EPA-R5-73-014
MAY 1973 Socioeconomic Environmental Studies Series
Data Acquisition Systems
in Water Quality Management
Office of Research and Monitoring
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and
Monitoring, Environmental Protection Agency, have
been grouped into five series. These five bread
categories were established to facilitate further
development and application of environmental
technology. Elimination of traditional grouping
was consciously planned to foster technology
transfer and a maximum interface in related
fields. The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
U. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the SOCIOECONOMIC
ENVIRONMENTAL STUDIES series. This series
describes research on the socioeconomic impact of
environmental problems. This covers recycling and
other recovery operations with emphasis on
monetary incentives. The non-scientific realms of
legal systems, cultural values, and business
systems are also involved. Because of their
interdisciplinary scope, system evaluations and
environmental management reports are included in
this series.
-------
EPA-R5-73-014
May 1973
DATA ACQUISITION SYSTEMS
IN
WATER QUALITY MANAGEMENT
by
Robert C. Ward
Colorado State University
Fort Collins, CO 80521
Project 16090 FUO
Project Officer
Mr. Dale B. Parke
Region VII
Environmental Protection Agency
173 S Baltimore Avenue
Kansas City, Missouri 64108
Prepared for
OFFICE OF RESEARCH AND MONITORING
ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D. C. 20460
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402
Price $2.85 domestic postpaid or $2.80 QPO Bookstore
-------
EPA Review Notice
This report has been reviewed by the Environ-
mental 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
The role of routine water quality surveillance in a water
quality management program was investigated. This included
a delineation of the objectives a state water quality pro-
gram based upon the state and federal laws. Seven specific
objectives are listed under the two general objectives of
prevention and abatement. These seven specific objectives
are planning, research, aid programs, technical assistance,
regulation, enforcement, and data collection, processing,
and dissemination.
The information requirements of the objectives were deline-
ated in general terms. Each objective was broken down into
the genreal activities required for its accomplishment and
the data needed for each activity was identified.
A survey of the current state-of-the-art of grab sampling,
automatic monitoring, and remote sensing was performed.
Each data acquisition technique was analyzed as to its
capabilities, reliability, and cost.
With the above information, a design procedure was developed
for designing a state water quality surveillance program
which is responsive to the objectives of the agency. The
design procedure has two major aspects: (1) determination
of the state agency's strategy with respect to its attack
on water pollution, and (2) characterization of the streams
in the state so that rational decisions with respect to
sampling location and frequency can be made. The optimum
grab sampling network was designed and then the possibilities
of substituting automatic monitoring and remote sensing and
various points in the surveillance system were explored.
The design procedure was applied to the State of Colorado
and a water quality monitoring system was developed for the
Water Pollution Control Division of the Colorado Department
of Health. Financial and manpower constraints were con-
sidered in the design.
This report was submitted in fulfillment of Project Number
1609 FUO, under the (partial) sponsorship of the Environ-
mental Protection Agency.
111
-------
CONTENTS
Section Page
I Conclusions 1
II Recommendations 3
III Introduction 5
IV General Objectives of State Water Quality
Control Programs 7
V Data Requirements of General Objectives 19
VI State-of-the-Art Summaries 31
VII Design Procedure for Water Quality Sur-
veillance Systems 39
VIII Application of Design Procedure to
Colorado 61
IX Acknowledgments 125
X Literature Cited 127
XI Appendices 129
-------
FIGURES
Paqe
1 BREAKDOWN OF THE OBJECTIVES OF A STATE WATER
QUALITY MANAGEMENT PROGRAM 10
2 SUMMARY OF DATA NEEDS AS RELATED TO THE WORK
ELEMENTS 28
3 STRATEGY EVALUATION FOR STATE WATER QUALITY
MANAGEMENT PROGRAMS 41
4 POLLUTION CONTROL STRATEGY ASSOCIATED WITH
DATA ACQUISITION STRATEGY 45
5 FLOW DIAGRAM FOR SURVEILLANCE NETWORK SIM-
ULATION 50
6 EXAMPLE INSTANTANEOUS SPILL DISPERSION CURVE 53
7 POLLUTION CONTROL STRATEGY DETERMINATION
FOR COLORADO AS COMPLETED BY A WPCD STAFF
MEMBER 6 6
8 CACHE LA POUDRE RIVER - SEPTEMBER 1970 73
9 DETECTION VERSUS NUMBER OF SAMPLES FOR ONE
STATION AT LOWER END OF REACH WITH 0-3 DAY
RANDOM SPILL DURATION 81
10 DETECTION VERSUS NUMBER OF SAMPLES PER STATION
FOR CONSTANT NUMBER OF TOTAL SAMPLES FOR 100
MILE REACH 83
11 SPILL DETECTION BY ONE STATION AT END OF 20
MILE REACH FOR ONE, TWO AND FOUR SAMPLES
DURING ONE MONTH PERIOD 85
12 SPILL DETECTION BY ONE STATION AT END OF 20
MILE REACH FOR EIGHT AND TWELVE SAMPLES
DURING ONE MONTH PERIOD 86
13 SPILL DETECTION BY ONE STATION AT END OF 20
MILE REACH FOR SMALL SPILL DURATION AND VARI-
OUS NUMBERS OF SAMPLES DURING ONE MONTH PERIOD 87
14 DETECTION OF INSTANTANEOUS SPILLS 88
VI
-------
FIGURES
Page
15 DETECTION VERSUS SAMPLING FREQUENCY FOR
VARIOUS SPILL DURATION RANGES 90
16 DETECTION EFFECTIVENESS VERSUS K.UMBER OF
SAMPLING STATIONS PER STREAM WITH A CONSTANT
SAMPLING FREQUENCY OF 12 SAMPLES PER YEAR
PER STREAM 91
17 NUMBER OF SAMPLES VERSUS ALLOWABLE ERROR FOR
a EQUAL ONE AND TWO 95
18 NUMBER OF SAMPLES VERSUS ALLOWABLE ERROR FOR
a EQUAL THREE, FIVE, AND TEN 97
19 NUMBER OF SAMPLES VERSUS ALLOWABLE ERROR FOR
a EQUAL 25 AND 50 98
20 NUMBER OF SAMPLES VERSUS ALLOWABLE ERROR FOR
a EQUAL 100 AND 500 99
21 OBSERVED DEVIATION FROM TRUE MEAN 103
22 RELATIONSHIP OF COST TO NUMBER OF SAMPLES
FOR COLORADO 108
23 COST-EFFECTIVENESS RESULTS FOR THE PRIMARY
NETWORK IN COLORADO FOR A 0-3 DAY SPILL LENGTH 111
24 ACCURACY LIMIT VERSUS COST FOR THE SECONDARY
NETWORK IN COLORADO 114
25 SECONDARY PARAMETER LIST FOR COLORADO AS
REPORTED FROM STORET 117
VI l
-------
TABLES
NO.
1 Application of Automatic Monitors 34
2 Effect of Number of Samples on Detection gg
3 Effect of Sampling Frequency and Number of
Stations on Event Detection with Constant
Number of Samples §2
4 Primary Network Analysis Results for Colorado 93
5 Sampling Frequency Compared to Effectiveness
Levels for Colorado's Primary Network 94
6 Base Level Surveillance Model Results 102
7 Number of Samples Required for Various
Accuracy Limits for Colorado's Secondary
Network 105
8 Effectiveness Versus Cost for the Primary
Surveillance Network in Colorado 110
9 Effectiveness Versus Cost for the Secondary
Surveillance Network in Colorado 113
Vlll
-------
SECTION I
CONCLUSIONS
A rational and practical routine water quality surveillance
design analysis can be performed using the procedures out-
lined in the report. The amount of refinement used in the
design analysis is dependent upon the information available
on the character of the streams involved. The systems anal-
ysis approach used in developing the design procedures pro-
vides an excellent basis for the development of useful and
practical surveillance system design information. This
information is extremely valuable to the water qualtiy
manager who must make surveillance system design decisions.
The systems approach used in the report also permitted the
viewing of the surveillance system as a whole rather than
as a series of parts. Perhaps the most important aspect
of the systematic approach used in the report is the sim-
ple fact that in order to analyze a surveillance system in
a broad context, it must be completely understood. This,
by itself, often leads to improvements in the surveillance
system.
The design or evaluation of a routine water quality sur-
veillance network must be related to the goals and ob-
jectives of the agency collecting and acting upon the data.
This helps to insure that the surveillance effort is di-
rectly supporting the agency's water quality control ef-
forts. Depending upon the goals of an agency, some of
the six general objectives discussed in this report may
be emphasized over others. This, in addition to the meth-
ods of water quality control (permits, site approvals,
etc.), forms the agency's strategy with respect to control
of water quality. The goals, and consequently the stra-
tegy, are established by the personnel, funding levels,
priorities, laws, etc., under which the agency must op-
erate. The strategy, therefore, however relative, must
be determined so that surveillance system design is re-
levant.
The use of automatic monitoring for routine surveillance
purposes is economically competitive with grab sampling
in the area of spill detection. Due to its limited param-
eter measurability, a parameter evaluation of the stream
involved will provide the final answer as to whether auto-
matic monitoring is practical for that stream.
Remote sensing is presently of questionable value in a
routine water quality surveillance system. It is ex-
tremely useful, however, in the special surveys that
-------
should precede the development of a routine surveillance sys-
tem containing grab sampling and automatic monitoring. Fur-
ther development and refinement in remote sensing procedures
will continue to make it a surveillance technique to be con-
sidered in both special surveys and routine monitoring.
Grab sampling will continue to be the backbone of most rou-
tine water quality surveillance systems. This is due to
parameter measurability and cost. Grab sampling, however,
can and should be made more effective through a thorough
analysis of sampling station locations, sampling frequencies,
and parameters measured. This can be accomplished through
procedures given in the report.
-------
SECTION II
RECOMMENDATIONS
1. Due to the unique self-evaluation required in the sur-
veillance design procedure, it. is recommended that
each state agency be encouraged to analyze its data
acquisition system using the results of this report as
a guide.
2. In order to facilitate the first recommendation, it is
suggested that a conference or short course be scheduled.
The report could be sent to the state agencies for their
review and then a representative of the agency would
attend the conference at which time the design procedure
would be thoroughly reviewed. The representative would
then be encouraged to return to his agency and see that
his surveillance network and agency are analyzed in a
comprehensive and systematic manner.
3. The subject matter of this report has been mainly con-
cerned with the actual acquisition of data. The results
of this report need to be tied in closely with a data
handling system such as those currently being developed
by the Pennsylvania Department of Environmental Resources
and the Michigan Water Resouces Commission. This would
provide a sound basis for action upon data rather than
reaction to a crisis.
4. There are many types and sources of data utilized by a
state agency other than data from its routine surveil-
lance network. For this reason, it is recommended that
further work be instituted in the area of optimum util-
ization and acquisition of all data from all available
sources. A project of this type may illustrate the
need of an agency to rely less upon data collected in-
ternally and a need to rely more upon data collected
externally. This would free agency personnel from the
collection of data for the sake of data collection and
would put more emphasis upon the action which is to
result from analysis of data.
5. As for the application to Colorado, it is recommended
that the Colorado agency use the results of the appli-
cation section to evaluate their current surveillance
efforts. Also, as further information becomes avail-
able the analysis should be expanded beyond.that given
in the report.
6. Since the efforts of any one person or group of people
can in no way encompass the whole problem completely,
-------
it is recommended that as more practical experience is
gained in applying the design procedure, the procedure
should be modified or refined so that it can remain a
valuable tool with which to manage a surveillance
system.
-------
SECTION III
INTRODUCTION
Water quality degradation has been widely recognized as one
of the urgent problems facing modern society. In response
to an increasing public awareness of this problem, and in
recognition of the importance of achieving and maintaining
water supplies of acceptable quality, the U.S. Congress
enacted a federal water quality act in 1965. The 1965 act
required individual states to establish water quality stand-
ards for interstate streams and to develop detailed plans
for implementing and enforcing these standards.
The Federal Water Pollution Control Act, as amended, stimu-
lated review and revision of state statutes and institutional
arrangements for the management of water quality. In many
states new commissions, boards, authorities, etc., were es-
tablished to handle water pollution control activities. In
other states the existing agencies were given new duties
and authority. These water pollution control agencies were
empowered to adopt comprehensive water pollution control
plans, identify sources of pollution, and to adopt and en-
force water quality standards. In many states the above
responsibilities presented a problem in that there was a
lack of basic data on which to base an assessment of the
water quality situation existing throughout the state. All
states faced the problem of limited manpower when the new
assignments were set forth.
In order to obtain basic data and to enforce water quality
standards, the states either established or expanded their
network of water quality monitoring stations. The networks
command the attention of engineers, chemists, aquatic samp-
lers, systems analysts, and other staff of the state agen-
cies. Yet, even with the commitment of a substantial
effort, it is apparent that these surveillance networks do
not meet the overall needs of the water quality management
programs. Many waste sources go undetected. Frequently,
pollution problems have passed before their existence is
noted by the surveillance system.
Thus, one must conclude that water quality sampling and
data analysis consume a large portion of the valuable
staff time without providing the information necessary to
support water quality management programs. At the very
least, one would expect that an analysis of the information
obtained since the Federal Water Pollution Control Act, as
amended, would indicate improvements which would make the
-------
existing systems more effective. On the other hand, it
may be possible to develop an entirely different system with
a much lower professional manpower requirement by turning
to more advanced methods of instrumentation and data analy-
sis. If so, the result would be to release manpower criti-
cally needed for other responsibilities.
With the above situation in mind, it will be the objective
of this study to determine the role of water quality infor-
mation collection and interpretation in the water quality
management program and then to identify the most effective
means of filling this role. In the process, the needs to
be served by the water quality information collection and
evaluation system will be identified, different means of
obtaining and evaluating water quality information will be
determined, and a water quality information system will be
designed for the State of Colorado.
The different methods of information collection and inter-
pretation to be considered will include: 1. traditional
sampling techniques with laboratory analysis to identify
materials in the water sample and statistical analysis of
the resulting water quality data; 2. remote sensing to
determine "in situ" the character of materials in the
stream coupled with automated screening and analysis of
data and automatic photographing of unusual events; and
3. automatic monitoring with wet-probe instrumentation
coupled with automatic screening and interpretation of data,
and automatic sampling and/or photography of unusual events.
Different systems will be evaluated in terms of their cost,
manpower and hardware requirements, and their ability to
meet the requirements of water quality management agencies.
Based on this evaluation, a model water quality monitoring
program will be recommended for the State of Colorado.
-------
SECTION IV
GENERAL OBJECTIVES OF STATE WATER QUALITY
CONTROL PROGRAMS
The objectives of a state water quality program must be
determined from the state laws which establish the program.
These laws, however, have undergone a tremendous review
and revision during the past five years as a result of
the Federal Water Pollution Control Act, as amended. Since
this act has stimulated review and revision of state statu-
tes, it must have had a large influence upon current state
water quality control programs and, consequently, the ob-
jectives associated with the program. Because of this in-
fluence, the federal impact upon state programs will be
evaluated.
Federal Influence
The Federal Water Pollution Control Act, as amended, states
that its purpose is to "establish a national policy for the
prevention, control, and abatement of water pollution."
This seemingly broad statement is actually a very concise
statement of the objectives of water pollution control. That
is, the objective is simply to prevent and abate water pollu-
tion. How this objective is to be obtained is spelled out
in the Federal Water Pollution Control Act, as amended. Also
spelled out is the relationship between state water pollution
control programs and the federal program.
The objectives stated in the federal law are to be obtained
mainly through action at the state level. The federal law
specifies that a state will adopt "(A) water quality criteria
applicable to interstate waters or portions thereof within
such State, and (B) a plan for the implementation and enforce-
ment of the water quality criteria adopted, ..." This then
is an objective of the state water quality program-~to set
stream standards and enforce them. If the state does not
satisfy the above federal stated duties, then the federal gov-
ernment will establish and enforce water quality standards in
the state. This then is a directly stated legal influence
upon the state program objectives.
Besides the above stated influence, the federal government
offers the states money to help support the states' pollu-
tion control activities. This financial support is in the
form of (1) direct program support (Section 7 Grants), and
-------
(2) waste treatment plant construction grants. To obtain
the federal money, grant applications must be completed.
These applications require specific information which the
state program must supply. This, therefore, is another
objective of a state program; perform the necessary tasks
and collect the necessary information which will insure
success in obtaining federal funds.
In summary, the federal influence on state water pollution
control objectives can be considered to be in two broad
areas: (1) the legal requirement that each state establish
and enforce water quality standards, and (2) the more in-
direct requirement that each state must supply the infor-
mation required to obtain federal financial support for
the state program. Each of these broad influences will be
analyzed in more detail later in order to determine the
data requirements necessary to obtain the objectives de-
lineated through the influence.
State Legal Influence
In order to evaluate the objectives of a state water pollu-
tion control program, six sample states were chosen and
their laws analyzed in detail. The objectives were deter-
mined in a general form into which all state programs could
be fitted. This is not to say that each state agency will
contain all the elements discussed; it is only an attempt
to relate data needs to program objectives. This general
form will necessarily include the general objectives im-
posed by the federal government since the state laws re-
flect the effects of the Federal Water Pollution Control
Act.
Before spelling out the general objectives, it will be
constructive to look at the declaration of policy and the
statement of purpose which are contained in all the state
laws. Each state has a declaration of policy of sorts in
which it is stated that the policy of the state with regard
to water pollution is to maintain the waters of the state
in a reasonable state of purity which is consistent with
the various uses of the water such as public health, in-
dustrial development, etc. A typical example is South
Carolina's declaration,
It is declared to be the public policy of
the State to maintain reasonable standards of
purity of the air and water resources of the State,
consistent with the public health, safety and wel-
fare of its citizens, maximum employment, the indus-
trial development of the State, the propagation
8
-------
and protection of terrestrial and marine flora
and fauna, and the protection of physical pro-
perty and other resources.
These policy declarations indicate that water pollution con-
trol must be accomplished under the constraints that are im-
posed by the complex relationships of water to most activities
of the state. This is saying that water is vital to many
aspects of a state's activities and control of water pollu-
tion must, therefore, be considered in a systems context.
These policy declarations make no statements as to the rela-
tive importance of each effect. The importance is stated
later in the law and at that time the objectives related
to the activities in the state can be determined.
The statement of objective or purpose of the law can be sum-
marized in two words—abatement and prevention. This is
exactly the same as the federal law statement of purpose.
Abatement and prevention may not be the exact words, but
these words usually summarize the intent. New York's law has
a concise statement of purpose which is as follows.
It is the purpose of this article to safeguard
the waters of the state from pollution by: (a)
preventing any new pollution and (b) abating pollu-
tion existing when this chapter is enacted, . . .
Objectives
By reviewing the duties of the water pollution control pro-
grams contained in the state laws, the broad objectives,
prevention and abatement, can be stated in more specific
terms. In general, the seven basic duties or objectives are:
1. Planning
2. Research
3. Aid Programs
4. Technical Assistance
5. Regulation
6. Legal Enforcement
7. Data Collection, Processing, and Dissemination
Planning, research, and aid programs can generally be classed
as prevention, while technical assistance, regulation and
legal enforcement can generally be classed as abatement.
Data collection, processing, and dissemination is an objec-
tive in that it supports the first six activities. Figure
1 illustrates the type of structure envisioned by the fore1-
going discussion.
-------
Planning
Pollution
Prevention
'!•
1
Research
Aid
Programs
1
Technical
Assistance
Pollution
Abatement
I
1
Regulatior
1
Legal
Assistance
- Establish Priorities
on Construction Grants
- Comprehensive Water
Quality Planning
- Interagency Planning
Activities
- Contingency Planning
and Implementation
- Regional Plans for
Construction Grant
Applications
- Facility Planning
- Program Planning
- Approve Research Proposals
- Evaluate Progress of Spon-
sored Research
- Develop Research Needs
- Coordinate Research Acti-
Federal Agencies
- In-House Research Activities
• Construction Grant
Administration
State Program Grant
ftdmi n is tr ation
• planning Grant Pro-
gram Administration
• Training Grand
Administration
• Recommend Abatement
Steps for Polluters
- Contact St
Violators
- Approve Sewage Treat-
ment Plant Plans and
Spec
- Inspect New Construction
- Recommend Sewage Treat-
ment Plant Site Approval
- Training of Sewage
Treatment Plant Operators
- Inspect Waste Treatment
Plants
- Advise in Drafting
Regulations
- Expert Witnes
Regulations
- Conduct Hearings
- Prepare Briefs for
Court
- Recommend Further Internal - Initiate Court Action
Action for Persistent
Violators - Testify
- Levy and Collect
Pines
- Sewage Treatment
Plant site Approval
H
Info
Collection
Info
Processing
1
Action
initiation
- Surveillance Network
*Collection
*Analysis
•Transmit
- Field Investigations
*Collection
•Analysis
'Transmit
- Other Sources
•Encourage
"Receive
•Analyze
•Transmit
• Screen
- Verify
• Interpret
• Index
• Store
• Retrieve
- Action Need Reports
- Inventory and Data
Summary
- Special Reports
*Water Pollution
•Index Report
•Annual Report
Figure 1. Breakdown of the Objectives of a State Water Quality Management Program.
-------
Each of the seven objectives will now be discussed as to
their legal background, and a general sub-structuring of
each objective will be established in order to better re-
late the data needs of the objective. Again, the sub-
structuring is necessarily accomplished by idealizing a
state's water pollution control program and is in no way an
attempt to detail the activities of any one agency. Hope-
fully, each agency can see its various activities in the
idealized structure, but at the same time some states em-
phasize certain areas over others. Consequently, not all
states will be actively participating in all activities.
Planning
The duty of planning in a state water pollution control agen-
cy has received increased emphasis over the past few years
from both the state and federal levels. The state level em-
phasis has been on program planning while the federal level
emphasis has primarily been on project planning.
The state laws generally require the development or adoption
of a general comprehensive program (program planning) for
the prevention and abatement of water pollution and from time
to time a review and modification of the program as necessary,
Typical of the statements related to program planning is the
following from the clean streams law of Pennsylvania (as
amended, July 1970). The law states that,
The board shall have the power and its duty shall
be to: (2) Establish policies for effective
water quality control and water management in the
Commonwealth of Pennsylvania and coordinate and be
responsible for the development and implementation
of comprehensive public water supply, waste manage-
ment and other water quality plans.
The comprehensive program plan is concerned with transmitting
broad goals into accomplishable units of work. The program
plan permits efficient organizational staffing and budgeting
and development of short and long range work plans. By so
structuring a state water pollution control agency through a
program plan, the activities of the agency can be geared to
the problem at hand and hopefully little effort will be
wasted on irrelevant activities.
States have just recently embarked upon comprehensive pro-
ject planning. This has been in response to federal regu-
lations (18 CFR 601.32 and 33) which state that,
11
-------
(a) No construction grant shall be made unless
the Commissioner determines, based on information
the State, . . . , furnishes to him pursuant to
paragraph (b) of this section that a project is
included in an effective current basinwide plan
for pollution abatement consistent with applic-
able water quality standards.
(b) In reaching such determination, the Commissioner
may require information in such manner as he pre-
scribes concerning the total basin plan, or portion
thereof, as he deems adequate to evaluate the effec-
tiveness of the project.
In evaluting the project plans, the Commissioner shall con-
sider whether it adequately covers the following: (1) sources
of pollution, (2) volume of discharge, (3) character of ef-
fluent, (4) present treatment, (5) water quality effect, and
(6) detailed abatement program.
Besides the project planning required by the Environmental
Protection Agency (EPA/WQO), state agencies are also involved
in some planning activities associated with the Water Resources
Commission and with the Department of Housing and Urban De-
velopment in their coordination of sewerage system design
with overall urban planning. Project planning also'involves
evaluating the water quality effects of water resources pro-
jects. This is especially true in many western states where
diversions greatly influence water quality.
Now that state water pollution control agencies are involved
in both comprehensive program and project planning, the
planning function is demanding more attention in terms of
both money and manpower. ' Also, the large planning effort
requires large amounts of information and, consequently, is
stimulating a closer look at current data acquisition techni-
ques and practices. The influence of a strong planning ef-
fort spreads throughout the agency, and, therefore, illustrates
the importance of planning in the overall structure of a state
water pollution control agency.
Research
The duty indicated under this general objective is that the
state agency will conduct studies, investigations and re-
search with respect to pollution abatement or prevention.
The state agency will determine what the research needs are
and then either perform the studies with agency personnel
or contract with outside organizations (public or private).
The research objective also implies that the state agency
will coordinate water-quality-related research investigations
12
-------
that are performed by other state or federal agencies. The
Wisconsin law will serve as an example of how this duty is
stated. The Wisconsin law states,
The department may conduct scientific experi-
ments, investigations, waste treatment demon-
strations and research on any matter under its
jurisdiction.
An important aspect of research investigations not deline-
ated in the state laws is the need to follow through on the
research results and see that they are put to an effective
use. All too often, research results are received by the
practitioner and are not utilized by him due to his failure
to appreciate the findings. This wastes funds since the
research results are not being used to meet the agency's
obligations or objectives and it reflects poorly upon the
researcher and the state agency. Implementation of research
findings should be an important objective of the research
function within a state water pollution control program.
Aid Programs
The duty denoted under aid programs is that associated with
the accepting and supervising of loans and grants from the
federal government and other sources. This involves pro-
cessing applications and the general administration of the
loan or grant. The primary activity would be associated
with sewage treatment plant construction grants, but there
are also state water pollution control agency support grants,
assistance programs under the environmental education act,
manpower and training grants of the Environmental Protec-
tion Agency (EPA), and facilities grant programs of the De-
partment of Housing and Urban Development (HUD) and Federal
Housing Administration (FHA) associated with construction
of community sewage works. Also, a duty of aid programs is
the need for the state water pollution control agency to
certify the need for various other types of projects such
as some research grants form EPA and to coordinate water
pollution control planning grants obtained under the Com-
prehensive Water Pollution Control Planning Act, Section
3(c), PL 84-660, as amended and the Urban Planning Assis-
tance Program authorized by Section 701 of the Housing Act
of 1954, as amended.
The majority of the activities of this objective will in-
volve the federal government. The California law states
this objective in the following manner,
13
-------
The state board shall administer any program of
financial assistance for water quality control
which may be delegated to it by law, and may
accept funds from the United States or any per-
son to that end.
The importance of the aid program objective grows as the fund-
ing agencies demand more and more information with the appli-
cations. As an example, the grant application for sewage
treatment plant construction entails the compilation of many
statements and certifying letters in addition to the require-
ment of a project plan;and a review of information supplied
with grant applications indicates that the various sections
of the state agency must fully cooperate with the aid pro-
gram function in order for the grant application to be suc-
cessful.
Technical Assistance
The duty implied by this general objective is to advise, con-
sult, and cooperate on technical matters with other agencies
and political subdivisions of the state and federal govern-
ment and with private enterprises which are polluting the
streams. The activities needed to obtain this objective will
involve the recommendation of abatement steps for stream
standard violators, review of the technical aspects of waste
treatment plans and specifications, inspection of new con-
struction of waste treatment plants, recommendation of site
approvals for construction of waste treatment plants, in-
spection of existing waste treatment plants, prediction of
water quality effects from projects which change the physi-
cal system, and training of waste treatment plant operators.
In regard to this objective, the Colorado law states that it
is the duty of the state water pollution control agency
To advise, consult, cooperate, and enter into
agreements with other agencies of the state,
the federal government, other states, and inter-
state agencies, and with groups, political sub-
division, and industries affected by provisions
. . . (of this law) and the policies of the
commission.
The law also states that the agency shall have the duty,
Upon request, to examine and approve or dis-
approve plans and specifications for the
construction and operation by a political sub-
division of: New sewerage systems, disposal
14
-------
systems, and treatment works; and extensions,
modifications of, or additions to new or ex-
isting sewerage systems, disposal systems, or
treatment works. :
Regulation
The duty denoted by this general objective is that the
state water pollution control program is responsible for
regulating the quality of water in the state's streams and
rivers. This objective has traditionally been the most im-
portant with respect to the money and manpower devoted to
its accomplishment. To better understand the traditional
importance, it will be instructive to look at some of the
relationships that exist between stream standards and regu-
lation. The Colorado water pollution control program has
the following as one of its duties.
i
In order to develop the comprehensive pro-
gram for the prevention, abatement, and control
of the pollution of the waters of the state, the
division of administration is authorized to recom-
mend the grouping of such waters into classes in
accordance with their present and future most
beneficial uses in the interest of the public, . . .
This statement very clearly illustrates the general rela-
tionship that says in order to establish a control over
water quality, the waters must have certain standards of
purity. These water quality standards are to reflect the
quality of water necessary to meet the beneficial uses of
the water. The logical sequence of action, once the stand-
ards are established, is to regulate the stream's quality
so that it is maintained within the limits set by the stream
standards. This is where the general objective of regulation
comes into play. South Carolina's law states this in the
following manner as a duty of the water pollution control
authority.
Conduct investigations of conditions in the
air or waters of the State to determine whether
or not standards are being contravened and the
origin of materials which are causing the pollu-
ted conditions; . . . ,
This implies some sort of routine monitoring to insure that
the stream standards are being met. In summary, then,
water pollution control implies stream standards, stream
standards imply monitoring, and monitoring suggests that
15
-------
regulation will follow a detected violation of stream
standards.
The actual activity of regulation is tied very closely to
the concept of permits. Conceptually, regulation involves
the following steps.
1. Establishing stream classifications and criteria;
2. Listing pollution sources which violate criteria;
3. Establishing abatement requirements in terms
of performance requirements and an implementa-
tion schedule.
4. Establishing procedures to assure that future
actions (flow depletions or waste additions)
do not cause violations.
Steps 1 and 2 are implicit in the water quality standards as
discussed above. Step 3 is the point at which a permit is
issued or an abatement plan is implemented or both are used.
The purpose of the permit would be to set limits on waste
discharges at current levels while the purpose of the abate-
ment schedule would be to reduce waste discharges to a
level consistent with stream standards. Once this level is
reached, the permit may or may not be used. Step 4 is carried
out by requiring site approval for future waste treatment
plants, requiring the acquisition of a permit to discharge
wastes, etc.
If the situation arises where a polluter does not or will not
follow the abatement schedule, the regulatory function ends
and the legal enforcement function begins. Also, if new dis-
charges fail to follow the established procedures of Step 4,
legal enforcement may be required to insure compliance.
Legal Enforcement
The duties denoted by this general objective deal with the
enforcement of stream standards by legal means. If the stream
standards cannot be maintained by persuasion, the state water
pollution control agency must be able to resort to legal means.
This therefore must be an objective associated with prevention
and abatement of water pollution, Pennsylvania's law states
that one of its water pollution control board's duties is to
Formulate, adopt, promulgate and repeal such rules
and regulations and issue such orders as are
necessary to implement the provision of this act.
16
-------
Data Collection Processing, and Dissemination
The duty denoted by this general objective is mainly that of
a support role for the first six objectives. The successful
attainment of the first six objectives depends upon good
data so the first six objectives may be considered to imply
the seventh. However, it is usually spelled out in the laws
that data shall be collected. For example, the Colorado
law states:
The commission shall cause samples to be
taken from the waters of the state periodically
and in a logical geographical manner so as to
advise the commission of the water quality stand-
ard of the waters of the state.
Data collection involves sampling of a surveillance network,
performing special stream surveys, receiving operating re-
ports on waste treatment plants, development of an inventory
on the sources of pollution and the water uses, determination
of present treatment, evaluation of volume and character of
discharge, etc. The primary aim of this study is to evaluate
the data requirements placed on the surveillance network and
determine the optimum manner in which to obtain this data.
Therefore, the main consideration will be the surveillance
network data.
The first step is obtaining good water quality data is to
select a good representative sample. This should be followed
by proper sample preparation and handling and by a form of
laboratory procedure quality control. The quality control
function of obtaining data must also follow the information
through generation of reports by having thorough verification
of reported results.
Once a wator sample has been taken and analyzed, it must be
followed by transmission of the results to a data processing
group. Data processing will involve screening, verification,
interpretation, indexing, storage, and retrieval of the data.
Data dissemination will best be accomplished through the
generation of action need reports to. be sent to the appropri-
ate part or section of the agency. Data dissemination may
also include inventory and data summary reports and special
reports. It will be these reports that will keep the entire
agency moving toward the objective of prevention and abate-
ment of water pollution.
17
-------
SECTION V
DATA REQUIREMENTS OF GENERAL OBJECTIVES
The routine collection and analysis of water quality samples
is relatively straight-forward when compared to the problem
associated with determining in detail what the water quality
data is to be used for once it is collected. The establish-
ment of general objectives and work elements was an initial
step in the determination of data use within a state water
pollution control agency. Figure 1 is a summary of the
objectives and work elements.
Delineation of Data Requirements for Each Objective
The purpose of this section is to evaluate the work elements
and determine the type of water quality data needed to ac-
complish the activity efficiently. To accomplish this the
basic activities associated with the objectives will be de-
lineated and the information requirements of each activity
will be identified. From the list of information requirements,
that sub-set derived from the water quality surveillance net-
work will be identified. Rather than detail each objective
in this manner, two or three examples will be given and then
all the results will be summarized in a table at the end of
the section.
Planning
Project and program planning are similar in that a plan for
future action is being developed, but they are different in
the amount of detail required. Program planning deals with
broad definition of programs and work plans and, consequently,
does not need specific data on water qualtiy. General, over-
all trends would best serve the program planning function.
Project planning embodies specific planning related to de-
tailed evaluation of future water quality effects as a result
of waste treatment plant location, degree of treatment,
number of pollution sources, changing physical stream char-
acteristics, etc. Project planning will be broken down into
its various activities for purposes of data requirement ident-
ification. This is one of several examples that will serve
to illustrate the process.
19
-------
In order to identify the water quality data requirements of
the project planning function, the planning activities will
be structured into three phases, simulation, analysis and
evaluation. The activities under each of these phases will
be identified and the specific data needs of the activity
will be delineated. An outline format is used to illustrate
the activities and their data needs.
I. Simulation Phase
1. Simulate natural hydrologic regime
a) water quantity
b) water quality
2. Simulate economy
a) Identify water needs versus time
b) Identify water quality requirements
c) Identify waste discharge amounts
(unconstrained by pollution controls)
II. Analysis Phase
1. Simulate modified hydrologic regime
a) Superimpose water demands
b) Superimpose waste loads
2. Compare water requirements with modified
projection of hydrologic regime
a) water supply versus time
b) water quality versus time
III. Evaluation Phase
1. Formulate alternatives for solving problems
identified in II-2.
a) zoning
b) water control
c) pollution control
d) combinations
2. Evaluate alternatives in terms of a
benefit-cost ratio
Several data requirements are alluded to in the above outline,
but to get these into a specific format, each sub-phase of
the outline will have its data needs spelled out.
20
-------
(1-1) 1. historic data on water supply-water quality
2. analytic relations between water supply
and water quality
(1-2) 1. population data and projections
2. economic data and projections
3. data and projections on water use,
production processes, and wastewater
quality
4. effects of artificial changes on analytic
relations identified in 2.
(II-l) none
(II-2) 1. projections of water uses (other than
economic)
2. water quality requirements of various
water uses
3. water volume requirements of various uses
(III-l) 1. information on legal and institutional
authorities and limitations
2. information on water storage potential
of hydrologic area, etc.
3. information on pollution control techno-
logy
(III-2) 1. cost data
When the planning data requirements are looked at in the above
manner, several observations can be made. Fir si; the planning
function requires data from many sources and this data has
various degrees of quality. Population growth and economic
activity can only be projected in broad generalities; there-
fore, detailed hydrologic simulation models are rarely of
much value in long range projections. The sum of the data
for planning is only as strong as its weakest link. Second,
the information associated with uncertainty or risk is not
readily available. For example, how much risk is associated
with DO level fluctuations in terms of effects on fish? It is
very difficult to plan action to protect fish when the risks
are not clearly understood and the value of such action is
not clearly delineated.
The above information outline for planning indicates that
many types of water quality information are needed. Of these
many types, only the historic water quality data is to be
obtained from the water quality surveillance network of samp-
ling stations. The other types (sources of pollution, waste
characteristics, effluent volumes, etc.) will be obtained
through special inventories or surveys. The current practice
21
-------
of monthly operating reports from waste treatment plants
could be expanded to increase the supply of needed informa-
tion. The permit system, currently under study by many
states and the federal government, will supply much of the
needed information.
The use of data from sources other than the surveillance
network should not obligate the state agency to enter the
business of supplying this data. It is logically a duty
of the waste discharger to monitor his own operations just
as it is the duty of the state to monitor the water quality
of streams which belong to the public. This, then, should
leave the state agency to collect and analyze only those
samples associated with instream water quality. The other
data should be collected and stored at the state agency,
but the agency should not commit resources to the actual
sample taking and lab analysis.
The state must, however, undertake those quality control func-
tions necessary to assure the validity of the data used. This
implies state review of (or control over) practices used in
collecting, handling, and analyzing samples; and perhaps the
occasional splitting of samples or other procedures to veri-
fy the overall reliability of the sampling, analysis, and
reporting process.
Research
The identification and solution of water quality problems
through research demands water quality data of two broad
categories. When a state water pollution control agency,
through performance of its relevant pollution control activ-
ities, meets a situation where it is constrained by limited
response options, it must have a means of expanding these
options. This will involve research—both technical and
non-technical, depending on the situation—which hopefully
can study the problem and develop recommendations for ex-
panding the response options. Identification of the limited
response situation may occur in a number of ways, but the
one of concern here is the identification through analysis
of the surveillance network data. This then is one category
of water quality data needed by the research function.
Identification of limited response situations by other means
does not place a direct data demand upon the surveillance
network and, consequently, is of little concern in designing
the surveillance system.
The other water quality data demand category for research
is that associated with the actual performance of the
22
-------
research. This data is usually of a highly specialized na-
ture and is not generally collected by the surveillance
network.
The research function, therefore, mainly places a data demand
on the surveillance in the form of problem identification.
The final decision that research is needed comes from an
analysis of the options open to solve the water quality prob-
lem.
Aid Programs
The attainment of this objective is primarily administrative
in nature, but does require information inputs from other
parts of the agency such as the technical assistance section
concerning approval of waste treatment plans and specifica-
tions and the planning section concerning the section seven
grant application. Therefore, the aid program section is
not directly concerned with water quality data but is serving
as a means to transmit federal influence to the other sections
of the agency. The aid program section is primarily concerned
with keeping abreast of the federal programs and transmitting
their data requirements to the sections responsible for that
data.
To elaborate further, under Section 7 of the Federal Water
Pollution Control Act (33 U.S.C. 466 et seq) funds are author-
ized,
for grants to States and interstate agencies to
assist them in meeting the costs of establishing
, and maintaining adequate measures for the pre-
vention and control of water pollution, . . .
In order to obtain these funds the state must submit to the
federal government a plan for the prevention and control of
water pollution and have the plan approved. In order to con-
tinue receiving funds the state must complete and submit to
the federal government a "State Program Grant Application."
This application is submitted yearly.
Completion of the application requires information on all
aspects of the state water pollution control program. Under
"pollution control surveillance and enforcement," the appli-
cation requires information on water quality standards and
compliance, water quality monitoring, effluent controls,
inspection of treatment plants, certification of waste treat-
ment plant operators, and enforcement and regulation. This
information must be collected from the surveillance and
enforcement sections of the agency.
23
-------
The construction grant applications are more specific in de-
fining the water quality data needed to obtain the grant.
The waste treatment plant must be evaluated in terms of its
effect upon the water quality of the river. This requires
that a water quality data base be established from which such
projections can be made. Also, the construction grant appli-
cation must include information on the other sources of pollu-
tion on the streams and the quality of effluent from each.
The engineering data required by the construction application
requires that the stream standards be evaluated in terms of
the effect the project will have upon them.
The collection of water quality data for the purpose of grant
applications should not entail extra effort since much of the
required data can be obtained from the routine monitoring.
There may be a need for special surveys if the stream being
considered has not been monitored very often.
After completion of a waste treatment plant under the con-
struction grant program, the state must inspect the plant at
least annually for three years and then as often as specified
in the State Program Plan submitted under Section 7 of the
Federal Water Pollution Control Act.
Technical Assistance
The need for technical assistance also carries the need for
valid data upon which technical decisions can be made. Tech-
nical assistance must necessarily be involved in the following.
I. Waste Treatment Plant Construction
1. waste treatment plant site approvals
2. waste treatment plant plans and specifi-
cation approval
3. inspection of new construction
II. Waste Treatment Plant Operation
1. inspection of plant operation
2. training plant operators
III. Abatement Recommendations for Stream Standard
Violators
1. process modification recommendations
2. waste treatment plant construction recom-
mendations
3. legal testimony
24
-------
Major data requirements and their sources for the above duties
are as follows.
(I) Historical record necessary to compute
future water quality effects as related
to plant location and design criteria.
This implies determination of the stream's
condition with respect to effluent added
and treatment needed.
(II) Review of waste treatment plant operating
reports as related to stream water quality
indicated by suveillance network.
(Ill) Historical record of stream quality as
related to possible process modifications
or waste treatment alternatives.
The use of historical record to establish models of river or
stream water quality for the purpose of abatement procedure
planning is quite similar to the establishment of water qual-
ity models for the purpose of planning for water pollution
prevention. Both involve the prediction of future effects.
The main difference lies in the fact that planning for abate-
ment is concerned with a specific polluter on a specific
stream at a specific time. Planning for prevention concerns
a more general plan which must provide guidelines for future
action at all levels of pollution control activity and for all
streams in the state. The implications are that the models
developed for prevention planning could be refined to a more
specific nature and used for abatement planning in the area
of technical assistance. This may require more specific data
obtained by special surveys.
The maintenance of control over the operation of waste treat-
ment plants will require routine evaluation of both surveillance
data and operating reports from the waste treatment plants
themselves. Periodic inspection of plant operation will also
be necessary to insure the reported conditions are correct.
Routine evaluation of the surveillance network data is accomp-
lished through a regulation section, but stream standard
violations must be traced down by an analysis of the waste
treatment plant operating characteristics. This requires close
coordination between routine surveillance and operation of
waste treatment plants.
Regulation
Regulation entails the routine surveillance of water quality
to insure that the stream standards are being met. This has
25
-------
been and is to a large extent today the main reason for
collecting instream water quality. To accomplish this ob-
jective, the water quality data from the surveillance network
serves to:
I. Determine compliance and noncompliance
with water quality standards.
II. Evaluate progress of the abatement program
toward meeting standards.
III. Identify emerging water quality problem
areas so that prevention can be effected
rather than abatement at the last moment.
The type of data necessary to insure compliance is determined
by the stream standards in effect for the particular body of
water sampled. The quantity of data needed depends upon the
fluctuations in water quality present in the stream. A statis-
tical analysis, including modeling of stream quality fluctua-
tions, will determine the frequency of sampling required to
adequately determine the instream water quality.
In general, regulation data will be reviewed after the lab re-
sults are reported, and if compliance is indicated the data is
stored. Once stored, the data is, in effect, partly establish-
ing the historical record nedded to meet other objectives. If
noncompliance is noted, the data should immediately promote
abatement action within the agency. The data will again be
stored, but it may later be reviewed in terms of abatement
progress of a polluter. If the data indicate that stream qual-
ity is near noncompliance, this may promote prevention action
within the agency.
Basically, the data used for regulation is serving to put into
action those parts of the state agency needed to keep stream
quality within the set limits. Beyond this purpose, the data
has no direct use for meeting immediate regulation objectives.
The main problem associated with data for regulation purposes
is the tendency to let data collection become an end in it-
self and the neglect of emphasis on followup action. Hope-
fully, by keeping the general objective of regulation in per-
spective, its data requirements can be optimized and more
effort devoted to the other objectives of a state water pollu-
tion control agency. This optimization of data collection
for regulation purposes will necessarily involve the statis-
tical analysis mentioned earlier.
26
-------
Legal Enforcement
Water quality data needs of this section are mainly those
associated with proving stream standard violations in court.
The data used for this purpose can be better utilized by
channeling it through other sections of the agency such as
technical assistance or regulation. This provides for a
technical evaluation of the raw data and puts it in a more
relevant form for the purpose of legal action. The actual
data needs of legal enforcement may be of a more detailed
nature than that which routine monitoring would supply. This
may result in special surveys to verify that certain conditions
do exist beyond the shadow of a doubt.
Summary
In Figure 2, the work elements are listed under each objec-
tive and the water quality data requirements of each work
element are summarized using a broad classification system.
The water quality data needs are broken down into waste
discharge data and stream quality data. Each of these is
in turn evaluated as to volume, character, and variability.
The last column summarizes the data needs as related to a
routine monitoring network. This does not include effluent
monitoring, but the waste discharge data classification was
included to illustrate the total water quality data need.
