WATER POLLUTION CONTROL RESEARCH SERIES 16020 FQT 12/70
Fluid Product Pipeline Leak
election From Airborne Platforms
U.S. ENVIRONMENTAL PROTECTION AGENCY
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes the
results and progress in the control and abatement of pollution
in our Nation's waters. They provide a central source of
information on the research, development and demonstration
activities in the Environmental Protection Agency, through
inhouse research and grants and contracts with Federal,
State, and local agencies, research institutions, and
industrial organizations.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed to the Chief, Publications Branch
(Water), Research Information Division, R&M, Environmental
Protection Agency, Washington, D.C. 20460.
-------
FLUID PRODUCT PIPELINE LEAK DETECTION
FROM AIRBORNE PLATFORMS
by
Resources Technology Corporation
1275 Space Park Dri ve
Houston, Texas 77058
for the
ENVIRONMENTAL PROTECTION AGENCY
Program #16020 FQT
December, 1970
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Prico $1.00
-------
EPA Review Notice
This report has been reviewed by the Environmental Protection
Agency and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the Environmental Protection Agency nor does
mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
ABSTRACT
A computer simulation program in conjunction with an instru-
ment systems analysis program lead to the conclusion that
microwave radiometry working in concert with thermal infra-
red systems could detect petroleum product pipeline leaks.
The utilization of these systems from an airborne platform
would result in a low false alarm rate and a high probabil
ity of leak detection. A demonstration experiment was de-
signed to test the simulation program. This demonstration
was carried out in west Texas along three different pipeline
sections with eighteen individual leak circumstances.
Three ground crews, one for each pipeline section, marked the
sections of pipeline and each individual leak as well as gath-
ering ground data consisting of oil saturation, soil moisture,
and thermal temperatures at both the surface and subsurface
to ten inches.
All data, airborne and ground, was reduced, correlated and
analyzed to demonstrate remote sensor capabilities. It was
found that the apparent microwave (13.7 GHz) temperature of a
leak increases significantly compared to surface material con-
taining no oil. Also, a soil containing oil caused a decrease
in polarization contrast. Thermal infrared showed a warm area
surrounded by a cool halo. When these circumstances all occur
red together a leak was identified, proving the correctness of
the original computer simulations.
This report was submitted in fulfillment of project 16020 FQT
under the sponsorship of the Water Quality Office, Environmen-
tal Protection Agency.
-------
CONTENTS
S e c t i on Page
I Summary and Introduction 1
II Conclusions and Recommendations 7
III Program Hi story and Development 9
IV Field Measurements Program 31
V Data Correlation and Reduction 47
VI Data Analysis and Interpretation 51
VII Problem Areas and Solutions 67
VIII Appendix A - Radiometer Data 75
Reduction Procedures
-------
FIGURES
No. Page
1 Microwave Temperature Variations 14
as a Function of Changing Moisture
Content, 3.0 GHz
2 Microwave Temperature Variations 17
as a Function of Changing Moisture
and Oil Content, 0.3 GHz
3 Microwave Temperature Variations 18
as a Function of Changing Moisture
and Oil Content, 3.0 GHz
4 Microwave Temperature Variations 19
as a Function of Changing Moisture
and Oi1 Content, 10 GHz
5 Microwave Temperature Variations 20
as a Function of Changing Moisture
and Oil Content, 0.3 GHz
6 Microwave Temperature Variations 21
as a Function of Changing Moisture
and Oil Content, 3.0 GHz
7 Microwave Temperature Variations 22
as a Function of Changing Moisture
and Oil Content, 10 GHz
8 Polarization Contrast, ATp, for Soil 23
Containing Oil and Moisture at 0.3
GHz
9 Polarization Contrast, ATp, for Soil 25
Containing Oil and Moisture at 3.0
GHz
10 Polarization Contrast, ATp, for Soil 26
Containing Oil and Moisture at 10 GHz
11 Computer Simulation of Pipeline Leak 28
Detecti on
12 End of Flight Line Markers 38
13 Aluminum Panel Electromagnetic 38
Markers
-------
FIGURES (continued)
N_p_. Page
14 Leak Site Identification Markers 39
15 Schematic Diagram of a Test Site 39
16 Site 1, West of Crane, Crane County, 43
Texas
17 Site 2, Southeast of Wink, Winkler 44
County, Texas
18 Site 3, East of Wink, Winkler 45
County, Texas
19 Multisensor Data Correlation, 53
Site 1
20 Multisensor Data Correlation, 58
Site 2
21 Multisensor Data Correlation, 66
Site 3
22 Profile Offset Due to Wind 68
23 Profile Ground Track, Site 1 70
24 Profile Ground Track, Site 2 71
25 Profile Ground Track, Site 3 72
26 Schematic of Incredata Tape 76
27 Definitions of Roll, Pitch and Yaw 80
28 Strip Chart Recording of Serial 85
Digital and Audio Channels
29 Strip Chart Recording Used for 87
Timing
VI i
-------
TABLES
No. Page
1 Dielectric Constants of Soil and 16
Water
Computed Radiometn'c Temperature 27
Differentials
Surface Measurements, Site No. 1 54
(Night)
Surface Measurements, Site No. 1 55
(Day)
Surface Measurements, Site No. 2 61
(Night)
Surface Measurements, Site No. 2 62
(Day)
7 Incredata Digital Tape Print-Out 82
8 Incredata Digital Tape Print-Out 83
Vlll
-------
SECTION I
FLUID PRODUCT PIPELINE LEAK DETECTION FROM
AIRBORNE PLATFORMS
SUMMARY AND INTRODUCTION
Initial research indicates that oil spilled from leaks in land
traversing pipelines is much more common than ordinarily an-
ticipated. In fact the quantities of oil that infiltrate con-
tinental soils from leaks in trunk and gathering lines may
equal that spilled on the oceans. In general it is the lack
of visible mobility that has been a deterent to identifying
pipeline leaks as a form of continental pollution. However,
these spills represent a source of pollution to all continental
water supplies including surface run-off, acquifers and general
ground water systems. In the process of examining various pipe-
line sections to select proper sites for the conduct of this
program, oil leaks and hydrocarbon surface trails were found
which flowed into local streams, saturated ground next to city
water supply wells and potentially contaminated near surface
gravels normally utilized for water supply. In addition large
areas contaminated by hydrocarbon saturation caused the demise
of local vegetation and for all practical purposes made local
areas unfit for any use. The areal extent of land affected by
pipeline leaks has not been determined. However, the ubiquity
of pipeline leaks indicate the numbers would be surprisingly
1arge.
However, the objective of this program is not to analyze pipe-
line leaks as a contaminate, but to test airborne sensors which
may detect oil leaks early in their history. As a precursor to
the reported survey program a systems analysis was performed
to determine which remote sensors would have the highest poten-
tial of detecting pipeline leaks from an airborne platform. It
was determined that a combination of microwave radiometry sup-
ported by infrared thermal imagery offered the greatest poten-
tial. Computer models of pipeline leak circumstances were gen-
erated to further test the feasibility of these systems prior
to conduct of the actual survey. Both the systems analysis and
the computer models indicated potential success and actual field
testing was initiated. The field tests were posivive for air-
borne detection of leaks in the environment tested. However,
problem areas in the utilization of such systems were identified
which require further study before operational systems can be
def i ned.
-------
Systems Analysis
The greatest potential for pipeline leak detection exists
using electromagnetic sensors. Unfortunately there are no
al1-spectrum detection devices. Therefore, the electromag-
netic spectrum was separated into spectral regions based on
instrument types. The following criteria were used to arrive
at a final selection:
The system should have day/night capabilities.
Detection should not be limited by weather con-
ditions.
The system should "penetrate" the surface to de-
tect the presence of subsurface oil.
The effects of rough ground should be minimized.
Detection through foliage is highly desirable.
The system should not interfere with local commu-
nications.
The system must be light weight; capable of instal
lation on a light aircraft.
t Data reduction and analysis should be simple, in-
expensive and accomplished rapidly.
These criteria were most nearly fulfilled by passive microwave
radiometers and thermal infrared imaging systems.
The Simulated Models
The basic premise is that the addition of oil to a soil would
cause a change in the dielectric constant of the soil. The
change in dielectric constant in turn results in a change in
electromagnetic emission characteristics. If these changes
are large enough then oil leaks should be detectable.
The calculations of the effects of adding oil to a soil is not
a simple task. First, not only are soil and oil involved but
the moisture content (water) in the soil must also be consid-
ered. This means that dielectric mixing formulas involving
three dielectrics must be considered.
The approach first developed a data set for soil containing
water only. This was followed by a data set involving only
-------
soil and oil. These were then combined to determine a data
set for oi1/water/soi1 systems.
These models showed that in all circumstances, using various
percentages of each constituent, that the addition of oil
caused perturbations that should be detectable. The emissiv-
ity is increased such that the detected apparent temperature
of the leak area is increased. Also, the differential emissiv-
ity for the horizontal and vertical polarized energy is de-
creased, resulting in a smaller differential between apparent
temperatures detected for the horizontal and vertical polarized
energy components.
Therefore, the prediction models indicate two mechanisms for
identifying petroleum product pipeline leaks: An increase in
the detected apparent temperature contrast; and a reduction in
polarization contrast. When both of these are combined in the
analysis phase a high probability of detection with a low false
alarm rate results.
The Field Survey
Three sections of pipeline in west Texas were selected for the
demonstration flights. Each section was approximately two miles
long and had four to seven identifiable leaks at each site.
Ground survey teams (3) marked the beginning and end of each of
the three pipeline sections such that positive identification
could be made by the pilot. Also each field team gathered
ground data which include ground temperature, ground moisture
content, hydrocarbon saturation, and general wind speed and
direction. These data were used in the analysis phase.
Two aircraft were utilized to obtain data, one for high alti-
tudes (2,000' above ground) and one low altitude (5001 above
ground). The instrumentation in the low altitude aircraft
proved faulty so that only the high altitude platform was used,
and the altitude was reduced to 1,000 feet above ground. No
further problems were encountered in the field program and the
following simulaneous data were obtained for each of the three
si tes :
Color Photography - Ektrachrome 2448
« RS-14 Infrared Imagery 9-14 microns
Black and White Photography - Plus X 2402
Black and White Infrared Photography - 2424
-------
Color Infrared Photography - 8443
Microwave Radiometric Profiles - 13.7 GHz
Each site was flown during the day (all sensors) and during
the night (infrared and microwave only).
There were no problems, other than the failure of the low
altitude aircraft, encountered in the field operation and all
data gathered were valid and ready for data processing and
analysi s.
Data Processing
All photographically recorded data were processed using stan-
dard development and printing techniques. Correlation between
various photographic and imagery sensors was easily accom-
plished by matching image characters such as roads, houses,
oil wells and vegetation groupings. Scale factors were deter-
mined by two techniques: measuring the film dimensions of
known ground targets, whose actual dimensions were known; and
using camera focal lengths and aircraft altitude formulas.
These scale determination functions were well within error
budgets and correspondance was excellent.
The correlation between microwave data and imagery data was
accomplished using two systems. The prime technique employed
aluminum foil panels on the ground. These caused cold anoma-
lies on both IR imagery and microwave profiles. These panels
in turn were identified on the photography and start and stop
points were identified for all sensors simultaneously. The
second technique was based on time and internally generated
400 Hz signals. At the beginning and end of each data run 400
Hz signals were impressed on the magnetic recording tapes. In
this manner all sensors could be simultaneously related to
start and stop points. Since the microwave radiometer was
viewing aft at an angle of 45°, it was necessary to slide the
profiles forward an amount equivalent to the altitude of the
aircraft. Also, because the microwave system produces only a
profile, it was necessary to compensate for roll, pitch and
yaw of the aircraft. These parameters were recorded inflight
by an Incredata system and geometric corrections were made such
that a profile was indeed identified as to actual ground inter-
cept position. All microwave data were corrected by computer
programs generated for a CDC-6600 computer system and coded in
FORTRAN IV. The computer output was played into a Cal-Comp
plotter and the corrected profiles plotted for use in data
analysis.
-------
Data Analysis
All the obtained data were mounted on data boards (one for
each site) such that point-to-point correspondence of data
was established. All data were then examined to determine
the response of each sensor.
The basic premise derived from the computer modeling program
was found to be correct. That is, the apparent microwave
temperature did increase and the polarization contrast did de-
crease when an oil leak was encountered.
The infrared imagery was also valuable in detecting leak
zones. The centers of active leak areas appeared hot, sur-
rounded by cool anomalies associated with hydrocarbon trails.
It appears that fresh oil in a soil appears warm and dead oil
(old) appears cool. However, there is a "greenhouse" effect
which causes an increase in subsurface water content beneath
the spill. This may have an effect on detection capabilities
which increase the probability of leak identification. Cer-
tainly it is something that should be considered.
All leaks were visible and detectable with color photography
except two leaks at site 2. These were identified as poten-
tial leaks by microwave radiometry and the color photographs
were re-examined for surface staining indicative of a leak.
Staining was found at one microwave anomaly and very slight
traces, which may be leak staining at the other. Neither of
these areas would have been called "leaks" without microwave
identification. Field sampling of these potential leaks has
not been accomplished and positive identification as leaks has
some uncertainty. However, the slight surface staining strong-
ly suggests a leak circumstance.
Black and white photography and black and white photographic
infrared and color infrared images were very erratic in being
useful for leak detection. Some of the larger leaks were
quite obvious on the images, but some were marginal and many
of the smaller leak "scars" were undetectable with these sys-
tems .
