EimiONMENTAl PROTECTION AGENCY
OFFICE OF ENFORCEMENT
EPA-330/3-75-001
An Application Of ERTS Technology
To The Evaluation Of
Coal Strip Mining And Reclamation
In The Northern Great Plains
NATIONAL FIELD INVESTIGATIONS CENTER-DENVER
REGION
AND
DENVER.COLORADO
FEBRUARY 1975
Maw
-------
ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF ENFORCEMENT
AN APPLICATION OF ERTS TECHNOLOGY
TO THE EVALUATION OF
COAL STRIP MINING AND RECLAMATION
IN THE NORTHERN GREAT PLAINS
National Field Investigations Center - Denver
and
Region VIII
Denver, Colorado
February 1975
-------
CONTENTS
I. INTRODUCTION 1
II. SUMMARY AND CONCLUSIONS 5
III. BACKGROUND 9
Climatology 9
Coal Resources 10
Production 12
IV. STUDY TECHNIQUES 15
Satellite Data 15
Ground Truth 20
Data Interpretation and Analysis 20
V. RESULTS AND EVALUATION OF
THE DATA ANALYSIS 39
Wyoming Mines (Nos. 1-10) 40
Montana Mines (Nos. 11-19) 57
North Dakota Mines (Nos. 20-30) 77
m
-------
TABLES
II-I Summary of Mine Classifications 6
IV-1 Spectral Bandwidths 18
IV-2 Standard Deviation Point Intensities 33
V-l Areal Distribution of Land Use (Mine 4) .... 49
V-2 Areal Distribution of Land Use (Mine 5) . . . . 53
V-3 Areal Distribution of Land Use (Near Mine 6). . 54
V-4 Areal Distribution of Land Use (Mine 11). ... 60
V-5 Areal Distribution of Land Use (Mine 12). . -. . 64
V-6 Areal Distribution of Land Use (Mine 13). ... 67
V-7 Areal Distribution of Land Use (Mine 14). ... 72
V-8 Areal Distribution of Land Use (Mine 15). ... 75
V-9 Areal Distribution of Land Use (Mine 20). ... 80
V-10 Areal Distribution of Land Use (Mine 21). ... 84
V-ll Areal Distribution of Land Use (Mine 22). ... 89
V-12 Areal Distribution of Land Use (Mine 23). ... 94
V-13 Areal Distribution of Land Use (Mine 26). ... 99
V-14 Areal Distribution of Land Use (Mine 29). ... 103
FIGURES
V-l Northern Great Plains Coal Field 2
IV-1 Typical ERTS Daily Ground Trace 16
IV-2, 3 Observatory Configurations; MSS Scanning. ... 17
I V-4 ERTS Coverage of Continental US 19
IV-5 Red Band Image of Mine No. 4 22
IV-6 Sample Symbolic Map of Mine No. 4 24
IV-7 Sample Classification Map of Mine No. 4 .... 26
IV-8 Sample Interpretive Overlay 27
IV-9 MSS-Geometric Corrections 30
IV-10 Comparison of Successive Classifications. ... 34
IV-11 Comparison of Mine No. 4 Maps 37
V-l Locations of Wyoming Coal Mines 105
V-2 Satellite Image of Mine No. 4 43
V-3 Dave Johnston Mine Reclamation 44
V-4, 5 Topsoil Borrow Areas 45
V-6, 7 Middle Area of Mine 46
V-8 Classification Map of Mine No. 4 48
V-9 Satellite Image of Mines No. 5 and 6 51
V-10 Classification Map of Mine No. 5 52
V-ll Classification Map of Abandoned Mine
Area Near Mine No. 6 55
V-12 Locations of Montana Coal Mines 106
V-13 Satellite Image of Mine No. 11 58
V-14 Classification Map of Mine No. 11 59
V-15 Satellite Image of Mine No. 12 62
V-16 Classification Map of Mine No. 12 63
V-17 Satellite Image of Mines No. 13 and 14 65
IV
-------
V-18 Classification Map of Mine No. 13 66
V-19 Classification Map of Southern Section
of Mine No. 14 69
V-20 Classification Map of Northern Section
of Mine No. 14 71
V-21 Classification Map of Mine No. 15 74
V-22 Locations of North Dakota Coal Mines 107
V-23 Satellite Image of Mine No. 20 . 78
V-24 Classification Map of Mine No. 20 79
V-25 Satellite Image of Mines No. 21, 22 82
V-26 Classification Map of Mine No. 21 83
V-27 Classification Map of the South Mine 86
V-28 Classification Map of the North Mine 88
V-29 Satellite Image of Mine No. 23 91
V-30 Classification Map of Mine No. 23 92
V-31 Satellite Image of Mine No. 25 93
V-32 Satellite Image of Mine No. 26 97
V-33 Classification Map of Mine No. 26 98
V-34 Satellite Image of Mine No. 29 101
V-35 Classification Map of Mine No. 29 102
-------
CONVERSIONS
multiply by to obtain
Metric Unit. English Unit
hectares* 2.471 acres
kilometers (km) 0.621 miles (mi)
meters (m) 3.280 feet (ft)
metric tons 1.102 tdns (short)
*Multiply pixels by: 0.486 to obtain hectares; or 1.199 to obtain acres.
ABBREVIATIONS
ERTS - earth resources technology satellite
LARS - Laboratory for the Application of Remote Sensing (Purdue University)
MSS - multispectral scanner
SCS - Soil Conservation Service
USGS - United States Geological Survey
GLOSSARY
borrow pit - an area of special excavation to provide material for an
embankment when there is insufficient excavation on or
near the job site to form the embankment
embankment - an artificial ridge of earth and broken rocks, or a fill
with a top higher than the adjoining surface
high wall - the unexcavated face of exposed overburden and coal in an
opencast mine, or the face or bank on the uphill side of a
contour strip mine excavation
overburden - the material that overlies a deposit of coal, especially as
mined from the surface by open cuts
pixel - the instantaneous field of view within the multispectral scanner
radiometric analysis - an analysis based on the intensity of the light as
determined by ERTS multispectral scanner
spectral signatures - a set of four radiant intensities in the ERTS spectral
bands defining a class of material
spoil pile - the area where mine waste is disposed of or piled
VI
-------
I. INTRODUCTION
Historically, the production of coal in the Northern Great Plains
has been minimal by national production standards, although reserves are
very large. The recent energy crisis, however, coupled with the ex-
tremely low sulfur content (averaging about 0.6 percent)'of the coal have
produced increased interest in coal mining in this region. But the
potential adverse environmental impacts of surface mining operations, as
demonstrated in other parts of the country, have offset this interest.
The Northern Great Plains Resources Program, initiated in 1973 by
the Department of the Interior, was to provide tools to quantify the
environmental impacts of actual and proposed development of coal resources
in Wyoming, Montana, North and South Dakota [Fig. 1-1]. As an integral
part of the program, the National Field Investigations Center-Denver
(NFIC-Denver), Office of Enforcement, EPA, was requested by EPA Region
VIII, Denver, Colo., to conduct a study of coal mines in Montana, North
Dakota, and Wyoming. The purpose of the study was to document the size,
shape and location of the following:
1. The actively mined area within each coal strip mine site
2. The untouched spoils piles within the mine site
3. The reclaimed or recontoured areas within the mine site
4. Newly vegetated recontoured areas within the mine
5. Abandoned spoils piles that have been naturally revegetated
with native plants.
-------
Figure I —1. Northern Great Plain* Coal Field
-------
The study was conducted with the use of remote sensing data from
the Earth Resources Technology Satellite (ERTS}, recorded from late June
through late August 1973. To verify mining data, a telephone survey of
companies operating active coal mines was conducted in.early 1974.
Other verification data included USGS and SCS aerial photographs of
mining areas, maps provided by mining companies, and aerial photographs
taken by NFIC-Denver personnel.
Besides presenting a synoptic inventory of Northern Great Plains
coal strip mines, this report demonstrates how computer classification
of satellite data can be applied to monitoring land use. Future ERTS
imagery of this area, when compared to the baseline information in this
report, can be used to detect various changes at the mine sites.
-------
II. SUMMARY AND CONCLUSIONS
Thirty active, inactive and proposed coal mine sites identified by
the Northern Great Plains Resources Program in the states of Montana,
North Dakota and Wyoming were evaluated in this study. Three under-
ground mines and three proposed mines were not computer classified.
Computer classification was not attempted for eight additional mines be-
cause they were small, inactive, or lacked ERTS data. Of the remaining
sixteen sites, computer classification was successful at fourteen.
Attempts to classify the other two mines were unsuccessful because of
the small size of one mine and the masking effects of cultivated crop-
land at the other.
Land use, or classification, at each mine evaluated was defined by
computer processing the ERTS data which was in the form of digital
magnetic tapes. A multipectral classification algorithm was used after
accepted procedures for smoothing, filtering and line drop-out correc-
tions were applied to the digital data.
In most cases, the computer classification techniques were success-
ful in separating mine areas into: active mining areas, graded spoils
piles, ungraded spoils piles and revegetated spoils areas. Other topo-
graphical and cultural features such as ponds, roads, railroads and
cropland in the mine vicinities were defined.
The results of the computer classifications are presented in Table
II-l. Disturbed areas of all types for the mines classified totaled
2 2
45.2 km (11,159 acres; 17.4 mi ). In addition to this total, several
-------
TABLE II-l. _
Summary of Mine Classifications
Mine*/
No
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
TOTALS
Spoils
Active Mine Ungraded Graded Revegetated Miscellaneous
Hectares Acres Hectares Acres Hectares Acres Hectares Acres Hectares Acres
Mine not computer classified
Mine not computer classified
Mine not computer classified
247^ 610-7 125 305 182 450
21 53 28 68 44 108
145 356 16 38 20 49
Proposed mine
Mine beginning operation
Proposed mine
Proposed mine
40 100 44 108 56 137
74-/ 182-/
42-' 104-7 25 61 23 56
61 151 508 1255 304 750 64 158
18 44 161-' 398-'
Mine not computer classified
Underground mine
Underground mine
Underground mine
36 90 81 199 6 14
135-7 334-' 162 401 136 337 138 341
129 320 536 ' 1324
362 892 164 405 32 78
Mine closed
Mine not classifiable
93-' 230-' 31 75
Mine not computer classified
Mine not computer classified
28 68 69 172 20 50
Mine not computer classified
896 1988 2055 5069 673 1661 348 856 434 1067
Total Disturbed-7
Hectares Acres
27
85
2
554
93
180
140
74
90
937
179
123
572
665
557
124
117
4519
66
210
4
1365
229
444
347
182
222
2315
442
304
1414
1644
1375
306
290
11,159
a/ Mine locations shown on Figures V-l, V-12, V-22 inside back cover.
b/ Area of mines not classified was determined from ERTS transparencies
c/ Includes ungraded spoils
d/ Includes roads and areas in preparation for mining
-------
square kilometers of land have been disturbed at the small mines not
classified. Ungraded spoils piles accounted for the largest portion
(45 percent) of the mine area. Active mine areas including some ungraded
spoils accounted for another 20 percent of the total mine area. Only
four mines had significant areas that had been revegetated, accounting
for 8 percent of the total area. An additional 15 percent of the area
contained in 5 mines had been graded but not vegetated. Some degree of
natural revegetation has occurred on both graded and ungraded spoils at
several mines.
