CA9200955

REMOTE SENSING TO MONITOR URANIUM TAILING SITES - A REVIEW

(AECB Project No. 5.142.1)

wyï-'y-iy: :- ..• INFO-0403 Atomic Energy Commission de contrôle 1*15 Control Board de l'énergie atomique P.O. Box 1046 C.P. 1046 Ottawa. Canada Ottawa. Canada K1P5S9 K1P5S9

REMOTE SENSING TO MONITOR URANIUM TAILING SITES - A REVIEW

(AECB Project No. 5.142.1)

by

Intera Renting Ltd.

A research report prepared for the Atomic Energy Control Board Ottawa, Canada

Published February 1992

Canada Research report REMOTE SENSING TO MONITOR URANIUM TAILINGS SITES - A REVIEW A report prepared by Intera Kenting under contract to the Atomic Energy Control Board.

ABSTRACT This report concerns the feasibility of using remotely-sensed data for long-term monitoring of uranium tailings. Decommissioning of uranium mine tailings sites may require long-term monitoring to confirm that no unanticipated release of contaminants occurs. Traditional ground-based monitoring of specific criteria of concern would be a significant expense depending on the nature and frequency of the monitoring. The objective of this study was to evaluate whether available remote-sensing data and techniques were applicable to the long-term monitoring of tailings sites. This objective was met by evaluating to what extent the data and techniques could be used to identify and discriminate information useful for monitoring tailings sites. The cost associated with obtaining and interpreting this information was also evaluated. Satellite and aircraft remote-sensing-based activities were evaluated. A monitoring programme based on annual coverage of Landsat Thematic Mapper data is recommended. Immediately prior to and for several years after decommissioning of the tailings sites, airborne multispectral and thermal infrared surveys com- bined with field verification data are required in order to establish a baseline for the long-term satellite-based monitoring programme. More frequent airborne surveys may be required if rapidly changing phenomena require monitoring. The use of a geographic information system is recommended for the effective storage and manipulation of data accumulated over a number of years. RESUME Le présent rapport traite de la possibilité de contrôler à distance et à long terme les résidus d'usines de concentration d'uranium. Le déclassement des aires de résidus miniers d'uranium peut exiger un contrôle long terme pour qu'il n'y ait pas de rejets imprévus de contaminants. Les moyens de contrôle traditionnels au sol de certains paramètres particuliers pourraient coûter cher dépendant de la nature et de la fréquence des mesures. La présente étude visait à évaluer si les données et techniques de télémesure disponibles étaient applicables au contrôle a long terme des aires de résidus miniers. Cet objectif a été atteint en examinant dans quelle mesure ces données et techniques pouvaient servir à déterminer et à identifier les renseignements utiles pour le contrôle des aires de résidus. Le coût de la collecte et de l'interprétation de ces données a été estimé et on a aussi évalué la pertinence de télémesures prises à partir d'avions ou de satellites. -11-

On recommande d'instaurer un programme de contrôle basé sur les données annuelles recueillies par le système Landsat Thematic Mapper. Immédiatement avant le déclassement d'une aire et pendant plusieurs années après le déclasse- ment, on recommande de procéder à des relevés multispectraux, dans l'infrarouge thermique, combinés avec des vérifications au sol, afin d'établir l'état initial qui servirad e référence pour le contrôle à long terme par satellite. Des mesures aéroportées plus fréquentes pourraient s'avérer nécessaires si des phénomènes à variations rapides l'exigeaient. Le recours à un système de données géographiques est recommandé pour le stockage et la manipulation des données accumulées au fil des ans.

DISCLAIMER The Atomic Energy Control Board is not responsible for the accuracy of the statements made or the opinions expressed in this publication and neither the Board nor the author assume liability with respect to any damage or loss incurred as a result of the use of the information contained in this publication. -111-

TABLE OF CONTENTS

Page

ABSTRACT i

LIST OF FIGURES vi

LIST OF TABLES vii

1. INTRODUCTION 1

2. RENOTE SENSING REVIEW 3 2.1 A Discussion on the Fundamentals of Remote Sensing 3 2.1.1 Fundamentals of Remote Sensing 3 2.1.1.1 Electromagnetic Radiation 3 2.1.1.2 Energy Interactions 4 2.1.2 Remote sensing systems 7 2.1.2.1 Non-Imaging Systems 7 2.1.2.2 Imaging Systems 7

3. TECHNICAL FEASIBILITY 9 3.1 Sensor Systems 9 3.1.1 Satellite-Based Sensors 9 3.1.1.1 Landsat Multispectral Scanner (MSS) 11 3.1.1.2 Landsat Thematic Mapper (TM) 12 3.1.1.3 SPOT Multispectral (HLA) and Panchromatic (PLA) 13 3.1.1.4 Soyuzkarta 14 3.1.1.5 Radarsat 14 3.1.2 Airborne Sensors 15 3.1.2.1 Multi Element Linear Array Imaging System (MEIS) 15 3.1.2.2 Compact Airborne Spectrographic Imager (CASI) 17 3.1.2.3 Laser Fluorosensor 17 3.1.2.4 Thermal Infrared 17 3.1.2.5 Aerial Photographs 18 3.1.2.6 Airborne Radar 20 3.2 Monitoring Evaluation Criteria 21 3.2.1 Vegetation 21 3.2.1.1 Encroachment 21 3.2.1.2 Die Off 24 3.2.1.3 Stress and Morbidity 23 3.2.1.4 Coverage/Type 25 -IV- TABLE OF CONTENTS fCont'd) Page 3.2.2 Moisture 26 3.2.2.1 Drainage 26 3.2.2.2 Ponding 27 3.2.2.3 Seepage 27 3.2.3 Soil and Rock 27 3.2.3.1 Dam Failure 28 3.2.3.2 Surface Erosion 28 3.2.3.3 Sub-aqueous Erosion 28 3.2.3.4 Diversion Channel Silting/Erosion 29 3.2.3.5 Waste Rock/Open Pits 29 3.2.4 Radioactivity 29 3.3 Monitoring Plan Design 30 3.3.1 Long-Term Monitoring 30 3.3.1.1 Data Requirements 31 3.3.1.2 Data Processing 31 3.3.2 Short-Term Monitoring 33 3.4 Information Extraction and Storage 34 3.4.1 Image Processing 34 3.4.2 Geoaraphic Information Systems 34

4. ECONOMIC FEASIBILITY 36 4.1 Satellite Based Monitoring 36 4.1.1 Data Acquisition 37 4.1.2 Image Processing 37 4.1.3 Data Output 38 4.1.4 Interpretation 39 4.1.5 Integration 40 4.2 Airborne Based Monitoring 41 4.2.1 Data Acquisition 41 4.2.2 Image Processing 43 4.2.3 Data Output 43 4.2.4 Interpretation 44 4.2.5 Integration 44 4.3 Computer Processing Facilities 44 4.3.1 Hardware 45 4.3.2 Software 46 4.3.3 Owned Facilities Vs Service Bureaus 47 -V-

TABLE OF CONTENTS fCont'd)

5. SATELLITE DEMONSTRATION 49 5.1 Study Area 49 5.2 Data Processing Methodology 49 5.2.1 Data Input 52 5.2.2 Geometric Correction and Resampling 52 5.2.3 Linear Contrast Stretch Enhancement 54 5.2.4 Linear Stretch Difference Enhancement 54 5.2.5 Principal Component Enhancement 54 5.3 Image Interpretation and Evaluation 62 5.3.1 Spatial Resolution 62 5.3.2 Vegetation/Moisture Enhancement 62 . 5.3.3 Enhancement of Tailings Area 63 5.4 Conclusions from Satellite Demonstration 67 6. CONCLUSIONS AND RECOMMENDATIONS 68

7. BIBLIOGRAPHY 69

APPENDIX A DEMONSTRATION IMAGES

APPENDIX B SELECTED PAPERS

APPENDIX C SUPPLIERS PUBLISHED LISTS -VI- LIST OF FIGURES Page Figure 2.1 The Electromagnetic Spectrum 5

Figure 2.2 Transmission of Energy Through the Atmosphere as a Function of Wavelength. Wavelength Regions of High Transmittance are Atmospheric Figure 3.1 Methodological Approach to Data Processing 32 Figure 4.1 Satellite Data Monitoring Steps 36 Figure 4.2 Airborne Data Monitoring Steps 41 Figure 5.1 Rabbit Lake Mine Site Plan 50 Figure 5.2 TM Bands 3, 4, 5 colour composite of entire quadrant 51 Figure 5.3 Data Processing Methodology 53 Figure 5.4 Histograms of TM Bands l(a), 2(b), 3(c), 4(d) 55 Figure 5.4 Histograms of TM Bands 5(c), 6(f), 7(g) 56 Figure 5.5 Landsat TM 30 metre colour composite image Bands 3, 4, 5 57 Figure 5.6 Landsat TM 20 metre colour composite image Bands 3, 4, 5 57 Figure 5.7 Landsat TM 10 metre colour composite image Bands 3, 4, 5 58 Figure 5.8 Landsat TM 10 metre colour composite image Bands 1, 3, 5 59 Figure 5.9 Landsat TM Bands 1, 3, 5, 10 metre difference enhancement 60 colour composite Figure 5.10 Principal Component Enhancement 10 metre resolution 61 Figure 5.11 Vegetation and Tailings Enhancement of the B-zone Mine Area 64 Figure 5.12 Vegetation and Tailings Enhancement of the Central Tailings 65 Disposal and Stockpile Areas Figure 5.13 Vegetation and Tailings Enhancement of the Tailings and Precipitation Pond Area 66 -Vll- LIST OF TABLES Page Table 3.1: Summary of Spaceborne Sensor Characteristics 10 Table 3.2: Spaceborne Sensors for Monitoring of Mine Tailings 22 Table 3.3: Airborne Sensors for Monitoring of Mine Tailings 23 Table 4.1: Satellite Data Acquisition Prices 37 Table 4.2: Satellite Data Processing Costs 38 Table 4.3: Satellite Output Product Costs 39 Table 4.4: Airborne Data Acquisition Prices 42 Table 4.5: Airborne Data Processing Costs 43 Table 4.6: Airborne Data Output Costs 44 Table 4.7: Owned Facilities versus Contracted Processing Services 48 1. INTRODUCTION

Mining activities have always been an important component of the Canadian non- renewable resource sector. One inevitable product of these activities is the waste material that results from processing of the ore. With increasing attention being focused on the environmental impacts of such operations, the industry has made great progress in ensuring that damage to the immediate environment is minimized.

In the case of uranium mining operations, the waste material is combined with water to form a slurry. This material is then transported by pipe a short distance from the mill (< 2km) and deposited in settling basins called tailings ponds. In Canada, these sites range from 14 to 160 ha in size, and estimates are that they receive 400 million tons of waste material annually and require over 4.5 billion litres of water daily (Ritcey, 1989). Once deposited, the tailings eventually settle out and consolidate into a fine grained granular residue while the water is decanted and either evaporates or drains off.

Decommissioning of uranium mills may require long-term monitoring of the tailings sites to confirm that no unanticipated or unacceptable release of contaminants occurs. In addition, there will probably be a desire to monitor the natural regeneration of vegetation on the sites as well as the quality of the runoff water. The cost of this monitoring will be borne by the party responsible for the tailings site after decommissioning. Such monitoring may be an expensive activity depending on the nature, extent, and frequency of the exercise. Conventional ground survey techniques are costly due to travel time, number of personnel required, size of area to cover, difficulty of access to sites, frequency of site visits, and complexity of tests available to inspectors. Remote sensing techniques may provide a cost effective alternative or supplement to the more traditional on-site inspection programs.

The tailings features in the Elliot Lake region of northern Ontario are considered typical of Canadian waste sites. These tailings are all stored in natural drainage systems. Containment is achieved by damming the outlet channels and constructing berms and dykes around the deposits. For the purposes of this study, the Atomic Energy Control Board (AECB) identified specific concerns that -2- would have to be addressed in a monitoring program. This report is intended to couple those concerns with the current, and near future, capabilities of the Canadian remote sensing industry:

The objectives of this report are:

1) to outline the advantages and limitations of available remote sensing techniques for long-term monitoring of tailings sites, and 2) to determine the feasibility of conducting a monitoring program using remotely sensed data. -3-

2. REMOTE SENSING REVIEW

2.1 A DISCUSSION ON THE FUNDAMENTALS OF REMOTE SENSING

The following is a brief discussion on the fundamental principles of remote sensing and remote sensing systems. It is intended as additional background material for the reader if required. For a more complete discussion on these topics, it is suggested the reader consult one of the many excellent text books available such as Lillesand and Kiefer (1979) and Siegal and Gillespie (1980).

2.1.1 Fundamentals of Remote Sensing

The definition of remote sensing in the context which it is normally applied, deals with obtaining information about a phenomenon through the analysis of data acquired by a device not in contact with the phenomenon. The quantities that are measured are the emitted, scattered, and reflected electromagnetic energy from the phenomena under investigation. Useful information is then extracted through interpretation of these measurements. In a broader context the term may be applied to the use of force fields such as gravity or magnetics, and mechanical vibrations of acoustical or seismic origin since these also involve the sensing of phenomena froir a distance. The accepted use of the term, however, concerns only the sensing of electromagnetic energy from the visible, infrared and microwave wavelengths. In this report, remote sensing refers to both electro- optical sensors which transform electromagnetic radiation into electrical signals to produce images, and imaging systems which measure target radiation or reflectance.

2.1.1.1 Electromagnetic Radiation

The spectral region of interest in remote sensing ranges from the blue end of the visible at approximately 0.4 /m to the microwave wavelengths of up to approximately I metre (Figure 2.1). The information about an object is contained in the electromagnetic radiation passing from the target to the sensor. The amount of radiation reflected or emitted by a particular substance or surface will vary at different wavelengths. -4-

The source of the energy used by most remote sensing systems is the sun. The amount of electromagnetic energy produced by t:.e sun reaches a maximum in the visible portion of the spectrum. Conversely, the amount of natural energy available in the microwave portion of the spectrum is limited. This has resulted in the necessity for most microwave sensor systems to supply their own source of energy. In the infrared portion of the spectrum, much of the energy received above the Earth's surface is absorbed by water vapour, carbon dioxide, and other substances in the atmosphere (Figure 2.2). As a result, infrared systems rely on several atmospheric transmittance windows whose limits are at specific, known wavelengths.

2.1.1.2 Energy Interactions

The analysis and interpretation of remotely sensed data requires an understanding of the nature of interaction between electromagnetic energy, the atmosphere, and Earth surface materials and features. The amount of solar energy, incident on, and emitted or reflected from, the ground surface is attenuated by scattering and absorption processes which occur throughout the atmospheric volume.

Electromagnetic radiation is either absorbed, transmitted or reflected depending upon the specific wavelengths involved and the physical properties of the material or surface receiving the radiation. For example, water generally appears blue to the observer because its reflectance peak occurs within the visible blue portion of the spectrum. On the other hand, water appears black in thermal infrared imagery because energy at these wavelengths is absorbed. Furthermore, water in glacier fed lakes appears greenish due to the effects of the fine suspended sediment which enhances reflection of green wavelengths.

Reflected solar radiation from water, soil, vegetation, and other cover types contains information which is usually unique to that surface material. These spectral 'signatures' can be used by remote sensing specialists to study environmental parameters of geological and hydrological importance. The known spectral properties of different vegetation types and species can be used to deduce information from these data. Figure 2.1: The Electromagnetic Spectrum (from Sabins, 1978)

ATMOSPHERIC ABSORPTION BANDS

5 6 7 8 9 10 11 M 13 WAVELENGTH, Aim

Figure 2.2: Transmission of Energy Through the Atmosphere as a Function of Wavelength. Wavelength Regions of High Transmittance are Atmospheric Windows. Gases Responsible for Absorption are Noted (from Sabins, 1978) -6-

Passive optical remote sensing techniques can be used to monitor changes in vegetation health and vigour. As an example, it has been found that, for many species, there is a high correlation between reflected radiation and leaf chlorophyll content (Horler, 1983). As chlorophyll content increases, the reflected light in the .70 /m range shifts to progressively longer wavelengths. The position of the edge of this curve then becomes a good measure of the chlorophyll content. The shifting of this reflectance curve towards shorter wavelengths has been termed the blue-shift or red-edge shift. The presence of vegetation can be determined from chlorophyll absorption bands which exist at .45 /m and .65 fm- The condition of vegetation can be assessed using data between 1.0 /m and 2.5 fm which denote the presence or absence of hydrous and carbonate minerals. In addition, the region from 1.5 /ym to 2.5/ym responds to leaf water content which allows for study of soil and plant moisture conditions. Changes in the chemical properties of soils can be manifested by changes in the vegetation cover characteristics. Furthermore, morphological and physiological changes are often due to anomalous concentrations of specific metals. Reflectance data acquired between 0.5 fm and 1.1 fm may be used to determine the presence or absence of iron-oxide or hydroxide minerals in soils.

Another aspect to consider, in terms of the interpretation of remotely sensed data, is the changing appearance of features due to the controlling influence of factors such as water content, sun angle, time of year, and condition. For example, an unvegetated sand area occurring beside a stream will appear much different, in terms of its characteristic reflectance, than a sand bar just a few metres away. Despite the homogeneity of the sand material, the controlling factor in this case is the amount of moisture in the sand. As a result, the interpreter must have some background knowledge of these factors rather than relying on comparison to characteristic reflectance curves. In order to properly analyze and interpret remote sensing data, it is critical to first have an adequate understanding and appreciation of surface and atmospheric processes and energy interactions. -7-

2.1.2 Remote sensing systems

Remote sensing devices or sensors measure the intensity of electromagnetic radiation (EMR) emitted or reflected from surface targets as a function of time, wavelength, space, geometry (including the angular orientation of the target with respect to the observer) and polarization of the radiation (Reeves, 1975). Given the complexity there is no single instrument which is appropriate for every need. The decision concerning which sensor is best suited to a particular application is made by first considering the parameters that must be measured, then evaluating all available sensors that are capable of making those measurements. Remote sensing systems may be divided into two classes: imaging and non-imaging systems.

2.1.2.1 Non-Imaging Systems

The non-imaging systems are primarily used for experimental work to determine precise characteristics of the interaction of energy with specific targets. Spectroradiometers, for example, are electro-optical instruments which will provide a spectral response measurement over a range of wavelengths for a single target; the output being either an analogue trace or digital counts. This type of instrument is best used for investigating the specific characteristics of natural phenomenon, from which broader models may be developed for use in other applications. In one application in the southwestern United States this type of instrument is being used to identify surface rocks and minerals. The absence of vegetation and soils in the arid areas where this work is being carried out provides optimal conditions for using this type of technology for real time exploration field work. In this case, a know!edge-base containing known mineral spectra is stored in the instrument and measurements are compared directly to elements of this database in order to facilitate identification.

2.1.2.2 Imaging Systems

Imaging systems are often divided into photographic and electro-optical systems. The photographic systems, which are most familiar to the average person, consist -8- of an enclosed chamber with a lens at one end and a light sensitive film at the other. The lens gathers the reflected energy and transmits it onto the light sensitive film. A shutter serves to regulate the amount and duration of energy reaching the film. These types of remote sensing systems are generally mounted on airborne platforms and operated at high, medium, and low altitudes depending on the required amount of spatial detail. Panchromatic and colour aerial photographs are produced in this way. The strengths of these types of systems lie in their simplicity and economical operation. They also sense in the visible portion of the spectrum, making interpretation of the imagery somewhat simpler for the human interpreter. When using data beyond the visible wavelengths, an understanding of how specific wavelengths behave with specific material types is essential before interpretation can take place.

Electro-optical imaging systems are most commonly used in remote sensing. These systems utilize an array of detectors which scan across portions of the earth's surface in a regular manner. The electrical responses of the detectors are modulated by the amount of incident electromagnetic radiation at specific band widths and location over the spectrum. A spatially gridded image (raster) is then produced by assigning integer values to each level of response. The advantages over photographic systems are the ability to readily transmit, record, process, and analyze the output signal since it is in an electrical form. -9-

3. TECHNICAL FEASIBILITY

3.1 SENSOR SYSTEMS

The following section provides a discussion on the technical feasibility of using remote sensing data and techniques for monitoring uranium tail ings sites. This is primarily based on a review of relevant published papers and reports and a review of different of sensor capabilities. Copies of the most relevant papers are provided in Appendix C.

Two main categories of sensors will be discussed. These are satellite-based and airborne sensors. Only those sensors which may be considered operational or commercially available are discussed.

3.1.1 Satellite-Based Sensors

For some years now, satellite remote sensing has been recognized as a valuable tool for monitoring the Earth's natural resources. Satellites can provide relatively inexpensive, repetitive datasets that can either augment or replace expensive, and often time consuming, ground based monitoring programs. Satellite data provide an instantaneous view of surface phenomena over large areas that are often inaccessible to ground personnel.

An examination of current operational platforms reveals that there is a distinct gap, in terms of spatial coverage and detail, between low and high resolution systems. For instance, the Advanced Very High Resolution Radiometer (AVHRR) onboard the NOAA (National Oceanic and Atmospheric Administration) series of satellites has a ground resolution of 1.1 km at nadir, and covers a ground swath of approximately 2500 km. The next coarsest resolution is found with the Landsat MSS system at 80 m ground resolution and a 185 km swath width. Clearly, given that tailings sites are on the order of 5-10 ha, 1.1 km data would not contain sufficient detail for an adequate monitoring program. Consequently, only satellites capable of providing data with a spatial resolution of at least 80 m were considered in this report. A summary of sensor characteristics is provided in Table 3.1. -10-

Tabie 3.1 Sumnary of Satellite Based Sensor Characteristics

Spatial Name Type Resolution Spectral Resolution

Landsat - MSS Multispectral 80 m Band 4 .5 - .6 fm Band 5 .6 - .7 fm Band 6 .7 - .8 fm Band 7 .8 - 1.1 JM - TM Multispectral 30 m Band 1 .45 - .53 fm Band 2 .52 - .57 fm Band 3 .63 - .69 fm Band 4 .76 - .90 fm Band 5 1.55 - 1.75 fm Band 7 2.08 - 2.35 fm Band 6 10.80 - 12.50 fm SPOT - HRV Multispectral 20 m Band 1 .50 - .59 fm Band 2 .61 - .68 tm Band 3 .79 - .89 urn - PLA Panchromatic 10 m Band 1 .51 - .73 fm Soyuzkarta - KFA-1000 Multispectral 5 m Band 1 .57 - .67 fm Band 2 .67 - .81 fm - MK-4 Multispectral 6 m Band 1 .46 - .50 fm Band 2 .51 - .56 fm Band 3 .63 - .69 fm Band 4 .81 - .90 fm Band 5 .40 - .70 fm Band 6 .58 - .80 urn Radarsat Radar 25 m Band C 3.8 - 7.5 cm -11-

3.1.1.1 Landsat Hultispectral Scanner

Since the first launch in 1972, the Landsat series of satellites has been collecting Multispectral Scanner (MSS) data that have been shown to be extremely useful for monitoring the Earth's resources. The MSS sensor acquires data in 4 spectral ranges: .50-.60 fm, corresponding to green reflectance; .60-.70 //m, corresponding to red reflectance; and two bands sensitive to reflected, near- infrared wavelength ranges of .70-.80//m and .80-1.10 im. The scanner provides spatial resolution of 80 m and covers a swath width of 185 km. Currently, data is being received from the fifth Landsat platform which has a revisit cycle of 16 days. The launch of Landsat 6 is scheduled for mid 1991. The historical data provides a valuable data archive, while the frequency of coverage is adequate for most monitoring efforts. MSS products consist of digital data on computer compatible tapes (CCT's) and colour or black and white photographic prints and transparencies.

MSS data have been evaluated for mining waste monitoring (Boldt and Scheibner, 1987; Gagnon et al., 1985) and for monitoring the environmental impact of mining activities (Moore et al., 1977). The latter study showed that these data could be used to determine approximate percent vegetative cover and surface area for tailings sites. However, study areas had to be on the order of at least 1-2 hectares in size. The error associated with area estimates was found to be approximately 10% for sites less than 20 hectares, and 5% for most larger sites. This error is attributed to the coarse spatial resolution of the data.

In terms of analysis procedures, Pickup and Nelson (1984) demonstrated that band ratios 4:6 and 5:6 could be used to categorize eroding, stable, and depositional environments around small streams and drainage channels. Moore et al. (1977) found visual interpretation to be the most useful technique for classifying surface features around Elliot Lake, Ontario mining sites. Water, tailings, waste rock, green vegetation, and brown vegetation could all be separated through this technique. That study also concluded that initial ground truthing is essential in order to establish the baseline classification scheme for use in long-term programs. Most studies have concluded that, although the spectral -12-

information provided by the MSS system is adequate, the limited spatial resolution of these data is not sufficient for most monitoring programs (Gregory, 1975).

3.1.1.2 Landsat Thematic Napper

In the early 1980's, a new sensor was placed onboard the Landsat platforms. The Thematic Mapper (TM) was a long awaited sensor as it has improved spatial, spectral, and radiometric characteristics over the MSS system. TM data is acquired at 30 m ground resolution in 7 spectral bands covering the range .45- 1.25//m. Three bands were selected to provide coverage of the blue (.45-.53 fm), green (.52-.57 fm), and red (.63-.69 fm) portions of the visible spectrum. Three more bands cover the near infrared spectrum at .76-.90 tm and the short-wave portions at 1.55-1.75 tm and 2.08-2.35 fm. The final band corresponds to the principal atmospheric window for thermal infrared wavelengths, 10.80-12.50 /m (see pages A2, A4 and A10).

TM offers four distinct bands in the infrared wavelengths which is the region most sensitive to changes in vegetation characteristics. In addition, bands 3,4, and 5 are well suited for differentiation of vegetation types when used in combination. Consequently, TM is often the data source used for studies in forestry and geobotany (Ahem and Leckie, 1987). Since green vegetation is inversely correlated to the red band due to chlorophyll absorption, and IR reflectance is positively correlated to biomass, a.vegetation index based on the difference over the sum of 2 bands is also useful for vegetation studies (Pitbaldo and Amiro, 1982).

Although a review of the literature has failed to locate studies which specifically used TM data for monitoring waste sites, there is evidence of the utility of TM data for exploration geology and geobotany (Hornsby and Bruce, 1986) in addition to the more traditional forestry and agriculture applications. Based on these studies, it can be assumed that TM data would be suitable for monitoring of vegetation conditions and absolute areas of cover on tailings sites. -13-

3.1.1.3 SPOT Multispectral and Panchromatic

SPOT (Systeme Pour 1'Observation de la Terre) is an earth observation satellite launched by France in 1986. The satellite platform contains two High Resolution Visible (HRV) sensors which are capable of obtaining data over swaths of 60 km in nadir mode and 80 km off-nadir. SPOT has a 26 day cycle, but has a pointing capability unlike other existing satellite platforms. This unique capability means that data can be obtained over a specific site from several overpasses. If imagery of a site is obtained from adjacent overpasses it will be from different view angles, providing the ability to obtain stereoscopic coverage. In addition, users can request coverage of specific sites over several adjacent orbits. This unique feature of the SPOT system reduces the probability that a site may be missed due to system malfunctions or weather conditions, and also provides for more user control over data acquisition.

Using the Multispectral Linear Array (MLA) mode, data are collected in three spectral bands: green (.50-.59 fjm) centred in the chlorophyll reflectance range, red (.61-.68//m) corresponding to chlorophyll absorption, and near infrared (.79- .89 fm). These data have a spatial resolution of 20 metres.

In the Panchromatic Linear Array mode (PLA), data are acquired in one spectral band sensitive to wavelengths between .51 im and .73 //m (see page A14). The ground resolution of these data is 10 metres. The PLA mode is similar to black and white aerial photography but has the advantage of being less costly to obtain. Data can be acquired from the Canada Centre for Remote Sensing (CCRS) as precision corrected geocoded products in digital and photographic format having a resolution of 12.5 m (MLA) or 6.25 m (PLA). Although SPOT data have only been available for 2 or 3 years, preliminary indications are that the spatial resolution capabilities are well suited to applications concerned with localized phenomena. Although the spectral capabilities are not as extensive as TM data, when combined with the improved spatial qualities, SPOT data are valuable to environmental monitoring programs. -14-

3.1.1.4 Soyuzkarta

Under terms of a trade agreement with the USSR, an American firm is now marketing high resolution satellite photography from the Soyuz series of satellites. This effort represents the first time the USSR has offered remotely sensed data to the public.

Soyuzkarta products are available from two camera systems. The KFA-1000 system produces photographs with resolution of 5 metres covering an area of approximately 60 X 60 miles. Imagery is available in panchromatic mode, or in what is termed spectrozonal colour. This product is sensitive to .57-.67 im (green/red) and .67-.81 JM (near IR) ranges. The second system, MK-4, utilizes a multispectral camera with a ground resolution of 6 metres. 4 band selections are available from 6 available spectral ranges: .46-.50//m, .51-.56 fm, .63-.69 fm, .81-.90//m, .40-.70 /m, and .58-.80//m. One feature of both systems is that, with 60% overlap between scenes, stereo coverage is possible. The disadvantage of this data is that it is only photographic and not digital.

Since Soyuzkarta data have only been available since 1989 there has not yet been sufficient time to evaluate the utility of the data for environmental applications. Nevertheless, it now appears as if the USSR will be offering more satellite services in the near future. A recent report announced that digital radar data with a resolution of 25 m should be available from the USSR in late summer 1990.

3.1.1.5 Radarsat

In 1994, Canada will launch a unique satellite known as Radarsat. This platform will carry a high resolution C-band (3.8-7.5 cm wavelength) synthetic aperture radar (SAR) system producing a ground resolution of 25 m over a swath of 500 km. (Note: for a discussion of radar fundamentals, see section 3.1.2.6) The SAR antenna will be able to point its signal anywhere within the swath between 20-50° to either side of the orbit track. In addition, the SAR will have the ability to zoom in and out for greater or lesser detail. The data received from -15-

Radarsat will be used for applications ranging form monitoring of agricultural conditions in the Prairies to ice conditions in the Beaufort Sea.

With a three day repeat cycle, and the inherent ability to image through all weather conditions, the probability of acquiring data over a specific site is quite high. Other advantages radar remote sensing has over other systems is that data acquisition is independent of local weather conditions, and it has the ability to penetrate greater depths below the surface; this increases with longer wavelengths. It should be noted that for the first time Canada will have control of satellite operations, which is an important consideration when evaluating national concerns.

3.1.2 Airborne Sensors

In many cases it is desirable to have more control over sensing systems than is provided with most satellite programs. As a result, a suite of airborne sensors has been developed, several of which are now commercially available. Most of these systems have great flexibility in terms of the spectral ranges and resolution available. Several of them allow the user to select the wavelength regions to be sensed. In addition, because they are acquired at much smaller distances from the surface, airborne remotely sensed data have much higher spatial resolutions, typically less than a few metres.

The discussion here covers only those proven systems which are currently operational.

3.1.2.1 Multi Element Linear Array Imaging System

In the last 10-15 years, Canada has been active in the development and operation of airborne electro-optical sensors for resource management applications. This effort has involved cooperation between government and industry, and has produced such systems as the Multi-Element linear array Imaging System (MEIS) and several line imaging systems (see pages A16 and A18). -16-

The MEIS system is a high performance digital multispectral imager which is capable of detecting small changes in target reflectance that may not be picked up by other sensors. In addition, because it is an airborne system it can provide very small ground resolutions, .4-7 m, depending on the flight altitude and focal length selected. The current model, MEIS II, can record eight spectral channels and incorporates real-time radiometric and geometric corrections. The silicon detectors respond to energy from .350 - 1.100/M, but special filters are used to determine which wavelength ranges are imaged. Stereo image data acquisition can also be provided, subject to some restrictions. In addition, the system owned by CCRS offers an airborne MSS scanning imager to complement the MEIS sensor. This instrument covers .39-13 tm with 10 band capability, and can be recorded in parallel with the MEIS data. It should be noted that there is only rne MEIS scanner in existence. This could be considered a disadvantage for a long-term monitoring program.

The ability to acquire image data in narrow spectral bands has made the MEIS an important tool for vegetation stress studies and forest damage surveys (Till, 1987). Use of the da:a in forestry applications has demonstrated the value of the high spatial resolution, flexibility of wavelength selection, and good radiometric resolution for species identification and condition assessment (Ahem and Leckie, 1987).