Identification and detection of oil leak zones using 13.7 GHz
microwave profiles along the surveyed pipeline sections were
positive when both the apparent temperature raise and polari-
zation decrease were used together. Either one separately
were of little value and had high false alarm rates. When used
together the false alarm rate was nearly zero for those leak
zones encountered by the horizontal and vertical beams. How-
ever, not all leaks were encountered by both beams and a total
analysis was not possible. This was due to the inability to
fly exactly the same flight line twice.
-------
Problem Areas
Three major problem areas developed in performing analysis of
the obtained data. These were errors in pilot judgement, air-
craft dynamics and polarization differential.
Pilot judgement was the dominant problem. When using profil-
ing instruments that require flying the exact same track sev-
eral times, proper visual alignment is critical. This is es-
pecially true if the instruments view the ground from some
angle other than nadir. Winds blowing at angles to the flight
path force the pilot to "crab" the aircraft into the wind in
order to fly a straight line. This means that instruments
viewing at some angle are not tracking along the path desired.
Aircraft dynamics such as pitch, roll, yaw and velocity changes
cause deviations from the desired ground track. These devia-
tions can be corrected for by recording the dynamic deviation
and geometrically determining where the true ground intercept
occurred. However, this only tells if the proper target was
encountered or not. If it was missed, after the fact knowl-
edge does not make the information recoverable.
The computer simulations program, which initiated this project,
predicted a decrease in polarization contrast (Ty - T^). The
actual polarization differential was very nearly zero, that is
TV - T^|, which was not predicted. Obviously the model is not
totally adequate or the acquired data included circumstances
which were not totally due to leak characteristics.
The solutions to these problems are relatively simple, but re-
quire instrumentation that is difficult to obtain. Both the
pilot judgement and aircraft dynamics problems can be solved
by using multibeam or imaging systems. This would alleviate
the problem of very accurate flying. The polarization problem
can be solved by using microwave systems which have simultan-
eous dual polarization data acquisition capabilities. These
systems are presently available and will be used on future pro-
jects .
-------
SECTION II
CONCLUSIONS AND RECOMMENDATIONS
The total project was successful in proving that microwave
and infrared systems are indeed valuable for detecting the
delineating leaks in petroleum product trunk lines, espe-
cially under the environmental circumstances encountered.
Basically the computer models were correct and the electro-
magnetic characteristics of soils are changed by oil to the
extent that existing instrumentation can be used to detect,
delineate and map leaks. The "all-weather" day/night capa-
bilities of microwave systems has already been proven. There-
fore, the potential of using microwave systems as a leak de-
tection system in areas of inclimate weather is real,
It is highly recommended that the program, as started, be con-
tinued to include other environmental type areas. The ability
to operate in forested areas, swamps and agricultural areas
has not been established. Experiments to test these environ-
ment types should be conducted to bring this entire demonstra-
tion program to a proper and logical conclusion.
It is also recommended that the suggested solutions to the
problem areas as outlined be incorporated in the next set of
experiments.
An expansion of the program to test water quality in areas
where leaks have occurred close to local or regional water
supplies is also in order. The fact that pipeline leaks are
much more prevalent than normally thought, leads to the con-
clusion that water quality may be affected much more than
present considerations anticipate.
-------
SECTION III
PROGRAM HISTORY AND DEVELOPMENT
The need to "clean up America" has been brought forceably to
the attention of the American public and government, simply
because pollution of the air and natural waters became so
apparent that it could no longer be ignored. The American
public is looking to the State and Federal Governments to
protect them from further pollution and to reduce pollution
that presently exists.
This report is concerned with a type of pollution that is not
generally considered because it is invisible in most cases and
does not irritate the eyes or affect the respiratory system.
This is the pollution of ground water and aquifers by leaks in
oil product pipelines. Oil from product pipelines migrating
into soil is not only a potential contaminant of ground water
supplies, but once subjected to saturation or even mild con-
tamination by oil, an area loses its agricultural potential
and all vegetation in the area dies very rapidly. Further-
more, once soil is contaminated by oil, it is almost impossible
to flush it out.
Thousands of miles of product pipelines crisscross the United
States carrying almost every conceivable kind of bulk fluid.
The total line-miles crossing Federal lands represent a major
part of all these pipelines. Aircraft are used to examine
pipeline right-of-ways on a routine basis. However, examina-
tion is by visual observation, generally only the pilot. As
a result leaks which are so gross that they saturate the sur-
rounding areas to the surface are the only ones detected.
As pipelines age, they become more and more subject to leakage.
Many major lines are decades old making continuous surveil-
lance a must, if only from a hazard point of view.
RESOURCES TECHNOLOGY CORPORATION completed a study to deter-
mine the feasibility of using modern electromagnetic sensors
for detecting subsurface pipeline leaks. The study indicates
that relatively simple instrumentation may be used on light
aircraft to detect pipeline leaks that are not readily apparent
Airborne electromagnetic radiation sensors were examined in de-
tail to determine which sensors had the highest probability of
achieving the set goals.
The natural development of instruments for remote sensing has
been an extension of human functions, primarily to increase
detectabi1ity. Optics aid the visible senses through magnifi-
cation of objects, and cameras were developed to permanently
-------
record those objects that could be visibly detected. "Seeing"
capabilities have been extended by orders of magnitude by using
many different devices. The spectra of human sight has even
been extended into the infrared and ultraviolet by clever de-
tection and display systems.
During and since World War II, extensive investigations have
been undertaken to extend the ability to "see", detect, and
discriminate objects in the microwave portion of the electro-
magnetic spectrum. The primary objective was to develop sys-
tems which could "see" for long distances, day and night, and
penetrate heavy weather conditions. The outcome of these ef-
forts was the development of RADAR.
In general, there are two modes of operation for these new
geophysical sensors, active and passive. In the active mode
electromagnetic energy is transmitted from a source and re-
flected energy is received, registered, and analyzed. In the
passive mode naturally emitted electromagnetic energy is re-
ceived, detected, and displayed for analysis.
It has been firmly established by a number of studies spon-
sored by the Federal Government and private industry, that
energy in the microwave region of the electromagnetic spectrum
(from 109 CPS to 10'' CPS) has some capability of penetrating
dense dielectric materials such as soil, rock and oil. Also
these systems operate efficiently during day or night and dur-
ing inclimate weather conditions. Since operations are not
dependent on time of day or weather conditions, airborne oper-
ations of passive microwave systems are extremely flexible and
useful.
Existing empirical data on soils and earth surface materials,
combined with available data on petroleum products were util-
ized to determine the practical capabilities of detecting leaks
in petroleum product pipelines using passive microwave radi-
ometers and known data analysis techniques. Initial models
and simulations established that microwave radiometry could be
used from an airborne platform to detect pipeline leaks, since
it appears that the radiometric anomalies for oil added to
various soils are large.
A field measurements program was designed and carried out to
test the developed models. This report covers both aspects,
modeling and the obtained field data.
Statement of the Problem
Leakage of underground pipes which transport refinery products
or crude oil needs to be detected rapidly if breakage occurs,
10
-------
not only to protect the owner of the pipeline and the public
but to prevent long term contamination of ground water sup-
plies. Testing the application of remote sensing methods to
the solution of the problem appears appropriate.
Electromagnetic Sensors
Electromagnetic radiation is energy produced by the oscilla-
tions of charged atomic and molecular particles. It is a con-
tinuum function in that particle vibration occurs at all possi-
ble vibrational frequencies. This radiation is characterized
chiefly by frequency (wavelength), intensity, polarization and
phase. Radiation can only react or change when photons inter-
act with matter. Electromagnetic energy in free space does not
interact with electromagnetic energy from the same or different
sources. This lack of interaction in free space is the essen-
tial ingredient utilized by the communications industry and
allows electromagnetic sensor systems such as infrared, photog-
raphy, radar and microwave radiometry, to be used as remote sen-
sor sys terns .
Unfortunately, there are no all-spectrum detection devices or
mechanisms, so the electromagnetic spectrum has been separated
into various spectral regions based primarily on instrument
type. Because of this, the requirements for a specific problem
area, such as the detection of pipeline leaks, must be analyzed
to determine which instrumentation is most suitable. This anal-
ysis for pipeline leak detection was performed using the follow-
ing criteria:
The systems utilized should have day/night capa-
bilities with little change in signal character-
istics. The detection of pipeline leaks should
not be limited in time.
t Detection of pipeline leaks should not be limited
by weather conditions. Fog, light rain, cloud
conditions or snow should not prevent detection
of leak areas.
The selected system should have penetration capa-
bilities or receive signals from sufficient depths
that primary and secondary subsurface anomalies
can be sensed.
Surface features such as rough ground should have
a minimum effect on detection capabilities.
Detection through foliage is highly desirable.
11
-------
The system should be passive so that interfer-
ences with local FM, TV, and other communica-
tions systems will not present problems. Also
frequency of operation should be separated from
weather and aircraft radar operating frequencies,
and from local communications systems, to avoid
interference with sensor response.
The system should be light in weight and require
low power so that it can be easily installed and
carried on a small, economical airborne platform.
Data reduction should be easy, relatively inex-
pensive and be accomplished in near real time.
These requirements are most nearly fulfilled by passive micro-
wave radiometers and thermal infrared systems. Having identi-
fied the sensor types, further investigations using models and
empirical data to determine the best system are warranted.
Microwave Applications Using Computer Models
The dielectric constant of soil is the single most important
function in microwave sensing and should be understood when
analyzing radiometric data. The general plan is to use soil
dielectric properties to generate models by computer analysis,
then combine the soil with water and oil using weighted aver-
ages and dielectric mixture theory for the included components.
This leads to a realistic model which defines the expected re-
sponse of microwave systems.
Emissivity, E, is a function of the complex dielectric constant,
frequency and polarization. Since the frequency of any system
used is set a^ priori only the complex dielectric constant and
the effects of polarization need to be examined in detail.
Dry soils of different types have only slightly different com-
plex dielectric constants at the same frequency. However, large
changes in the dielectric constant of natural soils are obser-
vable by changing the interstitial fluids, such as increasing
or decreasing the percentage of water saturation or changing
the salinity of the fluids. This fact has been proven by many
investigators and extensively studied by Lane, Saxton, von
Hippie, and Kennedy. The models and data presented in this
report are derived from, and based on measured dielectric prop-
erties of soil.
12
-------
Computer Simulation of Effects of Oil and Water
The basic problem of detecting oil leaks in pipelines was ap-
proached using the obvious fact that oil would seep out of the
pipeline and enter the surrounding soil and contaminate it to
varying degrees. Interstitial water within the soil may be
partially or totally replaced. This leads to an examination
of the effects of adding various quantities of oil to dry soils
and comparing these data with soils containing water. Then
make a determination of the effects of oil/water mixtures on
soils. Other variables are the frequency of operation, polari-
zation effects and the angle of observation. In order to ex-
amine the effects of all these variables, families of curves
were used along with the basic coordinate systems established
for the presentation of radiometric temperature data. The
system utilizes the radiometric brightness temperature as the
ordinate and the observation angle as the abscissa for a given
frequency. In this manner the behavior of different materials
under the same conditions can readily be compared.
For example, Figure 1 shows the theoretical variation of micro-
wave temperature obtained from a sandy soil as the interstitial
water content is varied. An examination of these curves shows
that two specific variations occur. As the water content in-
creases, the microwave radiometric temperature decreases and
the polarization contrast, (ATp = Ty - TH), becomes larger.
For instance the microwave temperatures at an observation angle
of 45° and 5% water have values of 278°K for vertical polariza-
tion (Ty) and 221°K for horizontal polarization (TH). The dif-
ferential, (AT = TV - TH), is 57°K. If the water content is
increased to 10%, the vertical and horizontal temperatures de-
crease to 250°K and 181°K respectively, or a AT of 69°K. Using
both these phenomena, the moisture content of subsurface soils
can be predicted with a high degree of accuracyO).
Since the mechanisms of changing microwave temperatures with
changes in moisture content are well known, what are the con-
sequences if all or part of the interstitial water is replaced
with oil?
Radiation in the microwave region is expressed as the radiomet-
ric brightness temperature, which is determined by:
(TB)P = (1 - |RP 2)T
g
Microwave Radiometric Sensing of Soil Moisture Content;
International Union of Geodesy and Geophysics, 14 General
Assembly. International Association of Scientific Hydrology
Division, Bern Switzerland September 1967.
13
-------
;=>
EH
a
EH
GO
01
W
o
o
M
a
300
275
250
225
200 -
175
150
125
100
Frequency
Surface Temp.
Vertical Pol.
Horizontal Pol.
10
20 30 40
OBSERVATION ANGLE, 6 (DEGREES)
50
60
Figure 1
Microwave Temperature Variations as a Function of Changing
Moisture Content.
-------
where Tg is the ground thermometri c temperature, subscript P
stands for either the horizontal or vertical polarization and
Rp is the complex Fresnel coefficient for horizontal or verti
cal polarization. The equations for the Fresnel reflection
coefficients are:
cos 8 - (e - sin 20) 1/2
R
H
cos 6 + (e - sin 26) l|^
and
e cos 6 - (e - sin 20) l
e cos 0 + (e - sin 20) l/z
In (2), e is the complex relative dielectric constant and e is
the nadir angle, i.e., the angle from the normal to the surface
or the radiation observation angle.