The application of ERTS technology to land use classification and
areal determination was most successful for the large mines containing
large areas with the same land use. The minimum size object or land
area that can be defined by the ERTS sensors is about 0.5 hectare
(1.2 acre), about the size of a football field. This limited spatial reso-
lution means that present ERTS technology can be effectively used only
for mines with homogeneous areas at least 200 m (660 ft) on a side and
preferably 340 m (1,100 ft) in minimum dimension. ERTS technology to be
implemented in the next few years will substantially improve the spatial
resolution, allowing classification of small as well as large mines.
The most efficient use of ERTS technology could be achieved by
collecting detailed ground truth or field data on one or two mines,
calibrating the computer procedures using this data, and then extrapo-
lating to other mines, thus completing the classification without addi-
tional data collection. In practice, this did not prove feasible. Sig-
nificant variations between mines, and even for the same mine, were
-------
8
observed in the ERTS data for the same type of mine area such as ungraded
spoils piles. This required that the classification algorithm be cali-
brated for each mine site. The accuracy of the classification at each
mine was thus largely dependent upon the availability of- ground truth
such as low-level aerial photographs, maps, or field observation.
Two types of problems were encountered in distinguishing areas dis-
turbed by mining activities from adjacent lands not disturbed by mining.
Due to the semi-arid climate, native vegetation near mines in Wyoming
and most of Montana is relatively sparse. If reclamation or natural
revegetation of spoils piles produced significant vegetative cover, it
was often difficult to distinguish between such disturbed areas and the
surrounding lands because both types of terrain yielded similar ERTS
imagery. A similar problem was encountered in distinguishing between
nonvegetated mine areas and adjacent lands with eroded or cultivated
soil. These problems also occurred for North Dakota mines where adjacent
lands often consisted of alternate strips of bare cropland and growing
crops.
-------
III. BACKGROUND
The area known as the Northern Great Plains encompasses portions of
the states of Montana, North Dakota, South Dakota, Nebraska and Wyoming.
At present, active coal mines are located only in Montana, Wyoming and
North Dakota. The area is sparsely settled and the principal activity
is ranching with a limited amount of farming.
Climatology
The climate of the states included in the study area may be de-
scribed as a typical mid-latitude continental climate. The Rocky Mountains,
located west of the study area, act as a barrier to the moist Pacific
air in the prevailing westerly flow. There are no major geographic
features to modify air masses originating in the cold polar regions of
Canada, or the warm moist regions of the Gulf of Mexico before they
overflow the study area.
Weather changes frequently occur over the area with a passage of
cold fronts and their associated pressure systems; these generally
follow an easterly track. Stagnant air masses which would adversely
affect atmospheric dispersion occur rather infrequently. Warm windy
periods may occur in Wyoming, Montana and the Black Hills of South
Dakota during the winter as a result of warm Chinook winds reaching 40
to 80 km/hr (25 to 50 mph). These periods may last several days and
tend to aid in the dispersion of any atmospheric pollutants.
-------
10
Within the study area, average wind speeds range from 13 to
23 km/hr (8 to 14 mph), while the average temperature varies between
6 and 9°C (42 and 49°F) with annual average highs of 10 to 16°C (50 to
60°F) and average lows of 1 to 2°C (30 to 35°F). The mean annual relative
humidity is 60 percent but averages 70 percent for the fall and
winter quarters. Annual precipitation in the region varies from 25 to
50 cm (10 to 20 in.) per year. Montana and Wyoming are generally more
arid than the three eastern states in the region. Usually 50 percent
or more of the rainfall occurs in spring while winter is the driest
time of the year. The Dakotas receive about 91 cm (36 in.) of snow
annually, except for the Black Hills area which receives 152 to 254 cm
(60 to 100 in.) per year. Wyoming receives 91 to 152 cm (36 to 60 in.)
annually.
Coal Resources
The Northern Great Plains Region contained about 46 percent of the
estimated remaining coal resources of the United States as of January 1,
1972. This reserve was contained in an area of 218,000 km2 (84,000 mi2),
representing only 18 percent of the total area of coal bearing rock in
the United States. The presence of 46 percent of the tonnage in 18
percent of the area of coal bearing rock provides convincing evidence of
the concentration of coal in the study area, a fact explained primarily
by the greater thickness of individual coalbeds in the region as
compared to the bed thickness in other parts of the United States.
The coal in the Northern Great Plains Region occurs in a sequence
of beds 460 to 910 m (1,500 to 3,000 ft) thick of late Cretaceous to
-------
11
Eocene Age. This sequence is divided from bottom to top into the Hell
Creek, Tullock, Fort Union and Wasatch formations. The coal resources
are concentrated in the Fort Union formation of Paleocene Age, and only
subordinate amounts are present in the underlying and overlying
formations. The rocks associated with the coal consist primarily of
poorly consolidated sandstone and siltstone with minor amounts of shale.
Many of the coal outcrops have been ignited by spontaneous combustion,
forest fires, lightning and acts of man. Areas thus effected are
characterized by the red color of the baked rocks over the burnout coal.
These baked rocks are locally termed "clinker," but are more properly
identified as "baked shale." Some of the burning took place thousands
of years ago, -ind some is recent.
The coal bearing strata of the Northern Great Plains are nearly
horizontal. Throughout most of the region, measurable dips are only 1
to 2°. However, these low dips are locally reversed and give rise to
three broad regional structural features: 1) the Wiliston Basin,
centered in western North Dakota; 2) the Cedar Creek Anticine, near the
Montana-North Dakota line; and 3) the Powder River Basin in northeastern
Wyoming and southeastern Montana. The steepest dips, commonly 5 to 10°
eastward, are along the western edge of the Powder River Basin in north-
central Wyoming. Because of the gentle dips and the reversals of dips
associated with the three structural features mentioned above, the coal
bearing rocks are near the surface in most parts of the region. In the
deepest part of the Powder River Basin in Wyoming, the coal bearing
-------
12
rocks are a little more than 610 m (2,000 ft) below the surface, but are
nearer the surface over most of the Wyoming Part of the basin. In
eastern Montana, most of the coal is less than 460 m (1,500 ft) below
the surface. In North Dakota all the coal is less than 370 m (1,200 ft)
below the surface, and 98 percent is less than 300 m (1,000 ft) below
the surface. In South Dakota, all the coal is less than 300 m (1,000 ft)
below the surface, and about 80 percent is less than 150 m (500 ft)
below the surface.
The Bull Mountain field, which lies mostly in Musselshell and
Yellowstone Counties, Montana is a broad shallow basin with a northwest
trending synclinal access. Throughout most of the basin, the coal
bearing seams dip 1 to 5° toward the center of the basin, but are nearly
flat to the east and south, and are steeper to the north. In the
northwest part of the basin, the dip steepens to a maximum of 30°. Most
of the past and current mining has been in the broad, central part of
the basin where dips are low.
Production
Throughout the years, production of coal in the Northern Great
Plains has been minimal by national production standards. In 1965, for
example, the region produced about 5 million tons, or 1 percent of the
national total of 512 million tons. In 1972, the region produced about
20 million tons, or about 3.3 percent of the national total of 595
million tons. This may be compared to 1972 production of 144 million
tons, or 24 percent of total national production, from the Illinois coal
basin in Illinois, Indiana and Western Kentucky. The Illinois basin is only
-------
13
half the size of the Northern Great Plains coal bearing region, and it
contains only one-fourth as much coal. The recent energy crisis,
and the extremely low sulfur content of coal in the Northern Great Plains
(averaging about 0.6 percent), however, have significantly increased
interest in mining of coal from this region.
-------
15
IV. STUDY TECHNIQUES
SATELLITE DATA
The source of the remote sensing data for the Northern Great Plains
study was the Earth Resources Technology Satellite. ERTS was launched
into earth orbit in July 1972. Its orbit around the earth is inclined
9° with respect to the true north and south poles. The altitude of the
satellite above the surface of the earth varies from 900 to 950 km
(560 to 590 mi) due to the slight flatness (oblateness) of the earth
and to the perturbing gravitational forces induced by the sun, moon and
earth. ERTS completes an orbit every 103 min, making fourteen complete
orbits per day. A typical ground trace for the daylight passes shown in
Figure IV-1. A particular target on the earth's surface is covered by
the satellite every 18 days with respect to the previous over-pass.
Likewise, it views the entire earth every 18 days. For every 18-day
period, the satellite ground trace repeats its earth coverage at the
same local time. The overlap registration of a current ERTS frame
(rectangular format for the recorded data) is within 37 km (23 mi) of
the frame recorded 18 days earlier.
ERTS is commonly referred to as an observatory containing various
sensors and communications systems [Fig. IV-2]. The sensor data used
for this study was obtained from the multispectral scanner (MSS). The
MSS is a line scanning device which uses an oscillating mirror to
continuously scan perpendicular to the spacecraft path of travel
-------
16
165 150 135 120 105 90 75 60 45 30 15 0 15 30 45 60 75 00 105 120 135 150 165
DAY1
(REPEATS EVERY 18 DAYS!
180 165 150 135 120 105 00 75 60 45 30 15 0 15 30 45 60 75 90 105 120 135 150 165 18C
Figure IV—1. Typical ERTS Daily Ground Trace
(Daylight Passes Only)
-------
SOLAR ARRAY
ORBIT ADJUST TANK
DATA COLLECTION ANTENNA
17
ATTITUDE CONTROL SUBSYSTEM
RETURN BEAM
VIDICON CAMERAS (3)
WIDEBAND
RECORDER
ELECTRONICS
WIDEBAND ANTENNA
Mgf*- ATTITUDE MEASUREMENT SENSOR
d "^-MULTISPECTRAL SCANNER
S-BAND ANTENNA
X,
Figure IV —2. Observatory Configuration
6 DETECTORS
PER BAND:
2» TOTAL.
->2FCn BANDS
(EflTSBI
SCAN MIRROR
IDSCILLiTfS
NOMINALLY
Figure IV—3. MSS Scanning Arrangement
-------
18
[Fig. IV-3]. Six lines, with the same bandpass, are scanned simultaneously
in each of four spectral bands for each mirror sweep. Spacecraft motion
provides the along-track progression of the six scanning lines. Optical
energy reflected from the earth's surface is sensed simultaneously by an
array of detectors in the four spectral bands with the bandwidths shown
below:
Table IV-1
Spectral Bandwidths
Channel Number
1
2
3
4
Color Designation
Green
Red
Infrared I
Infrared II
Bandwidth (microns)
0.5 to 0.6
0.6 to 0.7
0.7 to 0.8
0.8 to 1.1
At the ground station, the continuous strip of imagery from ERTS is
transformed to framed images with a 10 percent overlap of consecutive
frames measuring 185 x 185 km (100 x 100 nautical mi).
The complete orbital coverage of ERTS for the Continental United
States is shown in Figure IV-4. In this ground trace pattern, every
location within this country is imaged every 18 days at the same re-
spective local time. With a particular target in mind, one can check
the date of satellite coverage [Fig. IV-4] for orbit number and geo-
graphical coordinates. With this information, a search of the NASA
data base is instituted to find the exact satellite frame(s) (imagery)
desired. The imagery can be obtained in three forms:
-------
291 5"
tT \ V
• \ ' L-A-^
71" 7-r1
Figure IV — 4. ERTS Coverage of Continental United States
-------
20
1. Digital computer compatible tapes
2. 22.9 x 22.9 cm (9 x 9 in.) optical transparencies,
scale 1:1,000,000
3. 22.9 x 22.9 cm (9 x 9 in.) black and white prints,
scale 1:1,000,000.
GROUND TRUTH
The ground truth for this study consisted of the following items:
1. Large scale low-level aerial photographs of various mines
recorded by Soil Conservation Service and U. S. Geological
Survey from 1969 to summer 1973
2. Large scale detail maps of certain mines defining active,
recontoured and revegetated areas
3. Aerial photography of certain mines recorded by NFIC-Denver
personnel during the summer of 1974.