Gagnon et al. (1985) found that standard image analysis techniques on MEIS data provided good results for mapping vegetation zones and surface areas around gold tailings sites in Quebec. The spatial and spectral qualities of the data allowed for clear discrimination of plant cover types and unvegetated tailings materials. The misclassification of trees, shrubs, horsetails, sparse vegetation, grass, and bare tailings was less than 10%, and was attributed to factors such as cloud and structure shadows. The ability to select spectral ranges should allow for detection of such other phenomena as iron staining on vegetation on tailings sites. -17-

3.1.2.2 Compact Airborne Spectrographic Imager

The Compact Airborne Spectrographic Imager (CASI) is a commercially operated airborne instrument that produces digital multispectral data. This system is wall suited to applications where subtle differences in spectrum or reflectance on a small scale need to be detected. It has the capability to acquire 288 user- programmable channels across the wavelength range of .45-.90 fm with spectral resolution of up to .0018 fm and spatial resolution of .5 m (see pages A21 to A24). These capabilities provide for application specific mission planning. The system is easily transportable and operates from light aircraft or helicopters.

To date, the system has been used in applications such as detecting algae blooms and aquatic vegetation, and monitoring suspended sediment and vegetation stress.

3.1.2.3 Laser Fluorosensor

The Laser Fluorosensor is an imaging spectrometer that uses a pulse of ultraviolet laser energy to illuminate a small area on the ground. Any fluorescence that is excited by the energy pulse is then measured by a receiver on the aircraft. The system was designed to image ocean chlorophyll fluorescence and spectral reflectance changes in water caused by phytoplankton. The chlorophyll strongly absorbs blue light so that the colour of water is seen to change as phytoplankton levels change. This sensor is also used for measuring the fluorescence spectrum of other substances, such as oil slicks. The fluorescence level acts as a signature and is used the classify the substance using comparison to known spectra.

3.1.2.4 Thermal Infrared

All matter above 0 Kelvin radiates energy at thermal infrared wavelengths (30 fm - 140 //m). This technology exploits that characteristic and uses it to detect temperature differences between ground targets. This particular type of sensor is sensitive to thermal, or radiating, heat energy and is best used to detect targets which are significantly warmer or cooler than their surroundings. These -18-

products can be used to delineate boundaries between vegetation and water or exposed earth and grasses, and to locate internal drainage channels within tailings ponds (Aronoff, 1978). The spatial resolution of these systems is commonly less than 1 m, with thermal resolution of 0.2°C.

Thermal infrared sensing techniques are particularly well suited to studies concerned with phenomena which are associated with the presence or absence of water. Water has a high thermal inertia which means that it retains its heat level longer than most other materials. Consequently, it is slow to give up heat energy which has been absorbed during the daytime and, therefore, appears much warmer than its surroundings in nighttime imagery. This characteristic provides the basis for techniques which identify dam and drainage channel leakage, groundwater discharge, and heat loss locations. In one study, thermal infrared imagery collected during night flights were used to detect effluent plumes at deposition locations (Aronoff, 1978). Damp ground is cooler than dry ground throughout the day because of the cooling effect of evaporation. As a result, moist areas adjacent to dry land or water bodies have cooler signatures. This can be used to identify locations where seepage or leaks have occurred, or where the ground water table has reached, or is just beneath, the surface.

Thermal IR can also be used to monitor vegetation patterns. Imagery acquired at night can be used to determine moisture variations within plant canopies since the water will appear warmer than the surroundings. Daytime imagery can also be used to determine the amount of biomass present given that the volume of air trapped within vegetation is warmed by the soil and biomass and contributes to a higher overall canopy temperature.

3.1.2.5 Aerial Photographs

Aerial photography is the original form of remote sensing and is widely used in all areas of environmental monitoring. This particular technique covers the ultra violet (.30 -.40 //m), visible (.40 -.70 /m), and near-infrared (.70 - .90 fm) portions of the electromagnetic spectrum. The range from .70 - .90 /m is sensitive to reflected infrared energy, not thermal infrared, and is often -19- called the photographic IR range. The ground resolution of a photograph is determined by the aircraft height and the focal length of the camera. Typically, ground resolutions for aerial photography range from .25-1.0 m.

The most important characteristic of colour infrared imagery is that it is very sensitive to characteristics of vegetation. In the visible spectrum, blue and green light are absorbed and approximately 20% of the incident green light is reflected, hence the green colour of healthy vegetation. In the colour IR region, however, vegetation has a wider reflectance range containing higher levels of reflectance. This higher reflectance is caused by the internal cell structure. When vegetation is stressed for any reason, the internal structure may be damaged and the IR reflectance decreases. This can occur long before the visible green colour of the plant changes. Historically, the mining industry has had success using black and white aerial photography for monitoring the expansion of pit and waste areas, and in planning future development (Mine waste location by satellite imagery, US Dept of Interior). In the case of black and white photography, image interpretation is based on evaluating tone, contrast, and texture. In conjunction with colour infrared products, this type of photography has been used to delineate tailings features and to monitor slope movements (Boldt and Scheibner, 1987). In addition iron staining appears greenish in well exposed colour infrared photos (Boldt and Scheibner, 1987).

One technique that may be useful for erosion monitoring may be to place marker stakes in straight lines along several transects and then use multitemporal photography to detect any movement of these stakes. This technique has been used to detect slope movement (Boldt and Scheibner, 1987) and could be used to monitor the integrity of containment structures. This is not meant to be a very accurate technique quantitatively, but merely to indicate specific sites for potential problems or failures. One problem that has been noted is that when stakes are put in place early in the year, vegetation growth often obscures the markers later in the year (Boldt and Scheider, 1987). -20-

3.1.2.6 Airborne Radar

Synthetic Aperture Radar (SAR) data can be used to determine morphological and topographic characteristics of surface features. The ground resolution of these sensors varies between 5 to 30 m which is sufficient for identifying small scale features. Radar wavelength ranges are identified by letter designations, an artifact of the classified nature of early radar development during World War II. The 8 common bands used are: Ka (0.8-1.1 cm), K (1.1-1.7 cm), Ku (1.7-2.4 cm), X (2.4-3.8 cm), C (3.8-7.5 cm), S (7.5-15 cm), L (15-30 cm), and P (30-100 cm). The amount of radar return is determined by two properties of the terrain, one being the surface geometry and the other the dielectric properties of the surface material.

Radar is an active system, that is, it supplies its own energy and relies on measuring the returned component of this energy to indirectly measure surface phenomena. The electrical field vector of the transmitted energy is normally polarized in either the horizontal or vertical orientations. Upon striking the terrain, some of this energy is depolarized. This characteristic is useful in the interpretation phase. Receiving antennas are then designed to acquire the horizontal or vertical component of the reflected energy. This design setup then gives rise to like-polarized (HH and VV) and cross-polarized (HV and VH) systems.

As far as monitoring waste sites is concerned, radar data can provide information on soil moisture properties and surface morphology. The radar backscatter coefficient responds not only to changes in soil moisture but other factors such as surface roughness and vegetation cover will also affect the amount of backscatter. This should provide the capability to differentiate certain broad cover types such as waste rock, ponded water, bare soil, and vegetation. Benched tailings strips, slumping of outer slopes, and the roughness of unvegetated benches have been identified using this type of imagery (Gregory Geoscience, 1975). In that study, it was found that side-looking airborne (SLAR) was better able to delineate mining features than Landsat MSS data. Often used in geological applications, radar data can reveal geologic structures which are potential traps for ground water seepage (see pages A8 and A12). -21-

3.2 MONITORING EVALUATION CRITERIA

The following section addresses specifically the criteria of importance in monitoring uranium tailings sites. These were obtained from discussions with AECB personnel. The criteria are described under three different categories of vegetation, moisture and soil/rock. Tables 3.2 and 3.3 summarize the ability of the different airborne and satellite-based sensors for providing information on these criteria.

3.2.1 Vegetation

Five vegetation related criteria were identified as being important components of any long-term monitoring program. These included vegetation encroachment, die off, stress and morbidity, coverage and type. Although these phenomena are closely related, each one has unique qualities and can be addressed independently.

3.2.1.1 Encroachment

Vegetation encroachment refers to monitoring the extent of vegetation cover over time. The concern is in identifying either the presence or absence of vegetation. Each state can suggest something about the soil and water conditions of the site.

Sensor requirements for achieving this may be divided into long-term and short- term monitoring. Long-term monitoring involves annual surveys whereas short-term programs would be monitoring changes which occur within a year. In the long- term, Landsat MSS and TM, and SPOT HRV would be adequate. The scale of the expected amount of change should dictate the sensor to use. Landsat MSS provides 60 x 80 metre resolution, Landsat TM 30 x 30 metre and SPOT HRV 20 x 20 metre. Table 3.2: Spaceborne Sensors for Monitoring of Mine Tailings Sensor Landsat TM MSS SPOT PLA MLA Radarsat Soyuzkarta Spatial Resolution 30 m 80 m 10 m 20 m 10-100 m 5 -15 m

Tailing Parameter to be Monitored A. VEGETATION Encroachment 0 -- X Die-Off 0 X Stress and Morbidity 0 I i oox x Percent Coverage/Type 0 oox x X

B. MOISTURE Drainage Pattern 0 0 0 0 0 0 -22 - Ponding 0 0 0 0 0 0 Seepage 0 0 C. SOIL/ROCK Dam Failure 0 O i 0 0 0 0 Surface Erosion 0 0 0 0 0 Sub-Aqueous Erosion Diversion Channel 0 0 0 0 0 Silting Waste Rock/Open Pits X X X 0 X

X = Can be monitored 0 = Monitored only on limited scales or long-term basis -- • Can not be monitored Table 3.3: Airborne Sensors for Monitoring of Nine Tailings

Sensor Photo MEIS, FLI, CASI Thermal IR Radar Spatial Resolution <.3 m 0.5 m <.3 m 4 - 20 m

Tailing Parameter to be Monitored A. VEGETATION Encroachment X X 0 0 Die-Off X X 0 0 Stress and Morbidity X Percent Coverage/Type X

B. MOISTURE to Drainage Pattern X X X 0 Ponding X X X 0 Seepage X X C. SOIL/ROCK Dam Failure X X 0 0 Surface Erosion X X 0 0 Sub-Aqueous Erosion CASI ONLY Diversion Channel X X 0 0 Silting Waste Rock/Open Pits X 0 X

X = Can be monitored 0 = Monitored only on limited scales on long-term basis -- = Cannot be monitored -24-

If the scale of change is expected to be less than 60 metres in extent, Landsat MSS would not have sufficient spatial resolution for monitoring encroachment. Landsat TM would be the most appropriate satellite-based sensor for long-term monitoring. The relatively high spatial resolution combined with the high spectral resolution of the six bands would provide clear discrimination of the extent of the vegetation encroachments. The optimum TM bands for this monitoring would be bands 3, 4 and 5. These provide the most information in vegetated terrains due to the sensitivity of the red/infrared portions of the spectrum to vegetation characteristics. If Landsat MSS were used a combination of bands 4, 5 and 7 would be appropriate. When displayed through blue, green and red respectively, vegetation would appear red due to the high reflectivity in the infrared portion of the spectrum.

In the short-term, airborne sensors would be required since the amount of encroachment would be less than the satellite sensors spatial resolution. A spatial resolution of less than 5 metres would be required. Based on this criteria, either the MEIS or CASI sensors would be appropriate. The spatial resolution of up to 0.5 metres combined with the ability to select bands in red and reflected infrared portions of the spectrum would make these sensors ideal for monitoring very small changes in vegetation encroachment.

3.2.1.2 Die Off

Vegetation die-off is concerned with the health of the plants. It may be considered a localized phenomena, which suggests the use of an airborne system. If in the long-term relatively large areas can be expected to show die off then the same sensor recommendations as related to encroachment monitoring can be made. Either Landsat MSS, TM or SPOT MLA could be used depending upon the extent of the die off.

Airborne sensors would provide better discrimination of vegetation die-off particularly in the short-term. Use of MEIS or CASI systems employing bands in the green, red and infrared portions of the spectrum would be optimal for monitoring vegetation die-off. -25-

3.2.1.3 Stress and Morbidity

Stress and morbidity is concerned with plant vigour, similar to vegetation die off, with the difference being the level of severity. As far as detection from satellites or aircraft, vegetation stress is less apparent than die off. The importance of detecting stressed vegetation is that it may be indicative of leaks in tailings ponds or soil moisture and chemistry abnormalities. Stress and morbidity also is more localized than die-off. As a result satellite sensors are not appropriate for monitoring this criteria.

One condition that often occurs near or on tailings sites and results in vegetation stress is termed dusting. This occurs when the tailings surface undergoes significant wind erosion because it has not been stabilized by the presence of vegetation. Also, radium bearing dust can be scattered by winds from unvegetated, dry surfaces. This dusting should be minimized to prevent the transport of radioactive, and metal, particles. One solution may be to attempt reclamation by covering the tailings with a layer of top soil, preparing the surface, and then applying grass seed or planting grains. Another condition is iron staining which occurs when vegetation is flooded by water having abnormally high amounts of iron.

The MEIS and CASI airborne sensors would have sufficient spectral and spatial resolution for monitoring stress and morbidity. Bands selected along the red- edge of the vegetation reflectance curve would provide the ability to identify vegetation under stress.

3.2.1.4 Coverage/Type

Coverage/type refers to monitoring the extent of, and differences between, several species of vegetation. Satellite based sensors would only be appropriate for long-term monitoring of coverage and type. Landsat TM is best suited of all the satellite sensors due to its high spectral resolution which facilitates the discrimination of specific plant species. SPOT MLA data may also be useful because of its higher spatial detail, but the spectral capabilities of TM are superior. -26-

For short term monitoring or monitoring small spatial changes the CASI and/or MEIS sensors would be most appropriate. Although aerial photography would also be sufficient for area estimation, it does not possess multispectral capabilities. These sensors have the desirable qualities of both high spatial resolution and wide, selectable, spectral coverage. Processing and analysis of the data could involve classification of the data. This grouping of the data into classes of vegetation types would provide the ability to easily calculate area coverage of each class.

3.2.2 Moisture

There are three monitoring criteria which fall under the moisture category. These are drainage, ponding and seepage. For all of these, satellite sensors can provide information, but only at a relatively coarse scale due to the limitations of the spatial resolution.

3.2.2.1 Drainage

For both satellite and airborne sensors a single infrared band may be sufficient for identifying drainage patterns. Water absorbs reflected infrared wavelengths. Therefore, if a drainage network consists of a number of channels with standing or running water, these can be readily identified by isolating the low reflectance levels in an infrared band such as Landsat TM band 5 or SPOT band 3.

For higher spatial resolution, thermal infrared scanners are most appropriate, although the multispectral systems could also be sufficient. The difference in temperature between water and land, particularly at night time, is easily identified. Stereo coverage, such as is obtained from aerial photographs, can aid in the delineation of drainage patterns if surface water is not extensive. The three dimensional aspect provides the interpreter the ability to trace drainage patterns on the basis of channel morphology. -27-

3.2.2.2 Ponding

Ponding has the same sensor requirements as drainage monitoring although stereo viewing would not be a requirement. Either reflected or thermal infrared data would provide the ability to monitor this criteria. Single band data would be sufficient and little processing of the data would be required.

3.2.2.3 Seepage

A certain amount or seepage could be permitted from a tailings pond. This, of course, would depend on the location and amount involved. What would be of concern is large amounts, or several leaks in close proximity, which could indicate potential containment problems. The quality of the seepage is also important.

Moisture seepage or variable amounts of moisture in surface sediments should be monitored using thermal infrared data. Seepages are often not readily identifiable even using the reflected infrared portions of the spectrum. Thermal infrared will highlight areas of seepage due to the temperature differences associated with different water contents.

The problem of the quality of seepage is a difficult one given that, in most cases, the quantity of water involved would be small. In the case of larger leaks, techniques sensitive to UV wavelengths could be used to detect petroleum contamination.

3.2.3 Soil and Rock

The soil/rock category deals with a number of criteria associated with tailings structures. -28-

3.2.3.1 Dam Failure

The potential for dam failure may first be detected using thermal infrared data to detect seepage indicating a weakness or failure in the structure. Dam failure occurs when the structure itself is observed to have failed. If the monitoring program is successful, dam failure should not occur. A failure of this nature could be observed from any of the satellite sensors reviewed in Section 3.1 Targets placed on dam slopes would provide a means of monitoring the movement of materials.

In general, the portions of the spectrum best used would be the shorter visible wavelengths due to the greater dynamic reflectance characteristics of non- vegetated surfaces at these wavelengths. Spot panchromatic with its 10 metre resolution would be a good data set for monitoring tailings site dams.

Aerial photography with stereo overlap has the advantage of stereo viewing, high spatial resolution and sensing in the visible portions of the spectrum. The technique described in Section 3.1.2.5 concerning the use of stakes to monitor ground displacement may be appropriate for identifying potential dam failure sites.

3.2.3.2 Surface Erosion

Assuming a non-vegetated surface, stereo aerial photographs would be a good sensor for monitoring surface erosion. SPOT panchromatic data would be an appropriate satellite product for slightly lower resolution monitoring.

3.2.3.3 Sub-aqueous Erosion

Sub-aqueous erosion refers to drainage within sediments and sediment ooze. This type of monitoring would require an airborne electro-optical scanner in the visible and infrared portions of the spectrum. Airborne SAR data may also be effective since some surface penetration is possible depending on the water content in the sediments. The amount of surface penetration would be less than -29-

one wavelength. Aerial photography, and thermal infrared data would not be appropriate for monitoring this criteria.

3.2.3.4 Diversion Channel Silting/Erosion

The monitoring of diversion channels for silting and erosion can be accomplished using satellite provided the channel is of sufficient size. Relative amounts of silt in water can be detected using shorter wavelength bands such as Landsat TM bands 1, 2 and 3. The use of an airborne scanner such as CASI with filters in similar wavelength regions would enable the monitoring of smaller channels.

3.2.3.5 Haste Rock/Open Pits

Waste rock pits and open pits tend to be of a size where the use of_satellite data would be sufficient. SPOT panchromatic, Landsat TM, or Soyuzkarta data could be used for monitoring these features Landsat MSS could be used although the spatial resolution of 60 x 80 metres is a limiting factor.

3.2.4 Radioactivity

As uranium is a radioactive element it is possible to use airborne geophysical sensors to measure the parts per million (PPM) level of the uranium channel. Following a mine decommissioning and capping the PPM level detected from the air should be noticeably lower than when the mine is active. Consequently, if a leak or contamination occurs a corresponding increase in the PPM level may occur. Geophysical instruments are calibrated, so samples from one year to the next can be directly compared. Changes in the order of 1 PPM can, therefore, be detected. Although it should be noted that the systems are affected by the presence of surface moisture, so other methods involving ground truth would be Vequired. Furthermore, it may be possible to compare the PPM level of the post-operation mine with the pre-operation PPM level.

A uranium detector operating from a flight altitude of 300-500 feet will have an absolute sensitivity of 1/2 PPM gamma over a swath width of 300-500 feet. -30-

The geophysical systems provide useful input although they are not generally considered a remote sensing instrument. In addition, they are unlikely to detect minor changes affecting the morbidity and stress in the vegetation. For these reasons the system can be considered as a secondary sensor to an airborne data acquisition system. That is they may be used in conjunction with other sensors but not alone. Therefore, we have not considered this technology further in this report.

3.3 MONITORING PLAN DESIGN

The design of a monitoring program for uranium mine tailings may be broken into two categories, long-term and short-term. Long-term monitoring, annual programs being the most frequent, would be concerned with monitoring gross changes suitable for satellite based sensors. Short-term monitoring of less than a year, possible on a monthly basis would require higher spatial resolution to detect smaller changes. The nature of the individual phenomena to be monitored may dictate the types of system to be used.

Stress and morbidity for example would require high spectral and spatial resolution and therefore an airborne multispectral scanner system would be most appropriate. It is essential to assign levels of importance to the different criteria in order to design the most appropriate monitoring program. Some flexibility needs to be maintained in the program to accommodate criteria which may require short term monitoring. With these factors in mind, the following system discusses a possible monitoring program design.

3.3.1 Long-Term Monitoring

A long-term monitoring program would require data to be collected on an annual basis. Once a base line has been established satellite-based data should provide enough information to monitor most criteria. Flexibility should be monitored so that if a criteria which requires an airborne system is needed, ?uch as in monitoring stress in vegetation, then this could be introduced. -31-

An airborne program each year for up to five years would be recommended for establishing the base line for each of the criteria of concern. This would also require field work to coincide with the flights.

3.3.1.1 Data Requirements

Landsat TM and SPOT PLA data collected during the summer months would provide the optimal data set. The TM data would provide information on vegetation and moisture conditions while the SPOT PLA data would provide more detailed information on soil/rock criteria. If a single data set were required due to cost considerations the Landsat TM data should be used. This would provide information on all the criteria to varying degrees depending on site conditions. The shorter wavelength bands could be enhanced for soil/rock criteria while the longer wavelengths into the reflected-infrared region would be optimal for monitoring the vegetation and moisture criteria.

For airborne programs, either the CASI or MEIS systems are recommended. These systems provide high spectral and spatial resolution from which a good database may be established. If seepage is a major concern then a thermal infrared survey would be recommended. This type of survey may be worth implementing at regular intervals beyond the initial five year detailed work.

3.3.1.2 Data Processing

Figure 3.1 illustrates the steps involved in processing the digital data and incorporating the extracted information into a GIS database.

Input of the data into the system would be from computer compatible tape. Geometric correction is important in order that information extracted from the data may be transferred to the appropriate spectral database. -32-

1. Data Input Image Analysis System 2. Geometric Correction

3. Enhancement

4a. Classification 4b. Interpretation

5. Input Classified or Interpreted Polygon Geographic Information System 6. Input Associated Attribute Information

7. Compare to Database for Change Detection

Figure 3.1: Methodological Approach to Data Processing -33-

There are a variety of image enhancements which may be applied to the data. The best results would likely result from contrast stretching of three bands of data then displaying these as a colour composite. TM bands 3, 4 and 5 would be optimal for vegetation and moisture criteria. TH bands 1, 2, 3 and 2, 3, 7 would be more appropriate for soil/rock criteria.

Enhancement of either MEIS or CASI data would employ the same techniques. Reflected infrared bands for vegetation and shorter wavelengths for soil/rock criteria would be used. Some experimentation would be required on a demonstration image to determine optimal band combinations for isolating particular tailings criteria.

Results of visual interpretation of the enhanced imagery may be stored in a GIS database. Alternatively, digital classification of the remote sensing data into groups representing different criteria would enable direct transfer of digital polygon information into a GIS environment.

Once the tailings information is brought into a GIS, then comparisons may be made from one dataset to the next. During the first few years of a monitoring program it would be important to establish a comprehensive database in the GIS into which the long-term monitoring results could be input and compared. Information relating to different features would be assigned as attributes. Field information, for example, could be input into the database and catalogued for later access. Since all of this information would be spatially referenced in a GIS, simple overlay operations could be used to compare changes from one year to the next.

3.3.2 Short-Term Monitoring

At different times throughout the long-term monitoring program it may be necessary to monitor specific criteria at frequent intervals over a relatively short period of time. For example, if beavers were damming diversions channels or seepage is suspected in an area, several airborne flights may be required in a year. Once the problem is rectified the monitoring program could revert back to the long-term schedule. -34-

3.4 INFORMATION EXTRACTION AND STORAGE

The following section briefly reviews the findings of previous work using image processing and geographic information systems.

3.4.1 Image Processing

Standard image analysis techniques would be useful in any analysis effort. Several studies (Boldt and Scheider, 1987; All urn and Droisinger, 1986) have demonstrated the utility of contrast stretches, colour and edge enhancements, band ratioing, and band combinations for monitoring programs. Even simple band differencing could be used to perform change detection.

Visual interpretation of imagery has been found to be the best method of delineating waste areas (Moore et al., 1977). This is usually performed after some digital enhancement has been performed. Enhancements are used to accentuate selected features of interest while diminishing the appearance of irrelevant information. Edge enhancements are employed for detection of topographic or linear features. The physical and spectral contrasts between the tailings and the natural surroundings, the resolution limitations of the imagery, and the availability and quality of data all contribute to determining the success of this technique.

In addition, one can create classes based on grouping of spectral responses which could be used as a reference in comparison to multi-year data samples.

3.4.2 Geographic Information Systems

An often overlooked aspect of many long-term monitoring efforts is the effective storage and manipulation of the data accumulated over a number of years. Recently, the development of Geographic Information Systems (GIS) has created an extremely useful tool for data storage and manipulation. A GIS combines the ability to maintain graphic information, as in a digital mapping system, and attribute information, as in a database management system. The advantage is that -35- each spatially referenced graphic element can have associated attribute information which can be recalled or manipulated as needed.

The use of a GIS for mine waste monitoring applications has been demonstrated by (Jensen and Christensen, 1986). In the case of the current application, a GIS could be used to provide the initial base map for each study site. Firstly, the common physical and man-made features (roads, contours, water bodies) would be digitized from existing base maps. To these base maps, the boundaries of tailings ponds, vegetation, and other phenomena related to the waste sites would be added from interpretation of the remotely sensed data. With each repetition of a data acquisition phase, a new monitoring layer would be produced.

This type of monitoring could be accomplished with a relatively low cost, PC based system having both image analysis and GIS capabilities. -36-

4. ECONOMIC FEASIBILITY

This section reviews the costs associated with the collection and processing of remotely sensed data to allow analysis of mine tailing areas. Each of the sensors described in Section 3 above are listed along with the cost of acquisition, processing and interpretation. By combining this cost analysis with the technical suitability, one can derive the economic feasibility of each sensor to perform the monitoring. Where appropriate labour charges are based on $60 per hour and computer system charges at $75 per hour

4.1 SATELLITE BASED MONITORING

Data from environmental satellites is readily available over all portions of the globe. Coverage from Landsat, Radarsat or SPOT can be purchased from Canadian ground stations in either Prince Albert, Saskatchewan (PASS) or near Ottawa (Gatineau Station). Coverage from the Soviet satellites is available through a U.S. distributor.

Computation of the charges for monitoring the mine sites is relatively straight- forward and independent of the location of the mine. We have broken the cost analysis into several distinct steps to allow the processing chain shown in Figure 4.1.

Receipt of Image Processing Hard Copy or Human Computer Tape Computer Soft Copy Output Interpretation

Figure 4.1: Satellite Data Monitoring Steps -37-

4.1.1 Data Acquisition

Table 4.1 below lists the cost of data acquisition for the satellite coverage of a typical sized area from which analysis could be performed. All prices are from suppliers published lists attached as Appendix B and are current at time of writing. All products are digital unless otherwise stated. Table 4.1 Satellite Data Acquisition Prices

Sensor Features Price Comments LANDSAT TM 7-band precision $1,725 4,000 km2 corrected., geocoded SPOT MLA Precision corrected, $1,675 1/4 scene geocoded SPOT PLA Precision corrected, $1,970 3,600 km2 geocoded Soyuzkarta MK4 Analogue $4,890 Film negative Soyuzkarta KFA-1000 Analogue $1,100 Film negative Radarsat Black & White N.A. No published prices but expected to be $1,500 - $l,800/scene

4.1.2 Image Processing

To fully exploit the information contained in the digital satellite data it must be processed into an image format for subsequent analysis by computer. The cost of processing the various data sets will vary slightly depending on the particular sensor and the level of digital enhancement desired. However, the variations in the processing time and resulting costs will be relatively small and for the purposes of this study can be considered equal for all digital sensors and estimated at less than one day of labour and computer time per scene. The above estimate is based on the assumption that the user is working from an established set of guidelines for enhancing the data and not performing an experimental study. -38-

Sensors which are not available in digital format such as the Soyuzkarta data will not have any associated processing charges nor the advantages of custom computer enhancement methods as discussed in the technical analyses above. For comparison purposes only, a typical charge for an image analyst and computer system have been applied to the time estimates. The costs of processing are summarized below in Table 4.2. Table 4.2 Satellite Data Processing Costs Labour Computer Approx. Approx. Rate Rate Time Cost Sensor (S/hr) ($/hr) (Hrs) ($)

Landsat MSS 60 75 8 1080 Landsat TM 60 75 8 1080 SPOT MLA 60 75 8 1080 SPOT PLA 60 75 8 1080 Soyuzkarta MK4 60 -- 0 0 Soyuzkarta KFA-1OOO 60 -- 0 0 Radarsat 60 75 8 1080

4.1.3 Data Output

Following the processing of the data to an image format, using a computer where applicable, the data is normally output to a digital film writer. These systems are large format and relatively expensive. The remainder of this section deals only with the cost of producing data products once access to a system has been obtained.

The purpose of an output product is to allow interpretation and to provide historical documentation for comparison with future images. Both a hardcopy (photographic) product and a soft copy (computer tape) product will meet the objective, although much of the interpretation will be performed on a hardcopy -39-

product. The analyst will eventually want to have both forms available, that is, enhanced hard copy image(s) and a computer tape (CCT) of the processed data. Table 4.3 summarizes the cost of each of the sensor types and shows the cost of producing a single negative, one photographic enlargement and one CCT where applicable. Table 4.3 Satellite Output Product Costs CCT Negative Positive Total Production Production Print Output Cost Sensor ($) ($) ($) ($)

Landsat MSS 150 130 200 480 Landsat TM 150 130 200 480 SPOT MLA 150 130 200 480 SPOT PLA 150 65 100 315 Soyuzkarta MK4 N.A. Incl. 200 200 Soyuzkarta KFA-1000 N.A. Incl. 200 200 Radarsat 150 65 100 315

4.1.4 Interpretation

a) Human Interpretation

Each output image can be interpreted in a relatively short period of time. Approximately four hours should be sufficient. A comparison between the various sensors is somewhat arbitrary as the multi-spectral digital data can provide many different images from the same data set, each providing additional information. These factors are examined further in the technical feasibility and recommendations below. The computer system will be used to assist the human interpreter through an interactive process. In this manner, we consider the analysis to be done by humans as the operator will be making all decisions and instructing the computer on a step-by-step basis. For the purposes of this study -40- a charge of $60 per hour was used for an image interpreter. Based on the above estimate of four hours, the interpretation cost is $240 per image for human analysis.

b) Expert Systems

It is also possible to have the digital soft copy images interpreted by computer software using "Expert System" techniques. This involves the development of a standard interpretation process and decision making rules. These techniques show great promise for situations where an analysis is done very frequently, the answer is required quickly or the interpretation procedure is very difficult and a trained expert is not available. The monitoring of mine tailings using remote sensing techniques is embryonic and is expected to be a relatively infrequent process. Consequently, we do not recommend that expert system analysis or artificial intelligence be undertaken at this time. As experience is developed and the state-of-the-art improves, a re-examination of these interpretation techniques may be appropriate.

4.1.5 Integration

One of the advantages of digital data sets is their ability to be easily merged with other digital data sets. The costs of doing a digital merge is rather wide open due to the inherent flexibility and number of combinations available. However, for simplicity, the cost of merging two digital data sets together can be computed based upon an estimate of 12 hours of time for both an operator and a computer analysis system. At the rates used in previous examples the processing charges would be $720 labour plus $900 computer totalling $1,620 for processing.

Following the integration, the charges identified above for a colour image output and a CCT output, plus the interpretation would need to be applied. In summary, for a two data set merge the following costs should be considered for the example of Landsat TM plus SPOT PLA. -41-

Acquisition of data set one (TM) $ 1,725 Acquisition of data set two (SPOT) $ 1,970 Digital processing (integration) $ 1,620 Data output (CCT, neg, print) $ 480 Interpretation S 240 TOTAL COST $ 6,035

4.2 AIRBORNE BASED MONITORING

Airborne based monitoring charges will have many similar costs to that of satellite based monitoring except for one major area. The location of the mine will have a major impact on the total cost. Ferrying charges, or getting the sensor to the mine site, will be highly dependent on the remoteness of the mine and the base location of the aircraft.

To analyze these costs, we again look at the individual steps to process and interpret the data as illustrated in Figure 4.2.

Airborne Data Data Hard Copy or Human Acquisition Processing Soft Copy Output Interpretation

Figure 4.2: Airborne Data Monitoring Steps

4.2.1 Data Acquisition

The data acquisition phase involves three separate activities to make the data available; mission planning, ferrying and data collection.

A typical mission can be planned in a period of less than two hours using modern computer based packages. As these systems run on readily available PCs only a labour charge is considered totalling $120 per mission. -42-

As mentioned above the ferrying costs are highly dependent on the location of the mine and the base of operation for the sensor system. Most airborne systems operate from a typical aircraft suitable for aerial survey. For the purposes of this analysis a cost of $700/hour has been utilized for cameras, CASI, and thermal infrared. This is a wet lease charge including both aircraft and fuel.

Data collection charges are slightly higher than the ferrying costs as they also include sensor on time and consumables.

A summary of the data acquisition costs for the airborne sensors listed in Section 3.3 above is included in Table 4.4 below.