The imaginary components of the complex dielectric constant, e,
for oils and other petroleum products is less by several orders
of magnitudes than those for soil and water. A computer com-
parison in brightness temperatures, and polarization differen-
tials between soil with oil and soil without oil, will suffice
to show whether obvious local anomalies result from the pres-
ence of petroleum products.
For a sandy soil, at a representative ground temperature (25°C)
at three selected microwave frequencies (0.3, 3, and 10 GHz),
the horizontal and vertical brightness temperatures have been
computed as a function of the viewing angle 0. This was per-
formed for an array of combinations of soil, water and oil.
Initial data for a representative oil selected from 17 petrol-
eum oils is e1 = 2.09, and e" = .0014. e1 and e" are the real
and imaginary components of the complex dielectric constant,
with the convention that
e = e1 - e
Table 1 presents the measurements of soil and water as deter-
mined vy von Hippie (Reference (2)).
(2) Dielectric Materials and Applications, A. R. von Hippie
The M.I.T. Press, Cambridge Massachusetts 1966 PP 314-315.
15
-------
Dielectric Constants of Soil and Water
frequency (GHz) * -3 3 10
day sandy soi 1
water
sandy soil
3.88% H20
16.8% H20
e '
tan 6
e1
tan 6
e1
tan 6
e '
tan 6
2.55
.01
77.5
.016
4.50
.03
20.
.03
2.55
.0062
76.7
.157
4.40
.046
20.
.13
2.53
.0036
55.
.54
3.60
.12
13.
.29
TABLE 1
The "loss tangent", tan <5, is an alternative way of tabulating
= e
tan 6
In order to employ equations (1) and (2) to evaluate bright-
ness temperatures of sandy soil mixed with water and oil a
weighted average based on oil/water content was used.
To establish the weighted averages, the horizontal and verti-
cal microwave radiometric temperatures for soil with inter-
stitial oil were computed as functions of frequency and obser-
vation angle. These data are shown in Figures 2-8 for 0.3
GHz, 3.0 GHz, and 10.0 GHz respectively. It is obvious that
the horizontal and vertical microwave temperatures change
drastically when interstitial water is replaced by oil in a
soil. Another interesting point is that soil with oil main-
tains very nearly the same microwave temperatures, even though
the frequency changes by an order of magnitude.
16
-------
W
oc
CC
W
(X,
w
CO
w
12
O
M
CC
CQ
W
<
O
CC
o
300
275
250
225
200
175
150
125
100
~ 5% Wtr. & 5% Oil
Frequency
Surface Temperature
Vertical polarization
Horizontal polarization
I t
0.3 GH
25°C
I
I
10
20 30 40
OBSERVATION ANGLE, 6 (DEGREES)
50
60
Figure 2 Microwave Temperature Variations as a Function of Chanqina
Moisture and Oil Content
-------
00
W
(C
w
CO
CO
w
X
0
M
§
w
o
cc
o
M
2:
300
275
250
225
200
175
150
125
100
[_ 5% Wtr. & 5% Oil
Frequency
Surface Temperature
Vertical polarization -
Horizontal polarization
3. GH
25°C
I
1
I
I
10
20 30 40
OBSERVATION ANGLE, 6 (DEGREES)
50
60
Figure 3 Microwave Temperature Variations as a Function of Changing
Moisture and Oil Content
-------
300
275
250
225
200
175
150
125
-100
5% Wtr. & 5% Oil
15% Wtr. & 5% Oi
Frequency
Surface Temperature
Vertical polarization -
Horizontal polarization
I I
10. GHr
25°C '
10
20
30
50
60
OBSERVATION ANGLE, 8 (DEGREES)
Figure 4 Microwave Temperature Variations as a Function of Changing
Moisture and Oil Content
-------
ro
o
300
275
250
225
200
175
5% Wtr. & 10%
150
125
100
Frequency
Surface Temperature
Vertical polarization -
Horizontal polarization
I I
0.3 GH
25 °C 2
10
20
30
1(0
50
60
OBSERVATION ANGLE, 6 (DEGREES)
Figure 5 Microwave Temperature Variations as a Function of Changing
Moisture and Oil Content
-------
ro
w
a:
cr:
w
O,
s:
u
EH
co
CO
u
1
o
M
K
CO
w
o
cc
o
300
275
250
225
200
175
150
125
100
Frequency
Surface Temperature
Vertical polarization
Horizontal polarization
3. GH
25°C Z
I
I
I
I
10
20 30 40
OBSERVATION ANGLE, 9 (DEGREES)
50
60
Figure 6 Microwave Temperature Variations as a Function of
Changing Moisture and Oil Content
-------
ro
W
cc
CO
CO
W
2:
O
M
K
ffl
o
o
M
X
300
275
250
5 225
PH
200
175
150
125
100
10% Wtr. & 10% Oil
5% Wtr. & 10% Oil
Frequency
Surface Temperature
Vertical polarization
Horizontal polarization
10. GHr
25°C l
\
\
I
10
I
20 30 10
OBSERVATION ANGLE, 6 (DEGREES)
50
60
Figure 7 Microwave Temperature Variations as a Function of
Changing Moisture and oil Content
-------
ro
co
W
K
w
CO
CO
w
EH
K
O
H
«
03
W
O
ct;
o
M
160
140
120
100
80
60
40
20
0
KEY
ATp =
10% Wtr. & 5% Oil
5% Wtr. & 5% Oil
5% Wtr. & 10% Oil
10
20 30 40
OBSERVATION ANGLE, 8 (DEGREES)
50
60
Figure 8 Polarization Contrast, ATp, for Soil Containing Oil
and Moisture at 0.3 GH,
-------
Effects on Brightness Temperature and Polarization
Let us examine the effects of combining oil and water at 0.3
GHz (Figure 8). If we assume the sensor is pointed at 45° and
the soil contains 5% oil and 5% water, the brightness tempera-
ture would be 286°K for the vertically -polarized energy and
237°K for the horizontally polarized energy, or a ATp of 74°K.
These numbers are consistent with both theory and experiment
and show there are actually two techniques of data analysis
that can be used to define areas containing oil as an inter-
stitial fluid. First, if oil is added to a soil, the micro-
wave temperature increased sharply (in the above example from
254° to 286°K). Secondly, the polarization contrast is greatly
reduced (in the above example from 74°K to 49°K).
Either one of the above functions may be used to detect oil
leaks, but the possibilities of false alarms are greatly de-
creased if both functions are simultaneously observed. That
is if a temperature increase is observed in conjunction with a
decrease in polarization contrast, then it is almost certain
that a dielectric with properties very similar to oil has been
added to the soi1 . .
Effects of Frequency
An examination of Figures 8, 9, and 10 show that microwave tem-
peratures increase and polarization contrast decreases by the
addition of oil are not restricted in frequency or by varying
the oil/water content of the soil. In all cases the addition
of oil caused an increase in microwave brightness temperature
and a decrease in polarization contrast. It is interesting to
note that as the specified frequency increases, the observed
microwave temperatures become less variable and warmer. That
is, the envelope of variability narrows even though the per-
centage of contained oil varies drastically. The same is true
for polarization contrast. That is, the envelope of polariza-
tion contrast variability narrows considerably by increasing
the (observation) frequency.
Signal and Resolution Considerations
The detectabi1ity of local warm spots will vary somewhat with
the area or volume invaded by the oil. The invaded area must
occupy enough of the radiometer beam ground intercept to regis-
ter a significant change in radiometric temperature. For in-
stance, if the invaded area is only one percent of the total
area the radiometer receives radiation from, the anomaly will
probably be hidden in fluctuations arising from natural hetero-
geneity.
24
-------
en
GO
t/o
UJ
Qi
CO
=£
IS
O
o
160
140
120 _
< 100
80
60 _
20
KEY
AT_ = T.. - T
10% Wtr. & 5% Oil
5% Wtr. & 5% Oil
5% Wtr. & 10% Oil
10
20
30
60
OBSERVATION ANGLE, 6 (DEGREES)
Figure 9 Polarization Contrast ATp, for Soil Containing Oil and
Moisture at 3.0 GHZ
-------
ro
W
cc
w
CL,
s
w
CO
w
PC
CQ
o
o
M
s
160
140
120 __
100 _
80 _
60 _
40 _
20 -
10% Wtr. & 5% Oil
5% Wtr. & 5% Oil
5% Wtr. & 10% Oil
10
20 30 40
OBSERVATION ANGLE, 6 (DEGREES)
60
Figure 10 Polarization Contrast, ATp, for Soil Containing Oil
and Moisture at 10 GH2
-------
To estimate the size of warm areas required for a detectable
anomaly, let the beam-filled area of a radiometer be denoted
S, comprised of area A, with brightness temperature T/\, and
area B, with brightness temperature Tg. Then the effective
temperature combination the radiometer measures, J$, is given
by:
TS = (A/S) TA + (B/S) TB
Table 2 presents some examples of microwave radiometric tem-
perature predictions. In computing Table 2, an airborne
tern at 1000 feet altitude and having a 5° beam width
sumed (a "worst case" condition).
sys-
was as-
d
(feet)
20
40
60
80
Percent Oil
5%
10%
15%
20%
Temperature Anomaly (°K)
.5
1 .9
4.3
7.6
1 .4
5.5
12.3
21 .9
2.3
9.0
20.3
36.2
4.1
12.4
27.9
49.6
Computed Radiometric Temperature Differentials
Table 2
The magnitude of anomalies required to be distinguishable
from background fluctuations is approximately 2°K. However
this will vary with location and weather.
Effects of Varying Oil and Hater Concentrations
Figure 11 is a generalized cross section of an oil leak in a
pipeline, along with the microwave radiometric profile resul'
ting from a survey. The following assumptions are made for
modeling purposes:
27
-------
300
Vertical Polarization
256
258°
250
w
tf
o
w
Q
1229
200'
Horizontal Polarization
I I i
Figure 11 Computer Simulation of Pipeline
Leak Detection
28
-------
The ground surface temperature is 25°C.
The soil moisture is a variable 5 and 10 percent
along the pipeline track.
t The subsurface dimensions of the oil intrusion
fill the beam diameter of the microwave antenna.
The operating frequency is 10 GHz.
The observation angle is 45° (To obtain a polar-
ization contrast AT = TV -
t The soil moisture and oil variability are (see
figure) :
A to B - Soil with water only 10%
B to C - Soil with water only 5%
C to D - Soil with water 5%, oil 10%
D to E - Soil with water 5%, oil 15%
E to F - Soi 1 with water 10%
This profile of horizontal and vertical microwave temperatures
is a computer simulation of a flight along a pipeline and rep-
resents a model developed by using the developed data and theory.
The flight begins at point A and proceeds to point F with changes
in water and water/oil content encountered along the flight path.
Points B, C, D, and E represent the boundaries of these changes.
The zone from A to B has a soil moisture content of 10% and is
relatively cold both in the vertical and horizontal modes of
operation. In the zone B to C, the soil moisture is decreased
to 5% with the result that microwave temperatures increase both
in the horizontal and vertical modes and a decrease in polari-
zation contrast is observed. From C to D the total fluid con-
tent is 15% (5% water and 10% oil). If this were all water,
cold microwave temperatures would be registered along with a
large increase in polarization contrast. However, the addition
of oil has the effect of increasing both the horizontal and ver-
tical temperatutes , with the horizontal component increasing
much more than the vertical. Therefore, a large decrease in
polarization contrast is realized. From D to E, the fluid con-
tent is again increased by additional oil. While the vertical
temperature is increased very little, the horizontal tempera-
ture increases a significant amount, resulting in a further de-
crease in polarization contrast. From E to F the soil contains
only water (10% and the same microwave characteristics are
29
-------
observed as from A to B, which also contains 10% moisture.
Infrared Measurements
Other than photography, probably the most advanced imaging sys
terns, in a hardware sense, are thermal IR imaging systems.
However, these systems lack some desirable features for pet-
roleum product leak detection. First thermal IR does not
penetrate to significant depths and must depend on secondary
thermal effects that migrate to the surface for detectabi1ity.
Secondly, IR is not all weather, rain, fog, snow and haze
affects the data. Thirdly, the ability to detect any phenom-
ena by IR is time dependent. That is, most successful IR data
is obtained at night or in the early morning hours when solar
reflections are absent.
The value of IR imaging in conjunction with microwave survey-
ing for leaks comes from its ability to detect the secondary
thermal effects of a buried pipeline with almost photographic
resolution. This fact is firmly established and is very appar-
ent from the obtained field data.
30
-------
SECTION IV
FIELD MEASUREMENTS PROGRAM
The field measurements program was divided into three primary
phases :
t Area reconnaissance and site selection.
Experiment design.
Airborne and ground measurements.
Each of these major phases are sub-divided into subject groups
for discussion purposes.
Area Reconnaissance and Site Selection
A number of major pipeline companies were contacted in an
effort to determine areas that could be used as test sites.
Two major pipeline companies were very cooperative and not
only furnished detailed records of petroleum leaks but also
directed a number of field engineers to support the study. Six
areas were examined as potential test areas. Two were in Brazos
and Robertson Counties about one hundred miles northwest of
Houston, Texas. These areas were very densely vegetated with
both trees and underbrush. Since a multiple canopy exists,
this probably would represent a "worst case" situation for dem-
onstration purposes. The areas were not rejected, but set as-
ide for test flights at a later date.