The information contained in these types of ground truth data
was used during the computer analysis of the ERTS data.
DATA INTERPRETATION AND ANALYSIS
Data Source
The ERTS data was obtained from the NASA data base for the thirty
areas reported to be sites of coal strip mines. The data consisted of
nine sets of 22.9 x 22.9 cm (9 x 9 in.) transparencies and the respec-
tive sets of digital computer compatible tapes. Nine ERTS frames were
required to completely cover the Northern Great Plains area. The
selectivity of the data was based upon the time interval of late summer
1973, near-zero cloud cover over the respective mine locations, and
freedom from telemetry degradations and errors.
-------
21
The transparencies were used to visually locate the general area of
each mine. The precise geographical location of each mine was used in
cueing the computer for selection of the area(s) within the computer
tape to be radiometrically analyzed. The detailed analytical work was
subsequently performed using a general purpose digital computer (IBM
360-44) and the digital computer compatible tapes which provided an
accurate radiometric analysis.
The four separate spectral bands of the ERTS imagery afford a
multispectral identification and classification of a particular target
in question within the limits of the spatial resolution of the satellite
2
scanner. Its resolution capability is 4,860 m (1.2 acres) which by
definition is 1 pixel, the instantaneous field of view within the multi-
spectral scanner. A sample ERTS image constructed from the computer
compatible tapes is shown in Figure IV-5, a red band image of the Dave
Johnston Mine (Mine 4). The tick marks on the perimeter of the image
are guides to a coordinate system used in selecting training sights as
discussed below.
A multispectral statistical computer classification technique was
used to analyze the area in and around each coal mine. This program
was an adaptation of one developed at the Laboratory for the Application
of Remote Sensing (LARS), Purdue University, for the classification of
agricultural (farm) land from digital aircraft and satellite data. The
LARS program operates by reading the optical intensity (in this case
ground reflectance of sunlight) of each pixel in the four ERTS spectral
bands (tape) and then determines the statistical classification of that
element in relation to all other pixels in the target area or scene.
-------
22
-N-
J
:T?ETCH
MINE NO. 4
nUG !'?« 1974 124052 JPL-IPl
Figure IV-5. Red Band Image of Mine No. 4
-------
23
Data Categories
The satellite data for each coal mine were classified into the
following categories:
1. The actively mined area within each strip coal mine site
2. The untouched spoils piles within the area of the mine site
3. The reclaimed or recontoured areas within the area of the
mine site
4. Newly vegetated recontoured areas within the mine
5. Abandoned spoils piles that have been naturally revegetated
with native plants.
Training Areas
During the computer analysis-of satellite data, training areas
(areas of known target character such as active mine locations derived
from ground truth) were used to define the spectral characteristics of
the target classes into which the total scene is to be classified. The
LARS program classifies each pixel of the target area into one of the
previously defined training classes and then computes a number represent-
ing the statistical confidence with which the pixel falls into that
respective class. By examining the confidence numbers, the program
analyst can determine the effectiveness of the classification routine.
Data Forms
The output of data from the computer is presented in several forms.
A symbolic display map of the pixel intensities in each of the four
spectral bands is used initially to identify and select training areas.
-------
24
t eMTFP.it
4 »==_-pFg.
= *€ =«
i i£ «EE
8 F 6 * » « 4 4 . = « fg £= i • .
t= e=== * «4 = =
Peg t=t=CeFc
4444= 4** 4 1* 44 168BSGB18
444* = 4= . 4S*44(*t 4*eet>-
st =- =
-------
25
A symbolic map for Mine 4 is shown in Figure IV-6. Using maps or aerial
photographs as a reference, training sites are established on the symbolic
display and the coordinates are entered into the computer. The statistics
for the four bands are computed and displayed in the form of histograms
for the training areas that have been selected. After the selection of
the respective training areas has been finalized, a classification of
the image is made and a computer map is printed out in which the number
of the particular class is printed in place of each pixel [Fig. IV-7].
By inspection of the histogram for confidence levels assigned to
all the pixels or elements within the target area, a lower limit of
confidence may be selected which permits a second classification map to
be printed, on which confidence levels lower than the selected limit
are replaced by blank spaces. Lty observing this map, the analyst can
tell at a glance in which regions of the image his choice of classi-
fication categories is deficient. He can then combine similar classes,
determine new classes or alter the boundaries of training sites if
necessary. Then the process described above is repeated until an
acceptable confidence level is reached for the particular classifi-
cation scheme being employed. An interpretive overlay* [Fig. IV-8] is
drawn to illustrate the results of the final classification and anno-
tated to identify significant features.
ERTS Measurement Limitations
Limitations can be placed on the accuracy or uncertainty of the
* Some scale distortion is inherent in producing an overlay map (on the
computer line printer) which can be avoided only by additional processing.
In scaling the overlay map to a USGS topographic map, the longitudinal
axis of the former must be reduced to 80% of its nominal value.
-------
26-
1 73UM162M 11 1M1 11 113*13331 1 111 1133311111111
2 22*12*6661 111 1331 112 HI 331 131 13111 1 1 3311 1 1331-"
1 6? I 11221 1122013112263 3131111313)341 I 111 16U1
4 1111 12211 71133333311113)3311111111 133311 It 11 1 1
5 «>212->2 11226311 331331/3113111 111 11 11311221 11 11
* AMl221l?111313313im31l.lJllUllllllimilll
7 IM1 I22l62oll 11 123131131111111111111111111111
B 111116616211111221333111111111111111111111666 ~
f) 61 11 1221 12662S&2?5131 1113111 111 11 11 11 11 1 1 1111
10 21 11 12522645 54?hf./S 1331 11 1211 11 11 11 116611 11 111
11 2112112*55453136*111 111 1221 112121111 lllll 1111
12 6211 1232&4554331261 lllll 111112126111 lllll 1111
13 •><.22211212A'.241312:'3313U 1111111 11 11 llUlllll
-I* S',622U216655631<.46221 11 11 111 111 1 llllllllllll —
15 62&?222. "! 266 35434 ',4o66ll 11 II 11 11 11 116111111111
16 661122?22o333l33 1333 Jl llll 31
?fi 111 I 1 1 &f,64 2 1 1 33 3 311 > 5 -16 1 1 1 1 3 .J 3 3 I 2 1 1 3 1 1 1 1 1 1 1 .1 3
?«» mimii:22mil2226Mllin331121111U33?331 10 15 20 25 30 35 40
30 11111131 111 11 131 1 112521 11 1143324 1131 J333111 11
31 611llim31ll?3333111?5?lll3431ll333H33 311111 33 11111311111131133. 111113311133113
12 '.1 11 1 I II 111 12621111 I 116T6122'. 331. '3333331 111 311111111111 1111 Ull 11111331111333111
31 AV,l 11 Ul 111221 11 111 11 12512333313331)311 311 1.13331131 111 11 111) 1111331111111111331
32?l 1 1111 112621 16323331 13333373311 11 11 lllll 1111 1111 113311 311 111 13331
35 1111116111126611111111112222133333113331 1 1 1 I I 1 11 \ 1 1 1 1 11 3 1 1 1 11 1 1 11 1 1 1 I 1 1 1 11313331
If. Ill 1 11 1 16 11 2M 311 11 1111 11 1261 133333333 U 331 11 111 11 11111111133111 1111 I 11133311131
3/ 1311 1111 11 11211 13111 131 1 1161 211 13113 3 Ul II 33! Ill 1121 lllll II 13 31 133J1 1111 11111331
in lllllUlli;?'!'. ?113!llllll!12?6i24tJH1111l311116<>'i2i lllll ! 3331112J311111LU11111
l'» 1.UIM114I»'>111.I>1IM lJMlllll">M4'ilU l-H mi !255'.t4'ill 1111 331 11.11111 133)11 Jlllll
An 1111 u ii?22iii2ii u i J33Hiii22iim3 nil i:miJ22'."<,2mi limn iimm-'iimu
*! ll?21 I 11 11221 11133 11333 11 1111222226 3 3 111 , n 2^«4643J2? 1 1 H 1 1 I I I 1 1 1 1 1 3 11 1 I 1 1 1 11 1 1
*? 11 11 1311 I 1111111333111 H31I 11 162t?5V,n3i2, *,,<,,. 1? rJ,l2mlll,, 333, 3m
V, M11
«!> 1
*6 1
47 1
',b I
49 I 11 1211 111 1131131316662111 11331333311
50 111 HI II 111 111131131161131131 13313111131 , 1 1 1 J l',-',',^^ 32 11 1 16','o63 1 1 1 1 1 3333! 1 1 11
1 1 11 I 2 2 '. -3 <• «. ^ 22 (.?55/l^/t132llA',«3^4ab', I 111 111 12M31 III
11112M13313333H211 lllll 111 11
,
mil2&tl21U113»31?2lllJjmiM3ll22522i,f^,.j33?21lli:2M/.S511111333llllllll
i333 UJ3 31^31 12'.644^56t- 1 1333.' 11 3733)1 111
1333J3»33)3331U'.'-336ft61 2 J 1 311J3 »H 13
333313UJ53J313333*.'. 313
1113311131111113 MI3J3141 366 111113331111
llJltll'limn33i311U23M2H333333llll
11 1 3 I 1 13 1 1 H 1 1 3 14 3 I 1 1 ?f.<. 3 352 <. 1 1 1 1 1 1 1 26 1 1
1131 13112313121 121 1 lf>624 311 I 112621 112211
31111 1)1 111)11131 1111 116 JlbM 111 11666161
I M 1 11U t 1 I HI! 111 1 ?/?'•<> 4 3J 4 hi Iftrf 6AA611
lllll 111 I 11 11 11 II lbS6/.«.? 244ft U? 122261111
Figure IV — 7. Sample Classification Map - Mine 4
-------
27
_3PRING 1973
FALL 1 B7.O
SECOND HIGH WALL CUT
AND SPOILS PILES
HIGH WALL DEFINITION FROM
RATIO IMAGE RED BAND
INFRARED 1 if A N D
LEGEND
ACTIVE MINE AND UNGRADED SPOILS
FAUI972. SPUING 1973
REVEGETATION, SPUING 1972
HEVEGETATION , 196?, SPRING 1971
TOPSOIL BORRO
Figure IV — 8. Sample Interpretive Overlay
of a Classification Map - Mine 4
-------
28 .
physical measurements carried out on the ERTS (frame) transparencies and
the computer compatible tapes. Any measurements for linear distance and
surface area in the transparencies were made with scaling instruments
and light table microscopes.
The uncertainty for linear distance (ALO) is
ADL = ± 5 x 10 x image scale (m).
The image scale of the transparencies is 1:1,000,000. The value
for ALD is
ALD = ± 5 x 10"4 x 106 m = ± 500 m (1,640 ft)
A distance X, measured physically on the ERTS transparency is accurate
to within ±500 m (1,640 ft).
The uncertainty for the surface area (ASA) is (rectangular)
ASA = ALD (±X±Y) m2.
For the ERTS imagery
ASA = 500 (±X±Y) m2.
For example, a rectangular area with dimensions of X ±500 m and
Y ±500 m would have the value
Area = [XY + 500 (±X±Y)] m2.
The uncertainty in the computer analysis of the computer compatible
2
tape data is limited to one pixel which measures approximately 4,860 m
(1.2 acres). Due to the digital character of the recorded data the
computer can resolve two targets only if they are separated by a dis-
tance of at least 80 m (260 ft). In calculating the surface area of a
-------
29
target, the computer counts the number of pixels present in the image
which is significantly more precise than through visual measurements.