Table 4.4 Airborne Data Acquisition Prices Sensor Number Crew Ferrying Charges of Charges Sensor Aircraft Type ($/hour) ($/area) Crew ($)

Camera Survey 700 1,000 3 1950 MEIS Falcon Jet 3,60° 4,000 3 1950 CASI Survey 700 4,000 3 1950 Thermal IR Survey 700 4,000 3 1950 Radar Turbo-Jet 2,500 2,000 3/4 2600

Additional charges may be incurred by personnel and expenses as a typical mission would include a minimum crew of three persons for a full day - a pilot, a navigator/mission manager and an equipment operator at $650/day each.

For example, the cost of a one-day mission with the CASI sensor in a light aircraft is computed as follows: -43-

Aircraft $700 X 10 hours $ 7,000 Sensor Charge $ 4,000 Crew $650 X 3 crew $ 1,950

TOTAL DATA ACQUISITION CHARGE $12,950

4.2.2 Image Processing

The cost of processing the airborne data will be comparable to the satellite data sets for the digital sensors as a very similar process is required. Analogue products such as a camera's photography are significantly less. The following Table 4.5 compares the relative costs:

Table 4.S Ai rborne Data Processing Costs Labour Computer Approx. Approx. Rate Rate Time Cost Sensor Data Type $/hr $/hr (hrs) ($)

Camera Analogue 60 -- 2 120 HEIS Digital 60 75 8 1,080 CAS I Digital 60 75 8 1,080 Thermal IR Digital 60 75 8 1,080 Radar Digital 60 75 8 1,080

4.2.3 Data Output

Again the cost of producing output products for airborne data is directly comparable to that of satellite data and is summarized in table 4.6 below. -44-

Table 4.6 Airborne Data Output Costs CCT Negative Positive Total Output Production Production Print Cost Sensor ($) ($) ($) ($)

Camera N.A. Incl. 200 200 MEIS 150 130 200 480 CAS I 150 130 200 480 Thermal IR 150 65 100 315 Radar 150 65 100 315

4.2.4 Interpretation

Interpretation time has been estimated at four hours for each of the images. At the rate of $60/hour a cost estimate of $240 per image is estimated using manual techniques.

4.2.5 Integration

Integration of airborne digital data sets is also very similar to that of digital spaceborne data as identified in Section 4.1.5 above. Two digital data sets can be merged in 12 hours using an image analysis system. At the rate of $60 per hour for labour, and $75 per hour computer time a total charge of $1,620 exclusive of data acquisition, output or interpretation charges.

In summary, for an airborne survey over a site such as Elliot Lake, the following costs would be considered an example:

Data Acquisition {CASI, MEIS) 16,000 Data Processing 1,080 Data Output (2 colour images) 960 Interpretation (2 images) 880 Total Estimated Cost 18,920 -45-

It should be noted these costs, in particular the data processing costs, are calculated based on a production situation once methods and paths have been clearly identified.

4.3 COMPUTER PROCESSING FACILITIES

There are two main cost components to an image processing and geographic information system facility which are hardware and software. The hardware components include the computer as well as various input, output and display devices. A personal computer (PC) based system would provide sufficient computing power for the mine tailings monitoring.

The software component consists of a series of packages which perform the image processing and GIS functions. With the software it is important that the image processing software and GIS software have the means to transfer data from one environment to another.

4.3.1 Hardware

The following itemized breakdown of hardware would provide the functionality necessary for a monitoring program.

Approx. Value A. System Hardware to Purchase Personal Computer (IBM Compatible), 25 MHz o 80386 CPU o 80387 Co-processor o 3 mb total memory o 300 mb, drive o 2 parallel ports o 2 serial ports o 1.2 mb floppy drive o Monochrome monitor and card o DOS 4.X $ 11,000

B. Image Display 9,000 -46-

C. Data Input Tape Drive 1600/6250 bpi dual density 15,600 Digitizing table 9,700 D. Hardcopy Output Dot matrix printer 1,050 Colour Plotter 4,250 E. Consumables (for one year) 800 TOTAL ESTIMATED HARDWARE COST $ 51,400 Please note that the above hardware cost does not include the cost of a digital laser film writer. These systems can cost between $100,000 and $500,000 and sometimes more. Then one needs colour darkroom facilities as well. It is our recommendation that such film writers not be purchased but that the data be transferred via computer tape to a service bureau.

4.3.2 Software

There are a number of software packages on the market which provide similar image processing and GIS functionality. Each package tends to have its advantages as well as some disadvantages. The cost estimate provided is based on two software packages, one for image analysis and one for GIS. These have proven to be compatible and very flexible.

A. Image Analysis Software Multispectral Analysis $ 5,200 Geometric Correction 2,800 Vector Edit 1,400 Vector Digitization 1,400 Hardcopy Plotting 2,500 Tape 1/0 2,500 GIS Link 1.800 $ 17,600 -47-

B. GIS Software Mapping 7,500 Analysis 7,500 Image Analysis Link 4,000 Display 2,000 Terrain Analysis 5,000 Database 825 $ 26,825 TOTAL ESTIMATED SOFTWARE COST $ 44,425 TOTAL ESTIMATED SYSTEM COST $ 95,825

4.3.3 Owned Facilities Vs Service Bureaus

Access to the computer processing facilities identified in the above sections may be obtained in two different methods, either by purchasing the system for installation at AECB or by contracting the processing to the private sector. Each of the methods has distinct advantages and disadvantages.

From our understanding of the objectives, we believe it is not cost effective to purchase the required facilities to perform the monitoring of the tailing sites. Changes and hence the monitoring will occur at relatively infrequent periods. Consequently a system obtained for this purpose alone would be idle for much of the time. The state-of-the-art in both hardware and software systems changes at a relatively fast pace such that systems are often out of date within three years.

With a purchased system, the cost savings obtained by a lower operating cost will be minimal compared to the depreciation costs of the equipment. A discounted cash flow analysis would show a payout period of 10-15 years or more for the equipment which is excessive for high technology areas.

Table 4.7 lists the advantages and disadvantages of both owned and contracted services. -48-

Table 4.7 Owned Facilities versus Contracted Processing Services

OWNED FACILITIES Advantages

1. Direct control over analysis steps 2. Database may be used for other purposes 3. Equipment may be used for other purposes 4. Maintains confidentiality of data 5. Continuity of Service and Availability

Disadvantages

1. High purchase price 2. Requires trained equipment operators 3. Systems become obsolete quickly 4. Maintenance Costs 5. Space and Cooling Requirements

CONTRACTED PROCESSING SERVICES

Advantages

1. Access to state-of-the-art equipment 2. Flexibility to incorporate new methodology 3. No equipment purchase required 4. Trained specialists in remote sensing

Disadvantages

1. Higher processing costs 2. Do not have total control over each step 3. Risk of Disruption of Service -49-

5. SATELLITE DEMONSTRATION

A satellite demonstration study was conducted as part of an extension to the technical feasibility study. The objectives of this study were:

1) To investigate the ability to enhance and enlarge an area of interest using digital methods rather than photographic processes;

2) To identify and evaluate optimum digital enhancement techniques for extracting information on uranium tailings sites.

5.1 STUDY AREA

The study area chosen was the Rabbit Lake mine site in Saskatchewan (Figure 5.1). This site includes an open pit, waste stockpiles, waste rock areas, ore stockpile areas, tailing disposal area, tailings pond and a precipitation pond. These features are spread over approximately a 16 kilometres by 10 kilometres area.

5.2 DATA PROCESSING METHODOLOGY

A Landsat Thematic Mapper image was acquired from the Prince Albert Satellite Station in Saskatchewan. A single quadrant of data was acquired, providing more than adequate coverage of the Rabbit Lake mine area (Figure 5.2). The following are the scene specifications:

Date Acquired: August 2, 1986 Quadrant: 4 Number of Bands: 7 Path: 38 Row: 19 Rad: CAL 2 LIN Geo.: SYS-GEOREF CCT No.: M07361 -50-

Rabbit Lake

MINH SITE PLAN

Figure 5.1 -51-

Figure 5.2: TM Bands 3, 4, 5 colour composite of entire quadrant, Mine site is located in the centre of the image. -52-

Abundant cloud cover in the region prevented a more recent image from being acquired. Figure 5.3 illustrates the data processing methodology used in the demonstration study.

5.2.1 Data Input

The data were read onto the INTERA MAGIC Digital Image Analysis System. This system is designed to analyze and process remotely sensed digital images. The system is based on a MicroVax II mini computer with 16 megabytes of main memory. Images are displayed on a Metheus 3610 monitor with a resolution of 1280 by 1024 pixels. The software is an enhanced version of the Landsat digital image analysis system (LDIAS) originally developed at the Canada Centre for Remote Sensing.

The data, which was written on a single 6250 bpi computer compatible tape (CCT), was read onto the system disk storage from where the data could be more readily accessed and manipulated.

5.2.2 Geometric Correction and Resampling

Due to geometric distortions within the data, primarily from the atitude of the spacecraft and earth curvature and rotation, a geometric correction procedure was conducted in order to correct the geometry of the data to a UTM map base. This procedure involves identifying point locations on the imagery, and matching these with corresponding points on the topographic map. Once a sufficient number of randomly spaced points are collected over the image, a transformation is calculated between the image and map points from which the image may be resampled to match the geometry of the map base. A cubic convolution resampling algorithm was used for this study.

The resampled and corrected image was output to 30, 20 and 10 metre pixel sizes in order to investigate the first objective of the study. DATA INPUT

~n tQ c GEOMETRIC (D in CORRECTION OJ

o a>

-o RESAMPLED RESAMPLED RESAMPLED -s o o TO30M TO20M TO10M ess i 6 BANDS 6 BANDS 6 BANDS

3 IQ eth o

a. LINEAR LINEAR LINEAR LINEAR LINEAR o PRINCIPAL PRMOPAL PRINCIPAL —-* CONTRAST CONTRAST CONTRAST CONTRAST CONTRAST o COMPONENT COMPONENT COMPONENT fflHEICti Slkti'ui SIHblHi STRETCH STRETCH ENHANCEMENT ENHANCEMENT FNHANCFMFNT BANDS 3,45 BANDS 1 AS BANDS 3,4,5 BANDS 3,4^ BANDS 1 AS

BAND COVARIANCE CORRELATION DIFFERENCE COVARIANCE CORRELATION COVARIANCE CORRELATION ENHANCEMENT

Figure 5.3: Data Processing Methodology -54-

5.2.3 Linear Contrast Stretch Enhancement

Following an initial viewing of each of the bands, two contrast stretch enhancements were conducted on the data. The first on bands 3, 4, 5 to maximize the ability to interpret information pertaining to vegetation, and the second on bands 1, 3, 5 to maximize the ability to interpret information from the disturbed areas. Figures 5.4 (a to g) are histograms of the original data. These histograms plot intensity against pixel frequency. It is readily apparent that the data is distributed over a narrow range of intensity levels. In order to maximize contrast within the image, the data is remapped over the full dynamic range of 256 intensity levels.

The band 3, 4, 5 enhanced image was displayed with band 3 in blue, band 4 in green and band 5 in red (figures 5.5, 5.6 and 5.7).

In the case of the band 1, 3, 5 enhancement, only the disturbed areas were enhanced. Since the data for these areas were the higher intensity levels, they could be isolated by setting any other data of lower intensity (including vegetated areas) to 0 (Figure 5.8).

5.2.4 Linear Stretch Difference Enhancement

New image channels were created by calculating the difference between bands 1 and 3, 1 and 5, and 3 and 5. This was done to bring out subtle differences between the bands. These differences were then stretched over the 0 to 255 digital range and displayed as a colour composite (Figure 5.9).

5.2.5 Principal Component Enhancement

A standard principal component transformation was applied to the 6 channel imagery based on the scene correlation table. The first output channel consists of most of the variance within the imagery with channels 2 and three containing less and less of the total scene variance (Figure 5.10). -55-

HISTOGRRM COUNT (a) HISTOGRRM COUNT (b)

2

I I i—i—i—T—i—i—i—I—i—i INTENSITY ( 1 INTENSITY ( 0 TO 255 )

HISTOGRRM COUNT HISTOGRRM COUNT (c) (d) 3

INTENSITY (' o'TO 2*55') ' ' ' i ' ' ' I INTENSITY (' O'TO zfes' ) '

Figure 5.4: Histograms of TM Bands l(a), 2(b), 3(c), 4(d) -56-

HISTOGRHM COUNT HISTOGRAM COUNT (e) (f)

INTENSITY! C TO I INTENSITY ( 0 TO Z55 J

HISTOGRRM COUNT (9)

7

1 INTENSITY (' O'TO 2^55') ' ' '

Figure 5.4: Histograms of TM Bands 5(c), 6(f), 7(g) -57-

Figure 5.5: Landsat TM 30 metre colour composite image, bands 3, 4, 5

Figure 5.6: Land'^t TM 20 metre colour composite image, bands 3, 4, s Figure 5.7: Landsat TM 10 metre colour composite image, bands 3, 4, 5 -59-

Figure 5.8: Landsat TM 10 metre colour composite image, bands 1, 3, 5 Figure 5.9: Landsat TM 10 metre band difference enhancement -61-

Figure 5.10: Landsat TM 10 metre Principal Component Enhancement -62-

Although it would be possible to produce many more different enhancements by changing input and output parameters, the enhanced images produced were considered optimum for the purposes of the tailings monitoring. All the techniques used are standard image processing techniques, and the bands selected were identified based on their information content.

5.3 IMAGE INTERPRETATION AND EVALUATION

The images were evaluated based on visual interpretation made on each image. The interpretations were limited due to the lack of field verification in this study. The interpretations were, therefore, primarily based on the experience of the interpreter and the knowledge of how different phenomena on the earth's surface reflect different wavelengths of energy.

5.3.1 Spatial Resolution

The ability to discriminate detail was enhanced through resampling of the data to 20 and 10 metre resolutions. At 30 metres the pixels appear "blocky" when output at a scale of 1:50,000. At 10 metres edges appear blurred although the data may still be readily interpreted. At 20 metres the block appearance is reduced without the image becoming blurred. On this basis, the 20 metre data is the optimum pixel size at a scale of 1:50,000.

5.3.2 Vegetation/Moisture Enhancement

TM bands 3, 4, 5 were enhanced using a linear contrast stretch methodology in order to maximize the interpretability of the terrain for vegetation differences. This is readily apparent in the areas surrounding the disturbed mine workings (Figure 5.7). There is, however, evidence of vegetation growth within the tailings areas. At the B-zone open bit and the adjacent ore stockpile and waste stockpile areas, red, pink and orange coloured pixels suggest vegetation growth. High reflectance in the infrared wavelengths is characteristic of healthy vegetation. This can also be seen in the centrally located waste rock and ore stockpile area. The tailings and precipitation pond adjacent to Parks Lake is surrounded by vegetation, these likely being grasses. On this basis it is -63-

apparent that vegetation encroachment and die-off could be monitored in the vicinity of the mine tailings area. In the band 3, 4, 5 enhance image stressed vegetation will appear dark green (Figure 5.7). This is displayed in the burn area which crosses the central portion of the study area. Outcrops or non- vegetated ground appears purple. The area immediately surrounding the B-zone stockpile area and the tailings pond can be seen to be quite different. As a result of either environmental or man-induced changes the terrain surrounding the stockpile area lacks healthy vegetation.

With bands 3, 4, 5 very moist areas will appear dark on the image (Figure 5.7). Standing water appears black due to absorption of energy in the infrared wavelengths. Drainage patterns and ponding may be easily interpreted from the image. Two ponds located between the B-zone open pit and waste stockpile area are not contained and may represent an environmental hazard. Sedimentation in water appears as a lighter blue colour within the ponds. This is displayed in the tailing and precipitation pond. Fluid also appears to be flowing from the precipitation pond into the local drainage network to the soutwest.

5.3.3 Enhancement of Tailings Area

Three different enhancements were applied to the tailings and disturbed areas within the image. These included the linear contrast stretch of bands 1, 3 and 5 (Figure 5.8), the band difference enhancement (1-5, 3-5, 1-3) (Figure 5.9) and the principal component enhancement (Figure 5.10). Each of these enhancements were optimized for the tailings area data.

The band 1, 3, 5 contrast stretched image (Figure 5.8) provided some discrimination of features within the open pit and stockpile areas. The highest reflectance and most of the information was contained in band 1.

The band difference enhancement using bands 1, 3 and 5 (Figure 5.9) brought out some more subtle differences between the bands. Figures 5.11, 5.12 and 5.13 are enlargements of the tailings areas from both the band 3,4, 5 and band 1, 3, 5 difference enhancements. With the band difference enhancement it can be seen that more subtle differences are enhanced than with the 3, 4, 5 composite. The exact nature of the differences cannot be determined without field observati-ons. -64-

Figure 5.11: Vegetation and Tailings Enhancement of the B-zone Mine Area -65-

Figure 5.12: Vegetation and Tailings Enhancement of the Central Tailings Disposal and Stockpile Areas -66-

Figure 5.13: Vegetation and Tailings Enhancement of the Tailings and Precipitation Pond Area -67-

5.4 CONCLUSIONS FROM SATELLITE DEMONSTRATION

This satellite demonstration study has demonstrated the ability to digitally improve the spatial resolution of the data as an aid to interpretation. It has also demonstrated that vegetation, moisture and disturbed ground information may be obtained about a mine site. Due to cloud cover, the ability to collect cloud free images on a regular basis may prove to be a problem. This would be dependent upon the location of the mine tailings site and the local climate characteristics.

Interpretation of the data highlighted the need for field information to be collected in order to establish the characteristics of various phenomena in the imagery. Once established, the long term need for field data collection would be minimal. -68-

6. CONCLUSIONS AND RECOMMENDATIONS

It can be concluded from this review that satellite-based remote sensing data and analysis techniques supplemented with airborne surveys and field calibration data could provide a solution for long-term monitoring of uranium tailings. The optimum data set would include Landsat Thematic Mapper, SPOT panchromatic, thermal IR, and either MEIS or CASI airborne multispectral scanner data. This combination of data would provide the means to monitor, in detail, all criteria of concern outlined by AECB. A single operational mission using these data and including data acquisition, processing, output and interpretation would cost approximately $37,000.00

Processing of these data on a PC-based image analysis/geographical information system (GIS) would provide the ability to maintain a database of historic and current observations and interpretations of the various tailings site across the country. The costs associated with such a monitoring program should be within the resources of the Crown or the responsible party.

It is recommended that, in order to adequately complete this feasibility assessment, a demonstration project be undertaken in which the recommended data and techniques be tested. This would allow for an evaluation of data and techniques that have not yet been applied to the study of tailings. In addition, optimal data processing techniques would be identified for monitoring specific criteria which in turn would provide proof of the proposed methods. Specific examples would be required in order for an operational methodology to be accepted and implemented. Cost estimates would also be refined in a practical demonstration and operational constraints identified. The completion of such a study would set the stage for an operational monitoring program using remote sensing. -69-

7. BIBLIOGRAPHY Ahem, F. and D.G. Leckie. 1987. Digital Remote Sensing for Forestry: Requirements and Capabilities. Today and Tomorrow. Geocarto Int., Vol. 2, No. 3, pp. 43-52. Allum, J.A.E. and B.R. Dreisinger, 1986: Remote sensing of vegetation change near Inco's Sudbury mining complexes. International Journal of Remote Sensing, v. 8, pp 399-416. Amato, et al. 1975. Application of EREP, Landsat and Aircraft Image Data to Environmental Problems Related to Coal Mining. NASA Earth Resources Survey Symp., Proc, Vol. 1-A, June 1975, Houston Texas. Ambionics, Inc. 1974. Remote Sensing of Coal Mine Pollution in the Upper Potomac River Basin, Ambionics, Inc., Washington, D.C. Anuta, M.A. and O.P. Bahethi. 1982. Mine Waste Location by Satellite Imagery, Spokane Research Centre, Spokane, WA. Aronoff, S. 1978. The Application of Remote Sensing to Environmental Monitoring of the Athabasca Oil Sands Region, Volume 2. Univ. of Calgary, Dept. of Environmental Design, Calgary, Alta

Bisagni; J.J. 1986. Long Term Remote Sensing of Receiving Waters at an East Coast United States Offshore Dumpsite. Oceans '86, Conf., Proc. Washington, D.C. September 23-25, Vol. 3, pp. 822-827 Boldt, C.M.K. and B.J. Scheibner, 1987: Remote Sensing of Mine Wastes. U.S. Department of the Interior, Bureau of Mines Information Circular #9152. Borstad, G.A. and D.A. Hill, 1989: Using visible range imaging spectrometers to map ocean phenomena. Paper presented at the Conference on Advanced Optical Instrumentation for Remote Sensing of the Earth's Surface from Space, Paris, France, April 1989. Bowker, D.E. and S.R. Lecroy. Int. Journal of Remote Sensing, Vol. 6, No. 5, May 1985, pp. 759-771„ Brosse, J.M. 1985. Evolution du Littoral Sous 1'Effet de Rejets Miniers Dans la Region D'Albo (Corse du Nord). Etude Diachronique des Photograhies Aeriennes. Hydrogeologie, No. 2, pp. 105-111. Foresman, T.W. 1986. Mapping, Monitoring and Modelling of Hazardous Waste Sites. The Science of the Total Environment, Vol. 56, No. 15, pp. 255-263. Franks, L.A., G.A. Capelle, and D.A. Jessup. 1983. Aerial Testing of an N2 Laser Fluorosensor System. Applied Optics, Vol. 22, No. 11, June 1, 1983, pp. 1717-1721. Gagnon, J.Y., Bruce, B.D. and A.A. Crowder, 1985: The use of remote sensing for describing and evaluating the regeneration potential of an abandoned mining waste site, western Quebec, Canada. -70-

Gibson, L.J. 1974. The Application of Aerial Photographic Interpretation to Derelict Land Studies in South Wales. Symposium on Remote Sensing and Photo Interpretation, Proc, Banff, Alta, October 7-11. Gregory Geoscience Ltd. 1975. An Assessment of the Potential Use of Remote Sensing Technology for Monitoring Mine Waste Dumps, Vol. 1, Development of Techniques. Gregory Geoscience Ltd. 1975. An Assessment of the Potential Use of Remote Sensing Technology for Monitoring Mine Waste Dumps, Vol. 2. Experimental Inventory of Mine. Halstead, E.H. and D.A. Rennie. 1972. Experiment Program on Remote Sensing for Crop Identification and Pollution Detection. Can. Symp. on Remote Sensing, Proc, 1st Held, G. 1985. Measurements of Gaseous Emission from Smouldering Coal Dumps. Mesoclimate Air Pollution Related Research in the Eastern Transvaal Highveld, Seminar, Proc., Pretoria, South Africa, Sept., 3, pp. 9-11. Horler, D.N.H., M. Dockray, J. Barber, and A.R. Barringer, 1983. Red-edge measurements for remotely sensing plant chlorophyll content. Advanced Space Research, Vol. 3, No. 2, pp. 273-277. Hornsby, J.K. and B. Bruce, 1986. Regional geobotany with TM: A Sudbury case study. Proc. 10th Canadian Symposium on Remote Sensing, Edmonton, Alberta, pp. 611-621. Jensen, J.R. and E.J. Christensen. 1986. Soil and Hazardous Waste Disposal Site Selection using Digital Geographic Information Techniques. The Science of the Total Environment, Vol. 56, No. 15, pp 265-276. Johnson, R.E. and J.N. Mulvaney. 1978. The Use of Low-Level Oblique Aerial Photography for Environmental Management in Elliot Lake Area. Can. Symp. on Remote Sensing, Proc., 5th, Victoria, B.C. pp. 396-398. Johnson, R.W . 1980. Remote Sensing and Spectral Analysis of Plumes from Ocean Dumping in the New York Bight Apex. Remote Sensing of Environment, Vol. 9, No. 3, pp. 197-209. Kenny, J.F. and J.R. McCauley. 1982. Remote Sensing Investigations in the Coal Fields of Southeastern Kansas. Remote Sensing for Resource Management, Chapt. 29, pp. 338-346. Klemas, V. and W.D. Philpot. 1981. Drift and Dispersion Studies of Ocean-Dumped Waste Using Landsat Imagery and Current Drogues. Photogrammetric Engineering and Remote Sensing, Vol. 47, No. 4, pp. 533-542. Landers, R.W. 1981. Aeriial Photography Interpreted for Contingency Planning Spill Prevention, Compliance Monitoring and Spill Surveillance. Airborne Remote Sensing of Oil Spills in Coastal Waters, Workshop, Proc., Washington, D.C. April 18-20, pp. 98-108. -71-

Li1lesand, T.M. and R.W. Keifer. 1979. Remote Sensing and Image Interpretation. J. Wiley and Sons, 612 p. Milton, N.M. and D.A. Mouat, 1989: Remote sensing of vegetation responses to natural and cultural environmental conditions. Photogrammetric Engineering and Remote Sensing, v. 55, pp 1167-1173. Moore, H.D., Adams, J.H. and A.F. Gregory, 1977: Mapping mining wastes with Landsat images. Presented at the 4th Canadian Symposium on Remote Sensing, Quebec City. Morris, W.G. 1970. Photo Inventory of Fine Logging Slash. Photogrammetric Engineering/Coden_Pgme-A, Vol. 36, No. 12, pp. 1252 Murray, and Moffet, 1977: Vegetating the U Journal of Soil and Water Conservation, v. 32, pp 171-174. Nat. Field Investigations Centre, Denver, Co., 1974. Remote Sensing Report, Baltimore Harbor and Wicomico River Basin, Maryland. Ohlhorst, C.W. 1981. The Use of Landsat to Monitor Deep Water Dumpsite 106. Environmental Monitoring and Assessment, Vol. 1, No. 2, 1981, pp. 143- 153. Ohlhorst, C.W. and G.S. Bahn. 1979. Mapping of Particulate Iron in an Ocean Dump. Photogrammetric Engineering and Remote Sensing, Vol. 45, No. 8, pp. 1117-1122. Peplies, W.R., N.S. Fischman and C.E. Tanner. 1982. Detection of Abandoned Mine Lands: A Case Study of the Tug Fork Basin. Remote Sensing for Resource Management, Chap. 31, pp. 362-376. Petersen, G.W. et al. 1983. Remote Sensing of Energy Waste Transport Processes Using Simulated SPOT Data. ASP-ACSM Ann. Meet., Proc, 50th, Washington, D.C. Mar. 11-16, Vol. 2, pp. 447-455. Pickup, G. and D.J. Nelson. 1984. Use of Landsat Radiance Parameters to Distinguish Soil Erosion, Stability, and Deposition in Arid Central Australia. Remote Sensing of Environment, Vol. 16, No. 3, pp. 195-209. Pitbaldo, J.R. and B.D. Amiro, 1982: Landsat mapping of the industrially disturbed vegetation communities of Sudbury, Ontario. Canadian Journal of Remote Sensing, v. 8, pp 17-28. Reeves, G.R., (ed.) 1975. Manual of Remote Sensing. American Society of Photogrammetry, Falls Church, Va. 2144 p. Reinhold, G, J. Kirsch, H.W. Hurtig and D. Fothmann. 1985. Remote Sensing by Electromagnetic Induction (EMI), Magnetometer and Ground Penetrating Radar. Contaminated Soil, Conf., Proc, 1st, Utrecht, Netherlands, November 11-15, pp. 379-387. Ritcey, G.M., 1989: Tailings Management: Problems and Solutions in the Mining Industry. CANMET, Elsevier. -72-

Rogers, R.H. and W.A. Pettyjohn. 1975. Determine Utility of ERTS-1 to Detect and Monitor Area Strip Mining and Reclamation. Final Report. Rudder, C.L., C.J. Reinheimer. 1973. Detection of Water Pollution Sources with Aerial Imaging Sensors. Joint Conf. on Sensing of Environ. Piolet. 2nd, Washington, D.C., December, pp. 65-72. Sabins, F.F., Jr. 1978. Remote Sensing: Principles and Interpretation. W.H. Freeman and Co. 426 p. Scherz, J.P. 1972. Final Report on Infrared Photography Applied Research Program. Inst. for Environmental Studies, Univ. of Wisconsin, Madison, WI. Schmidt, D. Use of Aircraft Remote Sensing on Problems in Marine Chemistry and Pollution Research and Monitoring. Remote Sensing for Observation and Invertory of Earth Resources and the Endangered Environment, Int. Symp., Proc, Freiburg, W. Germany. Schmidt, D. 1977. Use of Aircraft Remote Sensing in Marine Chemistry and Pollution Research with Particular Reference to the Dumping of T102 Waste Solutions. Deutsches Hydrographisches Inst. Schreier, H. and L.M. Lavkulich. 1980. An Examination of the Overall Relationship between Spectral Reflectance and Chemical Composition of 58 Mine Tailings Samples. Machine Processing uf Remotely Sensed Data, Int. Symp., Proc, West Lafayette, In., June 3-6, pp. 126-134. Siegal, B.S. and A.R. Gillespie. 1980. Remote Sensing in Geology. J. Wiley and Sons. 70 p. Sydor, M.L. 1982. Remote Sensing Evidence for Cleanup of Lake Superior. A Study of Minnesota Land and Water Resources using Remote Sensing, December 1982. Section B., pp. B1-B12. Till, S.M. 1987. Airborne Electro-Optical Sensors for Resource Management. Geocarto Int., Vol. 2, No. 3, pp. 13-23. Titus, S.E. 1982. Survey and Analysis of Present or Potential Environmental Impact Sites in Woburn, Massachusetts. ASP-ASCM Ann. Meet., Proc., 48th, Denver, Co. Mar. 14-20. pp. 538-549. Whitlock, C.H. and T.A. Talay. 1977. Remote Sensing Observations of Industrial Plumes at Hopewell, Virginia. Kepone Seminar II, Proc., Easton, M.D. September 19-21, pp. 482-501. Wobber, F.J. et al. 1975. Coal Refuse Site Inventories. Photogrammetric Engineering and Remote Sensing. Vol. 46, No. 9, September 1975. -73-

LIST OF ACRONYMS AVHRR Advanced Very High Resolution Radiometer CAS I Compact Airborne Spectrographic Imager CCT Computer Compatible Tapes EMR Elc-ctromagnetic Radiation FLI Fuorescent Line Imager HRV High Resolution Visible HEIS Multi Element Linear Array Imaging System MLA Multispectral Linear Array MSS Multispectral Scanner NOAA Natioanl Oceanic and Atmospheric Administration PASS Prince Albert Satellite Station PLA Panchromatic Linear Array PPM Parts per Million SPOT Systeme Pour 1'Observation de la Terre TM Thematic Mapper APPENDIX A DEMONSTRATION IMAGES

Landsat TM Satellite Image of the Elliot Lake Area Scale: 1:500,000 (one Quadrant)

"A" is the centre of an enlargement for the next image example. This Landsat-TM 5 colour composite image of quadrant 1 within coordinates: path 20 and row 28 was acquired on May 10, 1984. The image area is centred within the 1:250,000 NTS map sheet 41J. The colour composite image is generated by inserting TM band 5 through the red gun, TM band 4 through the green gun and TM band 3 through the blue gun. Band 3 which is sensitive to the visible red portion of the electromagnetic spectrum enhancing areas of mine tailings and other less spatially significant features in the centre and lower portion of the image. TM band 4, sensitive to quantity of chlorophyll within vegetation, indicates the percentage of the image covered by vegetational growth. Variations in the green colour are a response to changes in vegetation type, density and/or health. TM band 5, sensitive to barren soil and/or bedrock outcrops, delineates areas of sparse or no vegetation. The sensitivity of the sensor to detect sparse vegetation is a function of sensor resolution. To answer questions regarding sensor discrimination of tailings from barren ground, field work measurements must be used. -A2- -A3-

Landsat TH 4 Times Enlargement of Elliot Lake Image Scale: 1:62,000

This is a photographic enlargement of the previous image showing the various mine tailings areas situated between Quirke Lake "Q" and the town of Elliot Lake "E". A digital enlargement is expected to show enhanced detail. Within the largest tailings area just west of Quirke Lake, changes in tone and texture information from TM band 3 indicate the flow direction of slurry input as well as changes in grain size of the deposited tailings material. -A4- -A5-

SEASAT SAR Satellite Image of the Elliot Lake Area Scale: 1:500,000

Imagery from SEASAT is similar to imagery that will be available from RADARSAT. "B" is the centre of an enlargement for the next image example. The image header indicates that this data was acquired on July 23, 1978 on orbit 378. This is the first radar data set ever imaged from a satellite and was operational for only 100 days. In contrast, Landsat sensors 1 through 5 have been continually imaging the earth for almost a decade, enabling an historic image record to be compiled. Landsat data is most useful when imaged during cloud-free days, whereas SAR data is considered an all-weather sensor, unaffected by cloud cover. This Synthetic Aperture Radar (SAR) image demonstrates the use of the sensor in delineating terrain topography. The more rough the ground surface becomes, brighter the image tone becomes. Patterning that results from variations in image tone (image texture) enables the discrimination of bedrock structure and general lithology. By rotating this image "B" is placed at the same location as "A" on the Landsat TM image. Although the spacial resolution is slightly better in the radar image, the spectral resolution of a TM colour composite image is superior for detection of mine -A6-