The other four areas examined were in west Texas about forty
miles west of Odessa, Texas. Two were in Crane County and two
in Winkler County. From these, two sections of pipeline, each
approximately two miles in length, were selected in Winkler
County and one, approximately one and one-half miles in length,
in Crane County. These areas were selected for several reasons,
but primarily because:
A variety of leaks, old and new, were present and
well documented.
The leak areas and pipeline sections had not been
disturbed to any great extent.
Some leaks are readily visible while others are
hard to detect visually.
31
-------
t Both compact and low density soils are repre-
sented .
Accessibility was relatively easy.
The flat terrain allowed easy location of the
sites from the air, especially at night.
The selected pipeline sections were relatively
straight trunk lines.
t Both small and large diameter pipelines were
represented.
Population density is very low (open range land).
Nearby airfield facilities are present.
Land owner permission to enter is easily obtained.
Terrain, Geology (surface) and Geography
Six individual leaks were identified along a one and one-half
mile section of pipeline right-of-way in Crane County. Actu-
ally there are two parallel eight inch trunk lines used pri-
marily to transport crude oil. The leaks occurred at various
times between November 1968 and March 1970.
The terrain is gently rolling sand hills with occasional inter-
mittent stream cuts. The sand dunes are recent, underlain by
quaternary alluvium which rests, unconformably, on Triassic
red beds. Together the dunes and quaternary alluvium are 30
to 100 feet thick. In general the regional soil maps indicate
simply "sand dunes" for the area. The dunes are very immature
and readily move in the direction of prevailing winds. They
are highly permeable and move on caliche beds five to thirty
feet below the surface. The thickness of soil covering the
caliche is highly dependent on dune development. The caliche
is generally flat and horizontally planar. In many places the
pipeline, which was originally buried, is now exposed due to
dune movement.
The base elevation for the region is about 2600 feet, with var-
iations due to sand dunes, rolling hills and intermittent
streams. The climate is semiarid with about fourteen inches
of annual precipitation, concentrated in late summer. In June,
when the survey was conducted, the average rainfall is less
than two inches. Low rainfall and high permeability were ex-
cellent for testing the sensors under dry, low soil moisture
condi ti ons.
32
-------
The population density is very low. Land use is confined to
open range cattle grazing (ranching) and oil production. No
farming occurs anywhere in the general area.
The other two pipelines selected are in Winkler County near
the town of Kemit. There are about fifteen identified petrol-
eum leaks with these two test sites. These occurred over a
period of time from November 1968 to April 1970.
The terrain is very flat with some gently rolling hills. The
surface_soil is a fine loamy sand known as the Springer unit.
This soil packs well and in general is relatively firm. How-
ever, when the surface crust is broken it becomes soft and
dusty. Sand dunes of the type encountered in Crane County are
rare. The Springer soil unit is a residual of Quaternary allu-
vium which is 80 to 200 feet thick and rests unconformably on
triassic red beds. Fifteen feet below the surface there is a
caliche horizon that is 10 to 20 feet thick. Soil maps of the
area are in conflict, some maps showing sand dunes, some bar-
ren of sand dunes. Area reconnaissance did not locate any
dune sand areas. Groundwater occurs unconfined in the Quater-
nary alluvium at 70 to 100 feet.
Other parameters such as rainfall, vegetation, land use and
general geography are the same as that found in Crane County.
Vegetation and Water
Vegetation is sparse and widely scattered as in most semiarid
areas. There are no trees of any kind in the area. Plants,
where present, are mostly sage brush, mesquite, shinoak, and
clumps of long bladed grass.
Water is sparse, highly mineralized, and derived mainly from
deep-water wells. Although the water table is only 60 feet at
the test sites, it occurs unconfined in Quaternary alluvium,
and is not potable because of mineralization.
Airport Faci1i ties
Two airfields of significant size are located in the general
area, one at Monahans and Midland International Airport. For
the instrumented aircraft used on this survey it was highly
desirable to use the complete facilities at Midland Inter-
national .
33
-------
Aircraft and Instrumentation Selection
Five requests for bids were sent out to various companies
known to have airborne remote sensor systems of the types de-
sired. Based on abilities to meet instrument and flight speci-
fications two (2) contractors were accepted, North American-
Rockwell, Downey, California and Remote Sensing Inc., of Houston,
Texas.
The North American-Rockwell aircraft was to fly low altitude
missions at the same time RSI was flying higher altitude mis-
sions. During the first pre-mission instrument checks the
North American-Rockwell instrumentation was found to be in-
operative and repairs could not be instigated in time to ob-
tain simultaneous airborne and ground data. Since the three
(3) ground crews and site markings were already in place it
was decided to use only the RSI instrumented aircraft at an
intermediate altitude, about 1000 feet.
The RSI aircraft and instrumentation specifications are as
fol1ows:
1 . Aircraft:
Fan Jet Falcon
Wing Span - 53.3 feet
Length - 56.25 feet
Height - 17.75 feet
Landing Weight - 26,956 Ibs.
Maximum Gross Weight - 27,300 Ibs.
Engine (2) thrust - 4,250 Ibs. each
Landing Speed - 125 mph
Cruise Speed - Mach .76
2. Instrumentation:
a. Ryan Model 703, 13.7 GHz Radiometer
Antenna Gain - 31 db
Beamwidth (3 db) - 5 degrees
34
-------
Polarization - Horizontal or Vertical (selected)
Viewing Angle - 45° aft (relative to horizontal
aircraft axis)
Overall Linearity - 0.1% max.
Operational Temperature - (-) 50 degrees C to
(+) 50 degrees C
Sensitivity- 0.3°K (one second integration
time)
Absolute Accuracy - 2 degrees K (nominal)
Baseline Stability - 2 degrees K/hr (max)
Temperature Measurement Range - 190 degrees K
to 450 degrees K
Texas Instruments RS-14 Infrared Scanner
Operational Temperature - (-) 40 degrees C to
(+) 50 degrees C
View Coverage - (± ) 40 degrees lateral
Noise Equivalent Temperature - 0.3 degrees C
Scan Rate - 200 Scans/second
Velocity to Height Ratio - 0.02 to 0.2
Radi ans/Second
Stabilization - Roll ± 8 degrees
Detector Cooling - Closed cycle, 26 degrees K
Calibration - Continuous (black body)
Camera System - Four 500 EL Hasselblad
70mm cluster
Film Format - 70mm
Field of View - 52 degrees
Lens - Zeiss Planar
Aperature - f/2.8
35
-------
Focal Length - 80mm
Shutter Speed - to 1/500 sec.
Magazine Capacity - 100 ft. (each camera)
Exposure - Four cameras simultaneously
Film types - Kodak 2402, 2424, 2448 and 8443
d. Recording System
The recording system is an individual unit
specially constructed to meet the demands of
all remote sensor systems on the aircraft.
All available instrumentation was not utilized
on this experiment. Basically, the recording
system is built around a Precision Instrument
fourteen-channel tape recorder. All necessary
timing and ancillary information is recorded
simultaneously with the data gathering func-
tion. This insures that all data can be prop-
erly oriented in time and space and data off-
sets are avoided during the reduction operation
Ground Correlation Instruments
Prior to performing the survey it was determined that the only
ground parameters that may be necessary to correct or verify
the airborne data was ground temperature, soil moisture content
and soil hydrocarbon content.
Ground temperature was obtained at the surface and ten (10)
inches below the surface using special steel shaft soil thermom-
eters. The surface temperature was obtained by horizontally in-
serting the sensor tip a fraction of an inch below the soil sur-
face so that it would not be perturbed by direct sunshine impin-
ging on the detector element. Three full minutes were allowed
to insure an equilibrium measurement. Adjacent to each surface
measurement a four inch hole was dug out and the six inch shaft
forced into the soil to obtain the temperature at ten (10)
inches below the surface-
Soil moisture was obtained at the surface and at a depth of ten
(10) inches by collecting a quantity of surface soil and placing
it in a sealed aluminum container. The ten inch grab samples
were obtained by digging a hole and taking soil samples from the
sides of the hole. These samples were then placed in individual
aluminum sample containers and labeled accordingly. Water con-
tent was determined by using well known heating and weighing
36
-------
techniques. All water content measurements are in percentage
by weight.
Hydrocarbon content was measured at several leak sites for each
of the three sections of pipeline overflown. The samples were
obtained in the same manner as the soil moisture samples. How-
ever, the hydrocarbon content was determined by Core Laboratory
Co., in Houston. The method used was standard combustion tech-
niques used by the oil industry, also in percentage by weight.
The obtained data are reported in Section VI, Data Analysis
and Interpretation.
Site Markings
The test sites were marked with a variety of materials to aid
both flight crews in identifying the test locations from the air
and to aid in the interpretation of data.
The three test sites were marked basically in the same manner.
"T" markers were used at the beginning and end of each test site
for flight crew orientation. The "T" markers were 80' x 50'
white paper towels that pointed towards the leak areas (Figure
12).
The next marker used was 2800 square feet of aluminum foil. The
aluminum foil was used to provide additional flight crew guidance
and as a distinctive marker for the microwave radiometer and
thermal IR imagery. The aluminum foil was laid out perpendic-
ular to the pipeline in 18 inch wide strips 25 feet long and
18 inches apart (Figure 13).
Each leak was marked by four 30 foot strips of white paper
towels at angles to the pipeline to form a tie mark for addi-
tional flight crew orientation (Figure 14). Strips of white
paper towels were laid out along the pipeline for flight crew
guidance. A schematic of the marking system is shown in Figure
15.
At night the right-of-way was marked by a series of lights. The
beginning of each line was marked by a flashing red strobe light
with white flashing strobe lights placed at regular intervals
along the pipeline. The end of the site was marked with three
highway flares placed in a triangle. All of the night markings
were for flight crew guidance with the exception of the aluminum
foil, which was used as a correlation point.
37
-------
leaks
flight line
= pipe line
80'
50'
end of
flight line
Figure 12 End of Flight Line Markers
end of
Towel .
\/
100 yd.
'16 rolls
flight line
leaks*-
= pipeline
>16 rolls
Figure 13 Aluminum Panel Electromagnetic Markers
38
-------
Flight line
Pipeline
Soil sample
Figure 14 Leak Site Identification Marker:
f1ight line
pipeline
Towel
H m li
Foil Panel Leaks
_\/ ,
^\ /\
Towel "X" Towel "X
Foil Panel
Towel
M rn II
Figure 15 Schematic Diagram of a Test Site
39
-------
J-0014 SCHEDULE
Tuesday. June 9, 1970
Leave Nassau Bay, 4:00 P.M.
Drive to Midland, Texas
Arrive approximately 11:30 P.M. - 12:30 A.M.
Stay at Holiday Inn Motel in Midland
Wednesday, June 10, 1970
Breakfast 6:00 A.M.
Proceed to test sites, mark route, take soil samples.
Party No. 1 - Crane Test Site
Si te 1 : 3-mile stretch of 8-inch and 10-inch pipeline
in northwest Crane County, southeast of county road
1223, halfway between Crane and Monahans (see map).
Place "T's" and foil panels at each end of line, mark
and sample six leaks.
Party No. 2 - Site Southeast of Wink
Site 2: 3-mile stretch of 6-inch pipeline 2-3 miles
southeast of Wink, between pipeline patrol stations S9
and S15. Place "T's" and foil panels at each end of
line, mark and sample five leaks (see map).
Party No. 3 - Site East of Wink
Site 3: 2-mile stretch of 8-inch pipeline northeast of
Wink, between railroad and black top road (see map).
Place "T's" and foil panels at each end of line, mark
and sample large leak area.
Thursday, June 11, 1970
8:00 A.M. - Brief airplane crews at airport on test site
locations, flight line markings, daylight flight pro-
cedures .
10:00 A.M. - Noon - ground parties proceed to test sites
take surficial and deep soil temperatures, record in
field books.
Noon - 1:00 P.M. - Daytime overflight. Afterwards, take
soil temperatures again, and record.
7:00 P.M. - Brief airplane crews at airport on night
overflight.
10:00 P.M. - 12:00 Midnight - Set up strobe lights at
ends and in middle of flight lines (3 per line).
40
-------
12:00 Midnight - 1:00 A.M. Take and record soil tempera-
tures .
Friday. June 12, 1970
1:00 - 2:00 A.M. - Night overflights
2:00 A.M. - 4:00 A.M. - Take and record soil tempera-
tures, pick up route markers, leak markers, strobe lights,
etc.
12:00 Noon - 5:00 P.M. - Return to test sites and clean
up foil, paper and other trash. Also remove survey
markers .
Friday evening or Saturday - Return to Houston
J-0014 PILOT ORIENTATION
The mission is to overfly three sites in Crane and Winkler Coun-
ties, Texas, about twenty-five to fifty miles south and west of
Odessa. Directions, distances, and orientation features are as
fol1ows.
Daytime Flights
Site 1 is in northern Crane County, about fifteen miles northwest
of the town of Crane. It is in the eastern edge of a sand dune
area. Flying along a bearing of 290°, the pilot will see a green
smoke bomb marking the start of the line. Next will be a "T"
made of white paper towels, with the stem of the "T" pointing
northwest. One hundred yards further will be a 50' x 50' alum-
inum foil marker. He will see
marking the flight line. Next
marker panel, a towel "T", and
of the line.