The integrity of the satellite data is greatly enhanced by fourteen
distinct geometric corrections [Fig. IV-9] carried out immediately on
the telemetered data received at the various ground stations. The data
is further improved through precise geometric and contrast corrections
before being analyzed.
A full technical description of the accuracy achieved in the anal-
ysis would require a mathematical treatment of the problems of separating
overlapping Gaussian distributions in a multidimensional spectral space.
Some of the practical problems affecting accuracy will be discussed.
A problem exists in obtaining registration between the satellite
image and the aerial photograph or map used as ground truth. Some
features such as the location of a pond or high wall in the infrared
bands are quite distinct, but others, such as the exact boundary of a
revegetated area, are not. The latter can result in the inclusion of an
erroneous area in a training site or in extending a training site to a
large enough size to avoid an area that is questionable.
Lack of adequate spatial resolution is the most significant problem
in using ERTS imagery. The individual picture elements are approximately
the size of a football field. For a large mine this is not significant
but for smaller mines, the size of the picture elements severely limits
the amount of detail that can be seen. This results in the blending of
-------
30
GEOMETRIC FOOTPRINT
MACmiUDE OF CORRECTION -
MEIERS ON THE GROU1D
GCOMETRIC FOOTPRINT
VACKITUD! OF CORRECTION -
METERS ON THE GROUnO
SCAIE ADJUSTMENT -
ALIITUOE VARIATIONS
MTHINAFIIAUI
B CORRECTIONS
WHIN THE
FBA.ME
AlIIIUBi VARIATIONS
KITHIN A FRAME
(CORRECTIONS
MIHINAFRA.ME
SCORRECriONS
(1IIHIN A FRA.V
•CORRECTIONS
AITHI.1 A FRAME
IMAGE SKEW
AUSED BY EARTH
ROTATION
(FUNCTION OF
LATITUDE!
AVERAGE VELOCITY
CHANJE FROVI
NOMINAL
CAUSED BY FINITE
SCAN Tl •:
ix-inxio* h,
(AT EDGE OF FRAME)
IT
-H t— J»
JX-IIIX lO'o
(AT IDF AND MTTW
OF FRAME)
lU-IISMEIEHS
SPECTRVMANO OFFSE TS
BAND 1 |
BAND I J
BAND ]
BAND!
1 .
JX- 1I7METERS
SCAN MIRROR
VELOCITY CORRECTION
PERSPECTIVE CORRECTIONS
SWEEP TO
FRAME IN DIRECTION
OF SCAN
JXMAX-lliMETERS
FRAME CORRECTION
1 DIRECTION OF
SPACECRAFT TRAVEL -
IEBR CORRECTION)
ALIGNMENT
CORRECTIONS -RELATIVE
AIIG1MENTOFMSS
TO AMS
ROLL
iX— t 1—
Figure IV —9. MSS Geometric Corrections
-------
31
picture elements that straddle the border between two different classes
of terrain. The resulting spectral signature is a blend of the two
types of material viewed. These effects are noticeable at the border
between two distinctly different classes of material where one or two
pixels on the border will drop into some intermediate spectral class.
For smaller mines, these can represent a significant portion of the
area.
The low ERTS spatial resolution also causes a statistical problem
in selecting a training site commonly composed of only 16 to 25 picture
elements. A much larger number of representative samples of a class is
desirable to adequately define the mean and standard deviation of that
class relative to other classes against which the classification algorithm
is trying to discriminate for the size mines that are being used in this
analysis.
Since the multispectral scanner detects only the optical properties
of the surface observed, there are circumstances where uncertainty can
arise. One of the most common is the extensive areas around a mine from
which the vegetation has been stripped by trucks and other equipment.
These are particularly noticeable in the area about to be mined in front
of the high wall. The exposed soil is difficult to separate from spoils
where the earth has been disturbed. Although graded and ungraded spoils
can be separated, in general, exposed dirt has about the same spectral
signature regardless of its form or physical state. Other sources of
-------
32
misclassification arise when spoils piles have naturally overgrown with
vegetation, in which case they resemble the spectral characteristics of
the natural landscape. In areas that are being revegetated the spectral
difference between grass planted at different times seems to be due to
the ratio of exposed soil to vegetation.
The accuracy and extent to which a mine can be classified also
depends upon the level of detail supplied in the ground truth available.
Three computer runs usually suffice to overcome the mathematical data
analysis problems involved in multispectral classification. The re-
sulting accuracy is then determined by the detail provided in the ground
truth and the ability to secure registration of the satellite image
to the ground truth.
As an example of the changes in the classification map that result
from small changes in definition of the training sites, Figure IV-10
displays two overlays of the northwest corner of the Dave Johnston Mine
that resulted from the second and third attempts at classification.
While small deviations in the revegetated areas defined as Classes 5 and
6 are noticed, the general features remain consistent. Class 6 has two
arms extending northward that enclose the Class 5 area on the east and
west. Two sections of revegetated area extend northward with the Class
6 area to the west of Class 5. A small finger of Class 6 extends south-
ward on the east side of yet another class of reclaimed spoils.
-------
33
The differences in the classification maps were caused by expanding
the size of the training areas by one additional column to include
additional representative samples of the different classes. To illus-
trate how small the differences between the classes are in the second
and third classification runs, the numerical values for the intensities
can be examined. The light intensities in each spectral band are ex-
pressed on an arbitrary scale from 0 to 256. The intensities of the plus
and minus one standard deviation points about the mean value are given
below.
Table IV-2
Standard Deviation Point Intensities
Class
5
6
Run
2nd
3rd
2nd
3rd
Green
Band
102
104
90
88
- 114
- 112
- 98
- 98
Red
Band
114 -
118 -
96 -
94 -
126
130
106
106
Infrared
Band 1
118 -
122 -
106 -
108 -
128
132
116
116
Infra-ed
Band 2
116 -
120 -
106 -
106 -
124
126
114
114
-------
REVECETATED SPRING 1973
TOPSOIL BORROW AREA
REVECETATEO SPRING 1972
HIGH WALl
2ND RUN
3RD RUN
Figure IV —10. Comparison of Successive Classifications
-------
35
To determine the accuracy of the computer classification, an overlay
drawing was prepared from a 1974 aerial photograph outlining the features
of the mine at the same scale as the computer line printer overlay. An
exact comparison could not be made because the horizontal and vertical
dimensions of the line printer output differ 20 percent in scale. The
two maps of Mine 4 were brought into optimum registration and over-
printed in Figure IV-11.
Comparison of the two maps shows that in the northwest section of
the mine roads were included as a part of the revegetated areas in the
computer classification. A small strip of land between the mine haulage
road and the automobile road to the west was included as part of the
mine. The northwesternmost tip of the mine seems to have been revegetated
recently where the computer classified a high wall area. In the center
of the mine a revegetated area just east of the road leading to the
second cut but unmarked on the company supplied map is discerned. A
road and some unreclaimed spoils cutting across the mine to the high
wall was classified as background. Revegetated areas planted in the
spring and fall of 1972 were classified as background; examination of
photographs showed that the revegetation that had grown in these areas
was more dense than in other revegetated plantings and more closely
resembled the natural vegetative cover of the background classification.
The 1974 high wall extends further to the northeast than the positions
determined by the computer analysis in 1973; this is the area currently
being mined.
-------
36
A numerical comparison of the areas allotted to high wall and
spoils shows that the computer recorded 86.3 percent of the acreage
shown in the aerial photo one year later. The computer counted
105.7 percent of the acreage devoted to revegetated land and topsoil
borrow. The overall area of the mine as determined by the computer
was 96.9 percent of the area recorded in the aerial photo one year
later.
-------
\
Aerlol Photo Mop of Min« 4 (1974)
Satollila Mop of Mine 4 (I973|
0 0.5 1.0 15 20 1.1
KIlOMITIti
Figur« IV —11. A Companion of Mine 4 Clarification Mop»
LEGEND
A.— ACTIVt MINI
§.— HlveOf TATID'SfOUS
C.— TOPSOIl 1OIHOW
0.— EXPOSED SOU
I.— IOADS
CO
-------
39
V. RESULTS AND EVALUATION OF THE DATA ANALYSIS
This section presents the results of the computer analysis of the
ERTS satellite data for coal strip mines in the Northern Great Plains.
Of the thirty mines identified via satellite, fourteen were successfully
classified using the techniques described in Section IV. Eight mines
were not classified because of their small size or lack of ERTS data.
Two of the mines were not classified because the necessary computer
compatible tape was not received from the NASA data base; classification
results will be reported when available. Three mines were not yet in
operation and three underground mines were not classified.
Satellite images are presented for each mine classified. A classi-
fication map is also presented showing the shape, location and relative
size of the active mine areas, graded and ungraded spoils piles, and
areas in various stages of revegetation. Where appropriate, adjacent
topographical and cultural features such as stream beds, cropland, roads
and railroads are shown. The area of each type of land use is tabu-
lated.
The problems encountered in applying ERTS technology to the classi-
fication of strip mines are also discussed. Such problems include
masking of disturbed areas by revegetation and similarities between
cropland and mine areas.
Coal production figures given for the mines for 1972, 1975 and 1980
were obtained from the mining companies for the Northern Great Plains
Research Committee.
-------
40
WYOMING MINES
MINE 1 - WYODAK RESOURCES DEVELOPMENT CO.. UYODAK MINE
The Wyodak mine, about 10 km (6 mi) east of Gillette [Fig. V-l,
inside back cover], produced 0.56 million metric tons (0.-62 million
tons) of coal in 1972. Coal production for 1975 and 1980 is projected
to be 0.9 and 1.4 million metric tons (1.0 and 1.5 million tons),
respectively. The thickness of the coal deposit averages about 27 m
(90 ft) within the mine.
Based on a satellite image of this area, the surface area of the
mine was about 27 hectares (66 acres) in 1973. The computer tape for
this mine was not received from the NASA data base; however, when
available, a classification map will be addended to this report.
MINE 2 - AMAX COAL CO., BELLE AYR MINE
About 22 km (14 mi) south of Gillette [Fig. V-l], this mine yielded
only 0.027 million metric tons (0.03 million tons) of coal in 1972. A
major increase in production is expected in the future with projected
levels of 1.6 and 8.6 million metric tons (1.75 and 9.5 million tons)
for 1975 and 1980, respectively. The coal deposit averages 21 m (70 ft)
thick.
The surface area of this mine, as determined from a satellite
transparency recorded in July 1973, was about 85 hectares (210 acres).
The computer tape for this mine was not received from the NASA data
base; however, when available, a classification map will be addended
to this report.
-------
41
MINE 3 - BEST COAL CO.. EAST ANTELOPE MINE
This mine is about 80 km (50 mi) north of Douglas in the Thunder
Basin National Grassland [Fig. V-l]. The mine is small with a 1972
production of only 910 metric tons (1,000 tons). No increase in pro-
duction is expected through 1980. The average thickness of the coal
deposit is about 12 m (40 ft).
The surface area of the mine, measured from a USGS aerial photo-
graph was about 1.5 hectares (3.7 acres) in 1969. The land surrounding
the mine is characterized by essentially bare soil with little vegetation.
The optical characteristics (spectral signature) of this land as re-
corded in the ERTS data were nearly identical to those of the mine area
precluding any meaningful computer classification of land use within the
mi ne.