This emphasizes the fact that not all water quality data
needed by the agency comes from a routine surveillance
network, i.e., stream quality monitoring.
By using Figure 2 as a summary of the previous discussion,
the overall water quality data needs of a state agency can
be visualized. The classification of data needs shown in
Figure 2 consists of four divisions: 1) Specific—specific
parameter data bits with high precision (±10% to 0%); 2}
General—trends, means, etc., of flow, dissolved oxygen,
temperature, TDS, etc., with a range of precision from
order of magnitude to ±10%; 3) Index—information that can
be used to estimate general data (basically this is quali-
tative information); 4) None—a general indication that no
direct input of routine water quality network data is needed.
This summarization technique will permit a general correla-
tion between data needed and the various data gathering
techniques.
27
-------
00
H-
^Q
c;
h
(D
tsj
Data Needed on
Waste Discharges
Vol Char Var
Prevention
A. Planning
- Program Planning
- Project Planning
- Planning Coordination
- Contingency Planning
B. Research
- Develop Research Needs S
- Approve Research Proposals
- Evaluate Progress of Sponsored
Research
- Research Coordination G
- Conduct In-house Research -
C. Aid Programs
- Grant and Loan Administration
- Coordination of Grants and Loans -
Abatement
A. Technical Assistance
- Approve WTP Plans & Specifi-
cations
- Inspect New Construction
- Recommend WTP Site Approval
- Train WTP Operators
- Inspect WTP Operations
- Recommend Abatement Steps for
Polluters
B. Regulation
- Issue Permits
- Approve WTP Sites
- Develop Abatement Schedules
for Polluters
- Review Stream Standards
Data Needed on
In-Stream Quality
Vol Char Var
Summarized
Surveillance
Network Data
Demand
G
G
G
I
G
S
G
I
G
G
G
I
G
S
G
I
G
G
G
I
G
G
G
I
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
I
S
S
S
I
S
S
S
I
S
S
S
General
General
General
Index
Specific
General
Specific
Specific
Specific
Specific
Index
Specific
Specific
Specific
-------
o
o
3
ft
H-
13
- Recommend Changes in Regu-
lation
- Initiate Further Internal
Action for Persistant Pollu-
ters
C. Legal Enforcement
- Issue Cease and Desist Orders
- Conduct Hearings
- Prepare Briefs for Court
- Initiate Court Action
- Prepare Laws and Regulations
- Levy and Collect Fines
Data Needed on
Waste Discharges
Vol Char Var
Data Needed on
In-Stream Quality
Vol Char Var
Summarized
Surveillance
Network Data
Demand
Specific
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
Specific
Specific
Specific
ro
<£>
-------
SECTION VI
STATE-OF-THE-ART SUMMARIES
Three major classifications of data acquisition techniques
are considered in this study. These are grab sampling,
automatic monitoring, and remote sensing. A state-of-the-
art paper has been prepared for each technique, and all
three are included in the Appendices. This section will
serve to present a short summary of each technique's state-
of-the-art. Each state-of-the-art paper has its references
presented at the end of the paper.
Development of a state-of-the-art paper is dependent upon
the published material available. For the established tech-
nique of grab sampling, the available literature is rather
small. This is primarily due to the assumed simplicity of
the sample collection. This assumption comes from a lack
of sophistication in grab sampling and, consequently, a
rendering of the technique less effective than it could be.
Hopefully, the results of this report will indicate the in-
creased effectiveness of a planned grab sampling surveil-
lance network.
Both automatic monitoring and remote sensing are developing
technologies, and there is a large amount of published ma-
terial related to the various development steps. Also new
and unique applications warrant published articles, thereby
increasing the volume of published material. The volume of
literature for automatic monitoring and remote sensing,
therefore, permits the development of more realistic state-
of-the-art papers. However, the volume of material also
requires decisions as to relevance which are sometimes
difficult to make.
Because of the rather small amount of literature dealing
specifically with grab sampling, many general articles that
relate to data collection in general (and, consequently,
grab sampling) have been reviewed in the grab sampling
paper. These articles could be included in all the papers,
but since the other two techniques have considerable lit-
erature available, it was decided not to repeat the reviews,
Grab Sampling Summary
Literature available on the design of routine grab samp-
ling water quality surveillance networks, as noted above,
is quite small. The available literature either speaks
31
-------
of surveillance system design in very general terms or it
dwells upon the actual grab sampling procedures. This
leaves the questions related to sampling station location,
number of stations, sampling frequency, and parameter mea-
surements unanswered.
The grab sample state-of-the-art paper in Appendix A first
reviews the papers which relate data acquisition to a spe-
cific function of a water quality management agency. This
includes reviews of papers which relate the data needs of
planning, legal proceedings, abatement, etc., to general^
suggestions for improving the ability of the data to satisfy
that need.
Next, papers are reviewed which attempt to identify those
parameters that should be measured by a grab sampling net-
work. This identification is the result of reviewing
stream standards, or the planning process. It also occurs
through the creation of an index. Some parameters may be
eliminated through correlation with a similar parameters.
Sampling frequency and number of stations are usually dis-
cussed at the same time. It seems to be generally agreed
that more samples from fewer stations is more desirable.
It is also agreed the sampling frequency depends upon the
variability of the constituent to be analyzed. The events,
therefore, that cause variability of a constituent are dis-
cussed as having an influence on the frequency of sampling.
There is little specific information on sampling station lo-
cations. Each situation presents a different set of condi-
tions, therefore, it is quite difficult to develop a set of
hard and fast rules.
Grab sampling procedures have received considerable atten-
tion in the literature; however, this literature is speci-
fically related to in-stream sampling, sample preservation,
sample storage, and sample analysis. Few articles appear
which attempt to relate all factors in an overall design for
a grab sampling surveillance system. Current developments
occurring in water quality laboratory analysis procedures
could have a profound effect upon the grab sampling method
of data collection. The cost of analyzing samples presents
a huge barrier to increasing sampling; therefore if the lab-
oratory analysis costs could be reduced (through sample anal-
ysis automation) this could greatly affect any cost-effec-
tiveness analysis which attempts to relate grab sampling to
other data acquisition techniques. For this reason articles
pertaining to grab sampling procedures are reviewed but not
with the depth given other aspects of grab sampling system
design. However, whenever a grab sampling system is created
there must be good system design and good sampling and anal-
ysis procedures if good, meaningful data is to be acquired.
32
-------
Cost of grab sampling is an elusive piece of information.
Almost no information appears in the literature. Estuary
sampling provides the designer of a surveillance system
with a different situation from that of streams. The mixing
of fresh and saline waters along with the complex hydrologic
and hydrodynamic characteristics of an estuary have been the
target of many papers on sampling.
Automatic Monitoring Summary
As noted earlier, much of the literature on automatic moni-
toring revolves around new technical developments or new
applications of existing equipment. The state-of-the-art
paper in Appendix B reviews these articles by classifying
them into applications at various levels of government and
in university research. For purposes of this summary,
Table 1 displays all the applications and their pertinent
facts in a very condensed form.
Beyond application oriented literature, there has been dis-
cussion of the ability of automatic monitors to satisfy the
data needs of its users. Automatic monitoring provides
high frequency sampling and real time data; however, it is
expensive and it is quite limited in the number of parameters
that can be measured reliably. A breakdown of the ability
of automatic monitoring data to satisfy various functions
within a water quality management agency depends upon the
function. For planning purposes, the parameter limitation
is quite restrictive. For abatement purposes the real time
data is extremely valuable.
The present instrumentation suffers from a lack of adequate
sensors, and there appears to have been little work in this
area. The actual recording and processing of the data ap-
pears to be the strongest link in the current automatic
water quality data acquisition system because it is fully
developed and relatively trouble free. The lack of adequate
sensors is due to the rough field conditions under which
they must operate (they were originally designed for pro-
cess control where an ideal environment can be maintained).
The sensor problem carries over into a maintenance problem
once the unit is in the field. Weekly service is highly
recommended for the units. Future needs point heavily to-
ward more sensors which are reliable.
Cost information on automatic monitors is available since
many systems are currently being sold, installed, and oper-
ated. Costs of each monitor run from $6,000 to $12,000
depending upon the technique of data recording and the sen-
sors involved. Computers can be added to accept the data
and control the monitor. Since the units require consider-
able maintenance, this is a critical cost factor.
33
-------
TABLE 1. Application of Automatic Monitors
U)
System Name
and/or
Agency Involved
uses
uses
EPA/WQO
EPA/ WOO
EPA/WQO
EPA/WQO
EPA/WQO &
Metro. St. Louis
Sanitation District
Bureau of
Reclamation
Ohio River Valley
Water Sanitation
Commission
(ORSANCO)
Where When
Employed Employed
Delaware 1955--
Estuary
Patuxent 19&3--
River
Estuary
Potomac
River &
Estuary
New York 1963--
HarboT
Hudson-
Delaware >
Basin
Oregan 1968--
Missis-
sippi River
in Missouri
C*Hf. Cen. '965--
Valley Pro-
ject
Ohio River 1960--
Valley
Type of Number
Water Of
Sampled Stations
Estuarine 8
Estuarine 1
Freshwater 4
&
Estuarine
Freshwater 5
Estuarine
Freshwater
&
Estuarine
Freshwater 3
Freshwater
Freshwater
fc
Eatu-arine
Freshwater 27
Type of
System
In situ
In situ
In situ
In situ
Mobile
Van
Mobile
Trailer
Mobile
Boat
Mobile
Trailer
In aitu
&
1 Mobile
Parameters
Measured
DO, pH.Cond,
T, Turb. S
Rad, Cl,
Stage, ORP
T.'DO, Cond,
Turb, Stage
DO, T, Cl
pH, Cond. T
DO Turb
GRP, S Rad
__
pH, Cond, DO,
T
DO, T, Cond,
PH
Cond, T. Turb,
DO, pH, S Rad
pH, T, Cond,
DO, Cl, ORP,
S Rad
Recording
Technique
Strip chart
w/o tele-
metry &•
ppt.
Strip chart
w/o tele-
metry
Telemetry
ppt at the
metry/twls
fa ppt
__
Strip chart
w/o tele-
metry
--
Strip chart
w/o tele-
metry
Tele/ppt
& Computer
control &
proces sing
Purpose
Suppl. grab sa.
program. De-
tect rapid flucu-
ations in w. q.
Suppl. grab sa.
program. Rapid
flucuations &
Diurnal variations
Provide real
time continuous
records of w, q.
Historical, con-
of water quality
Collect w, q.
data in remote
areas
Determine ex-
tent of pollution
in 100 mi. stretch
of Mississippi R.
Collect w. q.
data in remote
mine flucuations
in salinity
Suppl. grab sa.
program with
realtime system
Data Use
Regulation;
Water fc
Waste Treat-
ment control
Research;
Effects of
thermalpol-
lution on. bio-
logical life
of estuary
Planning:
mathemati-
cal models
Planning;
cal models
Regulation;
Water &
Waste treat-
ment control
Regulation;
Water quality
quality .
Planning;
New
re?e rvoi r(s)
Regulation;
Evaluate pol -
hitiun control
efforts
Reference
Source( s)
McCartney and
Eeamer, 1 962
Keyser, 1 964
Cory and Davis
1965. Nauman
and Cory, 1970
McDermott,
Ballinger and
Sayers, 1968
Brornberg and
Cararnes. 1 9 70
Dewling, 1«69
O'neal, 1971
Anon., 19o9
Anon., 1Q67
Marks. '966
Cleary. 1962
Klein et al. ,
1968
-------
TABLE 1. Continued
System Name
and/ or
Agency Involved
Interstate Com-
mission for the
Delaware River
Basin (1NCODEL)
'Empire State
System
New Jersey
State Dept. of
Environ Pro-
tection tf USGS
Pennsylvania
Dept. of Health
Wisconsin Dept.
of Natural
Resources
Wisconsin Dept.
of Natural
Resources
Dept. of
Ecology State
of Washington
Texas Water
Pollution Con-
trol Board
Where When
Employed Employed
Delaware 1959--
River
Basin
New York 1966--
State
New Jersey. I968--
Pennsylvania 1965--
Wisconsin 1968--
Wisconsin 1970- -
Washington 1969--
State
Galveston 1963--
Bay
Type of Number Type of
Water of
Sampled Stations System
Freshwater 12 In situ
&
Estuarine
Freshwater 12 In situ
Freshwater 10 In situ
Freshwater 2 In situ
Freshwater -- Mobile
Trailer
Freshwater 11 In situ
Freshwater -- Mobile
Trailer
Estuarine -- Mobile
Boat
Parameters
Measured
T, pH, Cond,
DO, Turb
pH, Cond, DO,
T, Turb, Stage
Cl, F, S Rad.
Cond, T, pH,
Turb, DO
pH, Cond, T
T, DO, pH.
Cl, ORP,
Cond, S Rad
DO, T, pH,
Turb
DO, T, Cond,
Turb, pH,
Cl
DO, Cond,
pH, T
Recording
Technique
Strip chart
w/o tele-
metry
Telemetry/
Computer
Control 8t
Processing
ppt w/o
telemetry
ppt 8,
printer w/o
telemetry
Strip chart
w/o tele.
metry
Telemetry/
Computer
Processing fe
control
--
Purpose
Suppl. grab sa.
program
Suppl. grab aa.
program & pro-
vide rapid intelli-
gence system to
protect state
waters
Suppl. grab sa.
program
Supply data on
Acid mine
drainage
Suppl. grab sa.
program. Special
surveys, diurnal
variations
Collect Historical
data
Determine affects
of Indus. V Munic.
pollution of w. q.
& fish
Data Use
Planning:
Mathemati-
cal models
Regulation:
Water &
Waste treat-
ment control
Planning:
Mathemati-
cal Models
Plan/ling;
Trends in
w.q.
--
Regulation;
Formulate
stds. Locate
pollution
source
Research;
--
Planning:
Water
pollution
manage-
ment
Reference
Source(s)
Parker, 1961
Smith and
Morris, 1969
Maylath, 1970a
Maylath, 19706
Anderson et.
al., 1970
Mentink, 1970
Anon., 1968
Schaufnagel,
1971
Anon., 1968
Schaufnagel,
1971
Palko, 1971
Davis, 19fe6
-------
TABLE I. Continued
a\
System Name Where
and/ or
Agency Involved Employed
Greater Chicago's Chicago,
Metropolitan Sanita- 111.
tion District
Detroit Sewer Detroit,
Monitoring fit Mich.
Remote Con-
trol System
University of Chapel
North Carolina Hill, N.C.
Georgia Institute Georgia
of Technology
Ontario Water Lake Erie
Resources
Commission
When Type of Number Type of
Water of
Employed Sampled Stations System
1968- - Effuent 11 In situ
ic
Freshwater
Collection -- In situ
system
I960-- Freshwater I Sub-
Reservoir mers-
ible
Freshwater -- Mobile
Van
1969-- Freshwater 2 Im-
mersion
Parameters
Measured
DO, T, Cond,
pH, ORP, Cl,
Turb, S Rad
_.
T, DO,
DO, T, Cond.
Turb, Stage,
Depth, S Rad
pH, T, Cond,
Turb, Depth,
Current (vel-
ocity & direc-
tion)
Recording
Technique
Strip chart
ppt & twls
__
Telemetry/
strip chart
& ppt
Strip chart
w/ o tele-
metry-
Magnetic
tape
Purpose
Obtain a better
appraisal of
w. q. conditions
Better under-
stand behavior
of sewer &
drainage systems
Study material
transfer through
metalimnion in
impoundment
reservoirs
Study the
affects of storms
on w. q.
Determine the
dispersion
characteristics
for near shore
areas of Great
Lakes
Data Use
Regulation:
Water 8,
Waste treat-
ment control
Regulation:
water &
waste treat-
ment control
Regulation:
Water
Impoundments
Research;
Relation
between water
flow &
quality
Planning;
Mathemati-
cal Models
Reference
Source(s)
Lanyonand
Kurland, 1971
Sahre, 1970
Weiss and
Oglesby, 1963
Ingols. !970
Palmer and
Izatt, 1970
Palmer, 1970
N.C. = North Carolina
II!. - Illinois
Mich. = Michigan
EPA/WQO = Environmental Protection
Agency/Water Quality Office
USGS = United States Geological Survey
Metro = Metropolitan
# = number
ppt = punched papertape
w, q. = water quality
Abbreviations
twls -• typewritten log sheet
«td« = «tandard<"
Suppl. = Supplement
sa. = sampling
R = rivers
mi = miles
w/.o = without
indus = industial
munic = municipal
Calif. = California
Cen = Central
r»H = negative lov of hydropen ion activity
DO = dissolved oxygen
T = temperature
Cond = Conductivity
ORP = Oxidation reduction potential
Cl = Chlorine (Dissolved Chloride)
Turb = turbidity
S Rad = Solar Radiation
F = Fluoride
-------
Remote Sensing Summary
Aerial remote sensing data acquisition techniques add to
but do not replace, existing field and laboratory analy-
sis methods. Laboratory methods, which depend on the use
of test tubes, water-sample bottles, and microscopes, are
made even more valuable. Critical areas that require
field work can be located more rapidly and accurately.
Both time and money can be saved by reducing field work,
leaving more of both for confirming laboratory analyses
and, more important, ,for coordinating and taking correc-
tive actions.
The available remote sensing techniques can generally be
classified into four major categories: 1) visible range,
2) thermal infrared, 3) microwave, and 4) multispectral
sensors. Color aerial photography (visible range) holds
high present potential for actual identification of kinds,
concentration, and distribution of pollutants. Further-
more, color aerial systems are the least expensive and
most versatile of all aerial remote systems to use. Un-
fortunately, interpretation has been historically more of
an art than quantitative measurement. It has been demon-
strated that concentration distributions of pollutants
can be derived from color aerial photos if appropriate
ground correlations are made. The success of aerial photo-
graphic systems is dependent upon two considerations:
1. Do pollutants and water quality parameters
affect light sufficiently to extract de-
sired information; and
2. Can remote sensing techniques be combined
with analytic procedures to extract avail-
able information.
Many investigators have used aerial thermal infrared systems
to very accurately determine the surface temperature of
water bodies. Temperature perhaps is the most quantita-
tive parameter presently capable of measurement with remote
sensors.
The great value of data collected with thermal systems is
its conjunctive use with other methods of aeiral surveil-
lance, such as color photography and coordination with
ground data.
Remote sensing apparatus, using the microwave region, has
successfully been applied in accurately measuring surface
water temperature through clouds and precipitation. The
37
-------
microwave region has been used to measure the extent and
thickness of oil slicks with respectable success. However,
care must be exercised when this technique is used, be-
cause weather conditions and sea state can render oil
slicks invisible.
Multispectral remote sensing systems hold the greatest po-
tential for actual distinguishing and determination of pol-
lutants. As is shown and discussed in Appendix C, each
pollutant differs spectrally from other pollutants at one
or more corresponding wavelengths. The procedure then is
to choose the interval, or intervals, at which the greatest
differences exist. By comparing spectral signature data
catalogs and ground data, various kinds and concentrations
of pollutants may be identified with reasonable certainty
and accuracy.
The cost of remote sensing equipment is quite high. Actual
purchase of equipment with a reasonable degree of sophisti-
cation quickly runs into a few hundred thousand dollars.
If larger aircraft are employed with highly sophisticated
electronic systems, costs may easily run into millions of
dollars. For state water pollution control budgets a con-
tract basis appears to be the only feasible means to apply
remote sensing. There are no reported applications of
remote sensing by a state agency on any routine basis.
38
-------
SECTION VII
DESIGN PROCEDURE FOR WATER QUALITY SURVEILLANCE SYSTEMS
In developing a design procedure which can be applied under
different conditions in different states, considerable flexi-
bility must be incorporated into the procedure. This flexibi-
lity must take into account the different program objectives,
the various capabilities of different agencies and the influ-
ence of interrelationships between such factors as number of
sampling stations and frequency of sampling. To contain the
desired flexibility, the design procedure must permit the
input of descriptive parameters so that each agency can desig-
nate its priorities, strengths and limitations.
Pollution Control Strategies
The strategy a state water pollution control agency uses to
control water quality has a very large influence upon the
type of data required for the agency to function properly.
For example, a state with little resources may tend to be
crisis oriented, waiting for the pollution crisis to arise
before acting. This, in effect, would require no routine
monitoring since the crisis would identify itself by the fact
that it is a crisis. On the other hand, a state may choose
to concentrate on planning to solve its water quality prob-
lems—implying the end justifies the means. This would
indicate a data need of general water quality trends. Or a
state agency may be interested in only enforcing its stream
standards. If the stream quality is to be the primary
source of information on regulation, then a strong surveil-
lance program is indicated. But if the stream quality is
only to be an indicator while effluent control is to be the
primary aim of regulation, then the surveillance network need
only indicate trends.
The above examples may indicate the extremes, but at the same
time they do illustrate the fact that the strategy a state
chooses will largely determine the general form of its sur-
veillance network. Therefore, for a set of design guidelines
to be developed which can be used under all strategies, a way
of classifying a state's strategy must be delineated.
The general policies of pollution control, prevention and
abatement yield a natural division of a state agency's control
efforts. Looking at Figure 2, we see that a prevention stra-
tegy largely needs general data which implies one type of
39
-------
surveillance system, while an abatement strategy needs
specific data which denotes a more intensive surveillance
effort than that for planning. Within prevention, an agency
may choose to emphasize planning over research, thereby elim-
inating any specific routine surveillance data needs for the
prevention effort. Under abatement, an agency may use permits
as a means of regulating water quality, thus putting the main
responsibility for information reporting upon the polluter.
This strategy would require only general indications on stream
quality to insure that the strategy is working. The surveil-
lance system would not be used as a means of catching polluters
as it would be if the strategy was only to approve waste treat-
ment plant sites as a means of regulating water quality. Under
this strategy it would be the agency's responsibility to prove
the pollution event can be traced to a certain outfall. This
would demand a comprehensive surveillance network.
Using the above line of reasoning, a state agency can use
Figure 2 as a means of determining where it is currently
putting its main emphasis and once this is known, the agency
can use the appropriate guidelines in the development of a
surveillance network. Strategy determination can be accomp-
lished by either of two methods. The first was would be to
look at the various program elements in an agency's budget
and determine what percentage of its funds go to each objec-
tive. If it is not difficult to relate budget items to the
objectives, this method of strategy determination is the more
simple and may be more accurate.
The second method that is proposed to determine the strategy
of an agency is to establish a chart containing the objectives
and their associated work elements. The object of the chart
is to allow agency personnel to check the appropriate level
of activity under each work element (see Figure 3). Each
level of activity has a number of points associated with it
and this number is to be placed in the point column. The
points are totaled for each objective and then for prevention,
abatement and program control. The total possible points for
each objective level is the same; consequently, the points
can be used to compare the involvement of the state agency in
each objective. Likewise, the totals for prevention, abate-
ment, and program control are comparable. If an agency had
the same level of activity in all areas, the point totals
would be identical. Obviously, very few states, if any, would
have identical point totals; therefore, the state agency can
evaluate from the varying point totals whether it is putting
the emphasis on the function it intended to. This may reveal
areas where increased or decreased emphasis should be placed.
It makes no difference which strategy determination method is
used. The main point is that an agency's strategy has a large
40
-------
Activity Level
Prevention
A. Planning
- Program Planning
- Project Planning
- Planning Coordination
- Contingency Planning
High Med Low None Points
30 20 10
B. Research
- Develop Research Needs
- Approve Research Proposals
- Evaluate Progress of Sponsored
Research
- Research Coordination
- Conduct In-House Research
24 16
C. Aid Programs
- Grant and Loan Administration
- Coordination of Grants and
Loans
60
40
20
Total for Prevention
Abatement
Technical Assistance
Approve WTP Plans and Speci-
fications
Inspect New Construction
Recommend WTP Site Approval
Train WTP operators
Inspect WTP Operations
Recommend Abatement Steps
for Polluters
20
13
0
B. Regulation
- Issue Permits
- Approve WTP Sites
- Develop Abatement Schedules
for Polluters
- Review Stream Standards
- Recommend Changes in Regula-
tions
- Initiate Further Internal
Action for Persistent
Polluters
20
13
Figure 3 . Strategy Evaluation for State Water Quality Management
Programs.
41
-------
Activity Level
HighMedLowNonePoints
Legal Enforcement 20 13 7 0
Issue Cease and Desist Orders
Conduct Hearings ~~~
Prepare Briefs for Court
Initiate Court Action
Prepare Laws and Regulations
Levy and Collect Fines ~
Total for Abatement
Program Control
A. Information Collection 40 27 13
- Surveillance Network
- Field Investigations
- Other Sources
B. Information Processing 20 13
- Screen
- Verify ~ ~
- Interpret
- Index
- Store ~ ~
- Retrieve
C. Action Initiation 30 20 10 0
- Action Need Reports
- Inventory and Data Summaries ~ ~
- Special Reports
- Publicity
Total for Program Control
Figure 3. (Continued)
42
-------
influence upon the design of the appropriate surveillance
system and, consequently, the strategy must be known.
Stream Characterization
In the logical development of a surveillance system, there
must be consideration given to two major elements: 1) the
strategy of the agency and 2) the character of the stream.
Characterization of a stream will basically involve a deter-
mination of water quality patterns along the stream. This
will necessarily involve the utilization of many types and
large quantities of information. There must be an identifi-
cation of the more important parameters and the variation
of these parameters over the length of the stream. Since
flow has a large influence on quality, this parameter should
definitely be included. Ideally, there should be a complete
inventory of the stream's water uses and of its waste outfalls
and a survey of the natural or uncontrollable pollution
sources. This would provide a basis for determining the
critical pollution parameters for that stream, and it would
help in relating pollution variation to flow patterns. The
word "ideally" was used because unfortunately, inventories
of water uses, waste outfalls, and natural conditions are
not readily available in all states. Hopefully, as the
emphasis on effluent controls spreads, the needed inventories
will be developed. The use of a permit system appears to
be an excellent way to obtain the needed information with a
minimum amount of effort on the part of a state agency.
The water quality data to be used in the characterizations
can be obtained from many sources. Some of these are the
state water pollution control agency's data books, water
treatment plant records, local health department files, or
special surveys performed as a comprehensive analysis of a
particular stream's situation. A statistical analysis of the
data can provide information on the variation of the water
quality. Flow data may or may not be available in an agency's
records and, consequently, other sources of this information
may be required. U.S. Geological Survey (USGS) records may
provide the flow information needed.
The information needed for the inventory of waste outfalls,
if not readily available, may be obtained through use of an
aerial survey, a review of industries listed in a telephone
book or business directory, or by an actual walking of the
river. Also, special stream surveys usually supply some of
this information.
43
-------
Plotting the above information on a river-mile graph permits
rapid visualization of the existing water quality situation.
Plotting may only be necessary on the more complex streams,
whereas a simple review of the data may suffice on shorter,
less developed streams. The information now available, along
with stream standards, can provide a good indication of where
the sampling stations should be located and it permits gen-
eral determination of sampling frequencies. In other words,
sampling frequencies need to be higher near high pollution
centers. Also, at this point the strategy of the agency
must be considered. A strategy requiring only trend data
may dictate low relativeity frequencies everywhere.
Relation of Pollution Control Strategy
to Data Acquisition Strategy
Before any detailed design of a surveillance system can take
place, the strategy of the agency must be related to the
general data acquisition approach. The assumption is made,
however, that every state agency will monitor a very "skele-
ton" network in order to justify the organization's existence
in terms of budget requests versus accomplishments. The
skeleton network could consist of whatever monitoring level
the agency feels is needed to verify its success or failure
in controlling a state's water quality. In general the
skeleton network would consist of sampling stations at state
lines on the major streams and stations below major popula-
tion centers in order to insure stream standards are being
met. Location of these skeleton stations would not require
the stream characterization.
Data acquisition beyond the skeleton network will depend upon
the agencies' pollution control strategy as depicted in
Figure 4. An emphasis on prevention would indicate a need
for general data that may be collected by either routine moni-
toring, special surveys, or both. An emphasis on program
planning over project planning would indicate a need to em-
phasize routine monitoring over special surveys. This re-
sults from the fact that program planning uses trend data to
a greater extent than project planning.
Abatement.has basically three data acquisition strategies.
If a state emphasizes permits as its abatement tool, the
waste treatment plant operating reports can be the prime
source of water quality data. The other means of data ac-
quisition are special surveys and routine monitoring. This
is not to say that these three strategies are all the data
acquisition means available. A state could support a public
report system whereby a citizen reported any water pollution
44
-------
Sampling Points at
State Lines and Major Pollution Centers
(grab sampling or automatic monitoring)
Prevention
Abatement
Routine
Data Input
Periodic
Data Input
Permits
Ul
1
Sampling Points
at Mouths of Major
Streams and Summary
Points Along Major
Rivers (grab samp-
ling)
I
Special Surveys
(all means)
*
Monthly Waste
Treatment Reports
Routine
Stream Monitoring
Sampling Points Be-
low Cities and In-
dustries (all means)
Periodic Data
Input
Special Surveys
(all means)
Figure 4. Pollution control strategy associated with data acquisition strategy.
-------
event he witnessed. However, this would be a strategy sup-
plemental to the basic data acquisition system.
The foregoing discussion was designed to emphasize the fact
that routine monitoring is only one way to input water qual-
ity data to a water pollution control agency. So before a
state jumps into the design of a routine monitoring surveil-
lance system, it must first decide that it has chosen the
right data acquisition strategy for its pollution control
strategy. The choice of one strategy over the other is dif-
ficult to make since there are advantages and disadvantages
with each one. Currently the arguments are centering around
permits at both the Federal and state levels. It is not the
purpose of this project to enter into the various controver-
sies, but to design a routine surveillance system. At the
same time, however, it is realized that the final total in-
formation acquisition program will probably be a combination
of various information acquisition techniques.
Surveillance System Design
The design of a surveillance network is dependent upon whether
the network is to be used primarily for prevention or abate-
ment (see Figure 4). This difference will be reflected in
the sample station location, sampling frequency and sampling
techniques. In order to make valid comparisons on effective-
ness and cost, a grab sampling network will be designed first
and then automatic monitors and remote sensing will be evaluted
as a way to provide the same information at reduced costs.
In designing the grab sampling network, stream's characteri-
zation (quality parameters of consequence, flow, waste outfall
inventories and stream standards applicable to the stream) can
be used to determine a set of sampling locations consistent
with the state strategy. This establishes a basis upon which
to analyze the sampling frequency. The important aspect of
frequency is the cost involved; therefore, there needs to be
a way for a surveillance system manager to compare cost against
effectiveness. Under stream simulation, a way to relate fre-
quency to effectiveness will be developed and this can then
be related to cost. The manager can then make rational de-
cisions based on cost-effectiveness rather than on pure logic.
Stream Characterization Procedures
To design a monitoring system for a stream, the individual
characteristics of the stream must first be known. This will
permit determination of the sample station location and the
46
-------
parameters which need to be measured. The following is a
list of the activities needed to characterize a stream.
I. Collection of existing water quality and
quantity data for the stream
II. Plotting the flow and quality values on a
river-mile graph
III. Indicate location on the graph of the waste
outfalls and determine the waste load imposed
on the stream (volume and contents). This
will require a waste outfall inventory
IV. Indicate on the graph the stream standards
in effect for various reaches of the stream
V. Locate sampling stations at indicated prob-
lem areas according to state strategy
1. Prevention—place sample station location
at the critical quality point along the
stream. The routine monitoring data will
then give a general indication of the
stream's overall quality. Parameters to
be measured should be general data asso-
ciated with the agencies' planning activities
2. Abatement—place sample station location
at the critical quality point below each
major area of pollution sources. Measure
those parameters denoted in the stream
standards and any other parameters which
are critical according to the waste out-
fall inventory.
The activities listed above do not consider a combination of
a prevention and abatement objective. Data generated from
an abatement surveillance network could be used for planning
purposes, but the opposite is not tyrue. Because of this,
it may be desirable to establish two or three networks, each
with a specific objective.
With the stream characterization complete, the sample station
location and parameters of consequence have been determined.
The next step is to determine the sampling frequency which
yields the most information at the least cost. It must be
noted at this point that some sampling frequencies are es-
tablished by law and therefore cannot be changed without
changes in the law. Hopefully, this would not be too diffi-
cult if it could be proven that another frequency is better.
47
-------
Surveillance Network Simulation
In order to logically analyze the benefits associated with
various levels of sampling frequency, some method for making
comparisons must be developed. For this purpose, a relatively
simple surveillance network simulation model has been estab-
lished by Vanderholm (1972). This model has two basic parts.
One is to be used in an analysis of the ability of a monitor-
ing network to detect pollution events (abatement oriented),
while the second part is designed to evaluate a network's
ability to detect long-term trends (prevention oriented).
The mathematical models used in the surveillance network
simulation were selected because of their characteristic
of allowing a large number of trials to be made simulating
various conditions using high speed computers. A field
study of surveillance systems would require the installation
of the necessary facilities or the observation of existing
ones. To evaluate many systems under many conditions would
be too costly and time consuming for a study of this nature.
When using models, however, many systems under many condi-
tions can be evaluated at a fraction of the cost and time
involved with field studies. Models also require that much
care be taken to insure the validity of the simulation for
the situations under consideration. For this particular
study, the models are developed to predict the performance
of water quality surveillance networks for various water
quality conditions. This implies that the study is concerned
with modeling water quality surveillance systems, not water
quality alone. The data obtained from the models is intended
for use in design and evaluation of actual surveillance systems,
The first part of the surveillance network model was developed
to study the effect of sampling location and frequency on
the ability of a monitoring system to detect significant
short-term quality variations or pollution events. The terms
"pollution event" and "spill" are used synonomously in this
study and refer to any short-term quality variation or ex-
treme value from any cause which may indicate a stream stand-
ard violation. The model generates a series of these events
at random times and locations on the stream reach under study.
Downstream movement and dispersion of the pollutant is cal-
culated. Various combinations of sampling times and locations
are read into the program for testing, and if sampling and
spill coincide at a certain point in time and space, detection
of the spill is assumed. By testing the various sampling
combinations with a large number of random spills, estimates
of the sampling effectiveness can be made.
To study base level type data acquisition by grab sampling,
both a statistical and model approach were used. In this
48
-------
second part of the surveillance network model, the objective
is not to detect extremes but rather to obtain representa-
tive mean values. Statistical theory contains methods for
estimating the number of samples necessary to predict a mean
within a given range of the true mean for a known confidence
level. This method requires only that some estimate of the
variability of the parameter under consideration is avail-
able. By then specifying the allowable error, the number of
samples necessary can be estimated. For the time period in
question, number of samples is converted to a sampling fre-
quency at the specified sampling stations.
The base level model was established in the following manner.
The model generates a set of water quality data for a hypo-
thetical parameter over a long time period. The character-
istics of the generated data,- such as mean and standard de-
viation are known. Various sampling schedules are tested by
inputing them to the program and obtaining a set of observed
data points from the generated data. An observed data point
is that value occurring in the generated data at the point
in time a sample is specified. The characteristics of the
observed data can then be compared to the characteristics
of the generated data. The model and the statistical pre-
sentation provide a means whereby the performance of a sur-
veillance system in acquisition of base level type data can
be evaluated.
The first part of the model which is concerned with pollu-
tion event detection will now be discussed. The effective-*-
ness of a surveillance system in detecting a pollution event
or stream quality standard violation is of primary impor-
tance for the regulation function. The state-of-the-art
paper on grab sampling did not reveal any information or
methods to quantify this characteristic for grab sampling,
or even that any attempts had been made to determine the
level of effectiveness. Since one of the objectives of this
study is to evaluate grab sampling methods for regulation
data needs, it was necessary to devise a method whereby a
numerical estimate of detection levels could be made.
The method developed involves the use of a simple mathemat-
ical stream model. A program was written for the CDC§400 di-
gital computer. A simplified flow diagram for the program is
shown in Figure 5, while a complete listing of the program is
found in the work by Vanderholm (1972). The basic program
inputs are the geometric and hydraulic characteristics of the
stream. For the initial analysis, a hypothetical stream was
assumed with the following characteristics: average width,
200 feet; average depth, 1.5 feet; average velocity, 2 fps;
average flow rate, 600 cfs; Manning "n" value, 0.025. These
values are similar to those that can be expected on the lower
reaches of the South Platte River in Colorado. Time of travel
49
-------
Input Stream
Characteristics
Input Sampling
Schedule
Generate Pollution
Events
Compute Movement
and Dispersion
of Polutant
Test Sampling Schedule
for Event Detection
Output Detection
Values
Figure 5. Flow diagram for suveillance network simulation,
50
-------
and dispersion of pollutants from a specific pollution event
are calculated using these values. By specifying these
values, the assumptions of uniform channel cross section and
steady uniform flow are made. The length of the stream reach
used was varied, depending upon the particular situation
under study.
The arbitrary selection and use of a single set of values
for stream characteristics is obviously not in accordance
with an actual situation. However, sensitivity tests were
performed on the model to check the effects of varying these
values. These are described in the application section.
Location and time of pollution events are assumed to be com-
pletely random occurrences. A uniform random number gener-
ating function is used to select each of these parameters.
For all of the analyses reported here, the location was a
uniform random point anywhere along the entire reach under
study. A base time period of 30 days was selected, while the
time of pollution event occurrence was allowed to vary within
the range of 0 to 30 days. Another pollution event charac-
teristic considered was the length of time discharge occurs,
or the spill duration. Since this may vary from an instantane-
ous spill to a continuous discharge over a long time period
under actual conditions, the event time was studied as a in-
dependent variable, either specified or random within a
specified range.
A method developed by Glover (1964) for the USGS was used to
compute the longitudinal dispersion curves for pollutants
during travel downstream. This method of solution requires
the stream geometry and hydraulic characteristics described
earlier.
Equation 1 is the form developed by Glover for calculating
longitudinal dispersion.
Qs W 4Kxt
S = __-_
This is also the form used in the model where S is pollutant
concentration in Ib per 1000 cubic feet of water, Qs is size
of spill in Ib, B is average stream width in feet, D is average
stream depth in feet? x is distance from spill point to samp-
ling point in feet, v is average stream velocity in feet per
second, t is elapsed time since spill in seconds, and KX is
shear velocity in square feet per second. A detailed
explanation on the use of this equation is found in the refer-
ence just cited.
51
-------
The final dispersion curve for the sampling point along the
stream reach in question is calculated by substituting values
for t into Equation 1 and solving. The initial t value must
be specified to occur just before the leading edge of the
dispersion curve reaches the sampling point. By then incre-
menting t in finite steps until the significant portion of
the pollutant has passed the sampling station, a curve such
as shown in Figure 6 is obtained. Glover's formula calculates
the longitudinal dispersion curve for a finite amount of con-
servative pollutant released instantaneously from a point
BC urce. A conservative pollutant can be described as essen-
tially inert, such that no reactions occur which significantly
add, remove, or change the form of the pollutant within the
time period considered. Complete lateral and vertical mixing
was assumed in all cases. If the distance from the spill to
sampling point is very small, this assumption is not valid.
In addition, this situation does not allow time for development
of a dispersion curve from an instantaneous spill. The pro-
gram contains provisions to compensate for this possibility.
The calculated dispersion curve is in terms of pollutant con-
centration versus time at a given location along the reach.
The initial quantity of pollutant introduced must be specified.
A value of 1000 pounds of pollutant was used in all cases
except when sensitivity of the solution to this value was being
checked. A background concentration of pollutant in the
stream is also specified with the actual concentration curve
being computed as the sum of background and pollution event
concentrations. By then specifying a minimum detectable
level for detection of a pollution event, a time period is
defined during which a sample must be taken for the event to
be detected. The term "minimum detectable level," as used
here, does not refer to the actual concentration which can be
detected by a given analysis technique. Rather, it refers
to some concentration indicative of a pollution event due to
its value being beyond the range of normal variation. The
actual value of the concentration at any point in time is
not important to the final desired solution, but is just a
tool to arrive at the detection interval. A hypothetical pol-
lutant that meets the requirements of this method is assumed
throughout the study, but many common pollutants very nearly
meet these requirements and the methods should be applicable
to these.
In the case of a continuous discharge, no practical method
of calculating longitudinal dispersion was found. For the
present model setup, if a continuous discharge occurs, its
duration is simply added to the detection interval for an
instantaneous spill occurring at the same time and location,
thereby increasing the detection interval by the duration
of the spill. The discharge diluted by stream flow is still
52
-------
c
O)
u
c
o
O
__ detection interval
o>
minimum detection
level
Time, minutes
Figure 6. Example instantaneous spill dispersion curve,
53
-------
assumed to be above the minimum detection concentration.
This creates what approximates a square wave dispersion
curve and as spill duration increases and the curve spreads
out, the calculated dispersion become less influential on
the detection interval length.
Sampling locations along a reach and sampling times at these
locations are read into the program. Sampling times are
specified within the same 0 to 30 day range as pollution
event occurrence. Any sampling frequency from once to many
times during this period may be selected. The model thus
permits any combination of sampling location and frequency
to be studied.
A brief explanation of the computation procedure is provided
below. All necessary stream characteristics, sampling times
and locations, and spill durations are read into the com-
puter program. Uniform random numbers are generated to
select the time and location of the pollution event. For
each sampling location downstream from the spill location,
a dispersion curve is calculated and the time interval for
detection is determined. The given times of sampling at
each station are tested against the calculated detection time
interval and the computations are made as to whether or not
detection occurs. Each pollution event is tested at each
sampling location for all sampling times, if necessary. If
detection occurs, no further locations and times are tested
for that event, so that detection will not be counted more
than once for the same event. When this is completed, a
new pollution event is generated and the process repeated.
Presently, the program generates 50 events and outputs the
number of events detected out of this total. This is repeated
20 times so the result is 20 sets of 50 events, or a total of
1000 random pollution events. The results are used to calcul-
ate the detection level for the specified sampling configur-
ation and spill duration. The results from different com-
binations are compared and estimates made of quantitative
values for detection.
Now the second part of the model concerning base level data
will be discussed. For planning and other related purposes,
it is necessary to have data that are representative of the
ranges in stream quality that may occur in time and space.
Short term variations such as diurnal are not considered here,
but seasonal and annual variations as well as long-term
trends are important. The state-of-the-art paper indicated
that grab sampling was well suited to obtaining this type of
data due to the less stringent sampling frequency requirements
and to the wide range of parameters measurable by this method.
In most instnaces, however, these observations were based on
54
-------
opinion and experience, rather than experimental evaluation.
This phase of the study is designed to develop a method
whereby the effectiveness of grab sampling in meeting data
needs of this type can be evaluated.
Data of this type can probably be best presented in terms of
some simple statistical parameters, such as means and standard
deviations. This suggested that a theoretical statistical
approach might be applicable. At the same time, it was felt
that the evaluation could also be accomplished using simula-
tion techniques. The end result was to use both so that the
statistical theory was used as the basic evaluation procedure
and a mathematical model developed for illustrative and com-
parative purposes.
A primary objective of employing statistics is to determine
the sampling frequency necessary for obtaining,representative
base level data. As was the case throughout this study, the
assumption is made that sufficient data is already available
for designing the surveillance system. Using the known data,
it was desired to establish a routine surveillance program
which would adequately perform at a given level. Snedecor
and Cochran (1967) present a simple method for answering the
question, "How large a sample do I need?" This method was
easily adapted to the problem at hand to answer the question,
"What sampling frequency do I need?"
The first assumption is that a normally distributed population
is being sampled. Since we are dealing with averages and
since the "Central Limit Theorem" states that sample averages
tend to become normal even if the original population is not,
this assumption is easily justified. Next, an estimate is
desired to be correct to within some limit, ±L. For 95 per-
cent probability, there is a 95 percent chance that the
observed mean x lies between the limits u - 1.96 a/ n and
u + 1.96 a/ n, where u is the mean, a is the standard deviation
and n is the number of observations. This gives the equation
^ - ':!£-£'
1.96a/ n = L ..... ^ (2)
Solving for n and replacing 1.96 by 2 for simplicity gives
n = 4ff2/L2 (3)
Equation 3 has the form used in statistical comuptations for
arriving at necessary sampling frequencies for given perform-
ance levels. Some estimate of the standard deviation, a,
for the data of concern must be available for the equation
to be used. This completes the statistical approach, and the
model approach will now be presented.
55
-------
The model approach uses a simple mathematical model of a
stream and surveillance system as the basic tool. Although
actual water quality data for an existing stream could be
used, the initial approach was to use hypothetical data.
The advantage of this procedure was that the hypothetical
data can be generated with known variations and trends. In
addition, use of hypothetical data prevented the problem of
trying to work with incomplete or inadequate data, which is
often the case in real situations. The evaluation of the
surveillance system will be accomplished by observing its
effectiveness in detecting the known data characteristics.