%MD!&?mMz, -A7-

SEASAT SAR 4 Times Enlargement of Elliot Lake Image Scale: 1:62,000

This is a photographic enlargement of the previous SEASAT image. Within Quirke Lake "Q" the water increases in image tone on the south side of every island within the lake, indicating a wind is blowing out of the north. The Elliot Lake town site "E" can not be discriminated as well as in the Landsat TM enlargement, as there little difference in surface roughness between the buildings of the town and the surrounding ground surface. -A8- -A9-

Landsat TM Satellite Image of the Baie Verte, Nfld. Area Scale: 1:1 Hill ion (one TM scene = 4 Quadrants)

"C" shows the location of the Airborne SAR image at a larger scale. This Landsat-TM 5 colour composite image can be located using the Latitude and Longitude coordinates shown along the image margins. The image is also identified by orbital coordinates: path 4 and row 26 and was acquired on August 25, 1988. The colour composite image is generated by inserting TM band 7 through the red gun, TM band 4 through the green gun and TM band 2 through the blue gun. Band 2 which is sensitive to the visible green portion of the electromagnetic spectrum enhancing areas of an unknown origin. TM band 4, sensitive to quantity of chlorophyll within vegetation, indicates the percentage of the image covered by vegetational growth. Variations in the green colour are a response to changes in vegetation type, density and/or health. TM band 7, sensitive to barren soil and/or bedrock outcrops, delineates areas of sparse or no vegetation. The sensitivity of the sensor to detect sparse vegetation is a function of sensor resolution. Generally areas of red in this image appear to be caused by the result cf resent forest fires. By comparing this to the Elliot lake TM image, an immediate conclusion is that there is far more vegetation coverage in the Baie Verte area than that of Elliot Lake. To answer questions regarding the cover type that band 2 is detecting field work measurements must be used. -A10- -All- Canada Centre for Remote Sensing Airborne SAR Baie Verte, Nfld. Scale: 1:86,000

"M" is the location of an open pit asbestos mine situated just north of the town of Baie Verte "V". The image header indicates that this data (Task 88-29) was acquired on September 6, 1988. This high resolution radar image allows the delineation of individual buildings within the town of Baie Verte, as well as highways, secondary roads, powerlines and small forest clearings not sensed by a satellite like SEASAT. This Synthetic Aperture Radar (SAR) image demonstrates the use of the sensor in delineating terrain topography. The more rough the ground surface becomes, brighter the image tone becomes. Patterning that results from variations in image tone ''~iage texture) enables the discrimination of wetland types. To view the image correctly, the image should be held with the image shadows facing down (facing towards the observer). Mine waste tailings dumped around the pit area show variations in roughness and dump truck routes are visible. -ziv- -A13-

SPOT Panchromatic Satellite Image of the Baie Verte, Nfld. Area Approximate Scale: 1:500,000

This image has three times the resolution of Landsat TM but being a single band, lacks the spectral resolution of a composite image enhancement. The data's ability to sense terrain albedo can be integrated digitally with Landsat TM to create a superior data product. -A14- -A15-

Canada Centre for Remote Sensing Airborne MEIS II Multispectral Array Imager Gatineau River, P.Q. Scale: unknown

The left portion of the image continues on to the right. This image clearly shows the potential for monitoring land surface features. With the added advantage of 8 selectable bands, spectral resolution is sufficient to monitor vegetation stress of individual trees. The size variation of common urban targets (ie. cars, logs in river and parking lot lines) can be used as a guide to deduce the detectability of mine tailings features. •686'. ri2 ^30 -Z JiVd' 19 • 3ma 'L • M33S3 ' •i - aaa :SI3NNWHD- """I 'a3A:a n»3n:_sa oi -S^a nana .NOW - na mos 3N:MOTIOJ

-:S-N3WWOD aasn

0'69 • ««d. 0'98 • 3N;aK3H I'lB • C3S/W) Q33dS WiL ' (H) '-It* B'.e- 3N0Z St'6 •IW)'3WS 3fj:i £S'6 -CM) 'DWdS X:d

NNNA- d*(W NNNNNNNNNNNNNNNA" saa • NNNNNNNNNNNNNNNA- 3a5

(WN) MS '3AW SdNkia..

e0!0i2S>N 2D2EZ68 3N:i.iH3PJ S:3W-aOSN35

-9TV- -A17-

Canada Centre for Remote Sensing Airborne HEIS II Multispectral Array Imager (4 times enlargement) Gatineau River, P.Q. Scale: unknown

This photographic enlargement shows the individual pixels or image elements used to create the image. Image resolution is determined by the size of individual pixels. -A18- -A19-

The Erie Canal at Otter Creek. Orleans County. Black and white aerial photography from May 1988.

The E'ie Canal at Otter Creek. Orleans County. Colour infrared photography from May 1988. Red colour is dje to high reflectance from vegetation in the infrared wavelengths. -A20-

The Erie Canal at Otter Creek. Thermal infrared imagery from November 1988. Streams clearly visible despite dense tree cover. -A21-

CASI Airborne Scanner Data

The following three pages contain example data from the CASI scanner. The data was collected late November so the sun angle was low. All data were collected using a "chlorophyll" band set.

Band 1 435.3 - 479.1 nm Band 2 538.9 - 551.3 nm Band 3 484.9 - 597.3 nm Band 4 631.0 - 645.3 nm Band 5 657.7 - 673.8 nm Band 6 673.8 - 688.1 nm Band 7 707.7 - 714.9 nm Band 8 745.4 - 759.8 nm Pixels are 2.5 m by 2.5 m in size. -A22-

Isle of Wight: 512 x 2800 pixel image decimated to display 512 x 512 pixels. Saturated shorelines correspond to chalk cliffs. True colour composite of bands 1. 3 and 5.

Isle of Wight:

False co'cur composite using bands 3. 5 and 8. -A23-

Isle of Wight: 512 x 512 cclour subscene. White dots in estuaries are moored sailboats. The reddish-brown colour corresponds to salt march. The lower right of the image is agricultural land-use behind a dyke.

Isle of Wight: Zoom of subscene to illustrate limit of the resolution. Individual pixels are visible. -A24-

False colour composite of the River Wyre.

Contrast stretch of false colour composite to enhance industrial plume, Subsurface features in the river indicate affect of feeder streams on bottom sediment. APPENDIX B SELECTED PAPERS B-l

Airborne Electro-Optical Sensors for Resource Management

S.M. Till Acting Head, Sensor Section Canada Centre for Remote Sensing 2464 Sheffield Road Ottawa, Ontario, K1A OY7 Canada

Abstract

There have been major advances in Canada in the development and operation of airborne electro-optical sensors for resource management Systems developed include the laser-based remote sensors far profiling and coastal mapping, and the multispectral array imagers for land and ocean monitoring. The paper describes the state-of-the-art systems and their applications, and addresses their role in resource management

1. Introduction — the Multi-Element linear array Imaging System, MEIS, a pushbroom imager, During the past decade, Canada has been active in the - the Fluorescence line Imager, an imaging spectro- development and operation of airborne electro-optical meter with two-dimensional detector arrays. sensors for resource management. Major developments These will be discussed in some detail in this article, as have occurred with laser-based systems and with digital examples of the capabilities offered by airborne electro- multispectral imagers. Hardware developments have been optical systems developed in Canada for management of accompanied by the parallel development of data proces- renewable and non-renewable resources. sing and analysis systems and methodologies, and by This list of remote sensors is by no means exhaustive, extensive data acquisition programs for remote sensing and does not include systems used for monitoring atmo- applications ranging from cartography and mapping to spheric constituents or those developed primarily for forest inventory and ocean monitoring. ground-based use. Such systems are described elsewhere The airborne electro-optical program in Canada has and the interested reader is referred to recent publications involved Canadian industry and the Canadian remote and trade journals. For example, commercially available sensing community, and has resulted in the development atmospheric monitors include the COSPEC, for remotely and evaluation of new technologies and of new applica- measuring sulphur dioxide and nitrogen dioxide for air tions. These state-of-the-art airborne electro-optical pollution studies (Moffat et al, 1971; Hoff ei al, 1982; Hoff systems have the potential to revolutionise resource and Gallant, 1985), the GASPEC for remote detection of surveys and the management of resources by providing trace gases (Ward et al, 1975; Reichle et al, 1982) and the efficient and cost-effective information in digital format that GASPILS for remote detection of methane (Lee it al, is readily compatible with computer-based information 1985). Several ground-based radiometers are available systems. Airborne sensors both complement and supple- such as the Hand-held Reflectance Radiometer, that is ment the role of the satellite sensors in resource manage- designed for field mineral exploration and fine REFSPEC ment. (Gladwell et al, 1983) for ground-truthing remotely-sensed Some of the airborne electro-optical systems developed data. There are also various ground-based and/or in recent years in Canada for remote sensing of land and sampling systems that use laser sources, such as the ocean include: Luminex (Seigel and Robbins, 1983) and the tunable laser - a lidar, known as LARSEN, for hydrographic surveys, diode absorption spectrometer developed by Unisearch - a laser fluorosensor, for marine pollution and hydro- Associates Inc. (Concord, Ontario) to sample and identify carbon monitoring, target gases of atmospheric interest.

Geocarlo International (3) 19S7 B-2

2. The Airborne Scanning Lidar - LARSEN mapping system that provides water depths, contours, shoreline and foreshore information. In addition to the 2.1 Coastal Mapping airborne component, the other major parts of the system are the data reduction system and the video mapping In August 1985, Canadian hydrographers became the system. first to use airborne laser scanning techniques for The airborne component contains the laser transmitter chartmaking purposes, when the LARSEN system was and receiver and the associated electronics, the guidance used to locate and survey shipping channels through parts system for positioning, a video camera/video disc system of the Northwest Passage in the Canadian Arctic (Casey to provide along-track imagery and a data logger for the and Vosburgh, 1986). Two high priority areas (2000 km2) storage of the digital data (Banic et a\, 1987). The lidar uses of the southern route were sounded, in order to compile a frequency-doubled Nd:YAG laser, operating simultane- new nautical charts. A second Arctic survey was carried out ously at 1064 and 532 nm, at an average repetition rate of in 1986, when an additional 1000 km2 of area were about 20Hz. The laser output beam is directed towards the surveyed. water by a scanning mirror which with the !?»ser firing The LARSEN system measures water depths of up to control allows a specific and uniform grid pattern to be 40m in clear coastal water, with an accuracy of 0.3m. generated on the water. The location of each laser From an altitude of 500m, the system covers a swath sounding is determined using aircraft position, altitude and 270m wide and generates a uniform sounding grid pattern, attitude information. Aircraft position is provided by a with the soundings spaced about 30m apart and a posi- microwave ranging system, using ground-based transpon- tioning accuracy of about 15m. ders of known location, and by the satellite-based Global The success of the two hydrographic surveys has Positioning System. Aircraft height above the water is demonstrated the advantages of the airborne lidar for measured by an infrared laser altimeter and attitude data surveying coastal areas. Of major importance are the by an inertia! reference system that is hard-mounted to the speed of data acquisition, with the accompanying cost- sensor. These positioning systems not only provide the savings compared to surveys using surface vessels, and the location of each lidar sounding, but also provide real-time fast mobilization and response capability; these advantages flight-line guidance to the pilot. The airborne processing is are of particular value in remote areas that are difficult and based on a PDP-11/73 processor. For each laser pulse, the expensive for survey vessels to access, and in Arctic areas received lidar wave form is digitised and recorded together where'the optimum conditions for surveying occur in short with positioning data. The airborne system is a nadir- and unpredictable time windows. looking video camera and recorder. The data reduction system includes the hardware and 2.2 LARSEN Concept and System Design software required to reduce the raw lidar data to a field sheet format. During this processing the depths may be The LARSEN is the result of a Canadian research and plotted as integers and colour-banded by depth, and any development program that started in the early 1970's with section of the area sounded can be enlarged and participation from Government agencies and industry. The examined in detail for quality control purposes. The data prototype system (O'Neil, 1980, Ryan and O'Neil, 1980) reduction system has also been designed to be operated in was used to demonstrate the concept, to evaluate the the field in a trailer during the survey. performance, and to develop data interpretation proce- The video mapping system provides shoreline and dures (Gluch et al, 1983). The LARSEN was developed by foreshore information, by correcting the video imagery Optech Inc. (Downsview. Ontario), the Canada Centre for using the processed position data. Remote Sensing, and the Canadian Hydrographic Service. Terra Surveys (Sidney, British Columbia) was contracted 2.3 LARSEN Performance to manage the 1985 and 1986 surveys, and contributed to the data processing and deployment strategies. As part of the LARSEN development program, the The system uses an optical radar (lidar) technique to system underwent numerous flight trials to evaluate its measure water depth from an aircraft. A short pulse of performance and to investigate the effects of changes in infrared radiation and a short pulse of green radiation are water quality, depth, surface wave structures, and ambient simultaneously transmitted from a laser on board the light conditions on the performance of the lidar. Right tests aircraft down towards the water surface. The infrared pulse took place over lakes and rivers with differing conditions, is scattered from the water surface, while the green pulse and depths measured with the lidar were compared with penetrates the surface and is scattered from the bottom. data from acoustic surveys. Measured depths ranged from Both scattered pulses are detected by a receiver in the less than 10m in highly turbid (river) waters to greater than aircraft and the elapsed time between the pulses is used to 20m in clearer lake waters. Analysis of the lidar depths and measure the water depth. The laser transmitter and acoustic depths obtained during trials in Lake Huron receiver comprise one part of the complete LARSEN revealed no consistent bias, and yielded a standard devia- B-3

Son of the lidar measurement error of 0.26m (Casey et al, tion in the application of the technology. The Mklll 1985). During the Arctic Survey, depth measurements version, which is still being operated for commercial appli- greater than 35m were common (Casey and Vosburgh, cations by Baninger Research Ltd. (Rexdale, Ontario) has 1986). now been pined by a system dedicated to hydrocarbon monitoring (the Fluoroscan), while the MklV version has 2.4 Summary been designed for operational monitoring of the Canadian offshore and Arctic, with emphasis on realtime display and The LARSEN coastal mapping system has proven to be flexibility of performance that can be optimised for a relatively fast and inexpensive hydrographic survey tool, different targets (Till and O'Neil, 1985). which has resulted in the acquisition of a tremendous volume of new and valuable sounding data. It promises 3.2 Laser Fluorosensor System Design greatly increased levels of productivity, provides the capa- bility to seek and explore shipping channels in frontier The airborne laser fluorosensor uses a pulsed ultraviolet areas, and provides survey managers with a new tool for laser to illuminate a small area beneath the aircraft. Any use in areas traditionally considered too expensive and fluorescence that is excited by the laser is collected by the time-consuming to survey. receiver on the aircraft, and is resolved into a spectrum. The airborne lidar system also promises to be of value Since the fluorescence spectrum is characteristic of the in land applications, such as terrain profiling and tree- target substance, the spectrum acts as a signature and is height determination. The lidar profiler of the LARSEN used to classify the substance by comparison with known has, for example, been used in the Kananaskis Mountain fluorescence spectra. For example, in the case of an oilspitl Range, Alberta. The operational range limit is greater than on water, the laser fluorosensor will excite oil fluorescence 10km and the range resolution is close to 0.15m, the accu- which may be used to detect the presence of oil and to racy of the measurement being affected by terrain vari- identify the class of oil (O'Neil et al, 1980). ations within the illuminated target spot on the ground The basic laser fluorosensor comprises a laser transmit- (Gibson, 1986). A lower powered lidar profiler, with a ter, telescope receiver, and a range-gated spectrometer range limit of about 400m, has been evaluated over plan- with dedicated data acquisition system and display unit. tation forests in Petawawa, Ontario (Schreier et al, 1984). The Mklll version, which is described in detail elsewhere (O'Neil et al, 1980), uses a nitrogen laser operating at 337nm and a pulse rate of lOOHz The induced fluores- 3. The Airborne Laser Fluorosensor cence is resolved into sixteen spectral channels covering the range 370nrn to 710nm. The spectra are displayed on 3.1 Development a video monitor, together with the correlations obtained with stored reference spectra. The data are also stored on The laser fluorosensor development program in digital tape for post-flight analysis. The early version of the Canada started in the mid 1970's, and has been largely Mklll included time-resolution of the fluorescence signal to related to the requirements for systems to monitor and provide additional target characterisation. This was found police marine discharges, and to detect hydrocarbons and to be at the expense of increased system complexity and minerals. It offers the resource manager ihe capability of reduced responsrvity (i.e. signal-to-noise ratio) and this not just detecting but classifying marine pollutants such as feature was omitted in the later models. oilspills (of value for pollution control activities). It can The Fluoroscan and the MklV are based on a higher- operate over a range of conditions such as sea-ice or power excimer laser, operating at 308nm. In the case cf coastline, and it is not limited to daytime use. the MWV, an array detector system is used to allow select- Following early prototype systems, the multi-channel able spectral channels so that it can be readily optimised laser fluorosensor known as the Mklll was developed for for targets other than oil, and the majority of the data the Canada Centre for Remote Sensing. It successfully processing is carried out in real-time (Till and O'Neil, participated in 1978 in oil spill trials off the East Coast of N. 1985). America, and was one of the major systems tested in the Arctic Marine Oilspill Program of the Department of Envi- 3.3 Applications ronment (O'Neil et al, 1980; O'Neil et al 1983). Since then, the system has continued to be used in oilspill monitoring, The airborne laser fluorosensor is recognised in Canada water quality and Coast Guard related activities over as a valuable component of a remote sensing system for marine and inland water features. The development monitoring oilspill and marine discharges. As in the case of project at CCRS has been accompanied by extensive the LARSEN, it has demonstrated advantages for laboratory measurements on oils, development of data monitoring coastal areas, it can be mobilised quickly in reduction and processing techniques, and system improve- response to emergencies, and it is a relatively inexpensive ments, which have helped place Canada in a unique posi- tool for remote or frontier areas. Of interest for Canadian B-4

industry and Canadian environmental agencies has been forward, a new line is imaged and sampled, and as this its potential to monitor oil on sea ice, of particular value for process continues, two-dimensional imagery is built up, Arctic regions where other systems may not be able to the aircraft motion providing the scanning in the forward successfully identify oil. Its application for monitoring direction. The spectral content of the imaged scene is oilspills and marine environments has been its major appli- selected by means of an optical filter placed in front of the cation to date in Canada. It is not limited to such applica- lens. tions, and offers promise for inland water quality MEIS I had two optical channels, each with detector monitoring and for mineral exploration (Seigel and array, lens and filter, and successfully demonstrated the Robbins, 1983). pushbroom imager concept (Zwick et al, 1978; Zwick, 1979). MEIS II has eight spatially registered imaging spectral channels, and incorporates sophisticated real-time processing to provide geometric and radiomelric correc- 4. The MEIS Pushbroom Imager tions to die output imagery. An onboard image display and analysis system provides scrolling colour imagery in 4.1 Introduction real-time, with operator selectable enhancements and hard copy (video tape) output. The digital imagery data, naviga- The MEIS u.as the first solid state muttispectral array tion and system data are recorded on tape, for post-flight imager to be developed and to provide data to the remote analysis and precision production. The system has been sensing community (Zwick, 1979; Neville et al, 1983). described in detail elsewhere (Neville et al, 1983). In brief, Following the development of the prototype MEIS I in the field-of-view imaged is 40° with 1024 pixels of 0.7 1978, the MEIS II was implemented and has been mrad size recorded across the swath. The detectors are operating as pa: • of the airborne program of CCRS since silicon linear array charge-coupled devices, with spectral 1982 (Till et a\, 1986). As a result of its demonstrated response from 350nm to llOOnm. Interchangeable performance and capability, it has generated new applica- spectral filters are used and these are selected to optimise tions in resource monitoring and has laid a solid founda- the performance for different remote sensing applications tion for the routine operational use of such systems for by matching the response to, for example, forestry signa- resource monitoring. It is seen to be a major new tool, for tures or water quality features. Note that the filters used are example, in forest surveys and mapping, and activities are "blue-shift-free" across the field-of-view; that is, unlike the already underway in Canada to implement dedicated usual interference filters, they have been developed to airborne systems and processing systems to provide prod- exhibit no change in passband with view-angle (Till, ucts on an operational basis to the resource manager. Neville et al 1986) and so greatly enhance the perfor- mance of the imager. MEIS (Multi-detector Electro-optical Imaging Sensor) was developed by MacDonald, Dettwiler and Associates (Richmond, British Columbia) under contract to the 4.3 MEIS Stereo Imager Canada Centre for Remote Sensing. It is a high perfor- mance digital multispectral imager that makes use of the In addition to the nadir-mode of operation, the MEIS pushbroom design to provide improved radiometric system is used to acquire continuous fore-aft stereo imag- sensitivity and geometric fidelity. These characteristics in ery, by the addition of a precision stereo minor module. In turn enable it to monitor small changes in target which may the stereo mode, two of the eight channels look fore and not be apparent with other systems such as opto-mechan- aft at angles of ± 35° while six channels look nadir. The ical multispectral scanners and survey cameras. The data may be combined with the aircraft attitude data to geometric quality of the output imagery and the digital provide digital terrain information and standard geometri- format make the sensor readily compatible with computer- cally corrected products. One of the standard products based inventories; these properties together with the capa- available through the CCRS transcription system is a geo- bility for continuous fore-aft stereo acquisition offer referenced image referenced to the Universal Transverse particular promise for applications requiring cartographic Mercator Grid, in digital or hard copy format (Gibson et al, quality products. 1983; Gibson, 1984; Gibson et al, 1987).

4.2 MEIS Concept and Design 4.4 MEIS Applications

The concept is straightforward. A linear multi-element The MEIS II completed its fourth year of operation in array detector is located in the focal plane of an imaging 1986 as part of the airborne electro-optical facility of the lens. The distant scene is focussed onto the array, and the Canada Centre for Remote Sensing and is now being signals generated by each element of the array are elec- operated for commercial applications of remote sensing by tronically sampled and digitized, to produce a line image in Innotech Aviation Ltd. (Montreal, Ottawa). digital format of the scene below. As the aircraft moves The airborne program of CCRS has provided services B-5

to acquire airborne digital imagery in order to support and tigating classification of coniferous species, insect damage encourage the development of new techniques and appli- mapping, and clear-cut mapping (see, for example the cations of remote sensing for federal and provincial agen- review by Ahem, this issue) have led to initiatives to imple- cies, industry and university. Data acquisition has been ment a MEIS-based system for operational forestry accompanied by the in-house development and evalua- surveys. Such a system will provide forest inventory tion of new sensors and systems, such as MEIS, the mapping, insect and disease damage assessment, inven- development of new data-handling and interpretive tory update and forest sampling, and will offer multi- methodologies, and the introduction of new applications. spectral imaging with stereoscopic and cartographic capa- The airborne electro-optical facility was developed to bility. provide state-of-the-art multi-spectral digital image data for Other applications where MEIS has been demonstrated Canadian research and development, and for pilot project include water quality monitoring, agricultural studies, and operations where industrial capability was not yet in place geological mapping. (Till, McCoU ef al, 1986). Airborne array imagers offer flexibility of performance, The MEIS II was installed in the CCRS Falcon in 1983. optimised to the target characteristics, as well as efficient This aircraft provides a fast-response and large coverage and cost-effective coverage. With their capability for high capability, well-matched to the requirements for efficient spectral and high spatial performance, they offer enhanced and speedy coverage of land and ocean resources. Since performance compared to satellite systems of low spectral 1983, the MEIS has flown on dose to 60 remote sensing and spatial characteristics. missions per year, and has acquired data for both applica- tion research and development and for sensor evaluation (Till et al, 1986b). The success of this system, developed 5. Fluorescence Line Imager initially as part of the research program of CCRS, can be gauged by the interest generated in its capability for oper- 5.1 Background ational resource management, and in the data acquisition service now being offered by Canadian industry on a The Fluorescence Line Imager is an airborne imaging commercial basis. Applications with the MEIS are spectrometer that was designed to image ocean described elsewhere (for example, Till, Neville et al 1986); chlorophyi) fluorescence and spectral reflectance changes particular interest has been generated by its capability to in water caused by phytoplankton in the sea. The mapping acquire image data in narrow spectral bands, for example of phytoplankton distribution and growth has important for vegetation stress studies and forest damage surveys, applications in fisheries and physical oceanographic and by its ability to acquire digital stereo data for carto- studies, since the phytoplankton constitute the first link in graphy and terrain profiling. An example of a MEIS D image marine food webs and play a major part in the conversion is shown in Figure 1. The early data for forestry applica- of solar energy into organic biomass. Chlorophyll is tions demonstrated that the high spatial resolution and commonly used as an index of phytoplankton abundance. good geometric characteristics of MEIS II offered excellent The chlorophyll of the phytoplankton strongly absorbs capability for tree species identification and information blue light so that the colour of water bodies is seen to suitable for inventory purposes. Other missions, inves- change as the phytoplankton concentration changes, that

Figure 1. MEIS II imagery

a) Geometrically corrected imagery of a suburb of Ottawa, from one channel of the MEIS stereo mode, looking forward at 35 degrees. The data was acquired by the CCRS Falcon aircraft on October 29, 1985. The pixel dimension is 0.4m, and the spectral band is centered at 842nm with a bandpass of 40nm. B-6

b) MEIS 11 images of logging activities and forest regeneration near Smithers, British Columbia. The images were processed on the FIRE image recorder from data acquired by the CCRS Falcon aircraft on August 20,1986 as part of an investigation of juvenile stand condition by the Canada Centre for Remote Sensing and the British Coiumbia Ministry of Forests. The first image (1) has a pixel dimension of about 5.5m and a swath width of close to 6km. The second image (2) was acquired in the same area with a pixel dimension of about lm. The region to the left of the area marked by the circle was logged in 1976, burned in 1977 and now has 6 year old regener- ation growth. Note that slash piles can be seen. The three MEIS channels printed are 777nm, width 37nm, printed red; 676nm, width 39nm, printed green; 549nm, width 31nm, printed blue. B-7

is as the amount of the chlorophyll pigment changes. This across the swath colour change can be exploited to remotely sense the The system design has been described elsewhere chlorophyll In addition, chlorophyll fluorescence is stimu- (Borstad et al, 1985). It comprises five separate optical lated by sunlight and provides a signature of the phyto- channels or camera modules, aligned to provide a field-of- plankton chlorophyll in near surface water, again providing view of about 70°, with 1925 detector elements and 1.3 a method for remote sensing of the chlorophyll and the mrad resolution. (Each module has a silicon diode array phytoplankton. The monitoring and interpretation of water with 385 elements across track by 288 elements.) The colour are important for monitoring phytoplankton distri- spectral response is from 430nm to 800nm, all bution on local and regional scales, for monitoring ecolo- wavelengths are accessible in steps of 1.4nm and the gical effects in lakes and coastal waters, and for deter- spectral resolution is 2.5nm. mining water quality. One disadvantage with the large number of detectors is The Fluorescence Line Imager development evolved associated with the large increase in data and data rate. from the studies in these areas and from the requirement to The hardware development program has been accom- monitor chlorophyll fluorescence for such applications panied by the development of the calibration techniques (Borstad et al, 1985). It was developed for the Department and the data processing methods required to handle the of Fisheries and Oceans under contract by Moniteq Ltd. large amount of data and to make full use of the system's (Concord, Ontario) with a subcontract to Itres Ltd. (Cal- performance capability. gary, Alberta). After delivery in 1984, it was flight tested, and participated in several missions over inland waters and 5.3 Fluorescence Line Imager Applications off the E. coast of N. America. It has also been used over land, and has acquired data for forestry and agriculture The Fluorescence Line Imager is one of the first projects. airborne imaging spectrometers to be available to the The major advantage of the multi-element array imager remote sensing community, and its spectral imaging capa- compared to the more conventional mechanical scanners bility offers promise for both ocean and land-based appli- is its increased sensitivity. As in the case of the pushbroom cations. This spectral imaging capability can be used to imager, the increase in detector number allows much determine spectral features characteristic of certain longer integration times, with the associated higher signal- phenomena, to monitor the occurrence of such features to-noise ratios, which in turn allow imaging of narrow and to redefine the optimum spectral bands. spectral bands or of low radiance targets such as water. Designed primarily for ocean applications, it has been Note that each detector stares at the target and is able to used in several missions for applications related to water integrate the signal over one pixel for a complete "scan" quality and aquatic vegetation monitoring, as described in period. In the case of the mechanical scanner, with a presentations by O'Neil et al and Mouchet et al in the rotating mirror to sweep the field-of-view of a few detec- 1986 Workshop co-ordinated by Edel and Bianchi. It has tors across the scene, the signal from one pixel is integrated also been demonstrated over land, for example for inves- for a fraction of the scan period. The imaging spectrometer tigating chlorophyll reflectance effects in vegetation and for also offers the advantages and flexibility of spectral band imaging areas of different forest type (Edel and Bianchi, selection, so that its spectral response can be matched to 1986). During 1986, it was used in a project in the Federal different feature signatures. Republic of Germany to monitor spectral reflectance from forests as part of a forest disease and forest damage survey 5.2 Fluorescence Line Imager System Design program. It is available for data acquisition through Moniteq Ltd. (Toronto). The system is an imaging spectrometer that makes use of a two-dimensional multi-element array of detectors in the focal plane of a dispersive optical system. The light 6.0 Electro-optical Support Systems from the distant line swath on the ground is collected by the objective lens, dispersed by a grating into a spectrum, The development of the high resolution digital imagers and focussed onto the array. The across-track spatial infor- has required the parallel development of systems for real- mation falls along one dimension, and spectral information time formatting and display, and of transcription and from each pixel is registered along the other dimension. analysis systems to provide data products. The system can operate in two modes. In the "spatial" mode, the system can perform high spatial resolution 6.1 ALICE and Real-time Monitoring in Remote mapping by forming pushbroom images in 8 spectral Sensing bands, the width and location of the spectral bands being selectable by software. Alternatively, in the "spectral One of the major developments in Canada in the mode", the system can provide low spatial resolution former category has been that of the ALICE family of mapping in 288 spectral bands, with 40 pixels located systems (McColl, 1985). developed under contract to B-8

CCRS by Knudsen Engineering Ltd. (Perth. Ontario). The Other systems in the ALICE family include the digital airborne real-time display features operator selectable ultraviolet/infrared dual channel scanner, with real-time display of multispectra) imagery from digital sensors such radiometric corrections, enhancements and display as the ME1S, the CCRS (Daedalus) MSS and the Fluores- (McColl el a/, 1984). cence Line Imager. It includes image analysis features such These systems with their capability for real-time assess- as scene-enhancements (linear stretch, histogram ment of the target conditions, provide a valuable tool for equalised or selectable) and scene histogram statistics, and the resource manager. For example, the airborne electro- provides colour scrolling imagery and output to video optical imager can be deployed rapidly in response to cassette. This allows useful quality assurance directly from emergencies, and by selecting and optimising the spectral Ihe airborne system, and rapid analysis either in real-time bands for the target can provide an immediate assessment in the air or on the ground using the recorded digital data. of the situation that is not possible by other means. A