-------
42
MINE 4 - PACIFIC POWER AND LIGHT CO.. DAVE JOHNSTON MINE
Background
Situated about 48 km (30 mi) northeast of Casper and 26 km (16 mi)
north of Glenrock [Fig. V-l], the Dave Johnston Mine produced 2.4 million
metric tons (2.6 million tons) of coal in 1972. Production is expected
to increase only slightly as reflected by projections of 2.4 and 2.5
million metric tons (2.7 and 2.8 million tons) for 1975 and 1980, re-
spectively.
As shown in the satellite image [Fig. V-2], the mine has a long,
narrow configuration. The 8 km (5 mi) long high wall lies in a south-
east direction. Excavation is advancing to the northeast into an arid
landscape thinly covered by sagebrush and prairie grass. Spoils piles
and revegetated areas lie to the southwest of the high wall.
Revegetation was begun in 1969 and closely follows the excavation activi-
ties. The area of unreclaimed spoils piles is thus relatively small.
The status of reclamation as of Spring 1973 is shown in Figure V-3
provided by the mining company.
For verification, aerial photographs of various areas of the mine
were taken by NFIC-D personnel, four of which are shown in Figures V-4
through V-7.
Classification of Mine Areas
As shown in Figure V-3, about 55 percent of the disturbed area of
the mine has been revegetated. The computer classification defined the
disturbed areas but difficulty was encountered in separating older
-------
43
-N-
•i^E_i_sa
I ((i! i •. I M 1111; \ i i i •! ' i' ' i 11' f • • • •• : •' '• •', i
- "*•-». ' *''"'** '" "
•- ,
MINE NO. 4
'H.fi.P.1
c ;'4::-i:-iii05-4;: : :
JBn
Figure V-2. Satellite Image of Mine No. 4
-------
>>v
//• .—/^v"
:- f J§* ^V
«^v-n
^
Figure V — 3. Revegetation of Dave Johnston Mine
(Map Provided by Pacific Power and Light Co.)
-------
Figure V-4. Figure V-5.
Topsoil Borrow Area in Foreground. View of Southeast End of -Mine Looking North,
Revegetated Area Between Road and Spoils Pile. Note Topsoil Borrow Area North of Mine.
en
-------
Figure V-6. Figure V-7.
Middle Area of Mine Showing Second Cut. Middle Area of Mine Showing Discontinuity
in High Wall.
-------
47
revegetated areas from undisturbed native vegetation. This separation
was achieved using Figure V-3. Difficulty was also encountered in
defining the high wall location in the northwestern area of the mine
during the initial classification. The high wall was then defined by
preparing a ratio image of the green band to the infrared 1 band. A
composite map [Fig. V-8]* of the classifications of various mine areas
was prepared using both the ratio image and the multispectral classi-
fication.
In the northwestern corner of the mine, land revegetated in Spring
1973 was discernible from grass in an adjacent area planted one year
earlier. A topsoil borrow area on the west side was also identified.
Proceeding to the southeast, revegetated areas planted in 1969, 1970 and
1971 as designated on the map produced spectral signatures the same as
undisturbed native vegetation thus indicating a comparable density of
vegetation had been achieved.
The southeastern portion of the mine contains much revegetated land
and many topsoil borrow areas. In the borrow areas, a scraper has
removed strips of topsoil leaving native vegetation in alternating
strips as shown in Figure V-4. Some natural revegetation has taken
place in the disturbed strips. The revegetated land and the borrow
areas were spectrally indistinguishable from one another as both consist
of areas of vegetation mingled with bare soil.
* Some scale distortion is inherent in producing an overlay map (on
the computer line printer) which can be avoided only by additional
processing. In scaling overlay maps to USGS topographic maps, the
longitidinal axis of the former must be reduced to 80% of its nominal
value.
-------
48
Figur* V —3. Classification Mop of Mine 4 — Pacific Power and Light Company,
Dave Johnson Mine, Glenrock, Wyoming
-------
49
The land north of the high wall in the southeastern part of the
mine had a signature characteristic of revegetation or topsoil borrow.
No activity in this area was indicated on the company map [Fig. V-3].
An NFIC-D aerial photograph [Fig. V-5] verified, however, that this was
a topsoil borrow area.
Characteristics of the most extensive spoils piles near the second
cut in the center of the mine are shown in Figures V-6 and V-7.
By counting pixels, the computer keeps track of the area in each
classification. The area of each classification for the Dave Johnston
Mine is presented in Table V-l. Of the 554 hectares (1,365 acres) of
disturbed area, about 55 percent has been revegetated.
Table V-l
Areal Distribution of Land Use
Mine 4
Land Use
High wall and spoils
Northwest Portion of Mine
Spring 1973 Revegetation
Topsoil Borrow
Spring 1972 Planting
Spring 1971 Planting
Fall 1970 Planting
Fall 1969 Planting
Southeast Portion of Mine
Spring 1972 Planting
Fall 1972 Planting
Fall 1972 Planting and Topsoil
Borrow in Southwesternmost
Corner
Total Mine Area
Pixels
508
26
7
67
17
33
46
196
30
36
368
434
1,138
Area
Hectares
247
13
3
33
8
16
22.
95
15
18
179
212
554
Acres
610
31
8
80
20
40
55
234
36
43
442
521
1,365
-------
50
MINE 5 - BIG HORN COAL CO.. BIG HORN MINE
Background
This mine, at Acme about 11 km (7 mi) north of Sheridan, produced
nearly 0.9 million metric tons (1.0 million tons) of coal in 1972
[Fig. V-l]. Production is expected to remain the same through 1980. The
mine lies in a valley at the confluence of Goose Creek and the Tongue
River. Mine 6, which is discussed in the next section, is a few kilo-
meters to the west.
Classification of Mine Areas
The satellite imagery [Fig. V-9] revealed a complex setting for
this mine. The mine is bound on the north by the Tongue River and on
the west by a small reservoir on Goose Creek. Graded spoils lie between
the reservoir and the active area of the mine [Fig. V-10]. On the east
is a bluff with a surface of eroded soil and brush that gives the same
spectral signature as the graded spoils. The mine is being extended
eastward into the bluff.
To the west of the mine, a gravel pit and a coal loading area are
distinguishable. Across the Tongue River Valley to the north of the
mine, an area of abandoned spoils piles and a pond from previous mining
activity were defined. The area disturbed by mining activities in this
vicinity is shown in Table V-2.
-------
51
ABANDONED MINE
SLATER CREEK
-N-
11 lltjItttj'lUjll II JIHI JHIIJIHU 11 t I j I I I I j I I ! I (III!
----
1330-17232-L-5 UED 17WJG73 E-14
-------
52
Tongul Rivtr Vollty-
Tongu* River Voll.y
ACTIVE MINE
SPOILS PIUS NORTH OF RIVER
GRADED SPOILS
I I POND
Figurt V —10. Claisificotion Mop of Mine 5— Big Horn Cool Company,
Big Horn Mint, Acmt Wyoming
-------
53
Table V-2
Area! Distribution of Land Use
Mine 5
Land Use
Active mine
Graded spoils
Total, Big Horn Mine
Abandoned spoils north of
Tongue River
Total, Acme, Wyoming
Pixels
44
90.
134
5!
191
Area
Hectares
21
44_
65
28
93
Acres
53
108
161
68
229
MINE 6 - WELCH COAL CO.
Background Information
Welch Coal Company operates a small strip mine about 16 km (10 mi)
northwest of Sheridan anc! a few kilometers west of the Big Horn Mine
[Fig. V-l]. Production in 1972 was only 18,100 metric tons (20,000 tons)
with no projected production through 1980.
The active mine is very small. Computer classification was thus
difficult and of little value. However, extensive spoil piles of an
abandoned mine in the area north of the Tongue River and the town of
Kleenburn [Fig. V-9] were classified and are discussed below.
Classification of Abandoned Mine Areas
Four areas of spoils piles with varying degrees of re-vegetation
are apparent [Fig. V-ll]. Several small ponds were defined in the three
major spoils areas. Exposed spoils in the three areas yielded differing
spectral signatures.
-------
54
Spoils in the northwestern area were overgrown with vegetation. In
the northeastern area, the spoils were uniformly graded and gave the
highest intensity signature because they were oriented in an optimum
direction (east-west) for reflecting sunlight.
Several ponds and areas of vegetation break up the structure of the
middle spoils pile. By carefully locating the training areas, the spectral
signature of the spoils alone was determined.
Much of the southeastern area is covered by ungraded spoils. Some
of the spoils are partially overgrown with vegetation.
To the west of the middle spoils is an area where much light-colored
soil is eroded and exposed. Such an area is frequently classified into
the same categories as the spoils piles because of similar spectral
signatures.
The areal distribution of each classification of the abandoned spoils
piles is summarized in Table V-3.
Table V-3
Areal Distribution of Land Use
Abandoned Spoils Near Mine 6
Land Use
Overgrown spoils
Lakes (total)
Northern section spoils
Middle section spoils
Southern section spoils
Total Area
Pixels
32
41
67
113
117
370
Area
Hectares
16
20
33
55
57
180
Acres
38
49
80
136
140
444
-------
55
LEGEND
•EVEGETATED SPOILS
IRREGULAR SPOILS
REGULAR SPOILS
B§?^ STREAM BED
' ' POND
EXPOSED SOIL
SLATER CREEK
Figur« V —11. Cloilif icotion Mop of Abandoned Spoilt Pil»« North o( Min« 6-
W«lch Coal Company, Kleonburn, Wyoming
-------
56
MINE 7 - CARTER OIL CO.
Carter Oil Company has proposed a mine about 12 km (7 mi) north
of Gillette [Fig. V-l]. It is doubtful that mining will begin before
1985.
MINE 8 - ATLANTIC RICHFIELD
Situated 68 km (42 mi) south of Gillette, this mine is just begin-
ning operations with 1975 production estimated to be 0.18 million metric
tons (0.2 million tons) [Fig. V-l]. Production is expected to increase
sharply, reaching 6.4 million metric tons (7.0 million tons) annually by
1980. No indication of the mine was visible on the ERTS imagery.
MINE 9 - REYNOLDS METALS, INC.
A mine has been proposed 32 km (20 mi) south of Sheridan [Fig. V-l].
Production plans are uncertain.
MINE 10 - KERR-McGEE
No 1975 production is planned at this proposed mine 63 km (39 mi)
south of Gillette and a few kilometers north of Mine 8 [Fig. V-l].
Coal production of 6.4 million metric tons (7.0 million tons) is planned
for 1980.
-------
57
MONTANA MINES
MINE 11 - DECKER COAL CO.. DECKER MINE
Background
West of the Tongue River Irrigation Reservoir 30 km (19 mi) north
of Sheridan, Wyoming, this mine is about 24 km (15 mi) northeast of
Mines 5 and 6 [Fig. V-12, inside back cover]. Coal production was only
0.7 million metric tons (0.8 million tons) in 1972; however, a major
increase in production to 5.4 and 11.8 million metric tons (6 and 13
million tons) is expected for 1975 and 1980, respectively, making this
the second largest mine in the Northern Great Plains.
The satellite imagery [Fig. V-13] shows that the disturbed area of
the mine is characterized by a circular pattern. The loading area is in
the center with radial roads extending to the active mine area. The
mine is advancing along a circular arc into the hillside to the west and
south.
Classification of Mine Areas
In addition to the high wall, two types of disturbed areas were
present in the mine area. The area to the west and south of the high
wall was being prepared for mining and the vegetation had been removed
[Fig. V-14]. However, the surface had not been disturbed to any great
extent. The land to the east and north of the high wall contained
spoils piles. Both types of disturbed areas displayed the same spectral
signature precluding separation by computer analysis.