A computer program has been written for development of the
model (Vanderholm, 1972). The initial function of the pro-
gram is to generate a set of long-term water quality data
for a given point in space along a stream. The data includes
known seasonal and annual variations, long-term trend vari-
ation, and random variation. The surveillance system char-
acteristic of concern is the sampling frequency, which is
handled as an independent variable. This is an input variable,
and it can easily be changed to study various combinations.
Again, a complete listing of the program can be found in
Vanderholm (1972). The model does not consider the effect
of sampling locations, since it deals with only one point
in space. In actual design of a surveillance system, it is
anticipated that sampling station locations will be established
on the basis of known river and pollution source character-
istics. After these locations are established, it will be
necessary to study the variability of the desired quality
parameters to arrive at a sampling frequency adequate to ob-
tain the representative data.
A bried description of the computation procedure by this
program is as follows. A hypothetical historical data set
for a desired parameter is generated for a sampling location.
For example, this could consist of ten years of weekly
observations of a particular parameter. The known data
characteristics could include annual means, a cyclic seasonal
variation and a random normal error term with a specified
standard deviation. A proposed sampling schedule is read
into the program for evaluation. Data values and sampling
schedule have time subscripts so that a sample taken at
any time will fall on a single data value, or between two
values. An observed data value is obtained either by inter-
polation or by direct reading, if sampling and data time
correspond. These observed values are retained and the
characteristics of the hypothetical data. By varying samp-
ling frequency, it is possible to estimate a frequency which
will result in reasonably representative data.
56
-------
Data obtained from this model is compared to that predicted
by the previously described statistical approach. While
it is realized that many trials should be made to obtain
values that are statistically valid, it is felt that the
model data serves as a useful illustration of what may be
happening in an actual situation.
The above described surveillance network simulation models
permit a quantitative evaluation of a surveillance network
design. Optimum as used here refers to that design which
best meets an agency's needs when the financial and man-
power constraints under which the agency must operate are
considered.
It is not proposed that every state go through the modeling
phase of the design procedure, although the modeling is
simple and mainly requires only access to a computer. The
analysis to be presented under the application section of
this report will illustrate the results of the model and
this information in itself will provide a state agency with
a much better feel for the relationships that exist between
effectiveness and frequency. This information, combined
with a stream's characterization, will greatly improve a
routine sampling network.
Up to this point in the design procedure, sampling location,
parameters to be measured, and sampling frequency have been
analyzed. Thr fourth and last aspect of the grab sample
design procedure is a cost analysis of the various designs.
Cost Analysis
The primary reason to establish a cost analysis is to deter-
mine the trade-offs between performance and cost so the
routine network manager can obtain the most information with
his budget. The performance evaluation has been developed
previously; therefore, this discussion will outline the
accumulation of cost information and relate this to perform-
ance .
The determination of costs associated with grab sampling
involves three major categories. These are collection, anal-
ysis, and handling costs. Costs will be determined on a
per sample basis. Using costs on a per sample basis will
permit easy analysis of alternative sampling frequencies,
number of samples, etc.
Obtaining costs on grab sampling will involve a determination
of actual expenses associated with collection, analysis,
and data handling. Collection costs include salaries, trans-
portation (both of collectors and shipping of samples to the
57
-------
laboratory), and field analysis equipment amortization.
Laboratory analysis costs will include salaries of labora-
tory personnel, laboratory equipment costs, and purchases
of expendable chemical reagents. Data handling costs can
vary widely, depending upon the level of data handling;
but in general, the costs will include salaries and expenses
related to automatic (computer) storage and retrieval. The
sum of the above costs should yield the total cost associated
with an agency's grab sampling data acquisition efforts.
This total can then be divided by the total number of_samples
collected within the time span considered in calculating
the costs. This then gives a cost per sample which can serve
to relate cost to effectiveness.
The effectiveness versus frequency relationships developed
in the previous sub-section provide a means of relating the
number of samples to some level of effectiveness. Now the
above cost information provides a relationship between number
of samples and cost. Therefore, by using the number of
samples as a common factor, the very important relationship
between cost and effectiveness can be developed. This will
be illustrated clearly in the application section.
Costs for automatic monitoring and remote sensing are devel-
oped in their respective state-of-the-art papers. These
costs will be fairly uniform across the country, and conse-
quently, they were not considered dependent upon the state's
location. This is not true for grab sampling costs since
they are dependent upon prevailing wage rates, size of state,
etc. This is the reason for the above discussion of grab
sampling cost determination.
The comparisons of cost-effectiveness of grab sampling versus
automatic monitoring or remote sensing are difficult to make
because of the different types of data acquired. For this
reason, the decisions relative to these techniques must be
based more on the goals of the agency relative to types of
data supplied and to the relative costs involved. These
aspects of the techniques will be discussed in the appli-
cation section.
Sensitivity Analysis
As was briefly mentioned in the above section, an analysis
of the performance versus cost is needed to provide the
necessary information to a surveillance network manager so
that he can make rational decisions concerning its design.
This will be accomplished by evaluating two levels of desing
alternatives: 1) alternate levels of grab sampling; and
2) alternate data acquisition techniques (automatic monitoring
58
-------
and remote sensing). A third alternative with respect to
data collection would be to consider different pollution
control strategies.
The levels of grab sampling will be evaluated by presenting
various degrees of effectiveness versus the costs involved.
The water quality manager can weigh this information against
other factors involved and then make his decision with res-
pect to grab sampling. By presenting various levels of
effectiveness versus cost, the decision maker can visualize
how sensitive the data acquisition system is to changes in
cost, manpower, etc. '
The alternate data acquisition techniques will be evaluated
as to their ability to replace a part of all of the grab
sampling network. Automatic monitoring is geared almost
exclusively to- supplying an abatement or enforcement type
of data. Thus, automatic monitoring will be evaluated with
respect to its ability to replace the abatement part of a
grab sampling network.
Evaluation of remote sensing as an alternate data acquisition
technique will involve a general comparison between a com-
prehensive space coverage, a severe limitation on parameters,
and a relatively high cost of operation. Relating this to
grab sampling and automatic monitoring will be accomplished
through a specific discussion of its merits, cost, advantages,
and disadvantages in the field of routine water quality
monitoring.
59
-------
SECTION VIII
APPLICATION OF DESIGN PROCEDURE TO COLORADO
In this section, the procedure developed earlier in the
report will be applied to the State of Colorado. The
application will be performed using the data available;
therefore, some aspects of the procedure application will
be weak due to a lack of available data. This is especially
true with respect to waste outfall inventories. With the
current concern over permit systems, however, this data
should generally be available to water management agencies
in the near future. Also, the federal requirement for
completed regional plans before waste treatment construc-
tion grants are funded is contributing to the information
needed to apply the surveillance design procedure. These
regional plans require information on waste outfall location,
flow, and character. This information will be necessary
in determining which parameters are to be measured.
Application Area and Conditions
Prior to the design of any system to collect data, the
entity to be described by the data must be understood. For
a water quality surveillance system, this includes an under-
standing of the effects of physical, economic, and legal
aspects upon water quality. This is especially true in
Colorado, where there are wide variations in climate, to-
pography, and population density. Each of these variations
has a large effect on Colorado's water quantity and quality.
Physical Conditions
Colorado's topographical uniqueness is very evident when
one observes a relief map. The eastern half of the state
consists of flat high plains and broad rolling prairies,
while the western half consists of alpine terrain. The
state is bisected by the Western Continental Divide of the
Rocky Mountains and contains 53 mountains higher than 14,000
feet. With an average altitude of 6,800 feet, Colorado is
the highest state in the Union.
In total area, Colorado is the eighth largest of the 50
states, with 104,247 square miles. The United States Gov-
ernment owns over one*-third of the land area and uses it
for National Forests, National Parks and Monuments, military
61
-------
reservations and federal installations. As for neighbors,
Colorado is bounded on the south by New Mexico and Okla-
homa; on the east by Kansas and Nebraska; on the north by
Nebraska and Wyoming; and on the west by Utah.
Rainfall in Colorado is light, and consequently water is a
most carefully guarded resource. The eastern plains have
an annual precipitation of from 6 to 18 inches, with the
Denver metropolitan area receiving an annual average of
from 12 to 14 inches. In the mountains, annual precipita-
tion varies from 18 to 26 inches, with a large part of this
in snowpack. Based on the above precipitation averages,
only pasture of limited value can be grown without irrigation.
Due to its high elevation, Colorado contains the headwaters
of four major river basins. These are the Colorado, Rio
Grande, Arkansas, and Platte Rivers. The rivers and streams
carry water out of Colorado in all directions into adjoining
states with no useable streams bringing water into the state.
Colorado, therefore, depends on water generated within its
boundaries.
The population of Colorado increased 25.8 percent between
1960 and 1970 to a current (1970) total of 2,207,259. Of
this total, approximately 74 percent live in a narrow band
along the eastern edge of the Rocky Mountains from Fort
Collins to Pueblo. By the year 2000, the population within
this "Front Range" area is expected to be 3.2 million—a
doubling of population in 30 years. The Front Range is
currently supplied with clear water as it flows down from
the mountains.
Economic Conditions^
Colorado's economy has been based on its natural resources.
Mining, and agriculture based on irrigationf were the original
economic endeavors, but recently the excellent environment
(crisp, dry air; pure mountain streams; etc.) has lured many
recreation minded and environmentally conscious people and
industries into the state. Also, many government installations
are located in the state. This has led to rapid growth as
illustrated in the above population figures. More water is
being demanded to satisfy the growing needs as more pressure
is being exerted upon environment.
As a result of the large growth on the Front Range, the Colo-
rado-Big Thompson project was developed and is currently
dapable of transferring 300,000 acre-feet per year from the
Colorado River to the South Platte River B,asin. The Frying
Pan-Arkansas project will also import some 67,000 acre-feet
62
-------
per year into the Arkansas 'River Basin (Colorado Water
Pollution Control Commission, 1967).
Legal Conditions
States in the eastern part of the United States use the
Riparian Doctrine with respect to water. Since this doctrine
is dependent upon adequate stream flow, it does not work
in semi-arid or arid regions. For this reason, Colorado uses
the doctrine of Prior Appropriation. Under Colorado law:
1) water in its natural course is the property of the public,
and is not subject to private ownership; 2) a vested right
to use the water may be acquired by appropriation and appli-
cation to nebeficial use; 3) the person first in time to use
the water is first in right; and 4) beneficial use is the
basis, the measure, and the limit of the right (Colorado
Water Pollution Control Commission, 1967).
There are certain preferences in use of water under Prior
Appropriation as stated in the Colorado Constitution. It
states in Section 6, Article XVI, that:
"The right to divert the unappropriated waters
of any natural stream to beneficial use shall never
be denied. Priority of appropriation shall give
the better right as between those using the water
for the same purpose; but when the waters of any
natural stream are not sufficient for the service
of all those desiring the use of the same, those
using the water for domestic purposes shall have
the preference over those claiming for any other
purpose and those using the water for agricultural
purposes shall have preference over those using
the same for manufacturing purposes."
From the above, it can be seen that an individual does not
actually own the water, but he has the right to take from
the source of supply, sufficient water to meet his daily
needs. "Sufficient" depends upon the limit of his decree.
Because of the prior appropriation doctrine, diversions
along the foothills take water from the streams and canals
deliver it to the users. This leaves the streams with little
or no flow until the return flow from agriculture, munici-
palities, and industry begin to enter the streams. Therefore,
streams, past the points of use, today run continuous where,
in early history, they were intermittent. As a consequence
of prior appropriation, the streams consist primarily of
return flows or wastewater—an important point to remember
when developing a strategy to control water quality.
63
-------
In summary, physically Colorado has many variations and
constrasts, and legally it allows water to be removed from
the streams and put to beneficial uses. These facts are
extremely important when considering a water quality man-
agement program for Colorado. They also have a very_large
influence on the data acquisition system to be used in
managing the water quality.
Stream Characterizations
The stream characterizations are performed by river basin
with the major main stems being analyzed in more detail
than their tributaries or other smaller streams. For the
main stems, water quality data available from the Water
Pollution Control Division (WPCD) of the Colorado Department
of Health is used to develop graphical representations of
the river's quality over the total length. For the tribu-
taries, the WPCD data is put into tables. Many of the
smaller rivers and tributaries have too few data points
to warrant graphical representation. Since the WPCD does
not currently have a waste outfall inventory for the state,
this information will not be included in the analysis.
Biochemical oxygen demand (BOD), dissolved oxygen (DO), pH,
total dissolved solids (TDS), and flow are the parameters
used in this particular characterization. This is not to
say that these are the only ones which should be used, but
they were chosen for this particular study. The WPCD has
been collecting data since 1968, and the means and standard
deviations used in the characterizations are obtained from
the three years, 1968-1970. The stream characterization
details are contained in Appendix D.
In summary, the characterizations indicate that the South
Platte River is the major problem area of the state. Here
the industrial, municipal, and agricultural uses of the
water cause the stream to be high in biochemical oxygen
demand (BOD), low in dissolved oxygen (DO), and high in
total dissolved solids (TDS). Also, since the majority
of the population of the state lives in this basin, it
also has high pollution potential as witnessed by the vari-
ation observed in the data presented in Appendix D.
The Arkansas River has general quality problems near the
population centers, but its main problem is salinity as
it leaves the state. This shows very clearly in the graphs
in Appendix D. The Colorado River and Rio Grande River also
have salinity as their major problems as the urban develop-
ment is minor in their basins. Acid mine drainage and
64
-------
natural geologic conditions in addition to agriculture con-
tribute to the salinity problems.
The data contained in Appendix D will be used in the follow-
ing analysis. When this data is used, references to it will
be noted.
State Agency Strategy
Colorado's Water Pollution Control Division's (WPCD) stra-
tegy was determined by having several of the Division's
staff complete the form in Figure 3. A completed form is
shown in Figure 7. The completed form shows totals of 206,
260, and 254 for prevention, abatement, and program control,
respectively. This results in percentages of 0.29, 0.36,
and 0.35 for each of the broad objectives. Other staff mem-
bers had percentages of 0.30, 0.38, and 0.32; and 0.30, 0.36,
and 0.34 for each of the objectives. This indicates general
agreement of the relative levels of effort. Point totals
for the total chart were 489, 720, and 851 for each of the
respondents out of a possible 1,080. This indicates the
view of the program available to the respondent since the
lower figure was supplied by an engineer in the program
while the top figure was supplied by the Division's director.
The important fact here, however, is the general agreement
of relative levels of effort among the Division's personnel.
It is these relative levels that determine the state agency
strategy. An analysis of budget allocations among the vari-
ous portions of the program should reveal similar results.
The numbers in Figure 7 and the percentages presented ear-
lier indicate that the Colorado Water Pollution Control
Division (WPCD), in general, emphasizes abatement over pre-
vention. In the area of prevention, Figure 7 indicates
medium activity in planning, low to no activity in research,
and high activity in aid programs. Therefore, within pre-
vention, aid programs receive the largest amount of effort.
Referring back to Figure 2, the data needs of aid programs
are nil, which implies that water quality data needed for
prevention will primarily be that associated with planning.
In the past, the WPCD has had very little planning activities
due to its limited size; but now that the federal government
requires regional plans with each sewage treatment plant
construction grant application, the WPCD can be expected to
increase its planning efforts. This in turn will place a
data demand upon the surveillance network. If the WPCD
were not currently getting into planning, it could be said
that the WPCD need not collect data for prevention purposes
since it was devoting its efforts to abatement, which would
be the strategy of the agency. However, since planning is
being emphasized, the surveillance network must reflect its
need of data.
65
-------
Activity Level
Prevention
A. Planning
- Program Planning
- Project Planning
- Planning Coordination
- Contingency Planning
B. Research
- Develop Research Needs
- Approve Research Pro-
posals
- Evaluate Progress of
Sponsored Research
- Research Coordination
- Conduct In-House Research
C. Aid Programs
- Grant and Loan Admin-
istration
- Coordination of Grants
and Loans
Total for Prevention
Abatement
A. Technical Assistance
- Approve WTP Plans and
Specifications
- Inspect New Construction
- Recommend WRP Site Ap-
proval
- Train WTP operators
- Inspect WTP Operations
- Recommend Abatement
Steps for Polluters
B. Regulation
- Issue Permits
- Approve WTP Sites
- Develop Abatement Sched-
ules for Polluters
- Review Stream Standards
- Recommend Changes in
Regulations
- Initiate Further Internal
Action for Persistent
Polluters
High Med Low None
30 20 10 0
X
X
X
X
24 16 8 0
X
X
X
X
x
60 40 20 0
X
X
20 13 7 0
X
X
X
X
X
X
20 13 7 0
X
X
X
X
Points
20
20
20
10
0_
8_
8
0
0
60
60
206
20
7
20
13
13
20
0
20
20
20
X
Figure 7- Pollution control strategy determination for
Colorado as completed by a WPCD staff member
66
-------
Activity Leve
High Med Low None Points
20
jx
X
c. Legal Enforcement
- Issue Cease and Desist
Orders
- Conduct Hearings
- Prepare Briefs for Court
- Initiate Court Action
- Prepare Laws and Regu-
lations
- Levy and Collect Fines
Total for Abatement
Program Control
A. Information Collection
- Surveillance Network
- Field Investigations
- Other Sources
B. Information Processing
- Screen
- Verify
- Interpret
- Index
- Store
- Retrieve
C. Action Initiation 30
- Action Need Reports x
- Inventory and Data
Summaries
- Special Reports x
- Publicity
Total for Program Control
40
IE.
20
13 7 0
X
27
x
13
x
_x_
X
20
X
X
X
X
13
_x
X
X
10
J-JL
JL
260
27
_13_
13
13
30
254
Figure 7. (Continued)
67
-------
As noted in Section V, planning requires large amounts of
data from many sources. The major data requirement that
planning places upon the routine monitoring program is
associated with base level trends. The other types of
water quality data are collected through special inventories,
surveys or a permit system. In addition to water quality
data, the planning function must have data on population
trends? economic conditions, legal and institutional arrange-
ments , etc.
The creation of a good historical record or the establish-
ment of base level water quality conditions does not require
the data acquisition sophistication that enforcement and
regulation require. The data need only indicate the means
and standard deviations, not the extremes in variation or
the relation of parameter extremes to waste outfalls required
for regulation. Because of this the data need not be as
precise or as thorough as specific data for abatement pur-
poses. It can simply be a representation of the summed
effect of all pollution loads on the stream.
To establish a water quality base level, the designer of a
surveillance system must consider parameters, sampling station
locations, and sampling frequencies. Each of these will be
discussed in detail with respect to prevention and then an
application to Colorado will illustrate the utilization of
the guidelines that are presented.
Looking at abatement in Figure 7, it is noted that technical
assistance and regulation both command high levels of activity
while legal assistance has a medium activity level. This
indicates that the WPCD is generally successful in the use
of "conference, conciliation, and persuasion" in abating pollu-
tion and, therefore, the legal activities reflect a lower
level of effort. The WPCD operates under the Colorado Water
Pollution Control Act in which paragraph 66-28-10 states:
"Whenever the division of administration or
county health department shall have cause to
believe that any person is engaging or threat-
ening to engage in any act or practice which
results in the violation of a waste discharge
or water quality standard, or which results in
the violation of any other rule or order of the
coHimission, the division or county health de-
partment shall so inform the commission and
promptly cause an investigation thereof to be
made and, if the division or county health
department shall find, after such investigation,
that a violation of any such waste discharge or
water aualtiy standard or other order exists,
it- shaJ 1 by conference, conciliation, and
68
-------
persuation, endeavor to eliminate the alleged
violation."
It is form this section of the law that the agency derives
its enforcement tool. Abating pollution using "conference,
conciliation, and persuasion" as the enforcement tool re-
quires large amounts of time and effort in regulation,
especially in the development of abatement schedules, and
in technical assistance as related to persuading a polluter
to stop polluting. A perfect example of this has been the
situation in Greeley, Colorado, which occurred during 1971.
The city was violating the regulations and the division
engineers had to spend many months persuading Greeley offi-
cials to remedy the problem. This required the engineers
to develop a report on the technical aspects of the problem
and possible technical solutions.
The question must be asked, if the agency were to devote more
effort to legal enforcement, could activity levels in techni-
cal assistance and regulation be reduced. If this is possible,
the data needs of the agency will necessarily reflect the
change in effort. Under current law, however, under which
this analysis is being performed, the emphasis is on "confer-
ence, conciliation, and persuasion" and, consequently, the
WPCD must concentrate its abatement efforts in regulation and
technical assistance. Therefore, the data collected for
abatement must support the regulation and technical assistance
efforts. Hager (1970) puts particular emphasis upon the fact
that regulation and enforcement are dependent upon surveillance
Elaborating further on data needs for regulation and enforce-
ment, it should be noted that the surveillance system must
yield data that verifies if stream standards are being met.
As noted earlier in the grab sampling state-of-the-art papers,
Velz (1950) expanded upon the fact that in order to accurately
analyze data for stream standard violations, the data must
have a statistical basis upon which conclusions can be drawn.
Otherwise, there is little basis for enforcement action.
The legal enforcement of stream standards revolves around the
parameters and conditions enunciated in the law. For the
Colorado WPCD, there needs to be data available to insure
compliance with the criteria on coliform, dissolved oxygen,
pH, taste and odor, dissolved solids, turbidity, temperature,
and toxic materials among others. The data should statistic-
ally state whether criteria are being satisfied for these
parameters as well as any others that may be crucial to that
particular stream.
If it is noted that the criteria are not being met, the WPCD
uses the data to "persuade" the violator to clean up. In
69
-------
this capacity, the data must be utilized to develop recommen-
dations for eliminating the problem (technical assistance).
As Bramer (1970) has noted, the need for data upon which to
base treatment facility recommendations becomes more acute
as the receiving streams become more sensitive to waste loads
due to flow and/or to water uses which require more consist-
ently high water quality. Therefore, the routine monitoring
should provide in-stream water quality data which will serve
to determine the variability of the rivers flow and quality,
and this should be tempered with the uses of the water. The
development of a relationship between stream criteria and
size and variability of a stream is crucial to determining
realistic alternatives for violators.
If "persuasion" is not successful, the data must be able to
stand up in court. Unless the data is obtained from an ex-
tremely well designed routine surveillance network or by well
planned special studies, there is little chance of the data
being able to withstand the courtroom procedures. In this
light then, the data must be accurate and precise.
As has been alluded to above and noted in the state-of-the-
art paper on grab sampling, the exact nature of the data
needed by a water quality management agency for any particu-
lar stream, will depend greatly upon the character of the
stream. This is especially true in Colorado where some moun-
tain streams have acid mine drainage to contend with while
others have quality problems associated with recreation while
others have problems related to natural pollution. On the
plains, some streams pass through large urban areas while
others are used entirely for irrigation purposes. Clearly
there has to be careful analysis and interpretation of water
quality data obtained from Colorado streams to insure appro-
priate use of the data.
In summary, the Colorado WPCD, under current enforcement pro-
cedures , needs specific data for abatement purposes and his-
toric or trend data for prevention. The routine surveillance
network supplies some of the specific data and all of the
treand data currently used. Special surveys and inventories
supply much of the specific data required.
Before leaving the discussion of state agency strategy, a
few comments will be made on the program control point totals
of the respondents. All respondents indicated a high level "
of activity in information collection, low in information
processing, and low to medium in action initiation. This
assessment could probably reflect the activity levels of
most agencies in the United States. There needs to be more
balance between data collection, data processing and action
generation based on the data. Colorado has taken one step
in this direction by making use of the STORET (a national
water quality data storage and retrieval system sponsored by the
70
-------
Environmental Protection Agency) system, but because of the
agency's limited size, no one person is in charge of data
manipulation. This prevents complete utlization of the data
because there is no complete analysis of the data. As plan-
ning activities within the agency increase, there will nec-
essarily be increased interest in and use of the surveillance
data.
Sampling Station Locations
As noted in the State-of-the-Art Section on grab sampling,
location of sampling stations can be influenced by many
factors and by as many opinions. Some of these factors are
present and potential pollution sources, water usage, physical
stream characteristics, sampling convenience, and response
to public clamor. Once the stations have been located and
data has been collected for a number of years, another factor
enters the picture as far as a re-evaluation of the sur-
veillance system--loss of continuity if a station is moved
or discontinued. Considering all these factors, sampling
station location must be a compromise--no one point will serve
or satisfy all factors involved. »
The first step in determining sampling station location is
to determine the factors involved for thd existing situation.
For Colorado, and as will be the case for most states, the
design of a surveillance system for the state is basically a
re-evaluation of the existing system. This results from the
fact that data from the existing system is used to design
the new system. This immediately presents problems with
determining sampling location since the data available is
dependent upon previous sampling locations. If only the rou-
tine surveillance data is used, it will be difficult to
justify moving a station from its present location since
nothing is known of the river until the next station. This
implies that a more intensive survey of the stream's water
quality is needed if adequate justification is to be developed
for moving, discontinuing, or creating new sampling stations.
Since Colorado does not currently have the information avail-
able to develop the justifications, no attempt will be made
to analyze station location with the exception of the Cache
la Poudre River. The WPCD is currently developing the nec-
essary reports and one has been completed for the Cache la
Poudre. This tributary of the South Platte will be analyzed
to illustrate the suggested procedure for determining station
location. Since the data is not available for the remainder
of the state, the station locations will be taken as given
and the frequency will be analyzed and determined.
71
-------
Cache LJI Poudre River Sampling Station Location Analysis
The comprehensive Cache La Poudre survey was performed to
determine existing water quality, to identify sources of
pollution, and to analyze effects of pollution on the Poudre's
quality. The Poudre was sampled for its entire length of
90 miles with a total of 110 sampling points used. The re-
port does not state such, but apparently only one grab sample
was taken at each point. The samples were taken between
May 5 and June 12, 1970. A followup survey was performed
from September 8 through October 13, 1970. The survey during
May and June occurred during the high water flow and conse-
quently, there was a large dilution factor which reduces the
effect of municipal and industrial effluents. Even so, there
were violations of stream standards at several points.
The survey during September and October was performed during
the low flow period of the year and consequently, its results
would show the more critical areas. Because of this, the
results of the Fall survey will be used in this analysis.
Also, since the water in the mountains is of very good quality,
there is little need to routinely sample there. Therefore,
the station locations will be considered only for the lower
stretches of the Poudre.
The results of the Fall survey are summarized in Figure 8
using the water quality index developed by Brown and others
(1970) . Also noted in the figure are the existing sampling
stations. The station at point 84 was established as a
result of the stream survey. The chart in Figure 8 presents
a very good picture of the water quality situation in the
Poudre River. A corresponding flow curve would be extremely
valuable, but none is available. The chart, however, does
point out the fact that the sampling station above Fort Collins
is monitoring water of very high quality, while the other two
are monitoring water quality problem areas. The upper station
was established to monitor the quality of water supplies to
Greeley and Fort Collins. Since the WPCD is concerned pri-
marily with controlling water pollution, the question must be
asked whether the WPCD can afford to monitor water supply
intakes. Should the water treatment plant do this, and could
they not do it better? How much information can one sample
per month supply to the treatment plant operator? Based on
these facts, it would appear justified to discontinue the
sampling station above Fort Collins and'leave the other two.
Also, according to the chart, there does not appear to be a
need for any other stations.
The comprehensive survey has indicated the problem areas and
the routine sampling stations have been located accordingly.
An analysis similar to this needs to be performed whenever
sampling stations are to be located or moved. Realistically,
72
-------
•98 *.
WATER
80
70
60 .
QUALITY
50 .
INDEX
KO
30
20 „
POTABLE V.'ATER
f*MOUTH OF CANYON
EiC iTY QL £1 COLLtNSS
N.
^COLLEGE STP
« BELOW FT
5COLLINS
91 STP
BELOV/ FT
3COLLINS
#2 STP
A \ A
"*— SAMPLING STATION LOCATION ^DOMESTIC SEV.'AGE
SREF.LE
«BELOW
^GREELEY STP
#58 *62
#92
#101 *103 *108
1110
SAMPLING POINTS
Figure 8. Cache La Poudre River - September, 1970.
(Courtesy of the WPCD, Misbach, 1972)
-------
the location of routine sampling stations should be an objec-
tive of every stream survey along with waste outfall inven-
tories, water quality determination, etc.
Grab Sampling Frequency
As with sampling station location, there are many factors
which affect sampling frequency. Some of these are pollu-
tion potential, water use, hydrological conditions, cyclic
pollution patterns, and available resources for sample col-
lection and analysis. The last factor mentioned is apparently
the most limiting with respect to increasing sampling fre-
quency. This is definitely the case in Colorado. It is,
however, very difficult to justify increased sampling fre-
quencies unless some increase level of effectiveness can be
associated with the increased frequency. Effectiveness, as
used in this report, will be a measure of how well the samp-
ling data represents the true water quality conditions in
the stream.
Colorado established a surveillance system in 1967, but did
not actually begin routine surveillance until 1968. There
were 77 stations which were sampled monthly. This network
and sampling frequency were in response to the Pollution
Control Act as amended. The state established stream stand-
ards with guidance from federal levels, but the surveillance
was established with little local experience or federal help.
Consequently, one sample a month was chosen largely because
at the time no information was available to indicate that it
should be otherwise.
Although one sample per month does not lend itself to a very
active enforcement program, it has established some valuable
trend data. It is this trend data that was used in the char-
acterization of Colorado's streams contained in Appendix D.
An analysis of the stream characterizations indicates very
clearly that some Colorado waters are likely to have their
stream standards violated more often than others. A mean
near the stream standard and a large standard deviation fore-
tells that violations are prone to happen at the point in
question. Likewise, a mean which is not near the stream
standard and a small standard deviation indicates that the
chance of a violation is quite small. Therefore, if one is
permitted to predict the future from the past, it can be
said that certain stations are very crucial to insuring no
stream standard violations occur while others are not. This
is not to imply, however, that some stations are not to be
sampled; it only indicates that there should be differences
in sampling frequency to match differences in quality vari-
ations and possible stream standard violations.
74
-------
Past water quality trends are not the only aspect to be con-
sidered. There is a greater possibility of violations in
heavily populated areas than in areas not heavily populated.
This factor may indicate a need for a larger sampling fre-
quency in urban areas than in rural or mountain areas. In a
mountian area it may be necessary to sample some stations
more often than others due to mine tailings or in rural areas
due to feedlots or irrigation return flows.
Secondary and Primary Sampling Networks
The above discussion indicates a need for several sampling
frequencies in a surveillance network, depending upon the
situation. Given a number of sampling stations, it would be
possible to have a different frequency for every station,
but this is not very practical. Also, since no one sampling
frequency is adequate for all stations, there must be some
compromise. For Colorado, the suggested compromise is two
sampling frequencies. The high sampling frequency will be
on a Primary network, while the low frequency will be on a
secondary network. This recommendation is made because Colo-
rado has a highly urbanized front range, where its greatest
pollution potential exists, while the remainder of the state
is rural and mountainous where there is less pollution po-
tential. This is not to say that all primary stations will
be in the urban areas and all secondary stations will be in
the rural areas, but in general, the primary stations will
be at critical points while the secondary stations will be
in less critical areas. This will permit the acquisition
of more data in areas of high variation or high probability
of stream standard violation, while reducing the effort
devoted to stations which are fairly constant. There is no
loss of information on the secondary stations since they are
fairly constant, and the money and manpower devoted to them
can better reflect their importance to the overall program.
Likewise, with the primary stations, where more effort is
being devoted to surveillance.
Another way to look at the primary-secondary concept is from
the objectives of the state agency. As disucssed earlier
in the report, a water quality management agency has two
broad objectives—prevention and abatement. As illustrated
in Figure 1, prevention, in general, requires trend data
while abatement, in general, requires specific data. In this
context then, the primary and secondary networks both supply
trend data for the entire state while the primary network
supplies specific data at the points where abatement is needed.
Hence, the surveillance network is geared to the data needs
of the agency.
75
-------
For Colorado, the designation of sampling stations as either
primary or secondary was accomplished as follows. Those
stations which had high variability and were located in a
highly populated area were given primary classification.
The stations which had high variability with small population
or low variability with high pollution potential were con-
sidered on an individual basis. Those stations which had
small variation and little pollution potential automatically
were put into the secondary classification.
In performing the evaluation, the figures and tables of
Appendix D were used to determine variability. Due to a
lack of information on waste outfalls, the pollution poten-
tial was determined from consideration of population-density
and industrial and agricultural activity.
For the Arkansas River, Figures D-l, D-2, D-3, and D-4 of
Appendix D illustrate graphically the variation in water qual-
ity. All parameters are fairly constant until the river passes
Pueblo. For the Nepesta and La Junta stations the biochemi-
cal oxygen demand (BOD) is quite high and, as shown in
Figure D-3, the variability is large. As the river nears
the state line, the total dissolved solids (TDS) reaches its
maximum and is quite variable. Because of the high pollution
potential from Pueblo and the variability at Nepesta, La
Junta, and Holly, these stations are designated as primary
stations with the remainder being secondary stations.
The tributary to the Arkansas which appears to have serious
problems is Fountain Creek. It passes through Colorado
Springs and enters the Arkansas at Pueblo. Its flow is
quite small to be serving such a high population/ but its
stream standards reflect the fact that it would be very diffi-
cult to maintain a high quality water in the creek. Due to
this extremely high pollution potential, the station below
Colorado Springs should be a primary station with the other
two being secondary. Due to the fact that none of the other
tributaries appears to be near violating its stream standards,
and to low pollution potential, the remainder of the stations
in the Arkansas River Rivsin should be secondary.
For the Colorado River, Figure D-5 indicates that everything
except the flow remains fairly constant until near the state
line, where the BOD rises. Figures D-6 and D-7 indicate that
there is some variability in dissolved oxygen (DO) and BOD,
especially at Dotsero and near the state line. At Dotsero,
the standard deviation curve for DO is well below the stream
standard, indicating violations may occur often. Population
density and industrial and agricultural activity on the upper
Colorado main stem are most critical in the Grand Valley.
76
-------
Much of the pollution probelm in the Colorado is due to
TDS buildup as the water passes over geologic formations
which contribute a heavy salt load. In Grand Valley, irri-
gation return flows contribute to the salt load. From this
analysis, the station at Dotsero and either the station at
Loma or the Environmental Protection Agency (EPA) station
at the state line should be primary, with the rest being
secondary.
For the tributaries to the Colorado River, there do not
appear to be any major problems besides the TDS situation.
Table D-4 indicates that the major points of TDS problems
are when the tributaries with large flows enter the Colo-
rado or other tributaries. The Uncompahgre, a tributary to
the Colorado, has high TDS at Delta. The Gunnison also has
high TDS and a very high flow as it enters the Colorado.
Other tributaries with high flows are the Eagle and Roaring
Fork Rivers. Both of these also have a small probability
of DO and pH problems. For these reasons, the four stations
mentioned should be primary stations, while the remainder
of the stations in the Colorado River Basin in Colorado
should be secondary.
Figures D-9, D-10, D-ll, and D-12 illustrate the water qual-
ity variation of the main stem of the South Platte River.
It is this basin which also has tHe largest concentration of
population density and industrial and agricultural activity
in Colorado. Denver is generally situated between Littleton
and Henderson, and as Figure D-9 illustrates it is between
these two points that the BOD is the largest and DO is the
lowest. Between Henderson and Kersey, the St. Vrain,
Thompson, and Cache la Poudre Rivers all enter the main stem,
so recovery from Denver is not as fast as possible. Beyond
Kersey, the river recovers and remains fairly stable to the
state line. Because of the variation and pollution poten-
tial, the stations at Henderson and Kersey should be primary
and the stations at Littleton and Julesburg should be pri-
mary to determine quality of inflow to Denver Metro area
and the quality of water leaving the state, respectively.
Also, the TDS standard at Julesburg may be violated, but
since flow measurements are not taken, it cannot be deter-
mined. It will be recommended later that flow measurements
be obtained so that the TDS stream standard can be checked.
By observing the variations of data in Table D-6 and con-
sidering the pollution potential, the primary tributary stations
were determined. These are the stations at the mouths of
the Cache la Poudre River, Big Thompson River, and Clear
Creek. Also, primary stations are those on the St. Vrain
below Longmont, Boulder Creek at the Weld-Boulder county
77
-------
line, the remaining two stations on Clear Creek, Bear Creek,
and a new station on Cherry Creek. This network of pri-
mary stations will supply data on the major streams in the
Denver area and on those major tributaries entering the
South Platte north of Denver.
Tables D-7 through D-13 of Appendix D contain the stream
characterization data for the remaining streams in Colorado.
Generally, the pollution potential in the areas of these
remaining streams is low, as is the water quality variation.
TDS is a problem at some stations, but there is no way to
tell how near it is to violating the stream standards since
no flow measurements are made. For these remaining stations,
therefore, a secondary designation will be applied.
Sampling Frequency versus Effectiveness
In order to determine the primary and secondary sampling
frequencies, there must be some relationship between samp-
ling frequency and effectiveness. As mentioned in the
previous section under "Stream Simulation," Vanderholm (1972)
has developed the needed relationship. It should be pointed
out, however, that the predictions made in the generation
of relationships should not be taken as absolute, but rather
as relative comparisons for various sampling combinations.
In developing these relationships; Vanderholm used two ap-
proaches . One set of relationships will determine the
ability of a monitoring network to detect spills, or in
general, to detect rapid changes in water quality. The other
set will determine the ability of a station to detect trends.
These two analyses roughly follow the breakdown of the net-
work into primary and secondary stations.
In the area of spill detection, relationships have been de-
veloped to illustrate the effect of number of samples,
number of sampling stations, and spill duration on the samp-
ling frequency. In trend analysis, relationships have been
developed to illustrate number of samples versus the error
between sample and true value which can be tolerated.
These relationships, with cost data, will be used to relate
cost to effectiveness.
The relationship between number of samples and pollution
event detection was determined by establishing a stream
reach length of 100 miles in the stream simulation model and
establishing a sampling point at the downstream end of the
reach. Spill location was completely random along the
entire reach and time of occurrence- random in the range 0 to
78
-------
30 days. A random spill duration from 0 to 3 days was used.
Sampling times were uniformly spaced over the 30 day period
and the number of samples varied from 2 to 30. The results
of this trial are shown in Table 2. The total number of
samples can also be described as the total number of chances
to detect each pollution event. The data indicates that the
detection probability is proportioned to the number of samples
taken. Figure 9 illustrates this graphically. The relation-
ship is not linear. Since a random spill duration is used,
a certain number of short duration spills will remain un-
detected until the time interval between samples is smaller
than the detection interval of the smallest spills. This
accounts for the decreasing change in detection with change
in sampling frequency. To obtain a detection level of 100
percent under these conditions would require a very high
sampling frequency which would probably not be feasible.
Still considering spill detection, a relationship is de-
veloped between the number of sampling stations and sampling
frequency while holding the total number of samples constant.
The stream reach under consideration was again set at 100
miles and the spill characteristics established in the same
manner as above. In this case, however, the number of
sampling stations and the number of samples monthly per sta-
tion were varied to hold the total number of samples con-
stant at 12. From the plot in Figure 9, we can see that this
should give a detection value of approximately 50 percent.
The actual model output data for this trial in shown in
Table 3.
The data shown in Table 3 is for two different situations:
1) with pollution events originating on the upper 20 miles
of the reach; and 2) with events originating over the entire
100 mile reach. Sample locations were uniformly spaced
along the 100 mile reach (for example, for four stations
the mileage locations would be 25, 50, 75, and 100 miles
downstream). Where only one station was used, it was loca-
ted at the downstream end of the reach. Sampling times
were arbitarily selected at uniform time intervals over a
30 day period. Compensation was made for the fact that
the time period is finite. This insured against events
going undetected due to the end of the time period.
Figure 10 is a plot of the data in Table 3. The most notice-
able feature is the higher detection levels when events ori-
ginate in the upper 20 miles. This is logical when it is
considered that events pass a large number of sampling sta-
tions when all events originate in the upper 20 miles. The
plot also illustrates the effect of number of sampling sta-
tions and frequency. However, the effect is different for
the two situations considered. When spills originate over
79
-------
Table 2., Effect of number of samples on detection.
Total Number
of Samples
Number of
Pollution Events
Percent of
Events Detected
Trial
8 12 16 20 25 30
4000 4000 4000 4000 4000 4000 4000 4000
6.2 16.7 39.2 52.1 60.0 71.3 75.1 83.9
80
-------
0)
o
0)
100
80
60
oo
o
•H
•M
O
0)
-p
0)
Q
40
20
10 15 20
Number of Samples per Month
25
30
Figure 9. Detection vs number of samples for one station at lower
end of reach with 0-3 day random spill duration.
-------
Table 3. Effect of sampling frequency fnd number of stations
on event detection with constant number of samples.
Number of Sampling
Stations
Number of Samples
per Station
Total Number of
Samples
Total Number of
Events Tested
Average Detection,
Per Cent
*
Case 1
Case 2
Trial
1.2 3 4 5 6
12 6 4 3 2 1
1 2 3 4 6 12
12 12 12 12 12 12
4000 4000 4000 4000 4000 4000
•
32.2 32.3 34.6 38.9 40.3 49.2
59.9 61.4 61.3 62.2 62.8 56.6
*Case 1 - events originating over entire 100 mile reach
Case 2 - events originating over upper 20 miles of 100 mile
reach
82
-------
00
U)
4J
fi
(1)
U
Q)
O
-H
-P
O
(1)
4J
-------
entire reach, detection increases almost linearly as samples
per station increase. When spills originate over the upper
20 miles, there is a slight increase, then a decrease as
samples per station increase.
As the number of stations is reduced, a greater number of
samples are taken towards the lower end of the reach, and as
the frequency per station increases, the sampling occurs
more often during the time period. Both of these factors
explain why the detection increases as the number of sta-
tions decreases and frequency of sampling increases. For
the events originating in the upper 20 miles, detection falls
off as the events have time to dissipate when a few stations
are located far downstream from the sources of pollution.
From this discussion, it can be concluded that fewer stations
wit.Ii higher sampling frequencies would better serve to detect
spills than many stations with low frequency. The many
stations, however, would provide a better basis for locating
the pollution source once a spill has been detected.
Now an analysis will be made of the effect of spill duration
on detection percentages. This information is helpful if
some knowledge is known of what pollution events to expect.
For this trial, a stream reach length of 20 miles was used
with one sampling station specified at the lower end of the
reach. Spill location was allowed to vary randomly through-
out the reach, and the time interval used was 0 to 30 days.
Figures 11, 12, and 13 show the results obtained from the
model for this trial. Each plotted point represents the
percent of events detected from a total of 1000 events. For
example, 1000 events with a spill duration of 24 hours were
generated occurring at random times and locations. When a
sampling frequency of twice monthly was used, a detection
value of 5.9 percent resulted. This process was repeated
many times to obtain the plots shown. The scales used in
Figures 11, 12, and 13 do not permit clear presentation of
zero detection or instantaneous spills, so Figure 14 has
been included to better illustrate these results. The 100
percent detection point was assumed to exist when spill
duration was equal to the time interval between samples. In
all cases, this plotted almost exactly on the observed line.
If information is avialble on spill durations, Figures 11,
12, 13, and 14 can be used to estimate the surveillance
schedule necessary for a desired detection level. Again,
however, it must be stressed that the values presented are
approximations for the physical situations used in the
model, but they can be useful when considered in a relative
context.
To elaborate a little further on spill duration, another
trial was run where spill duration was allowed to vary
84
-------
100 _
oo
en
-P
fl
0)
ft
O
•H
4J
U
OJ
-p
(U
Q
80
60
1 sample
40
20
100
200
300 400 500
Spill Duration, hours
600
700
800
Figure 11 . Spill detection by one station at end of 20 mile reach for one, two,
and four samples during one month period.
-------
100
-p
c
0)
o
•H
.(->
u
0)
4J
-------
25
oo
0)
o
M
OJ
fi
O
•H
-P
0
-------
00
00
-p
C
Q)
O
M
0)
fi
O
-H
-P
0
0)
-P
(U
Q
468
Number of Samples, monthly
10
12
Figure 14. Detection of instantaneous spills
-------
randomly in several ranges rather than specified as in
the preceding trial. All other factors were the same.
The results of this trial are shown in Figure 15. The plot
can be used in the same way as Figures 11, 12, 13, and 14,
but it is not necessary to specify an exact spill duration.
Figure 15 indicates, as did Figure 9, that there is, for
the longer spills, a decreasing return as sampling fre-
quency increases. The principle here is very similar to
the production function concept in economics—the first
few samples return considerably more information than the
last few samples. The short duration spills will require
sampling on the order of minutes or hours instead of days;
consequently, their curves remain in low detection per-
centages.
The concept of information return for sampling effort can
also be related to the number of sampling stations. Except
for the discussion where sample totals were held constant,
there has been no attempt to consider multiple sampling
stations. This must be considered if there are to be gen-
eral relationships developed for an entire river basin
or state. It has been noted already that increasing the
number of sampling points while holding the total number
of samples constant reduces detection effectiveness. How-
ever, if sampling points are increased with a corresponding
increase in total number of samples, the detection could
be expected to increase. This relation will not necessarily
be linear, since the addition of extra effort will return
less and less information per unit of effort as the number
of sample points and samples increases.