Figure 2. Airborne mulbspectral scanner image of an oilspill in the Beaufort Sea. acquired by the CCRS Falcon jet on August 14, 1986. The extent of the oilspill is dearly visible B-9

recent demonstration of this was during the Beaufort Sea forestry targets. The MEIS was deployed rapidly once the Oilspill Dispersant trials in 1986, when the CCRS Falcon colour change started, and monitored a province-wide aircraft equipped with an electro-optical sensor configura- swath at the rate of close to 5000 km2 per hour. In such tion was used to monitor the oilspills and to evaluate the situations, where the bio-window is short, the airbome effect of the clean-up procedures (McColl et a\, 1987). The system with its flexibility and speed of response, offers prime sensor was the Daedalus multispectral scanner with considerable advantages over satellite systems. As well, the the ALICE processor and display. The aircraft monitored high spectral resolution combined with the high spatial the path of the oil throughout the trials, and the real-time resolution of the airbome imager offers improved capa- imagery was used to map the extent and the relative thick- bility for the monitoring of the subtle target signature ness of the oil. An example of the imagery is shown in effects associated with tree damage. Immediately following Figure 2. The outputs from three spectral channels the flight, the ALICE system was used to display the centered at 525, 573 and 618 nm have been printed as imagery to the survey managers, and a number of affected red, green and blue respectively. The oil slick is dearly areas were detected and later verified. The MEIS imagery visible with the thin oil appearing pink, and the thicker oil clearly showed spruce budworm defoliation in real-time. red, against the blue of the background water. Indeed, not More detailed evaluation was carried out, resulting in the only was the multispectral imager able to map the spills transfer to the forestry map base of the defoliation informa- during the trials, but two days after the trials it was able to tion derived from the MEIS imager. This signified an detect and map slicks believed to be the remnants of the important advance in the use of electro-optical imagery for trial spills. A similar demonstration was carried out in resource management. response to the Vinland drilling platform natural gas blowout in 1984. On that occasion, the dual channel 6.2 Data Processing Facility digital ultraviolet/infrared imager was used with the ALICE processor; the strong contrast in the ultraviolet between In order for airbome electro-optical systems to be of the thin oil and the water, together with the infrared signa- value for resource management on an operational, routine ture of the water background provided excellent oil basis, it is essential that the systems provide products in a imaging and oil thickness discrimination of the slick form matched to the manager's requirements, whether this (McColl el al, 1984). be a map product for surveys or an enhanced colour The real-time display has also been demonstrated in a image for agricultural monitoring. forest management role, in particular in the monitoring of In Canada, the development of the electro-optical insect disease. A forest insect problem encountered in New airbome sensors with their digital data output recorded on Brunswick is that of the spruce budworm, which infests the tape, has been accompanied by the development of spruce and causes eventual defoliaton with loss of timber systems to handle the data and to provide standard prod- revenues. The effects of the spruce budworm are apparent ucts. The systems developed for CCRS in support of the to the eye first as a change in the colour of the needles, for airbome program provide data in both hardcopy and tape a short variable period in early summer. This colour formats. The hardcopy may be in the form of quicklook change has been used by experienced observers to sketch black and white paper print, video tape, or precision map from aircraft the extent and depth of the colour colour hardcopy from a high-resolution film recorder, the change in order to estimate the damage and to help in the FIRE-240 (MacDonald, Dettwiler and Associates, Ltd.). planning of the spraying program. The use of a multi- Digital products are in standard computer compatible tape spectral imager of high spatial and spectral resoution, such format. The most recent transcription system at CCRS is as MEIS II, promises improved surveys of the spruce the AIR-2 system, developed by CCRS and Moniteq Ltd. budworm defoliation, in terms of detection of early stages as an integrated turnkey VAX computer based system for of the infestation, and discrimination between the levels of transcription of the airbome high density instrument tapes defoliation. In addition to the real-time video output, to standard format computer compatible tapes and for output is also recorded in a digital form, compatible with geometric correction of airbome line imager data, such as map products or inventory bases and this allows for ready MEIS and MSS, (Gibson & a\, 1987). Based on the and cost-effective comparison with, and update of, existing algorithms developed by Gibson (1984), aircraft inertial inventories. navigation data are used to compute positions for all input The potential advantages for forest damage mapping pixels and to. resample the imagery to a user selectable were first demonstrated using MEIS II in the early summer geo-referenced output grid. Producis are available as a of 1983, when areas affected by spruce budworm were standard geometric product (correct but not absolutely monitored. More recently, the CCRS Falcon aircraft positioned), as a geo-referenced imagery product, by the participated in a pilot project with die Province of New use of a few ground control points, and as a comparative Brunswick Timber Management Branch, during the 1986 geometric product that references one image to a second spruce budworm mapping program. The prime sensor image for comparative studies. flown was the MEIS II with spectral bands optimised for There are further developments in Canada of computer B-10

based processing systems for airborne digital data and of data," Proc. 11th Canadian Symposium on Remote Sensing, interfaces to computer based geographic information 1987. systems. The emphasis is on data throughput requirements Gladwell, D.R., Lett, RE. and Lawrence, P., "Application of and product outputs for operational cartographic, reflectance spectrometry to mineral exploration using portable mapping and forestry management applications. These radiometers," Econ. Geol. 78, 1983, 699-710. offer great promise for the resource manager. Gower, J.F.R., Borstad, G.A. and Truax, D., "Optical imaging of the sea surface with high spectral resolution," fVoc. 9th Cana- dian Symposium on Remote Sensing, 1984, 145-154. Gluch, T., Piwowar, J., Till, S.M. and O'Neil, R.A., The 7. Summary bathymetric estimator search technique for processing airborne lidar hydrography," Proc 8th Canadian Symposium To adequately address its own resource management on Remote Sensing, 1983. challenges, Canada has become one of the leaders in the Hoff, R.M. et al, "Ground-based air quality measurements," development and application of airborne electro-optical Atmos. Environ., 16, 1982, 439. sensors. Airborne systems provide flexibilityan d efficiency Hoff, R.M. and Gallant, A.J., "The use of an available SO2 tracer of operation, and offer a cost effective tool for the resource during the 1983 CAPTEX experiment," Almos. Enuiron 19, manager. The high performance spectral, spatial and 1985, 1573-1575. radiometric characteristics provide information content that Lee, H.S., Zwick, H.H. and Till, S.M., "Gas filter correlation is not available from satellite systems, while the parameters instrument for the remote sensing of gas leaks," Rev. Sci. Instrum. 56, 1985, 1812-1819. of the airborne system such as pixel size and spectral band McColl, W.D., Neville, R.A., Bonke, C.A. and Knudsen, DC, can be readily optimised for the specific application. The "Remote sensing of the Vinland natural gas blowout," Proc. digital format of the data allows for efficient processing and 9th Canadian Symposium on Remote Sensing, 1984, 245- interpretation, and is readily compatible with geographic 251. information systems. McColl, W.D., "Enhanced CCRS data acquisition capabilities," Important developments have occurred in Canada with Remote Sensing in Canada, 13 (4), 1985. both laser-based systems and multi-spectral imagers, and McColl, W.D., McKibbon, R.A.E. and Till, S.M., "CCRS Remote these systems offer the potential of revolutionising the Sensing of the Beaufort Sea Dispersant Trial", Proc. 10th resource survey industry worldwide. Arctic Marine OilspiV Program Technical Seminar, 1987, 291- 306. Moffat, A.M., Robbins, J.R. and Bamnger, A.R., "Electro-optical sensing of environmental pollutants," Amos. Environ 5, References 1971, 511-525. Neville, R.A., McColl, W.D. and Till, S.M.. "Development and Banic. J., Sizgoric, S. and O'Neil, R., " Airborne scanning lidar evaluation of the MEIS D multi-detector electro-optical bathymeter measures water depth" Laser Focus/Electro- imaging scanner," Proc. SPIE, Advanced infrared sensor Optics, February 1987, 48-52. technology, 395, 1983, 101-108. Borstad, G.A., Edel, H.R., Gower, J.F.R., and Hollinger, A.B., O'Neil, R.A., Buja-Bijunas, L. and Rayner, D.M., "Field perfor- "Analysis of test and flight data from the fluorescence line mance of a laser fluorosensor for the detection of oilspills," imager," Canadian Special Publication of Fisheries and Appl. Opt, 19, 1980, 863-870. Aquatic Sciences, 1985, 83, 38 pp. O'Neil, R.A., "Field trials of a lidar bathymeter in the Magdalen Casey, M.J. and Vosburgh, J., "Chartmaking with LARSEN." The Islands," Proc. 4th Laser Hydrography Symposium, 1980. Canadian Surveyor, 40, 1986, 251-260. Salisbury, South Australia Casey. M.J , O'Nei!, R.A. and Conrad, P., The advent of O'Neil, R.A., Neville, R.A. and Thomson, V., The Arctic marine LARSEN," Proc. 1st Biennial Canadian Hydrographic Confer- oilspill program remote sensing study". Environmental Protec- ence. 1985 tion Service Report. EPS 4-EC-83-3. 1983, 1-257, Canada Edel, H and Bianchi, H. (Co-ordinators). Summary Report of the Reichle, H.G., Beck. S.M et al, "Carbon monoxide measure- Workshop on Remote Sensing of Fluorescence Signals, ments in the troposphere," Science. 218. 1982, 1024-1026. October, 1986, Department of Fisheries and Oceans, Canada. Ryan, J.S. and O'Neil, R.A.,"Field trials of an airborne lidar Gibson. JR.. O'Neil, R.A., Neville, R.A., Till, S.M. and McColl. bathymeter," Proc 19th Annual Canadian Hydrographic W.D.. "A stereo electro-optical line imager for automated Conference, 1980. mapping", Proc.6th International Symposium on Automated Schreier, H., Lougheed, J., Gibson. J.R and Russell, J.. * Calib- Cartography, II, 1983, 165-176. rating an airborne laser profiling system," Photogr. Eng. and Gibson. J.R., "Processing stereo imagery from line imagers," Bern. Sens., 50. 1984, 1591-1598 Proc 9th Canadian Symposium on Remote Sensing, 1984, Seigel, HO. and Robbins, J.C., "Exploring with Luminex," 471-488. Mineral Engineering, July 1983. Gibson, J.R., "The use of auxiliary data in photogrammetiic Till, S.M. and O'Neil, R.A.. "Develooment of an airborne opera- adjustments, Progress in Imaging Sensors," Proc. ISPRS tional laser fluorosensor for use in the Canadian offshore and Symposium, ESA SP- 252, 1986. 583-588. Arctic.'Proc. 8th Arctic Marine Oilspill Program Technical Gibson, J.R. Park. W.M., Hollinger, A.B., Dunlop, J.D. and Seminar, 1985,287-291. O'Neil, NT., "Geometric correction of airborne line scanner Till. S.M., McColl, W.D. and Neville, R.A, "CCRS Airborne B-ll I'll

electro-optical facility." Proc. lOth Canadian Symposium on GASPEC," Appl. Opt 14, 1975, 2896-2903. Remote Sensing, 1986, 497-503. Zwick, H.H., de Vdliers, J.N. and McCoU, W.D., Laboratory Tdl, S.M., Neville, R.A., McColl, W.D. and Gauthier, R.P., "The evaluation of the prototype MBS, Report 78-5, Canada MEIS II pushbroom imager- four years of operation, Progress Centre for Remote Sensing, Department of Energy, Mines in Imaging Sensor," Proc. ISPRS Symposium, ESA SP-252, and Resources, 1978, Canada. 1986, 247-253. Zwick, H.H., "Evaluation results from a pushbroom imager for Ward, T.V. and Zwick, H.H., "Gas cell correlation spectrometer. remote sensing," Can. J. Remote Sensing, 5,1979,101-116. B-13

HAPPING MINE WASTES WITH LANDSAT IMAGES CARTOGRAPHY OES DECHETS DE MINES AU MOYEN D1 IMAGES LANDSAT COPY H.D. Moore, J.H. Adams and A.F. Gregory Gregory Geoscience Limited Ottawa, Ontario RESORS ABSTRACT

Techniques for visual classification and mapping of mine wastes from LANOSAT and supplementary data were developed and tested by the authors in 1974-75- These techniques were subsequently used to complete an inventory of mine wastes across Canada. The work was completed under contract to the Canada Centre for Mineral and Energy Technology.

Surficial materials at mine sites were subdivided into four classes of mine waste (overburden, tailings, waste rock and slag), two classes of water, two classes of vegetational cover and two classes of mining facilities. 718 mine sites were studied and classifiable wastes were detected at 399 sites. Inventory sheets were prepared for all detectable disposal areas larger than one hectare and 1:50 000 maps were prepared for all such areas in excess of 10 hectares.

The total area of mine wastes in the inventory is '(7 233 hectares, which is 0.004% of the area of Canada. Of that total, about 46.8% is overburden, 37-3% tailings, 15-3% waste rock and 0.6% slag. About 14.8% of those wastes have vegetational cover.

Total cost of development and inventory was about $91 per mine site in the inventory, The i nven- tory alone cost about $60 per mine site and required about 266 man days of work.

RESUME

Des techniques de classification visue lie et de cartographie des dechets de mines a partir d1ima- ges LANDSAT et autres donnees ont 6t6 developp6es et testees par les auteurs en 1974-1975. Ces techniques ont ensuite 6t€ utilisees pour dresser un inventaire de tous les dechets de mines a la surface du Canada. Ce travail a ete ex£cut£ sous contrat passe avec le Centre canadien de techno- logie mingrale et energetique.

Les materiaux de surface des sites miniers ont ete subdivises en quatre classes de dgchets de mi- nes (morts-terrains, residus, dechets de roches et scories), deux classes d'eau, deux classes de couverture vegetale et deux classes d'installations minieres. Au total, 7'8 sites miniers ont ete etudies, et 1'on a dfecele des dechets class ifiables 3 399 sites. Des releves d'inventaire ont fete prepares pour toutes les zones d'fevacuation d'une superficie superieure a un hectare et des cartes au 1:50 000 ont ete dressees pour les zones d'evacuation d'une superficie depassant 10 hec- tares.

La superficie totale des dechets de mines inventories est egale a 47 233 hectares, c'est-d-dire 0,004% de la superficie du Canada. Sur ce total, on compte environ 46,8% de morts-terrains, 37,3% de residus, 15-3% de dechets de roches et 0,6% de scories. Environ 14,8% de ces dechets ont une couverture vegetale.

Le coDt total de developpement et d'inventaire s'est eleve a $91 par site minier inventorie. A lui seul, 1'inventaire a coGtfe $60 par site minier et a entraTne ('equivalent d'environ 266 jours- hommes de travaiI.

"Presented to the 4th Canadian Symposium on Remote Sensing, Quebec City, May 1977". B-14

INTRODUCTION Narionn] Inventory

Mine v.istos comprise a paradox in that they A specific LANDSAT-bascd project, as warranted nrc an inevitable product of successful by the results of the Calibration Phase, to mining operations which serve the public need; provide an initial base-line study of all however, the disposal and rehabilitation of mining areas in Canada, including former such wnstes is an operating expense which productive sites. may decrease the economic viability of that mine. Specific mine wastes may be relatively HIKING WASTES large in area and, locally may be considered a public nuisance or even a hazard. On the Mining waste is a collective term that refers other hand, some wastes are reclaimed and to the mineral and rock products that are modified to serve as airports, recreational rejected from extraction and processing oper- sites and farmlands. ations. Such rejects are commonly dumped in piles or disposal areas adjacent to the Rehabilitation of mine wastes is nov being facilities that produced them. These rejects practised by many mining companies 1.1 Canada are: tailings, spoil or transported over- and waste management is under study by the burden, slag and waste rock. They all may mineral industry and several levels of govern- contain admixtures of other refuse also. In ment. The total area of mine wastes is addition, gaseous or particulate wastes may be believed to comprise a relatively small dispersed widely in smoke plumes from smelters fraction of the land area of Canada (c.f. and waste waters may be discharged into Leroy, 1973). However, the actual area of natural drainage systems. such waste and subtotals for different types of waste and amount of vegetation cover were Tailings, the principal waste product, are imprecisely known in 1975. Thus a general produced as a slurry of sand, slimes and water inventory of the location, size, types of during the physical and/or chemical concentra- waste and degree of vegetative cover was an tion of the valuable products. The water is essential input to defining the scale of the decanted in the disposal areas and the tailings problem for either national or local manage- eventually solidify into a granular residue. ment of mine wastes. Being man-made, they are much more uniform in their physical properties than are most natural Remote sensing from Earth-orbiting satellites deposits. The quantity of tailings produced has a recognized potential for providing annually by the Canadian mining industry is information about the size and type of wastes estimated to be in excess of 400 million tons and the amount of vegetative cover. A number (Hoare, 1972). of mapping techniques were tested on three test sites by Gregory Geoscience Limited. Spoil or transported overburden occurs exten- From these tests, specific techniques were sively in the vicinity of open pits and strip developed and used, first in an experiental mines. Such mining operations T< ^uire the inventory, and then in a national inventory of prior removal and storage of large quantities mine wastes. of oveiburden before mining of the valuable product can begin. OBJECTIVES OF STUDY Slag is a vitreous mass that solidifies from Calibration Phase molten wastes resulting from snelting or blast furnace operations. Pyrometallurgical plants (a) Two parallel studies (LANDSAT and air- are not as common as mines; hence, slag dumps borne) to define the advantages and limita- occur much less frequently, though often tions of remote sensing for monitoring mine prominently. waste dumps in three test areas at Elliot Lake, Sudbury and Bristol. Waste rock is barren or uneconomic rock pro- duced during the extraction of ore and other (b) Temporal, spectral aiid spatial analyses valuable mineral products. It may be used to to determine the minimal size and contrast refill openings in the mine or as fill around that can be recognized and the optimal scale the plant site. for such studies. Also to determine the feasibility of measuring area, vegetative Waste disposal areas are commonly located ad- cover, water cover, other surficial character- jacent to the facilities that produced them istics and rate of change. because it is not economic to transport wastes B-15

further than is necessary to protect the is no indication that various types of waste mining operation and the public. Commonly, might be classified except very broadly (e.g. this distance is less than two miles. Dis- stripped land). Further, none of the tech- posal areas frequently fill in topographic niques had been developed for mine wastes in depressions. This may require the construc- the Canadian environment nor applied systemat- tion of dams, dykes or berms to retain the ically to such analyses. waste. Occasionally waste may be contained wholly within retaining structures where suit- Accordingly, three test areas were selected to able topography is not available. Most waste serve as a basis for preliminary assessment of dumps are less than one hundred feet high but the published techniques, using both visual a few have attained heights of over 1500 feet. intepretation and available computer programs. Field surveys and airborne sensing within Barren waste disposal areas have long been these test areas served to calibrate and test considered unavoidable evils associated with the relevance of preliminary interpretations any mining operation. However, it is now of LANDSAT data. As a result of these becoming standard practice for mining compa- studies, new visual techniques were developed nies to plan and establish permanent, self- to use LANDSAT data for mapping mine wastes. sustaining vegetation on rebuilt landscape These techniques were then tested in a proto- sculptured from mining wastes. type inventory which extended to the full frame for each LANDSAT image that included test DEVELOPMENT OF TECHNIQUES FOR MAPPING MINE sites. WASTES BY REMOTE SENSING Description of Test Areas Introduction Three test areas were chosen for the initial A literature search revealed that current evaluation of the use of remotely—sensed data techniques for assessing mine wastes using in the inventory and subsequent monitoring of LANDSAT data can provide the following types mine wastes. Separate sites within the test of information under specified conditions and areas, were selected at four uranium mines at with some prior knowledge of the area: Elliot Lake and six Ni-Cu sulfide deposits in the Sudbury area, as well as the Hilton iron (1) location of probable mine dumps; mine (magnetite) near Bristol, Quebec. A wide range of waste materials and mining methods (2) area of dump (minimum, about 2 to 5 are to be found within these three areas. acres); Examples of reclamation of wastes in various stages are included also. Mining companies in (3) percent of vegetative cover on the each of these test areas assisted in the field larger dumps? studies and, in some cases, provided detailed maps. In addition, about 200 field photo- (W) location and size of mine-related water graphs in colour were talcen during relevant bodies; field studies.

(5) differentiation between deciduous and Hilton Mine Area: The Hilton Mine, which is coniferous cover; and located on the north side of the Ottawa River near Bristol, Quebec, is an open pit operation (6) recognition of changes with time (active producing iron ore. Waste rock, ore, mill vs derelict, revegetation, etc.) reject, and tailings are stored around the mine. Large areas of pit waste, as well as All the techniques, however, appeared to part of the tailings pond, had been sown with require further development for operational grass. use. For example, mine wastes and disposal areas have not been shown to have spectral Elliot Lake Mining Area: Four underground characteristics that are clearly separable uranium mines were selected as test sites in from other types of wastes and clearings the Elliot Lake area. Because of the mining (e.g. lumber mills and sawdust). This is methods used, very little waste rock is particularly true for small disposal sites brought to the surface. which are most numerous. Further, consistent measurements of vegetative cover, classes and Tailings is, by far, the most abundant waste areas has not been demonstrated, especially from these mining operations. They are all for different seasons nor have climatic stored in the natural drainage system which effects, such as prior rainfall or haze con- has been controlled by dams. current with imaging, been assessed. There B-16

The Sudbury Area: In the Sudbury area throe NATIONAL INVENTORY sub-study areas verc used: separate mining and snelting operations at Copper Cliff and The test inventory was found to be a clear at Falconbridgc, both on the south side of the presentation of important data about the dis- Sudbury Basin, and a complex of mining and tribucion of mine waste. With these favourable milling operations in the Levack area on the results it was decided that a national inven- north side of the Basin. tory of mine wastes in Canada should be under- taken. At Copper Cliff and Falconbridge, three types of mine waste are stored at disposal sites: Outline of Methodology tailings, slag and waste rock. At Levack the major waste is tailings. A complete search of Canadian LANDSAT data for mine wastes was not attempted. Rather, 718 DEVELOPMENT AND TESTING OF TECHNIQUES FOR mine sites were chosen from published maps MAPPING MINE WASTES (GSC 1252A, EMR 900A) and LANDSAT images were selected and acquired for subsequent analysis. During the calibration phase many different The sites analyzed include 680 mines with types of visual and digital classification "production + reserves" in excess of 36,000 and enhancement techniques were evaluated for tons (GSC, 1968) and 38 other mines (EMR, 1974). the purpose of mine waste mapping: The inventory thus covers sites that had had significant production as of 1974 plus a few 1. Visual classification of imagery; recent developments in the selected areas. A total of 718 sites was analyzed. 2. Density slicing; The visual interpretation of LANDSAT images for 3. Alphanumeric printouts; classifying and mapping mine wastes is based on the projection of selected transparencies 4. Digital colour composits; onto a topographic base map at a scale of 1:50,000. The common base comprised false 5. Principal component enhancement; colour composites at a scale of 1:1 million in 9" x 9" format. In the few cases where such 6. Spatial frequency filtering; composites were not available, black-and-white transparencies in the four MSS bands at similar 7. Unsupervised classification; and scale were used. For parts of eastern Canada, it was necessary to use 70 mm black-and-white 8. Supervised classification. transparencies at a scale of 1:3.6 million because other transparencies were not available From the evaluation work it was concluded at the time of interpretation. that the optimum method for mapping mine wastes was the visual interpretation of Boundaries were mapped directly from the pro- LANDSAT imagery. The 1:50,000 topographic jection onto the base map. Wastes were maps which were used as a base for the classified by integration of several different interpretation supplied geographic and topo- sources of information, e.g. calibration data graphic control to the interpretation. The from field investigations at representative criteria used in selecting this method were waste sites, previous interpretive experience cost, speed of interpretation, and ability and supplementary maps and reports that are to classify small areas which have non- commonly available. uniform spectral characteristics. Limitations of Inventory Once the optimum mapping method was selected a test inventory of nine wastes for sixty The accuracy of the classifications and maps mines was undertaken. The sixty mines were of mine wastes is affected by the quality and located in the areas surrounding the original availability of the relevant LANDSAT data and three test sites. topographic maps, by the physical and spectral contrasts between the waste and its physical The data derived for each of the mine sites surroundings and by the geometric limitations were presented on separate data sheets inherent in both the LANDSAT data and the (Fig. 1). If the mine was of sufficient size interpretive methodology. Collectively, these and complexity (usually greater than 10 hec- techniques limit the inventory to waste sites tares in area) a map of the mine site was that have an area larger than about 1 hectare produced at 1:50,000 showing the distribution and a detectable contrast relative to back- of waste material (Fig. 2). ground. Similarly, these factors limit the B-17

precision for defining boundaries and measur- mines with "production + reserves" in excess ing areas. Topographic maps were not avail- of 1 million tons and 32% of those with able at 1:50,000 scale for about a dozen sites "production + reserves" between 36,000 and In each of Ontario and Quebec. In such cases, 1 million tons. Selected field checks during base maps were prepared from the 1:50,000 calibration, suggest that the remaining mines projection of the LANDSAT image. are undetectable for one or more of the following reasons: (1) area of waste is Specific limitations are discussed in the smaller than the minimum resolution element; following paragraphs. (2) the waste is overgrown with vegetation; or (3) the mine has little or no waste because Minimum Area of low production or processing at another location. Prospect shafts, trenches and small LANDSAT images have a finite limit of piles of rock waste related to preliminary resolution which is dependent on the size of exploration are not usually detectable. the picture element (or pixel) and the alti- tude of the satellite. One LANDSAT pixel Accuracy in Measurement of Area represents an area of approximately 80 m x SO m on the ground, or about 0.5 hectares. The absolute accuracy of measurement is difficult to establish (Gregory Geoscience Contrast Limited, 1975, vol. I). For tailings disposal sites, accuracies appear to be about ± 10% for All the classified materials e.g. water, areas between 8 and 20 hectares and less than barren tailings and waste rock, vegetated + 5% for larger areas. A wide range of error waste, spoil banks and natural vegetation appears to hold for areas less than 8 hectares. have different spectral signatures (e.g. It may commonly be less than + 15% but may brightness levels in each of the 4 LANDSAT exceed + 30Z, especially for small areas bands). If two signatures are distinctly approaching the minimum of about 1 hectare. different (i.e. high contrast), the materials are separable; but if they are similar (i.e. Accuracy of Classification low contrast), they may be difficult to separate. Hater and green vegetation, for Classes of waste are based initially on the exanple, have signatures that are readily spectral signature and geometric pattern as recognized and separated, from each other presented on the LANDSAT image and as and from mine wastes, because of their high extended from calibration data for represent- contrast and spectral characteristics. On ative wastes. Confirmation is attained from the other hand, there may be little or no topographic maps and other published reports contrast between outcrop and waste rock of the and maps. same type. Experience to date suggests that, with the Detectability of Waste Sites exception of a few minor points of confusion the classifications represent real classes of Contrast and minimun area combine to limit materials with reasonable accuracy. the detectability of waste site. The brightness of a waste dump that is smaller in Surficial Characteristics of Wastes area than 1 pixel (0.5 hectares) is averaged in with the brightness of the surrounding The configuration of the waste surface is background and hence may not be detected difficult to assess with the data used in this unless it is of sufficiently high contrast to inventory. More detailed information could affect the average brightness recorded for the be obtained from relevant aerial photographs. pixel. On the other hand, if the waste is However, the present methodology, in conjunc- covered with vegetation, it may not be tion with the topographic base naps, can possible to separate the dump from the provide general information about the slope vegetated background on the basis of pixel of the surface and the adjacent topography. brightness, regardless of area. Terracing of waste rock and overburden, how- • ever, is not usually detectable. Image quality is also an important factor in detectability. For this reason, the optimal The bulk colour of the waste is estimated from images were selected for each site although the spectral brightnesses of the false colour all selected images are not of equal high composite and/or the four separate bands. quality. Vegetational cover is obvious but separation of natural vegetation from revegetation depends During this inventory, a total of 399 mines largely on patterns of cultivation or supple- were detected. These included 82% of the mentary information. However, an estimate of B-18

the cmount of vegetative cover on the wastes ed gypsuin quarry at Dingwall, N.S., could not is feasible in terms of the two broad classes: bo classified, presumably because it has been heavy and light cover. overgrown with vegetation. In addition, cur- tain wastes are associated with mineral pro- Cross-Reference Table cessing plants that are not located at mine sites e.g. slag at the former smelter in The data for the national inventory were Coniston near Sudbury, Ontario. Such wastes present on daca nheets and maps similar to are not included in this inventory. Further, those in the test inventory. To help a user as specified in the contract, the disturbed of the inventory find all the mines he is ground in the immediate vicinity of plant and interested in a cross-reference table was set mine sites was not considered as waste. New up. areas of plant site were, however, indicated on the maps of the larger sites. The cross-reference table summarizes infor- mation interpreted from the LANDSAT and The detectable wastes of this inventory com- ancillary data. The table has a hierarchical prise those which are most obvious and hence subdivision with the following levels: are most likely to attract future attention. The distribution of these wastes by province 1. Provincial or territorial jurisdiction, and territory is given in Table 1. The calcu- from east to west to north. lated total area of mine wastes in Canada is 46,916 hectares. If we assume that each non- 2. Commodity extracted at the mine or pro- detectable mine has wastes averaging 1 hectare, cessing site. Seven categories were then the total area of wastes as defined in requested: (1) asbestos; (2) coal; this inventory is estimated to be 46,914 + (3) copper, lead, zinc, nickel; (4) gold, (319 x 1) = 47,233 hectares. Our calibration silver; (5) iron; (6) potash, and (7) studies suggest that many of the undetected others, including uranium, waste sites are much smaller than 1 hectare and hence this average will probably compensate 3. Type of waste produced. Four categories for the undetected large areas of waste such as were requested: (1) tailings; (2) waste Noranda, Coniston and Dingwall mentioned above. rock; (3) overburden; and (4) slag. On a national basis, 46.8% of the wastes com- 4. Operational status of the waste disposal prise exposed overburden, 37.3% tailings, 15.3% site i.e. open or closed. waste rock and 0.6% slag. About 14.8% of these mine wastes have vegetational cover. However, In the special case of Ontario gold mine the averages vary grsatly within the provinces tailings, the closed waste sites are further and territories. subdivided on the basis of whether or not the wastes are vegetated. The average area of tailings or waste rock or slag at a mine site is about 70 hectares or All wastes at the last level of classification twice the size of the current garbage disposal are listed on one page in the cross-reference area (39 hectares) for the city of Ottawa. table. Summary descriptions of the wastes are Disposal sites for overburden are larger and the associated mine site. average about 147 hectares (c.f. Table 2). Surface areas of conical and other waste dumps The wastes are identified by association with with significant relief will be somewhat specific mine sites which are listed by narae. greater than the averages tabulated here. Company names are given for all operating mines and for those closed sites for which The estimated total area of mine wastes as current ownership was reported in the limited defined in this study (47,233 hectares) amount of literature used in the study. comprises about 0.004% of the surface area of Canada. This area of waste is about four Results of Inventory times the area of the city of Ottawa, about the same size as the area covered by railroad The inventory of Canadian mine wastes, as track in Canada and significantly smaller than presented herein, is not a complete inventory the areas covered by paved municipal streets, of all mine wastes in Canada. Rather, it is or paved highways or railroad right-of-way in an inventory of those wastes which were de- Canada (c.f. Table 3). tectable on LANDSAT images within the estab- lished limitations of the interpretive tech- Total cost of development and inventory was niques. It is known, for example, that about $91 per mine site in the inventory. The relatively large slag dumps are associated inventory above cost about 60$ per mine site with the smelter at the Home mine in Noranda, and required about 266 man days to complete. Quebec, although the slag per se could not be separated from the disturbed ground compris- ing the plant site. Also the large, abandon- Heller, R.C. (1970) Imaging with Photographic Sensors, chapter 2 in "Remote Sensing, Alexander, S.S., Dein, J. and Gold, D.P. with Special Reference to Agriculture and The Use of ERTS-1 MSS Data for Mapping Forestry"; National Academy of Sciences, Strip Mines and Acid Mine Drainage in Washington, D.C., 424 pp. Pennsylvania; "Symp. on Significant Results from ERTS-1", NASA SP-327, pp. Henkes, W.C. (1971) Satellite Monitoring of 569-576. Open Pit Mining Operations; U.S. Depart- ment of Interior - Burearu of Mines Infor- American Society of Photogrammetry (1952) mation Circular IC-8530. Manual of Photogrammetry, Banta Publish- ing Co., Menasha, Wis., 876 pp. Hoare, B. (1972) The Disposal of Mine Tailings Material, unpublished Ph.D. thesis, Dept. (1960) Manual of Photographic of Civil Engineering, Univ. of Waterloo, Interpretation, Banta Publishing Co., Waterloo, Ont., 192 pp. + appendices. Menasha, Wis., 868 pp. Leroy, J.C. (1973) How to Establish and Anderson, A.T. and Schubert, J. (1974) dialog Maintain Growth on Tailings in Canada - and Digital Techniques Applied to Strip Cold Winters and Short Growing Seasons; Mining in Maryland and West Virginia; in "Tailing Disposal Today", edited by NASA-GSEC X-923-74-313. C.L. Alpin and G.O. Argall, Jr.; Miller- Freeman Publications, San Francisco. Eush, P.W. and Collins, W.G. (1974) The Application of Aerial Photography to Lyon, R.J.P. (1965) Analysis of Rocks by Spec- Surveys of Derelict Land in the United tral Infrared Emmission (8 - 25 Microns); Kingdom; "Environmental Remote Sensing: Economic Geology, vol. 60, pp. 715-736. Applications and Achievements", ed. Barrett, E.C. and Curtis, L.F., Pub. Lyon, R.J.P. and Burns, E.A. (1963) Analysis Edward Arnold, London, pp. 167-184. of Rocks and Minerals by Reflected Infrared Radiation; Economic Geology, vol. 58, Canney, F.C., Wenderoth, S. and Yost, E. pp. 274-284. (1970) Relationship Between Vegetation Reflectance Spectra and Soil Geochemistry; Murtha, P.A. (1974) SO2 Damage to Forest New Data from Catheart.. Mountain, Maine; Recorded by ERTS-1, "Third Earth Resources "Third Annual Earth Resources Program Technology Satellite Symp.", NASA SP-356, Review", NASA-MSC 03742, section 18. vol. I, pp. 137-139.