-------
58
MINE NO. 11
ItmmMimi iiiimmiun
"""" * -^**-*fe^fi«i4»«* •• •• MMttMHBBMIMi
, -pTv, ,- -•.taw
II litiltttitill 11M 111111111
1390-17S3M-5 FED 17AUG73 £-14
(N.G.P.)
C N44-37/HIGS-37 C M44-34/U106-24 HM 151
•SUM EL51 «135 5433
APPLIED :KDJ = £.74 DEGREE:. ft^PECT =
1.392930
»* CTPP MINE HO, 11 »«
KTRTCH2
STPETCH 5 32-112
JDA AiJG 19» 1974 124416 JPL/IPL
TONGUE RIVER
RRIGATION RESERVOIR
-N-
Figure V-13. Satellite Image of Mine No. 11
-------
59
LEGEND
ACTIVE MINE AREA (HIGH WALL)
UNGRADED SPOILS
li-UJ STREAM BOTTOM
II SHALLOW WATER AND PONDS
^H DEEP WATER
Tongut Siv.r Irrigalio
Mint Offlc«l
Figure V—14. Clai iificati on Mop of Mine 11— Decker Cool Company, D e e k e r, Monte
-------
60
Because the active area is a narrow cut, the picture elements blend
the active area with the disturbed areas on both sides. The pure
spectral signature of the active area was present for the north-south
cut, but the majority of the picture elements covering the active area
fell into a different class because of the blending. This effect was
particularly noticeable in the southern portion of the mine. Small ponds of
water were defined in the active area.
The upstream end of the Tongue River Irrigation Reservoir is east
of the mine [Fig. V-13]. Two distinctly different spectral signatures were
observed in this portion of the reservoir. Either the end of the reservoir
has been shoaled by silt deposition giving the light image area, or more
probably, turbid river inflow had discolored the southern end of the reservoir.
The v&lley of the Tongue River south of the reservoir was clearly
distinguishable.
The areal distribution of land use at this mine is summarized in
Table V-4.
Table V-4
Areal Distribution of Land Use
Mine 11
Land Use
Active mine
Ponds in mine
Spoils
Area in preparation
for mining
Total Area
Pixels
83
9
90
107
289
Area
Hectares
40
4
44
52
140
Acres
100
11
108
128
347
-------
61
MINE 12 - WESTMORELAND RESOURCES. SARPY CREEK MINE
Background
Westmoreland Resources operates this strip mine about 40 km (25 mi)
northeast of Hardin in the upper Sarpy Creek drainage area [Fig. V-12].
The mine began operation in 1973. Projected annual production for 1975
and 1980 is 3.6 million metric tons (4 million tons).
A small tributary of Sarpy Creek bisects the mine site. A substan-
tial area of cropland (light-colored areas) is present in the vicinity
of the mine as shown in the satellite image [Fig. V-15].
Classification of Mine Areas
This mine had just begun operation when the ERTS image was taken
during the summer of 1973. A large area in the vicinity of the mine was
distinguishable as disturbed area, different from the native rangeland
[Fig. V-16]. Separation of this area into mined areas and cropland was
difficult.
Those areas classified as mined areas exhibited spectral signatures
similar to the high wall and spoils areas of Mine 14 about 32 km (20 mi)
to the east on the same ERTS frame. Close examination of a 1971 US6S
aerial photograph of the mine vicinity revealed that part of the area
south of the tributary stream, which showed the same spectral signature
in 1973 as the mined area, was actually cropland in 1971. The area to
the north of the stream classified as mined area does not appear to be
cropland in the 1971 aerial photograph. Whether this area is being
mined or has been placed in cultivation since 1971 is unknown.
A summary of the land classification is shown in Table V-5.
-------
62
SARPY CREEK
1373-1729Q-R-5 IO 31.JUL73 .£-9
(H.G.F'.)
C N45-58/U107-2e S H45-5S4iiO?-17 Hffi 1
SUM EL54 A2132 5201
APPLIED SKEW = £.67 DEGKEC'. ASPECT =
• SIMP HUE NO. 12 »»
iriSEPT
A1TPTCH2
ITRETCH I 42- 99
TRIBUTARY
MINE NO. 12
-N-
JDA AU6 19t117i_123SOS JPL/IPL
Figure V-15. Satellite Image of Mine No. 12
-------
63
LEGEND
ACTIVE MINE
BACKGROUND VEGETATION
I I IACKGIOUND VEGETATION
STREAM IED
Figur. V-16. Cloilllicalion Map ol Min. 12— W.itmor. Land Re1ourc«l.
Sorpy Cf««k Min«, Hardin, Wyoming
-------
64
Table V-5
Areal Distribution of Land Use
Mine 12
Land Use
Mined area north of tributary
Mined area south of tributary
Total area
Pixels
99
53
152
Area
Hectares
48
26
74
Acres
119
64
182
MINE 13 - PEABODY COAL CO.. BIG SKY MINE
Background
Peabody Coal Co. operates the Big Sky Mine about 8 km (5 mi)
southwest of Colstrip [Fig. V-12]. Annual coal production is estimated
to be 3.6 million metric tons (4 million tons) during the 1975 to 1980
time period, a significant increase over 1972 production of 1.4 million
metric tons (1.6 million tons).
The Big Sky Mine and the large Rosebud Mine at Colstrip are visible
in the satellite image [Fig. V-17]. Colstrip is in the headwaters of
Armells Creek, a small tributary of the Yellowstone River.
C1 assi fi cati on of Mi ne Areas
Comparison of the classified image with a USGS aerial photograph indv
cated that excellent discrimination was achieved between graded and
ungraded spoils [Fig. V-18]. Disturbed ground in the railroad car
loading area and an area where turning trucks had eroded the grass from
the soil yielded the same spectral signatures as the spoils.
Two classes were identified for surrounding undisturbed areas. One
was grass covered, the other grassy areas interspersed with patches of
-------
65
ARMELLS CREEK
-N-
MINE NO. 13
1373-17290-R-5 FED 31JUL73 E-?
(H.G.P.)
C fM5-53.-M107-28 £ N45-5&/M07-17 MM 1
iUM £154 6Z132 5^01
APPLIED SKEW = t.V PEGPiE:. AIPECT =
i.'3^059
•» iTRIP !1IflE: MQ. 12U4 **
IfCEPT
(WTPTCH2
:TPETCH 45-11^
JDA MJG T?» W74 123530 JPL
MINE NO. 14
RAILROAD
Figure V-17. Satellite Image of Mines No. 13 and 14
-------
66
-N-
Truck Turning Area
Railroad Loading Area
ACTIVE MINE AND
UNGRADED SPOILS
GRADED SPOILS
Figure V—18. Classification Map of Mine 13— Peabody Coal Company,
Big Sky Mine, Colstrip, Montana
-------
67
evergreens. Comparison with the aerial photograph indicated that these
types of vegetative cover were successfully separated.
The area of disturbed land is summarized in Table V-6.
Table V-6
Areal Distribution of Land Use
Mine 13
1 anrl Ifco
Railroad loading and turnaround
Graded spoils
Active mine and ungraded spoils
Total area
Pixels
47
51
87
185
Area
Hectares
23
25
42
90
Acres
56
61
104
222
-------
68
MINE 14 - WESTERN ENERGY CO., ROSEBUD MINE
Background
The Rosebud Mine at Colstrip is the largest coal strip mine in the
Northern Great Plains [Fig. V-12]. It produced 5 million metric tons
(5.5 million tons) of coal in 1972 with the same value projected for
1975. By 1980 the yield of the mine is projected to increase nearly
four-fold to 17.5 million metric tons (19.3 million tons) per year.
Because of its size, the mine is an excellent target for satellite
imagery and does not pose the problems of small size training areas and
blending of spectral signatures that are associated with smaller mines.
The mine is clearly visible in the satellite image [Fig. V-17]. Geo-
metrically the mine is complicated, consisting of three arms of spoils
piles in the northern section, a northwest-southeast central arm, a
goose neck and the southern active mining area.
Classification of Mine Areas
The mine was classified in two steps since the initial aerial
photograph obtained from the Company covered only the southern section.
Classification of the southern section defined the high wall, the spoils
piles, an older revegetated area, and a more recent attempt at revegetation
[Fig. V-19]. Also identified were the areas in preparation for mining
south of the high wall and a railroad and highway corridor skirting the
east side of the mine. Distinguishing the mined area from eroded soil
west of the southern section and the goose neck proved difficult.
-------
KZI
LEGEND
ACTIVE MINE
•(VEGETATED
GtADED SPOILS
UNGRADED SPOILS
NATUHAUY REVEGETATiD
MISCELLANEOUS
EXPOSED HOCK OUTCIOPSING
HAILIOAD AND HIGHWAY CO««IDO«
Figure V —19. Classification Map of the Southern Section of Mine 14 — Western Energy Company,
Rosebud Mine, Colstrip, Montana
cn
-------
70
When the spectral signatures derived from training sites in the
southern part of the mine were applied to the northern part, the results
were poor due to the changing character of the spoils. New training
sites were thus established in the northern part of the mine. Exam-
ination of aerial photographs covering the northern section shows that
the spoils piles occur in well-ordered rows that are oriented along the
direction of the arms of the mine. The orientation of the spoils piles
to the incident solar radiation thus changes with location, also changing
the spectral signature.
Figure V-20 illustrates the classification derived from training
sites on spoils piles in the northern section of the mine, nlus a few of
the original training sites from the southern section of the mine. Some
overlapping occurred since several different classes of spoils piles
were used that closely resemble each other spectrally. It was possible,
however, to distinguish between graded, irregular ungraded, regular
(uniform) graded and partially overgrown spoils.
The high wall areas in the northern section of the mine did not
classify separately because they were too narrow. In the one successful
attempt at classifying the northern section highwall, only a few pixels
one column wide were detected. Due to natural revegetation, the western
part of the spoils (shown by the dashed line) near the landing strip on
the southwestern arm classified as background. The area northeast of
the drainage ditch has naturally revegetated. The full extent of the
area mined was thus not classified as disturbed area.
-------
71
DIAINAOf OUCH
Figure V —20. Cloilifieotion Mop of the Northern Section of Mint U — W«it«rn En«rgy Company,
Roi*bud Mine, Colttrip, Montana
-------
72
In the diagonal middle part of the mine, the drainage ditches and
roads that cut across the graded spoils [Fig. V-20] were identifiable.
Table V-7 summarizes the classification of mine areas.
Table V-7
Areal Distribution of Land Use
Mine 14
Land Use
Northern section
High wall
Ungraded spoils
Partially overgrown spoils
Subtotal
Middle section
Graded spoils
Ungraded spoils
Roads and drainage ditches
Subtotal
Southern section
High wall
Graded and seeded
Ungraded
Graded for special studies
Subtotal
Total high wall
Total graded
Total ungraded
Road and ditches
TOTAL AREA
Pixels
26
479
530
299
335
132
766
100
85
207
241
633
126
625
1046
132
1,929
Area
Hectares
13
233
257
145
163
64
372
49
41
101
117
308
61
304
508
64
937
Acres
31
575
636
359
402
158
919
120
102
248
289
759
151
750
1255
158
2,315
-------
73
MINE 15 - KNIFE RIVER COAL CO.. SAVAGE MINE
Background
The small Savage Mine is about 13 km (8 mi) northwest of Savage
near the North Dakota border [Fig. V-12]. Annual coal production from
1972 to 1980 is expected to average 0.29 million metric tons
(0.32 million tons).
Classification of Mine Areas
Four east-west cuts and a smaller cut to the south comprise the
mine [Fig. V-21]. In the north area, the two cuts have advanced toward
each other, meeting in the high wall area shown. A similar mining
pattern was present in the middle section. The small southern cut and
its highwall were separately discernible. Much of the disturbed area
may not have been excavated, but may represent roads and areas stripped
of vegetation in preparation for mining. The classification of mine
areas is summarized in Table V-8.