In order to analyze multiple sampling stations, the previous
results will be utilized in the following manner. The
results for one station will be modified to depict results
for several stations. Due to the decreasing information
returns to surveillance scale, the addition of a second
station on a river will add a certain fraction of the first
station's information to.the total of the two stations.
Figure 10 indicated that the highest detection level for
one station occurs when it is sampled frequently. It is
now assumed that this arrangement would supply the maxi-
mum information for one station and any others added to
the river will return some fraction. Therefore, the first
station is assigned a relative detection effectiveness
multiplier of 1.0 and the next station will be assigned
a value less than 1.0. By using relative detection effec-
tiveness multipliers derived from Figure 10, values for
effectiveness were calculated and plotted in Figure 16.
This figure can now be used, along with Figure 9, to esti-
mate the detection for a multi-station network on a given
stream. The following examples are presented to illustrate
89
-------
ioo L
u
M
-------
1.0
CO
CO
•H
-P
(0
H
(U
PS
0.8
0.6
0.4
0.2
2 46 8 10
Number of Sampling Stations per Stream
12
Figure 16- Detection effectiveness versus number of sampling stations
per stream with a constant sampling frequency of 12 samples
per year per stream.
-------
the procedures involved. In each example, it is assumed that
a minimum number of sampling points can be selected to ade-
quately monitor the stream. This was accomplished in the
stream characterization and station location sections.
For the first example, assume it has been decided to locate
six sampling stations on a hypothetical stream. Each samp-
ling station is to be sampled twice monthly, and it is im-
portant to know the detection level to be expected. From
Figure 9, it can be seen that for a total of 12 samples
monthly, the level of detection is approximately 52 percent.
However, Figure 9 is based on the ideal situation of one
station at the end of a reach. Since six stations are now
being considered, the effectiveness per additional station
and for the end station is reduced and the 52 percent detec-
tion value must be modified accordingly. From Figure 16,
the relative effectiveness for six stations is about 0.67
(termed "relative effectiveness factor"). Multiplying 52
percent by 0.67 results in an estimated detection level for
the proposed system of about 35 percent for spills in the
0 to 3 day random duration range. If spills of different
durations than these were expected, the above percentages
would have to be modified further using Figures 11 through
15 as guidelines.
Approaching the same situation from a different viewpoint,
the question may be to determine how often the six desig-
nated stations would have to be sampled to obtain an approxi-
mate detection level of 50 percent for 0 to 3 day random ••'
duration spills. Since the relative effectiveness is again
0.67, one must divide 50 percent by 0.67 which yields a
value of 75 percent. Entering Figure 9 with a 75 percent
detection value indicates that approximately 24 samples
monthly are necessary. Each of the six stations would be
sampled about 4 times monthly to obtain the desired detection
level. The decreasing return as sampling frequency increases
is obvious when the two examples are compared. It was
necessary to double the sampling frequency to increase de-
tection from 35 percent to 50 percent.
Continuing to expand the analysis for spill detection, the
primary network for Colorado will now be considered. The
analysis will still be approached from a stream-by-stream
basis. Colorado has 14 streams which have been assigned
primary sampling stations. The assumption is made that a
spill detection level of 50 percent is desired on these
primary stations. Table 4 lists the streams and the number
of primary stations on each. From the table it is noted
that 191 samples per month are needed from the 22 primary
stations. This averages to 8.68 samples per station monthly.
92
-------
Table 4. Primary network analysis results for Colorado.
VD
U)
River or Stream
Arkansas River
Fountain Creek
Colorado River
Eagle River
Roaring Fork River
Gunnison River
Uncompahgre River
South Platte River
Cache La Poudre River
Big Thompson
St. Vrain
Boulder Creek
Clear Creek
Beajr Creek
Number of
Stations
3
1
2
1
1
1
1
4
1
1
1
1
3
1
22
Relative
Effectiveness
(Figure 40)
0.75
1.00
0.83
1.00
1.00
1.00
1.00
0.72
1.00
1.00
1.00
1.00
0.75
1.00
Absolute
Detection
Necessary
67
50
60
50
50
50
50
70
50
50
50
50
67
50
Total
Samples
Per Month
(Figure 33)
18
12
15
12
12
12
12
20
12
12
12
12
18
12
191
-------
Table 5 . Sampling frequency compared to effectiveness
levels for Colorado's primary network.
Effectiveness Total Samples Samples Per
Level Per Month Station Per Month
10 32 1
20 63 3
30 106 5
40 141 6
50 191 9
60 251 11
70 346 16
94
-------
W
(D
H
XI
(0
8
0 5 10 15 20
Number of Samples per Time Interval
Figure 17.. Number of samples vs allowable
error for a equal one and two.
95
-------
On a statewide basis for Colorado, approximately 9 samples
per station per month are needed to achieve a detection of
50 percent. Obviously, some streams will be above and some
below the desired level if an average figure such as this
is used. Those streams with several stations will have
higher detection levels. This may be highly desirable in
looking at the total situation.
Nine samples per month seems rather high; therefore, the
questions becomes what level of detection can be obtained
with one, two, four or more samples per month. The same
analysis used in Table 4 has been completed for other levels
of detection, and the results are presented in Table 5.
Table 5 spells out very clearly the trade offs between the
level of 0 to 3 day spill detection (effectiveness level)
and frequency of sampling for the primary surveillance net-
work in Colorado. These results will be analyzed further in
the next section, but before leaving the topic of frequency
versus effectiveness, the subject of the trend data needs to
be analyzed in order to evaluate the effectiveness of the
secondary stations.
As mentioned in the previous section, two approaches are
used in analyzing base level or trend data surveillance
systems: 1) a statistical approach; and 2) a model approach.
For the statistical approach, the annual mean is used as the
representative characteristic for base level quality data.
Any time period could be used, but for purposes of illustra-
tion, the annual mean was selected.
Figures 17, 18, 19, and 20 are plots of points computed
using the statistical formula described in the previous sec-
tion. As with the previous frequency versus effectiveness
plots, these Figures do not refer to any specific parameter
but are generalized to permit broad application. Use of
the curves requires some previous knowledge of the variabil-
ity of the data to be collected. This was the purpose of
the stream characterizations.
Knowing the variance or standard deviation for a parameter
of concern, the proper curve can be selected from the Figures.
For a given number of samples, then, the ordinate value for
the allowable error at a 95 percent confidence level can be
read. For that number of samples, one can expect to be within
the allowable error of the true mean 95 percent of the time.
At this point, a hypothetical example might aid in use of
the Figures. Consider a parameter for which the available
data shows an annual mean of 50 mg/1 and a standard deviation
of 25 mg/1. The question becomes how many observations yearly
are necessary to estimate the annual mean within tlO mg/1
with a 5 percent risk that the error will exceed 10 mg/1.
96
-------
25 _
vo
0)
iH
XI
n3
O
0
20 40 60 80 100
Number of Samples per Time Interval
120
Figure IB. Number of samples vs allowable error for a equal three,
five, and ten.
-------
00
tn
e
O
5-1
M
W
0)
•H
Xi
fO
g
100
80
60
40
20
0
20
120
40 60 80 100
Number of Samples per Time Interval
Figure 19. Number of samples vs allowable error for a equal 25 and 50
-------
1000
10
IO
H
\
5
O
w
-------
Selecting the appropriate curve, it can be found that 24
observations yearly or a frequency of 2 samples monthly is
necessary. If this were raised to an allowable error of
15 mg/1, it would only be necessary to sample once monthly.
Note that it was not necessary to use the value for the
mean. The standard deviation is the only data character-
istic necessary and the values obtained from the curve are
valid if the mean is 50 mg/1 or 50,000 mg/1. The mean is
really only of concern in deciding what a reasonable value
for the allowable error might be. It should also be empha-
sized that individual observations will commonly exceed the
allowable'error and that only the mean of these observations
is being considered.
These curves can also be used to evaluate a monitoring system
or to evaluate data obtained from a system. Considering
the latter, one can make the statement that if an observed
mean greatly exceeds the allowable error for the given data
and system characteristics, additional observation is de-
sirable. The large deviation may be due to an actual sig-
nificant change in water quality. Since the annual means
are obtained from historical data, deviation for a given
year from previous annual means may be due to long-term trends,
Judgment and additional study would be necessary to determine
the cause of the deviation.
To estimate the performance of an established surveillance
system or schedule, the Figures are entered with the known
sampling frequency and standard deviation for the parameter
of concern. The allowable error can then be read directly.
This allowable error value is an estimate of the accuracy of
the system in obtaining mean values for the parameter for the
time period.
The model approach will now basically serve to further illus-
trate the results of the statistical approach. A time period
of one year was selected and the mean for that period or
the annual mean was used as a basis of comparison. For each
run, a data set was generated consisting of daily values
for a total of 10 years. The generated data was random,
normally distributed about a specified mean, and had a spec-
ified standard deviation. The specified mean of the gener- ,
ated data was shifted each year to simulate a long-term trend.
Various sampling schedules were tested to determine their
ability to predict the annual mean within the allowable error
for the given conditions. Sampling internals tested were 5,
10, 15, 30, 60, and 90 days. The observed data values were
those values of the generated data occurring on a day when
sampling was scheduled. For each year, an annual mean was
calculated from the observed values and compared to the mean
100
-------
of the generated data. For the results shown, the specified
mean and standard deviation of the generated data for the
first year were 35 mg/1 and 3 mg/1 respectively. The
specified mean was increased by 1 mg/1 each year and the
standard deviation held constant.
Table 6 is a summary of the results of these tests. The
specified standard deviation of the population was used to
compute the allowable error. The allowable error could also
have been read directly from Figure 18. The deviation of
the observed annual means from the true mean of the gener-
ated data is compared to the allowable error. The statis-
tical approach stated that it should not exceed the allowable
error over 5 percent of the time. As shown in the table,
the allowable error was exceeded 3 times in 60 trials, which
is exactly 5 percent.
Figure 21 has been included to show how the deviation of
the observed from the true mean decreases with increased
sampling frequency. Although the data shown in this figure
is only applicable to the particular situation specified in
the model, it does point out the type of relationship to be
expected between sampling frequency and accuracy.
The results presented in the foregoing discussion will now
be applied to Colorado to analyze sampling frequency versus
effectiveness for the proposed secondary network. The re-
sults presented are for one station only; therefore, the
analysis will have to be summarized in order to consider a
statewide network of secondary sampling stations.
The analysis requires that parameter's expected means and
standard deviations be used as a basis for sampling frequency
selection. Ideally, all parameters should be included in
the analysis, but this would make the procedure extremely
laborious. Therefore, the procedure proposed is to select
three or four important parameters and to revolve the analy-
sis around these. More could be chosen if more accuracy is
sought.
Knowing the variability of a given parameter permits the
selection of a desired accuracy limit or allowable error
to be permitted in the data collection. With these arguments,
the Figures are entered and a value for the number of samples
at a station for the time period under consideration is read.
This value is converted to a .sampling frequency. If four
parameters are considered, the average of the frequencies
obtained is the compromise sampling 'frequency for the network
under consideration.
101
-------
Table 6. Base level surveillance model results.
Sampling
Interval ,
Days
5
10
15
30
60
90 .
Number of
Samples
Per Year
73
37
25
13
7
5
S . D .
Of
Population
3.0
3.0
3.0
3.0
3.0
3.0
Allowable
Error
mg/1
0.70
0.99
1.20
1.66
2.27
2.68
Number
of
Trials
10
10
10
10
10
10
Number of Times
Allowable Error
Exceeded
0
0
1
1
0
1
Total 60
-------
0)
O
M
Q)
Oi
i
-------
As mentioned earlier, the analysis is for one station,
therefore to consider all stations in a state will require
the selection of one of two approaches. Either each sta-
tion must be analyzed and the resulting frequencies averaged,
or the means and standard deviations must be averaged and
the analysis performed once. The latter approach was chosen
for this application to Colorado. For this particular anal-
ysis, DO, pH, and TDS are chosen as the parameters to be
considered. By averaging the means and standard deviations
listed under the stream characterizations for Colorado, an
average mean and standard deviation for the state for each
parameter can be obtained. The results are as follows:
Standard Accuracy
Parameter Mean Deviation Limit
DO 7.62
pH 8.01
TDS 813.25
1.90
.44
314.86
3.81
4.01
406.63
Necessary
No. of Samples
(From Figures 17&20)
1
1
For purposes of illustration an accuracy limit of 50 percent
of the mean and a period of study of one year will be selected.
Therefore, by taking one-half of each mean, the accuracy
limit or allowable error is determined. The total number
of samples needed for all parameters (five in this case) is
then divided by the number of parameters (3) to give the
average number of samples necessary for the accuracy limit
and period involved. For the above situation, this is 1.67
or 2 samples per year, or one sample each six months. It
can now be said that for these parameters, sampling twice a
year will result in annual mean values within 50 percent of
the true mean with a 5 percent chance of exceeding this limit.
By calculating the sampling frequencies for various limits
of accuracy, the water quality manager can obtain an excellent
view of how his decisions on frequency influence his infor-
mation gain. In Table 7, the frequencies are presented for
accuracy limits from 10 to 100 percent. The frequencies are
presented in whole numbers which required rounding of the
numbers. The results of Table 7 illustrate very clearly that
the closer to the true mean one gets, the many more samples
he will have to collect. The needed relationship between
frequency and "effectiveness" has now been obtained for the
secondary network.
In summary, the Figures and Tables presented in this sub-
section are included for use in relating sampling frequency
104
-------
Table 7. Number of samples required for various
accuracy limits for Colorado's secondary
network.
Accuracy
Limit
10%
20
30
40
50
60
70
80
90
100
Number of Samples
Per
Year Per Station
25
7
4
3
2
1
1
1
1
1
105
-------
to effectiveness. This should be a very useful aid in
itself in planning and evaluating water quality monitoring
systems. The Figures and Tables are highly generalized
and, consequently, should be applicable to a wide variety
of situations.
As with any generalized guidelines, however, use must be
combined with a certain amount of judgment. Individual
situations may have characteristics which dictate more or
less stringent sampling requirements than those indicated
by this data.
Certain arbitrary decisions are also necessary for use of
these guidelines. For example, if one is trying to decide
what sampling frequency is necessary in a given situation,
the decision must first be made as to what the magnitude
of the allowable error will be. Different quality param-
eters may indicate different sampling frequencies, so the
relative importance of the various parameters must be
weighed. When sampling frequency decreases, the decision
of when the samples should be taken becomes more important.
This particular decision can be aided by the use of his-
torical data indicating when extremes or approximately
average conditions are likely to occur.
With the grab sampling
ness relationships now available, the grab sampling
frequency versus cost must be obtained. Analyses of auto-
matic monitoring and remote sensing will follow the grab
sampling discussion.
Cost Analysis
In order to obtain accurate cost estimates for the grab
sampling cost-effectiveness analysis, cost data was obtained
from the Water Pollution Control Division (WPCD). Since
the WPCD is in the Colorado Department of Health, the
laboratory analyses are performed in the Department's gen-
eral labs. The WPCD contributes to the staff and some to
laboratory supplies but in no way pays for the entire
laboratory operation. Because of this, the cost figures
may be low when compared to a laboratory established only
for water quality sample analysis with all costs being
assigned to these samples. Cost figures for automatic
monitoring and remote sensing will be presented when they
are discussed as possible replacements for parts of the
grab sampling network.
Determination of the total grab sampling cost was accomp-
lished in three parts. The first part, collection, included
106
-------
costs for salaries, mileage, equipment amortization, travel
expenses, and shipping of samples to the laboratory. The
total annual cost, $33,460, for collection was then divided
by 1560, the total annual number of samples per year, to
arrive at a general cost estimate of $21.40 per sample per
year for collection.
The estimate for laboratory cost per sample included annual
yearly costs for three chemist salaries and a fixed sum
($4,400) for laboratory support. Since the WPCD utilizes
bacterial and radiological expertise of the state lab, cost
estimates were also included for these items. The total
annual lab cost, $54,210, was then divided by the total
number of samples, 1560, to obtain a general laboratory
analysis cost estimate of $34.80 per sample.
*
The third portion of the total cost is that associated with
posting and storing the data. At the WPCD, this includes
the full time of one engineer and one-half of a secretary's
time. Also, the WPCD is currently using the STORET system
and is renting a terminal. Total annual cost is $15,880
with a cost per sample of $10.18. This third cost factor
is highly variable from agency to agency, depending upon
the effort devoted to data analysis.
The total cost per grab sample, therefore, is $66.38. Again,
it cannot be overemphasized that this cost is for Colorado,
and there are certain aspects of the Colorado situation which
play a very important role in the final figures.
In the following analysis, it will be assumed that there is
a straight line relationship between cost and number of
samples. There are many factors which could affect this
but to determine the true nature of these effects would
require a detailed economic analysis. A detailed economic
analysis at this point would be of marginal value when used
in conjunction with the general water quality trend data.
For the analysis, the cost per sample will be rounded to
$70 per sample. In Figure 22 the cost, using $70 per sample,
is plotted against total number of samples. This now yields
a cost versus number of samples (frequency) which can be
combined with the frequency versus effectiveness relationship
to give the desired cost-effectiveness results.
Surveillance Effectiveness Versus Cost
The results of the previous two sub-sections on grab sampling
can now be combined to develop the information necessary to
provide the basis for rational decision making associated with
107
-------
-P
to
O
u
250,000 I
200,000 L
150,000 L
100,000 L
50,000 L
0
1000
3000
Figure 22.
2000
Number of Samples
Relationship of cost to number of samples for Colorado,
4000
-------
water quality management at the state level with respect
to grab sampling data acquisition. The results of Table 5
can be combined with the results of Figure 9 in order to
associate effectiveness to cost for the primary network.
The results are shown in Table 8. The calculations were
not carried beyond the 70 percent effectiveness level since
the number of samples is increasing past the point of re-
ality. The exponential nature of the curve indicates
that the extra cost for each 10 percent increase in effec-
tiveness is so large that an alternative must be considered.
In other words, the cost involved is not worth the returns
generated. This illustrates very clearly the principle of
diminishing returns. It must also be remembered that the
costs discussed here are for the primary grab sampling
network only.
Figure 23 illustrates the relationship between effort
expended and benefits gained for the primary network in
Colorado for a 0 to 3 day spill length. The information
presented in the figure can be very helpful in deciding
where to spend limited funds, especially if similar curves
were available for other aspects of the total program. The
water quality manager would then have a very good basis for
expending funds in certain areas to the exclusion of others.
He would concentrate his effort in those areas where he had
the greatest return. This would then yield the most effec-
tive overall program for the money expended.
The 0 to 3 day spill length used in the analysis was chosen
because it was felt that this length would be most represen-
tative of the majority of spills. If more specific informa-
tion were available on expected spill lengths, then the
analysis could be repeated. If a longer spill length is
used, the curve would tend to flatten out. This would result
in more effectiveness at a given level of cost. Specific
information on expected spill length could be obtained in a
comprehensive waste outfall inventory.
Considering the high costs involved in obtaining the last
percentage points of effectiveness, it would be very appro-
priate to seek alternatives which, hopefully, would supply
the higher effectiveness levels at a lower cost. It is
at this point that automatic monitoring and remote sensing
may prove valuable in a surveillance effort. However,
before getting into the analysis of these other techniques,
the secondary grab sampling network must be considered.
In analyzing cost-effectiveness for the secondary network,
the results of the past two sub-sections are again used.
Using the number of samples per station needed for certain
levels of accuracy, the total number of secondary samples
109
-------
Table 8. Effectiveness versus cost for the primary
surveillance netowrk in Colorado.
Effectiveness Yearly Number v -.,-,+.
Level of Samples Yearly Cost
10 384 $26,880
20 756 52,920
30 1272 89,040
40 1692 118,440
50 2292 160,440
60 3012 217,840
70 4152 290,640
110
-------
300,000 .
O
O
250,000 .
200,000 .
150,000 -
100,000 -
50,000 -
0 10 20 30 40 50 60 70 80 90 100
Effectiveness Level
(Spill Detection Percentages)
Figure 23, Cost-effectiveness results for the primary
network in Colorado for a 0-3 day spill
length.
Ill
-------
per year can be obtained by multiplying by the number of
secondary stations. For this analysis, Colorado is assumed
to have the same number of stations as listed in the stream
characterizations. This is not entirely correct as Colo-
rado has recently expanded their network. In the stream
characterization sub-section, 82 stations are listed since
all primary stations can also be used as secondary stations.
Using the figure of 82 and the sampling frequencies obtained
earlier for the secondary network, the total yearly samples
necessary for various accuracy levels can be calculated.
The results of these calculations are given in Table 9 along
with the yearly cost associated with the number of samples.
The cost versus accuracy level is presented in Figure 24.
This figure again presents a valuable tool to the water qual-
ity manager in that he has an excellent view of the tradeoffs
available to him with respect to the secondary network.
The results presented in Figures 23 and 24 permit the design
of a grab sampling surveillance network for Colorado with
due consideration of the cost constraint involved. It is
not very realistic for a researcher to sit at his desk and
decide the optimum network for Colorado s'ince the real life
situation is removed. It is for this reason that the results
of the research are presented in a form that a manager can
use. In order to illustrate the use of these results, assume
that Colorado has $350,000 to spend on surveillance for the
next year. The question to be answered is how should this
money be spent in order to best meet the data needs of the
WPCD and how should it be allotted to obtain the most infor-
mation for the given resources. According to the strategy
determination, approximately 45 percent of this should go
to prevention (the secondary network) while 55 percent or
$192,500 should go to abatement (the primary network).
Entering Figure 23 at the $192,500 cost figure indicates
that the effectiveness level will be approximately 57 percent
for the primary network. This implies that the total number
of samples for the primary network should be approximately
2790 or approximately 127 samples per station per year.
This breaks down to an average of 10 samples per month for
each primary station.
Continuing the assumed budget example, the secondary network
would receive $157,500 for the year. Entering Figure 24
at this figure indicates an accuracy level near 10 percent
for the secondary network. This transforms into approximately
2050 samples per year, or 25 samples per year per station
for a frequency of 2 samples per month.
The above example assumed a given budget and then the prob-
lem was to maximize returns. If the level of returns was
specified, say 60 percent effectiveness for the primary
112
-------
Table 9. Effectiveness versus cost for the secondary
surveillance network in Colorado.
Accuracy Yearly Nunber Yearlv m^t
Limit of Samples Yearly Cost
10 2050 $143,500
20 574 37,180
30 328 22,960
40 246 17,220
50 164 11,480
60 82 5,740
70 82 5,740
80 82 5,740
90 82 5,740
100 82 5,740
113
-------
-p
w
o
u
rt
-------
network and a 20 percent accuracy level for the secondary
network, the analysis could be run the other way and a cost
figure could be developed. The approach would be similar
to the above example.
The above example, besides illustrating the technique of
using the report's results, also indicates that as an
agency reaches the maximum effectiveness level in both net-
works, it must re-evaluate the technique used. In other
words, the assumed budget in the example pushes the agency
near the maximum effectiveness level for grab sampling in
both networks. Any additional money obtained would return
little additional information if grab sampling is maintained
as the only data gathering technique. Therefore, it would
be at this point that the agency should implement other
techniques of data gathering such as automatic monitors or
remote sensing.
Grab Sampling Parameters
The decision of which parameters to measure can be influenced
by many factors. The technique of sampling can be the most
restricting element in deciding which parameters to measure.
Automatic monitoring, as noted in the state-of-the-art
paper, has technological limitations on the parameters to
be measured. So does remote sensing. Grab sampling, how-
ever, is not limited in the sense of automatic monitoring
and remote sensing. The sample is analyzed in a lab where
the determinations that can be performed are almost endless.
For grab sampling, therefore, there has to be a decision
made as to which parameters are to be measured and which
are not. The decision must be based on the purpose of the
sampling station (specific or trend data), the character of
the stream, the waste discharged to the stream, the use of
the stream's water, and the personnel and facilities avail-
able for surveillance. Also, the use to be made of the
data must be considered in deciding which parameters are
to be measured.
At one time or another, it is conceivable that almost every
parameter measurable will be needed on a particular stream.
It is impossible, however, to measure every parameter, and
consequently, sound judgment must be used in making the
final decision. This decision must relate to the particular
situation associated with the sampling point. Again, however,
it may not be very practical to have a different set of
parameters to be measured at every sampling station. Some
compromise must be developed with respect to the various
situations that occur in a state.
115
-------
Another very important aspect of parameter selection is
public relations. This point was brought out very clearly
by Brown and others (1970) in the following quote.
"For more than twenty years, demands for billions
of dollars of expenditures for water pollution
control have been made on American municipali-
ties and industires; but no provision has yet
been made for keeping the public informed, in
simple and understanable terms, as to what this
effort and investment is achieving—or not
achieving—in water quality enhancement,"
Brown follows with the development of a water quality index
whicl"> would help remedy the problem mentioned in the quote.
For parposes of public relations, each sampling station that
is sampled should have an index computed for that sample.
If Brown's index is to be used, this would require the mea-
surement of DO (in percent saturation), fecal coliform
density (16/100 ml), pH, BODS (mg/1), nitrate (mg/1), phos-
phate (mg/1), temperature (°C departure from equilibrium),
turbidity (units), and total solids (mg/1). This then would
be a standard set of parameters to be measured on every
sample.
For Colorado, the proposed establishment of primary and
secondary stations presents the opportunity to establish
two general parameter sets with the index parameters serving
as a basis of both. Each set would reflect the purpose of
the station and, therefore, better meet the data needs of the
agency. The secondary parameter set will be long, since
the purpose is to establish trends of various water quality
conditions. Each time the secondary stations are sampled and
analyzed, the primary parameter list will also be analyzed
since the primary list is a subset of the secondary list.
This will establish trend data on many parameters at the pri-
mary stations. The primary parameter set, for practical
reasons, cannot be too long since the number of samples
will be large. The parameter set, however, needs to provide
an indication of various aspects of water quality so that
changing conditions can be noted and abatement action initiated.
Getting more specific, the secondary parameter set for Colo-
rado should contain those parameters currently measured by
the WPCD. Flow should also be included, but since its
measurement may be performed by another agency (USGS, for
example), some arrangements should be made to obtain the
other agency's data. The list of secondary parameters is
shown in Figure 25. The WPCD parameter set has evolved over
the life of the agency and reflects the concern for existing
water quality conditions in the state. This is not to say,
however, that the list should be considered fixed; it should
116
-------
Parameters
Temperature
Turbidity
Conductivity
Dissolved Oxygen
Biochemical Oxygen Demand5
PH
Total Volatile Residue
Total Dissolved Solids
Total Nonfiltrable Residue
Total Settleable Residue
Ammonia
Nitrite
Nitrate
Phosphate
Cyanide
Total Hardness
Calcium
Magnesium
Sodium
Sodium Adsorption
Chloride
Sulfate
Fluoride
Arsenic
Boron
Cadmium
Chromium
Copper
Iron
Lead
Manganese
Silver
Zinc
Selenium
Alpha Total
Alpha-T Total
Beta Total
Beta-T Total
Dissolved Radium 226
Dissolved Radium 226 Counting Error
Total Coliform
Fecal Coliform
Alkyl Benzene Sulfonate
Mercury
Figure 25.
Secondary parameter list for Colorado as
reported from STORET.
117
-------
remain flexible in order to reflect any changes that may
occur.
For the primary parameter set, the list should contain param-
eter measurements which give indications of any physical,
chemical, or biological quality changes. The parameters
mentioned earlier for the water quality index, plus flow,
would provide this basic information. And since the data is
being gathered with a high frequency, it will give specific
indications of a stream's water quality condition. Of
course, the old argument of whether BOD should be included
will arise, but regardless of the validity of the measurement,
it is still considered by many to be very important as wit-
nessed by its inclusion in the index. Questions will also
arise concerning other parameters, but it must be remembered
that this is a compromise.
In order to make the data more useful, several other bits of
information need to be available. Time of sampling would be
very useful in analyzing DO measurements and in developing
general conclusions from the laboratory results. As mentioned
above, flow data is needed to really make a quantitative
analysis of pollution levels. Since Colorado is the highest
state in the nation, elevation is a large factor in some of
the parameter readings. The point in the stream where the
sample was taken may be critical if complete dispersion of
upstream wastes has not occurred. Other bits of information
which may be useful are the sampler's name, technique of
sampling (composite or single), field or lab analysis, how
transported to the lab, and chemist's name. It is not nec-
essary to store all of this in the data storage system.
Instead, it may exist on the sample sheet which would follow
the sample through the complete system and then be filed
after the primary information was removed to be stored in a
computer. The main point of this discussion is to emphasize
the need for a logical, well-documented, and carefully executed
procedure for collecting and analyzing water samples.
The parameter sets detailed above are to serve as the basic
parameter lists for the samples. Adverse local situations
may dictate that other parameters need to be measured at
particular points. The sample sheet which follows the sample
should provide a place for these additional samples, and any
determinations different from the standard set should be
decided on by the sample collector and noted on the sheet.
This illustrates that the collection of samples is critical
to obtaining descriptive information and, consequently, its
importance should be emphasized.
118
-------
Automatic Monitoring
As illustrated in Figure 23, automatic monitoring must be
considered as an alternative to continued expansion of a
grab sampling network in order to increase the effectiveness
levels without pushing cost beyond reality. The ultimate,
of course, would be to have automatic monitors at all 22
primary stations measuring the parameters technologically
feasible and then to use grab sampling to supplement for the
remaining parameters. This would increase the frequency of
sampling by several magnitudes and would correspondingly
increase the effectiveness levels nearer to 100 percent
with respect to spill detection.
The parameter restrictions, discussed in the state-of-the-
art paper, indicate that grab sampling must continue to be
a part of the surveillance network. The importance of grab
sampling, however, may be put on a secondary sampling basis
providing the trend data, while the automatic monitors supply
the enforcement data. The value of automatic monitoring
data as enforcement data must be considered in deciding to
rely completely upon it. This is especially true with res-
pect to the parameter limitations and the ability of auto-
matic monitoring data to stand up in court. This particular
aspect has not been taken into consideration in the analysis
to follow since the depend upon the stream character-
ization and the legal situation that will result when the
data is first used in a court case. The final answer can
only be obtained after the data has been presented in a
local courtroom.
The truly major advantage of automatic monitoring over grab
sampling is the real time nature of the technique. This
permits immediate enforcement action to take place, but
again, the problem arises as to how will this fact be ex-
pressed in a quantitative fashion. Agency action depends
largely upon the agency personnel and the legal structure
under which they operate. The action generated by the data
acquisition technique can, therefore, not be considered
within the analysis framework established in this report.
The fact, however, remains that automatic monitoring pro-
vides a greater potential for faster abatement action than
available through grab sampling for those parameters mea-
sured. The water quality manager must make his decisions
with the above facts in mind.
The real basis in this report for comparison between auto-
matic monitoring and grab sampling comes with respect to
cost and spill detection. The 22 stations of the primary
network in Colorado have automatic monitoring costs associated
119
-------
with them. The costs are original equipment purchase and
operation and maintenance. In the state-of-the-art paper,
these costs have been reviewed, but specific costs for
specific situations will of course be different. For the
costs quoted below, average figures have been used.
The cost of installing and operating automatic monitors
at all 22 of Colorado's primary stations will be used to
compare the technique with grab sampling. The purpose of
the primary network is enforcement, and this is where auto-
matic monitoring has its advantages. The effectiveness
levels (spill detection) using automatic monitoring will
be close to 100 percent; therefore, the question breaks
down to one of cost. By determining the total cost involved
for automatic monitors, the trade offs between cost and
^effectiveness can be visualized.
The average of the total yearly costs presented in the
automatic monitoring state-of-the-art paper is $5,930. For
the 22 primary stations in Colorado, this results in an
annual cost of $130,460. Referring back to Figure 23, this
would buy an effectiveness level of approximately 43 percent
using grab sampling. With automatic monitoring, the levels
of effectiveness are much closer to 100 percent for the
prarmeters measured. This indicates that Colorado, for
purposes of abatement and enforcement, would be'getting more
information for the money if automatic monitoring was to be
used. Of course this must be tempered with a waste consti-
tuent analysis in order to determine if automatic monitors
can detect a majority of the wastes emitted.
Another extremely important factor in the above analysis is
the initial costs involved. The above figures are a result
of a long-term view. To implement the 22 automatic monitors
would require an initial investment of $176,000 with strip
chart recorders. Again, this is based on an average figure
of $8,000 per unit obtained from the state-of-the-art paper
on automatic monitoring. However, strip chart recording
removes one of the big advantages of automatic monitoring—
real time data. With a telemetry system and a data logging
device, the initial cost per unit goes to $12,000 for a
total initial cost of $264,000. The use of computers pushes
the initial costs up even higher.
Considering the high initial costs, the establishment of a
complete automatic monitor network must be considered on
the basis of installing a certain number (exact number es-
tablished by budget consideration) each year. The effort
(money and manpower) devoted to the automatic monitor net-
work must not, however, detract from the secondary network
which is also very important in providing data on all param-
eters .
120
-------
The comparisons just made must be considered in the light of
the assumptions necessary to arrive at the figures. The
primary grab sampling network is sampling more parameters
than an automatic monitor is capable of. Consequently, the
spill detection of automatic monitors must be considered in
light of the limited parameters. Also, the cost of grab
sampling includes these extra parameter determinations. The
figures quoted for automatic monitoring are average figures
derived from the literature, and specific costs at specific
geographic locations may vary greatly. Also, many of the
dollar figures quoted in the state-of-the-art paper are
subject to increases due to inflation, etc. This is espec-
ially true with some of the earlier articles. For a more
detailed analysis of the cost-effectiveness for automatic
monitoring, refer to Sylvester (1972) .
Remote Sensing
Remote sensing, as a quantitative data acquisition techni-
que, is the least comparable of the three techniques con-
sidered. This was brought out very clearly in the state-of-
the-art paper when the comparison was made between sight
and the other, more quantitative, senses. However, some
rough analyses can be performed to supply the water quality
manager with an understanding of where remote sensing may
fit into his overall program.
The first and major advantage of remote sensing is the
spatial coverage of surveillance. This can be envisioned
as an almost infinite number of sampling stations on the
streams to be photographed. The sampling frequency can be
as often as desirable, realizing however that the cost of
sampling will be high. The parameters which can be mea-
sured from each sample depend heavily upon the equipment
used. For this reason, remote sensing surveillance is used
more for studying a specific element or elements than for
routinely flying over an area and looking for whatever may
be causing a water pollution problem.
121
-------
In order to perform a cost-effectiveness analysis, however,
it will be assumed that the remote sensor can detect pol-
lution events on a routine flight basis. Also, for purposes
of analysis, it will be assumed that equipment similar to
that at Colorado State University (CSU) is utilized. (CSU's
equipment is listed in the remote sensing state-of-the-art
paper.) It costs approximately $200/day to use this equip-
ment. The plane costs $100/hr, and the operator is paid
$100/day. The plane flies at an average 200 miles per hour.
Flying at 2000 feet with 60 percent overlap, it will take
approximately five pictures to cover each mile. Film, pro-
cessing, and interpretation cost around $10 per picture.
Colorado has 14,503 miles of streams within its boundaries.
Of this total, it will be assumed that only the main stems
and major tributaries will be routinely surveyed by the
remote sensing equipment. This will be estimated at 3600
miles or a total of 18 hours of air time per routine sample.
Eighteen hours of operation will involve a cost of $2400.
The mileage (3600 miles) will require, for pictures, a
cost of $180,000. It can immediately be seen that this
cost is beyond reality for routine sampling. For a special
survey of the entire state, this figure may be more practi-
cal .
In order to create a cost-effective analysis, assume that
only 100 miles are to be surveyed routinely. These would
be the more polluted rivers of the state and the different
reaches will probably be in different parts of the state.
This will involve using the equipment for one-half day or
a cost of $550. For 100 miles, 500 pictures will be needed
for a cost of $5,000. Thus, each routine remote sensing
sample would cost $5,550. Sampling monthly would require
an annual cost of $66,600 for only 100 miles of stream.
If the remote sensor detects every pollution event at the
time of the picture, the detection percentage can be ap-
proximated by dividing the number of events noted by the
number that occur during the month. Theoretically, if
3000 events occur during a month (100 each day) and their
duration is between 0 and 3 days, at most 300 events could
be detected. This yields a percentage of 10 percent.
Comparing the effectiveness percent of 10 percent with the
cost of $66,600, indicates that for a routine basis, re-
mote sensing is expensive. For $66,600, grab sampling is
better than 20 percent effective, and this is for the en-
tire state, not just 100 miles of streams. Also, the 10
percent detection may be high when it is considered that
remote sensing cannot detect every pollutant.
122
-------
The above discussion indicates that remote sensing is very
expensive if it is to be used in the surveillance of water
quality on a routine basis. The cost, therefore, tends to
preclude the use of remote sensing for routine water qual-
ity surveillance. However, the spatial coverage gained
through its use would indicate a distinct advantage in
special surveys or inventory work. The information gained
in this case would be worth the cost. Given that remote
sensing is not practical on a routine basis, the question,
becomes, how would remote sensing fit into a state program
such that its spatial advantage could be utilized.
As noted earlier in the report, the design of a routine
surveillance network id dependent upon adequate information
on stream characteristics and an inventory of existing
waste loading and water use conditions. Since remote sens-
ing can speed the acquisition of this information, its
major contribution to a state agency will be in the area
of special surveys which would then be used to optimize the
development of routine surveillance. This implies that re-
mote sensing, with its great spatial coverage, would serve
to help identify problems, while the other two techniques
would then be used to selectively quantify the situation.
A perfect example of this for Colorado is the identifica-
tion of acid mine drainage areas. Once the sources were
identified, an action program could be started concentra-
ting on the worst offenders first. The program would en-
tail more refinement in the evidence through more quanti-
tative analyses, but the inventory of the problem would have
been greatly simplified.
Before actually discussing the use of remote sensing, the
reliability of the results must be assured. The state-of-
the-art paper indicated the great care needed in using
remote sensing if reliable results are to be obtained.
The survey must be well planned and well executed relative
to agencial goals if the inventory is to supply the data
sought.
Considering the facts presented above, it must be con-
cluded that remote sensing would not be practical for
Colorado in its routine surveillance efforts. The deci-
sion of whether or not to use remote sensing in special
surveys must depend upon an evaluation of the type of data
needed for the survey, an analysis of the ability of_re-
mote sensing to supply the needed data, and the ability
to get the data at a cost that can be tolerated by the
agency involved.
123
-------
SECTION IX
ACKNOWLEDGMENTS
The author wishes to express his sincere appreciation to
the following:
Mr. Dale E. Parke, Chief, Technical Data and Information
Section of EPA in Kansas City, for his support as project
officer and for the constructive criticism in preparing
the report.
Mr. L. Russell Freeman, Director, Pacific Islands Basin
Office of EPA, for assistance with many aspects of the
research and for reviewing the report.
Mr. Gaylord V. Skogerboe, Associate Professor, Colorado
State University, for his help in the execution of the
project.
Mr. Dale H. Vanderholm, Graduate Research Assistant, Colo-
rado State University, for preparation of the grab sampling
state-of-the-art paper, for development of the surveillance
network simulation models, and for the assistance rendered
on many other aspects of the study.
Mr. Marc A. Sylvester, Graduate Research Assistant, Colo-
rado State University, for preparation of the automatic
monitoring state-of-the-art paper and for the support on
other aspects of the work.
Mr. Steven R. Nichols, Graduate Research Assistant, Colo-
rado State University, for preparation of the remote sensing
state-of-the-art paper and for the assistance in the evalua-
tion of the existing water quality data.
Miss Kevin Feigen and Miss Paula C. White, for typing and
assistance in the preparation of the final manuscript.
Mr. Douglas Brophy and Mr. Michael Boetger, for their
help in preparation of many of the tables and figures which
appear in the report.
The author is also indebted to the following organizations
and their staff members for the great amount of information
obtained through personal communication:
Water Pollution Control Division of the Colorado
Department of Health, Denver, Colorado
125
-------
Analytical Quality Control Laboratory, EPA,
Cincinnati, Ohio
Water Quality Surveys Branch, Div. of Sanitary
Engineering, Ontario Water Resources Commission,
Toronto, Ontario, Canada
Division of Pure Waters, New York State Department
of Environmental Conservation, Albany, New York
Bureau of Sanitary Engineering, Pennsylvania
Department of Health, Harrisburg, Pennsylvania
California State Water Resources Board, Sacramento,
California
South Carolina Pollution Control Authority,
Columbia, South Carolina
Division of Environmental Protection, Wisconsin
Department of Natural Resources, Madison, Wisconsin
Ohio River Valley Water Sanitation Commission
(ORSANCO), Cincinnati, Ohio
Denver Regional Office of EPA, Denver, Colorado
126
-------
SECTION X
LITERATURE CITED
Bramer, H.C., "Data Needs in Engineering Design," Paper
presented at the National Symposium on Data and Instru-
mentation for Water Quality Management, Madison, Wiscon-
sin (1970).
Brown, Robert M., McClelland, Nina I., Deininger, Rolf A.,
and Tozer, Ronald G., "A Water Quality Index - Do We
Dare?" Water and Sewage Works, (1970).
Colorado Water Pollution Control Commission, "Stream Clas-
sification for Surface Waters of Colorado," Adopted June,
1967.
Glover, R. F., "Dispersion of Dissolved or Suspended
Materials in Flowing Streams," U.S. Geological Survey
Professional Paper 433B (1964).
Hager, Walter R., "Guidelines for Estimating Personnel
Requirements for State Water Quality Control Agencies,"
Office of State Program Review and Assistance, Environmen-
tal Protection Agency, Washington, D.C. 20242 (1970).
Misback, Gregory, "Comprehensive Survey of Cache La Poudre
River," Prepared as a Special Report of the Colorado Depart-
ment of Health, Water Pollution Control Division, December
(1970) .
Snedecor, G.W., and Cochran, W.G., Statistical Methods,
Iowa State University Press, Ames, Iowa (1967).
Sylvester, Marc A., "Application of Automatic Monitoring to
State Water Quality Surveillance Programs," Unpublished M.S.
Thesis, Department of Zoology, Colorado State University,
Fort Collins (1972) .
Vanderholm, Dale H., "Planning Water Quality Surveillance,"
Unpublished PhD Thesis, Department Agricultural Engineering,
Colorado State University, Fort Collins (1972).
Velz, C.J., "Sampling for Effective Evaluation of Stream
Pollution," Sewage and Industrial Wastes, 22:666-682 (1950).
127
-------
SECTION XI
APPENDICES
Page
A. Grab Sampling: State-of-the-Art 131
B. Automatic Monitoring: State-of-
the-Art 149
C. Remote Sensing: State-of-the-Art .... 191
D. Colorado's Stream Characterizations .... 225
129
-------
APPENDIX A
GRAB SAMPLING: STATE-OF-THE-ART
Grab sampling has traditionally been the sampling techni-
que used in water quality management. Grab sampling is a
term used to describe the process of going to a specified,
spot on a stream, removing a sample of water (grab sample),
carrying the sample back to a laboratory and perform-
ing tests on the sample in order to determine the values
of various parameters used to quantify water quality.
Since the sample is analyzed in a laboratory, there is
essentially an unlimited number of parameters which may be
measured.
As noted in Section VI, there are many articles in the lit-
erature concerning data acquisition in general while there
are few relating specifically to grab sampling. Since
these general articles do include grab sampling, they are
reviewed in the grab sampling section and are not repeated
in the automatic monitoring or remote sensing state-of-the-
art papers.
The following discussion is concerned primarily with rou-
tine water quality surveillance data acquisition; however,
it is recognized that many other data types (special sur-
veys, waste inventories, etc.) are necessary for the suc-
cessful operation of a water quality management organization,
Functional Data Needs
There are many ways in which a water quality management or-
ganization can be divided by functions. Seven functional
objectives were listed at the beginning of this report.
McDermott (1968) states that water quality monitoring is
a support activity for three functions: (1) water quality
standards enforcement and revision; (2) water quality base-
line and trend evaluation; and (3) planning and management
programs. DeFalco (1964) classifies data needs as follows:
1. To determine compliance with a given .standard
2. To determine or forecast the effect of a water
resource project
3. To determine treatment needs in the use of the
water
4. To provide water quality control.
131
-------
How the purposes are grouped is not important. To realize,
however, that the data required varies greatly with the
purpose of collection is important. This should be consid-
ered in designing a quality surveillance system. Some of
the general problems encountered in surveillance system
planning are discussed by Pomeroy and Orlob (1967) . They
state that minimum data needs are dependent upon drainage
area, length of water course, slope, surveillance cycle,
and pollution characteristics, along with other factors.