"X'x"
a number of white towel
will be another aluminum foil
a red smoke bomb to mark the end
Proceeding thirty miles west and north, passing over the town of
Monahans, he will come to Site 2. Flying on a bearing of 320°,
he will see another green smoke bomb followed by a towel "T" ,
next a 50' x 50' aluminum foil panel. Again, a number of white
"X's" will further delineate the flight line. Coming to the end
of the line, he will see another aluminum foil panel, a white
towel "T", and a red smoke bomb which marks the end of the line.
Site 3 extends along a bearing of 49°, and is about five miles
northeast of the end of Site 2. The pilot will see a green smoke
bomb just east of a railroad, marking the start of the line.
Next will be a towel "T", then an aluminum foil panel, then one
41
-------
large white "X". Approaching the end of the line, he will see
an aluminum foil panel, a white towel "T", and finally a red
smoke bomb. Figures 16, 17 and 18 are detailed lay-outs of
each site to be surveyed.
Nighttime Flights
Approaching Site 1 from Crane along a bearing of 290°, the
pilot will see a flashing red strobe light, which marks the
start of the line. Further along the line he will see three
blinking white lights. Red flares mark the end of the line-
Flying north and west over Monahans as during the day, the
pilot will see a flashing red strobe light marking the start
of Site 2. A line of three blinking white lights will extend
along a bearing of 320°. Red flares will mark the end of
Site 3.
Turning northeast and flying along a bearing of 49°, the
pilot will see a flashing red strobe light, marking the start
of Site 3. Three blinking white lights will further mark the
site. Red flares will mark the end of Site 3.
Flight Tests and Ground Data
The flight schedules and ground data gathering proceeded as
scheduled with no problems or significant errors. The desired
results were completely realized with exception of site #3,
where night ground data were not obtained because of vehicle
di ffi culties.
The outcome of the flights and comparisons with ground data
are contained in Section VI, Data Analysis and Interpretation.
42
-------
Si!
rtJH
*K-01f
i! 300
ip-Yds-
Gate Plant
"D"
Towel "T
/Red
/ Strobe
/ Light
azimuth
"90°
Figure 16 SITE 1 West of Crane, Crane County, Texas
-------
-pi
-pi
White
Strobe
Light
White
Strobe
Lights
azimuth
Mark
Sample
Mark
Only
Mark &
Sample
Green Smoke
Bomb
Figure 17 SITE 2 Southeast of Wink, Winkler County, Texas
-------
\
\
Green Smoke
Bomb
azimuth 49
\
\
\
\
Lights
Road
Red Smoke
Bomb
Wh i t e
Strobe
Light
100
Yds
azimuth 49'
Figure 18 SITE 3 East of Wink, Winkler County, Texas
-------
SECTION V
DATA CORRELATION AND REDUCTION
The correlation of the data from the six sensors is accom-
plished using both photogrammetrics and elapsed time. The
black and white photography, the black and white photographic
IR, the color photography and the photographic color IR are
individually mosaiced into uncontrolled strips. The four
strip mosaics are then photogrammetrically tied together using
numerous checkpoints to assure accurate alignment. The corre-
lation of the thermal infrared with the photographic sensors
is accomplished by dividing the former into high interest sec-
tions, such as a leak area, and then matching the sections to
the corresponding areas on the photograph.
The correlation of microwave profiles with thermal infrared
and photographic images was attained by a 400 Hz signal gen-
erator which drives a light source internal to the infrared
system and simultaneously puts a 400 Hz signal across all 14
tracks of the recorder system. Two of these timing dots occur
at the beginning of each data run and three at the end. This
permits an exact correlation between a particular infrared scan
line and a particular microwave radiometric temperature value.
Also, each data strip of thermal infrared imagery is coded by
a margin light source. This coding identifies mission, time
and flight number. The infrared and photographic images are
visually aligned so that all required data is correlated. In
order to doubly insure proper correlation between all systems,
large aluminum panels were placed on the ground. These panels,
at the beginning and end of each section of pipeline surveyed,
caused large cold anomalies to occur in both the infrared and
microwave radiometric night data and readily observed in day-
time photography. Although the aluminum panel system works
well for both day and night operations, it was designed pri-
marily to ensure both data correlation and target acquisition
during the night missions. Also, the measured distances be-
tween markers and the size of the markers were used to estab-
lish accurate scale factors.
The entire system worked very well with the exception that the
aluminum panels were not, in all cases, intercepted by the
microwave radiometer beam pattern.
Calculation of Scale
An important factor in photographic interpretation is accurate
determination of scale. Scale is the relationship between dis-
tances on maps, graphs or photographs and the actual ground
47
-------
distance. It is normally expressed in one of three ways; either
as a representative fraction in which the numerator is unity,
such as 1/10,000, or as a ratio 1:10,000 or in dissimilar units
1 inch equals 1 mile. The representative fraction and the ratio
one unit on the map or photo equals 10,000 units on the ground.
The scale of the photography was determined by ratioing known
distances on the ground to the distance on the photograph. The
primary instruments used to calculate scale factors for each
flight line was the measured distances between ground targets.
The accuracy of the scale determinations were verified by com-
puting the actual size of a four door sedan and two station
wagons used for ground transport. The measured and calculated
lengths of the vehicles corresponded very well. In the case
of the nighttime thermal images measured distances between roads
were used for confirmation. Again the correspondance was ex-
cellent. The daytime photographic scale factors compared to
thermal infrared night flights are:
Site 1 8-14 Infrared Thermal Image (night) - 1:7600
All Photographic Images (day) - 1:3150
Site 2 8-14 Infrared Thermal Image (night) - 1:8200
All Photographic Images (day) - 1:2800
Site 3 8-14 Infrared Thermal Image (night) - 1:7600
All Photographic Images (day) - 1:2450
These scale factors show that aircraft altitude was maintained
better during the infrared and microwave night missions than
during the photographic day missions. This was probably due
to high daytime wind velocities.
These scale factors were supplied to the computer and the micro-
wave radiometric data was graphically printed out to match the
individual scales. Therefore, all data is correlatable on a
point-to-point basis.
Radiometer Data Reduction Procedure
RESOURCES TECHNOLOGY CORPORATION designed a procedure for the
reduction and interpretation of ground brightness temperatures
as detected by an airborne radiometer. The flights over oil
pipelines resulted in automatically plotted profiles of bright-
ness temperature against distance. Corrections for aircraft
motion (pitch, roll, yaw) during flight are made and locations
of intersection of radiometer antenna axis with the ground are
accurately determined.
48
-------
Inputs to the data reduction procedure consist of digital
recordings of the aircraft flight dynamics on a one-half
inch, seven-track digital tape which is computer compatible;
a continuous magnetic field proportional to brightness tem-
peratures on a one inch, fourteen-track, analog tape; and
film strips from an airborne infrared scanner. Outputs are
profiles of brightness temperature, resulting from data which
has been numerically processed.
The equipment employed for entry into a general purpose digi-
tal computer, consists of the following instrumentations 1:
Two Ampex 1900 tape recorder/play-back machines, Clevite
chart recorder, 1400 Preston analog-to-digital converter and
a Control Data Corporation 1700 computer. Photographs and
infrared imagery are processed by photogrammetric methods as
requi red.
There are three main computer programs involved in the data
processing procedure, all coded in FORTRAN IV. One program
accepts the digital one-half inch tape containing the aircraft
history of flight motion parameters and produces a tape of
angles, sines and cosines versus times. The second program
accepts the radiometric data and produces a tape containing
aircraft locations versus temperatures. The third program
accepts the tapes produced by the first two programs and com-
putes the radiometric antenna temperature as a function of an-
tenna beam axis ground intercept. The details of the data
reduction procedures are contained in Appendix A, Section VIII
49
-------
SECTION VI
DATA ANALYSIS AND INTERPRETATION
The basic techniques used to analyze and interpret the data
were the same for all three pipeline sections. Primary to
proper analysis and interpretation was correlation of the
data in time and space.
The data for each of the three pipeline sections examined were
mounted on a board such that point-to-point correlation was
possible. Each data board has six strips of data, four obtained
from cameras, one infrared scanner image and one set of micro-
wave radiometer profiles. In each case the data are mounted
in the same sequence. Starting from the top of the data board
they are as follows :
t Color - Ektachrome MS Aerographic 2448
RS-14 Infrared Image (night) 8-14y
3 milliradian resolution
Black and White - Plux-X Aerographic 2402
Black and White - Infrared Aerographic 2424
Color - Ektachrome Infrared Aero 8443
t Microwave Profiles - Vertical and Horizontal (night)
All these data are very nearly the same scale for all three
pipeline sections examined, with the exception of the RS-14
images which are about one half the scale of the photography
and microwave profiles.
The basic premise derived from the theoretical modeling dis-
cussed previously is shown to be correct and valid by the ob-
tained field data. That is, when a petroleum product pipeline
leak is viewed with a microwave radiometer, there is an in-
crease in microwave radiometric temperature values along with
a decrease in the polarization contrast. The principle of de-
tection is correct and the magnitude of the anomalies is about
as expected for a 5° beam system at 1000 feet. However, the
polarization contrast (AT = Ty - T^) is much smaller than an-
ticipated. The leak areas almost assume the characteristics
of a microwave absorber. They appear to be very diffuse, with
the horizontal and vertical temperature values being very
nearly equal. This enhances detection capabilities and may
greatly reduce false alarm rates. When the horizontal and
vertical microwave profiles are coincident (observe the same
target area), the polarization contrast is nearly zero, in
51
-------
conjunction with an increase in microwave radiometric tempera-
ture, a petroleum product leak is present. These conditions
are found to be true only when a petroleum leak was encoun-
tered .
Site_Number One
Pipeline test site number 1 is located in southern Crane County
Five small leaks and one large leak occurred along the section
of pipeline examined. The five small leaks were about 5-10
barrels and the large leak about 250-300 barrels. Most of the
leaks occurred in 1968-69 and were one or more years old, ex-
cept for a small 6 barrel leak which occurred in March of 1970.
Figure 19 is the pictorial and graphic data obtained from all
the instruments flown. Surface and subsurface temperature data
were gathered at all leak sites, but soil samples were obtained
from only four leak areas. Since the thermal IR and microwave
data shown in Figure 19 are from the night missions the ground
data in Table 3 is also that obtained at night during the mis-
sion overflight. These nighttime temperature values show that
the leak areas absorb more thermal energy than barren soil and
the thermal inertia of the leak zone is higher than barren soil
This is more apparent when the daytime values (Table 4) are
compared with those obtained at night. These two sets of data
show that the thermal temperature of the leak
higher than soil barren of oil. Insufficient
an actual calculation of thermal diffusivity,
sets
zones are always
data exists for
but the data are
sufficient to show the presence of oil
electromagnetic albedo.
does cause a decrease in
Photographic Imagery
All the leak zones are detectable with color photography, but
they are not all detectable with black and white photography,
black and white IR, or color IR. Even with color photography
several of the smaller leaks (4, 5, 6) are barely discernible.
This was the primary reason for selecting this site, the fact
that leak size ranges from very small to very large.
It is interesting to note that along the pipeline right-of-way,
those areas free of vegetation and where the sand is under dense
(very loosely packed) the solar albedo is very high. This is
apparent in both the visible and photographic IR. In this case
the photographic flights were made at about 1300 hours and the
solar angle was nearly vertical.
52
-------
en
CO
HORIZONTAL -/'
COIWC-DEMCH O
LEAK ZONES
HORIZONTAL DATA
Figure 19 Multisensor Data Correlation, Site 1
-------
PIPELINE SURVEY SITE NO. 1 (NIGHT)
Ground Crew: Shover & Brown
Leak Site
1
2
3
4
5
6
Time
(local )
2245
2250
2300
2300
2310
2320
Surface
Temp °C
25.0
27.0
25.0
27.0
25.0
26.0
23.7
26.0
23.0
22.5
Sub-
Surface
Temp °C
30.0
33.5
30.0
31.0
29.5
32.1
28.3
31 .5
29.1
29.5
Surface
Water %
_
_
0.5
0.3
0.4
0.3
0.2
0.3
0.2
0.3
Sub-
Surface
Water %
_
-
0.9
9.7
0.6
0.1
1 .6
0.4
7.4
0.3
Surface
Oil %
_
_
0.0
6.0
0.1
2.8
0.1
2.5
0.1
2.6
Sub-
Surface
Oil %
_
_
0.0
7.3
0.0
2.5
0.1
3.7
0.1
2.5
Comments
Taken from oil zone
(temperature only)
Taken from oil zone
(temperature only)
Samp! e taken next to
oil zone
Sample taken from oil
spill area
Sample taken from sand
next to spill
Sample taken
Soil sample obtained
near leak area
Soil sample obtained
from leak area
Soil sample near leak
Sample from leak
SURFACE MEASUREMENTS, SITE NO. 1
Table 3
-------
PIPELINE SURVEY SITE NO. 1 (DAY)
Ground Crew: Shover & Brown
Leak Site
1
2
3
4
5
6
Time
(local )
143C
1433
1440
1444
1330
1340
1210
1210
1240
1232
Surface
Temp °C
46.0
45.0
42.0
46.2
40.0
48.0
34.0
39.0
33.0
40.8
Sub-
Surface
Temp °C
28.8
32.0
28.5
31 .0
28.0
32.0
26.5
28.2
27.5
28.2
Surface
Water %
-
-
0.5
0.3
0.4
0.3
0.2
0.3
0.2
0.3
Sub-
Surface
Water %
-
-
0.9
9.7
0.6
0.1
1 .6
0.4
7.4
0.3
Surface
Oil %
-
-
0.0
6.0
0.1
2.8
0.1
2.5
0.1
2.6
Sub-
Surface
Oil %
-
-
0.0
7.3
0.0
2.5
0.1
3.7
0.1
2.5
Comments
Taken from oil zone
Taken from oil zone
Taken from soil
Taken from oil zone
Taken from soil
Taken f rom oil zone
Taken from soil
Taken from oil zone
Ta ken from soil
Ta ken from oil zone
en
en
* All water and oil saturation data for both day and night are taken from daytime samples
Saturations were assumed not to change significantly in the 10 hour period between tem-
perature measurement.