-------
74 ,
-N-
NORTHERN
TWO CUTS
LEGEND
ACTIVE MINE
(HIGH WALL)
UNGRADED SPOILS
ROAD
MIDDLE
TWO CUTS
SOUTHERN SECTION
Figure V — 21. Classification Map of Mine 15 —
Knife River Coal Company, Savage, Montana
-------
75
Table V-8
Area! Distribution of Land Use
Mine 15
Land Use
Northern two cuts
Spoils, roads and areas in
preparation for mining
High wall
Subtotal
Middle two cuts
Spoils, roads and areas in
preparation for mining
High wall
Subtotal
Southern section
Spoils, roads and areas in
preparation for mining
High wall
Subtotal
TOTAL AREA
Pixels
145
15
160
154
18
172
33
3
36
368
Area
Hectares
70
7
78
75
9
84
16
2
18
179
Acres
174
18
192
185
22
206
40
4
43
442
-------
76
MINE 16 - RELIABLE COAL CO.. RELIABLE MINE
Located 87 km (54 mi) northeast of Billings [Fig. V-12], this mine
produced less than 910 metric tons (1,000 tons) of coal in 1972; 1975 and
1980 projections were not available. Because of its small size this
mine was not classified.
MINE 17 - DIVIDE COAL CO., STORM KING MINE
This small mine is 48 km (30 mi) north of Billings. Previously
an underground mine, surface operations were begun in late 1972, producing
less than 0.11 million metric tons (0.12 million tons) of coal. No
estimates of future production were available. Because of its small
size this mine was not classified.
MINE 18 - WESTERN COAL CO.. WESTERN MINE
This small underground mine, 64 km (40 mi) north of Billings,
produced less than 910 metric tons (1,000 tons) of coal in 1972;
projections for 1975 and 1980 were not available. Since this is an
underground mine, ERTS technology does not apply and the mine was not
classified.
MINE 19 - NIES COAL CO., NIES COAL MINE
Also an underground operation, this small mine 74 km (46 mi) north
of Billings produced less than 0.11 million metric tons (0.12 million
tons) of coal in 1972; projections were not available. The mine was not
classified.
-------
77
NORTH DAKOTA MINES
MINE 20 - BAUKOL NOONAN INC.. CENTER MINE
Background
This mine is about 5 km (3 mi) east of Center [Fig. V-22, inside
back cover] and 2 km west of an impoundment on Square Butte Creek. Coal
production in 1972 was 1.3 million metric tons (1.4 million tons) and
will be similar (1.4 million metric tons) in 1975. By 1980 production
is expected to increase to 4.1 million metric tons (4.5 million tons).
The mine is in two sections with only the north one active. These
areas are clearly visible in the satellite image [Fig. V-23].
Classification of Mine Areas
A thermal electric generating plant adjacent to the impoundment
[Fig. V-24] was defined. An adjacent coal pile was also defined and had
the same spectral signature as dark areas of the mine. The east end of
the road connecting the plant and mine was identified. A parking area
surrounding a large building is alongside this road.
The northern section of the mine is the smaller but is rapidly
expanding. Mining is proceeding to the north and west. The southern
mine section is in the shape of a long arc curving to the southwest.
Some revegetation may have occurred to the east of the high wall and
spoils piles in the southern section. Without ground observations, it
was impossible to distinguish this revegetated land from the surrounding
cropland.
The areal distribution of land use is summarized in Table V-9.
-------
78.
SQUARE BUTTE CREEK.
MINE NO. 20
I Mil I IIII111111111 11 II1111111111 [ 1111 [ III111111 j IIII
I I I I I I I I II I tl I I ! I I III Mill I II I I I I I I
13-^-17054-1-5 RED 2?Jit?: M.j.F. E-7
C rW7-££/U101-l2 '. M'-SmCO-?? HDQ 1?£
iUH EL54 n:i;J !:i!
Irr.ECT
•• :TPI? ni?,r ••:. ^; .»
nff.::: ;•:.. = z.w :ES=E-::. -:=E:T =
i.3»l5Q
ftiWcna
:*:"•> ? 44- 25
Jft WG K.' 1?74 033421 jR/IPl."
-N-
Figure V-23. Satellite Image of Mine No. 20
-------
79
LEGEND
(ffi&l ACTIVE MINE
[£53 UNGRADED SPOILS
K53 IOADS AND POWER PLANT
ti&ft STREAM IED
1^81 RESERVOIR
Figure V — 24. Clauificotion Mop of Mint 24— Boukol Noonan Inc.
C»nl«r Mine, C*nl«r, North Dakota
-------
80
Table V-9
Areal Distribution of Land Use
Mine 20
Land Use
Northern section
High wall
Spoils
Subtotal
Southern section
High wall
Spoils
Road
Subtotal
TOTAL AREA
Pixels
19
66
85
56
100
12
168
253
Area
Hectares
9
32
41
27
49
6
82
123
Acres
23
79
102
67
120
14
202
304
-------
81
MINE 21 - NORTH AMERICAN COAL. INDIANHEAD MINE
Background
Coal production from this mine at Beulah [Fig. V-22] was 0.9 million
metric tons (1.0 million tons) in 1972. Expected production for 1975
will be similar (1.1 million metric tons) but may increase to 4.1 million
metric tons (4.5 million tons) by 1980.
The Knife River a few kilometers to the southeast is between the
Indianhead Mine and the Knife River Coal Mining Co. operation (Mine 22)
[Fig. V-25]. A substantial area in the vicinity of these mines is
cropland.
Classification of Mine Areas
Several examples of the problems involved in the classification of
strip mines using satellite multispectral data were encountered at this
site. In the easternmost toe of the mine [Fig. V-26], the spectral
classification of the graded and seeded area changed in the northern
part due to a dark discoloration of the soil. In the middle of the
mine, an old spoils pile that was overgrown with vegetation had a spectral
signature indistinguishable from vegetation outside the mine. It was
difficult to obtain a representative spoils pile signature anywhere in
the eastern portion of the mine. Close examination of Soil Conservation
Service aerial photographs indicated that there was little consistency
in the size and angular orientation of the spoils piles. Consequently,
the spectral signature varied with location in the mine. The effect
became more noticeable at the northern end of the mine at the site of
current activity. There the spoils were piled up in unusually high and
-------
82
MINE NO. 21
KNIFE RIVER
NORTH MINE NO. 22
-N-
1352T7114-R-5 RED 10JUL73- E-30
; C N47-£54JiO£-34 S H47-25/U102-24 HDG 192
SUH EL57 AZ131 4903
APPLIED SKEW = £.59 DEGREES. ASPECT = ,
1.391117
** STF'IP MINE fffiS. 21i22 »«
ASTRTCH2
STRETCH 36- 83
t T ' f F B
•^^^- -.;, .•JW.^&r
SOUTH MINE NO. 22
Figure V-25. Satellite Image of Mines No. 21 and 22
-------
LEGEND
Figure V — 26. Classification Map of Mine 21 — North American Coal Company,
Indian Head, North Dakota
00
OJ
-------
84
large mounds at varying orientations to the sun, producing an extremely
large statistical variation in the spectral signature. The area was
very difficult to classify and was outlined from pixels that did not
fall into any of the other classifications.
The orientation of the spoils in the western cut of the mine was
relatively uniform and produced a consistent terrain that classified
easily into graded and ungraded spoils.
The classification of mine areas is summarized in Table V-10.
Table V-10
Areal Distribution of Land Use
Mine 21
Land Use
Western arm
Ungraded spoils
Graded spoils
Subtotal
Eastern section
Graded and seeded
Graded and seeded (dark soil)
Graded only
Ungraded spoils
Active mine and large spoils
piles (Northern part)
Subtotal
TOTAL AREA
Pixels
155
62
217
170
114
219
180
278
961
1,178
Area
Hectares
75
30
105
83
55
106
87
135
447
572
Acres
186
74
260
204
137
263
215
334
1153
1,414
-------
85
MINE 22 - KNIFE RIVER COAL MINING CO.. BEULAH MINE
Background
This Company operates two mines near Beulah on the Knife River east
of Mine 21 [Fig. V-22]. One mine consists of two long and narrow cuts
about 8 km (5 mi) south of Beulah clearly visible in the satellite image
[Fig. V-25]. The second mine is a similar distance northeast of Beulah
and north of the Knife River. Coal production through 1980 is expected
to remain at the 1972 level of 1.4 million metric tons (1.5 million
tons).
Classification of Mine Areas
The south mine consists of two long, parallel east-west cuts with
small north-south cuts on the western edge. The spoils have not been
revegetated or graded. The mine was classified into high wall (dark
material) and ungraded spoils (light material) [Fig. V-27].
At the time of the initial computer analysis, the existence of the
north mine was unknown. Therefore, no training sites were selected from
the north mine. Upon realizing that the north mine existed, a computer
classification was performed using the statistics from the training
sites in the south mine. Comparison of the resulting classification map
with a 1971 aerial photograph showed that the areas on the northern and
eastern portions of the mine were fairly well defined but the spoils
piles on the southwestern corner were largely missed. The explanation
for this is not known, but experience with Mine 21 suggests that if the
spoils piles were overgrown with vegetation, they would give the same
classification as the surrounding agricultural land. Questions
-------
86
Figur* V-27. Clot i f i c otio n Map of Min« 22 —
South Mint of Knife •:»•- Coal Company, louloh. North Dakota
-------
87
such as this can only be resolved by more extensive ground observations.
Mine 21 is on the same small element of the image extracted from
the Earth Resources Technology Satellite frame and would be expected to
have spectral characteristics similar to the north mine. The north mine
was classified using the statistics from training sites at Mine 21.
Comparison of the resulting classification map [Fig. V-28] with the
similar map derived using training sites from the south mine showed that
the use of statistics from Mine 21 resulted in systematically larger
areas classified as part of the mine. However, the southwest corner of
the mine was still largely classified as agricultural land. The classi-
fication of the north mine based on Mine 21 training sites was selected
as the most representative. Because of the difficulty in extrapolating
classifications from mine to mine, no further sub-classification of the
disturbed areas of the north mine was attempted. Table V-ll summarizes
the classification of disturbed areas for both mines.
-------
-N-
Figure V — 28. Classification Map of North Mine of
Knife River Coal Mining Co. Beulah, North Dakota
-------
89
Table V-ll
Area! Distribution of Land Use
Mine 22
Land Use
Pixels Hectares Acres
South Mine
North arm
High wall 116 56 139
Spoils 138 67 166
254 123 305
South arm
High wall 151 73 181
Spoils 202 98 242
353 171 424
Subtotal south mine 607 295 728
North mine
Using training sites
from Mine 21 763 371 916
Using training sites
from southern part of
Mine 22 488 237 586
TOTAL AREA 1,370 665 1,644
-------
90.
MINE 23 - CONSOLIDATION COAL CO., GLEN HAROLD MINE
Background
This mine 8 km (5 mi) southeast of Stanton and 64 km (40 mi)
northwest of Bismark [Fig. V-22] produced 1.3 million metric tons (1.4
million tons) of coal in 1972. An annual yield of 3.6 million metric
tons (4 million tons) is forecast for the 1975 to 1980 period.
The mine is in very hilly terrain in the bluffs south of the
Missouri River. The disturbed area is large, as shown in the satellite
image [Fig. V-29]. The physical configuration is heavily influenced by
the terrain that has divided the mine-into a north section and two south
sections separated by a northeast-southwest ridge.