Sayers (1971) discusses data needs from water quality sur-
veillance. He notes that there are two ways to obtain the
data: 1) through a vast network of strategically located
long-term stations, or 2) through use of repeated short-
term surveys. A combination of the two can supply the
most efficient source of data. Gannon and Wezernak (1967)
say that water quality information should be collected
with a definite purpose in mind.
Moody (1970) has described the data requirements for plan-
ning purposes. These are primarily data for evaluating
base level stream quality conditions and quality trends
and for identifying problem areas. These data are usually
then incorporated into a planning model to aid in the de-
cision-making process. Moody points out errors in the
data itself, or in its analysis, may cause over or under
design and result in loss of economic efficiency for a
project. Petri (1970) notes that planning data is often
insufficient for several reasons, one of which is that
data needs are not anticipated far enough in advance.
Anderson, et al (1968) points out the lack of water qual-
ity trend analysis in the literature and blames this
primarily on a lack of reliable and extensive data of
sufficient time length. Allen (1971) states that no gen-
eral procedures for the establishment of a satisfactory
sampling program applicable to all situations can be stated.
Data collected under a routine surveillance program may
indicate areas where additional study or research is nec-
essary. Research may then be performed by the agency con-
cerned, or contracted outside. As noted above, this point
was brought out by Sayers (1971) .
With respect to data needs for design purposes, Bramer
(1970) notes that since treatment facilities are instal-
led to maintain water quality levels in streams, the
basic water quality data requirements are concerned with
the nature of the receiving stream and the applicable
water quality criteria. These requirements are used to
establish effluent requirements and evaluate the proposed
treatment facility. Receiving stream data requirements
will vary greatly with the stream characteristics and the
v-a,ter usage, according to Bramer.
132
-------
In order to supply data needed for stream standard en-
forcement, a surveillance system must have a high prob-
ability of detecting stream standard violations. If
enforcement of stream standards results in court action,
the data must prove that violations have occurred and
the source responsible. Ballinger (1967) indicates that
the methods used and the quality control employed are
quite important in obtaining data which will withstand
vigorous challenges during legal proceedings. An in-
tensive water quality survey with strict quality control
measures is necessary to obtain data which will meet
these requirements.
Parameter Measurement
McDermott (1968) states that if the primary purpose of
data collection is to determine standards compliance,
then the quality parameters listed in the standards
should be given primary consideration in a surveillance
system. In most instances, the parameters listed in
the standards are far fewer than those available.
McDermott has summarized the frequency of parameter
usage in state standards. He indicates that only nine
of the many available parameters appear in the standards
of all states.
Five of the nine parameters appearing in state stream
standards are termed the "Five Freedoms:" freedom from
the presence of floating solids, settleable solids, un-
natural turbidity or color, unnatural taste or odor, and
toxic substances. The other four are DO, ppH, coliforms,
and temperature. McDermott notes that the first four
freedoms require use of man's senses and judgment as op-
posed to determination by analytical methods solely.
Morgan (1970) discusses parameters of interest for basic
water quality data, particularly with respect to water
usage. He indicates that selection of parameters for
monitoring should be on this basis. '
If long term base level data is the data collection
purpose, it is desirable to measure a wide range of para-
meters, according to McDermott (1968). Parameters not
presently included in water quality standards may be
added to standards at some later date, or may be needed
for current or future planning purposes. McDermott lists
seventy parameters which may be used to quantify stream
water qualtiy. Of these, all but four are listed in
Standard Methods (1965) and can be determined using grab
sampling and traditional wet chemistry methods. Obviously,
grab sampling has a capability of almost one hundred per-
cent with regard to parameter measurement. Ballinger (1967)
133
-------
indicates that only those procedures listed in Standard
Methods have been acceptable for obtaining data for use
in legal proceedings, so far. This implies that inten-
sive surveillance using grab sampling methods is neces-
sary for obtaining legally acceptable data.
Several studies have been conducted on the use of index
parameters for water quality surveillance. The primary
purpose of using such indexes is the possible reduction
of sample analyses necessary to obtain the needed data.
In addition, index parameters may be useful in estima-
tion of missing data points and in determining sampling
schedules. Wang and Evans (1970) used regression analy-
sis and found good relationships using streamflow as
the independent variable and various nutrient concentra-
tions as dependent variables. Durum (1953) also used
streamflow as the independent, or index, variable with
mineral concentrations as the dependent variables.
Gunnerson (1967) tried both streamflow and TDS as in-
dicator variables on a Columbia River study with fairly
good success. Steele (1969) used both streamflow and
specific conductance as independent variables with vari-
ous ionic constituents as dependent variables. He found
specific conductance to be the best indicator parameter,
Notably, all of these studies have selected a parameter
for the independent variable which is easily measured
and for which long term records are commonly available.
Sampling Frequency
The frequency of sampling necessary to obtain data sat-
isfactory for its intended purpose is difficult to de-
fine. Kittrell (1969) says frequency of sampling varies
with water use, the urgency of developing a representa-
tive record of quality, and the capacity of the respon-
sible agency for sample collection and analysis. He
points out that due to logistics, control agencies often
sample at monthly or longer intervals, thereby allowing
many standard violations to occur without being detected
because of such infrequent sampling. Kittrell questions
the practice of routine monitoring at regular intervals
throughout the year and suggests greater economy might
be achieved by limiting the sampling to periods of po-
tential damage. Allen (1971) agrees with Kittrell in
that the number of samples required to describe a nat-
ural water depends on the variability of the constituent
to be analyzed. Allen also notes that too frequently
the samples are divided among too many stations. The
collection of more samples and fewer stations permits
134
-------
statistical analysis of the data with much more reliable
results. Deininger (1971) discusses the "statistics" of
water quality surveillance.
Stream water quality may exhibit diurnal, seasonal, annual,
or other cyclic patterns (Fair, et a^, 1968). In addition,
it also will show random variation from meteorological
and hydrological events. Pollution occurrences may be
random or cyclic in nature. An example of a cyclic type
might be the effluent discharge from a seasonal industry
such as food processing. An accidental pollutant spill
would more likely fall into a random uniform pattern.
A California study (Gunnerson, 1962) points out that grab
sampling will miss, or detect, pollution events by chance.
Quality of an individual stream may range from highly
variable to fairly stable depending upon local hydrologic
and meteorologic conditions, the nature of the stream it-
self, and the contributing pollution sources. McDermott
(1968) states that the variability of the stream quality
dictates to a large extent the frequency of sampling nec-
essary to characterize the stream. In designing a sur-
veillance system, it is desirable to use the minimum
sampling frequency which will supply the necessary infor-
mation. In an Ontario study, Rizvi (1969) used analysis
of variance on historical lake water quality data to
determine minimum sampling frequencies. Since lake con-
ditions were relatively stable, he found that previous
sampling frequencies could often be significantly reduced
with no loss of information. Gunnerson (1966) employed
power spectra analysis to determine optimum sampling
frequency for DO and specific conductance in a tidal es-
tuary. He noted that for study of periodic harmonic mo-
tion, observations must be at less than one-half the wave
period. His studies showed that the optimum sampling
frequency for the parameters studied was about two hours.
Moody (1970) also discusses the use of power spectra analy-
sis for determining sampling frequency. Thomann (1967)
used time series analysis to study the periodicity of
DO and temperature in an Delaware estuary. All of these
studies have been concerned with short term periodic vari-
ation (e.g., diurnal) for which very frequent observations
are necessary. Obviously, routine surveillance by grab
sampling is impractical for this purpose.
Steele (1970) investigated the effects of sampling fre-
quency using regression analysis. He simulated short
term quality variations by use of their relations to
index variables as described earlier. The simulated
data was then compared to observed data. He notes that
time-dependent variations may be masked by this method.
135
-------
Pomeroy and Orlob (1967) present minimum surveillance re-
quirements for streams. They recommend the minimum num-
bers of stations and sampling frequencies for given situa-
tions based primarily on stream slopes and flow variability,
They do not explain how these recommended numbers were
obtained. Weisbecker, et a.1 (1970) propose a methodology
for planning monitoring programs assuming that preliminary
data such as water usage, base level quality and important
parameter data are available. They recommend sampling
frequencies for various parameters based on the variation
expected and the importance of the parameter.
Sampling Station Locations
Sample site selection is, in most cases, somewhat arbi-
trary. Sites should be located at the most representative
points, according to Ball (1970). Kittrell (1969) states
that sites should account for present and potential pol-
lution sources, water usage, and physical stream character-
istics. Kittrell and West (1967) emphasize that stations
should be located to obtain the most representative data,
rather than for convenience. A general discussion on
representative sampling is given by Roskopf (1968), who
states that there is no ideal time and place for sampling
for all purposes. Velz (1950) discusses several factors
to consider in station location. He notes that stations
should not be located immediately below tributary mouths,
or pollution sources. Colorado stations are based on
water quality standard change points, among other con-
siderations (Misbach, 1971). At least one study has
been conducted using statistical techniques to analyze
station locations. Palmer and Sato (1968) used multiple
and pair testing on data from lake sampling stations in
Ontario to test similarity between station results. They
found that stations could be reduced to prevent duplica-
tion of data and increase efficiency. McDermott (1968)
feels that a minimum of sampling stations with a high
sampling frequency will provide the best data. He also
states that the adequacy of a sampling schedule should
be judged by the ninety-five percent confidence interval
for correctly defining the stream quality. Obviously,
a very high sampling frequency would be necessary to meet
this criterion.
General Surveillance Design Studies
A few authors have described general procedures for design
of water quality surveillance systems.- Roche (1970)
136
-------
emphasizes the need for cooperative effort among concerned
agencies to achieve an effective monitoring program.
Kittrell (1969) discusses general guidelines for estab-
lishing a surveillance system. Haney and Schmidt (1958)
point out the importance of proper planning and present
several points to be considered in sampling program de-
sign.
Recommendations were made regarding a complete surveil-
lance system for the Sacramento River in a California
study (Gunnerson, 1962) but the basis for some of the
recommendations, such as sampling frequencies, was not
made clear. A systems analysis approach was used in a
study conducted for FWQA (1970) . The objective of this
study was to develop a general method for the design
of surveillance systems for major river basins. The
selection of parameters to be measured was primarily
based on the capabilities of automatic monitors and on
stream standards at the particular points sampled. A
sampling frequency of once or twice monthly, depending
upon location, was recommended for those parameters not
continuously monitored. The report states that these
frequencies are based on the needs to protect water uses,
as well as to provide information necessary for future
evaluation. However, the report does not explain how
the sampling frequencies were selected on this basis.
Surveillance stations were located on the basis of their
relation to (1) interstate borders, (2) potable water
supplies of major population centers, (3) major pollution
sources, and (4) tributary streams.
Several papers describe and recommend increased use of
the STORET system for data handling and retrieval. A
basic description of the system and its use is found in
a publication published by FWPCA (1966). Taylor (1970)
has explained some of the STORET characteristics and
mentions some experiences with its use.
Sampling Procedures
As noted earlier, the design of a grab sampling surveil-
lance network involves a determination of station location,
frequency of sampling and parameters of consequence. Once
the above design criteria are established, implementation
becomes a crucial factor. Implementation involves the
establishment of sampling routes, field measurement of
certain parameters, sample collection techniques, sample
preservation and storage techniques, and shipping proce-
dures to assure rapid laboratory analysis of the sample.
In general, these can be referred to as sampling procedures
137
-------
Kittrell (1969) in the preface to his guide to water qual-
ity surveys, notes that Standard Methods (1965), a 769-
page volume, has been prepared to guide the analysis and
examination of water samples, but that there is very little
published information which can be used to guide the ac-
quisition of samples that represent stream conditions.
He indicates that the variability that exists from stream
to stream may be the major reason for this. Kittrell hopes
that his discussion of the subject will help fill this
need.
Although Standard Methods (1971) is concerned mainly with
analysis and examination of water samples, it does very
briefly discuss sampling procedures. Rather than repeat
all the facts here, just some high points will be noted.
In the actual sampling, no attempt should be made to use
the same sample for chemical, bacteriologic, and micro-
scopic examinations. The methods of collection and hand-
ling are too different to permit this. Standard Methods
(1971) reviews the differences at the beginning of each
major section and then notes particular aspects of samp-
ling and storage associated with each parameter in that
section.
In general, the laboratory analysis of a sample will be
more reliable for those samples that have the shorter
time elapse between collection and analysis. No hard
and fast times may be stated as to time elapse and effect
on the sample. The effect depends upon the character of
the sample, the particular analysis, and the conditions
of storage. Standard Methods (1971) suggests the follow-
ing as maximum limits for samples for physical and
chemical analysis:
Unpolluted Waters ." 72 hours
Slightly Polluted Waters . . . .48 hours
Polluted Waters 12 hours
For bacteriological examinations, it is recommended that
samples be analyzed within one hour of collection. If
this is not possible, the sample should be iced and in
transport for not more than six hours. If times are
greater than this, alternatives should be considered.
In all cases the time elapsed between sample collection
and sample analysis should be recorded.
Certain constituents and physical values require field
analysis in order to obtain accurate results. Dissolved
gases (oxygen, carbon dioxide) may be gained or lost;
temperature can change rapidly; pH may change signifi-
cantly in a few minutes. Because of this, these
138
-------
parameters must be measured in the field. There is a
developing technology revolving around portable meters
for measuring pH, temperature, dissolved oxygen, speci-
fic conductance, etc. Scott (1964) discusses the de-
velopment of an instrument to read dissolved oxygen.
Just six years later, Stack (1970) indicates that mea-
surements of water quality have been greatly facilitated
by portable instrumentation. He bemoans the fact that
there is no compact multiparameter kit to make field
measurements much easier. He, therefore, proposes and
presents a multiparameter kit but indicates that the
design of such a system should include careful evalua-
tion of prarmeters to be measured and the accuracy re-
quired for various uses of the data.
Ficken (1970) presents a state-of-the-art paper on sensors
used to measure electrical conductivity, temperature,
dissolved oxygen, pH, turbidity, and specific ions. He
notes that considerable time is spent cleaning and check-
ing calibration of sensors. Mancy (1971) also gives a
very thorough discussion of sensors and instrumentation.
Grab sampling usually refers to the obtaining of a sample
at a single point in space and time. Various references
(Kittrell, 1969; Roskopf, 1968; USGS, 1960) describe pro-
cedures such as the compositing of samples and the use
of depth integrating samplers to obtain more representa-
tive samples. Ball (1970) discusses errors to avoid in
sample collection. Among these are errors in site selec-
tion, sample collection and field measurements, and
sample preservation and storage. He describes a cross
sectional sampling procedure used by the Bureau of Recla-
mation for determining the most representative sampling
sites.
Haney and Schmidt (1958) and Velz (1950) both elaborate
on representative sampling. They mention the importance
of understanding the major factors that affect stream
sampling before one converts a survey strategy into a
tactical plan of implementation. The major factors are:
1) sampling as related to the daily hydrograph, 2) samp-
ling in relation to sources of pollution and tributaries,
3) sampling in relation to physical characteristics and
river developments, 4) sampling in relation to waste
discharge, and 5) abnormalities in natural purification.
Haney and Schmidt (1968) also mention that composite
sampling has its greatest usefulness in connection with
correlative studies of pollution loads. They also note
that composite samples automatically mask quality abnor-
malities in the stream being sampled. It is often these
variations that are of importance. Tarazi and others
139
-------
(1970) discuss the advantages and disadvantages of auto-
matically obtaining daily composite samples of flow-
proportioned aliquots from a stream. They also discuss
a second technique which consists of automatically taking
three grab samples daily.
If samples cannot be analyzed within the prescribed maxi-
mum time limits, it will be necessary to utilize preser-
vatives and adequate storage conditions. Standard Methods
(1971) and FWPCA (1969) both give instructions for preser-
vation and storage of samples for particular unstable
constituents. Kittrell (1969) notes that much research
is needed to develop the most reliable preservation methods
for many constituents. Schaumburg (1971) discusses a new
concept in sample preservation—poisoning and depoisoning.
For more concentrated studies, there is an assortment of
sampling equipment available. There are automatic com-
posite samplers and automatic grab samplers, both refrig-
erated and non-refrigerated. For routine grab sampling
surveillance where only one sample per month is collected,
there is little need for this equipment. For the manual
collection of one grab sample, there is equipment avail-
able. Both Standard Methods (1971) and Kittrell (1969)
describe special devices to collect a sample for DO de-
terminations. They describe an Ohio-type sampler and
the Juday bottle (and some of its modifications, the
Kemmerer or Van Dorn). These devices also may be used
to collect samples for other determinations .
In the collection of any sample, the knowledge, skill,
and concern of the collector is crucial. Kittrell (1969)
also emphasizes the point that the person analyzing
the data and making interpretations from the data should
also be familiar with the sampling locations and pro-
cedures. There is very little published information
relating to manpower requirements for a routine grab
sampling network. The manpower level would depend upon
the surveillance effort and design. FWQA (1970) presents
the manpower levels of all the state agencies. One of
the classifications is pollution control surveillance
and enforcement.
Estuary Sampling
Estuaries present very complex problems in terms of water
quality. Their hydrologic and hydrodynamic characteristics
are complex and the mixing of fresh and saline waters
compounds this complexity. Considering that many estuary
shores are industrial centers producing a wide variety of
pollutants, one sees that the problems can become enormous.
140
-------
The Delaware River Estuary has been the subject of con-
siderable study. O'Connor, e^ a^ (1968) indicate that
several agencies sample this estuary regularly, but give
no background as to the planning of the sampling program.
Thomann (1967) used data from the Delaware River Estuary
in his study of time series analysis of water quality
data.
The problems involved in estuary water quality surveil-
lance are discussed by Pomeroy and Orlob (1967). Among
the factors they state should be considered are: 1) type
of quality constituent; 2) size of estuary; 3) shape of
estuary; 4) relation between runoff, tidal action, and
mixing potential; 5) degree of stratification; 6) qual-
ity, hydrologic, and hydrodynamic cycles; and 7) periodic
versus random phenomena. They present a set of minimum
surveillance requirements for various types of estuaries.
These recommendations are based on what the authors term
the minimum surveillance to obtain statistically signi-
ficant data. They do not, however, describe any of the
methods used to arrive at these numbers, other than to
state that at least six samples per cycle are necessary
to characterize a periodic phenomenon.
Gunnerson (1966) used time series analysis to determine
optimum sampling intervals in tidal estuaries. His
study was primarily concerned with short term periodic
phenomena, although the method can be applied to any
time period for which adequate data is available. The
use of spectral analysis for analyzing water quality
data from streams and estuaries is described by Wastler
(1963) .
Gibbs and Isaac (1968) discuss the water quality moni-
toring program for the Duwamish Estuary near Seattle,
Washington. They state that manual sampling was con-
sidered impractical in this situation due to complexities
of the estuary system. They also note that manual samp-
ling is used for parameters not available from automatic
monitors.
Velz (1950) notes that the semi-daily tidal cycle is a
major factor controlling estuary quality, necessitating
sampling at frequent intervals (one to two hours). He
states that the tidal influence tends to prevent sudden
changes in water quality but that random sampling is of
little value. Therefore, frequent, controlled sampling
is required to properly define estuary quality conditions
141
-------
Costs
Costs of water quality data collection vary with the
methods used and with locale, but it is generally ac-
cepted that costs are high. Petri (1970) states that
one of the basic reasons for insufficient planning
data, in many instances, is the high cost of sample
collection, laboratory analysis, and data reporting.
Gunnerson (1962) feels that four man-days of handling
and evaluation are needed for every one man-day of
data collection, not including laboratory requirements.
De Falco (1964) gives the cost of manual surveillance
as $25 to $50 for sample collection and the same for
laboratory analysis. ORSANCO records (Klein, et al,
1968) indicate that the cost of manual sampling and
analysis in that area is about $2.20 per data item.
Since the term "data item" is not defined by Klein,
it is assumed to be the value of one parameter from one
sample. Maylath (1970) states that the costs for manual
sampling and analysis in New York State are about $6.00
per parameter value. The term "parameter value" is
assumed to be the same as "data item" used above.
Manual sampling is the only method used presently in
Colorado for routine quality monitoring. The actual costs
for this program in Colorado during the 1970-71 fiscal
year were assembled by Misbach (1971) and used to calcu-
late costs per sample. His figures show these values:
sample collection, $21.40 per sample; chemical, bacterio-
logical, and radiological analyses, $34.80 per sample;
data handling and analysis, $10.18 per sample. These
total up to a cost of $66.38 per sample. The Colorado
Water Pollution Control Division shares laboratory faci-
lities with the state Public Health Department and con-
tributes an amount of financial support proportional to
the work load due to analysis of water samples. This
includes equipment costs. For this reason, Misbach (1971)
indicates that laboratory analysis costs were based on
their contribution rather than actual compilation of
laboratory costs.
There is not enough information given on the basis of
these costs for routine surveillance to allow direct
comparison, or to give reasons for the variation. The
point is made by almost all authors that due to the high
costs involved, proper planning of surveillance systems
is very important.
142
-------
Literature Cited
Allen, H. E., "Design of Sampling Programs," in Instru-
mental Analysis for Water Pollution Control by K. H.
Manoy, Ann Arbor Science Publ., Inc., Ann Arbor, Michi-
gan (1971) .
Anderson, P. W., Faust, S. D., and Zogorski, J. S.,
"Analyses of Long Term Trends in Water Quality Para-
meters," Proc. Fourth American Water Resources Conf.,
New York, N.Y. (1968).
Ball, M. D., "Errors to Avoid in Water Quality Col-
lection and Sampling," Paper presented at the National
Symposium on Data and Instrumentation for Water Qaul-
tiy Management, Madison, Wisconsin (1970).
Ballinger, D. G. , "Effective Water Pollution Control
requires Accurate Data," Environmental Sci. and Tech.,
No. 1, pp 612-616 (1967).
Bramer, H. C., "Data Needs in Engineering Design,"
Paper presented at the National Symposium on Data and
Instrumentation for Water Quality Management, Madison,
Wisconsin (1970) .
Deininger, R., "Analysis of Data," in Instrumental
Analysis for Water Pollution Control by K. H. Manoy,
Ann Arbor Science Publ, Inc., Ann Arbor, Michigan (1971)
De Falco, P., Jr., "Data Acquisition Systems in Water
Quality Control," in Symposium, Environmental Measure-
ments, Valid Data and Logical Interpretation, U.S.
Public Health Service Pub. No. 999-AP-15 (1964).
Durum, W. H., "Relationship of the Mineral Constituents
in Solution to Stream Flow, Saline River near Russel,
Kansas," Trans. American Geophysical Union, No. 34,
pp 435-442 (1953) .
Environmental Protection Agency, Water Quality Office,
"Guidelines for Water Quality Management Planning,"
Government Printing Office, Washington, D.C. (1971).
Fair, G. M., Geyer, J. C., and Okun, D. A., Water and .,
Wastewater Engineering, Volume 2, Water Purification
and Wastewater Treatment and Disposal, John Wiley and
Sons, Inc., New York, N.Y. (1968).
Federal Water Pollution Control Administration, "FWPCA
Methods for Chemical Analysis of Water and Wastes,"
U.S. Department of the Interior, FWPCA, Division of
Water Quality Research Analytical Quality Control Lab-
oratory, Cincinnati, Ohio (1969).
143
-------
Federal Water Quality Administration, FY 1970 State
Program Plans," U.S. Department of Interior, FWQA,
Washington, D.C. (1970).
Ficken, J. H., "Sensors—State-of-the-Art," Paper
presented at the National Symposium on Data and Instru-
mentation for Water Quality Management, Madison, Wis-
consin (1970) .
Gannon, J. J., and Wezernak, C. T., "Collection and In-
terpretation of Stream Water Quality Data," Proc.
Third American Water Resources Conf., San Francisco,
California (1967).
Gibbs, C. V., and Isaac, G. W., "Seattle Metro's
Duwamish Estuary Water Quality Program," J_._ Water
Pollution Control Fed., No. 40, pp 385-394 (1968).
Glover, R. F., "Dispersion of Dissolved or Suspended
Materials in Flowing Streams," U.S. Geological Survey
Professional Paper 433 B (1964) .
Gunnerson, C. G., "Optimizing Sampling Intervals in
Tidal Estuaries," J. San. Engr. Div. ASCE, No. 92,
pp!03-125 (1966).
Gunnerson, C. G., "Sacramento River Water Pollution
Survey," California Department of Water Resources
Bill, No. Ill, Sacramento, California
Gunnerson, C. G., "Streamflow and Quality in the
Columbia River Basin," J^ San. Engr. Div. ASCE, No. 93,
pp 1-16 (1967) .
Haney, P. D., and Schmidt, R., "Representative Samp-
ling and Analytical Methods in Stream Studies,"
Sewage and Ind. Wastes, No. 30, pp 812-820 (1958) .
Kittrell, F. W., "A Practical Guide to Water Quality
Studies of Streams," U.S. Department of the Interior,
FWPCA Publication CWR-5 (1969) .
Kittrell, F. W., and West, A. W., "Stream Survey
Procedures," J^ Water Pollution Control Federation,
NO. 39, pp 627-641 (1967).
Klein, W. L., Dunsmore, D. A., and Horton, R. K., "An
Integrated Monitoring System for Water Quality Manage-
ment in the Ohio Valley," Environmental Science and
Technology, No. 2, pp 764-771 (1968).
144
-------
Mancy, K. H., Instrumental Analysis for Water Pol-
lution Control, Ann Arbor Science Publ., Inc.,
Ann Arbor, Michigan (1971).
Maylath, R. E., "Water Quality Surveillance in New
York State," New York State Department of Environ-
mental Conservation, Albany, N.Y. (1970).
McDermott, J. H. , "Water Quality Surveillance in
Water Resource Development and Control," Paper Pre-
sented at the Pan American Health Organization Water
Quality Control Laboratory Symposium, Rio De Janeiro,
Brazil, September (1968).
Misbach, G., Private Communication, Data on Colorado
Sampling Station Locations, Denver, Colorado (1971).
Misbach, G., Private Communication, Sampling Cost
Data for Colorado, Denver, Colorado (1971).
Moody, D, W., "Organizating a Data Collection Program
for Water Resources Planning, Development, and Manage-
ment," Paper Presented at the National Symposium on
Data and Instrumentation for Water Quality Management,
Madison, Wisconsin (1970).
Morgan, P. V., "Basic Data Requirements to Evaluate
Water Pollution in Quality Control Programs," Paper
Presented at the National Symposium on Data and
Instrumentation for Water Quality Management, Madi-
son, Wisconsin (July 1970) .
O'Connor, D. J., St. John, J. P., and DiTor, D. M., .
"Water Quality Analysis of the Delaware River Estuary,"
J. Sanitary Engr. Div. ASCE, No. 94, pp 1225-1252 (1968).
Palmer, M. D., and Sato, G. K., "Review of Water Qual-
ity Station Density Statistical Procedures," Ontario
Water Resources Commission, Toronto, Ontario (1968).
Petri, L. R., "Need for Water Quality Data in Plan-
ning Water Resource Developments," Paper Presented
at the National Symposium on Data and Instrumentation
for Water Quality Management, Madison,) Wisconsin
(July 1970) .
Pomeroy, R. D., and Orlob, G. T., "Problems of Set-
ting Standards and of Surveillance for Water Quality
Control," California State Water Quality Control Com-
mission Pub. No. 36, Sacramento, California (1967).
145
-------
Rizvi, S. S. A., "Statistical Approach to Water Qual-
ity Monitoring Programs," Ontario Water Resources
Commission, Toronto, Ontario (1969) .
Roche, W. M., "Organizing a Data Collection Program,"
Paper Presented at the National Symposium on Data
and Instrumentation for Water Quality Management,
Madison, Wisconsin (July 1970) .
Roskopf, R. F., "A Composite-Grab of Water Pollution
Control Sampling," J_._ Water Pollution Control Feder-
ation, No. 40, pp 492-498 (1968).
Sayers, W.' T., "Water Quality Surveillance: The
Federal-State Network," Environmental Science and
Technology, No. 5, pp 114-119 (1971) .
Scott, R. H., "Miniaturized Dissolved Oxygen Field
Kit," J_._ Water Pollution Control Federation, No. 36,
pp 342-4 (1964).
Schaumberg, Frank D., "A New Concept in Sample Preser-
vation—Poisoning and Depoisoning," J. Water Pollution
Control Federation, N .. 43, pp 1671-1680 (1971).
Standard Methods for the Examination of Water and Waste-
water, 12th Ed., American Public Health Association,
Inc., New York, N.Y. (1965).
Standard Methods for the Examination of Water and Waste-
water, 13th Ed., American Public Health Association,
Inc. (1971).
Stack, Keith V., "A Portable Multiparameter Water
Quality Kit," Water Pollution Control Federation
Journal, No. 42, pp 1395-1399 (1970).
Steele, T. D., "A Technique for Improving Estimates
of Long-Term Chemical Quality Characteristics of
Streams," Paper Presented at the National Fall'Meet-
ing, American Geophysical Union, San Francisco, Cali-
fornia, December (1969).
Steele, T. D., "Beneficial Uses and Pitfalls of His-
torical Water Quality Data," Paper Presented at the
National Symposium on Data and Instrumentation for
Water Quality Management, Madison, Wisconsin (July
1970).
Tarazi, D. S., Hiser, L. L., Childers, R. E., and
Boldt, C. A., "Comparison of Wastewater Sampling
Techniques," J. Water Pollution Control Federation,
No. 42, pp 708-730 (1970) .
146
-------
Taylor, P. L., "Experiences with a Water Pollution
Control Storage and Retrieval System (STORET)," Paper
Presented at the National Symposium on Data and Instru-
mentation for Water Quality Management, Madison, Wis-
consin, July (1970) .
Thomann, R. V., "Time Series Analysis of Water Qual-
ity Data," J. San. Engr. Div. ASCE, No. 93, pp 1-23
(1967) .
U.S. Department of the Interior, FWPCA STORET II,
"Storage and Retrieval of Data for Open Water and
Land Areas," Government Printing Office, Washington,
D.C. (1966).
U.S. Department of the Interior, FWQA, "Design of
Water Quality Surveillance Systems," Government
Printing Office, Washington, D.C. (1970).
U.S. Geological Survey, "Methods for Collection and
Analysis of Water Samples," USGS Water Supply Paper
1454, Government Printing Office, Washington, D.C.
(1960) .
Velz, C. J., "Sampling for Effective Evaluation of
Stream Pollution," Sewage and Industrial Wastes,
No. 22, pp 666-682 (1950).
Wang, W., and Evans, R. L., "Dynamics of Nutrient
Concentrations in the Illinois River," J. Water Pol-
lution Control Federation, No. 42, pp 2117-2123 (1970)
Wastler, T. A., "Application of Spectral Analysis to
Stream and Estuary Field Surveys - I. Individual Power
Spectra," U.S. Public Health Service Publication No.
999-WP-7, Government Printing Office, Washington, D.C.
(1963) .
Weisbecker, L. W., Mackin, J. L., Knight, A. W., and
Brocksen, R. W., "An Environmental Monitoring Program
for the Sacramento-San Joaquin Delta and Suisun Bay,"
Stanford Research Institute, Menlo Park, California
(1970).
147
-------
APPENDIX B
AUTOMATIC MONITORING: STATE-OF-THE-ART
A concise definition of automatic monitoring is, "continu-
ous multiparameter measurement of water quality character-
istics" (Ballinger, 1968). The design specifications of
systems fulfilling the above requirements are described by
Mentink (1968). In general, such a system requires three
functional modules: 1) flow chamber; 2) analyzer; and 3)
output component.
Immersion monitors do not require a flow chamber as analyzer
and output components are submerged in the water body being
sampled. For the same reason, support equipment (shelter
and pump) is also not necessary.
Support equipment is necessary unless an immersion monitor
is employed. Usually a shelter is needed to house the
monitor and provide environmental protection. Also, a posi-
tive displacement submersible pump is required(A Program
Guide to Automated Instrumentation for Water Pollution Sur-
veillance, 1966) .
Operation of a system entails sample collection by submer-
sible pump and intake line, analysis by parametric sensors
located in the flow chamber, signal conditioning in the
analyzer phase, and data recording and/or transmission in
the output module (Cleary, 1967, pp 201-202).
A fair amount of flexibility is allowed in the choice of
recording mode. The methods available include on-site
strip chart recording, punched paper tape, or magnetic
tape (Smoot, 1970; Anderson, 1970). Recording format can
be either analog or digital (Mentink, 1968). When tele-
metry is employed, output devices are located at a central
receiving station containing the computer. The same choice
of recording mode is available except that here it is
advisable to perform teletype log sheet display in addition
to other data logging. Thus, rapid detection of water qual-
ity changes as well as data storage for later statistical
analyses is provided (A Program Guide to Automated Instru-
mentation for Water Pollution Surveillance, 1966).
Telemetry requires transmitter components, communications
link, interface, computer, and output devices (Anderson,
1970? Metink, 1966). The transmitter module is composed
of a programmer which sequences sensor recording, an input
addresser which assigns sensor signals to the appropriate
149
-------
memory storage module, and an output addresser which calls
for transmission of values contained in the memory storage
modules. The communications link may be telegraph, radio,
microwave or a combination of these (Smoot, 1970). The
interface contains the receiver which accepts signals from
the monitors and converts these for display or recording
(Anderson, 1970; Metink, 1966). For a discussion of the
computer facilities available, see Anderson (1970) and
Burroughs Corporation Product Brochures on electronic data
processing systems.
Like data recording, telemetry may employ either analog or
digital formats for data transmission (Shubrooks, 1968).
Also, a telemetry system may be adapted for simultaneous
two-way communication (duplexing) (Smoot, 1970). Two-way
communications allow functional commands (pump and auto-
matic sampler activation, A.C. power control, etc.) to be
sent to the monitor from a central receiving station. Thus,
computer control over the operation of the entire telemetry
and monitoring network is possible (Anderson, 1970) .
There are two approaches to continuous measurement of water
quality characteristics. One employs electro-chemical probe
type sensors while the other utilizes automated wet-chemical
analyses (Ballinger, 1968; Jones & Joyce, 1960). For a
discussion on the basics of operation of electrochemical
probes, see Ficken (1970).
Most of the following discussion on the application of auto-
matic monitoring systems will be restricted to electrochemical
probe type monitors because present wet-chemical monitors
are not suitable for field use requiring unattemded operation
(O'Brien & Olson, 1970; Ballinger, Personal Communication,
1971; Maylath, 1970) .
Historical Review of the Use of Automatic Instrumentation
for Water Pollution Control
The following discussion on the application of automatic
monitors is divided according to level of government. The
federal, interstate, and state applications have been aimed
primarily toward stream or estuary surveillance while the
municipal applications have centered around influent and
effluent monitoring. Accordingly, the discussion of the
various types of automatic monitors is broken down by appli-
cation. If the reader is interested in effluent monitoring
only, the discussion can be found under municipal applications.
150
-------
Development of Instrumentation
Initial development of automatic monitoring instrumentation
was undertaken by various industries to provide a means of
process control (Marks, 1966). Beginning in the mid-1940's
with the introduction of the laboratory pH meter for on-
line use, many laboratory instruments have been adapted for
process stream monitoring (Considine, 1965). A list of
available process stream monitors and their application in
various industries is given by Kehoe (1965).
Federal Applications
Actual field use of automatic instrumentation did not take
place until the mid-1950's. Much of the initial work on
the application of automatic monitors to measure water
quality characteristics was performed by the U.S. Geological
Survey (USGS). Initial systems applied in 1955 were only
capable of measuring one or two water quality characteris-
tics: conductivity and/or water temperature. Strip chart
recording was standard which required manual extraction of
the data. Because this process was very time consuming
and costly, conversion to punch paper tape and/or telemetry
was commenced in 1965. Both methods have the advantage
that recorded data is computer compatible; that is, data
recorded is ready for computer processing without prior
manual manipulation. Most of the some 300 automatic moni-
toring stations now operated by the USGS employ this pre-
ferred type of data recording (Blakey, 1970).
One of the first applications by the USGS of automatic
monitors was made on the Delaware Estuary. In 1955,
conductivity recorders were employed to aid in a study of
salinity fluctuations in the estuary. Because of the
rapid fluctuations in other water quality characteristics
besides salinity, temperature, pH, DO,' and turbidity re-
corders were later added. Eventually, an automatic monitor-
ing network was set up corresponding to the already estab-
lished grab sampling system (McCartney & Beamer, 1962) .
The automatic system was only meant to supplement monthly
grab sampling, but it was assumed that the monitor data
would be useful for detecting and alerting downstream users
of the occurrence of accidental spills. Also, it was hoped
that the automatic data could be used for predicting future
water quality trends and for more effective checking and
evaluating of municipal and industrial waste control efforts
(Anon., Wastes Engineering, 1961).
In 1963, attempts were made to improve on this system by
evaluating an electronic "Sentinel" analyzer manufactured
151
-------
by Honeywell. Results of tests were favorable and indicated
that the "Sentinel" was reliable and relatively trouble-free
(Keyser, 1964). Similar results were claimed by Honeywell
by testing a "Sentinel" unit on the Schuylkill River, trib-
utary to the Delaware (Anon, Chemistry and Engineering News,
1963). Advantages of the Honeywell monitor were that the
pH, DO, T, cond, and turbidity sensors were all contained
in one cabinet, two new sensors were added, air temperature
and solar radiation, and punched paper tape recording was
possible. Further testing resulted in improving the "senti-
nel" until it was capable of measuring chlorides, oxidation
reduction potential, stage, and wind direction in addition
to the before-mentioned parameters (Keyser, 1964).
Another early USGS installation was located on the Patuxent
River Estuary in Maryland. In 1963, a Honeywell "sentinel"
was installed here to help determine the effects thermal
pollution might have on the biological life of the estuary.
Because of the comprehensive nature of the investigation
requiring a large number of samples over 24-hour periods,
automatic monitoring was considered of primary importance
for water quality data acquisition. Rapid and variable
fluctuations in water quality required temperature profiles
and continuous measurement by conductivity, DO, temperature,
turbidity, stage and wind speed and direction (Cory & Davis,
1965). The grab sampling program was not scrapped and
proved beneficial because of the correlation of automatic
data with spot samples taken elsewhere in the estuary
(Nauman & Cory, 1970). Since only one "Sentinel" was employed,
location was very important. Thus, the monitor was installed
at the center pier of Benedict Bridge in a small laboratory.
Since a continuous record of water quality data was required,
strip chart recording was chosen (Cory & Davis, 1965) . A
good account of the operating experience, reliability of data,
and cost of this automatic system for a six-year period from
1963-1969 is given by Nauman and Cory (1970).
Other agencies within the Department of Interior have also
been interest in automatic water quality surveillance. The
Environmental Protection Agency (EPA) has made extensive use
of automatic monitors. As of 1969, it had 56 automatic
water quality data acquisition systems in use (McDermott,
Ballinger, and Sayers, 1968).
The Potomac system is one of EPA's biggest ventures in auto- '
matic monitoring. Data necessary to develop and verify a
mathematical model of the Potomac River and Estuary could
not be supplied except by the use of automatic monitors.
Thus, an integrated automatic system consisting of four stations
was established. Data on DO, temperature, and chlorides
required by the model was telemetered to EPA headquarters
152
-------
in Washington, B.C. By testing alternate programs for water
pollution control, the model will be helpful in Potomac
W^Sf <2uality management (McDermott, Ballinger, and Savers,
1968) .
EPA has also operated an automatic system in the New York
Harbor area since 1963. Monitoring is performed to obtain
a historical record on the water quality of the Raritan
River and Estuary. The Hudson-Delaware Basins Office, EPA
feels this data will be necessary for water quality models
of the estuary (Bromberg & Carames, 1970).
The historical record was originally provided by a network
of four monitoring stations. A fifth station was added in
1969. All stations measure pH, conductivity, temperature,
and DO. Some also measure turbidity, ORP and/or solar radi-
ation, as well as wind velocity and direction. On-site
punched paper tape was initially employed to record the data.
However, due to poor performance resulting in over 50 per-
cent loss of usable record, a telemetry system was installed
in 1967.
Experience with the telemetry system has been much more
favorable with a 91 percent usuable record obtained. Tele-
type is used to transmit data from the monitors to the
central receiving station where data is displayed on type-
written log sheets and recorded on punched paper tape. Daily
statistical summaries are preapred from the punched paper
tape and then data from the punched paper tape is transferred
to magnetic tape for storage and future analyses. Since
samples are analyzed from three depths, data retrieval is
fairly complicated. A complete interrogation of each sta-
tion takes 45 minutes with values from one level being trans-
ferred every 15 minutes.
Operation of an automatic monitoring network in an estuarine
environment was particularly difficult due to additional
maintenance problems produced by increased salinity. Tri-
level sampling , although possibly valuable for obtaining
a more representative picture of estuarine water quality,
produced additional maintenance problems because three sub-
mersible pumps and a valve network were required at each
station to keep samples separate. Pumps and valve manifolds
had to be rebuilt or replaced every three months due to salt
water corrosion. Shelters housing monitors suffered a
similar corrosion and had to be replaced every three years.
In addition, salt bridging on monitor sensors produced a
greater than normal electrical-mechanical drift. Hence,
routine washing with distilled water and increasing the
distance between sensors and their electrical connections
153
-------
were executed. To perform the necessary servicing a main-
tenance crew of three technicians was required (Bromberg &
Carames, 1970}.
Mobile monitors have also been employed by the EPA to survey
water quality. A mobile laboratory used in the Hudson-
Delaware Basin was designed with the capability of operating
monitors in remote locations. Either probe or wet-chemical
type analyzers can be incorporated into the lab. Since this
laboratory was designed with multi-utility in mind, monitoring
equipment is not always carried. When sucn instrumentation
is employed, it is effectively strapped in place by elastic
lash cords. The lab does not require on-site utilities as
it carries its own power and water supply. Thus, monitors
may be operated for extended periods in remote locations
(Dewling, 1969).
In Oregon, trailer-mounted automatic monitors have been in
operation since 1968. Two trailers using Schneider Robot
Monitors have been operated seasonally at three locations:
1) on the Columbia River at Bradwood, Oregon
2) summer use on the Willamette River
3) and winter use on the Snake River at Milner
Dam in Idaho
Both units are capable of measuring pH, conductivity, DO,
and temperature. Experience has shown that these monitors
obtain very good accuracy (one percent of full scale) . Most
of the problems are associated with the support equipment.
Since trailers are not designed with self-contained utilities,
site locations are restricted to areas where power lines or
some other power source is available. It has been found
that power lines must also be close to the river sampled
because of voltage loss due to line resistance. Also,
intake pipelines must be kept to a minimum length due to
possible change in water quality induced by passage through
an extensive length of pipe. As usual, the submersible
pump presented the biggest headaches and maintenance expense.
The only problem with the actual monitoring assembly was
due to DO probe fouling and aging. A partial solution
was found by purchasing DO probes with automatic cleaning
devices and larger electrolyte reservoirs (Gary O'Neal,
personal communication, 1971).
A unique approach to mobile monitoring has been tried on the
Mississippi River. The Honeywell monitors employed are
installed in specially equipped boats. Conductivity, DO,
pH, and temperature are measured continuously in order to
determine the extent of pollution in a hundred-mile stretch
of the river. EPA in conjunction with the Metropolitan St.
Louis Sewer District are conducting the survey (Honeywell
Instrumentation News, Oct., 1969).
154
-------
The Bureau of Reclamation is mainly concerned with water
quantity projects. However, in the California Central
Valley Project, water quantity for irrigation and other
uses is dependent upon water quality. Thus, beginning in
1965 a series of impressive studies using a Honeywell mobile
monitor were performed. Since irrigation water must meet
salinity requirements, most of these studies were aimed at
gathering knowledge on the effect of tides, freshwater in-
flow, and weather on the salinity of the Sacramento-San
Joaquin River delta water. Two large cities are affected
by the water quality of this area: San Francisco and
Sacramento. Mobility and operation in remote areas for
extended periods (up to 30 days) were two important features
of the Honeywell unit. The monitor capable of measuring
six parameters—conductivity, turbidity, DO, pH, temperature,
and sunlight intensity—was transported in a trailer equipped
with a small lab (Anon., Public Works, 1967). The trailer
contained its own generator and 500-gallon propane storage
tank but also was provided with connections for tapping on
to a power line (Marks, 1966). Data recording was accomp-
lished via a strip chart recorder. Extracted data is then
computer processed to give maximum, minimum, and mean values
for each parameter (Anon., Public Works, 1967).
From 1965 to 1967, monitored data aided the Bureau in con-
trolling stream flows to deter the encroachment of salt
water from the San Francisco Bay into the Sacramento and
San Joaquin Rivers, to flush salt marshes producing a more
favorable winter habitat for ducks and geese, and to dilute
industrial pollution during low flow periods so that salmon
would again spawn in the San Joaquin River. Other uses
included collecting data to help in the planning of new
water impoundment projects on the American River above
Sacramento and supplementing California's grab sampling pro-
gram by performing intensive surveys of potential and
existing pollution problem areas (Anon., Public Works, 1967).