Chase, P.E. and Pettyjohn, W. (1973) ERTS-1 Pettyjohn, W.A., Rogers, R.H. and Reed, L.E. Investigation of Ecological Effects of (1974) Automated Strip Mine and Reclama- Strip Mining in Eastern Ohio; "Symp. on tion Mapping from ERTS; "Third Earth Significant Results from ERTS-1", NASA Resources Technology Satellite Symp.", SP-327, vol. I, pp. 561-568. NASA SP-356, vol. II, pp. 87-101.

Cilbertson, B.P. (1973) Growth and Decline of Piech, K.R. and Walker, J.E. (1974) Interpret- Vegetation on Mine Dumps; Type II Report ation of Soils; Photogrammetric Eng., vol. to NASA (ERTS-1 Investigation SR. No. 577). 40, pp. 87-94.

(1973) Monitoring Vegetation Schubert, J.S. and MacLeod, N.M. (1973) Digital Cover on Mine Dumps with ERTS-1 Imagery: Analysis of Potomac Basin ERTS Data: Sedi- Some Initial Results; "Symp. on Signifi- mentation Levels at the Potoraac-Anacostia cant Results fron ERTS-1", NASA SP-327, Confluence and Strip Mining in Allegheny vol. I, pp. 577-584. County, Maryland; "Symp. on Significant Results from ERTS-1"", NASA SP-327, vol. I, Gregory, A.F. (1971) Remote Sensing: A New pp. 659-664. Look at the Canadian Environment; Can. Surveyor, vol. 25 t2, pp. 131-143. Thie, J. and Wachniann, C (1974) Remote Sensing for Environmental Monitoring and Impact Gregory, *.F. and Moore, H.D. (1975) The Assessment, preprint of paper presented at Role of Remote Sensing in Mineral Explor- ISP Symposium, Banff, 1974. ation with Special Reference to ERTS-1; Bull. Can. Inst. Min. 8 Met., vol. 68, Tiessen, H. (1975) Mining Prairie Coal and 0757. Healing Land; Canadian Geographical Jour- nal, vol. 90(1), pp. 29-37. B-21

REMOTE SENSING OF MINE WASTE

By C. M. K. Boldt1 and B. J. Scheibner2

ABSTRACT

This report summarizes five separate Bureau of Mines contract studies on the use of aerial photogrammetry, satellite transmission of in situ instrumentation information, and satellite imagery to monitor and update mine waste embankment data. The equipment used, methods applied, re- sults, recommendations, and cost analyses are presented along with a bibliography of related investigations.

'Civil engineer. ^Geologist. Spokane Research Center, Bureau of Mines, Spokane, WA. B-22

INTRODUCTION

Remote sensing can encompass a broad movement. Satellite monitoring allows spectrum of techniques including (1) pho- the embankment conditions data to be read togrammetry, which uses aerial pho- on demand at even the most remote sites, tography to obtain cadastral surveys, but image detection resolution is lim- (2) electro-optical systems, which trans- ited. Even with such limitations, remote form electromagnetic radiation into elec- sensing offers many advantages; there- trical signals to produce images, and fore, remote sensing studies were ini- (3) imaging and nonimaging sensors, which tiated by the Bureau to determine their measure an object's radiation O.)-3 Re- value for improving inspection techniques mote sensing, as it relates to mine waste and monitoring effectiveness. embankment monitoring, is the gathering This summary of remote sensing investi- of information without direct human con- gations is divided into three major sec- tact. This report concerns itself with tions: Aerial Monitoring, Remote Data aerial photogrammetry and the use of sat- Transmission, and Satellite Imagery. In ellites either as a communications trans- the "Aerial Monitoring" section, study 1 mitter of in situ instrumentation data or describes the use and results of aerial for imagery. The studies discussed in photogrammetry on an actively moving this report (2_, 4, j6-_8) were completed landslide in Oregon and on two coal ref- under contract with the Bureau of Mines, use sites in West Virginia; study 2 used Spokane Research Center, Spokane, WA. a different technique to monitor 15 coal Existing inspection techniques used by waste sites in West Virginia and Ken- Mine Safety and Health Administration tucky. Under "Remote Data Transmission," (MSHA) personnel consist of individual phases 1 and 2 describe the use and re- on-site visits. Typically, only a lim- sults of various in-place instruments, ited number of sites per day can be in- such as inclinometers and piezometers, spected owing to travel time requirements and the effectiveness with which their and the size or ruggedness of the ter- data can be transmitted from a remote rain. Aerial photogrammetry allows in- site to a collection center anywhere in spection of a number of sites and pro- the country via satellite. The "Satel- vides documented, sequential evidence lite Imagery" section assesses the effec- over time of an embankment's surface tiveness of using Landsat photos of mine changes, such as erosion, volume changes, waste locations to update and upgrade ex- and drainage maintenance; however, it may isting mine waste inventory data. not always detect sites of minor ground

AERIAL MONITORING

STUDY 1 tabulated according to their ability to meet these requirements (fig. 1). Aerial Since 1974, the Bureau of Mines has photogrammetry was determined to be the been interested in using remote sensing most promising technique for monitoring techniques to improve coal waste site coal waste embankments. monitoring capabilities (8). In 197A, the Bureau awarded a contract to CH2M Description of Work Hill to look at the feasibility of devel- oping a fast, reliable, and effective After the aerial photogrammetrie method method of measuring the stability of had been chosen, the technique was ap- coal waste embankments by remote means. plied to actual field conditions. To be Various techniques were categorized and able to obtain on-site measurements as well as the aerial photographs, it was •^Underlined numbers in parentheses re- necessary to have a site that was ac- fer to items in the list of references at cessible without disturbing production. the end of this report. Consequently, an active landslide near Monitoring methods 9lmpllll

• KE r* o NAp Not tppllci c • Suitable )(« or unknown e £ e a A Posilbly au e X Not sultabl II • bf« a 9

a c a a * a m • t t a 3 c t t 3 a * c H 3 a e u 3 a a c • 9 a * t tl * e m \ B a m 3 C 1 • u 9 tt a a \ n » • t « » e On Hit NAp NAp NAp NAp NAp

Ort alta NAp NAp NAp NAP NAP NAp • NAp NAp • NAp NAp NAo NAp • NAo A A A ft NAD MAC NAp NAo NAp NAp NAD NAD NAD NAD NAD NAD NAp NAp NAD NAo o NAp NAp NAp NAD • NAp NAp NAD NAp NAD NAD NAo NAp NAD NAD NAD NAD NAD c Primary tt pa r aonna i T • e hnlc is ft NAp NAD NAP NAp NAp NAp NAp NAp

primary NAp NAD NAp NAp NAp NAP NAp NAD NAp NAp NAo P a r•pn n a I (A NAp NAp NAp NAp NAp NAp NAP O Horli onial • A X X X X X X X A R) •

D Sloe* V a r 11 c a 1 • • X X ffl Tilling X X X X X X A UJ A A X • • • • • X X X X X X • • X X X X

Flow mi X X X X • • NAD

A A X X X X X X NAD S a a p•0* Haw ouTI• 1 • X X X X X X X X X X

1 n 1 • r Itftfl »• o* la 1 la • • X A X X X X X X X X X X X • X X X X • X NAo X X A Pla i oma I Me Iftf • ' Iron tulUli • • A A A X X X X X X X X X X • X E la v a 1 • UJ • • X A X X X X X X X X X X X •

FIGURE 1.—Matrix of monitoring methods. Monitoring method*

Slmpllf t.d R.mot. t.n.lno

i KE^i N e e » 3uMsbl« t a k Poi slbly au tabi« m X Not suitable 0 H High *A Medium • Low 1 o non 0 None

« a f k k • c \ 1 , * • • 1 t 9 • • " I e m ) • * " t • u 1 • • • • o m * J 0 • M t

• A A A X X X _ X X • X X X X X A X AD

X X • X NAD

ftt|an X X X X X X X X X • C • Itrnil *'* •' •. NAD • lability Crtchlfig Ji A X X X X X X X X X X X X X • X NAD fldiCi tor• Impending , 0*»r 1 Opplng * X X • X X A X X A X NAD indicator s

•flfleV noi«* • X X X X X X NAD

fit* X X X X X X X X X X A X NAD Embankman l stabilit y

X • X X A X X A X NAD

I ww.v.'r" •"" < o inll.il eon A M H H H H M M L H H U I I H 1. M M L L I I NAD Coal 3 Operating coil * M M H H H H H H M H M H M M M M M M M M M NAo ffl Ottt e O..OC Hon H H H H H H L L L L L M L M M M H M M M M M NAe LU ...... M M M M H H L I M M L H M I L M H M H H M H NAD

M M H M M M M M M U M M NAP ...... M M H H H H H H NAD

n*llabiMl T (long* vi 11) H H H H H M H H H H NAD

AlUviilhn e • D • b > • 11 y M H M NAe Practicalit y NAD

Ouanltf y vatum* • • • NAD

9 ucc * * * TO *••••• • tftBlllt y 0 H H 1 0 0 0 0 0 0 L 0 I

FIGURE 1.—Matrix of monitoring mtttiodt—Continued. B-25

Roseburg, OR, was selected as the primary period of monitoring, July through De- site. Two stable coal refuse sites were cember 1976. selected for later evaluation. The landslide was monitored with 22 Equipment and Instrumentation targets situated on the actively moving surface and 13 targets acting as control The aerial monitoring technique incor- points off the active area. All targets porated the least-squares computation for and control points were surveyed to calculating ground coordinates of targets first-order accuracy, and the coordinate on an embankment from measured coordi- system was determined with a least- nates of their images on three overlap- squares computation. The site was moni- ping aerial photographs. The flight of tored once a month from February to May the fixed-wing, low-altitude aircraft 1976 by comparing aerially derived co- used vertical and, for more accuracy, ordinates for each target to actual sec- convergent photographs (fig- 2), as op- ond-order (±0.02 ft) ground surveys. Us- posed to the more conventional, vertical ing a higher order accuracy, 20Z of the only, 602 overlap photographs. aerial system's coordinates were within Typical flight lines were developed to ±0.05 ft of the coordinates determined by optimize mapping accuracy of coal refuse ground surveys, 802 of the total read- sites (fig. 3). Ground targets 11 in. in ings were within ±0.10 ft, and 992 of the diameter, painted nonreflecting white and readings were within ±0.25 ft. with an anchor pipe, afforded high visi- Data from the Roseburg landslide indi- bility (fig. 4). However, these targets cated that aerial monitoring could work would be unsuitable in winter snows. on active, steep terrain; therefore, two Other equipment used during the study coal refuse embankments in West Virginia included— were selected as the next evaluation sites. 1. Wild RC-8,4 6-in focal length map- The Wharton refuse embankment is 500 ft ping camera with a 9- by 9-in format high with a 1,500-ft crest length and size. covers approximately 50 acres. It was 2. Monocomparator which measured X built between 1956 and 1976 with aerial- and Y image coordinates to the nearest tram-deposited material and contained an 1 x 10~6 m. upstream pond. Thirty-six targets were 3. Optical stereoscope to gather qual- installed and ground-surveyed on the em- itative information from the aerial pho- bankment and at seven control points. tographs; for example, cracks, bulges, During monitoring, 20 targets were de- slumps, erosion, and seeps. stroyed by mine activity. Ground surveys 4. Fixed-wing aircraft. were conducted at the beginning and end 5. Minicomputer to reduce data points of the monitoring period. to coordinates. The Stirrat refuse embankment is 490 ft 6. Kodak Double-X Aerographic 2405 high with a 1,700-ft crest length and (estar base) black and white (B+W) film. covers approximately 40 acres. It forms 7. Kodak Aerochrome Infrared 2443 (es- an impoundment with 250 ft of freeboard tar base) color infrared (CIR) film. and was constructed via aerial tram be- tween 1945 and 1970. Since 1970, mixed Results coal waste and fine slurry have been hy- draulically discharged behind the dam. The data and results of this aerial Thirteen targets were installed on the monitoring effort are described in more embankment and at four control points. detail in the full report (J5); however, Stirrat was used as a regular inspection the significant factors that must be con- prototype- Ground surveys of targets were conducted only at the beginning and ^Reference to specific products does end of the monitoring period to determine not imply endorsement by the Bureau of if any had moved during the 6-month Mines. B-26

Left convergent Right convergent photo Vertical photo photo

FIGURE Z—Covergent aerial flight scheme.

sidered when using aerial monitoring 4. CIR film is sensitive to tempera- of waste embankments are summarized as ture variations and humidity and must be follows: kept in a refrigerator or freezer with 1. Depending on the area to be moni- adequate thawing out time prior to use tored, the camera, and the flying height, to avoid moisture condensation during the average scale of the aerial photo- exposure. The color resolution also de- graphs upon which all photogrammetric teriorates in poor weather, and the measurements were based was 1:4500 to longer shutter speeds needed for low-sun- 1:5400. The maximum area encompassed by angle shots in the fall and winter cause a flying height of 2,700 ft above the blurred targets under magnification. mean terrain would be 4,000 by 4,000 ft. 5. Neither B+W nor CIR film proved 2. The accuracy of producing topo- clearly superior to the other for ei- graphic contours from aerial measurements ther quantitative or qualitative was 1:500 to 1:2000 of the flying height. interpretation. 3. B+W film needs no special storage 6. The theoretical accuracy of consideration and is conducive to repro- 1:35,000 for convergent photogrammetry duction. However, because of the eye's was not achieved. limited response to varying tones of 7. The photogrammetric method of moni- gray, it is difficult to differentiate toring produces a permanent, visual, and, variations in, for example, vegetation most importantly, objective record of the and moisture. site. B-27

Lett Right convergent convergent exposure exposure

Typical Typical perimeter of flight line Vertical I embankment exposure \

LEGEND •*> Direction of flight lines Aerial photograph taken at this location

FIGURE 3.—Typical exposure layout.

3/4-in bolt by 1-1/2 In long

Flat washer

11-in-diam aluminum disk, painted flat white

Multiple-set expansion anchor

1-in galvanized pipe by 3

FIGURE 4.—Ground target detail. B-28

8. The cost of aerial monitoring 10. Total costs for aerially monitor- ranged from A0% of conventional ground ing a coal waste site would vary greatly surveys on the Oregon landslide to 15Z on depending on the number of sites to be the two coal refuse sites. monitored, the number of flights, the 9. Aerial monitoring costs were 175% area to be covered, the target installa- to 300% more than costs of the current tion, and the ground survey (table 2). Mine Safety and Health Administration (MSHA) routine (table 1).

TABLE 1. - Comparison of possible inspection programs (based on 1975-76 costs)

Staff capability, Average cost Program embankments per month per embankment (3-nember staff) 60- 90 $150-$200 3- 5 2,000-3,500 90-130 350- 500 D Combination, current methods and rapid 70-100 230- 330 *Based on data from MESA District A (now MSHA). 2Using 1 film type.

Program A remarks: Program B remarks:

1. Inspections are qualitative only. 1. Capable of detecting and monitoring 2. Written descriptions of conditions movement. are recorded on standard forms. 2. Capable of mapping and determining 3. Possible unsafe conditions may be quantities. overlooked or misinterpreted be- 3. Provides written record. cause of lack of physical measure- A. Time consuming. ments or lack of experience of the 5. Labor intensive. inspector. 6. Costly. A. Economical.

Program C remarks: Program D remarks:

1. Capable of detecting and monitoring 1. Capable of monitoring more embank- movement. ments than current inspection 2. Capable of mapping and determining methods. quantities. 2. Provides qualitative information 3. Provides permanent, visual record. on all embankments and quantita- A. Provides qualitative data. tive information on selected 5. Economical in comparison with con- embankments. ventional ground survey. 3. Inspections would allow for better 6. Enables experienced people to view allocation of field inspection re- (via photos) many embankments per sources for suspect embankments. month. A. Method would be more economical than rapid monitoring alone. B-29

TABLE 2. - Initial cost6 of monitoring are recommended in order to obtain opti- (based on 1975-76 costs) mum results:

Rapid monitoring system: 1. More reliable results can be Airplane $55, 000 achieved if the camera is fixed on a ro- Aerial mapping camera...... 75,000 tating mount within the aircraft than by Monocomparator...... 30,000 attempting to tilt the aircraft to obtain Total cost 160,000 convergent angles. Conventional ground survey: 2. Ground targets should be a minimum Theodolites (2), at $5,000 of 1:2000 of the flying height above the each 10,000 mean elevation of the embankment to opti- Engineer's level 1,500 mize monocomparator results. Electronic distance-measuring 3. An accounting system to replace equipment 8,000 lost or damaged targets must be included Miscellaneous equipment 3,500 in a monitoring scheme of this type. Total cost 23,000 4. Control points should be surveyed after the flight lines are determined so Certain disadvantages of the aerial that they can be located as near the four monitoring technique became evident as corners of the photos as possible. the project proceeded: 5. To obtain the best target read- ings, the aircraft should fly directly 1. To achieve the accuracy desired, toward the face of the slope; that is, convergent photography was used. the flight lines should be perpendicular 2. This type of photography required to the crest of the embankment. four flybys per site (three for conver- 6. To obtain the best results for gent and one for stereo), and the accu- stereophotography, the flight lines racy was dependent on the capability of should be parallel to the crest of the the aircraft to tilt at a specified angle embankment. twice over each site. 3. Movement on the embankment, indi- STUDY 2 cating instability, could only be de- tected at the target itself. Study 2 of the Bureau's aerial photo- 4. The targets, especially those grammetric investigations began in 1979, placed on the embankment, were highly monitoring 15 coal waste sites once a susceptible to damage and to surface month for 10 months through seasonal movement due to construction activity, changes (6). This contract investigation vandalism, or looseness of the embankment was performed by Chicago Aerial Survey. material. The method for taking photos and the 5. Aerial monitoring is also highly technique for obtaining elevations of the dependent on weather. targeted site were different from those 6. Haze, cloud cover, snow, low sun of the first aerial study. Specifically, angles, and other problems decrease accu- this contract was intended to determine racy and hamper or preclude the ability the best procedure, the level of accu- to take photos. racy, and the costs of using aerial 'pho- togrammetry to monitor coal refuse dispo- Recommendations sal sites in comparison with current MSHA inspection practices. As described in the full report (8), the following procedures and conditions B-30

Description of Work at 1:5400 and to 2,700 by 5,250 ft at 1:9000. (The ground control targets out- In this study, large-scale, low-alti- side the refuse pile area must be in- tude, aerial photos were analyzed for cluded in the dimensions.) vertical elevations through an analyti- From the aerial photographs, orthopho- cal stereoplotter- In this technique, a tographs were produced. These are aerial stereoplotter operator benchmarks known photos of fixed scale in which all dis- elevation coordinates on control targets tortions and displacements have been cor- surrounding the coal refuse site find thus rected (camera tilt, terrain-relief dis- can determine the elevation of any other placement, etc.). The orthophotograph is point on successive orthophotos. This then a scaled picture map on which one technique is not dependent upon targets can directly measure distances and, when being placed on the investigation site, overlaid with contour information, pro- as was the case in the first study. In- duce elevation readings at any point. stead, control targets off the sites in Figure 5 shows an orthophoto flight map question were ground-surveyed at the be- with topographic contours overlaid. The ginning of the monitoring period for ref- contours were generated from compilations erence X and Y coordinates and eleva- of the stereoscopic models. tions. Use of such reference points out X, Y, and Z coordinates were read off of the movement area was a distinct ad- the aerial photos using an analytical vantage because they were not affected by stereoplotter. For this project, a grid any movement on the target area or by system of 100-ft squares was optically other disturbances such as construction overlain on the orthophoto, and coordi- activity- nates were taken only on the cross grids Fifteen coal waste sites in West Vir- for monthly comparison of movement (fig. ginia and Kentucky were monitored once a 6). month for 10 months using B4W aerial pho- Various configurations for displaying tography and four separate times using the produced data were attempted. One CIR. Four field inspections of the option consisted of computer listings of sites, using procedures similar to MSHA's each grid point with values greater than inspection procedures, were made. The 5 ft highlighted. Another method graphi- inspections familiarized the observers cally displayed the displacement mesh with the sites, enabling them to compare isometrically, displaying vertical move- the results of the stereo and photogram- ment (fig. 7). The displacement mesh metric observations with ground observa- tended to exaggerate small amounts of tions. The ground inspections were con- movement. Also, it was very difficult ducted on a seasonal basis, as was the to orient the isometric mesh with any of CIR photography. This format allowed a the photographic products (orthophotos, determination of the effects of seasonal contact prints, etc.) to indicate at a changes and the correlation of aerial to glance where the movement was occurring ground observations. on the embankment itself. Low-altitude flights were conducted Another technique involved suppressing over each site to obtain the aerial pho- the line printer information and produc- tos. Because the refuse piles varied in ing a computer-drawn plot consisting pf size, various scales of photography were circular symbols which represented ver- implemented to encompass each site with tical movement. The type of symbol dis- the surrounding ground control targets in played represented positive (upward) or a single stereoscopic model. The scales negative (downward) movement. The diame- ranged from l:540C (1 in = 450 ft) to ter of the circles was proportional to 1:9000 (1 in = 750 ft), using flight al- the magnitude of the movement and was or- titudes of 2,700 to 4,500 ft above the thogonal at the same scale as the ortho- average teirain elevation. This scaling photo. The movement values were printed restriction limits the dimensions of the next to the circles (fig. 8). site to be monitored to 1,620 by 3,150 ft w i

Contour Interval 10 ft N 200 406 i Scale, ft \

FIGURE 6.—Orthophoto site map with contours. B-32

'NHCS*! B-33

/////////

//////////// ////S////S / S S S / / / S / / S S X S / / 7 / / / /////////// / / / / / / 7 .///// s / ////S////////7

0 50 100 0 5 10 1 I I 1 I I Horizontal scale, ft Vertical scale, ft

FIGURE 7.—Isometric displacement mesh.

21.8 N32

Sj r12.7 -2J 5.3 N31

5j 23.8 -12.7 -0.8 5.3 82J N30 5J 34.5 -56.0 0.8 5.3 31.5 98.7 N29 32.6 12.9 -4.8 32J 87.5 N28 -8.0 12.9 -1.4 -0.7 .8 0.6 N27 2.6 2.5 1.3 0.6 2.8 0.6 0.8 77.7 N26 34.6 66.2 2.2 3.7 0.5 U -0.7 -3.8 1.9 N25 -< 34.6 66.2 7.6 1.8 1.5 0.7 14.4 18.7 71J N24

0 25 50 1 II Approximate Vertical differential in feet horizontal scale, ft FIGURE 8.—Enlarged view ol displacement vector plot. B-34

Results 8. A flight altitude of 1,800 ft above the mean terrain will produce opti- A more detailed discussion of the in- mal photography at an average scale of vestigation, as well as complete data and 1 in «= 300 ft and expected 6pot elevation results, may be found in the full report readings as precise as ±0.15 ft, using a (_6). However, the following observations stereoplotter with a C-factor rating of were considered significant factors to 3,000 or greater. this type of monitoring: 9. Vegetation can be a serious prob- lem. By the end of the summer, identifi- 1. Sites of greater dimensions than cation of field control points was ex- that described would require either tremely difficult, and even impossible at higher altitude photography or additional some of the sites, because the vegetation models. Higher altitude photography had overgrown the targets. Also affected would lose resolution of ground features, were photo identification features used decreasing the accuracy of the system. as control points, such as bases of util- 2. Aerotriangulation, necessary to ity poles. The memory function of the combine the coordinate systems of sepa- stereoplotter allowed the operator to rate models into one model, introduces eliminate sending supplemental field deviations in horizontal positions, which crews to the sites. Auxiliary control are extremely difficult to maintain or points were derived from the stereo model reconstruct in precisely the same way using points visible despite the vegeta- month after month. tion. These points were then used to 3. Useful aerial photography was dif- control successive stereo setups in the ficult to acquire on a monthly basis, following months. since the area (eastern Kentucky and 10. Erosion is easily seen on aerial southwestern West Virginia) has below- photos, and its severity is also general- average weather conditions for year-round ly apparent. However, flattened vegeta- flying, and acceptable eun angles (30c tion can give an erroneous impression of or higher above the horizon) are at serious erosion. best available for 5 h daily (9:30 a.m. 11. Tension cracks are generally not to 2:30 p.m.) in June, dropping to 2 h apparent in the photos. Cracks are some- (11:00 a.m. to 1:00 p.m.) in December. times visible when enhanced by erosion, A. Because of the great differences in deposits left from condensation of vola- high and low elevations on the piles and tiles, or scarps. Scarps having more because of the normally steep slope on than 3 in of vertical displacement are the downhill side, low sun angle causes visible on barren piles. The slumps as- unusually long and very black shadows. sociated with scarps are generally visi- Crevassed or eroded areas filled with ble on the photos whether or not vegeta- deep, black shadows toake precision ele- tion is present. vation readings nearly impossible to ac- 12. Using enlargements, seepage can be quire in any stereoplotter. seen on barren piles as darker areas on 5. Sites situated on the northern the refuse. On vegetated piles, seepage slopes of hillsides or mountains are to- is sometimes hidden by the vegetation. tally in shadow during the winter months. In other instances, seepage may be in- This reduces photo definition by lowering dicated by the vegetation having been image contrast. washed away or by abundant growth. 6. Snowfall immediately prior to an 13. Aerial photography is an excel- aerial survey made it necessary to expose lent method for monitoring diversion and paint some of the control targets ditch systems and water impoundments be- black and white. cause an entire system can be viewed 7. Forward overlap of 90% in the pho- simultaneously. tography ensured the least amount of fly- 14. Actual lift heights cannot be de- ing by helping to maintain complete termined from photos alone. This can stereo coverage. cause problems because MSHA requires that B-35

refuse piles be constructed in compacted on the MSHA report form. The photogram- layers that do not exceed 2 ft in thick- metric system used in study 2 is best for ness, unless otherwise approved. locating surface movements involving 15. Equipment tracks observed in the large areas because a 100-ft grid can photos on spread refuse can Indicate that easily miss movement occurring in areas some compaction has taken place, although between intersection points. the actual degree of compaction cannot be 19. Orthophotos with contours can be determined. used to determine slope, initially from 16. Stereophoto analysis can be used the contours and later by adding changes to detect changes in slope; however, the in elevation from the computer plots. angle of the slope is not obtainable from 20. Deposition is visible on the the photos alone. orthophotos. 17. CIR photography can, at times, 21. Air photo analysis and photogram- provide more information than B+W prints. metry do not completely answer all of the Iron staining appears as a greenish hue questions on the MSHA inspection form. in well-exposed CIR photographs. The CIR also has better resolution under high The advantages and disadvantages of us- magnification. Seasonal variation in ing aerial photointerpretation in moni- conditions, particularly vegetation and toring coal waste sites are listed on the s-now, affect visibility of features, next page. Table 3 tabulates cost esti- sometimes hiding them but sometimes high- mates for aerial monitoring of coal waste lighting them. sites as compared with MSHA inspection 18. Photogrammetry (fig. 9) can be techniques. used to answer many of the questions TABLE 3. - Inspection costs per site (based on 1980 costs) 1 site 2 sites 10+ sites Labor|Direct|Total!Labor]Pirect|Total Labor|Direct|Total PHOTOGRAMMETRIC METHOD Ground control targeting 1,600 400 2,000 1,600 400 2,000 1,600 400 2,000 25 0 25 25 0 25 25 0 25 Photographic flights using black and white film 80 140 220 60 110 170 50 100 150 Color infrared film in ad- dition to black and white (assumes a dual-camera aircraft or interchange- 0 10 10 0 8 8 0 8 8 Photo 1 ah...... 17 23 40 16 22 38 15 22 37 Photogrammetry contours 80 120 200 80 120 200 80 120 200 Data processing (including 20 25 45 20 25 45 20 25 45 40 0 40 40 0 40 40 0 45 Aerial photo interpretation (by MSHA), 2 h 46 0 46 46 0 46 46 0 46 Total, recurring costs 308 318 626 287 285 572 276 275 551 GROUND INSPECTION METHOD MSHA site inspection. \ 450] 20 47O| 450] 20| 470| 450] 20 450 Purvey costs represent a 1-time expense of $2,000 per site, which will not in- crease or change as monitoring time is extended. 2Average maintenance costs per monitoring flight (variable). Locations with light or no snow require little or no target maintenance during winter months. Heavy brush must be cut during summer or autumn months. B-36

Advantages and Disadvantages of Using Aerial Photointerpretation in Monitoring Coal Waste Sites

Advantages Disadvantages

Rapid return of interpretable data. Sufficient targets of photoidentifiable control points are required for use in Results are "time-frozen" - the aerial controlling the stereo model setup; how- film preserves a record of the site at ever, inaccessibility of many sites and the time of flying. The film can be re- activity on and around the sites make the set in a stereoplotter at any time to targets vulnerable to disappearance and verify derived data relative to monitor- destruction. ing. This becomes more of an advantage over a period of time, when not only data Accuracy of elevation readings was dis- from consecutive monitoring periods can appointing. Accuracy problems are en- be compared, but also data from monitor- countered due to lack or loss of control ings months or years apart. New types of points, terrain slope, vegetation growth, data may also be determined from earlier shadows, atmospheric conditions, enforce- photography. ment of readings at specific locations, or any combination of these. Aerial monitoring reduces field time. The 100-ft grid readings are not suffi- Aerial monitoring provides the inspector, cient to define true surface characteris- through typical stereoplotting instru- tics of the pile. Predefined grid inter- ments, the opportunity to measure any or section points present only comparison every visible point on the embankment readings at these points. High and low surface. surface points (peaks and valleys) are not well defined. Aerial monitoring offers reliable rela- tive measurements economically. The selection of stereo instruments is restricted to those with a large range Minimal ground survey is required for of elevation-ineasuring capability because control of the photography. of the great differences in pile eleva- tion. Instruments should be equipped Accuracies of ground readings can be a with 3-axis digital readout to facili- function of the photogrammetric equip- tate numerical model setup and stereo ment, typically ranging from 1:2000 to observation. 1:4000, or the aircraft altitude. The cost is greater than for on-site vis- its; see table 3. B-37

I*

i.' - • •

is-

o o

k t o I • \ -/. e> UJ a. => 03 a u. - - 0)

CO " E B-38

Recommendations 10. Consistent reconstruction of the original stereo model setup is the most Based on the results from study 2 of critical operation in the photogrammetric the aerial monitoring investigation, the procedures. It is essential that control following recommendations are made: points be sufficient in number and clear- ly visible at all times. 1. Monthly aerial analysis is not jus- 11. Targets that were not survey- tified; the minimal amounts of surface coordinated during field operations movement or change detectable from month should be used as auxiliary control to month suggest it is necessary to re- points. view surface conditions periodically, 12. It is possible to use premarked but semiannual aerial inspections seem auxiliary control points. These are to provide sufficient sampling of most points created by drilling tiny holes in stable sites. Inactive sites would gen- the photo emulsion using a point marking erally require less frequent aerial and transfer device. These points can be monitoring. transferred accurately from one month's 2. Flights should occur at the lowest flying to the next, as long as the images possible, nonhazardous altitude in order are compatible. to take photographs that include one in- 13. Profiles should be determined dividual site in a single flight line. transversely and perpendicularly to the 3. Atmospheric conditions such as haze crest of the embankments. Elevation and cloud cover should be closely watched points for these profiles should be taken because they can greatly influence the at a maximum distance of 25 ft, using the photo resolution. "most favorable location for reading" 4. All targets should be placed or technique. A minimum of three parallel made visible before each flight. transverse profiles should be determined 5. Wherever possible, the surrounding for each embankment. Large embankments embankment should be saturated with con- should have five of these profiles. In- trol targets to ensure adequate reference formation derived from the profiles in- point survival through the monitoring cludes determination of slope of the em- period. bankment face, determination of actual 6. Trigonometric levels are sufficient buildup on active piles, verification of for determination of target elevations. trends noticed in examination of the ba- Because only a relative datum is impor- sic profiles, and slippage at the crest tant, spot elevations from U.S. Geolog- and buildup at the base of piles. ical Survey (USGS) quadrangle maps may be Both studies 1 and 2 were based on ele- used to provide a vertical datum. These vations determined at specific locations should be as near to the corners of the on the surface of an embankment. Because stereo model as possible. of the unreliability of elevation read- 7. Targets should be referenced by ings, enforcing point-reading locations sketch and description to three physical does not appear to be effective. objects to aid in accurate repositioning A more effective system of surface should panels be moved or destroyed. definition would allow point selection 8. Once horizontal and vertical con- by visual means, using a "most favorable trols are determined for model setup (or- location for reading" technique. That thophoto, stereophoto analysis), field is, a stereoplotter operator reads points personnel are not required for aerial that he or she can see well to ensure procedures except for maintenance of the precision and reliability of point read- targets. ings. The operator would also be able 9. Targets should be durable and se- to read an indefinite number of points, curely anchored and may be of any shape reading high and low spots and other easily recognizable in the aerial points on the surface not farther apart photographs. than 25 ft. Figure 10 compares the B-39

fixed-grid method with the "most favor- this system may be found in the study 2 able location" technique. Details of report (6).