Classification of Mine Areas
At the time of data collection, reclamation had progressed to the
point where most of the mine area could be separated into graded or
ungraded spoils. About half of the west section had been graded with
the rest of the irregular or ungraded spoils [Fig. V-30]. Most of the
east section is ungraded with small irregular sections. The highwall is
the south edge of these two sections but was not detected in the classifi-
cation. Separation of spoils types was not clear in the north area.
Table V-12 summarizes land use in the mine.
-------
91
>
MISSOURI RIVER
MINE NO. 23
-N-
1&M7054-L-5 RED 27JUL73 H.G.P. £-7
C f!47-E£'m01-i£ S H47-2i4ttQ8-S9 HOG Itt
SUM EL54 S2133 5145
flPPLlED SHU = £.60 I'EGREt;. ASPECT =
1.391150
** STRIP HM MQ. 23 »«
ASTPTCH2
STRETCH 0 35- %
JUA AUG 26. 1974 033321 JPL/IPl
Figure V-29. Satellite Image of Mine No. 23
-------
92
LEGEND
GRADED SPOILS
UNGRADED SPOILS
MIXED SPOILS, ROADS
ROAD
Figure V —30. Clonificotion Mop of Mint 23— Ccniolidation Coal Company,
GI»n/Harold Mine, Stanton, North Dakota
-------
93
MINE NO. 25
-N-
1333-17104-R-5 REB 15flUG73 M.G.P. E-ll
C N48-50/U101-58 S N43-4S4I101-47 HUG 192
SUfl 0.49 A2140 5410
APPLIED SKEU = 2.52 DEGREE',. ACPECT =
1.390208
M STRIP MINE 1C. 25 «»
ASTRTCH2
STRETCH 9 32- 67
JM m 26» 1974 03i5£t JPL'IPL
Figure V-31. Satellite Image of Mine No. 25
-------
94
Table V-12
Areal Distribution of Land Use
Mine 23
Land Use
North Section
Assorted spoils
West Section
Graded spoils
Ungraded spoils
East section
Graded spoils
Ungraded spoils
Roads
TOTAL AREA
Pixels
170
209
167
128
407
65
1,146
Area
Hectares
83
102
81
62
198
32
557
Acres
204
251
200
154
488
78
1,375
MINE 24 - BAUKOL NOONAN, INC.. NOONAN STRIP MINE
This mine 5 km (3 mi) east of Columbus [Fig. V-22] produced
0,4 million metric tons (0.5 million tons) in 1972. An aerial
photograph of the mine indicated that, its two large areas of spoils
piles measure 5 km (3 mi) by 1.3 km (0.8 mi) and 2,4 km (1.5 mi)
by 2.i km (1.3 mi). As the mine is closed with no future production
expected, no classification was performed.
-------
95
MINE 25 - CONSOLIDATION COAL CO.. VELNA MINE
Background
Located 40 km (24 mi) southeast of Minot [Fig. V-22], this mine
produced 0.43 million metric tons (0.47 million tons) of coal in 1972.
Production for 1975 and 1980 is estimated to remain at this level.
Classification of Mine Areas
The satellite image of this mine [Fig. V-31] was very difficult to
interpret visually. But two sources assured that the mine is located
within this particular part of the ERTS frame: one was an accurate
examination of the ERTS frame at the known geographical coordinates of
the mine; another was the location of the mine at the northern edge of
an easily identifiable strip of land that represents the terminal moraine
of the continental ice sheet during the Glacial Age. A careful comparison
of the satellite image with aerial photographs showed two bright areas
at the ends of the mines that can be identified, and dark vegetation is
growing in stream bottoms immediately southwest of the mine.
All attempts at multispectral classification failed to separate the
middle area of the mine lying between the two bright areas at the ends
from the surrounding cropland. In classifying other mines it was observed
that when the land surrounding the mines was planted in the rectangular
pattern common in midwestern agricultural practice, an effective camouflage
pattern was formed. That factor made the mine very difficult to distinguish
with the multispectral recognition algorithms currently used.
-------
96
MINE 26 - KNIFE RIVER COAL CO.. GASCOYNE MINE
Background
Located 8 km (5 mi) northeast of Gascoyne [Fig. V-22], this mine
produced only 0.15 million metric tons (0.16 million tons) of coal in
1972. A major expansion of mining activity is planned, however, with
annual production for the 1975 to 1980 period forecast at 2.7 million
metric tons (3 million tons).
Classification of Mine Areas
A majority of the land in the vicinity of the mine is cultivated
farmland as shown in the satellite inage [Fig. V-32]. Discrimination
between mine areas and the adjacent striped cropland proved to be diffi-
cult. Only two classes of spoils (light and dark) could be separated
[Fig. V-33]. The mid-section of the mine was very irregular and contain-
ed several small ponds. The computer classification map of this area
was also irregular and difficult to interpret. Several small remotely
located pits to the east and south of the mine were picked up in the
computerized classification. The coal loading area to the south of the
mine was distinct. Extensive enlargement of the mine had occurred
compared to a 1957 aerial photograph. The areal distribution of land
use is summarized in Table V-13.
-------
97
i I»hi n n i ti i iimn H Jimum j JIM
II III I I) li i I Ml II Mill HIM I I I II Illllll [ I IIHl (I tilt I
. 135£-171£0-L-5 PED 10JUL73 M.G.P, E-i
: C H46-014I1W-09 3 H4t-uO/U10i-5'? HM 192
JUM EL53 AZ129 4903
IfCECT - ^FLIED 'KIU = £.67 DEGREES.
ft'PECT = 1.3920&
*• STRIP HIIC HD. 26 ««
STRETCH J 35- 93
JWt SER.27. 1974,084033 JPL/IRU
MINE NO. 26
-N-
Figure V-32. Satellite Image of Mine No. 26
-------
98
LEGEND
UNGRADED LIGHT SPOILS
MIXED LIGHT AND DARK SPOILS
COAL LOADING AREA
SPOILS WITH NUMEROUS SMALL PONDS
Figure V—33. Classification Map of Mine 26 —Knife River Coal Mining Company,
Gascoyne, North Dakota
-------
99
Table V-13
Areal Distribution of Land Use
Mine 26
Land Use
Active mine and spoils
Small ponds
Railroad loading area
TOTAL AREA
Pixels
192
37
26
255
Area
Hectares
93
18
13
124
Acres
230
44
31
306
MINE 27 - BONSNESS COAL MINE
Production in 1972 from this mine 5 km (3 mi) south of Columbus
[Fig. V-22] was less than 0.11 million metric tons (0.12 million tons).
The mine was closed by 1973 and it was therefore not classified. An
2
aerial photograph shows that spoils piles cover an area of about 5.8 km
(2.3 mi2) in a strip 7.2 km (4.5 mi) by 0.8 km (0.5 mi).
MINE 28 - UNDERWOOD COAL CO.. INC.. UNDERWOOD MINE
This company operates a mine 3 km (2 mi) southeast of Underwood
[Fig. V-22]. Production in 1972 was less than 0.11 million metric tons
(0.12 million tons). Projections for 1975 and 1980 were not available.
This mine was not located on the ERTS imagery and, therefore, was
not classified. On a Soil Conservation Service aerial photograph, the
mine dimensions were roughly 400 m (1300 ft) by 270 m (900 ft).
-------
100
MINE 29 - HUSKY BRIQUETTING INC.. LEHIGH STRIP MINE
Background
This small mine 5 km (3 mi) east of Dickinson [Fig. V-22] produced
less than 0.11 million metric tons (0.12 million tons) in. 1972.
Production estimates for 1975 and 1980 were not available.
Classification of Mine Areas
The small size of this mine compared to the spatial resolution of
the ERTS imagery [Fig. V-34] and the presence of cropland in the vicinity
limited the accuracy of the classification. The mine is divided into
three sections by the Heart River and'the Northern Pacific Railroad
[Fig. V-35].
Inspection of an aerial photograph indicated that spoils in the
north area are overgrown with vegetation. Two of three ponds were
identified in the classification map.
In the west section, the bright spectral signature of the spoils
was indicative of recent mining activity. Two of three ponds in this
area were also identified. A 17 hectare (7 acre) area that has been
revegetated was not identified in the computer classification map due to
its small size.
In the southern part of the mine, the large pond and mine buildings
were dominant. The spoils piles in the northern part of the southern
segment were overgrown with vegetation and did not classify uniquely.
-------
101
135M7114-L-5 PEB 10JUL73 E-30
C N47-254J1Q2-34 S N47-25/II102-S4 HDG 192
SUN EL57 ^2131 490S
INSECT
«* STRIP MItlE fO. 2? «
^PPLIEB SKEU = £.59 I'EGPIE:, ASPECT =
1,391117
ftSTRTCH2
STRETCH t 33- 37
JDA SEP 27» 1574 084222 JPL/IPL
MINE NO. 29
-N-
Figure V-34. Satellite Image of Mine No. 29
-------
102
-N-
ERODED SOIL
MINE BUILD INGS
LEGEND
GRADED SPOILS
LIGHT COLORED SPOILS AND BARE SOIL
NATURALLY REVEQETATED SPOILS
LISHT COLORED QROUND
STREAM BED
PONDS AND MINE BUILDINGS
HIGH WALL
Figure V —35. Clauif ica tion Map of Mine 29— Huiky Briquetting Inc.
Lthigh Strip Mint, Dickinion, North Dakota
-------
103
The two revegetated areas in this section are too small to identify;
however, the bright spoils behind the high wall showed up distinctly.
Spoils were being graded so close to the high wall that the spatial
resolution of the image was inadequate to define the high wall. Table
V-14 summarizes the classification of land areas.
Table V-14
Areal Distribution of Land Use
Mine 29
Land Use
Northern section
Ponds
Overgrown spoils
Subtotal
Western section
Ponds
Bright spoils
Pixels
2
33
35
5
26
Area
Hectares
1
16
17
2
13
Acres
2
40
42
6
31
Subtotal 31 15 37
Southeastern section
Pond and overgrown spoils 35 17 42
Overgrown spoils and
revegetated spoils 110 53 132
Bright spoils 31 15 37
Subtotal 176 85 211
TOTAL AREA 242 118 290
-------
104
MINE 30 - ARROWHEAD COAL CO., ADAMS MINE
Production in 1972 from this small mine was less than 0.11 million
metric tons (0.12 million tons). Production estimates for 1975 and 1980
were not available.
Overall mine dimensions on an aerial photograph were 4.6 km (2.8 mi)
by 340 m (1,100 ft). The mine was too small to locate on the ERTS image
and therefore was not classified.
-------
105
w. 6H
^v
^SHERIDAN
\
%
\
"\\
^
°\B9
i
V
BUFFALO*^
\ \
\ ^^
\
?
|
1
xx
"" WYOMING j£
I * O
II I*
41 .<
.. -N- IQ
i1
^
IS
-!S
/ i
7H /X j
1 ^-^
GILLETrEaJ ^-i I
^-\ : 1^9°
— "^ \ 1
0 i j
i i
o> 1 NEWCASTLE* j
10 I
• ' •
J B
ii 8 !
N
\
,eo
lui
Figure V — 1. Locations of Wyoming Coal Mines
-------
106
CANADA
USA
SHERIDAN*
Figure V —12. Locations of Montana Coal Mines
-------
107
i NORTH DAKOTA
24,
CANADA
'•COLUMBUS
27
H H
JS
1
CO
00
j
"S
-N-
M/NOT
25
SOUTH DAKOTA
GP0857-646
Fiqure V — 22. Locations of North Dakota Coal Mines
------- |