Interstate Applications
The early 60's saw the beginning of the application of
automatic instrumentation to monitor water quality on a re-
gional, drainage basin basis. First, developments in this
area were undertaken by the Ohio River Valley Sanitation
Commission, ORSANCO (Anon., Chemical Engineering, 1963).
Much preliminary thought and investigation was given to an
automatic water quality data acquisition before a formal
program was initiated in 1958. The first phase of this
project was to investigate instrument availability, capa-
bility and reliability. This was followed by the selection
of instrument type and the installation of a prototype
155
-------
system. Electrochemical probe type analyzers were found
to be preferable to automatic wet-chemical units because
of the signal requirements of the telemetry component and
the remote location of the monitors making reagent replen-
ishment difficult. The probe type unit chosen was designed
by Schneider Instrument Company and was termed "Robot Moni-
tor." These analyzers were capable of measuring pH, temp-
erature, conductivity, DO, chlorides, ORP, and solar radi-
ation. The prototype unit was an integrated system con-
sisting of the three basic modules mentioned earlier.
Operational as of September, 1960, the prototype unit
employed telemetry with data being transmitted to ORSANCO
headquarters in Cincinnati, Ohio. There, a data logging
facility was provided. Successful operation of the proto-
type unit allowed for expansion of the system. The result-
ing network consisted of six telemetry units on the upper
Ohio and five on-site strip chart recording units stationed
on the lower Ohio (Cleary, 1962).
The ORSANCO system now comprises 27 automatic monitoring
stations (1.7 on the Ohio and 10 more on tributaries)
(ORSANCO Twenty-First Yearbook, 1969). In addition, a
mobile version of the robot analyzer has been used in spec-
ial studies. This unit employs on-site strip chart re-
cording (Anon., Chemical Engineering, 1963).
The in situ system is fully integrated with data telemetered
to a central receiving station where an IBM 1130, 2B compu-
ter processes it (Klein, et. al., 1968). Data is not just
compiled and stored. Statistical analyses are performed
resulting in monthly and annual reports. Reports contain
tables, graphs, charts, and qualigrams showing maximum,
minimum, and average values for each parameter in relation
to station, time of year, and water quality criteria.
Naturally, these reports are excellent for public relations
and aiding state and ORSANCO officials in evaluating the
effectiveness of their pollution control efforts (ORSANCO
Quality Monitor, Oct., 1970; ORSANCO Twenty-First Yearbook,
1969) .
Paralleling ORSANCO's developments, the Interstate Commission
for the Delaware River Basin, INCODEL, in cooperation with
the Delaware Water Pollution Commission, the USGS and the
City of Philadelphia Water Department, initiated an auto-
matic water quality monitoring system for the Delaware River
and Estuary (Anon., Water ans Wastes Engineering, 1961).
The USGS role in this overall network has already been
explained (see section on Federal Applications).
While the USGS was working on the estuary, INCODEL, follow-
ing a similar path to ORSANCO, contracted with the Lehigh
Water Resources Research Council to research the feasibility
156
-------
of establishing a monitoring system on the Delaware River.
Preliminary investigations involved researching instrument
availability, capability, and reliability. A prototype
system^was installed in 1959 at Riegelsville, New Jersey.
Employing on-site strip charts, temperature, pH, conductivity,
DO, and turbidity were continuously recorded. Only a few
months' operation was necessary to warrant the construction
of eleven additional stations to be constructed between
Hancock, New York, and Trenton, New Jersey. Plans were made
to have these integrated into a network employing telemetry
and a central receiving station (Anon., ASCE 28th Progress
Report, 1960; Parker, 1961).
Data from the USGS and the INCODEL systems provide necessary
data to the subsequently formed Delaware River Basin Com-
mission in which many agencies cooperate to manage the
Delaware River and Estuary. (Agencies involved are: EPA,
USGS, Delaware Water Pollution Commission, New Jersey De-
partment of Health, Pennsylvania Department of Health, and
the City of Philadelphia.) Along with grab sampling data,
continuous records from the automatic systems helped to
supply the data necessary for formulating a mathematical
model of the River and Estuary. This model has been very
useful in guiding water quality management practices on the
River and Estuary (Smith & Morris, 1969).
State Applications
Many states on their own have entered the field of automatic
water quality surveillance. Most noteworthy is New York's
Empire State system. The main purpose of this automatic
network is to "provide a rapid intelligence system" to
protect the waters and water users of the state. Emphasis
is placed on rapid retrieval and analysis of data to yield
instantaneous reports on the prevailing water quality con-
ditions (Maylath, November 1970).
Like the interstate systems mentioned previously, much pre-
liminary research was necessary before any installations
could be made (research began in 1960) . A prototype network
was installed in 1966 with the purchase of two electrochemical
probe type monitors (Maylath, 1970). An exhaustive study
of wet-chemical monitors revealed that they were really not
satisfactory for unattended field use (O'Brien & Olsen, 1970).
Field experience using the prototype probe system yielded
the following conclusions:
1) Automatic monitors, if applied in a true
real-time computer approach, are necessary
to supplement a grab sampling program.
157
-------
2) A well-trained team of engineers and techni-
cians is necessary to maintain the system.
3) The monitoring system should be expanded
(Maylath, 1970) .
Thus, steps were taken in 1969 to produce a small-scale
monitoring network consisting of 12 monitors, with telemetry
and central computer processing of data. A Burroughs B3500
computer was the heart of the system (Burroughs Corp, News
Feature). Every attempt was made to incorporate the latest
design features into the monitoring system. Monitors were
equipped with sensors measuring pH, DO, conductivity, water
temperature, turbidity, stage, dissolved chlorides, dissolved
fluorides, solar radiation and air temperature. In addition,
environmental parameter alarm sensors, automatic samplers,
equipment status sensors, and functional command equipment
were included for better control of analyzer operation.
Monitors were housed in trailers equipped with a lab plus
air conditioning and heating. Leased telephone lines were
employed to transmit data to the central computer station
(Maylath, 1970) .
Monitoring stations were polled every hour with data travel-
ing to the computer and remote terminals. Remote terminals
are an integral part of the overall system since they are
installed at water and waste treatment plants. Data received
here is instantaneously available for altering plant pro-
cesses in relation to prevailing water quality and quantity.
Data received at the computer center is also displayed for
rapid control of the water resource. However, statistical
analyses are also performed which are the basis for daily,
monthly, and annual reports (Burroughs News Feature, 1970;
Maylath, November, 1970).
Future plans are to enlarge the small scale system into a
basin by basin network all telemetering data to the central
computer processing station. Mathematical models of each
basin are being prepared. Data from the basin networks will
then be inputed into these models to ascertain:
1) the water quality at all points in a stream
(not just at the monitoring station)
2) prediction times for pollutant spills to
travel downstream
3) the source of a pollutant
4) the best locationr design, and operation
of water and waste treatment plants
(Maylath, November, 1970).
158
-------
A less sophisticated monitoring system was initiated in
New Jersey in 1968. Of the 10 monitors purchased, four
were two parameter resistance type analyzers, while the
rest were capable of measuring five water quality charac-
teristics. Five parameter units monitored conductivity,
water temperature, pH, turbidity, and DO. They were housed
in_specially built structures for environmental protection.
Utility lines provided power for the system. Data retrieval
was via punched paper tape. Tapes were picked up once a
week, edited, and processed by computer. Processed data
consisting of maximums, minimums and means was used to
establish trends in water quality of the streams monitored.
Trends were displayed graphically.
The two parameter systems were much simpler. Conductivity
and temperature were measured with data being registered
on a Fisher Porter automatic digital recorder. Units did
not require specially built houses but were installed in
location. One advantage of the two parameter system was
its ease of installation and relatively maintenance-free
operation. The unit also requires no outside power source,
as it is battery powered (Anderson, et al, 1970).
Automatic monitors were needed in Pennsylvania on a special
survey of acid mine drainage. ProTech Company supplied
the automatic system in 1965. The system consisted of two
three-parameter units capable of measuring pH, conductivity,
and temperature. Units were operated on the Susquehana River
at two sites, United Gas Industries and Merc Chemical Company.
Data logging was accomplished in two ways via printer
(Presin Company) and punched paper tape (Fisher & Porter).
As designed, the system did not meet Federal Water Pollution
Control Administration (FWPCA) requirements (see Mentink,
Specifications for an Integrated Water Quality Data Acqui-
sition System, 1968, eighth edition). Thus, many operational
problems were encountered which eventually led to the return
of equipment to ProTech. Nearly all of the poor performance
was attributed to lack of design continuity and poor design
features of the analyzer which did not provide for adequate
sensor cleaning, minimal settling in the flow chamber,
modularity, chassis mounting, cabling, panel meter display,
internal electronic calibration check or color coded wiring.
Attempts at improving the system still resulted in unacceptable
operation. Thus, the Pennsylvania Department of Health
asked for assistance from the FWPCA. As a result, the
FWPCA performed a complete investigation of the design and
operation of the ProTech system and also researched alternate
159
-------
methods of obtaining the desired data. It was concluded
that nothing could be done to salvage the ProTech system
and that it should be phased out in favor of one of the
alternate methods examined. The alternate methods exam-
ined were other automatic data acquisition systems meeting
FWPCA specifications and various manners of grab sampling.
Information from this research was supplied to the Pennsyl-
vania Department of Health which would have to decide on
the method to be used, based on the level of the state's
commitment to water quality surveillance (Mentink, 1970).
Mobile monitors have been selected by other states for water
quality surveillance. Wisconsin's State Division of Re-
source Development chose a mobile water quality monitor
to supplement their grab sampling program (Anon., Public
Works, 1968). The Division (now the Department of Natural
Resources) felt that such a unit would be very useful in
special surveys and in determining the diurnal variations
at monthly grab sampling stations (Schraufnagel, Personal
Communication, 1971) . Such information would eventually
help to formulate the desired water quality standards,
locate sources of pollution, and find the cause of algae
blooms in Wisconsin lakes (Anon., Public Works, 1968).
The Honeywell mobile unit is housed in a 14-foot trailer
and measures water temperature, DO, pH, ORP, conductivity,
chlorides and solar radiation. The trailer contains lab-
oratory space, a sink, and cabinets so that on-location
chemical analyses of grab samples can also be performed.
The trailer is environmentally controlled, being supplied
with air conditioning and heating. Power is provided by
a propane-fueled 5 kw generator. A 500-pound propane
storage tank is attached to the trailer (Anon., Public
Works, 1968).
During the few years it has been in operation, the mobile
unit has supplied reliable data at eight-minute intervals
sampling frequency. All data is recorded on an electronile
15 potentiometer (strip chart recorder). Operational
experience has shown that cold weather use presents a prob-
lem due to freezing of intake and outlet pipes. Also, the
chloride probe has not performed satisfactorily.
Eleven new in situ monitors have now, been purchased from
Automated Enviornmental Systems, Inc. (AES). They will be
located on two of the most troublesome rivers. The AES
monitors will measure DO, temperature, pH, and turbidity,
and will telemeter data via telephone lines to a central
computer processing center. Here, data will be displayed
by teletype and functional commands will be delivered from
the computer to the monitors (Schraufnagel, Personal
Communication, 1971).
160
-------
In Washington, the Department of Ecology in 1969-70 employed
an AES mobile monitor to collect historical data on Wash-
ington waters. The AES unit employed is housed in a trailer
equipped with two propane storage tanks and a generator.
Thus, the unit may be operated in remote areas away from
power sources. Parameter^ measured are: DO, temperature,
conductivity, turbidity, pH, and chlorides. A regular
half hp submersible pump supplies water to the flow chamber
of the monitor.
In 1970, summer operation for 55 days yielded a 33-day
usable record. Some trouble with sensor operation was
noticed with only pH, temperature, and DO sensors working
consistently. However, most of the downtime (responsible
for loss of usable record) was due to support equipment
failure. The submersible pump was primarily at fault, but
the generator was also a source of problems. It was deter-
mined that increased efficiency could be obtained by
providing a separate trailer for carrying the propane tanks
and generator. Also, daily servicing was found necessary
to insure proper operation of the unit (Palko, Personal
Communication, 1971) .
A relatively unique approach to mobile water quality moni-
toring was tried by the Texas Water Pollution Control Board.
Because of .pressure from a sportsman's organization, a
comprehensive water quality study of Galveston Bay was
initiated in 1963. Due to the large area involved, it was
determined that a mobile, automatic monitoring device should
be employed. The first unit, called the Stargazer, was
an inboard 38- by 11.5-foot boat equipped with a 250 hp
motor. An on-board laboratory was provided containing an
improvised automatic monitoring system with DO, conductivity,
pH, and temperature sensors and separate recorders. Power
for the system was provided by a 1.2 kilowatt gasoline
generator. A one-fourth hp bilge pump supplied_water to
the sampling line which was manifolded at the site of_the
sensors. Later, the system was transferred to the "Little
Dipper," a 150 hp inboard-outboard boat.
During a 34-month period, 17,000 field analyses were per-
formed using the above setup. One hundred and eighty
sampling stations were established requiring 418 miles of
boat travel. Readings were taken at the sampling stations
during the regular run and special surveys for 26-hour
periods were accomplished in 1964 at points other than the
regular sampling stations. Data collected will aid the
Texas Water Pollution Control Board in its effort to develop
a long-range plan for controlling water pollution in the
Galveston Bay area (Davis, 1966).
161
-------
Municipal Applications
While federal, interstate, and state applications are aimed
at obtaining data on the quality of fresh and saline waters,
municipal applications are primarily involved with acquiring
data on the influent and effluent character of water entering
water and wastewater treatment plants. Regardless of this
difference in objectives, the same automatic monitoring instru-
mentation is employed.
Urban monitoring systems have been employed to better manage
water and waste treatment plants and to evaluate and improve
treatment plant processes. Salvatorelli (1968) gives a good
rundown of the application of automatic monitors to process
control in a waste water treatment plant. Sensors typically
employed are DO, pH, ORP, T, and Cl. Dissolved oxygen and
ORP are used to monitor the oxidation process occurring in
the aeration tank. Digestor operation is monitored by pH
and temperature sensors. The chlorine sensor is utilized to
measure the chlorine concentration in the contact chamber
and to check residual chlorine levels. Salvatorelli states
that such systems are a must for optimal treatment plant op-
eration. He maintains that instrument costs only run 3.5 to
4.5 percent of the total treatment plant construction costs
and that after proper personnel training for monitor mainten-
ance yields a completely dependable and reliable monitoring
system.
Application to an urban water supply system is well exempli-
field by the integrated network established in Chicago. The
Greater Chicago Metropolitan Sanitation District was forced
into installing an automatic monitoring system for four rea-
sons. One, population was increasing rapidly with a present
waste load from 5-1/2 million and an industrial equivalent to
2-1/2 million more. Two, a large collection system was in-
volved (860 square mile area). Three, enactment of state
water quality standards made standards compliance necessary.
Arid four, two U.S. Supreme Court rulings had limited the
amount of water Illinois could divert from Lake Michigan
and further stipulated that no additional requests for di-
version water would be accepted until all possible methods
of efficiently using existing water had been exhausted.
Installed in July 1968, the system consisted of an integrated
network of eleven monitors relays data to a central computing
facility and directly to water users at the three main treat-.
ment plants. Continuous data is received at the treatment
plants and is recorded on strip charts. Hourly values are
transmitted to the central computing facility which is staffed
24 hours a day. Here, data is available for instantaneous
162
-------
display by teletype printed on log sheets and also is re-
corded via punched paper tape for statistical review.
The_system was designed by Bristol Company, with instrumen-
tation supplied by Fairchild Space and Defense System.
Monitors are housed in either existing buildings or specially
built structures. A composite sampler is provided so that
laboratory analyses can be performed for water quality char-
acteristics not capable of automatic measurement. Parameters
measured automatically are DO, water temperature, conductivity,
pH, ORP, chlorides, turbidity, and solar radiation. Sub-
mersible pumps supply water to the monitors.
Because of the highly turbid and polluted waters sampled, a
well disciplined maintenance program was particularly im-
portant for this system. Servicing was on either a weekly
or biweekly basis, depending on the station. Calibration
was accomplished at intervals of two to four weeks, depending
on the sensor and station. Automatic water jet cleaning
devices were also intalled at some locations. Jets would
operate hourly just before parameter measurement. A main-
tenance crew of five people was needed to perform monitor
servicing. Two of these were electrical engineers, and the
other three were instrument mechanics.
Information gathered will aid the Greater Chicago Metropolitan
Sanitation District in determining the optimal distribution
of water from diversions to maintain minimum water quality
standards in receiving waterways (Lanyon & Kurland, 1971).
The benefits gained by continuously monitoring urban water
collection systems is explained by Sahre (1970). Reference
is made to the Detroit system which was designed to monitor
both qualitative and quantitative characteristics of treat-
ment plant receiving waters. He emphasizes the role auto-
matic monitoring plays in detecting improper plant operation
and thus in preventing unnecessary spills. He also points
out that the information gained will be valuable for improving
the design and operation of sewer and drainage systems.
Some quasi-automatic systems employing Technicon Auto Ana-
lyzers have been developed to survey and improve treatment
plant processes. These systems are composed of two parts:
automatic samplers followed by automated wet-chemical ana-
lyses with an Auto Analyzer. The Minneapolis-Saint Paul
study is an example of one such system. Here, the Technicon
system was used to measure MBAS, COD, chlorides, iron,
ammonia nitrogen, phosphates, and urea. Information obtained
was valuable for evaluating this combined sewer system, yet
only data from sewer level and rain guage monitors were
163
-------
treated on a real-time computer processing level employ-
ing telemetry (James J. Anderson, et al, 1967).
A similar investigation involving only one treatment plant
(Hyperion in Los Angeles Count, California) is described
by Garber, ejt al_ (1965) . Here, Technicon Auto Analyzers
were used to evalute the aeration efficiency and the quality
of secondary effluent.
Applications in University Research
In a few instances, automatic monitors have served as a tool
aiding the university researcher in a solution to a particu-
lar problem. Weiss and Oglesby (1963) utilized automatic
monitors to help them determine whether vertical oscillations
in the metalimnion of an impoundment reservoirs permit mat-
erial transfer into either the hyper or hypolimnion. The
system described consists of a two-parameter unit containing
several temperature probes and a galvanic cell oxygen anal-
yzer. Temperature and DO profile data were collected and
telemetered to the researcher's lab where recording was
accomplished by strip charts and punched paper tape. The
researchers believe that information gleaned from surveys
such as theirs will be valuable for selecting water of op-
timum quality from storage reservoirs.
Ingols (1967) used a Honeywell W10 monitor to study the
effect of storms on water quality. Parameters recorded were
DO, water temperature, conductivity, turbidity, pH, stage,
air temperature, and solar radiation. A continuous record
was provided using a strip chart. The W10 monitor was
housed in a mobile van supplied with a gasoline generator.
Stations were selected near utility poles, however, since
the generator required maintenance up to three times per day.
Data collected were analyzed for comparisons between water
quality and flow.
Applications in Canada
The Ontario Water Resources Commission has taken a unique
approach to automatic monitoring. In 1969, the Commission
initiated a monitoring program using two submersible water
quality meters. These immersion monitors were provided by,
Plessey Canada, Ltd. The units were completely self-con-
tained. They are battery powered with sensors, analyzers
and output component packaged together in a cylindrical
submersible housing. Parameters measured were time, depth,
temperature, turbidity, conductivity, pH, and DO. A current
meter was operated in conjunction with the immersion assembly.
164
-------
Water_quality data were correlated with current velocities
and directions (Palmer & Izatt, 1970). Data recording was
accomplished by magnetic tape cassette at fixed intervals
from one to six hours (Palmer, Personal Communication, 1971).
The objective of the monitoring program is to provide water
quality data for math models which attempt to predict the
dispersion characteristics for nearshore areas of the Great
Lakes. Since the water quality of these areas is known to
be highly complicated, variable continuous data records are
needed. Markov chain and time series analysis are used in
analyzing the data (Palmer & Izatt, 1970).
The 1969 survey was performed on Lake Erie at nearshore areas
around Nanticoke. Results indicate that conductivity is
directly related to water movements in these areas (conduc-
tivity can be treated as a passive contaminant). Dissolved
oxygen and pH exhibit diurnal relationships and seem uncor-
related with water currents (Palmer & Izatt, 1970).
Math model of the dispersion characteristics will be useful
for determining:
1) the best location for water intakes and
waste outfalls from the dilution point
of view;
2) the acceptable discharge concentrations and
flows on the basis of not exceeding desir-
able values at locations in the priximity of
the discharge point on a probability basis
(Palmer, 1970) .
The Ability of Automatic Monitoring
to Satisfy the Data Needs of
State Water Pollution Control Agencies
The previous discussion has shown how automatic monitors
are applied in water quality surveillance programs. However,
two very basic and important questions remain unaswered:
1) why is automatic water quality data collected, and 2)
what is the data used for once it has been collected. Answer-
Tng~these~questions will help to define the ability of auto-
matic monitors to satisfy the data needs of state water
pollution control agencies.
The foundation for understanding the need for and the utility
of automatic monitoring data is based on_two facts: 1) grab
sampling has many shortcomings when applied to situations
165
-------
where water quality fluctuates rapidly, and 2) water quality
management is becoming more complex, requiring real-time data.
To begin, Klein, et. al_ (1968) states that there are three
deficiencies in grab sampling programs:
1) cost
2) limitations on the frequency with which
analyses can be made
3) time lag between collection of samples and
receipt of analytical results for stream
evaluation purposes
Cost and limitations on frequency must be considered together.
Frequency limitations are mainly related to manpower limita-
tions. As applied to grab sampling, increased sampling rate
means increased field and laboratory personnel to collect and
analyze samples. A concurrent increase in operational ex-
penses of course follows (Anon., Electronics, 1970 - quoting
William T. Sayers).
In comparison, the major advantage of automatic monitoring
is the increased capability for high frequency sampling
(Ballinger, 1971). Nevertheless, automatic instrumentation
does not eliminate field and laboratory personnel because of
the necessity of periodic maintenance on automatic equipment
(Sayers, 1971). Indeed, personnel requirements are such that
higher salaries are required to obtain people with the right
qualifications (Sayers, 1971).
While automatic monitoring has some advantages with respect
to manpower limitations, the wealth of data produced increases
operational costs due to the necessity of expensive data
handling procedures (Sayers, 1971). Effective data interpre-
tation and analysis requires digital computers (Ballinger,
1971). ORSANCO's robot monitor system collects so much data
that it would take 150 man-years to process all the data
collected during only one year (Klein, et al, 1968).
Taking the above factors into consideration, as well as
installation and maintenance equipment cost associated with
automatic monitoring, automatic instrumentation is favored
on a least cost basis when a sampling frequency greater than
daily is desired (Ballinger, 1971).
However, the trade-off between automatic and manual sampling
involves other considerations also. According to Klein, et
al^ (1968), the third deficiency of grab sampling was the
time lag between collection and receipt of analytical results.
This topic will also be discussed later in relation to the
need for real time data for water quality management. For
166
-------
the purposes here, it will be sufficient to note that grab
sampling programs entail time consuming collection analysis
procedures (Cleary, 1962). Grab sampling programs applied
to situations involving rapid fluctuations in water quality
may be very inefficient and expensive due to the large pro-
portion of time spent in collection of samples and travel
between stations and to the laboratory (Mentink, 1970) .
Another problem in grab sampling associated with time con-
suming collection and transportation is the retrieval of a
representative sample. "In general, the sooner the samples
are analyzed after the collection, the more reliable the
data" (Ball, 1970). This is true because after sample col-
lection, changes in composition may occur rapidly (Glenn,
1970). Parameters such as temperature, dissolved gases,
and pH are subject to rapid alteration upon collection and
"must be measured in the field" (Ball, 1970). Other param-
eters such as BOD, NO?, and PO^ can change significantly
with time and precautions (e.g., refrigeration) should be
taken to assure minimum sample degradation between field
collection and laboratory analysis (Ball, 1970).
Another important consideration in comparing manual and
automatic monitoring is the relative compatibility of a
particular water to either method of evaluation. To explain,
some waters exhibit rapid fluctuations in water qualities
while others show relatively slow changes (Thomann, 1970).
"In any sampling, the frequency of sampling is directly
related to the rate of change of a given paramter " (Ballin-
ger, 1971). Thus, some water are naturally more conducive
to automatic monitoring than others (Thomann, 1970; Ballinger,
1971; Ballinger, 1968). The proper sampling frequency for
any water can be determined by spectral analyses. Ironically,
this requires a "relatively large amount of data" (Ballinger,
1971) .
The above discussion has described some of the deficiencies
of grab sampling. Most of these limitations are related to
monitoring programs requiring detection of rapid fluctuations
in water quality. As pointed out, herein lies the major
value of automatic monitoring, the capability for high fre-
quency sampling. The question now is: Why is it necessary
to detect rapid fluctuations in water quality? Smoot(1970)
supplies a concise answer to this question:
Today many complex water-quality problems
associated with protecting and improving
our environment frequently require immedi-
ate evaluation and prompt action.
This applies to the field of water supply and pollution con-
trol where "more and more, real time data networks are being
167
-------
recognized as essential in providing the current information
necessary for good water-quality management" (Smoot, 1970) .
For some time, water resource managers have desired to keep
continuous tabs on upstream conditions so that good water
quality could be maintained for downstream users (Jones, 1961).
The need for continuous tabs on water quality is a direct
result of the increased quantity and variety of waste matter
entering our nation's waters (Jones & Joyce, 1961; Cross,
1968). True, some waters do not fit this description and
are still blessed with good quality due to sparse population
and large stream flows in relation to pollution loads. In
these areas, only modest sampling programs may be sufficient
(McDermott, Ballinger and Sayers, 1968). However, Elving
(1967) states that any flowing water constitutes a constantly
changing and dynamic system which must be monitored continu-
ously. Moreover, in some areas of dense population, complex
waste sources, intensive water use, and rapidly fluctuating
water quantity, day-to-day water quality management requiring
continuous automatic monitoring is needed (McDermott, Ballinger
and Sayers, 1968) .
Ability to Satisfy Planning Data Needs
Long-range planning connotes the use of math models. Data
needed for math models "must be perfect continuous records
with no sags or bad readings" (Palmer, 1970). This basic
requirement is satisfied by automatic monitoring which
supplies large quantities of continuous inexpensive data
(Palmer, 1970). Ballinger (1971) agrees in that automatic
monitors provide large amounts of data in short time inter-
vals which are valuable as input into predictive river basin
models. Also, automatic data is very suitable to certain
statistical analyses often used in water quality models:
time series analysis and Markov chain (Palmer, 1970; McCartney
& Beamer, 1962) .
River basin models based on automatic data have much planning
utility. They are useful as an aid in locating municipal
and industrial intakes and outfalls (Palmer, 1970; Glenn,
1970) . In New York, the Empire system provides pertinent
data to "improve the precision of math models predictions"
(Maylath, 1970) . These predictions are the basis for formu-
lating design and operational criteria guiding the construc-
tion and operation of water and waste treatment plants. New
York also plans using monitor data in the preparation of
stream models useful for water quality forecasting (Maylath,
1970). Using deterministic and probabilistic BOD-DO models
based on continuous DO measurements and BOD waste load
determinations, downstream DO predictions can be made knowing
upstream conditions and events (Boes, 1970). Automatic data
168
-------
w m°dels can also Yield information
Si !*5 quality conditions at non-sampled sites and at
any time of day (Glenn, 1970; Maylath, 1970).
Ability to Satisfy the Data Needs for Regulation
A state water pollution control agency concerned with reg-
ulating water quality is interested in detecting pollution,
determining its source, and seeing that the pollution is
abated. There is little question about the ability of auto-
matic monitors to detect pollution. One of the main assets
of automatic monitoring lies in its ability to detect ab-
normal water qualities (Klein, et al, 1968). In this res-
pect, automatic monitoring surpasses manual monitoring as
"grab sampling, even at frequent internals, may not detect
undesirable levels in time to permit effective counter
measures" (Ballinger, 1971). Automatic instrumentation,
however, has the capability of sounding an alarm and/or
collecting a sample for further analysis once an abnormal
variation in water quality has been detected (Keyser, 1965;
Parker, 1961). Also, as mentioned earlier, one of the
deficiencies of grab sampling lies in the difficulty of
maintaining a representative sample. Sample degradation
can occur due to the time lag between collection and anal-
ysis (Ball, 1970) .
The ability of monitors to detect abnormal changes in water
quality and collect a sample at these times has significance
for determining pollution sources. Analysis of the collected
sample may indicate the particular waste causing" the abnormal
water quality. Knowing the nature of the pollutant may be
valuable for locating its source (Ballinger, 1971) . A good
example is the tracing of two fishkills to a common cause
by the ORSANCO Robot Monitoring System (Klein, et al, 1968) .
The advantage of the large spacial distribution of the
ORSANCO network was brought out in tracing the source of
widespread low DO readings to natural cuases (Klein, et al,
1968). A similar situation existed on the USGS Delaware
system in April, 1963, when DO values were depressed below
3 ppm. Again, the monitoring network traced the source to
natural causes (Anon., Chemistry and Engineering News, 1963).
Electronic monitors are also very amenable for detecting
diurnal fluctuations in water quality characteristics; es-
pecially for documenting DO sags (Ballinger, 1968; Kleinert,
Personal Communication, 1971).
Making data readily available to water quality agencies is
an important part of pollution abatement. "Those who are
active in water resources management agree that any pollution
abatement program must be preceded with the acquisition of
a continuing intelligence on the quality of all our rivers
and streams (Shubrooks, 1968) . Merely collecting data is
169
-------
not enough; it must also be made readily accessible to those
concerned with controlling water quality (Shubrooks, 1968).
Once the data is made available, it must also be used. A
rapid intelligence system of automatic monitors, telemetry,
remote terminals and central data logging and analysis
facilities provides a highly effective means of water qual-
ity regulation (Maylath, 1970). As mentioned earlier, in-
stantaneous relay of water quality conditions allows for
alternation of treatment plant processes to conform to the
quality of receiving waters (Maylath, 1970; Langon & Kur-
land, 1971). Monitored data is also useful for determining
low flows and the amount of water necessary to supplement
stream flows so that standards are not violated. Using
automatic data discharges of pollutants can be traced and
their effect on the entire water system evaluated (Ballin-
ger, 1971) .
Monitored data coupled with math models can be useful for
establishing effluent concentrations for industries and
waste treatment plants. In this way, the assimilative
capacity of receiving waters is not exceeded and compliance
with water quality standards is assured (Palmer, 1970).
Data processing of telemetered data allows for rapid report
generation. Such reports on an hourly, daily, weekly,
monthly, or annual basis are extremely valuable as an aid
in regulation, since graphs, charts, and tables can be de-
vised to show the percent of time water quality parameters
are in violation of standards. In addition, computer anal-
ysis can determine the extent of daily pollution loading,
including: dissolved oxygen deficits, suspended solids
loads, salt water intrusion, acid loads from mines, caloric
loading from thermal pollution and waste treatment plant
effluents (Ballinger, 1971; Elving, 1967) .
Needed Improvements in Automatic Instrumentation
The previous section has cited the advantages of using auto-
matic instrumentation for water surveillance. Nevertheless,
even with sophisticated telemetry systems, present-day water
quality surveillance leaves much to be desired. The fact is
that most state and federal agencies are still not able to
determine whether our nation's waters are getting better or
worse (Cleary, 1970) .
Disadvantages of Present Instrumentation
Agencies are handicapped because monitoring systems still
fail to supply the right data (Cleary, 1970). Automatic
170
-------
r manY short«>»ings in this respect. In I960,
rt ?°£f?jerice On Water Quality Management and Instrumen-
tation held in Cincinnati, Ohio, it was pointed out that
data processing technology was sufficient for water quality
monitoring but that much research on sensing devices was
needed. Ten years later at a similar conference (National
Symposium on Data and Instrumentation for Water Quality
Management) held in Madison, Wisconsin, the same conclusions
were reached (Lyons, 1970). It is now generally recognized
that data recording and processing represents the strongest
link in the automatic water quality data acquisition system
because it is fully developed and relatively trouble free
(Smith, 1970) . Similar words cannot be spoken for sensing
devices. Most experts agree that in the last ten years,
little progress has been made in developing new automatic
sensors (Lyons, 1970) .
The reason for this lag in the development of new sensors
is twofold. One, as mentioned earlier, initial emphasis on
developing automatic water monitoring instrumentation was
focused on supplying monitors for industrial process control
(see section on application of automatic monitors). Two,
instruments devised for this purpose were merely adaptations
of existing laboratory instruments (Considine, 1965). Thus,
development of automatic field equipment paralleled and was
dependent on the development of laboratory and process con-
trol instrumentation. This process has resulted in the
deplorable situation where field sensors "represent param-
eters for which electrode systems are readily available—
and do not include many measurements vital to the adequate
characterization of water quality" (Ballinger, 1968). (See
Figures B-1 andB-2.)
The past ten years have not been completely fruitless. De-
velopment of specific ion electrodes and wet chemistry moni-
tors has had some impact on alleviating the above problem.
Specific ion electrodes can broaden the spectrum of sensor
detection to include some specific water quality parameters.
In general, ion selective electrodes detect ions in solution,
thereby developing an electrical potential in proportion to
the activity of the ion (Riseman, 1970); Kaminski, 1969).
The chloride ion is an example of such an electrode presently
used in automatic water quality monitoring. However, other
specific ion electrodes available for laboratory use are
"not satisfactory for monitoring purposes" (Ballinger,
Personal Communication, 1971) . This includes electrodes for
measuring important nutrient parameters (e.g., nitrates and
phosphorus). Electrodes for these parameters do not pro-
cess adequate sensitivity and fail to truly measure the de-
sired constituents (Ballinger, Personal Communication, 1971).
Another serious problem is interference from like ions.
Pretreatment of a water sample can be performed in the labora-
tory to minimize these effects, but this is difficult under
field conditions (Riseman, 1970).
171
-------
Acidity
Alkalinity
Ammonia
Arsenic
Barium
BOD
Boron
Bottom Deposits
Cadmium
CCE
Coliform
Iron
Lead
Manganese
MBAS
Nitrate
Pesticides
Phenols
Phosphates
Plankton
Radioac tivi ty
Selenium
Chloride
Chromium
Color
Copper
Cyanide
Dissolved Oxygen
Electrical Conductance
Floating Solids
Fluoride
Hardness
Hydrogen Sulfide
Settleable Solids
Silver
Sodium
Sulfate
Suspended Solids
Taste & Odor
Temperature
Total Dissolved Solids
Toxic Substances
Turbidity
Zinc
Figure B-l. Water Quality Criteria in State Standards.
(From Ballinger, 1971....Underlined criteria
represent parameters for which sensors are
already available.)
172
-------
Ranges of
Concentration
Desired
rag/1
Organic nitrogen
Ammonia nitrogen
Nitrate nitrogen
Nitrite nitrogen
Inorganic phosphorus
Organic phosphorus
COD
MBAS*
Acidity or alkalinity
Hardness
Sulfate
Phenols
Calcium
Cyanide
Manganese
Zinc
Sodium
Potassium
Copper
L
0-1
0-1
0-1
0-0.1
0-2
0-2
0-50
0-1
0-250
0-250
0-100
0-0.5
0-100
0-0.1
0-0.5
0-2
0-100
0-10
0-0.5
M H
0-10
0-10
0-10
0-2
0-20
0-20
0-500
1-10
0-1000
0-1000
0-1000
0-5 0-50
0-1000
0-1.0 0-10
0-5
0-10
0-500 0-5000
0-100 0-1000
0-5.0
Precision
Desirable
mg/1
L
0.01
0.01
0.01
0.01
0.01
0.01
1
0.01
5
5
2
0.01
2
0.005
0.01
0.01
2
0.5
0.01
M H
0.5
0.5
0.5
0.1
0.5
0.5
10
0.1
50
50
20
0.1
20
0.05 0.5
0.1
0.5
10 100
5 50
0.1
*Methylene blue active substances
Figure B-2. Parameters for which Sensors are not Available
but are Needed. (From Green, 1966)
173
-------
Wet chemistry monitors can also broaden the scope of sensor
detection. However, severe handicaps must be overcome be-
fore this system is suitable for field application (Ballin-
ger, Personal Communication, 1971). Some of these disadvan-
tages are interference from color and turbidity, biological
fouling of sampling lines, failure of solenoid valves, re-
agent supply and disposal, buildup of color complexes on
the measurement cell, and slow response of the system—
typical time is two hours from sampling to recording (O'Brien
& Olson, 1970; Feltz & Smoot, 1969; Ballinger, 1968). Sample
filtration has generally been employed to minimize some of
these problems. However, this "renders the sample not entirely
representative of the source" (O'Brien & Olson, 1970).
Besides not supplying entirely the right data, automatic mon-
itors are still plagued with maintenance problems. Charac-
teristically "automatic monitoring equipment is delicate and
unstable" (Blakey, 1970). This is evident from operational
experiences in which typical monitoring systems require weekly
maintenance. Sensors require cleaning and calibration due to
biological and sediment fouling causing electronic drift
(Blakey, 1970). Self-cleaning with automatic wipers, water
jets or ultrasonics may eliminate this problem (Ficken, 1970) .
Pump failure remains the number one maintenance problem. Semi-
positive displacement submersible pumps are advantageous
since they do not aerate or deaerate the water sample. How-
ever, they are disadvantageous because of frequent breakdown
due to electric motors burning up and metal fatigue producing
pump strator breakage (Maylath, 1970). The problem arises
from placing pumps in the harsh stream environment where they
are naturally subject to strain, abrasion, and clogging from
sediment, organic debris and biological growths on the pump
itself (Feltz & Smoot, 1969).
Another limitation of significance is aging and depreciation
of equipment. Solid state circuitry, amplification of sensor
signals, maintenance of proper velocity of sample through the
flow cells, and large electrolyte reservoirs are helping to
extend the life of and provide long-term stability for auto-
matic sensors (Bromberg & Carames, 1970; Klein, et. al, 1968).
Nevertheless, sensors and other monitor components do wear
out and require replacement (Bromberg & Carames, 1970). (See
Figure B-3.)
Rising costs of automatic monitors are another drawback. The
trend of instrument manufacturers is to produce more compli-
cated, costly systems, which is in direct opposition to the
needs of the users who desire simpler, more reliable and less
expensive instrumentation (Klein, et al, 1968) .
Other considerations and limitations include freezing of pump
and sampling lines during winter operation, salt bridging and
corrosion of mechanical parts during operation on saline
174
-------
Parameter Replacement Tolerance
Interval
Dissolved Oxygen 12 weeks +0.25 mg/1
pH 4-5 months +0.1 units
Temperature 6-12 months + 1 °F
Oxidation-reduction 6-12 months 12 mv
Potential
Figure B-3. Replacement Interval and Tolerance for Some
Automatic Sensors. (From Bromberg & Carames,
1970)
175
-------
waters, vandalism, and high initial costs of land acquisi-
tion and installation '(Maylath, 1970; Bromberg & Carames,
1970; Ballinger, 1971).
Future Needs
The above discussion has already pointed out many of the
improvements needed by explaining the disadvantages of pres-
ent systems. Some specific suggestions for improvement are:
1) Lessen frequency of cleaning by changing the
configuration of the flow chamber so that
it has the same internal shape as the water
supply lines (possibly have the sensor sur-
round the sample instead of the converse).
2) Develop a standard calibration procedure for
sensors (Maylath, 1970).
Other modifications are needed to make automatic monitors
more reliable and versatile. The trend in sensor technology
is toward developing microcircuit, digital sensors. These
sensors would eliminate analog to digital conversion before
signal transmission. This not only simplifies data trans-
mission and logging but also provides sensors of greater
reliability through the use of microcircuits (Tajima, 1969).
Monitor versatility can be increased by developing satellite
and mobile monitors for supplementing major stations and
for use in intensive surveys (Maylath, 1970; Ballinger, 1971).
Some progress has already been made in supplying monitors
suitable for both of these uses. One such system is the
immersion or submersible water quality recording monitor (May-
lath, Nov. 1970). Portability is an important feature of
this unit. The monitor is small, compact and battery operated.
In addition, advantage is gained by using submersible sensing
and recording instrumentation. The necessity of pumps, in-
take and outlet lines, and a shelter to house the monitor is
eliminated. Also, a representative sample is assured because
measurements are taken and recorded in stream. On shore data
logging and telemetry is also possible via external terminals
attached to the submerged monitor (Palmer, 1969) .
One disadvantage is the difficulty in maintenance due to
in-stream placement of the monitor. Such a location is awk-
ward for servicing since removal from the stream is required.
Under times of cold weather or high water, this can become
quite a chore. Also, in-stream placement allows for increased
possibility of damage due to floating debris, change in water
flow, biological and sediment fouling, and vandalism (Ballin-
ger, 1971; Ballinger, 1968).
176
-------
Mobile adaptations of in situ monitors can also be employed
as satellite monitors and in intensive surveys. Most in situ
systems can be modified for mobile operation without sI?nIf1T
cant change in design (Ballinger, Personal Communication,
1971; Ballinger, 1968). Mobile operation has some distinct
advantages. Most obvious is the fact that such units "provide
maximum flexibility in sampling location" (Ballinger, 1971) .
It isv this asset which makes mobile monitors valuable in
intensive surveys because of their capability to sample for
short periods at a number of critical points. Other advantages
are: 1) ease in maintenance, and 2) reduction in the initial
expenditures for land acquisition and installation (Ballinger,
. / •
One disadvantage of mobile adaptations of In situ systems
is the problem of inadequate operation of support facilities
(i.e., propane storage tank and generator) (Ballinger, 1968).
The benefits derived from mobile operation could be increased
if mobile monitoring were expanded to include buoy-type or
float systems (Anon., Electronics 1970 — quoting William T.
Sayers) . Two applications have been cited under the section
on Historical Review of the Use of Automatic Instrumentation
for Water Pollution Control (St. Louis and Texas Water Pollu-
tion Control Board projects) . In addition, research on a
floating system housed in a small boat has been conducted by
Raible and Testerman (1969) at the University of Arkansas
Graduate Institute of Technology.
The last point to be made, but an important one, is that
automatic monitoring data must be used to be effective (Bal-
linger, 1971; Cleary, 1970) . Automatic monitoring produces
a wealth of data which requires computer storage and analysis
(Klein, et al, 1968) . Moreover, report generation is ex-
tremely important and all water users should be informed of
water quality conditions on a real time, daily, weekly, monthly,
annual, and long-term basis (Maylath, 1970).
Cost Analysis t
A cost analysis of automatic water quality data acquisition
systems will vary according to the type of system (in situ,
mobile, immersion) , the recording technique (strip chart,
punched paper tape, magnetic tape) , and whether or not tel-
emtry is performed. The cost of telemetry systems varies
according to the type of transmission, communications link
and whether merely data logging or a computer facility is
employed. To obtain a feeling for the kinds of expenditures
involved, an analysis of three principal cost categories
will be examined first:
177
-------
1) purchase price of instrumentation
2) installation
3) operation and maintenance
Purchase Price of Instrumentation
The initial cost ;of developing an automatic monitoring sys-
tem is purchasing the equipment. Monitors conforming to EPA
specifications (Mentink, 1968) "cost from $6,000 to $12,000
depending upon the sensors required and the data acquisition
employed" (Ballinger, 1971). The five parameter system em-
ployed by the USGS on the Patuxent River costs approximately
$8,000 (Nauman & Cory, 1971). James J. Anderson (June, 1970),
for his case study on the design of urban water data acqui-
sition systems, figured a cost of $8,000 for a six-parameter
"Robot Monitor." Mentink (1970) estimates a cost of $7,000
for a four-parameter monitor.
The cost of the monitor usually includes the price of a
strip chart recorder, which runs approximately $2,200. If
another recording technique is desired, a slightly higher
expenditure is necessary. For eight channel punched paper '
tape, a cost of $6,000 can be expected (Mentink, 1970). Al-
though less expensive in itself ($4,000), sixteen channel
tape does not record data in computer compatible format.
If data processing is desired, an additional $7,500 for a
translator is required, along with $1,600 for a digitals
clock plus analog to digital converter. Also, a spare read-
ing head for the translator is advisable from a maintenance
standpoint. The price of this part is $2,000 (Mentink, 1964;
Mentink, 1970). Recording on magnetic tape is also possible.
Incremental magnetic tapes run around $5,000 (Mentink, 1970).
In addition to the price of the monitor and recording, the
expense of accessories and support equipment must be con-
sidered. The sum cost of submersible pumps, intake and out-
let lines, sample takers and sensor cleaning devices can be
significant. Sensor cleaning by high velocity water jet
costs $970, while ultrasonic cleaning runs $2,200 (Mentink,
1970).