SURFACE MEASUREMENTS, SITE NO. 1
Table 4
-------
The only photographic sensor that could be used to document
the occurance of a leak along this section of pipeline was
color photography.
It should also be pointed out, that even in this topographic-
ally flat country, oil flows away from the pipeline following
topographic features. Higher slope angles would result in
larger areas of contamination and very probably stream con-
tamination. The source of such contamination would be almost
impossible to identify visually, in areas with topographic
relief and vegetation cover.
Infrared Imagery
The 8-14 micron infrared imagery was obtained at 0230 hours
on 12 June 1970 simultaneously with microwave radiometer (13.7
GHz) data. The ground data show that the thermal temperatures
of the leak zones are always (day and night) higher than the
adjacent soil. This should make the oil stain areas appear
warmer than the surrounding soil.
An interesting phenomenon occurs in the infrared imagery. The
active area, or area of highest saturation, causes a warm an-
omaly. However, the surrounding "scar" or "oil trail" area
which consists primarily of dead oil, in more or less solid or
plastic state, appears cool in the infrared. Measured thermal
temperatures in these regions of dead oil do not indicate a
cool anomaly should exist. Therefore, it must be a change in
the emissivity function which makes these areas appear cool in
the infrared. Temperature anomalies associated with vegetation
and exposed pipe, among other things, have equal or larger IR
temperatures. This results in a high false alarm rate under
the conditions encountered. It would be extremely difficult
to identify particular leaks using only thermal infrared im-
agery. However, the pipeline right-of-ways are easily idenfi-
fied by thermal IR. This imagery shows, quite well, a number
of other pipelines crossing the main trunk line. These cross-
ing lines are part of the gathering system associated with a
nearby oil fi eld.
Microwave Radiometric Profiles
Two separate day and two separate night flights were made so
that both vertical and horizontal polarized microwave profiles
could be obtained for day and night conditions. The two pro-
files displayed in Figure 19 are those obtained during the
night flights. Ground track coincidence of the two profiles
56
-------
occurred less than 60% of the time, and even then, exact beam
coincidence was probably less than 20%. This problem is dis-
cussed under Problem Areas, Section V.
Almost all of the leaks, except the one at 7820 feet, were too
small to affect very large changes in the microwave tempera-
ture. This is because of the 56 beam diameter of the micro-
wave antenna. With a look angle of approximately 35° at 1000
feet altitude the beam diameter for the ground intercept would
be about 140 feet or an area of 16,000 ft. As shown in the
theoretical section, this would amount to microwave anomalies
of less than 2°K for the leaks at 3388 feet (6), 3831 feet (5),
8504 feet (3), 8689 feet (2), and 8745 feet (1). Therefore,
the only anomaly available for microwave analysis is that occur-
ring at 7820 feet (4).
The average microwave background temperature in the region of
leak No. 4 is 245°K which results in an emissivity (e) of 0.80,
using the simplified expression:
TA = e Tg + (1 - e) Ts
and 0.81 for the oil leak zone.
The most interesting feature associated with this leak is the
fact that the polarization contrast (AT = Ty - TH) does to
zero. That is both the horizontal and vertical microwave tem-
peratures are a measured 247°K. This means that oil leak zones
behave like diffuse materials and the reduction in polarization
differential predicted theoretically is greater than expected.
A great deal of bulldozer activity has been conducted in the
area of leak #4 in an effort to cover the saturated soil. This
is done to keep local cattle from developing sore feet and possi
bly suffering permanent injury. Even with thick cover (8-12
inches) the leak area is apparent in the microwave profile.
S|te Number Two
Pipeline test site number 2 is located about 2 miles southeast
of Wink, Texas. Six relatively large leaks were visible along
the stretch of pipeline examined. These leaks ranged in size
classification from 500 barrels to 30 barrels and stained the
surface.
Figure 20 is the pictorial and graphic data obtained from all
the remote sensors including photography. Starting from the
left and using the footage markers along the base of the micro
wave temperature profile the visible leaks are: Number 1 -
57
-------
en
CO
^**^^
HORIZONTAL
COINCIDENCE or VERTICAL AND HOSlZONTM- DATA
LEAK ZONES
' tixiii ' tees*
Figure 20 Multisensor Data Correlation, Site 2
-------
3499 feet, Number 2 - 4120 feet, Number 3 - 4540 feet, Number
4 - 7285 feet, Number 5 - 8350 feet and Number 6 centered at
8920 feet. Of these leaks, leak Number 4 - 7285 feet was en-
count^red by both the horizontal and vertical microwave beams.
All visible leak areas are identifiable to some degree by
color photography, infrared imagery (8-14y), black and white
photography, black and white infrared and color infrared. How-
ever, there are two leaks, one at 5300 feet and one at 8166 feet
that were discovered after reduction of the microwave radiometer
data and closer inspection of the color photography.
The leak at 5300 feet is slightly visible in the color photog-
raphy and even here only very slight light brown staining is
visible on very close and detailed inspection. The ground crew,
which inspected the entire measured pipeline section, did not
notice this leak area even though they walked over it several
times. No ground correlation data were obtained so saturation
and ground temperature data are not available. The leak area
at 8166 feet is evident on all the imagery, but very weak in
the thermal IR.
Photographic Imagery
All the leak zones were detected with all the photographic sen-
sors with the exception of the leak at 5300 feet which was only
visible with color photography. Listing the photographic sen-
sors in their order of "best for detection", they are:
Color Photography - Aerographic 2448
Color Infrared - Aero 8443
Black and White - Plus X Aerographic 2402
Black and White Infrared - Aerographic 2424
An interesting side issue is the very obvious change in vege-
tation color beginning at about 9799 feet. To the left (south)
the vegetation is light green and sparse compared to the area
north of 9799 feet. This is only obvious in the color photog-
raphy among the imaging instruments. However, there is very
obvious changes in the microwave radiometric profile data. The
microwave radiometric base temperature drops about two degrees
and the polarization differential very obviously increases.
This change will be more thoroughly discussed under microwave
radi ometry.
59
-------
Infrared Imagery
The 8-14 micron infrared imagery was obtained at night (0300
hours, 12 June 1970) simultaneously with the microwave radio-
metric profiles. In support of both the infrared and micro-
wave data, the ground correlation data were gathered (Table
5).
In all cases the nighttime thermometer temperatures measured
within the areas saturated with oil, both surface and sub-
surface (10 inches), were higher than those temperatures
measured in unsaturated areas. The one exception is the sur-
face temperatures measured "on" and "off" the leak at 8920
feet. In this case identical surface temperatures were re-
corded. These temperature differentials are evident in the
infrared imagery and certainly the leak areas where thermal
temperatures were measured show up as bright regions. The
indication is that the dark (visual) oil saturated ground has
a higher thermal inertia than unsaturated ground. This be-
comes evident when the daytime thermal measurements (Table 6)
are compared with the night measurements.
The surface temperature of the oil free area has an average
of about 45.0°C during the day and only 20.1°C at night. A
drop of about 23.8°C occurs for the unsaturated area. How-
ever, the surface temperatures of the leak areas average about
45.4°C during the day and 22.8°C during the night. This drop
of 22.6°C, shows characteristically higher thermal inertia for
fluid saturated soils. This is evident in the IR imagery and
the oil saturated soils are warm compared to the surrounding
soil. However, the cool halo observed in site 1 imagery also
exists around site 2 leaks.
Leak site Number 4 (7285 feet) was bulldozed over. That is,
the oil was covered up by scraping dirt from the surrounding
area over the oil area. The dozed area appears cool as com-
pared to both the undistrubed soil and those oil saturated
areas that were not quite covered. In this regard it acts
much the same as the road adjacent to the pipeline. This
cooling is due to the rapid loss of heat from the smoother,
unvegetated bulldozed area.
All of the leak areas which produce surface staining are visi-
ble to the infrared imager as warm anomalous spots along what
is normally a cool track created by the pipeline or the sur-
face scar of the pipeline. It is difficult to tell from this
pipeline right-of-way which is creating the cool anomaly, the
scar or cooling due to some other effect associated with pipe-
line presence. The subsurface temperature (10 ) is close to
90°F which is probably warmer than the oil in the line itself
and some heat is most likely transferred from the ground to
60
-------
PIPELINE SURVEY SITE NO. 2 (NIGHT)
Ground Crew: Kennedy & Shows
Leak Site
4120 ft.
(No. 2)
4540 ft.
(No. 3)
7285 ft.
(No. 4)
8920 ft.
(No. 6)
Time
(local )
0340
0350
0330
0336
0315
0310
0300
0308
Surface
Temp °C
19.5
23.4
19.0
20.2
19.4
25.0
22.5
22.5
Sub-
Surface
Temp °C
29.8
32.5
30.0
31 .6
29.8
31 .2
29.5
31 .0
Surface
Water %
0.9
1 .0
1 .4
1 .0
1 .9
5.1
0.7
1 .6
Sub-
Surface
Water %
2.1
13.6
2.5
13.0
5.1
20.9
1 .9
10.2
Surface
Oil %
0.1
17.5
0.0
18.8
0.1
10.2
0.0
27.6
Sub-
Surface
Oil %
0.1
6.0
0.0
9.0
0.2
4.8
0.0
6.9
Comments
Sample taken 75'SE
of leak center (no oil)
Sample taken from
center of visible leak
area
Sample taken 75 E
of leak center
Sampl e taken from
center of visible leak
Sampl e taken 75 ' W
of leak center
Sample taken from
di rectl y over pipe-
line
Sampl e taken 75 ' W
of leak center
Sample taken from
center of visible
leak area
SURFACE MEASUREMENTS, SITE # 2
Table 5
-------
PIPELINE SURVEY SITE NO. 2 (DAY)
Ground Crew: Kennedy & Shows
Leak Site
4120 ft.
(No. 2)
4540 ft.
(No. 3)
7285 ft.
(No. 4)
8920 ft.
(No. 6)
Time
(local )
1305
1310
1320
1330
1438
1447
1418
1425
Surface
Temp °C
47.5
46.5
45.2
46.2
40.0
44.5
43.0
44.5
Sub-
Surface
Temp °C
29.0
33.0
30.0
32.2
32.5
36.0
31 .0
31 .5
Surface
Water %
0.8
1 .1
1 .2
0.8
-
-
Sub-
Surface
Water %
2.2
13.6
2.2
9.0
-
-
Surface
Oil %
0.0
18.0
0.1
16.2
-
Sub-
Surface
Oil %
0.1
6.2
0.1
9.4
-
-
Comments
Sample taken 75 SE
of leak center
Sample taken from
center of leak area
Sample taken 75 E
of leak center
Sample taken at
leak center
In soil area) No
In oil zone ) sampl es
In soil area) No
In oil zone ) samples
en
ro
SURFACE MEASUREMENTS, SITE NO. 2
Table 6
-------
the pipeline causing secondary effects that are detectable
at the surface.
Microwave Radiometric Profiles
The microwave radiometric profiles, one vertically polarized
and one horizontally polarized, were obtained on two separate
daylight flights separated by about 10 minutes and two sepa-
rate night flights separated by about 10 minutes. The two
profiles displayed in Figure 20 are those obtained during the
night missions. The major problem areas, discussed later,
were maintaining a beam intercept of the pipeline along its
entire length and coincident of the horizontal and vertical
profiles flown at two separate times. Only short discon-
nected sections of the pipeline have coincident data.
If the apparent microwave brightness temperature is averaged
from 3966 feet to 9799 feet (Figure 20) a temperature of about
246.5°K is realized. Using the simple formula for estimating
emissivity, T/\ = e Tg + (1 - e) T£ , where:
TA = Apparent microwave temperature
e = Microwave emissivity
Tg = Thermal ground temperature
Tc; = Microwave sky temperature
the microwave emissivity is about 0.83, which is very nearly
the theoretical value for a look angle of 35°, ground tempera-
ture of 293°K, and a sky temperature of about 20°K. Also the
vertical polarized value is slightly greater than the horizon-
tal polarized values, which is necessarily true both empiric-
ally and theoretically. From 9799 feet to the end of the
flight line the vertical temperature remains at an average tem-
perature of 246.5°K but the horizontal temperature drops to
244°K, a decrease of 2.5°K. These conditions, a decrease in
horizontal temperature and an increase in polarization differ-
ential are generally caused by increased moisture content. In
this case the increase in moisture content is supported by a
definite change in vegetation color. A decrease of 2.4°K
means an increase of about 3% in soil moisture content (Kennedy
1967). In this arid west Texas area this is more than suffi-
cient to increase vegetation vigor.