REMOTE DATA TRANSMISSION

PHASE 1. - IN SITU INSTRUMENTATION WITH Description of Work REMOTE DATA COLLECTION BY TELEPHONE Factors considered in site selection In phase 1, a contract was awarded to were— Shannon & Wilson, Inc. (*_), to develop and demonstrate an instrumentation system 1. Location in a populated area with that could be wired to a remote data col- high precipitation rates. lection station for the purpose of cen- 2. Some certainty of measuring parame- trally monitoring the stability and seep- ter changes such as deformation, water age of one or more coal waste impound- pressure, and fluid levels. ments. Costs of 6ysten installation and 3. Height of structure in the range of long-term monitoring were also deter- 100 to 200 ft. mined. Specifically, the instrumentation 4. Cooperation of mine owner and per- system was designed to remotely monitor mission to use the waste impoundment. horizontal and vertical deformation, 5. Electrical power and telephone pore-water pressure changes, pond-water available close to site. levels, seepage through the embank- ment, and environmental factors such as These particular factors were selected rainfall, temperature, and barometric on the basis of the project requirements, pressure. the cost estimates made in the proposal,

KEY Recorded profile True surface Shadow area Recorded point

Profile by study technique

Maximum 25' spacing

Profile by recommended technique

FIGURE 10.—Typical surface profile comparing 100-ft grid to favorable point readings- B-40

and the ability to best demonstrate the local geology is composed of interbedded full capabilities of a remote monitoring layers of sandstone, shale, and coal. system. The site selected was the Lower Big Equipment and Instrumentation Branch Impoundment at Montcoal, WV, oper- ated by ARMCO Material Resources. It is Available instrumentation was evalu- a cross-valley coarse refuse embankment ated and selected for its capability of impounding about 5 acres of fine coal monitoring various parameters, its suit- refuse slurry pumped 1 mile from the No- ability to long-term measurements, and 7 Mine preparation plant (fig. 11). Its whether or not it was an electrical sen- height was about 190 ft (1,120-ft eleva- sor amenable to automatic remote monitor- tion) and will eventually be raised to ing. A complete remote instrumentation 1,175-ft elevation. The Lower Big Branch system for monitoring the stability of a Creek has been relocated to the south waste embankment was designed (fig. 12). side of the impoundment and flows into Final instrumentation installed in the March Fork Creek, which occupies the val- embankment included 7 vibrating wire pi- ley directly below the embankment. The ezometers, 2 resistance piezometers, 3 topography of the area is relatively biaxial tiltmeters (2 sensors each), 3 steep with heavily wooded hillsides. The multiple-position borehole extensometers

Scale, ft LEGEND V/yy) Coarse coal refuse embankment area # New Shannon and Wilson boring O Existing borehole and piezometer — T— Telephone line —c— Trenched cable — p— Electrical power line

FIGURE 11.—Lower Big Branch impoundment site plan. B-41

Parameter -

Signal conditioning I and pow«r ~j

Data acquisition

Communication link -

Off-«tf« c«nir«l station data monitoring and analysis facilities (in S-atll*)

FIGURE 12.—Block diagram of remote instrumentation system.

(1 with 4 sensors, 2 with 2 sensors and an Acurex Autodata 9 data logger con- each), 1 uniaxial in-place inclinometer nected to the telephone. Data were re- with 8 sensors, 1 fluid-level-monitor- corded automatically at the site on paper ing device to measure seepage at the tape and on request at the contractor's weir, 1 barometer, 1 rain monitor, and 2 office via the telephone. The pond lavel thermocouples, for a total of 21 instru- sensor initially installed in the up- ments and 37 sensors (table 4, figures stream impoundment was destroyed early in 13-14). These instruments (figs. 15-17) the program when it was covered with were connected via two Junction boxes coarse refuse as the embankment was being (fig. 18) and buried cable to the auto- raised. Other instruments and data ac- matic data acquisition system (DAS) quisition systems are discussed and com- located in a trailer at the test site. pared in the report (4_). The DAS consisted of power conditioning The goals of the field monitoring weie equipment, a signal conditioning unit, to collect -sufficient data to determine B-42

TABLE 4. - List of sensors and locations

Sensor, manufacturer, Sensor Serial Borehole Elevation, Depth, and model number No. No. location ft ft Extensometer, multiple-position MPBX-l(l) L.P.I B-9 978.8 140.7 borehole, Slope Indicator, model MPBX-K2) L.P.3 B-9 1,010.5 109.0 51891. MPBX-1(3) L.P.A B-9 1.0A4.9 7A.6 MPBX-l(A) L.P.5 B-9 1,098.5 21.0 MPBX-2(1) L.P.7 B-15 9A2.9 176.3 MPBX-2(2) L.P.6 B-15 1,042.6 76.6 MPBX-3O) L.P.10 B-6 946.7 171.5 MPBX-3(2) L.P.8 B-6 1,043.7 7A.5 1 Fluid-level-monitoring device, Weir 98172 C ) NAp NAp Leupold & Stevens, model A-71.

Inclinometer, uniaxial in place, II-l 021 B-5 985.1 134.2 Slope Indicator, model P/N50432. II-2 020 B-5 993.1 126.2 II-3 022 B-5 1,001.1 112.2 11-4 019 B-5 1,009.1 110.2 11-5 017 B-5 1,017.1 102.2 II-6 016 B-5 1,025.1 9A.2 11-7 018 B-5 1,033.1 86.2 II-8 015 B-5 1,041.1 78.2 Piezometer, electrical resistance, RP-1 A1107 B-l 942.7 176.9 Slope Indicator, model P/N 56A42. RP-2 41109 B-7 1,013.8 105.2 1 Pond level A1108 C ) 1,105 NAp sensor Piezometer, vibrating wire, Irad VWP-1 14-2 B-4 1,015.2 104.6 Gage Co., model PW-100. VWP-2 14-7 B-l 943.8 175.7 VWP-3 14-6 B-3 1,002.7 117.0 WP-A 14-9 B-7 1,015.0 104.0 VWP-5 14-3 B-l 2 942.8 139.0 VWP-6 14-8 B-l 3 930.5 125.7 VWP-7 1A-1 B-1A 920.0 73.2

Pressure transducer, Setra, model Barometer 24174 (2) 1,085 NAp 250. Rain monitor, Leupold & Stevens, NA 91871 (2) NAp NAp model A-71. Thermocouple, Pyrometric Service, TC-1 NA (2) NAp NAp model type "T" thermocouple. TC-2 NA (2) NAp NAp

Tiltmeter, biaxial, Slope Indica- TM-3(A) NA B-8 1,118.8 2.0 tor, model P/N50327-1. TM-3(B) NA B-8 1,118.8 2.0

Tiltmeter, biaxial, Terra Technol- TM-l(A) 101 B-ll 1,118.0 2.0 ogy, model 85-2032. TM-l(B) 101 B-ll 1,118.0 2.0 TM-2(A) 102 B-10 1,119.3 2.0 TM-2(B) 102 B-10 1,119.3 2.0 NA Not available. NAp Not applicable. 'See figure 11. 2At instrument trailer B-43

Overflow decant area 1.136-(t-elevation bench

8-6 (MPBX-3H rB"7 (Vwp-< «"«" RP-2) / bi1nCh"M"leV"t!On /"6"4 (Vwp-1' 8-11 (TM-I) l CVWP-2 and RP-1)-7B-3 (VWP-3)

LEGEND 20 O Cable junction station (CJS) * Borehole with instruments Scale. It MPBX Mut Iple-po s i t Ion borehole extensometer VWP Vibrating-wire piezometer RP Remittance piezometer TM Tiltrne ter Tl Trav«rsing probe inclinometer II In-place inclinometer Buried cable O Data collection platform (DCP)

FIGURE 13.—Site instrumentation location. B-44

VWP-2 and RP-1 TM-2

VWP-4 and RP-2 1.200 MPBX-3

CO TM -3 5 Bench elevation 1,120 ft

O 1.100 XI a

Z. 1.000 < > LU

eoo 100 200 300 400 soo 600 7 00 DISTANCE, fl

LEGEND ln-place inclinometer with sensor 0 o ° C oa r s e o o locations (II) coal refuse

Inclinometer casing for traversing Fine coal probe (Tl) m. r ef u se Multiple-position borehole extensometer M>x e d coarse with anchor locations (MPBX) m and fine refuse Vibrating wire piezometer or

resistance piezometer (VWP or RP) Approximate

top of rock Tiltmeter (TM)

FIGURE 14.—Cross-section instrumentation location. Derived in part from 1978 D'Appolonia report entitled "Modifica- tion to Existing Coal Refuse Disposal Facility, Lower Big Branch, Montcoal Raleigh Co., WV." (3) B-45

Steel lid

775fl Support pin Corruoated metal pipe cover

Upper pivot assembly

Portland cement and lime grout backfill

2.75-in-OD grooved ABS plastic casing

Minimum 6-in boring Universal joint

3/4-in-OD gauge length tubing

Gauge length *

Four grooves equally spaced Section A-A'

Grout valve

FIGURE 15.—In-place inclinometer installation. B-46

« in * ' "•{?MilSpec connector -Cap Corrugated metal pipe cover

Backfilled cable trench

Site material backfill or grout

-—Water-filled open standpipe

-Minimum 4-in-diam borehole 1-to 2-ft-thick bentonite seal

2-Jn PVC casing slotted Sand or pea - over lowermost 3 ft gravel backfill and wrapped with filter fabric

-Electrical piezometer sensor

Bottom cap

FIGURE 16.—Piezometer Installation. B-47

Data were to be collected at 1- or 2-day intervals. However, owing to various problems with the data logger, data had to be collected manually by calling the site to activate the instruments and take readings. Other problems included downed telephone lines, loss of the source of power from a nearby mine (a generator was substituted for a short time), damaged or nonfunctioing equipment, and site con- struction and maintenance activity. The data were processed by a DEC PDP 11/34 minicomputer and plotted with pro- grams written specifically for this par- ticular sensor configuration.

Results

Complete data and results of this study are described in more detail in the con- tract report (4_). Significant factors FIGURE 17.—TMmeter Installation. noted were—

1. Despite numerous disruptions in the monitoring process which made it dif- ficult to obtain long-term continuous data (fig. 19), several events could be observed: A. Consolidation appeared to be occurring as the embankment was raised from the 1,120-ft elevation to the 1,175- ft elevation and as the pond extended several hundred feet upstream. B. A 1.1-in deflection occurred at an elevation of 1,043 ft in the fine refuse in the embankment, probably owing to the additional fill added to the embankment. C. Pore-water pressures within the embankment remained relatively constant throughout the test. The piezometer lev- els ranged from 15 ft above bedrock in the old portion of the embankment to FIGURE 18.—Electrical cable junction box. about the fine-coarse coal interface at 1,035-ft elevation. the applicability and reliability of the 2. An approximate cost analysis for system, to provide typical embankment the system indicates that an automatic data and some site-specific data, and to system is justified if the impoundment is provide precursory data in case of an in- monitored three times a week for 8 yr stability developing in the embankment. (table 5, figure 20). A manual system B-48

Ground surlace elevation 1.119 f <

Elevation 1.120 ft Coarse retus

Mixed coarse and fine refuse •o" after filling Cross section

1,035 ft

Fine refuse

Tip elevation 1,0 14 ft »*

KEY o Water level indicator (manual) o Vibrating-wire piezometer \i.03s 18 24 6 12 18 2* S VWP-4 A M Resistance piezometer RP-2

9 8 5 JJASONDJ F M A M J JASONDJFMAMJ 1979 1980 198 1 TIME

FIGURE 19.-Water levels for piezometer B-7.

TABLE 5. - Cost analysis of manual and remote monitoring systems (1982 costs) (4)

Cost item Manual1 Automatic System design •.. $10,000 $20,000 Capital (equipment, instruments, cable, etc.). 62,000 85,000 Installation labor. 60,000 73,000 Subtotal installation costs.. 132,000 178,000 Maintenance costs .per year.. 2,000 20,000 Monitoring labor, per set2...... 116 15 Data processing labor, per set2 58 NAp Data processing labor (1st year only)3 NAp 5,000 Data interpretation .per year.. 10,000 10,000 NAp Not applicable. 'Manual costs calculated from automated system costs. 2Based on labor rate of $29/h. 3No data processing labor costs after initial year of operation.

NOTE.—All costs based on 37-sensor system. B-49

would be economical for less frequent and maintenance and instrumentation prob- monitoring. A large number of instru- lems of phases 1 and 2 were compared. ments were used for this project to determine possible installation prob- Description of Work lems and the long-term reliability of the various instruments on the market. Costs Because the instrumentation from phase in a real situation would be lower be- 1 was already in place on the embankment, cause only three to six instruments would only the data collection and transmission be required. system had to be selected and installed on-site, together with the instrument in- PHASE 2. - IN SITU INSTRIWENTATION terface circuitry. The GOES East satel- WITH SATELLITE TRANSMISSION OF DATA lite was selected to relay data from the test site DCP to the user's terminal via In phase 2, a contract was awarded to the Command and Data Acquisition Station Energy, Inc. (.7), to demonstrate a self- (CDA) and the Data Collection System-Data contained satellite communication link Processing System (DCS-DPS). The data and data collection platform (DCP) that were then reduced and plotted for 6tudy. remotely monitored the stability and seepage instrumentation located on the Equipment and Instrumentation coal waste embankment at the Armco No. 7 coal mine, Montcoal, WV. The costs, A Sutron 8004B DCP was selected because ease, and accuracy of data transmission, of it6 low power consumption, 16 analog or digital parameter inputs, and avail- ability. Additional equipment included a 12-v storage battery, two Solarex 4200EG 400 solar panels, and a six-element Yagi an- KEY tenna. The DCP, solar panels, and anten- 300 Once a month Once a week na were mounted on a support pole with Twice a week Once a day the solar panels aligned for optimum Manual system solar illumination supply (fig. 21) and CO o Automatic system the antenna aligned with the GOES East z satellite. O 100 < 90 The previously installed instrumenta- UJ SO tion continued to monitor the coal waste

FIGURE 21.-0ata collection platlorm, solar panels, Yagi antenna. B-51

operated without charge by the National channel frequency in a self-timed trans- Oceanic and Atmospheric Administration- mission mode at A-h intervals. The GOES National Earth Satellite Service (NOAA- satellite in turn transmitted to the CDA NESS) (_5) for the acquisition of envi- station ground equipment, which decoded ronmental data, provided that the data the data and checked for errors. The can be disseminated to other interested data were then transmitted via condi- parties- The system data flow consisted tioned leased lines to the DCS-DPS, which of four major subsystems: (1) sensor acted as a central data distribution fa- data collection, (2) data transmission cility by storing the data in user queues from the DCP to GOES, (3) data retrieval and providing the user with interfacing from GOES and storage by NOAA-NESS, and for data requests. The user acquired the (4) dissemination and processing of re- data from the DCS-DPS via a standard trieved data. telephone link (fig. 22). The DCP received and stored data from Data from the DCS-DPS were reproduced the various analog sensor data lines on a as hard copy and manually reduced, until preprogrammed collection interval. Data the programming for the automated data transmission to the GOES satellite oc- dissemination could be completed for the curred on a specific NOAA-NESS allocated Chromatics GC1999 computer. Completion

GOES (East)

Data transmission from DCP

User site CDA 300-bps and MODEM DCS- and data DPS processing equipment

3) Retrieval and (4) Dissemination storage of data and processing of by NOAA-NESS retrieved data

M) Data collection from sensors

FIGURE 22.—System data How. B-52

of this task allowed for data storage on harsh environment and in part because the 8-in floppy disks from which graphs could impoundment was in a continual state of be plotted in engineering units. construction and maintenance.

Results RECOMMENDATIONS—PHASES 1 AND 2

Complete data and results of this study As a result of the data compiled in are described in more detail in the con- phases 1 and 2, there are five recommen- tract report (_7_)« Although data collec- dations with regard to the overall system tion systems consisting of the DCP satel- and its geotechnical instrumentation: lite and ground station receivers have been widely used to successfully monitor 1. The system should be flexible environmental conditions (rain, wind, and enough to accept a number of different temperature) for flood control and on sensor types. buoys at sea, their use for monitoring 2. Further development in geophysical embankment instrumentation had never pre- instruments is needed, specifically in viously been documented. Major findings inclinometers, tiltmeters, extensometers, included— and piezometers, to make them more reli- able, cheaper, and easier to install and 1. The reliability of the system was maintain. very high. The DCP required no preven- 3. A standard remote monitoring system tive maintenance, and trips to the site applicable to coal and metal and nonmetal in the event of a failure were rare. waste impoundments and embankments should Data transmission had an error rate of be developed. 0.312, mostly due to a duplicate channel 4. Satellite data transmission should frequency time slot assignment. Other- be investigated if more than 10 sites are wise, the error rate wculd have been to be instrumented. 0.06352. 5. This method of remote data collec- 2. Though the initial cost of the sys- tion could be used by MSHA, mining com- tem is high (table 6), it is still lower panies, or public utilities to centrally than that of the phase 1 system; also, monitor the stability of one or more im- the system is easier to operate and main- poundments. It would be most effective tain, and it requires neither power lines either in populated areas with high over- or-' telephone lines into the site nor a all precipitation rates or local periods building to house the DAS. Therefore, of sudden extreme precipitation rates solar-powered data collection is a reli- (hazardous situations), or in remote able and cost-effective method to monitor areas where travel to and from a site is sensors at a remote waste embankment difficult. The system could act as an site. early warning device by monitoring pres- 3. Great care must be taken in select- sure changes and movements within an em- ing the site sensors and placing them in bankment caused by rising water levels. the embankment. This appears to be the It would provide more frequent readout of limiting factor in the use of this sys- coal waste impoundment stability and en- tem, and the major reason for downtime vironmental factors and would also aid in and the limited amount of data collected. inspection and control of other dispo- The tiltmeters, inclinometer, and piezom- sal sites such as metal and nonmetal eters were difficult to maintain over a waste embankments or even city reservoir long-term period, in part because these impoundments. were electrical instruments working in a B-53

TABLE 6. - Cost comparison of satellite and telephone data transmission (1984 costs) (7)

Cost item Unit cost Number Total cost of units Satellite system: Data collection platform $3,725 1 $3,725 DCP hand terminal* 630 1 630 Environmental enclosure. 220 1 220 Yagi antenna 195 1 195 Antenna cable. 55 1 55 Solar panels. 299 2 598 Power supply cable. 28 1 28 Batteries (100 A-h) 130 2 260 Wattmeter and load coil.... 250 1 250 Interface for biaxial sensors 3,000 1 3,000 Total for satellite system NAp NAp 8,961 Telephone system: Autodata 9 14,591 1 14,591 Signal conditioner. 4,320 1 4,320 AC power conditioner and filtering with temper- ature sensor and delay. 1,985 1 1,985 AC power conditioner 439 1 439 Anixter-pruzan metro tele PL1-2-2 I/F device... 123 1 123 Autodialer...... ^.. 167 1 167 Telephone modem. 570 1 570 Power line installation and lease3 NA NA NA Telephone line installation and monthly mainte- nance cost14 20- 100 112 240- 1,200 Trailer installation and monthly rental cost... 100 112 1,200 Total for telephone system2..... NAp NAp 23,635-24,595 NA Not available. NAp Not applicable. 112 months. 2Plus costs of power line installation and lease. 3In the case of this project this system was already in place; therefore the cost was not included. **Costs can vary according to length of telephone lines and weather, which can af- fect maintenance needs. SATELLITE IMAGERY A contract was awarded to Science Sys- DESCRIPTION OF WORK tems and Applications, Inc. (_2), to eval- uate the potential for using digital Four mine waste disposal sites were se- Landsat satellite data (fig. 23) for de- lected: a Florida phosphate strip mine,, tecting active metal, nonmetal, and coal an Arizona open pit copper mine, an Idaho waste and tailings disposal sites to up- underground silver mine, and a West Vir- date acreage and land use information ginia underground coal mine. The crite- previously collected for mine waste em- ria for choosing these mine sites were— bankment inventories (2). B-54

iloa* (USB) N T TF (USB) tt (USG and VMF> USB statloni

KEY ocs EOC EROS O.I. Outer EROS System*

GSFC H5S

Facility RBV

TRKG Tracking USB Upper sideband V HF

FIGURE 23.—Overall Landsat system. Source Landsat Data Users Handbook. (9)

1. An active surface or underground angle, season, currency of the informa- coal, metal, or nonmetal mining operation tion, and how closely the dates could be producing more than 500 st ore per day. matched for the two types of data. 2. A diverse set of climatic condi- 6. Availability of updated topographic tions (wet, dry, moderate) because these maps. could have different effects on the de- 7. Availability of road maps. tectability of mine waste areas. 8. Availability of USGS orthophoto- 3. Topography because of its effect quadrangle maps. on the size and type of disposal sites available to mining operations in differ- Based on the site selection data, the ent parts of the country. chosen sites ranged from a valley fill in 4. The ability to visit a selected West Virginia to a flat diked embankment site. in Florida to a terraced embankment in 5. Ease of obtaining aerial photogra- Arizona. Other mine waste sites were phy and accompanying Landsat data; these also located nearby for signature exten- can be affected by cloud cover, sun sion testing. This technique was used to B-55

check that the procedures used to study scenes for February and December 1979 and one site could be extended to other mine aerial photos for January and December sites in the same locality. 1979 were used Co study an area of 7,607 Selection of aerial photography and ha (fig. 24). Landsat data was based on the image qual- The change detection studies used image ity of the data, the time of year, the differencing to note changes at the site time of day, and the degree of cloud cov- (figs. 25-26). The greatest number of er- Most important was the need for re- changes were noted in the active waste cent coverage (after Feb. 1, 1979) be- disposal areas in the expansion of dry cause of site changes at active mine waste areas and changes in the water locations. Also necessary was the avail- areas. ability of at least same-year coverage by Table 7 shows the classification accu- both aerial photography and Landsat, pre- racies of the methods used in studying ferably as close together chronologically the Florida phosphate miring site. From as possible. this table and from details given in ref- erence 2, it is evident that the HT meth- EQUIPMENT AND PROCEDURES od can classify all the land-use catego- ries, but its accuracy was poor for the Digital data from the Landsat satellite "active mining" category (51%) and only multispectral scanner (MSS) and the Gen- fair for the "waste" category (63%). Ex- eral Electric Co.fs 1MALE-100 system in cept for the 'Vaste" category, the raw GE's Digital Image Analysis Laboratory data method classified all categories (DIAL) In Lanham, MD, were used to deter- with better than 50% accuracy. The prin- mine the capability of various image- cipal component analysis method was more processing techniques to monitor the accurate for land-use categories other waste sites. These techniques and clas- than "waste" and "reclaimed." sification methods improve the visual appearance of the image and accentuate selected features. For this study, piecewise linear contrast stretching, edge and color enhancement, and normal- ized and simple ratio techniques were used. Means and standard deviation were obtained for the land-use categories (waste, water, active, inactive, re- claimed) by using the raw data, enhanced by two-dimensional axis rotation, Hada- mard transformation (HT), and principal- component analysis. Change detection techniques were used in the Florida phos- phate region to determine their useful- ness for noting changes occurring at the site during a specified time- Each test site had four or five land- use categories, depending on the nature of the tailings and the site itself. Data analyses then consisted of comparing the various land-use categories with four classification methods using means and Scale, miles standard deviation, pixel count and area LEGEND A Active P Processing plant estimates, a classification matrix (pixel BC Bare ground W Reclaimed count), and classification accuracy. H Waste water V Vegetation The Florida phosphate mine waste test / Inactive W Waste site was 40 miles east of Tampa, FL, in FIGURE 24.—Fort Green, FL, study area: derived irom aerial an area of extensive mining. Landsat photography. B-56

Scele. miles Scale, miles LEGEND LEGEND Changed areas BHchanged areas Unchanged areas F ^unchanged areas

"1GURE 25.—Fort Green, FL, study area: automated change FIGURE 26.—Fort Green, FL, study area: manually inter- detection. preted.

Two tailings areas were studied at cop- the most inconsistently classified cate- per mines in Arizona: one at Hayden and gories owing to their spectral reflec- one at Miami. A Landsat scene for July tance complexity. 19/9 and aerial photos for April 1979 and A West Virginia coal mine having a 9.7- February 1980 were used to study areas of ha, valley-fill waste embankment with a 500 ha at Hayden and 2,000 ha at Miami. water impoundment was selected for the The classification accuracy tables (table coal waste test site. Landsat scenes for 7) for each site showed that the overall July 1980 and aerial photos for April accuracies for all classification methods 1980 were used for the study. The raw were not very good. For the Hayden site, data method (table 7) was the most accu- simple and normalized ratio methods rate classification method for this site. worked best, while for the Miami site the raw data and HT methods were the best. RESULTS The waste embankments from two Coeur d'Alene silver mines in Idaho were also Complete data and results of this study used for this study. Data from these em- are described in more detail in the con- bankments were combined since one of the tract report (2). The most significant embankments was too small for individual factors were— analysis. Landsat scenes for June 1979 and aerial photos for July 1980 and 1. No single automated digital image August 1977 created inconsistencies due processing technique would work consis- to the large time interval between the tently and accurately at each site. The dates. The most accurate method for this wide range of waste materials at a site, study area (table 7) was two-dimensional the use of processed wastes for construc- axis rotation; tailings and water were tion of roads and fill, color variations B-57

TABLE 7. - Classification accuracy of methods used to study five waste areas, percent (_2_)

Land-use category Raw Sample Normalized 2-D Hadamard Principal compo- data ratio ratio rotation transform nent analysis FLORIDA PHOSPHATE MINING AND WASTE AREA Waste 21.0 NAp NAp 50.1 62.7 49.1 Water 86.0 NAp NAp 94.8 75.9 91.2 Active mining.... 55.9 NAp NAp 44.4 51.0 66.1 Inactive mining.. 68.8 NAp NAp 63.2 81.1 85.3 Reclaimed...... 86.9 NAp NAp 0 60.0 0 COPPER MINING AND WASTE AREA, MIAMI-CLAYP00L, AZ 67.9 41.3 40.6 66.8 67. 3 71.7 Dark tailings... 43.3 79.0 43.5 52.2 23. 3 6.4 26.7 12.2 49.5 14.8 45. 1 22.4 Water 75.8 56.2 71.1 66.3 69. 5 67.7 COPPER MINING AND WASTE AREA, HAYDEN, AZ Dark tailincs... 55.6 68.1 64.5 66.6 60. 0 52.8 Water 72.9 98.6 100.0 74.3 75. 7 70.0 IDAHO SILVER MINING AND WASTE AREA Tailings. 25. 0 81. 6 66.3 85.2 11.7 74.5 77. 1 51. 4 62.9 100.0 74.3 97.1 Gypsum...... 100.0 78. 3 73.9 84.8 76.1 78.3 Water 70. 1 61. 2 67.2 80.6 59.7 61.2 Bare ground. 90. 3 80. 8 83.9 95.1 79.2 89.1 WEST VIRGINIA COAL MINING WASTE AREA Refuse. 89.5 80.7 84.2 89.5 75.4 56.1 Water.. 88.9 66.7 55.6 66.7 55.6 55.6 NAp Not applicable.

in the ponds and embankments due to the RECOMMENDATIONS mining of different ores as well as vary- ing moisture content, and reflectance Based on the results from this satel- variance in the ponds themselves due to lite imagery study, the following recom- silting or depth all contribute to creat- mendations are made: ing a complex and dynamic environment. The subtle differences that result make 1. Use of digital data from the ther- it difficult to extend a signature from matic mapper in Landsat 4 could be more one mine site to another. useful because the mapper has a higher 2. It is not possible with current resolution. state-of-the-art digital processing tech- 2. Manual interpretation of enlarged niques to inventory mine waste embank- Landsat images used in conjunction with ments on a national basis using satellite auxiliary information might be a use- imagery. ful supplement to updating mine waste inventories.

SIMMARY AND CONCLUSIONS

This report summarizes five contracted study 1 investigated the use of aerial projects dealing with remote sensing of photography and photogrammetry on an coal waste embankments. Three different actively moving landslide and on two forms of remote sensing were studied: coal embankments. Survey targets were aerial monitoring, remote data transmis- installed on the moving face, and subse- sion from in situ instrumentation, and quent photoreconnaissance measured the satellite imagery. Aerial monitoring targets' movements. Aerial monitoring B-58

study 2 investigated the use of survey telephone. This project suffered numer- targets installed off the embankment ous disruptions caused by local power faces of 15 coal waste sites in West Vir- failures and downed telephone lines. ginia and Kentucky. Placing the targets Cost estimates for such a technique, in- off the active areas of the embankments cluding internal instrumentation in the required more complex equipment: and data embankment consisting of 37 sensors, analyses but protected the targets from ranged from $144,174 for a manually read incidental movement and destruction. system to $213,015 for an automatic Phase 1 of in situ instrumentation inves- system. tigated the use of internally emplaced Phase 2 used the same internal instru- instruments to monitor embankment con- ments and embankment as phase 1, but the ditions and transmitted the data by data were relayed via satellite to a re- telephone to another geographic region. ceiving station which was then accessed Phase 2 used the same internally instru- by the user via telephone. The solar- mented embankment but transmitted data powered data collection platform, the through a satellite link to a receiving satellite antenna, and the use of the station which then could be accessed by satellite to relay data proved very telephone. Satellite imagery used digi- reliable. Costs to install a satellite tal hands at satellite data to evaluate system equalled $8,961, compared to active coal, metal, and nonmetal waste $24,595 for a telephone system. These sites to update mine waste embankment estimates did not include costs for the inventories. system design and the embankment instru- In aerial monitoring study 1 it was ments or their installation, and access found that the costs of such a monitoring to the satellite was free. technique were up to three times those of Mine waste location by satellite imag- existing conventional inspections. How- ery was an ineffective means to update ever, the benefits of having objective mine waste inventory data. At the time documentation, such as aerial photographs of the study, satellite sensors were not or photogrammetrie maps, produced over a sufficiently sensitive to detect color period of time could not be estimated. changes due to differing mineral composi- Because the targets were located within tions at the various waste locations. the area under study, they proved suscep- The remote sensing techniques studied tible to incidental movement, damage, or could be used to supplement existing mon- loss. itoring efforts and to provide visual, Aerial monitoring study 2 differed from historical documentation. Aerial photog- study 1 in the placement of the targets raphy can be used to characterize the and the mode of calculating movement. overall conditions at the embankment sur- The targets were placed outside the area face (seeps, slumps, fire, drainage ef- of interest, and computer-aided stereo- fectiveness, etc.}• Use of aerial pho- plotters were used to measure move- togrammetry can quantitatively document ment on the embankment. This protected surface movement over time. This could targets from damage and loss due to inci- be especially useful in determining em- dental movement or construction activ- bankment creep or swelling, or in esti- ity. The disadvantage of placing the mating volume. Internal embankment con- target off the active area proved to be ditions can be closely monitored using decreased visibility due to snow and in situ instrumentation (extensometer, growing vegetation. The cost of this piezometer, inclinometer, thermocouple, technique was estimated to be 17% to etc.) in conjunction with the GOES satel- 30% higher than existing inspection lite. It would then be possible to costs. initiate internal embankment readings and to transmit instrument data whenever In in situ instrumentation, phase 1 needed. featured remote data collection by B-59

REFERENCES

1. American Society of Photogrammetry Environmental Satellite Data Collection (Falls Church, VA). Manual of Remote System. NOAA Tech. Memo. NESDIS, 1983, Sensing. V. 1-2, 1975, 2144 pp. 49 pp. 2. Anuta, M. A., and 0. P. Bahethi. 6. Meisher, R. A., and R. L. Hoffman. Mine Waste Location by Satellite Imagery Improving Surface Coal Refuse Disposal (contract J0208030, Science Systems and Site Inspections (contract J018E027, Chi- Applications, Inc.). BuMines OFR 134-83, cago Aerial Survey). BuMines OFR 54-81, 1982, 100 pp.; NTIS PB 83-238519. 1980, 297 pp.; NTIS PB 81-215402. 3. Campbell, P. M., and R. G. Alines. 7. Prokoski, F. J., J. T. Byrne, and Modification to Existing Coal Refuse Dis- D. J. Bryant. Satellite Monitoring for a posal Facility, Lower Big Branch, Mont- Coal Waste Embankment (contract HO212O17, coal, Raleigh Co., WV. D'Appolonia Con- Energy, Inc.). BuMines OFR 102-85, 1984, sulting Engineers, 1978, 23 pp. 100 pp. 4. Green, G. E., and D. A. Roberts. 8. Roth, L. H., J. A. Cesare, and Remote Monitoring of a Coal Waste Im- G. S. Allison. Rapid Monitoring of Coal poundment in West Virginia (contract Refuse Embankments (contract H0262009, H0282041, Shannon & Wilson, Inc.). Bu- CH2M Hill). BuMines OFR 11-78, 1977, 113 Mines OFR 79-83, 1982, 175 pp.; NTIS PB pp.; NTIS PB 277 975/AS. 83-196584. 9. U.S. Geological Survey. Source 5. MacCallum, D. H., and M. J. Nestle- Landsat Data Users Handbook. Revised bush. The Geostationary Operational edition, 1979.