If telemetry is "not desired, above expenditures constitute
the initial capital investment on equipment. With telemetry
a higher initial capitalization is necessary. Initial
telemetry expenses can be divided into three areas: 1) trans-
mission; 2) interface; and 3) data logging and processing.
Telemetry may be by either analog or digital transmission.
For analog transmission, there are two modes: 1) pulse dur-
ation, and 2) millivolt to frequency (Smoot, 1970) . Both
modes require a transmitter at the monitor. For either mode,
analog transmitters cost $2,250 (Mentink, 1964). However,
a lower figure of $600/location is given by James J. Anderson
(June, 1970) .
178
-------
Digital transmission requires higher cost transmitter pack-
ages. Mentink (1964) gives a figure of $4,500. AgainT
Anderson (June, 1970) gives a lower cost of $4,000.
Telemetry also entails expense at the receiving end of the
communications system. This phase of the system is referred
to as the interface since it receives the transmitted signal
from the monitoring stations and converts them into a form
suitable for display, recording, data logging, and/or com-
puter processing (Anderson, June 1970).
If pulse duration analog transmission is employed, a cost of
$11,500 is possible at the interface phase (receiver and
analog to digital converter plus packaging plus engineering).
For millivolt to frequency analog transmission, the expense
is reduced to $7,500 (receiver + analog to digital converter)
(Mentink, 1964). Anderson (June, 1970) figures a cost of
$10,000 for this phase, using analog transmission.
Since digital transmission does not require analog to digital
conversion at the interface, the cost here is nominal (Anderson,
June, 1970) .
The interface is located at the central receiving station
which can be either a data logging unit or a computer process-
ing and control facility. Data logging units may display
and record data on typewritten log sheets and record data
for future analysis on punched paper tape or magnetic tape.
(Smoot, 1970) . If analog transmission is employed, the cost
of the data logging unit is $20,000. With digital transmis-
sion, data logging units cost $22,000 (Mentink, 1964).
Computer processing facilities, are employed when more than
data logging is desired. Anderson (June, 1970) gives a
rundown on the cost of a computer facility with data process-
ing, monitor control, and math modeling capabilities. The
The cost of a minicomputer with 4K core memory is given as
$15,000. Additional core is available at a cost of $6,000
per 4K words. The disk file costs $40,000 and is necessary
if foreground/background programming is desired. Other
peripheral equipment needed includes:
1) paper tape punch-reader, $6,500
2) magnetic tape, $28,000 first unit
3) magnetic tape, $18,000 added units
4) line printer, $30,000
5) cathode ray tube display, $4,000 each
6) logging typers, $5,200
In addition, special software systems are important and cost
10 percent of hardware costs, plus $100/remote point.. Appli-
cations software cost is $5,000/program.
179
-------
Installation
After purchasing monitoring equipment, the next expense
is installation. The literature with respect to this topic
is very sparse. The only figures available appear in a
recent article by Ballinger (1971). He indicates that an
expenditure of from $1,000 to $3,000 in construction costs
is necessary for installation.
Operation and Maintenance
Automatic monitors are intended for unattended operation
in the harsh stream environment. Yet, the present state-of-
the-art indicates that they are unstable and dilicate in-
struments subject to degradation due to biological fouling,
etc., and depreciation because of aging (see discussion on
Needed Improvements in Automatic Instrumentation). Given
these limitations, it is no wonder that a large portion of
the expense of running an automatic monitoring system is
due to maintenance expenditure.
Considering only maintenance; on the monitors, a figure of
$1,250/station/year was determined for the ORSANCO system
of fourteen monitors. This is a total of $17,500 for all
the stations combined. When maintenance costs at the cen-
tral receiving station are added, the figure is only in-
creased to $18,600. Thus, only a small portion of the
total maintenance cost was attributed to the central re-
ceiving station. A breakdown of the maintenance costs on
the monitors shows approximately 35 percent for travel,
30 percent labor, 20 percent servicing, and 15 percent
parts and supplies (Donnelly, Personal Communication, 1971).
Maintenance costs are based on a two-week service schedule
(Klein, et al^, 1968).
It is interesting to note that the maintenance cost per
station per year ($1,250) has not changed despite infla-
tion and holds for ORSANCO's present system of 27 stations.
An analysis of costs for a single monitor system on the
Patuxent River shows an annual maintenance expense of $2,700.
This figure is comparable to ORSANCO's results, since ser-
vicing on the Patuxent River monitor was on a weekly basis.
The $2,700 figure includes servicing, recalibration, trans-
portation, and salaries (Nauman & Cory, 1970). Wages paid
ranged from $7.50 to $8.00 per hour (Nauman, Personal Com-
munication, 1971) .
Besides the maintenance expenses, operational costs include
expenditures for telemetry and data processing. Operational
telemetry costs are mainly attributable to the communications
link. Charges vary according to the type of leased line
employed. For telegraph grade (schedule 1001)' lines a charge
of $ .75 to $1.00/month/mile can be expected (Shubrooks, 1968)
180
-------
ORSANCO s system utilizers telegraph grade lines and has
experienced a yearly cost of $11,300 (611 miles at $1.00/
month/mile plus a monthly terminal service charge per lo-
cation) (Klein, et al_, 1968) .
EPA experience with the New York Harbor monitoring network
showed a cost of $l,980/year for leased telegraph lines
for a five-station network and $3,170 for an eight station
system (Bromberg & Carames, 1970).
For voice grade lines, the charge is $3.00 to $4.00/month/
mile (Shubrooks, 1968). A comparable figure of $3 for the
first one-fourth mile plus $1 for each additional one-fourth
mile is given by Anderson (James J., June, 1970). In addi-
tion, he indicates that an installation charge of $10 per
terminal is necessary.
If tone multiplexing is desired with voice grade lines, an
additional charge can be expected. For the central receiv-
ing station, a charge of $10,000 per eight remote sites per
party line is made. For remote sites, a $1,500 charge per
site per eight points is assessed (Anderson, James J., June,
1970) .
When data processing equipment is leased, it should be con-
sidered as an operational expense. ORSANCO leases its data
processing equipment at a charge of $23,600 per year. The
computer facility contains "an IBM 1130, 2B computing sys-
tem, with 8K and disk pack, 1134 paper tape reader and 1055
punch, 1442 card-read punch unit, 1132 medium speed printer,
and 029 interpreting key punch (Klein, et al, 1968).
Other operational expenses include electric service and
salaries. Electric service for the New York Harbor system
of five stations was $2,400. Charges figured for an eight-
station network were $3,840/year. Salaries for the five-
station system were $25,250/year, very comparable to an
eight-station network—$25,280/year (Bromberg & Carames,
1970) .
An appreciation of the constituent operation and maintenance
expenses mentioned above is important. However, a discussion
of the total operation and maintenance expenditures is prob-
ably more valuable. For the New York Harbor monitoring
system of five stations (6 to 8 parameters per station),
the total operation and maintenance cost was $31,684 per year
(includes salaries, servicing, replacement equipment, leased
lines, electric service, and transportation). This is
$6,340/station/year. Figures for an eight-station system
were $35,189/year, or $4,400/station/year (Bromberg & Car-
ames, 1970).
181
-------
Anderson (James J., 1970) estimates a total operation and
maintenance for an eight-station system (7 parameters per
station) at $33,000/year, or $4,125 per station/year.
A total annual operation and maintenance cost for the some-
what larger ORSANCO system (14 monitors—7 parameters per
station) is lower, at $53,500 or $3,821/station. This fig-
ure includes servicing, telegraph line charges, and leased
data processing equipment (Klein, et al, 1968).
Life Expectancy of Instrumentation
Another important consideration which has significant im-
pact on the cost of an automatic monitoring system is the
depreciation of equipment. Due to aging (increase in main-
tenance necessary) and obsolescence, replacement of equip-
ment may be necessary every seven years (Klein, et al, 1968)
Total Annual Cost
Given the seven-year life expectancy, operation and main-
tenance costs, initial capital investment for equipment,
installation, and amortization, the total annual cost of an
automatic monitoring system can be determined. For the
ORSANCO system (14 monitors), a figure of $74,000 or $5,300/
station was computed (Klein, et al, 1968) .
Anderson (James J., 1970) estimates a cost of $60,000 or
$7,500/station for an eight station telemetry network over
a one hundred mile stretch of river. Included in the fig-
ure is an amortization rate of 10 percent for a ten-year
period.
Without telemetry the cost is only slightly lower for a one
station system. Total annual costs for the Patuxent River
system averaged $5,000 (Nauman & Cory, 1970) .
Mobile Monitors
Actually, the cost analysis for a mobile system will be
very similar to that for the in situ systems mentioned
above. The same monitors can be employed since no signi-
ficant design change is needed (Ballinger, Personal Commun-
ication, 1971) . The only real difference is the increased
cost for installation since mobile packaging is necessary.
Wisconsin's mobile unit intitally cost $20,000, including
monitor and trailer packaging. Assuming that the price of
the monitor alone was $8,000, the cost of the trailer and
installation must have been around $12,000. Total operation
182
-------
and maintenance expenses for this trailer monitor have been
$5,000_per year (includes assumed cost of $2,000/year for
depreciation of the mobile laboratory) (Kleinert, Personal
Communication, 1971).
Immersion Monitor
Since immersion monitors are a fairly recent addition to
monitoring technology, little data is available on the costs
of such units. However, Palmer (Personal Communication,
1971) has provided the following figures:
Purchase price ....
Remooring electronic check
out and calibration
Grab sample check on
calibration ....
Postmooring electronic
check out and calibration
Data processing .
Further data analysis
Support system
$10,000/unit
approx. $ 2,000
approx. $20-travel
30-labor
12-analysis
(each time)
. . $80
$65-computer time
10-labor
approx. $400-computer time
400-labor
(interpretation
and report writing
. . < $700
Palmer also commented that immersion units due to their
portability and compactness allow easy handling in the field
and electronic servicing in the laboratory. This may have
a beneficial and significant impact on the maintenance costs
associated with these monitors.
183
-------
Literature Cited
Anderson, James J., "Minicomputers and Water Quality Sur-
veillance," Presented at the National Symposium on Data and
Instrumentation for Water Quality Management, University of
Wisconsin Water Resources Center, Madison (July 1970) .
Anderson, James J., "Case Study of Design of Urban Water
Data Acquisition Systems," Presented at the Institute on
Urban Water Systems, Colorado State University, Fort Collins
(June 22, 1970).
i
Anderson, James J., Bruss, Orville E., Hamilton, Loeman A.,
and Robbins, Maurice L., "Application of the Auto Analyzer
to Combined Sanitary and Storm Sewer Problems," Presented
at the Technicon Symposium on Automation in Analytical Chem-
istry, New York, N.Y. (October 3, 1967).
Anderson, Peter W., Murphy, John J., and Faust, Samuel D.,
"Automated Stream Quality Sensing Network in New Jersey,"
Presented at the National Symposium on Data and Instrumen-
tation for Water Quality Management, University of Wisconsin
Water Resources Center, Madison (July 1970) .
Anon., "Bye-bye Bucket," Electronics, 43(Feb. 16): 48
(1970).
Anon., "Wisconsin's Mobile Water Quality Monitor," Public
Works, 99(Aug.): 99 (1968).
Anon., "A Water Quality Monitor with Mobility," Public Works,
98(Aug.): 118-120 (1967).
Anon., "Monitor System Continuously Records River Water
Quality," Chem. and Eng. News, 41(May 6): 54-56 (1963).
Anon., "Automatic Analyzers Help Combat River Pollution,"
Chemical Engineering, 70(Feb. 4): 48-50 (1963).
Anon., "Delaware River Pollution to be Monitored Electron-
ically," Waste Eng., 32(July): 337 (1961).
Anon., "Automatic System for Monitoring Water Quality," ASCE
San. Eng. Div., 86(July): 25-40.
184
-------
Ball, Melvin Douglas, "Errors to Avoid in Water Quality
Collection and Sampling," Presented at the National Sympos-
ium on Data and Instrumentation for Water Quality Manage-
ment, University of Wisconsin Water Resources Center,
Madison (July 1970).
Ballinger, Dwight G., "Decisions to be Made in the Use of
Automatic Water Quality Monitors," Presented at the Inter-
national Symposium on Identification and Measurement of
Environmental Pollutants, Ottawa, Canada (June 14-17, 1971).
Ballinger, Dwight G., "Automated Water Quality Monitoring,"
Environmental.Science and Technology, 2(Aug.): 606-610
(1968) . ~
Ballinger, Dwight G., Personal Communication, Director
Analytical Quality Control Laboratory, EPA/WQO, Cincinnati,
Ohio (1971).
Blakey, James F., "Design of an Automatic Monitoring System,"
Presented at the National Symposium on Data and Instrumenta-
tion for Water Quality Management, University of Wisconsin
Water Resources Center, Madison (July 1970)
Boes, R. J., "Application of Electronic Monitors to River
Quality Forecasting," Presented at the National Symposium on
Data and Instrumentation for Water Quality Management, Uni-
versity of Wisconsin Water Resources Center, Madison (July
1970) .
Bromberg, Albert W. , and Carames, Manuel, "Experiences with
Operating an Automatic Water Quality Monitoring System in an
Estuarine Environment," Presented at the National Symposium
on Data and Instrumentation for Water Quality Management,
University of Wisconsin Water Resources Center, Madison
(July 1970) .
Burroughs Corporation News Feature, "New York State Using
B3500 Computer to Check Air and Water Pollution," (July 1,
1970) .
Cleary, E. J., "Symposium Summation Review of Panel Sessions,"
National Symposium on Data and Instrumentation for Water
Quality Management, University of Wisconsin Water Resources
Center, Madison (July 1970).
Cleary, E. J., The ORSANCO Story. John Hopkins Press, Balti-
more, Maryland, pp 201-202 (1967).
Cleary, E. J., "Robot Monitor System for the Ohio Valley,"
Journal AWWA, 54(Nov.): 1347-1352 (1962).
185
-------
Considine, Douglas M., "What Lies Ahead in Process Control,"
Chem. Eng., 72(June): 142-165 (1965).
Cory, Robert L., and Davis, Henry F., "Automatic Data System
Aids Thermal Pollution Study of Patuxent River," Water and
Sewage Works, 112(April): 129-134 (1965).
Cross, Ernest F., "Sampling and Analysis," Chemical Engin-
eering, 75(Oct. 14): 75-77 (1968).
Davis, N. E., "Stream Monitoring in Galveston Bay," Public
Works, 97(May): 118-120 (1966).
Dewling, Richard T., "Versatility and Flexibility are Key
Elements of FWPCA's Mobile Laboratory," Water and Wastes
Engineering, 6(May): 88, 90 (1969).
Donnelly, John P., Personal Communication, Electronics Spec-
ialist, Ohio River Valley Water Sanitation Commission
(ORSANCO), Cincinnati, Ohio (1971).
Elving, Philip J., "The Need for Instrumental and Automated
Analytical Techniques," Journal WPCF, 39(Dec.): 2055-2061
(1967) .
Feltz, Herman R. , and Smoot, George* F., "An Evaluation of
Hydrologic Monitoring Equipment," Presented at the 158th
National Meeting, American Chemical Society, New York City,
September 11 (1969) .
Fichen, James H., "Sensors - State of the Art," Presented at
the National Symposium on Data and Instrumentation for Water
Quality Management, University of Wisconsin Water Resources
Center, Madison (July 1970).
Glenn, Thomas R., "Errors to Avoid in Collection and Sampling,"
Presented at the National Symposium on Data and Instrumenta-
tion for Water Quality Management, University of Wisconsin,
Water Resources Center, Madison (July 1970).
Honeywell Instrumentation News, "New Water Quality Monitoring
Systems Broadest Product Line Infield," Instrumenation Tech-
nology, 16(Oct.): 17-18, No. 41 (1969).
Ingols, Robert S., "A Mobile River Monitor," Water and Sewage
Works, 114(Nov.): 420-421 (1967).
Jones, Robert H., "Instrumentation in Water Pollution Control,"
Water and Sewage Works, 108(June): 244-45 (1961).
Jones, Robert H., and Joyce, Robert J., "Instrumentation for
Continuous Analysis," Journal AWWA, 53(June): 713-724 (1961).
186
-------
Kaminski, R. K., "The Basics of Ion-Selective Electrodes,"
Instrumentation Technology, 16 (Sept.): 41-45 (1969).
Kehoe, Thomas J., "On Line Analysis Now Imperative," Chemical
Engineering, 72(June 7): 165-170 (1965).
Keyser, A. H., "Instrumentation for Continuous Water Quality
Measurement," Chem. Eng. Prog., 60(4): 53-56 (1964).
Klein, William L., Dunsmore, David A., and Horton, Robert K.,
"An Integrated Monitoring System for Water Quality Management
in the Ohio Valley," Environmental Science and Technology,
2(Oct.): 764-771 (1968~T
Kleinert, Stanton J., Personal Communication, State of Wis-
consin, Dept. of Natural Resources, Bureau of Standards and
Surveys, Surveillance Section, Madison, Wisconsin (1971).
Lanyon, Richard, and Kurland, Jerome, "Monitoring System
Boasts Chicago Water Data Collection," Water and Wastes Engin-
eering, 8(Jan.): 20-23 (1971).
Lyon, Walter A., "Water Quality Information Systems the Key
to Cleaner Waters," Presented at the National Symposium on
Data and Instrumentation for Water Quality Management, Uni-
versity of Wisconsin Water Resources Center, Madison
(July 1970) .
Marks, R. H., "Automatic Water Quality Measurement," Power,
110 (Oct.): 92-93 (1966).
Maylath, Ronald E., "New York State Automatic Water Quality
Monitoring System," Presented at the National Symposium on
Data and Instrumentation for Water Quality Management, Uni-
versity of Wisconsin Water Resources Center, Madison
(July 1970) .
Maylath, Ronald E.,"Water Quality Surveillance in New York
State," Division of Pure Waters, New York State Dept. of
Environmental Conservation, November (1970).
McDermott, James H., Ballinger, Dwight G., and.Sayers, William
T., "The Future of Instrumentation in Water Pollution Control,"
Presented before the Instrument Society of America, Analysis
Instrumentation Division, 14th National Symposium, Philadelphia,
Penna., May 21 (1968).
McCartney, David, and Beamer, Norman H., "Continuous Recording
of Water Quality in the Delaware Estuary," Journal AWWA,
54(Oct.): 1193-1200 (1962).
187
-------
Mentink, A. F., "Specifications for an Integrated Water
Quality Data Acquisition System," U.S. Dept. of Interior,
FWPCA, Div. of Pollution Surveillance (1968) .
Mentink, A. F., "An Opinion on the Water Quality Instrumen-
tation Systems Installed at the United Gas Industries and
Merc Chemical Plants," Analytical Quality Control Laboratory,
Div. of Water Quality Research, Dept. of Interior, Cincinnati,
Ohio (1970).
Mentink, A. F., "Specifications for a Slow Speed Telemetry
System for Water Pollution Surveillance," Dept. of HEW, FWPCA,
Basic'Data Program, Water Quality Activities, Engineering and
Field Operations (1966).
Mentink, A. F., "Recommendation for Digitizing Output Data
from Water Quality Monitors to be Installed near Elking, West
Virginia for Acid Mine Drainage Project," Water Quality
Section, Basic Data Branch, Div. of Water Supply and Pollu-
tion Control, U.S. Public Health Service, Cincinnati, Ohio
(May, 1964).
Nauman, Jon W., Personal Communication, Geological Survey,
Water Resources Division, Anchorage, Alaska (1971).
Nauman, Jon W., and Cory, Robert L., "Cost of Patuxent River
Quality Monitoring," Journal AWWA, 62 (Dec.); 796-98 (1970).
O'Brien, James E., and Olsen, Rolf A., "Evaluation of
Automatic Chemical Analysis Monitor for Water Quality Para-
meters," Journal WPCF, 42(3): 380-90 (1970).
O'Neal, Gary L., Personal Communication, EPA/WQO Chief,
Technical Assistance and Investigations, Portland, Oregon
(1971) .
Ohio River Valley Sanitation Commission, "ORSANCO Quality
Monitor," October issue (1970).
Ohio River Valley Sanitation Commission, "Twenty-First Yearbook"
(1969).
Palko, Michael F., Personal Communication, State of Washing-
ton, Dept. of Ecology, Water Pollution Control Commission,
Olympia, Washington (1971).
Palmer, M. D., Personal Communication, Environmental Engin-
eer, Water Quality Surveys Branch, Ontario Water Resources
Commission, Toronto, Ontario (1971).
Palmer, M. D., "Submersible Recording Current and Water Qual-
ity Meters," Water and Sewage Works, 117(Nov. 28): R64-R70
(1970) .
188
-------
Palmer, M. D., "User Trials of a Submersible Water Quality
Recording Meter/' Great Lakes Survey Program, Toronto, Ontario
\ -L y O y ) •
Palmer, Merv D., and Izatt, J. Bryan, "Determination of Some
Chemical and Physical Relationships from Recording Meters
in Lakes," Water Research, Pergamon Press, Vol. 4: 773-786
(Great Britain)(Sept. 1970).
Palmer, M. D., "Water Quality Modelling with Submersible
Recording Water Quality and Current Meters," Presented at
the National Symposium on Data and Instrumentation for Water
Quality Management, University of Wisconsin Water Resources
Center, Madison (July 1970).
Parker, W. B., "Automatic Analyses as Applied to Water Quality
Determinations and Water Treatment," Ann. New York Academy of
Science, 91: 909-914 (1961).
Raible, Raymond W., and Testerman, M. K., "Development and
Performance of a Mobile Water Quality Monitor," Advances in
Instrumentation, 24(3), Proc. 24th Ann. ISA Conf., Houston,
Oct. 27-30 (1969).
Riseman, Jean M., "Specific Ion Electrodes Versatile New
Analytical Tools," Water and Sewage Works, 117(9): IW 12, 14,
16, 18, Sept./Oct. (1970).
Sahre, Darrel, "Cleaner Streams from Busier Sewers," Water
and Sewage Works, 117(Nov. 28): R109-R112 (1970).
Sayers, William T., "Water Quality Surveillance," Environ-
mental Science and Technology, 5(2): 114-119 (1971).
Schraufnagel, F. H. , Personal Communication, Director, Bureau
of Standards and Surveys, Div. of Environmental Protection,
Dept. of Natural Resources, Madison, Wisconsin (1971).
Shubrooks, H. L., "Systems of Telemetry for Water Quality
Data Collection," Advances in Instrumentation, Proc. 23rd
Ann. ISA Conf. (1968).
Smith, Paul Ferris, "What's in the Water," Oceanology Inter-
national, 5(Oct.): P10-P11 (1970).
Smith, Ethan T., and Morris, Alvin R., "Systems Analysis for
Optimal Water Quality Management," Journal WPCF, 47 (Sept.):
1635-1646 (1969).
189
-------
Smoot, George F., "Telemetry - State of the Art," Presented at
the National Symposium on Data and Instrumentation for Water
Quality Management, University of Wisconsin Water Resources
Center, Madison (July 1970).
Tajima, G. K., "Factors Affecting Ocean Data Quality,"
Advances in Instrumentation, Proc. 24th Annual ISA Conf.,
Houston, Oct. 27-30 (1969).
Thomann, Robert N., "Lessons to be Learned when Collecting
Valid Data," Presented at the National Symposium on Data
and Instrumentation for Water Quality Management, University
of Wisconsin Water Resources Center, Madison (July 1970).
U.S. Dept. of Interior, FWPCA Div. of Pollution Surveillance,
"A Program Guide to Automated Instrumentation for Water Pollu-
tion Surveillance," October (1966).
Weiss, Charles M., and Oglesby, Ray T., "Instrumentation for
Monitoring Water Quality in Reservoirs," Journal AWWA
55 (Sept.): 1213-1219 (1963).
190
-------
APPENDIX C
REMOTE SENSING: STATE-OF-THE-ART
It is important from the outset to clearly establish the
character of remote sensing as applied to solving water
quality problems. Remote sensing basically is the acqui-
sition of information about an object without intimate
contact between the object and the data-acquisition de-
vice. The data are contained in emitted, reflected, or
scattered electromagnetic energy that is registered in
various ways for subsequent extraction and conversion to
intelligent information. A variety of airborne sensing
devices (operating during both day and night) and data-
analysis techniques are successfully used to gather this
information (Chandler, et al, 1970).
This points up the primary difference between remote
sensing and grab sampling or automatic monitoring. The
number of individual data bits gathered by remote techniques
is infinitely greater than the number gathered by either
of the other two techniques. Each of the millions of data
bits gathered remotely, whether photographically or electron-
ically, is not necessarily quantitative individually, but
may be when collectively considered. In contrast, each
data bit taken by either grab sampling or automatic moni-
toring is individually quantitative. The point then is that
in terms of quantitative bits of data, it is difficult to
compare remote sensing with other sampling techniques.
However, as will be shown later, rough comparisons can be
developed.
To gain the full benefit of remote systems with technology
and interpretative abilities currently available, objectives
for gathering data must be clearly and carefully designed.
Flights sent out with any combination of sensors to "see
what they can do" generally result in an expensive gather-
ing of meaningless and subsequently useless data. Coor-
dination of flights, types of sensors, time of day and
many other factors must be carefully calculated and incor-
porated into a scheme to effectively monitor water quality
(Schneider and Kolipinski, 1969) .
The most probable areas of application of remote sensing
techniques to water quality are divided into six areas by
Burnett and White. These six major water pollution prob-
lems are as follows:
1. Eutrophication;
2. Metalic ions and toxic metals;
3. Chemicals and oil;
191
-------
4. Thermal pollution;
5. Suspended solids from domestic sources
and mining agriculture; and
6. Saltwater intrusion into fresh groundwater
sources.
Correspondingly, there are various remote devices which seem
suited to solving these certain categories of water quality
problems. Horn (1968) has suggested the following eight
regions of application.
1. Photography. Various black and white and
color films, in combination with selective
filtering devices, provide visual spectral
analysis of given pollution scenes.
2• Thermal infrared. Thermal scanning devices
in about the 8-14 micron (y) range indicate
within a ± 0.5 C range the apparent tempera-
ture of the scene.
3. Microwaves. Comparatively long wave lengths
give groundwater temperatures and to some
degree ground water levels.
4. Airborne magnetics. Nuclear procession-type
magnetometers reveal ground water conditions
to some degree.
5. Electromagnetics. Input systems using broad-
cast radio signals have broad, undetermined
application to water quality problems.
6. Gamma-ray spectrometer. This instrument is
being developed for effective use as a good
indicator of water quality.
7 - Chemical vapors. Chemical vapors emitted
from the earth's surface may be measured
down to seven parts per billion. This fact
holds potential for airborne sampling systems.
8. Fluorescent processes. As is the case with
chemical vapors, causing hydrocarbons to fluo-
rescing materials may also be recorded to seven
parts per billion.
192
-------
Scope of Remote Sensing
A vast amount of aquatic information may be recorded with
the many sensors and interpretation techniques. Perhaps
in no other way can large areas containing water resources
be examined and the aquatic and ecologic interactions
which occur within them be studied (Horn, 1968).
The extraction of the information from the remote systems,
however, requires highly qualified personnel for the ef-
fectuation of the systems. Literally every phase of the
remote sensing operation requires experience and expertise.
The planning of the studies must be carefully and imagina-
tively tailored to accomplish specific monitoring goals.
The actual in-flight operation must be precisely conducted
to record information as designed in the pre-flight plan.
Finally, the interpretation and use of the recorded data
must be carried out by skillful and innovative personnel
to gain the full potential use from the data.
Because of the highly sophisticated nature of equipment
and personnel for the specific application of remote sens-
ing systems to water quality control, it is impractical
for most states to consider actual purchase and hiring
requirements. For example, the aerial-photographic service
at Colorado State University, which may be considered a very
complete photographic facility, has in the order of $300,000
invested in aircraft, cameras and interpretation equipment
alone (Skinner, 1970). This figure makes no account for
the personnel required to operate the equipment. Radar sys-
tems and thermal scanning systems may require many times
the cost for photographic equipment due to increased size
of aircraft and instrumentation necessary for efficient op-
eration.
The expense of purchasing these systems is not, however, a
limiting factor for use in state programs. There exist a
number of governmental, educational, and private consulting
entities through whom all the necessary functions to carry
out remote sensing investigations may be contracted. At
present, our ability to monitor pollution transport and
diffusion mechanisms is confined to known sources and speci-
fic types of pollutants, for the most part (Burnett and
White, 1970). Therefore, the effective incorporation of
remote sensing techniques into state programs must be ac-
companied by precise ground-truth correlations.
A state which desires to evaluate and improve its present
water quality monitoring system may use remote sensing to
great advantage. As previously mentioned, remote sensing
can, first of all, provide an entire drainage basin
193
-------
perspective of water systems. Water-resource systems and
the numerous subsystems can be logically identified and
incorporated into planning water quality control programs.
This fact in itself provides a necessary basis to solicit
accurate data and cooperation from the managers of all facets
of the water cycle (Welch, 1969). Intuitively, remote
sensing of water systems does not necessarily become ex-
clusive to water quality management.
If remote sensing is used for intensive studies, site se-
lection for later routine ground monitoring will be tremen-
dously refined. The true value of combining grab samp-
ling and automatic monitoring with remote sensing will
then quickly fall into proper perspective for use in state
programs. At the same time, careful evaluation of data
derived,from all three techniques will contribute to the
growing validity of using remote sensing as a quantitative
pollution monitoring device to later displace many labor-
ious ground techniques.
Physical Basis for Remote Sensing
Rempte sensing as a generic term consists of obtaining in-
formation in various portions of the electromagnetic spec-
turm using mechanical, optical, and electronic apparatus
at some distance from the target. Devices utilized for the
collection of this information are broadly classified as
either passive or active. Conventional aerial camera sys-
tems are by definition passive, whereas systems such as
side looking radar (SLAR) are active types. Passive sys-
tems sense naturally emitted or reflected energy origina-
ting from solar sources or from the object itself. In
contrast, active systems propagate energy signals which
are in turn reflected by the object or scene and electron-
ically recorded (Chandler, ejt al, 1970) .
Every natural object above absolute zero temperature emits
energy in a distribution as a function of its temperature
and its wave length. Radiation laws describing the nature
of this emitted energy are formulated with reference to a
perfect energy emitter known as a black body (Wolfe, 1965).
As the temperature of a black body increases, the emitted
energy also increases for any given wavelength. At the
same time, the wavelength at which peak energy is emitted
decreases with increasing temperature of the black body.
Interestingly, the sun delivers its peak energy in a wave
band width which corresponds to the visible range of 0.4y
to 0.7y. A lower temperature black body, 200°K for ex-
ample, delivers its peak energy at about 12y or 13y.
194
-------
No real objects, with the possible exception of the sun
which emits at about 6000 K, radiate energy as perfectly as
a black body. By definition, a black body absorbs ail
energy incident on it and, conversely, the energy which
it emits is the maximum possible at that given tempera-
ture and wavelength.
Therefore, the term describing this emitted energy—emis-
sive ity (e)— is appropriately defined as unity, or 100
percent.
These non-perfect emitters, or "gray bodies", are described
as a ratio of the actual surface emittance to the black
body emittance at a given temperature and wavelength:
e, f (wavelength) = N f(wavelength and temperature)
A B f(wavelength and temperature)
where N = real surface radiance or emittance; B = black body
radiance at emittance at the identical temperature and wave-
length; and e-^ = spectral emissivity at a given wavelength.
Now, because of the small variation in radiance for a small
wavelength band at a given temperature, the above relation
also holds for small intervals of wavelength regions. The
effect is, in essence, an average emittance for a small
interval of wavelengths.
There are three other terms which describe the remaining
properties of radiation. These three are:
p - reflected energy or the surface property
reflectivity
T, - transmitted energy or the surface property
transmissivity
a, - absorbed energy or the surface property
absorbtivity
All these parameters are again a function of wavelength.
By the conservation of evergy, the sum of these properties
must be equal to the energy incident to a surface which is
unity. This may be expressed in equation form,
1 = PA + ax + Tx
This sketch depicts the phenomena (Figure C-l).
195
-------
1 incident energy
^2-surface of an
object
Figure C-l. Energy incident upon a surface.
For a black body, the transmissivity by definition must be
zero, and Kirchoff's Law therefore applies and the emissiv-
ity must be equal to absorbtivity .
1 = PA + aX
and e, = a,
A A
then 1 = p, + e,
°r PA = 1 - £X
Technically, these equations hold when no fluorescence
occurs.
For each natural gray body surface, the total radiated power
per unit of surface area is described by the Stefan - Boltz-
man Law:
W
= J
e,W,d\ =
A A
where W = total radiated power; e = effective (or average)
total emissivity, e = 1 for a black body; a = Stefan-Boltzman
constant which is equal to 5.67 x 10~12 Watts/cm2 - °Kl
T = absolute surface temperature.
and
This total energy per unit of surface is actually the ef-
fective power delivered to a sensor which has been radiated
from an object.
196
-------
senSina svsiL USSed' Snergy sensed bv Passive remote
refle£?edYh\ S rej°rds energy emitted from an object or
the amount of ^ei°bDeCt' Whlch is Primarily a function of
™ r Snergy the ob^ect ^ceives. In a
o»T*I 2°*ive rem°te Sensin9 systems record the
pr°pagat^d fro* the system and reflected to a re-
ar te f ^ either instance, passive or active, the char-
energy surface governs the nature of the reflected
All natural surfaces are described as bidirectional re-
flectors which are actually a combination of specular
reflectance and Lambertain reflectance. Specular reflec-
tance is primarily directional, such as would occur from
concrete and water (Figure C-2). Lambertian (or cosine)
reflectors return a diffuse or scattered energy form
(Figure C-3) . Bidirectional reflectors then combine the
two, specular and Lambertian, to describe the manner in
which energy is returned from an object or scene (Figure
***•mm 4 / •
incident
energy
surface
Figure C-2. Specular reflectance.
symmetric
distribution/
L) L / > t 11 t } \ M
incident
I n 1 n n ( i\ t n
regular surface
Figure C-3. Lambertian reflectance
197
-------
incident
energy
i , ^\\\/^^*r i M ^7
(itfcl J.Ut * (^(UvyS/^TJA.^!J4 ^vV/AViRf .A/ ^L. rough, irregular
surface
Figure C-4. Bidirectional reflectance.
In summary, our success in identifying a particular sub-
stance or pollutant is dependent on our ability to measure
and detect all the above physical properties of that sub-
stance and interpret and decode the recorded energy.
The foregoing discussion was developed from Miller (1970)
and Chandler, et al (1970).
Environmental Considerations
A record of reflected and emitted energy from a scene as
sensed by an airborne remote sensing system actually de-
pends upon the physical properties of the sensor, the inter-
vening medium through which the energy must pass, and the
target with its background. The most variable of these
three factors are of course the intervening medium and
the target's characteristics.
The intervening medium (air, water vapor, dust, etc.)
scatter and attenuate the signal. The nature of the trans-
mission of the sun's energy through the atmosphere has a
crucial influence on the wavelengths of energy to reach the
earth and, in turn, the signal transmitted back to the air-
borne sensor. In the ultraviolet region of the electro-
magnetic spectrum, atmospheric absorption of radiation is
nearly complete out to about 0.3y. Our small window in the
adjacent visible region (0.4-0.7y) is relatively free from
absorption, but one must still cope with considerable scat-
tering by atmospheric particles. In the infrared out to
25y there are many absorption bands of water vapor, carbon
dioxide, carbon monoxide, nitrous oxide and ozone. So
complete is the absorption in some of these regions that
198
-------
they are for all practical purposes opaque. Absorption by
water vapor virtually closes the atmosphere from 25y to the
~°r~leSS arbitrary start of the microwave region at
Figure C-5 depicts the regions of absorption and transmission
for the various atmospheric constituents (Parker and Wolff,
10 12 14 16
Figure C-5.
Wavelength (micron?)
Percent solar transmission as influenced by
atmospheric constituents.
The target's physical characteristic is the second crucial
influence on the signal received by a remote sensor. Most
important of these characteristics is the emissivity (e-,)
of the target surface. Every natural substance or combin-
ation of substances has a unique emissivity. Depending on
the emissivity, the amounts of energy reflected or emitted
(for a given wavelength) vary considerably. Our ability
to distinguish substances remotely, then, is also subject
to our abilities to distinguish between the effects of
emitted and reflected energies.
Further, sky radiance, cloud cover, wind, and surface
roughness modify various wavelength intervals of the tar-
get signal.
The emphasis of the discussion is to acquaint the reader
with the myriad of factors which affect our ability to
detect water quality conditions.
Remote Sensors for Spectral Regions
To this point, the characteristics of remotely sensed targets
and the media through which the energies from those targets
must pass have been discussed. The final topic of the physics
of remote sensing is the actual description of the applica-
bility of sensors to the various spectral regions.
199
-------
Basically, there are nine regions of the electromagnetic
spectrum. The reason for these divisions is primarily due
to the types of sensors available for detection of energy
within a spectral region. Technology has provided many
means by which the interpretive powers of the eye may be
aided and amplified. Parker and Wolff (1965) have made a
simple description of these nine divisions and the classes
of sensors which record in those ranges (Figure C-6}.
Wavelength
Y
rays
x rays
r
UV:
1
i
|M";MXC
Infrared
Medium
And
Far
licrowave
i!
vhf to If
«
) <; o <;
ro ro
o o
O O
o o
O
0
o
in
00
o
Scintill-
ation
counters ;
-ray
spectro-
meters
Scanners
with
filtered
photomul-
tipliers ;
image
orthicons
& cameras
with
filtered
ir film»
2900 A
Film
in
cam-
eras
X
Cameras
with ir-
sensitive
film;
solid-
state
detectors
in scann-
ers &
radio-
meters
Solid-
state
detector;
in scann'
ers &
radio-
meters
i
Radar ;
rf re-
sceivers
-in
scann-
ers &
radio-
meters
Electro-
magnetic
pulse
techni-
ques
Figure C-6. Remote sensors for the electromagnetic spectrum.
200
-------
Due to the nature of electromagnetic radiation, obervation-
al frequencies cannot be chosen at random. Nature dictates
the spectral regions where the atmosphere will transmit
sufficient radiation to the detector. Rabchevsky (1970,
Part II) has described seven of these regions which may be
applicable to use in water quality programs.
Low, Medium and High Frequencies (103-106cm). These freq-
uency bands have been allocated primarily for communica-
tion services. Little has been done in utilizing this
part of the spectrum for earth sciences investigations;
however, considerable research is now under way in this
area.
Very High Frequency (102-103cm). This frequency range
is useful for remote sensing particularly because the size
of the antennas is small enough to be practical for air-
craft and satellites. The penetrability of ice at these
frequencies is quite excellent, allowing profiling of ice
cover over the ocean by utilizing the information from
reflectors between the interface of ice and water.
Ultra High Frequency (10-102cm). This frequency range
offers all-weather and system-size design advantages. This
may be called a true microwave, all-weather, day-night band.
The antenna size required, to produce a very narrow beam
width for resolving earth materials, however, is still too
large for installation on most aircraft and spacecraft.
Microwaves (Super High Frequency; 0,1-lOcm). This has been
the most active part of the spectrum for the design of search-
ing, ranging, tracking, imaging and research-type radars.
The low frequency of this spectrum provides all-weather
advantages and the high frequency makes the design of small
size, high resolution, antennas possible.
Infrared (0.7y-lcm). All electromagnetic radiation is pres-
ent not only in the infrared region, but also in the micro-
wave and UHF regions, with sufficient intensity to be de-
tected with appropriate radiometers.
Visible Spectrum (0.4-0.7y). Photography is the most widely
used method for remote sensing in the visible portion of
the electromagnetic spectrum. Other promising devices for
obtaining surface information are lasers, vidicon (tele-
vision) systems, and optical-mechanical scanners and sim-
ilar devices. One of the most promising techniques is
splitting the visible portion of the spectrum into narrow
multiple bands, by suitable filtering and recording tech-
niques, using cameras and other above mentioned devices.
201
-------
0
Ultraviolet (102A-Q.3y). This is another area of increas-'
ing scientific interest for the study of surface materials.
In this region greater than anticipated amounts of ultra-
violet radiation may be recorded photographically or by
optical-mechanical scanners.
Little evidence was discovered in the literature of the
possible use of x-ray and y-ray techniques for airborne
sensing of water quality conditions.
Rabchevsky (1970, Part II) has condensed the methods of
remote sensing, the corresponding working range, and the
technique of application of the sensors into Figure C-7.
In conclusion, there are many single influences which af-
fect the ability of remote sensing devices to depict
water quality conditions. Basically, however, there are
five physical considerations which must be made to opti-
mize a remotely sensed signal (Horn, 1968) - the variability
of:
1. the intensity of the spectral composition
of the illuminant;
2. the reflectivity and emissivity of the
target;
3. the sensitivity of the film or the sensor;
4. the transmissivity of filters used (optical
or mechanical); and
5. the optical and electromagnetic effects of
the intervening medium between the target
and the remote sensor.
Applications of Remote Sensing
to Water Quality
Before discussing the particular state-of-the-art water
quality applications, a perspective in using remote sens-
ing system to solve water quality problems will be reit-
erated. At the outset, the reader should recognize that
remote sensing techniques are new and therefore have had
very limited application and investigation relative to
other monitoring techniques. This literature search has
discovered no instances where remote sensing has been
actually applied with success to solving a state water
202
-------
METHOD
Scintillation Counters
Gamma Ray Spec-
trometers
(Geiger Counters, etc.)
Scanners with Filtered
Photomultipliers;
Image Orthicons &
Cameras with Filtered
Infrared Film •
>2900 A
Photography:
Conventional Film
(B&W, and Color)
Infrared Film
(B&W, and Color)
Multispectral Bands
Lasers
Photometers
Spectrophotometers
Solid State Detectors
in Scanners &
Radiometers
(Thermal Infrartc!)
Radars:
Radio Frequency
Receivers in Scanners
& Radiometers
Electromagnetic Pulse
Techniques (Sonar)
Geophysical:
Gravimeters
Magnetometers
Seismographs
Electric Logs
Geochemical
WORKING
RANGE
<.03-100A
100A-0.4ju
4000-7000&
(0.40.7 ,u)
6000-9 000&
(0.6-0.9/u )
3000-9000A
(0.3-0.9 p )
4000-7000A
(0.4-0.7/j)
4000-7000A
(0.4-0.7 p )
In Any
Spectral
Region.
Ip -1mm
lmm-0.8mm
Sonic Waves
0.8-3xlO''mm
Gravity Field
Magnetic Field
Sound Waves
Electric Field
Chemical
Reactions
TECHNIQUE
Measurement of reflected/emitted natural
radiation by means of characteristic response
of detectors.
Records incident natural radiation by "Looking"
at elements in sequence;
Imaging ultraviolet spectroscopy.
Wide Band Recording on emulsions;
Converts small color differences to larger ones.
Records absorption/reflection patterns on film.
Narrow Band Recording on emulsion.
Records backscattered radiation on photographic
film and other media.
Narrow and/or wide band recording on emulsions
and other media.
Narrow and/or wide band recording on film and
other media.
Records incident natural radiation by "Looking"
at elements in sequence.
Records backscattered radiation; various media
used for recording.
Measurement of pressure waves reflected from
terrain or objects (primary use in water).
Measurement of acceleration of gravity.
Measurement of "local" magnetic induction.
Measurement of elastic waves in earth.
Measurement of electric properties of subsurface
geologic formations.
Measurement of relative and absolute abundances
of the elements in the earth.
Figure C-7. Remote sensing methods.
203
-------
quality problem. To be sure, there are a number of appli-
cations, particularly at a research level, which give
great promise of being applied to state problems. Nowhere
was there indication that aerial remote sensing techniques
are being applied on a regular basis for data acquisition
in any state program.
Therefore, to build a framework which places the use of
remote sensing in proper prospective is crucial to con-
sidering its potential. If the reader tries to conform re-
mote sensing to the traditional basis built around the
abilities of grab sampling and automatic monitoring, the
benefit of the following discussion will be greatly dimin-
ished.
In other words, the reader should not ask himself questions
like: Does remote sensing provide dissolved solids in-
formation in parts per million? or Does remote sensing
provide an exact pH reading? The reader should, however,
begin to consider questions such as: Since remote sensing
can locate the distribution and relative concentration of
an effluent plume in a stream, where can grab samples best
be taken?
A simple analogy at this point will point up why remote
sensing is difficult to compare, in terms of quantitative
information, to grab sampling or automatic monitoring.
Every individual is equipped with a very precise remote
sensor--a pair of eyes. We use our vision to give us a
rapid, spatially large, evaluation of a given situation.
