 17, p. 115-120.

Brinsmade, R.B., 1908, Mining and Milling at Stockton, Utah: Engineering and Mining Journal, v. 85.

Hall, R.B., 1985, Dragon halloysite deposit, Tintic Mining District, Juab County, Utah, in Field Trip Guidebook, Clay and Clay Minerals, Western Colorado and Eastern and Central Utah, 1985 International Clay Conference, Denver, CO, July 28-August 2, 1985: p. 31-19.

Hilton, Lynn M., 1994, Stockton's ghost towns and abandoned mines--Lecture & self-guided field trip (1998 Ed.): Stockton, Utah, 21 pp.

James, L.P., 1982, Sulfide ore deposits related to thrust faults in northern Utah, in D.L. Nielson, ed., Overthrust Belt of Utah: Utah Geol. Assoc., Publication 10, p. 91-99.

Kildale, M.B., and R.C. Thomas, 1957, Geology of the halloysite deposit at the Dragon Mine, in Geology of the East Tintic Mountains and Ore Deposits of the Tintic Mining Districts: Utah Geological Society, Guidebook to the Geology of Utah, v. 12, p. 94-96.

Lufkin, J.L., 1965, Geology of the Stockton stock and related intrusives, Tooele County, Utah: Brigham Young University Research Studies, Geological Series 12, p. 149-164.

Morris, H.T., 1985, Geology, ore bodies and halloysite deposits of the Tintic Mining District, in Field Trip Guidebook, Clay and Clay Minerals, Western Colorado and Eastern and Central Utah, 1985 International Clay Conference, Denver, CO, July 28-August 2, 1985: p. 40-53.