BIBLIOGRAPHY

Ahmad, M. V., D. A. Kanter, and J. V. Alexander, S. S., J. Dien, and D. P. Antalovich. Mapping of Spoil Banks Using Gold. The Use of ERTS-1 MSS Data for ERTS-A Pictures. Paper in Remote Sensing Mapping Strip Mines and Acid Mine Drain- of Earth Resources, ed. by F. Shahrokhi age in Pennsylvania. Paper in Proceed- (Proc. 2d Conf. on Earth Resources Obser- ings of Symposium on Significant Results vation and Information Analysis Systems, Obtained From ERTS-1 (Goddard Space Tullahoma, TN, Mar. 26-28, 1973). Univ. Flight Center, New Carrollton, MD, Mar. TN Space Inst., v. 2, 1973, pp. 1073- 5-9, 1973). NASA SP-327, v. 18, 1973, 1093. pp. 569-576.

Albright, J. N., P. M. Halleck, Alger, L. H., P. W. Mausel, and R. R. C. Pearson, and M. Fehler. Subsurface Herner. Identification of Surface-Dis- Subsidence Damage Monitoring: Seismic turbed Features Through Isursl Non-Para- Tomography and Microgravimetry. Paper metric Analysis of Landsat MSS Data. Pa- given at workshop on Surface Subsidence per in Machine Processing of Remotely due to Underground Mining, Morgantown, Sensed Data (5th Annu. Symp., Purdue WV, Nov. 24, 1981, unpublished, CONF- Univ., West Lafayette, IN, June 27-29, 811173-2, 9 pp.; NTIS PC A02/MF A01, DOE 1979). IEEE (Cat. No. 79CH1430-8 MPRSD), DE82010395. New York, 1979, pp. 172-182. B-60

Amato, R. V., 0. R. Russell, C. E. Ballew, G. Alteration Mapping at Gold- Weir, and F. J. Wobber- Application of field, Nevada, by Cluster and Discrimi- ERTS-A Imagery to Fracture Related Mine nant Analysis of Landsat Digital Data. Safety Hazards in the Coal Mining Indus- Paper in Proceedings of the Eleventh In- try. Coal Sec, Indiana Geol- Surv. , ternational Symposium on Remote Sensing Bloomington, IN, ERTS Prog. Off., NASA of the Environment. Environ. Res. lust. Goddard Space Flight Center, Greenbelt, MI, Ann Arbor, MI, Apr. 1977, pp. 783- MD, Rep. NAS5-21795, Jan.-July 1973, 136 790. pp.; NTIS N73-27252. Barr, D. J. Landsat Applications to American Society of Civil Engineers. Surface Mining. Paper in Proceedings of Recent Developments in Geotechnical Engi- the First International Conference on neering for Hydro Projects: Engineering Computers in Civil Engineering (New York, Geology Aspects, Rock Mechanics Studies NY, May 12-14, 1981). ASCE, 1981, and Embankment Dam Instrumentation. New pp. 384-396. York, 1981, 250 pp. Bayne. J. N., and H. Lawrence. Appli- Anderson, A. T. Evaluating the Envi- cation of Satellite Data to Surface Mine ronmental Effects of Past and Present Monitoring in Selected Counties of South Surface Mining - A Remote Sensing Applied Carolina. BuMines OFR 11-80, 1979, 141 Research Review. Paper in Proceedings of pp.; NTIS PB 80-144629. the Fourteenth International Symposium on Remote Sensing of the Environment. Envi- Boldt, C. M. K., and B. J. Scheibner. ron. Res. Inst. MI, Ann Arbor, MI, v. 1, Application of Remote Sensing for Coal Apr. 1980, pp. 275-278. Waste Embankment Monitoring. Paper in Mine Waste Disposal Technology (Proc. Bu- Anderson, A. T., D. T. Schultz, and Mines Technol. Transfer Workshop, Denver, N. Buchman. Landsat Inventory of Sur- CO, July 16, 1981). BuMines IC 8857, face-Mined Areas Using Extendable Digital 1981, pp. 40-45. l Techniques: Monitoring Surface Mines From Space. Final Report, NASA/Goddard Borden, F. Y., B. F. Mermbeck, D. N. Space Flight Center, Greenbelt, MD, June Thompson, B. J. Turner, and D. L. Wil- 30, 1975, 44 pp. liams. Classification and Mapping of Coal Refuse, Vegetative Cover Types, and Anderson, A. T., D. Schultz, N. Buch- Forest Types by Digital Processing of man, and M. Nock. Satellite Data for ERTS-1 Data. Paper in Proceedings of the Surface-Mine Inventory. Photogrammetry Ninth International Symposium on Remote Eng. and Remote Sensing, v. 43, 1977, Sensing of the Environment. Environ. pp. 1027-1036. Res. Inst. MI, Ann Arbor, MI, v. 1, 1975, pp. 133-151. Anderson, A. T., and J. Schubert. ERTS-1 Data Applied to Strip Mining. Budge, T. K. Development of an linage Photogrammetry Eng. and Remote Sensing, Processing System at the Technology Ap- v. 42, 1976, pp. 211-219. plications Center, UNM: Landsat Image Processing in Mineral Exploration and Re- Aronoff, S., G. A. Ross, and W. A. lated Activities. NM Energy & Minerals Ross. Environmental Monitoring of the Dep., Santa Fe, NM, Final Rep. EMD-78- Athabasca Oil Sands Using Landsat Data. 1138, Sept. 1980, 73 pp. Photogrammetria, v. 38, Oct. 1982, pp. 77-86. Butterworths. Field Instrumentation in Geotechnical Engineering. London, 1974, Bailey, D. J. Land Movement Monitoring 720 pp. System. Bull. Assoc. Eng. Geol., v. 17, No. 4, 1980, pp. 213-221. B-61

Carneggie, D. M., and D. 0. Ohlen. Mined Lands Reclamation. Soc. Min. Eng. A Selected Bibliography: Remote Sensing AIME, 1979, pp. 16-24. Techniques for Evaluating the Effects of Surface Mining. Technicolor Graphic Ser- Glass, C. H., R. A. Schowengerdt, and vices, Inc., Sioux Falls, SD, Mar. 1979, J. R. Carr. Inventory and Hazards Moni- 15 pp. toring of Mined Lands Using Automated Processing of Satellite Imagery and Col- Carneggie, D. M., J. A. Sturdevant, lateral Data. Office of Surface Mining, W. G. Rohde, D. 0. Ohlen, C. A. Nelson, Reclamation and Enforcement, Final Rep. J. C. Taranik, J. R. Lucas, and F. C. MGEN-79-16F, 1980, 115 pp. Billingsley. Assessing the Impact of Surface Mining. U.S. Geol. Surv. Prof. Gonzales, R. W., and M. H. Inglis. Re- Paper 1100, 1978, pp. 304-305. mote Sensing of Natural Resources. Quar- terly Literature Review, July-Sept. 1980. Chase, P. E., and W. Pettyjohn. ERTS-1 Univ. NM, Albuquerque, NM, Technol. Ap- Investigation of Ecological Effect of plication Center, Report TAX-RS-80-003, Strip Mining in Eastern Ohio. Paper in 1980. Proceedings of Symposium on Significant Results Obtained From ERTS-1 (Goddard Green, G., and D. A. Roberts. Remote Space Flight Center, New Carrollton, Monitoring of a Coal Waste Impoundment in MD, Mar. 5-9, 1973). NASA SP-327, v. 1, West Virginia, (contract H0282041, Shan- 1973, pp. 561-569. non & Wilson, Inc.). BuMines OFR 79-83, 1980, 173 pp.; NTIS PB 83-196584. Curtis, E. I. Strategy for Remote Sen- sor Use in Mining, Hydrology, and Pollu- Hampel, D. Advanced Remote Expendable tion Control. Paper in Technology Utili- Sensor/Processing Techniques. Paper in zation Ideas for the 70's and Beyond Proceedings, AIAA Sensor Systems for the (Proc. Spec. AAS-AIAA Tech. Event, Win- 80's Conference (Colorado Springs, CO, rock, AR, Oct. 30, 1970). AAS Sci. Tech- Dec. 2-4, 1980). AIAA, New York, 1980, nol. Ser., v. 26, 1971, pp. 301-315. pp. 104-108.

Earth Satellite Corp. Use of Photo In- Hardaway, T. E. Surface Coal Mining in terpretation and Geotechnical Data in the Northern Great Plains of the Western the Identification of Surface Damage and United States. An Introduction and In- Subsidence. Apr. 1975, 246 pp.; NTIS, PB ventory Utilizing Aerial Photography Col- 242468. lected in 1974 and 1975. EPA, Rep. EPA/ 0EA-76/1, June 1976, 198 pp. Elifrits, C. D. Study of Subsidence Over a Room and Pillar Coal Mine. Ph.D. Henkes, W. C. Satellite Monitoring Thesis, Univ. M0, Rolla, M0, 1980, 130 of Open Pit Mining Operations. BuMines pp.; Univ. Microfilms Order No. 81- IC 8530, 1971, 28 pp. 02,271. Hundemann, A. S. Strip Mining (cita- Gilbertson, B. Monitoring Vegetation tions from the NTIS Data Base, v. 2,, Cover on Mine Dumps With ERTS-1 Imagery: 1976-Nov. 1979). IL Nat. History Surv., Some Initial Results. Paper in Proceed- Jan. 1980, 255 pp.; NTIS, PB-80-803570. ings of a Symposium on Significant Re- sults From ERTS-1 (Goddard Space Flight Inglis, M. H. Development of a Coal Center, New Carrollton, MD). NASA SP- Surface Mine Monitoring Capability Uti- 327, v. 1, 1973, pp. 577-584. lizing Landsat Satellite Technology. Available from Technol. Application Cen- Glass, C. E., and R. A. Schowengerdt. ter, Univ. NM, Albuquerque, NM, 1980, Application of Digital Image Analysis to 39 pp. B-62

Inglis, M. H., H. W. Sheffer, R. J. P. Koerner, R. M., A. E. Lord, and W. M. Lyon, and A. E. Prelat. Landsat Monitor- McCabe. Remote Sensing Applications for ing of the Navajo Coal Surface Mine. Mine Waste Stability Monitoring Using Paper in Proceedings of the American So- the Acoustic Emission Method. Paper in ciety of Photogrammetry (Fall Tech. Meet- Proceedings International Geoscience and ing, Albuquerque, NM, Oct. 15-20, 1978). Remote Sensing Symposium (Washington, Am. Soc. Photogramm., Falls Church, VA, DC, June 8-10, 1981). IEEE Service Cen- 1978, pp. 523-539. ter (Cat. No. 81CH1656-8), v. 1, 1981, pp. 355-359. Irons, J. R., H. Lachowski, and C. Pe- terson. Remote Sensing of Surface Mines; Krumwiede, D. D. Remote Sensing Appli- A Comparative Study of Sensor Systems. cations as Related to Surface Mining. Paper in Proceedings of the Fourteenth Paper in Proceedings Symposium on Surface International Symposium on Remote Sensing Mining Hydrology, Sedimentology and Rec- of the Environment. Environ. Res. Inst. lamation (Lexington, KY, Dec. 1, 1980). MI, Ann Arbor, MI, v. 2, Apr. 1980, Univ. KY Office Eng. Serv., 1980, pp. 1041-1053. pp. 159-162.

Ishikawa, P., Jr., and G. A. Shelton. Mamula, N., Jr. Remote-Sensing Method Summary of the Western Energy Overhead for Monitoring Surface Coal Mining in the Monitoring Project (EPA contract 68-03- Northern Great Plains. J. Res. U.S. 2636, Environ. Monitoring and Support Geol. Surv., v. 6, No. 2, Mar.-Apr. 1978, Lab., Las Vegas, NV). EPA Rep. EPA-600/ pp. 149-160. 4-80-051, Oct. 1979, 38 pp. Marcus, P. M. Remote Sensing Inventory Johannsen, C. J., R. W. Blancher, and of Mining Activity in the Pennsylvania D. J. Barr. Characteristics of Proper- Anthracite Region. BuMines Research 76, ties of Strip Mine Spoils as Related to 1976, p. 81. Remote Sensing Measurement (Univ. M0, Columbia Campus Agr. Exp. Station, Dep. Maxim, L. D., and D. E. Cullen. Cost Agronomy, Columbia, M0, contract/grant Model for Remote Inspection of Ground No. 0073683, M000354-1). U.S. Dep. Agri- Sites. Photogramm. Eng. Remote Sensing, culture, Cooperative Res. Office, Colum- v. 43, 1977, pp. 1009-1025. bia, MO, 1981. Moik, J. G. Digital Processing of Re- Knuth, W. M., E. L. Fritz, and J. A. motely Sensed Images. NASA SP-431, 1980, Schad. Investigation of Color and Color 330 pp. Infrared Aerial Photographic Techniques for Mining and Reclamation Planning and Moore, 0. H., J. H. Adams, and A. F. Monitoring (contract J0155041, HRB- Gregory. Mapping Mine Wastes With Land- Singer, Inc., State College, PA). Bu- sat Images. Paper in Proceedings of the Mines OFR 37-79, 1978, 215 pp. ; NTIS PB- Fourth Canadian Symposium on Remote Sens- 294707. ing (Quebec, Canada, May 16-18, 1977). Canada Aeronaut, and Space Inst. Rep. Knuth, W. M., Jr., and H. B. Charmbury. A78-433 31943, 1977, pp. 294-304. Remote Sensing Techniques for Analysis of Burning in Coal Refuse Banks. Paper in National Aeronautics and Space Admini- Proceedings of the First Symposium on stration (Cleveland, OH). LANDSAT Remote Mine and Preparation Plant Refuse Dispo- Sensing: Observations of an Appalachian sal (Louisville, KY, Oct. 22-24, 1974). Mountaintop Surface Coal Mining and Rec- National Coal Association, Washington, lamation Operation. NASA-TM-84194, Oct. DC, 1974, pp. 38-43. 1979, 7 pp. B-63

National Field Investigations Center- 3-6, 1980). IEEE (Cat. No. 80CHI533-9 Application of ERTS Technology to the MPRSD), Piscataway, NJ, 1980, pp. 126- Evaluation of Coal Strip Mining and Rec- 134. lamation in the Northern Great Plains. Denver, CO, Feb. 1975, 120 pp. ; NTIS PB- Sharber, L. A., and F. Shahrokhi. Ap- 255590. plication of Satellite Data in Monitoring Strip Mines. Univ. TN Inst. Space, Re- National Technical Information Service mote Sensing Earth Resour., v. 6, 1977, (Springfield, VA). Strip Mining. 1980- pp. 499-514. Feb. 1982 (citations from the NTIS Data Base). PB-82-807447, May 1982, 156 pp. Solomon, J. L., W. F. Miller, and D. A. Quattrochi. Development of a Tree Clas- Patterson, D. B., and K. M. Campbell. sifier for Discrimination of Surface The Effectiveness of Multi-Date, Multi- Mine Activity From Landsat Digital Data. Scale Aerial Remote Sensing Imagery for Paper in Proceedings of the 45th Annual Monitoring Coal Mining Operations and Meeting of the American Society of Photo- Reclamation Efforts in Alberta. Paper in grammetry (Washington, DC, Mar. 18-24, Proceedings of the Fifth Canadian Sympo- 1979). Am. Soc. Photogramm., Falls sium on Remote Sensing (Victoria, Canada, Church, VA, v. 2, 1979, pp. 607-613. Aug. 28, 1978). Canada Aeronaut, and Space Inst., 1979, pp. 165-173. Tanner, C. E. Computer Processing of Multispectral Scanner Data Over Coal Rehder, J. B. Changes in Landscape Due Strip Mines. Lockheed Electronics Co., to Strip Mining. Paper in ERTS-1, A New Inc., Las Vegas, NV, Mar. 1979, 62 pp.; Window on Our Planet. U.S. Geol. Surv. NTIS PB 80-111677. Prof. Paper 929, 1976, pp. 254-257. U.S. Bureau of Mines. Fast Monitoring Rogers, R. H., W. A. Pettyjohn, and of Mine Waste Embankments. Technol. L. E. Reed. Automated Strip Mine and News, No. 67, 1979, pp. 1-2. Reclamation Mapping From ERTS. Paper in Proceedings, Third Earth Resources Tech- U.S. Environmental Protection Agency, nological Satellite-1 Symposium, ed. by Office of Enforcement. Remote Sensing S. Freden. NASA Sci. Tech. and Inf. Investigation Solid/Liquid Waste Disposal Off., Washington, DC, NASA SP-351, v. 1, Sites. Nat. Enforcement Inv. Ctr., Den- sec. B, 1974, pp. 1519-1531. ver, CO, EPA-330/1-80-002, May 1980. Roth, L. H., J. A. Cesare, and G. S. U.S. Geological Survey. Use of Landsat Allison. Rapid Monitoring of Coal Refuse CCT's To Inventory Kaolin Mines. Prof. Embankments (contract H0262009, CH2M Paper 1000, 1976, 271 pp. Hill). BuMines OFR 11-78, 1977, 113 pp.; NTIS PB-277975. Vanghan, P. R. The Measurement of Pore Pressure With Piezometers. Ch. in Field Russell, 0. R., V. Amato, and T. V. Instrumentation in Geotechnical Engineer- Leshendok. Remote Sensing and Mine Sub- ing. Wiley, 1974, pp. 411-422. sidence. Transp. Eng. J., v. 105, No. 2, Mar. 1979, pp. 185-198. Wobber, F. J., C. E. Wier, T. Leshen- dok, and W. Beeman. Survey of Coal Ref- Schreier, H., and L. M. Lavkulich. Ex- use Banks and Slurry Ponds for the Indi- amination of the Overall Relationship Be- ana State Legislature Using Aerial and tween Spectral Reflectance and Chemical Orbital Inventory Techniques. Paper in Composition of 58 Mine Tailings Samples. Proceedings of the First Symposium on Paper in Proceedings of the Sixth Annual Mine and Preparation Plant Refuse Dispo- Symposium—Machine Processing of Remotely sal (Louisville, KY, Oct. 22-24, 1974). Sensed Data and Soil Information Sys- National Coal Association, Washington, tems and Remote Sensing and Soil Survey DC, 1974, pp. 64-77. (Purdue Univ., West Lafayette, IN, June

US GOVERNMENT PRINTING OFFO 1987 60^0:! ->C Of C INT.-BU.OF MINES,PGH.,PA. 28546 APPENDIX c SUPPLIERS PUBLISHED LISTS C-l

RADARSAT International Tnc. SATELLITE DATA DISTRIBUTION CENTRE FEES AND CHARGES FOR SATELLITE REMOTE SENSING IMAGERY, TAPES AND SERVICES

1. MULTISPECTRAL SCANNER DIGITAL PRODUCTS, PER SCENE A) RAWFULLSCENE $ 69500 B) BUtK FULL SCENE $ 795.OO C) SYSTEMATIC GEOREFERENCED FULL SCENE $ 795.00 D) SYSTEM CORRECTED, GEOCODED $ 525.00 E) PRECISION CORRECTED, GEOCODED $ 650.00

2. THEMATIC MAPPER 7 BAND DIGITAL PRODUCTS, PER SCENE A) RAW QUADRANT $ 1,525.00 B) BULKQUADRANT $ 1,580.00 Q SYSTEMATIC GEOREFERENCED QUADRANT $ 1.580.00 D) RAW FULL SCENE $ 3.600.00 E) BULKFULL SCENE $ 3,840.00 F) SYSTEMATIC GEOREFERENCED FULL SCENE $ 3340.00 G) SYSTEM CORRECTED. GEOCODED $ 1,525.00 H) PRECISION CORRECTED. GEOCODED $ 1.725.00

3. THEMATIC MAPPER 3 BAND DIGITAL PRODUCTS, PER SCENE A) RAW QUADRANT $ 720.00 B) BULKQUADRANT $ 740.00 Q SYSTEMATIC GEOREFERENCED QUADRANT S 740.00 D) RAWFULLSCENE $ 1,400.00 E) BULKFULL SCENE S 1,480.00 F) SYSTEMATIC GEOREFERENCED FULL SCENE $ 1,480.00 G) SYSTEM CORRECTED. GEOCODED $ 720.00 H) PRECISION CORRECTED. GEOCODED $ 860.00

4. THEMATIC MAPPER NIGHT I BAND DIGITAL PRODUCTS, PER SCENE A) RAW QUADRANT $ 430.00 B) BULKQUADRANT S 430.00 Q SYSTEMATIC GEOREFERENCED QUADRANT $ 430.00 D) RAWFULLSCENE S 450.00 E) BULK FULL SCENE $ 450.00 F) SYSTEMATIC GEOREFERENCED FULL SCENE S 450.00 G) SYSTEM CORRECTED. GEOCODED $ 430-00 5. SPOT - MULTISPECTRAL LINEAR ARRAY DIGITAL PRODUCTS, PER SCENE A) RAWFULLSCENE S 1.780.00 B) BULK FULL SCENE S 1.855.00 Q SYSTEMATIC GEOREFERENCED FULL SCENE S 1,855.00 D) SYSTEM CORRECTED, GEOCODED S 1385.00 E) PRECISION CORRECTED, GEOCODED $ 1.675.00

6. SPOT - PANCHROMATIC LINEAR ARRAY DIGITAL PRODUCTS, PER SCENE A) RAWFULLSCENE $ 2,110.00 B) BULK FULL SCENE S 2,205.00 Q SYSTEMATIC GEOREFERENCED FULL SCENE S 2,205.00 D) SYSTEM CORRECTED, GEOCODED S 1,580.00 E) PRECISION CORRECTED, GEOCODED S 1,970.00 C-2

7. MULTISPECTRAL SCANNER FILM PRODUCTS, PER SCENE A) 185mm TRANSPARENCY. BLACK AND WHITE SYSTEMATIC GEOREEERENCED 75.00 B) I85mm TRANSPARENCY. COLOR SYSTEMATIC GEOREFERENCED s MOJOO 8. THEMATIC MAPPER FILM PRODUCTS, PER QUADRANT A) 185mn> TRANSPARENCY, BLACK AND WHITE SYSTEMATIC GEOREFERENCED s 120.00 B) 185mm TRANSPARENCY. COLOR SYSTEMATIC GEOREFERENCED $ 300.00 9. THEMATIC MAPPER FILM PRODUCT, PER FULL SCENE A) 185mm TRANSPARENCY. BLACK AND WHITE SYSTEMATIC GEOREFERENCED $ 150.00 B) 185mm TRANSPARENCY. COLOR SYSTEMATIC GEOREFERENCED $ 335.00 10. SPOT - MULTISPECTRAL LINEAR ARRAY FILM PRODUCTS, PER SCENE A) 185mm TRANSPARENCY, BLACK AND WHITE SYSTEMATIC GEOREFERENCED S 855-00 B) 185mm TRANSPARENCY. COLOR SYSTEMATIC GEOREFERENCED S 930.00

11. SPOT - PANCHROMATIC LINEAR ARRAY FILM PRODUCTS, PER SCENE A) I85mm TRANSPARENCY. BLACK AND WHITE $ 930.00 SYSTEMATIC GEOREFERENCED

NOTE: GEOCODED FILM PRODUCTS AVAILABLE ONLY WITH PURCHASE OF GEOCODED DIGITAL PRODUCTS

12. LANDSAT MICROFICHE CATALOGUE SUBSCRIPTION PER SATELLITE YEAR, NO ROYALTY OR BASIC CHARGE

SPECIFIC CARD REQUESTS PER CARD $ 2.89

STANDING ORDERS:

A) ONE TRACK (1 TO 23 HCHE CARDS) PER CARD S 2.89

B) 2 TO 10 TRACKS (24 TO 228 FICHE CARDS) PER CARD $ 2.46

C) 11 TO 20 TRACKS (229 TO 456 FICHE CARDS) PER CARD $ 2.06

D) 21 TO 40 TRACKS (457 TO 913 HCHE CARDS) PER CARD s 1.65

E) COMPLETE CANADIAN COVERAGE (OVER 914 FICHE CARDS) PER CARD s 1.36 C-3

13. SPOT MICROFICHE CATALOGUE SUBSCRIPTION PER SATELLITE YEAR, NO ROYALTY OR BASIC CHARGE

SPHC1F1C CARD REQUESTS PER CARD $ 24J9

STANDING ORDERS (ANY NUMBER OF gs> PER CARD $ 136

NOTE: AT J. MTCROFICHE ARE PRODUCED ON A PER CARD BASIS

14. HANDLING CHARGE FOR OTHER THAN MICROFICHE PRODUCTS, PER ORDER $ 10.00

15. SURCHARGE FOR RUSH ORDER TWICE UNIT PRICE

16. NON-STANDARD BAND OR COLOUR ASSIGNMENT, $ 150.00 PER SCENE C-4

17. TRANSPARENCIES ORDERED AT TIME OP CCT PRODUCTION. INTRODUCTORY OFFER

LS + -MSS S 17.00 -7M $ 45.00 SPOT - FULL SCENE -MLA-B&W $ 43.00 -COLOR $142.00 -PLA-B&W SI 10.00 •GEOCODED -MLA-BAW $ 21.00 -COLOR S 46.00 -PLA-B&W S 37.00 C-5 Energy. Mines and £nergie. Mines el Resources Canada Ressources Canada

Canada Centra for Ramota Sanalng

FEES AND CHARGES POR SATELLITE REMOTE SENSING IMAGERY, TAPES, ASD SERVICES

COLUMN 1 COLUMN II ITEM DESCRIPTION FEES AMD CHARGES

1. MDLTISPECTRAL SCANNER DIGITAL PRODUCTS, FER SCENE (a) raw full scene, 34,000 km2 i«t-^^- $ 695.00 (b) bulk full scene, 34,000 km2 795.00 (c) system corrected, geocoded, 4,000 km2 525.00

2. THEMATIC MAPPER 7 BAND DIGITAL PRODUCTS, *ER SCENE (a) raw quadrant, 8,360 km2 /•*$ »/% 1,525-00 (b) bulk quadrant, 8,360 km2 1,580.00 (c) raw full scene, 32,000 km2 3,600.00 (d) bulk full scene, 32,000 km2 3,840.00 (e) system corrected, geocoded, 4,000 km2 1,525.00 3. THEMATIC MAPPER 3 BAND DIGITAL PRODUCTS, PER SCENE (a) raw quadrant, 8,360 km2 720.00

4. THEMATIC MAPPER MIGHT 1 BAND DIGITAL PRODUCTS, PER SCENE (a) raw quadrant, 8,360 km2 430.00 (b) bulk quadrant, 8,360 km2 430.00 (c) raw full scene, 32,000 km2 450.00 (d) bulk full scene, 32,000 km2 450.00 (e) system corrected, geocoded, 4,000 km2 430.00

5. NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION ADVANCED VERT HIGH RESOLUTION RADIOMETER DIGITAL PRODUCTS (a) one scene 185.00 (b) two scenes 215.00 (c) three scenes 240.00 (d) four scenes 270.00 (e) five scenes 295.00

6. "ST8TEME POUR I'OBSERVATION DE LA TERSE" (SPOT) MULTI- SPECTRAL LINEAR ARRAY DIGITAL PRODUCTS, PER SCENE (a) raw full scene, 3,600 km2 1,780.00 (b) bulk full scene, 3,600 km2 1,855.00 (c) system corrected, geocoded, 1,000 km2 1,385.00

7. "STSTEME POUR L'OBSERVATION DE LA TERSE" (8P0T) PANCHRO- MATIC LINEAR ARRAY DIGITAL PRODUCTS, PER SCENE (a) raw full scene, 3,600 km2 2,110.00 (b) bulk full scene, 3,600 km2 2,205.00 (c) system corrected, geocoded, 1,000 km2 1,580.00 C-6

COLUMN II COLUMN 1 FEES AND ITEM DESCRIPTION CHARGES

8. PATLOAD CORRECTION DATA, PER 5 SCENES ON EITHER SIDE Of TARGET SCENE (1 TAPE, 9 TRACK, 6,250 BPI, 2,400 IT. REEL) 425.00

9. PROCESSING OP SIGNAL COMPOTES COMPATIBLE TAPE TO SEASAT OR SHUTTLE IMAGING RADAR-B IMAGE COMPUTER COMPATIBLE TAPE (1 TAPE, 9 TRACK, 6,250 BPI, 2,400 FT. KEEL) 1,050.00

10. SURCHARGE TOR HANDLING OF SHUTTLE IMAGING RADAR-B IMAGERY TRANSCRIPTION FROM JET PROPULSION LABORATORY AND COPT OF IMAGE COMPUTER COMPATIBLE TAPE TO JET PROPULSION LABORATORT 50.00

11. RETURN BEAM VIDICON, OR ADVANCED VERT HIGH RESOLUTION RADIOMETER PBOTOGRAPHIC IMAGERY, PER IMAGE (a) 185 ma transparency, black and white 75.00

12. MDLTISPECTRAL SCANNER (a) 185 an transparency, black and white 75.00 (b) 185 m transparency, colour 140.00

13. THEMATIC MAPPER PHOTOGRAPHIC IMAGERT BULK CORRECTED QUADRANT, PER IMAGE (a) 185 am transparency, black and white 120.00 (b) 185 on transparency, colour 300.00

14. THEMATIC MAPPER PHOTOGRAPHIC IMAGERT BULK CORRECTED FULL SCENE, PER IMAGE (a) 185 mm transparency, black and white 150.00 (b) 185 mm transparency, colour 335.00 15. "SYSTEME POUR L'OBSERVATION DE LA TERRE" (SPOT) MULTI- SPECTRAL LINEAR ARRAT FULL SCENE, PER IMAGE (a) 185 am transparency, black and white 855.00 (b) 185 mm transparency, colour 930.00

16. "STSTEME POUR L'OBSERVATION DE LA TERRE" (SPOT) PANCHRO- MATIC LINEAR ARRAY FULL SCENE, PER IMAGE (a) 185 mm transparency, black and white 930.00

17. COLOUR ENHANCEMENT, PER IMAGE 65.00

18. FACSIMILE TRANSMISSION CHARGES IN ADDITION TO TRANSMISSION LINES COSTS (THE ORIGINAL IMAGE IS SENT TO CLIENT), PER SCENE 5.00 C-7

COLUMN II COLUMN 1 FEES AND ITEK DESCRIPTION CHARGES

19. LANDSAT AND SPOT KICROFICHE CATALOGUE SUBSCRIPTION PER SATELLITE TEAR, RO ROYALTY OR BASIC CHARGE

NOTE: SPOT flche are produced on a per segment/per eensor basis.

(a) one track (1 to 23 fiche cards) $2.89 per card (b) 2 to 10 tracks (24 to 228 flche cards) $2.46 per card (c) 11 to 20 tracks (229 to 456 fiche cards) $2.06 per card

20. TRANSCRIPTION TO 9 TRACK, 6,250 BPI, 2,400 FT. REEL, PER OUTPUT KEEL 30.00 21. MASTER NEGATIVE PROM USER SUPPLIED DATA ON TAPE (NATIONAL AIR PHOTO LIBRARY CHARGES EXTRA), PER NEGATIVE (a) 185 ran black and white 65.00 (b) 185 am colour 130.00

22. HANDLING CHARGE FOR OTHER THAN CATALOGUE SUBSCRIPTION. PER ORDER 10.00 23. SURCHARGE FOR RUSE ORDERS twice unit price 24. NON-STANDARD BAND OR COLOUR ASSIGNMENT, PER SCENE 150.00