Three leak zones were encountered and traversed by both the
horizontal and vertical beams. Two leaks are obvious and one
is not truly visible at the surface. However, once it was
63
-------
identified by microwave profile data, some slight staining was
found on the color photography. These three leaks represent
a nice spectrum of leak occurrances . Beginning at the right
in Figure 20 and moving to the left, the leak at 8166 feet
is relatively new, and was not visible at the surface, but is
evident from the air. The next leak at 7285 feet is old (1968),
has been partially covered by bulldozing, and is primarily
visible because of the large land scar. The third leak, yet
to be identified as a leak circumstance, is not visible on the
ground or from the air, but only by microwave observation. It
may be a false alarm but the indications are that a leak does
exi st.
A comparison of the measured data with the original theoretical
predications show relatively close agreement. In this case the
closest theoretical values are those calculated for 10% water
and 10% oil in a sandy loam soil at 25°C. Allowances must be
made for the cooling effects of increased fluid content, de-
crease in thermal temperature and temperature increase due to
surface roughness. The water content is a measured 21%, 11%
higher than that used in theory, and the thermal temperature
(night) is a measured 20°C as opposed to 25°C used in the
theoretical predictions. These would result in a decrease in
microwave temperature of about 22°K, or reduce the theoret-
ical prediction for horizontal polarization of 270° to about
248°K for the oil spill areas. The roughness of the surface
would increase the apparent temperature somewhat, but this
increase is difficult to estimate.
It should also be mentioned that the beam diameter of the very
low resolution system was about 140 feet at the ground (5°
beam, viewing angle 35° at 1000 feet altitude). This means
that the encountered leak zone did not fill the entire beam.
Therefore, the microwave thermal anomalies created by the addi-
tion of oil to the native soil are increases superimposed on
normal values. For instance the leak at 7285 feet only occu-
pied 56% of the beam area. Therefore, a microwave temperature
increase of 10 degrees would show up as a 5 degree increase.
The leak at 8166 feet occupied only 34% of the beam area and a
10° microwave anomaly would be about 3.4°. Higher resolution
systems would represent a significant increase in ability to
detect leak areas. All three leak areas encountered along
pipeline section #2 have anomalies of the proper magnitude.
Site Number Three
Test site Number 3 is located above five miles east of Wink,
Texas. Six relatively large leaks are visible along the one
and one quarter mile section of line selected for examination
64
-------
The most conspicuous leak centered at 8932 feet was over
1000 barrels and occurred at the junction of two trunk lines.
These leaks follow the same pattern as found in the other two
sites. That is, best detection is by color photography fol-
lowed by color infrared, infrared imagery, black and white
photography and black and white infrared, in the order pre-
sented. This is especially evident in Figure 21 which shows
the data arranged in the same order as the previous sites.
The correlation of pipeline leak data in the infrared and
microwave portions of the spectra is not possible for this
particular site because the ground track of the microwave
radiometer is centered about 300 feet south of the pipeline
right-of-way. Although the microwave radiometer ground track
did impinge the pipeline in some cases, neither a ground sta-
tion location nor a leak were encountered simultaneously.
This was due to the pilot's inability to judge ground inter-
cept with aircraft flight track and is discussed under Sec-
tion VI, Problem Areas. In the case of Site No. 3, it was a
case of over compensation for wind.
Also, difficulties with the ground vehicle during the night
overflight prevented obtaining ground information. Therefore
further analysis of the data is unwarranted.
65
-------
$ * *" ^" (^ife^fS" ^ * '^*v
L,
Figure 21 Multisensor Data Correlation, Site 3
-------
SECTION VII
PROBLEM AREAS AND SOLUTIONS
Most of the problem areas encountered in the performance of
this project were discovered during the data reduction phase
and are primarily mechanical.
Pilot Judgment
The predominant problem can be identified as pilot judgment.
If, as in this project, there is a requirement to fly the same
ground track several times such that the instruments sense the
same areas several times, a critical problem of visual align-
ment is apparent. The ability to fly the same surface track
is made more difficult if the instruments view the ground at
an angle off the vertical and during periods of high gusty
winds. High winds, approximately 30 knots due east at ground
level, were especially troublesome during the daytime flights.
This is apparent from the aircraft shadow which is visible
on the photography. During the night flights ground level
winds were "dead-still".
The pilot visually aligns the nose of the aircraft with the
ground markers for a continuous line of sight track. However,
if the aircraft is "crabbed" into the wind the alignment geo-
metrically offsets the aircraft path in the direction in which
the wind is blowing. This is graphically shown in Figure 22.
When the measuring instrument points aft, as was the case,
the problem is amplified and the actual ground track is further
displaced in the direction in which the wind is blowing. Cer-
tainly the pilot cannot pre-judge the offsets involved and
make the necessary corrections by performing mental gymnastics.
Since the instrument is fixed, that is, hard mounted to the
aircraft frame, it appears that the best solution is to fly a
series of lines offset from each other in the direction from
which the wind is blowing. The distance to offset can be
"guessed at" by observing the drift angle obtained from the
doppler navigator. It may be necessary to develop some simple
nomographs which show the expected offset as functions of wind
velocity, wind direction and altitude. This would increase the
probability of a proper or usable ground track, but will not
eliminate the necessity of flying the same general line at
least three or four times with progressive offsets in the di-
rection from which the wind is blowing. Of course this prob-
lem is completely eliminated if microwave imaging systems were
used.
67
-------
Wind
Direction
Desired Track
'a Crab Angle
\
Aircraft Track
Figure 22 Profile Offset Due to Wind
68
-------
Aircraft Dynamics
In addition to the tracking error mentioned above, aircraft
dynamics pitch, roll, yaw and velocity changes also cause
deviations from the desired ground track because the meas-
uring instruments are hard mounted. All of these parameters
are continuously recorded on magnetic tape during the data
runs and are correlated with all other data using the 400 Hz
signal discussed in a previous report. These data are in-
puts to the computer data reduction program and the entire
flight line is corrected to display the actual ground track
from which data is obtained. However, knowing that the prob-
lem exists and performing error free corrections does not
make target acquisition any more accurate. This is after
the fact information and only informative in that it deter-
mines whether or not the sensor beams passed over the desired
area. In this particular case, the identified and marked
pipeline leak sites. Since realtime data reduction in the
aircraft is impractical, the solution to this problem is to
increase the number of over-passes to be relatively certain
that the leak areas were indeed observed. Of course, in an
operational system a multibeam system of sufficient path width
would be used to insure observation. However, in the case of
a research project where a single beam system is used multi-
ple flights are required.
The actual ground tracks of microwave radiometer data acquired
during performance of this field project are shown in Figures
23, 24, and 25.
Polarization Differential
The computer simulation program which initiated this project,
predicted that when a microwave radiometer observed a leak
circumstance the microwave temperature would increase and the
polarization contrast (AT = TV - Tn) would decrease. This
determination was based on the assumption that the horizontal
and vertical target areas were exactly the same. As explained
earlier, exact coincidence is rather rare for a system which
does not measure both components simultaneously and redundant
flights are required. However, when partial coincidence occurs,
that is major parts of observing beams cover the same area, and
an oil leak is present the polarization contrast should de-
crease significantly along with temperature increases for both
the horizontal and vertical microwave antenna temperatures.
This was found to be true in almost every case of coincident
leak abservations . However, it is possible that non-coincident
portions of the observed areas could contain objects which cause
69
-------
Red - Horizontal Polarization Beam
Blue - Vertical Polarization Beam
Figure 23 Profile Ground Track, Site 1
-------
21
28
22
27
_L±.
23
26
Red
Blue
Horizontal Polarization Beam
Vertical Polarization Beam
Fiif'iire 24 Profile Ground Track, Site 2
-------
Red - Horizontal Polarization Beam
Blue - Vertical Polarization Beam
Figure 25 Profile Ground Track, Site 3
-------
an increase in polarization contrast and even make one or the
other polarized components tend to be cooler. In this case a
leak would be passed over and not detected. While this prob-
lem was apparently not encountered in this program it should
be considered in future projects.
The obvious solution is to always take data using a dual polar-
ized microwave radiometer system, which simultaneously obtains
both horizontal and vertical data.
73
-------
SECTION VIII
APPENDIX A
RADIOMETER DATA REDUCTION PROCEDURES
Flight Dynamics Parameters
In order to correlate brightness temperatures with ground lo-
cation, a time profile of the airplane's position and its
angles of roll, pitch and yaw is required. The angle data are
provided by an Incredata recording system. This system accepts
voltages from the aircraft motion systems, such as the doppler
navigator, magnetic compass, etc., and digitizes the informa-
tion, which is stored in binary on a seven-track tape, with
odd parity. The system accepts an aircraft clock output and
at t, e beginning of each record the time of day is recorded in
hours, minutes and seconds to an accuracy ± .001 seconds. When
the system is first turned on, values of twelve thumbwheel set-
tings are first written on the tape in binary-coded decimal for
mat, odd parity. The thumbwheels are manually set to values
which identify the month, day, site number, flight line number,
the number of the pass over the flight line, and whether the
flight is day or night.
At the end of a mission an Incredata tape is available which
contains thumbwheel settings identifying the data to follow,
time of day, and the angles of roll, pitch and yaw at 400
millisecond intervals.
The first two words of every record on the tape are printed
to obtain thumbwheel settings, to count records and to locate
end-of-files , if any. This information is needed for input to
a computer program which sorts the data, unpacks the binary
representation and forms FORTRAN word representation. If more
than one input Incredata tape was required it also merges the
tapes. Finally, from the calibration curves of voltage versus
angle and the binary word length versus voltage limits, the
data are converted to tables of roll, pitch and yaw in radians
versus time of day.
The tape representation at the top of Figure 26 shows thumb-
wheel record followed by a number of DATA records. Between
records there is a 1-inch unrecorded area denoted IRG (inter-
record gap) .
A thumbwheel block contains 12 characters which are set short-
ly before beginning a flight line. The 12 characters, from
left to right, are in binary coded decimal and having the
followi ng meanings:
75
-------
TAPE
STARTS
X
RECORD
TIME TO RECORD
10 SEC.
IRC
12 BCD
CHARACTERS
DATA RECORD
IRG
DATA RECORD
IR3 \
\ IRG
\
120 MILSEC TI.1E TO RECORD
CTl
1/1
CO
IRG
' IM E
V
^
10
HRS.
1
HRS.
10
M I N .
M I N .
10
SEC.
SEC.
\
400 MILSEC TIME TO RECORD
Sl
\
X
1
\
\
\
V 55 TIME
Dl
40
D2
D3
MILSEC TIME TO RECORD
D4
D5
D6
D7
D8
°9
D10
\
Dl
D2
D3
D4
DS
X
IRG
IRG
IRG
TAPE
START
\
DATA
DATA
DATA
\ \
A V
IRG IRG
A V
\ V
DATA
T
DATA
IRG\
\
Figure 26 Schematic of Incredata Tape
-------
Character Meaning
1 10 ' s of months
2 units of months
3 10's of days
4 units of days
5 not used
6 site number, even 10's digit
7 site number, units digit
8 flight line number,
even 10's digit
9 flight line number,
units digit
10 refly* number, even 10's
digit
11 refly number, units digit
12 day flight = 4,
night flight = 8
The first four tape tracks record the values in binary, the next
two tracks are blanks, and the seventh track is employed for
parity.
The second tape representation of Figure 26 is an expansion of
a DATA record block. The six left-most tape frames are in the
same binary coded decimal format as is the TW record and con-
tain the time. This is clock time and although only shown to
one second in significant figures, it is actually accurate to
one mi 11i second.
Following the time there are 24 sets of parameter values, in
the figure denoted S] , $2 D]Q. The D's are two-charac-
ter, 11-binary-digit numbers which are the values of flight
* Refly number refers to how many times a line was flown before
successful data collecting occurred.
77
-------
dynamics parameters, as follows:
D-1 = ground speed
D2 = drift angle
Do = miles to go along track
D, = not used
D5 = magnetic heading
D, = pitch
D? = roll
Dg = course altitude
Dg = fine altitude
DIQ = signal ground
To convert angles from digital counts to radians, calibration
curves are used. These curves relate voltage to angle, and
angle to counts. The range of 0 to 5 volts is digitized into
a register of eleven bits, that is, a range of 0 to 2^' - 1 -
2047. This yields a conversion factor:
2047 counts = 5 volts
The pitch calibration provided the following voltage-to-angle
relationships:
2.72 volts = 20°, nose up
2.15 volts = 20°, nose down
from which the pitch is given in radians by the equation
3 = 2.982368 - .0029916770
where D is in counts, and 8 to pitch.
The roll calibration yields
2.61 volts = 20°, right wing down
2.04 volts = 20°, right wing up
from which the algorithm for roll (a) is derived:
78
-------
a = -2.847641 + .0029916770
again D in counts, a in radians.
The magnetic heading calibration yields the set
0 volts - 20° azimuth
5 volts = 370° azimuth
from which the equation to compute magnetic heading, H", is de-
termi ned
H" - 20 + . 1709819D (degrees)
Denoting the magnetic declination AH, the true heading, H, is
given by
H = H" + AH, modulus 360 (degrees)
from which the yaw, can be calculated using
y = .017453293 (H' - H) (radians)
where H' is the azimuth of the flight line.
The computer subroutine which makes these conversions from
counts to angles also determines the times associated with each
angle.
A ground-track coordinate system is required for remote sensor
system ground intercepts and is defined as follows: The X-axis
is horizontal and coincides with the flight line, positive in
the direction of flight; the Z-axis is vertical, positive up-
wards; the Y-axis is that required to form a right-handed co-
ordinate system. This, the ground track system and the air-
craft-fixed system, Figure 27, coincide when roll, pitch and
yaw equal zero.
The aircraft attitude measuring system is mechanized such that
roll, pitch and yaw may be considered unordered, simultaneous
rotations. Furthermore, careful operation during the flight
constrains these to be small angles so that infinitesimal ro-
tation theory may be utilized in data reduction. Accordingly,
the transformation of the unit vector along the sensor beam
axis from aircraft-fixed coordinates to ground track coordin-
ates is performed with the following equations:
Q! = PI - YP2 + 3 P3
Q = yP + P " a P
79
-------
CO
O
X ,Y ,Z = Ground Coordinate
axes
x,y,z = Aircraft Coor-
dinate axes
Figure 27 Definitions of Roll, Pitch and Yaw
-------
Q3 = 3P] + aP2 + P3
where (p-|, P2, P3)T and (Q-j , Q2, Q3)T are the X, Y and Z com-
ponents of the unit vector in the aircraft-fixed and ground
track coordinate systems, respectively; and a, 6 and Y are
the roll, pitch and yaw angles, respectively, in radians,
positive when counterclockwise.
Reduced data was accomplished with the assumption that flight
lines took place over flat terrain. Accordingly, the coordin-
ates of the ground intercept of the antenna axis are given by
x = d - (h / Q3) Q1
Y = - (h / Q3) Q2
where x and y are distances down and cross track of the ground
intercept, and d and h are the distance down track and altitude
of the aircraft.
The set of triplets (T,x,y) comprise the final output of data
processing. (T denotes brightness temperature.) This infor-
mation, a table of ground locations and corresponding radio-
metric temperatures, is written on magnetic tape and also
printed for permanent hard copy.
Establishment of a Common Origin of Time
The absolute reference for time is the aircraft clock. To re-
late a measurement of radiometric brightness temperature to
the corresponding position and attitude of the airplane, the
times of the two data sources must be expressed in the refer-
ence timing system. That is, microwave temperature is given
as a function of time, and aircraft movement is given as a
function of time, and it is required that both functions of
time have the same epoch. Also, to calculate the ground inter-
cept of the radiometer antenna axis it is necessary that the
times of the beginning and ending of aerial photography and
infrared scanning over a flight line be determined in the same
timing reference system. The roll, pitch and yaw of the air-
craft are given directly in terms of the time of day. Two
representative print-outs are illustrated in Tables 7 and 8.
The first entry, which is:
TW = 06110030104
is the thumbwheel setting, mentioned earlier, and is a record
of data, line number and other identifying information. The
81
-------
T*= 'ft! 100301004
TIML = o-^jz
' I l? . o_
4 --IT 707 0 1=>5 1121 *<*? 941 33= 60t> 0
5 ^27 il'.\ 4 104 Ii4l 9/7 924 321 5/9 0
6 -ij 76 j j ]b4 1149 10 OH 9't-sj 332 5'W 0
7 '! T.'5 5 lt>l- llbl H'06 946 3/-6i 60s; 0
b . / 061 2 1^2 1123 9v^ 93o 339 (-12 y
9_ < r.4 /i'- b.._. 10Q .._1 12'j .... ^^0 ... 917' 32H b^2 0
10 :" 73;< 3 Ihfc i!49 luLi3 9J4 317 bt>9 0
11 -.' ~ f- 4 lob ii5/J- IUUP 9"3b 343 599 0
12 "Ov ti-i 4 loo lij.:' 10U4 9b5 34 S 615 0
13 f-7 .7i- 4 )b5 ill./ 9a^j y^2 326 5^0 0
14 c- /^^ 5 lt>5 ii3b 9^2 9u/ 323 5/4 0
15 i- ^"-2 ... 4 its/ ..11+2 _. l«U-f 9jQ 335 5'^'j _.....0_
10 /' - 5 lb^ 1133 ^96 944 343 61? o
I/ >. i'-*.J: I l^c ill1- 9Hh 917 337 h'}t: 9
Id -.. (*;> 4 1'04 U3/ ^"~> d9b 319 bo;.) 0
1-^ .'- /'+J 4 io5 iibj I'.H)" 92 J 33fc 5/2 0
2r '- ooS 3 1"! ilb4 10U4 9J2 34b" t> :' 0
21 /"4 _ 2 13'-- H3c._ '^94 _ 924 345 '?'.? 0
22 :. /ii 3 13' 1123 'H-4 916 327 Lj-''- " 0
23 ^ ' ^2 ? ': It t 11 s i 100? 929 318 '-,!'> 0
24 , /'") ' 106 11S2 1UUS 9bb 342 599 o
INCREDATA DIGITAL TAPE PRINT-OUT
TABLE 7
82
-------
TW= 061100301004
TIME = &4278
/
. 1
2
3
4
5
6
7
B
9
10
11
12
13
14
_lb
1ft
17
lb
19
20
2i
22
23
24
1
»13
42H
406
411
<+2*
4lb
3'^d
426
4dl
4 0 '->
+lr<
*24
40b
M.J
"32
"Jc
u 'jo
"2D
ul
-» '.) '
4 !'»
415
4 o 3
" 1*
2
730
709
bH5
b99
731
/19
/6(l
732
/32
736
731
/68
/2l
/17
74 y
0/5
b07
/16
723
b.o4b 0006010
INCREDATA DIGITAL TAPE PRINT-OUT
TABLE 8
83
-------
second entry in the tables is a time, e.g., in Table 7, the
entry
TIME = 84202
is the aircraft clock time, that is, time of day (in seconds).
The remaining entries relate to aircraft motion.
This same information was simultaneously recorded on the serial
digital channel of the 14-track analog tape. Figure 28 illus-
trates a strip chart recording of this channel, along with a
recording of the audio channel.*
The serial digital channel appears as broken ribbons of inked
paper. Point A is the beginning of one bar. The reason the
graph immediately to the right of A is a solid bar is that
the data on the tape consists of zeros and ones, since original
Incredata data is binary. The zeros and ones are so closely
packed that the chart recording pen-line width appears to pro-
duce an unbroken bar.
At point A all systems are in a steady state. The numbers in
Table 7 are the values (in decimal notation) of the bar star-
ting at Point A. Therefore, the time of Table 7 is the time at
Point A.
The audio channel, recorded at the same time as the serial dig-
ital channel, and is in synchronization with the serial digital
track. Counting the bars on the chart recording of the serial
digital channel is one-to-one with counting tables in the print
out of the Incredata tape. Thus, the time of day of Point C is
readily established, from which the time of the end of valid
data, Point D, can be measured.
At the times of insertion of the 400 Hz pulses on the audio
channel, some of this energy is diverted to the infrared scan-
ner and light flashes, exposing small dots on the edge of the
scanner film. This establishes a time-tie between all systems.
The numbers representing microwave temperatures are assigned
reference times in the following way:
When the ana 1og-to-digital conversion takes place the radio-
meter channel is connected to one pen of the Clevite chart
A reproduction of an actual strip chart recording is not
shown, because noise would complicate this explanation.
84
-------
TIME INCREASES
A
SERIAL DIGITAL CHANNEL
oo
en
BEGINNING
OF LINE
END OF
LINE
TIME, FROM INCREDATA
PRINT-OUT
(TABLE 1)
TIME, FROM INCREDATA
PRINT-OUT
(TABLE 2)
Figure 28 Strip Chart Recording of Serial Digital and Audio Channels
-------
recorder and the digital instrumentation is connected to the
other pen. The Ampex machine is turned on in play-back mode
and the chart recorder is turned on and one pen begins plot-
ting the random voltages at the beginning of the tape. Then
the analog-to-digital coverter is turned on, and the second
chart recorder pen changes from drawing a line at zero volts
to drawing a line at some non-zero voltage. This chart, show-
ing voltages from the radiometer tape along with the start of
analog-to-digital conversion is the key to determining the
time of day corresponding to word number one written on a com-
puter tape.*
Figure 29 is a schematic of three strip chart graphs. Point
A represents the beginning digitizing. Point B represents the
time when the tape recorder aboard the aircraft began recor-
ding outputs from the radiometer. Points E and G are the be-
ginning and end of used data.
Spikes on the data as at points D, E and G occur at the ends
of one or more of ;the 400 Hz bursts. These are "wild points"
in the digital version and are used to corroborate the corre-
lation between word sound on the digital tape and time on the
analog tape. Times during radiometric "events", such as at
Point F, are readily found in the digital print-out.
Brightness Temperatures
The fourteen-track analog tape has three channels assigned to
microwave radiometer data recording, an audio channel, a ser-
ial digital channel, and the radiometer (video) channel. The
audio track records the flight log information as spoken by the
pilot. The serial digital channel records the Incredata out-
put at the same time the Incredata tape records itidentical
information, but the fourteen-track analog tape records the
data in serial mode on seven-tracks. Play-back with strip-
chart plotting of the serial digital channel and the audio
channel simultaneously utilized with the computer print-out
of the Incredata digital tape yields the tie between the time
of day and the 400 Hz pulses on the audio tape and on the in-
frared film.
The radiometer channel records a continuously varying voltage
proportional to the brightness temperature at the antenna.
Plotting this channel and the audio channel at the same time
The word "word" is used in its computer context. For ex-
ample, "The CDC 6600 has a 60-bit word length."
-------
CHANNEL FOR BEGINNING OF DIGITIZING
A_A
AUDIO CHANNEL
00
RADIOMETER CHANNEL
ABC D E
Figure 29 Strip Chart Recording Used For Timing
-------
shows the time-correspondence between the two channels, in
particular, the beginning of radiometer data relative to the
400 Hz pulses.
The analog to digital process is performed with a Preston 1400
converter connected with an IBM System 360, Mod 44 computer,
and results in an IBM nine-track digital computer tape with
record lengths of 512 words.
At this state in the procedure, two digital computer compati-
ble magnetic tapes have been prepared, one containing a his-
tory of aircraft attitudes and one containing digital counts
proportional to radiometric temperatures.
Data Processing
There are two magnetic tape data sets utilized by the computer,
one of aircraft roll, pitch and heading versus time of day,
and one of brightness temperature in digital counts versus
time of day. The first step in datj processing is to pro-
duce a third tape for each flight line containing aircraft
position versus brightness temperature in degrees Kelvin.
Input to the computer program which
sists of the following quantities:
writes the new tapes con-
number of files and records from the beginning
of the temperature tape to the second 400 Hz
burst word
along-line speed of the aircraft, derived from
the doppler system and checked by imagery meas-
urements
number of words of original data averaged to make
one output value
flag to employ the option to compute the stan-
dard deviations of output values
flag to employ the option to filter the output
values
filter length, if output values are to be fil-
tered
flag to employ the option to print the output
values , if desi red
88
-------
flag to employ the option to reverse the order
of the input data, if required
t site number and line number
factor to convert digital counts to degrees
Kelvi n
correction to zeroset of radiometer, if required
effective realtime sampling rate used in the
analog-to-digital process
time of the first data point
The data is compressed by averaging a number of input tempera-
tures and considering the average to be an improved tempera-
ture located at the midpoint of the number of values averaged.
Data compression is also employed to reduce the number of
points which must be processed. The corresponding ground
distances between temperatures are of the order of magnitude
of one foot and one-tenth feet, respectively.
The algorithm for converting digital counts to brightness tem-
peratures is derived in the following way: The Prescott 1400
Analog-to-Digital Converter is adjusted to that five volts
from the analog tape would fill its 13-digit register with 1's
This means that
0 volts = 0 = 0°K
5 volts = 213 -1 - 8191 = 500°K
since the radiometer circuit is calibrated at 1 volt equals
100°K. Therefore
1 count = 500 / 8191 = .06104261°K
Flight lines processed with this equation yield brightness tern-
peratures with reasonable values for the thermometric ground
temperatures measured by ground crews.
Profiles which are plotted on an X-Y plotter provide a graph
of temperature versus distance down line. The distance down
line is the perpendicular projection of the actual ground in-
tercept on the flight line.
89
-------
j 1 A< crs.sjfw; Number
W
5
2
Snbjei. ( Field &. Group
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Organization
Title
Fluid Product Pipeline Leak Detection from Airborne Platforms
10
Authori ^
Joseph Kennedy
16
Project Designation
21
Program #16020 FQT / 12/70
Note
O O Citation
December 1970, Final Report, Demonstration Grant 16020 FQT, 89 pp
Descriptors (Starred First)
Microwave Radiometry, Remote Sensing, Infra-red Imagery.
Pipeline Leak
25
Identifiers (Starred First)
Pipeline Leak
27 Abstract
A project to demonstrate the applicability of microwave rad'iometry to pipe detection
of leaks in pipelines. The demonstration involved flights over three different
pipeline sections containing known leaks determination of the apparent microwave
temperature, microwave polarization contrast and taking of infrared imagery
was carried out and correlated with ground data.
It was found that the apparent microwave (13.7 GHz) temperature increased
significantly at the site of a leak. Also, the polarization contrast de-
creased and the infrared imagery showed a warm area surrounded by a cool
halo. When these three circumstances occurred together a leak was posi-
tively identified.
A 6.s tractor
Louis G. Swaby
Environmental Protection Agency
/; R 102 < R E V JULY 1969)
W R S I C
SEND, WITH COPY OF DOCUMENT, TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.-S, DEPARTMENT OF THE INTERIOR
WASHINGTON, D, C, 20240
«U.S. GOVERNMENT PRINTING OFFICE: 1972 484-483/61 1-3
------- |