For example, we wish to know if a pot of water on an
electric stove is hot and how hot it is. We know from
experience quite a number of prarmeters which will pro-
vide this information by using our eyes only. First, we
see the pot on the right-front coil. Immediately, we
know not to consider any other coils. Next, we see that
the control knob labeled "right front" is positioned on
a "high" setting. Again, we need not consider any other
knobs, and we may infer the possibility of the water in
the pot on the right-front coil to be potentially hot be-
cause of the setting. Further, if we know how long the pot
has been on that coil and how much water is to be heated,
we have a much better chance of inferring the water's tem-
perature at that moment. Also, if we know the water's
previous condition prior to being in the pot, whether
frozen or boiling, our chance for accurate inference in-
creases. Going further, we "see" (or remote sense) that
the water is giving off steam, the water is highly agitated,
and the coil is glowing red. From our previous knowledge
and experience of the fact that we are at sea level and
204
-------
water boils at 212°F at that altitude, we may very accurately
infer the water's temperature is very close to 212°F. The
water is therefore hot!
Of course, we don't go through such a detailed deductive
process in real life, but we see the value of not actually
having to measure a given parameter by contact. The infor-
mation we derive from our eyes is not in itself quantitative,
but our inference from decinding first what we want to know
and second, what factors indicate what we want to know,
provides us with a quantitative, accurate answer.
In a similar manner, when put in the context of the "eyes"
analogy, remote sensing may provide overall perspective
and information not available from other techniques. For
example, if a color infrared photo reveals a healthy bed
of underwater plants, we actually know a great deal more
than that fact. The presence of those plants indicates
fairly high dissolved oxygen content, a tolerable pH range,
low sediment or settleable solids load (that is, the silt
is not blocking the light), and the absence of choking
bottom sludge or undegraded material competing with those
plants for oxygen.
Ground knowledge of plant varieties may indeed delimit the
range of these mentioned parameters and further aid our
data collection.
A color photograph of a scene is another example of the
use to a water manager. For example, the fact that water
is brown, indicating a turbid condition, may be sufficient
knowledge to answer a question of whether or not a farmer
is using poor irrigation practices as indicated by a muddy
surface return flow to a river. The fact that water has
2000 ppm settleable solids, or 2750 ppm settleable solids
is immaterial to answering that question. In another ap-
plication, a parts per million determination may be crucial.
By using remote sensing in this case, the expense of a lab
sample has been saved.
In summary, it is difficult to compare the value of remote
sensing, in terms of quantitative information rendered,
with information provided by grab sampling and automatic
monitoring techniques. The question, stated negatively,
then becomes, what information will remote sensing provide
that the others cannot. Thereupon, the dollar-cost eval-
uation comes into proper perspective. Therefore, the water
quality manager must decide which techniques will provide
at least costly information necessary to solve a water quality
problem within corresponding space and time constraints.
205
-------
In the previous section, the various remote sensors for
wavelength regions in the electromagnetic spectrum were
discussed. The state-of-the-art applications for these
various spectral regions will now be generally discussed
in order of increasing wavelength. The discussion for
each region is determined by appropriate investigations
found in the literature search.
The following regions will be discussed:
1. Visible Region -4-7y;
2. Thermal Infrared Region 1.0-14y;
3. Microwave Region 0.1-lOcm; and
4. Multispectral Sensor Systems
Visible Region (.4yi - .7y, photographic)
Silvestro (1969) has demonstrated that quantitative con-
centration distributions of particular pollutants can be
constructed from repeated microdensiometer scans parallel
and perpendicular to the shoreline of a river-pollutant
plume situation (Figure C-8). Silvestro examined two
120' ^
100'
80'
;-.•:: f==^
60-100 ppm I-'.V.:| 20-50 ppm E3 10-20 ppm
ix>l 6-10 ppm
1-4 ppm
20' 40'
60
100' 120' 140' 160' 180' 200'
220'
Figure C-8. Iso-concentration map of effluent flow under rough water
conditions. (After Silvestro, 1969)
categories of pollutants - molecular and particulate (in-
cluding algae). He examined each category at a theoretical,
206
-------
laboratory experimental, and field level. The spectral range
of his investigation included 0.3y to 0.9y and ten distinct
channels (spectral bands) in the visible range. His inves-
tigation revealed that ambient light used for imaging water
and water contaminants reaches the camera primarily by
the scattering mechanism. Sunlight is scattered into the
camera by the solution and thus scattered, light is the imag-
ing light. Silvestro also resolved that the concentration
of phytoplankton density generally increased reflectance
significantly in the 0.6-0.7y range.
Silvestro concluded that the applicability of aerial remote
photographic systems is dependent on the two following con-
siderations:
1. Do pollutants and water quality parameters
affect light (energy) sufficiently to ex-
tract the desired information; and
2. Can remote sensing techniques be combined
with analytic procedures to extract the
available information?
Wezernak and Polcyn (1970) have demonstrated on a gross ba-
sis, that Silvestro"s first condition is widely true. Fig-
ure C-9 demonstrated for roughly over the same spectral in-
terval , that significant differences do exist between water
bodies and types of pollutants in general. Specifically,
spectral differences exist for suspended solids concentrations
0.1 _
0.4
0.5 0.6 0.7
Wavelength (ym)
Figure c-9. Spectra comparison:
207
0.9
Raisin River.
-------
Scherz, Boyle, and Graff (1969) lend further support to
Silvestro's findings. These authors studied five classes
of fluid wastes; namely, corn cannery waste, milk waste,
municipal sewage, tannery waste, and sulfite liquor and
percent solids. Their findings show, for example, that
at 0.2y differences in percentage reflectance are about
4 percent between municipal sewage and milk waste, while
differences are about 48 percent between milk waste and
cannery waste. Differences in spectral signatures are
the keys to distinguishing between pollutants.
Further, these investigations recorded the variation of
the reflectance of distilled water with temperature over
the visible range. They reported, for example, that at
0.5y, the reflectance for 7°C water is about 10 percent
greater than is 83°C water, while at 0.35y, their reflect-
ance characteristics are identical. Unfortunately, the
authors found that lab methods may not correlate well with
field studies. Various film-filter combinations were
studied, with no combination apparently best for all cases
for depicting fluid wastes.
Cooper (1969) used two types of color film to study the
characteristics of water quality. Kodak color infrared
film type 8443 detected algal growths with ease along
river beds, "in some cases, acid stains showed brilliant
green along the shoreline through a moderate tree canopy.
Ektachrome MS film did not depict these phenomena. In
some cases, low pH (3.5-4.0) was a definite green to blue
green. Other streams with comparable pH showed blue-green
to light blue. Correlations were attempted and these
results were identified with a "Craftint Color Match De-
signer's Color Pack" selection of 219 colors.
Schneider and Kolipinski (1969) have also used color aerial
photography to identify the pH of a water body. They con-
cluded that color aerial photography is useful in identi-
fying some pollutants in streams. Streams polluted by acid
mine drainage can be identified on color.photos by bright
orange ferric oxide. The acidity of the stream can be in-
ferred. The pH above the deposits will be acid with iron
in suspension. With mixing in the region of the deposits,
the pH is about 6.0 where it meets with relatively alkaline
waters.
Thermal Infrared Region (l.Oy - 14y)
The basic physical considerations for detection of pollu-
tants within the thermal infrared region are similar to
208
-------
considerations for other regions. Colwell (1967) suggests,
among others, the following considerations:
1. Objects on the surface of the earth receive
most of their radiant energy from the sun
(therefore, time of day is crucial to the
interpretation of the data);
2. Various spectral zones of spectral bands
are recognized within the electromagnetic
spectrum;
3. Different objects reflect, absorb and re-
emit energy to differing degrees in various
spectral zones;
4. The earth's atmosphere (primarily the CO ,
O3, and H20 molecules) selectively absorbs
certain wavelengths of energy emitted by
the sun which create "atmospheric windows"
where energy of certain wavelengths passes
freely;
5. The 8y - 14 y range is one such window called
the "thermal infrared." In this region, the
reflectance is low and the emissivity high. :
This is a fact which contrasts with the
"photographic infrared" (0.7y - l.Oy range)
which has the property of high reflectance
and low emissivity; and
6. The 3y - 5y region is characterized by re-
flectance and emission being about equal.
Thermal infrared scan systems can locate thermal anomalies
day or night. A thermal infrared record in conjunction
with other records, such as aerial photography, is a valu-
able supplement to measuring specific phenomena. Some of
the most valuable information is obtained by acquiring
records simultaneously in two or more wave bands. For ex-
ample, the 3y - 5y band with one sensor and the 8y - 14y
band with another sensor (Colwell, 1967).
Richards and Massey (1966) installed an infrared thermo-
meter (IRT) on an aircraft.as an airborne radiation thermo-
meter (ART) and subjected it to a program of test and evalu-
ation as an instrument for measurement of surface water
temperatures in the Great Lakes area. The first phase of
the program dealt with the mounting and exposure of the
sensor. The second phase, in conjunction with the research
209
-------
vessel "Porte Dauphine," dealt mainly with operational
and theoretical aspects of the technique.
With some limitations, it was found that the ART provides
a good method of making fast and reliable water tempera-
ture surveys. The present instrument and installation pro-
duced an absolute accuracy of the order of 1.0°C and a
relative accuracy within 0.5°C.
Using a similar system of instrumentation, Van Lopik, Rambie,
and Pressman (1969) concluded that airborne infrared map-
ping systems, operating in the 8y - 14y portion of the elec-
tromagnetic spectrum, provide a capability for rapid sur-
veillance and monitoring of water-surface temperatures.
Specific applications include (a) determination of thermal
circulation and diffusion patterns produced in water bodies
by coolant discharge from power plants, (b) assisting in
the development of a thermal pollution index based on aerial
thermal data and water body dynamics/variations, (c) de-
tection of thermal pollution and measurement of river tem-
perature and temperature decline downstream from surface
and subaqueous heated water discharge, and (d) establish-
ment of natural seasonal and diurnal ranges of water-surface
temperatures before construction of plants that will produce
liquid effluents. One approach to the problem involves
identification of temperature-independent emissivity prop-
erties and relationships and aerial recording of energy
levels within selected narrow bands which adequately char-
acterize 4y to 14y emission curves characteristic of pollu-
tants.
Microwave Region (,0cm - 10cm)
The use of remote sensors in the microwave region has shown
some prospect as an indicator of water quality conditions.
Rabinovich, Shchukin, and Melent'yev (1968) successfully
used radiometric equipment in the 3cm range to measure the
temperature of the water surface. Ground truth data was
taken simultaneously with remotely sensed data, the average
difference between the two being only 0.3°C. The mean
square error of measurements with respect to average tem-
perature was 0.7°C. They concluded that information may be'
taken through clouds and precipitation, interference de-
creasing with increasing^ wavelength. Use of the 8cm - 10cm
range almost eliminates interference from the atmosphere
and clouds.
Aukland and Conway (1969) successfully used the microwave
region to detect the extent and thickness of oil slicks.
Examination of the1 theoretical and experimental body of
information that is presently available leads to the
210
-------
conclusion_that there are two mechanisms by which the pre-
sence of oil on a water surface may be detected. Both of
these mechanisms create an apparent temperature anomaly
when oil is present.
The first phenomenon to be considered is measuring the
local change in sea state due to the presence of the oil
pollution. This phenomenon presents very strong signals
to microwave radiometers when winds of 6 knots or more are
blowing. It is felt that this will be the primary detec-
tion mechanism for thin oil films. The second mechanism
to be considered is the direct change in the emissivity
of the water surface due to the presence of oil. This phe-
nomenon is slightly the weaker of the two but offers the
promise of measuring oil thickness. Because of the in-
dependence of these two potential detection mechanisms,
they are described separately in the following paragraphs.
The effects of oil slicks upon sea state were difficult to
determine in measurements from pier installations, as the
oil was often accompanied by detergent foam and debris.
The presence of foam caused an increase in radiometric
temperature, instead of the decrease expected from oil
alone calming the water. For relatively calm seas with
wind ripple, radiometric temperature decreased by approxi-
mately 4°K. In higher sea state,-decreases up to 10°K occur.
When oil film becomes thick, the apparent temperature in-
creases due to the dielectric layer on the water surface.
Thick films caused signatures of up to 100°K at 1.0mm
thickness.
Aukland and Conway concluded that microwave radiometers
can detect oil spills on water day or night. Care must be
taken when these techniques are applied since oil thick-
ness, incidence angle, and wind speed could render spills
undetectable.
Multispectral Sensor Systems
Wezernak and Polcyn (1970) present a good system descrip-
tion and discussion of design capacity for the multispectral
system. The passive multichannel scanner system is capable
of sensing reflected and emitted energy simultaneously in a
number of narrow spectral intervals in the ultraviolet,
visible and infrared region of the solar spectrum. The
system basically consists of a spectrometer for dispersing
the radiation spectrally and filtered detector arrays
placed at the focal points of a double ended optical-mechan-
ical scanner. Figure C-10 shows schematically the system
211
-------
operation. The scanner uses reflective optics to collect
energy over a wide spectral band between ultraviolet and
Reference/
Radiation
'Scene Radiation
Figure C-10. Schematic of multispectral scanner
and data processor.
thermal infrared. The entrance slit of the spectrometer is
placed at the focal plane of the scanner to insure that
space and time simultaneity is achieved for all channels
between . 4y and ly. Filtered detectors for the infrared
region and the use of short electronic time delays provide
coincident data for the region beyond ly. Additional de-
tectors and filters can be provided for the ultraviolet
region. Signals from the detectors are stored on magnetic
tape for later image reconstruction and data processing.
As an integral feature of the system, reference lamps, an
input proportional to the sun energy, and adjustable tem-
perature plates are viewed each revolution of the scanner
mirror. Thus, for the first time (with an optical-mechani-
cal scanner) reproducible data can be obtained which is
referenced to known radiation sources and is capable of
accounting for changes in solar illumination. These con-
trolled points in each scan line permit the use of sophis-
ticated electronic pattern recognition techniques with both
analog and digital computers (Nalepka, 1970) .
The generation of a spectrum for each element permits
machine recognition of classes of objects whose spectral
characteristics are similar within certain electronic
decision levels. In the field of water pollution control,
it is believed that remote spectral measurements can be
used to detect and identify effluent discharges.
212
-------
In order to determine spectral characteristics, the air-
borne data recorded on magnetic tape is electronically
digitized and then stored in a computer. The analog to
digital converter consists of a high accuracy (13 bit and
sign) successive approximation converter capable of pro-
viding samples in 10 microseconds. The data is formated
and coded into 48 bit words for a CDC-1604 computer. Spec-
ial computer programs have been written to edit the digi-
tal tapes and format the data for use with FORTRAN programs.
Selected portions of the objects mapped are used as
training areas for which the spectral characteristics are
computed. Once computed, a computer generated spectral
curve can be made that is referenced to the incoming solar
spectrum or to other objects in the scene. In some cases,
ground reflectance standards are used to provide absolute
reflectance calibration.
The spectral properties are then available for use with
the computer to perform spectral matching operations to
enhance the detection of selected objects of similar pro-
perties or to map the different concentrations of an ef-
fluent .
The electronic format of the data also permits a voltage
sampling operation (for any selected band) in order to
display small density differences. For example, the mix-
ing zones of a thermal discharge into a river or lake can
be studied readily by quantizing the data in the 8y - 13.5y
band into one-half degree steps, color coding each step,
and combining the set into one color composite display.
Each color represents the isotherm for a chosen temperature
level. The use of two adjustable reference plates in the
scanner system insures optimum recording and accurate re-
construction of the apparent temperature pattern.
Many other computer programs have been devised and are
employed in processing the multispectral data for particu-
lar applications.
One technique which is certain to be used extensively in
the future for noncontact data acquisition involves analy-
sis on the basis of fluorescence. This analytical technique
could be used for the detection and identification of sub-
tances such as rhodamine dye, chlorophyll, lignin-sulfonates,
and many other substances which possess the property of
fluorescence. Recent reports which describe the Fraunhofer
line discriminator (Stoertz, et al, 1969) and laser induced
emissions (Hickmann and More, 1970) indicate that airborne
fluorescence measurements are feasible.
213
-------
Carnes (1970) has spoken to the applicability of multispec-
tral scanning for solving water quality problems. Carnes
separates water pollutants into four categories:
1. thermal;
2. nutrient;
3. chemical; and
4. petro-chemical (oil)
Each pollutant has a spectral signature which may, or may
not, be capable of detection. He views the detection of
these four categories of pollutants as dependent on these
four facts:
1. Thermal maps show anomalies and pollutants
have categories, or thermal anomalies;
2. High nutrient concentrations have point sources
and an exothermic effect;
3. chemical pollutants are point sources and
are usually a different temperature than
receiving waters; and
4. Oil slicks inhibit evaporation, which pro-
duces a warmer apparent temperature than
surrounding waters.
Carnes points out that the primary pollutant is not always,
of itself, the agent that damages ecology. Similarly, the
primary pollutant may or may not provide the remotely ob-
servable effect which permits detection.
Lowe and Hasell (1969) observed the Santa Barbara oil spill
with multispectral equipment and determined the slick was
best detectable in the ultraviolet and thermal infrared
ranges.
Costs of Remote Sensing
The cost of remote sensing systems is perhaps the most
elusive aspect of applying such methods in the field. As
was previously discussed in the earlier portions of this
report, the expense of a remote sensing application is
entirely dependent upon the needs and goals of the under-
taking. However, to present the reader with a perspective
214
-------
on the costs, the following discussion will identify the
possible latitude of cost and present the few specific
costs available.
Sensing^ Equipment
A tremendous range of expenses exists for remote sensing
instrumentation. Simple systems, such as hand-held common
35-mm cameras with a two place, single engine aircraft,
may be as low as $30 - $40 per hour (Cooper, 1969). How-
ever, this makes absolutely no account for interpretation
of any kind, nor does this account for the quality or quant-
ity information produced.
In contrast, a sophisticated aerial photo system, like the
one at Colorado State University, is quite a different story,
This flight facility and interpretive laboratory utilizes
the Wild RC-8/6 inch Universal Aviogen Lense with automa-
tic trip and roll compensator. This camera, in combination
with the twin Beechcraft Aero-Commander 500-B research air-
craft, produces a 9 x 9 inch high-resolution photograph of
incomparable quality. Interpretive devices include a
Kelsh Plotter and a Wild STK-1 precision stereocomparator/
digitizer/IBM card punch. Cost of the flight facility in-
cluding pilot and operator with film, instruments and
supplies is about $200 per hour. A CDC 6400 digital com-
puter is available at an additional cost of $430 per hour.
Photointerpretation with this facility costs roughly $16
per hour.
For emphasis, it must be reiterated that information needed
and degree of interpretation dictate cost.
Radar and scanning-spot systems (thermal infrared) may have
a cost magnitude of $300 per hour for flight facilities
and equipment, alone. Interpretation or image enhancement
costs are additive to flight costs.
Application of Remote Sensing Systems
to State Programs
As an introduction to the actual form that a remote sensing
program might take at the state level, an investigation
undertaken by Neumaier and Silvestro (1969) will be dis-
cussed at some length. The purpose of their investigation
was to show how remote sensing can be used as a tool for
detecting water pollution, both qualitatively and quanti-
tatively.
215
-------
In a program sponsored by the New York State Conservation
Department, laboratory experiments were performed to de-
termine the spectral characteristics of effluents from a
paper industry, an oil industry and a sewage plant. These
data, along with theoretical studies, were used to extract
the desired information from multiband aerial photographs
taken over these discharges. The techniques employed
allow estimation of pollution concentration in a number of
cases.
The study was undertaken in three stages. The first stage
made a qualitative assessment of the potential of spectral
photography for detection of effluents. The second stage
performed laboratory experiments to determine the charac-
teristics of reflected signals from pollutants. In the
third and final stage, laboratory information was implemented
in field studies, with mathematical and physical relations
being developed to describe the nature of light reflected
from molecular and particulate pollutants.
The carry out the qualitative phase of the study, 22 miles
of shoreline along the Niagra River and Buffalo Harbor
were photographed by an aerial system. Twelve-hundred
photographs were taken in five narrow bands plus color
photos. Seventy-four waste discharges were located. Twenty
discharges were enhanced in at least one photographic band,
while fifty of seventy-four had improved contrast. De-
tection of pollutants could be improved by selection of
band widths.
The laboratory phase was then carried out. The spectral
signatures of different industrial effluents were examined
with a spectrometer over the entire visible spectrum at
different concentrations and depths with various external
illuminating conditions. The spectral reflectance char-
acteristics of discharges from an oil refinery, a paper
mill, and a primary treatment plant were measured at five
concentrations and at least six different depths.
The final field measurements revealed some significant
statistics. Neumaier and Silvestro stated with 90 percent
confidence that the precision of the photographic system
(versus ground data) was ±6 percent. Quantitative informa-
tion for pollution studies not easily available by other
means can be extracted from aerial photos. High potential
exists for the application of color photography to identi-
fying water pollutants.
Neumaier and Silvestro reached a significant conclusion<
In order to use aerial photography for quantitative
216
-------
measurements, rigid controls must, be maintained. Film his-
tory must be noted before processing. Film must be cali-
brated with gray scales of know density. A computer pro-
gram is used to obtain corrections for variations in expo-
sure with position in the format, variations in filter
transmittance with wavelength and viewing angle, camera
flare, and transmission factors of the atmosphere. In ad-
dition, the signal detected from the air varies with sur-
face characteristics of the water, type and concentration
of pollutant, nature of the bottom, depth, irradiance,
wavelength, the angle of illumination, and the observation
angle. When all these factors are taken into account,
quantitative aerial data can be obtained which is_ repeat-
able and accurate.
Further, the behavior of the effluent plume, as determined
from a quantitative analysis of aerial photography, can be
used to determine mixing characteristics of receiving
waters. Contours of pollutant concentrations may be plot-
ted manually or automatically by microdensitometer scans
parallel and perpendicular to the shoreline.
The techniques used in the preceding study conducted by
Neumaier and Silvestro is indicative of the thinking nec-
essary to use remote sensing in state programs. Since
states have some form of ground monitoring systems, the
correlative field data for many potential application is
already established.
Water uses vary considerably throughout the United States.
For this reason, the particular water quality requirements
for the states are difficult to quantify in terms of use.
In this light, a general discussion of how remote sensing
could be used in state water quality monitoring problems
would have little specific value for designing programs.
For example, Colorado would have little interest in water
quality criteria for protection of shellfish, whereas
Hawaii would have little concern over stream quality when
all the water is diverted for irrigation. On the other
hand, quality of drinking water supplied by wells would
be common to both states' interests.
So then, in discussing the applicability of remote sensing
to state water quality programs, the emphasis must be on
techniques of data acquisition, and the application will
be dependent on the imagination of whoever designs the
program. In order to present the breadth of possible
water quality problems throughout the states, Oswald (1969)
has developed a table of water quality criteria versus
water use (Figure C-ll).
217
-------
Figure C-ll. Quantitative estimates of water quality criteria for water resources
Use
Criteria
Epidemiol-
ogical
(E. coif MPN)
Dissolved
Oxygen
D.O.
pH
Biochemical
Oxygen Demand
(BOD)
Algal Growth
Potential (AGP)
Grease &
j_j Floating
CO Solids
Turbidity
(Non-biologi-
cal)
Color
Temperature
Toxic
Materials
Drinking
<1.0 org/100rnl
<10% of 10ml
Portions Exam-
ined per Month
Adequate to
Prevent Odors
5-10 mg/1
0.8-7.2
OK up to 8.0
<2-3 mg/1
5-10 mg/1
None Allowable
None Allowable
None Allowable
10-20°C
None Detri-
mental to Human
Pb, NO3", CN, F,
Zp Heavy Metals
Industrial
Depends on
Industry
Nill for
Boiler Feed
Water
6.8-7.2
Process
Dependent
Low, Depends
on Industry
5-10 mg/1
Low
Usually None
Depends on
Process
10-20°C
Depends on
Process
r
Agricultural
Low for Truck
Crops
Must not
Deplete D.O.
in Soil
None
Low to Prevent
• Loss of D.O.
Soil Sour or
Clogging
50 mg/1
Low to Pre-
vent Soil
Clogging
Not too
Important
Not too
Important
Not
Important
Low Mono-
valent Cations
Low Boron
Swimming
240-2400 org/
100ml 20%
N.T.E. 1000
Adequate to
Prevent Odors
6.5-8.5 Some
Alk. takes
up to 10
Low Enough to
Maintain D.O.
10-20 mg/1
None Allowed
Low to Permit
Light Trans-
mission
Should Not
Absorb Light
30-35°C
None Detri-
mental to
Humans
Recreation
100-5000 per
100ml*
Adequate to
Prevent Odors
6.5-8.5
Low Enough .to
Maintain D.O.
10-20 mg/1
.Low, Wild Fowl
Protection
Low to Permit
Light Trans-
mission
Not too
Important
25-30°C
None Harmful
to Equipment
Acids, etc.
Fish &
Wildlife
None
5 ppm or
Greater, De-
pends on Spe-
cies, pH, Temp.
6.5-8.5
Depends on
Species, D.O.
Low Enough to
Maintain 5 ppm
D.O.
<20 mg/1
Low, May Limit
Oxygen Transfer
Low to Prevent
Silting or Gill
Obstruction
Not too
Important
5-20°C
Cu-0.001 ppm
H2S-0.05 ppm
CR-0.05 SO ~S
C12-0.3
Phenol 0.1
Shellfish
<70/100ml OK
70-700/100ml bad
>700/100ml Imposs-
ible
Never Totally
Absent, 2-5 mg/
1 min Depends on
Temp .
6.5-8.5, Depends
on Species
Low Enough to
Maintain D.O.
Prevent Closeup
<20 mg/1
Low to Prevent
Gill Clogging
Low to Prevent
Silting or Gill
Obstruction
Not too
Important
8-15°C
Very Sensitive
to Org Acids Cu,
CR & High Bac-
terial or Algal
Populations
*California Board of Public Health STDS for Water Contact Sports - Salt Water
1. No more than 10% of samples in one month in excess of 1000 MPN
2. No single sample in excess of 10,000 if confirmed by duplicate sample within
hrs.
-------
There^have been a number of applications of remote sensing
techniques throughout the United States. Many of these
applications have potential for wide-spread use in state
programs of specific types. The salt-water marsh mapping
studies in North Carolina would have application to most
coastal states, for example. The following paragraphs
discuss a variety of present state applications.
Schneider and Kolipinski (1969) have applied color photo-
graphy to locating fresh water springs which produce
turbulence in offshore sea areas. In a North Dakota study,
species of grasses served as indicators of salinity in
potholes. Further, color photos of the Great Salt Lake
yielded sediment circulation and movement. In still other
studies in the Florida Everglades, photos revealed inland
penetration of salt water. There existed a marked change
in flora from fresh to salt water.
Stroud and Cooper (1968) mapped in North Carolina relatively
pure stands of salt marsh vegetation with color photographs.
Many states have coastal estuaries which could be investi-
gated by such techniques.
Lipley and Palmer (1967) successfully used remote sensors
to locate off-shore fresh water springs in the Hawaiian
Islands. In this project, spectral photography and infra-
red thermometry are being evaluated as sensors to develop
inexpensive and reliable methods of detecting and measur-
ing these springs. Photo techniques did not yield specific
results.
Coastal springs are anticipated to be particularly amenable
to detection and mapping by synoptic sensors due to the
large surface salinity and temperature anomalies found in
field investigations. The infrared thermometer is appli-
cable to location of some coastal springs.
Some of the suspected physical differences between spring
water and sea water supporting the search for instrument-
ally detectable optical contrasts are, according to Lepley
and Palmer (1967):
1. Fresh water contains different dissolved and
suspended organic matter, i.e., tanning dyes
from land vegetation in fresh water as opposed
to yellow "gelbstoff" from chlorophyl-bearing
phytoplankton in sea water;
2. Spectral interference effects from thin organic
films on either water surface may be especially
detectable in the near infrared (visible sea
slicks from such films are known to sometimes
align themselves with subsurface water
219
-------
structures, such as internal waves);
3. Exsolving gasses bubbling from cold float-
ing spring water warmed by sea water may
throw small droplets of water into the air.
This effervescent mist may be detectable by
its scattering of infrared light (an infra-
red "fog" layer);
4. Capillary surface waves on the sea water
surface may be smoothed by the spreading
surface layer of fresh water. The conse-
quent modification of surface-reflected
skylight caused by the smooth surface can
be enhanced by a comparison of images photo-
graphed through polarizing filters at vari-
ous orientations.
Howell (1969) has successfully conducted airborne measure-
ments of relative humidity, temperature and air currents.
These data are telemetered to a ground-based station and
recorded. A radar unit tracks the aircraft and electronic-
ally plots its position on a base map of the area being
studied. Thus, the distribution of atmospheric conditions
can be directly related to the underlying terrain and veg-
etation features. Consequently, such systems could be
indicative of large scale water quality features in the
states.
Application of remote systems has been made in California.
Studies were conducted at Lake Tahoe, La Jolla, San Diego
Bay, and the University of California at Davis Campus.
Sea dyes were evaluated for tracing tidal and current move-
ment by multiband and color aerial photography. Preliminary
studies were made into depth penetration capabilities of
aerial photos for mapping and identifying aquatic vegetation
and bottom sediments. The studies identified black and
white spectral bands useful for tracing Rhodamine dye and
evaluating sediment conditions. The studies indicated that
color film was more convenient to use than multiband photo-
graphy for water studies (Welch, 1968) .
State Application of_ Satellite Programs
Imagery taken by existing and proposed satellite reconnais-
sance programs may provide a valuable input to the remote
sensing programs within the states. Jamison (1970) reports
that the utility of the proposed Earth Resources Observa-
tion Satellite (EROS) Program as an information source for
operational state agencies and academic institutions has
been explored in a one-year study within the State of
220
-------
Washington. Operational personnel from most natural re-
sources and land use oriented state agencies participated
in the project. Research personnel from the two major
state academic institutions (University of Washington and
Washington State University) carried out studies of the
imagery.
High altitude, small scale photographs, both singly and in
mosaic form, were used to simulate Earth Resources Techno-
logy Satellite (ERTS) imagery. Participants were asked to
relate their data needs to the proposed ERTS imagery and
the timeliness of Satellite surveillance.
It was found that significant use of ERTS-A imagery can be
made by state agencies and researchers at state educational
institutions within the State of Washington. Without ques-
tion, additional uses for the imagery will be discovered
when the actual satellite images become available and a
cost is known. One major new use to be anticipated, if the
cost is low, is reducing the expense of existing methods of
field surveys by directing sampling only to areas with a
high probability of a change of specific interest. Another
major use is the acquisition of data (such as snowline and
rate of melt) that has, heretofore, been difficult or costly
to obtain.
221
-------
Literature Cited
Aukland, J. C. , and Conway, W. H., "Detection of Oil slick
Pollution on Water Surfaces with Microwave Radiometer Sys-
tems," Proceedings of the Sixth Symposium on Remote Sensing
of Environment, Univ. of Michigan, Ann Arbor, pp 789-796
(1969).
Burnett, E. S., and White, P. G. , "Water Pollution - the Role
of Remote Sensing," Journal of Remote Sensing, Vol. 1, No. 3,
June-July, pp 9-14 (1970).
Carnes, Richard C., "Multispectral Scanning for Water Pollu-
tion," Bendix Aerospace Systems Division, Ann Arbor, Michigan,
February, 3 pp (1970).
Chandler, P. B., Dowdy, W. L., and Hodder, D. T., "Study
to Evaluate the Utility of Aerial Surveillance Methods in
Water Quality Monitoring," Prepared for State Water Res-
ources Control Board, The Resources Agency, State of Cal-
ifornia, Pub. No. 41, Sept., 98 pp (1970).
Colwell, Robert N., "Agricultural and Forestry Uses of
Thermal Infrared Data Obtained by Remote Sensing," Thermo-
physics of Spacecraft and Planetary Bodies, Progress in
Aeronautics, Gerhard B. Heller, ed., Vol. 20, pp 517-540
(1967).
Cooper, Robert A., "Application of Color Aerial Photography
in Water Pollution Studies," American Society of Photo-
grammetry, Falls Church, Virginia, pp 99-106(1969).
Hickmann, G. D., and Moore, R. B., "Laser Induced Fluor-
escence in Rhodamine B and Algae," Presented at the 13th Con-
ference on Great Lakes Research, Buffalo, New York (1970).
Horn, Leonard W., "Remote Sensing of Water Pollution," Journal
of Water Pollution Control Federation, Vol. 40, October, 1728-
1738 (1968).
Howell, Ralph L., "Equipment and Techniques for Low-Alti-
tude Aerial Sensing of Water-Vapor Concentration and Movement,"
Remote Sensing of Environment (an Interdisciplinary Journal),
Vol. 1, No. 1, March, pp 13-30 (1969).
Jamison, David W., "Earth Resources Observation Satellite -
Image Utility at State Level," State of Washington, Dept. of
Natural Resources, April, 138 pp (1970).
Lepley, Larry K., and Palmer, Leonard A., "Remote Sensing of
Hawaiian Coastal Springs Using Multispectral and Infrared
Techniques," Technical Report No. 18, Water Resources Res-
earch Center, University of Hawaii, Honolulu, Hawaii, August,
39 pp (1967).
222
-------
Lowe, D. S., and Hasell, P. G-, "Multispectral Sensing of
Oil Pollution," Proceedings of the Sixth Symposium on Remote
Sensing of Environment, University of Michigan, Ann Arbor,
Michigan, pp 755-765 (1969).
Miller, Lee D., "Notes on the Physics of Remote Sensing,"
Dept. of Watershed Sciences, Colorado State University,
Fort Collins, Colorado 80521 (1970).
Nalepka, R. F., "Investigation of Multispectral Discrimina-
tion Techniques," Final Report No. 2264-12-F, Willow Run
Laboratories, Institute of Science and Technology, the Uni-
versity of Michigan, Ann Arbor, Michigan (1970) .
Neumaier, Gerhard, and Silvestro, Frank, "Measurement of
Pollution Using Multiband and Color Photography," New Hori-
zons in Aerial Photography, American Society of Photo-
grammetry., Falls Church, Virginia, pp 47-58 (1969).
Oswald, William J., "Remote Sensing Data and Evaluation of
Water Quality," Proceedings First Annual International
Remote Sensing Institute Symposium, Sacramento, California
Vol. 2, pp 142-153 (1969).
Parker, Dana C., and Wolff, Michael F., Associate Editor,
"Remote Sensing," International Science and Technology, July,
pp 20-31 (1965) .
Rabchevsky, George A., "Remote Sensing of the Earth's Sur-
face - Fundamentals of Remote Sensing," Part I of III Parts,
Journal of Remote Sensing, Vol. 1, No. 3, June-July, pp 15-17
(1970).
Rabchevsky, George A., "Remote Sensing of the Earth's Sur-
face - Remote Sensing and Sensors," Part II of III Parts,
Journal of Remote Sensing, Vol. 1, No. 4, August-Sept.,,
pp 5, 15-17 (1970) .
Rabinovich, Yv. I., Shchukin, G. G., and Melent'yev, V. V.,
"Determination of Water Surface Temperature from Radio;
Emission in the Centimeter Range," Soviet Hydrology; Selected
Papers, Issue No. 3, pp 266-270, from Transactions of the
Voyeykov Main Geophysical Observatory (Trudy GGO), No. 222,
pp 49-53 (1968) .
Richards, T. L., and Massey, D. G., "An Evaluation of the
Infrared Thermometer as an Airborne Indicator of Surface
Water Temperature," Canada Dept. of Transport, Meteorologi-
cal Branch, January 4, 20 pp (1966).
Scherz, James P., Boyle, William C., and Graff, Donald R. ,
"Photographic Characteristics of Water Pollution," Photo-
crrammetric Engineering, Vol. 35, No. 1, pp 118-120 (1969).
223
-------
Schneider, William J., and Kolipinski, Milton C., "Appli-
cations of Color Aerial Photography to Water Resources
Studies," American Society of Photogrammetry, Falls Church,
Virginia, pp 257-262 (1969).
Silvestro, Frank B., "Quantitative Remote Sensing of Water
Pollution," Proceedings of the 15th Annual Tecnnical Meeting,
"Man in His Environment," Institute of Environmental Science,
Anaheim, California, April 20-24, pp 611-618 (1969).
Skinner, M. M., Personal Communication with M. M. Skinner,
Assistant Professor of Civil Engineering, Engineering Res-
earch Center, Colorado State University, Fort Collins,
Colorado 80521 (1970).
Stoertz, G. E., Hemphill, W. R., and Markle, D. A., "Air-
borne Fluorometer Applicable to Marine and Estuarine Studies,"
Marine Technology Society Journal, Vol. 3, No. 6, pp 11-26
(1969) .
Stroud, Linda M., and Cooper, Arthur W., "Color-Infrared
Aerial Photographic Interpretation and Net Primary Product-
ivity of a Regularly-Flooded North Carolina Salt Marsh,"
Report No. 14, Water Resources Research Institute of the
University of North Carolina, November, 86 pp (1968).
Van Lopik, J. R., Rambie, G. S., and Pressman, A. E., "Pollu-
tion Surveillance by Noncontact Infrared Techniques," Journal
of the Water Pollution Control Federation, March, pp 425-438
(1968) .
Welch, Robin I., "Remote Sensing in California Water Resource
Management," Papers of the 34th Annual Meeting, American
Society of Photogrammetry, Falls Church, Virginia, pp 330-
344 (1968).
Welch, Robin I., "Remote Sensing Systems in Water Quality
Control Programs," Proceedings, First Annual Symposium of
the International Remote Sensing Institute, Sacramento,
California, Vol. 2, pp 46-65 (1969).
Wezernak, Chester T., and Polcyn, Fabian C., "Pollution Sur-
veillance and Data Acquisition Using Multispectral Remote
Sensing," Water Resources Bulletin, Vol. 6, No. 6, November-
December, pp 920-934 (1970).
Wolfe, W. L. (ed.), "Handbook of Military Infrared Photo-
graphy," Office of Naval Research (1965).
224
-------
APPENDIX D
COLORADO'S STREAM CHARACTERIZATIONS
The following river basin descriptions and data sum-
maries represent the stream characterizations utilized
for this study. The Arkansas, Colorado, and South Platte
Rivers have a subsection presenting their data. The other
rivers are covered in a final subsection.
Arkansas River Basin
The upper Arkansas River Basin lies entirely within Colo-
rado where it drains approximately 26,140 square miles.
The river runs through Colorado for approximately 350 miles
and its maximum drainage width is 170 miles.
The Arkansas River begins in the Rocky Mountains near Lead-
ville at an elevation in excess of 13,000 feet. It flows
south by Buena Vista and Salida, turns east just after
Salida, emerges from the mountains near Canon City, enters
the high plains at Pueblo, and, if the water escapes being
used for irrigation, it leaves the state near Holly. Below
Pueblo the major tributaries are Fountain Creek, St. Charles,
Huerfano, Apishapa, and Purgatoire Rivers. John Martin
Reservoir is located on the main stem just below its junc-
tion with the Purgatoire River.
In the mountain headwater region of the Arkansas River, the
precipitation averages over 20 inches per year, while down-
stream from Pueblo in the high plains the rainfall averages
less than 15 inches per year. The water originating in the
mountains is the result of light rainfall or melting snow,
while on the plains intense rainfall produces flash floods
and intermittent stream flow.
Arkansas River water in Colorado is used primarily for
irrigation, municipal and industrial water supplies, and
for recreational purposes. Limited water supplies in the
basin have resulted in water being imported from the western
slope of Colorado. This indicates the importance of water
quality control in the Arkansas River Basin.
Water quality of the main stem of the Arkansas River is char-
acterized in Figures D-l, D-2, D-3, and p-4. Figure D-l is
a composite, while Figure D-2, D-3, D-4 illustrate the
variations. Data for these graphs is given in Table D-l.
Table D-2 contains similar data for the tributaries of the
Arkansas.
225
-------
to
to
o
_l
u.
3000
2000
1000
4000
2000
1000
MILES
FLOW(cfs) - *
PH
TDS(mg/l) - O
DO(mg/l)
BOD(mg/l| . -ft
10 20 30 40
Figure D-l. Arkansas Main Stem Water Quality Characterization with
Proposed Sampling Point Locations.
-------
to
ro
O
Q
Stream Standard
0 10 20 30
Miles
1
1
0)
H
i-H
-H
>
(d
i
4J
•H
CJ
a
0
c
u
i
1
o
^
(U
^j
P!
1
1
(d
-P
W
0)
a
o
s
i
1
id
^j
c
^3
(-3
OS
1-3
l
1
0)
(d
e
-H
c
<
u
rd
i-q
1
1
S-l
td
e
(d
i^l
i
1
>i
•-i
r-l
o
K
Figure D-2.Mean Dissolved Oxygen for the Arkansas Main Stream with Standard Deviation
and Stream Standard Illustrated.
-------
13 .
NJ
ho
oo
Figure D-3. Mean Biochemical Oxygen Demand for the Arkansas Main Stream with Standard
Deviation Illustrated.
-------
to
N>
Stream Standard
Figure D-4. Mean Total Dissolved Solids for the Arkansas Main Stream with Standard
Deviation and Stream Standard Illustrated.
-------
Table D-l. Arkansas Main Stem Stream Characterization Data
5ta
10
IP5
2S6
3S
4P
5P
6S
7S
8S
9S
Location
Holly
Lamar
Las Animas
La Junta
Nepesta
Pueblo
Canon City
Salida
Leadville
DO (mg/1)
M1 S D2 S S3
7.5 1.9 5.0
8.2 2.0 3.0
7.9 2.1 3.0
7.4 1.9 3.0
6.4 1.7 3.0
7.6 1.1 5.0
8.0 2.4 5.0
8.1 2.1 6.0
7.9 1.3 6.0
BOD (mg/1)
M S D
2.0 0.8
1.8 0.8
2.1 1.0
6.5 5.6
6.1 2.4
1.8 1.0
1.5 0.7
1.6 0.5
2.6 2.8
PH
M S D S S
8.0 0.5 6.5-8.5
7.9 0.3 5.9-9.0
7.9 0.3 5.9-9.0
8.0 0.4 5.9-9.0
7.9 0.4 5.9-9.0
8.1 0.4 6.5-8.5
8.1 0.4 6.5-8.5
7.9 0.4 6.5-8.5
7.6 0.7 6.5-8.5
TDS (mg/1)
M S D S S
3700 640 Agric
3500 1300 Agric
2500 1000 Agric
1300 360 Agric
650 220 Agric
480 220 <500
180 58 <500
130 34 <500
120 34 <500
Flow1* (cfs)
M
233
224
214
251
683
707
718
626
70.8
ro
ui
o
JMean
2Standard Deviation
3 Stream Standard
"*USGS Data, Surface Water Records
5WPCD Station Number, with P Indicating Primary Station Designation
6WPCD Station Number, with S Indicating Secondary Station Designation
-------
Table D-2. Arkansas Tributary Stream Characterization Data.
Sta
No
10 S
US
12S
13S
14S
Tributary
Purgatoire
(near Las Animas)
Purgatoire
(below Trinidad)
Apishapa
(near Fowler)
Huerfano
(near Boone)
Cucharas
(below Walsenburg)
Parameter
DO (mg/1)
BOD "
pH
TDS
Flow (cfs)
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
M
8.37
1.74
8.13
3038.19
134.00
6.85
1.39
8.17
2784.00
87.50
7.36
4.01
7.93
1489.64
34.30
6.84
3.61
8.13
3366.30
—
8.04
2.28
8.41
2776.62
S D
2.41
0.93
0.33
957.81
1.55
0.52
0.26
863.82
1.95
2.15
0.32
808.46
2.37
6.67
0.45
1655.36
1.63
1.24
0.28
1374.49
S S
3.0
5.0-9.0
Agric.
3.0
5.0-9.0
Agr ic .
3.0
5.0-9.0
Agric .
3.0
5.0-9.0
Agric .
3.0
5.0-9.0
Agric .
231
-------
Table D-2. (Continued}
Sta
No
15S
16P
17S
Tributary
Fountain Creek
(at Pueblo)
Fountain Creek
(below Colo Sp)
Fountain Creek
(near Manitou)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
7.85
3.37
8.06
1588.56
-
5.21
29.10
7.85
491.89
11.8
7.71
1.44
8.03
126.57
S D
2.23
3.20
.42
427.64
1.88
22.23
.39
304.01
2.16
.63
.45
17.57
S S
Agric .
Agric .
6ppm
6.5-8.5
<500rag/l
232
-------
Colorado River Basin
The Colorado River and its tributaries which originate in
the state represent the major undeveloped water supplies in
Colorado. The river begins in the Rocky Mountain National
Park where peaks are more than 14,000 feet high. The river
runs south for a short distance and then flows southwest
until it enters Utah at approximately 4,500 feet. The
Colorado River drainage area in Colorado is approximately
17,900 square miles and has an estimated average annual
yield of 11,150,000 acre-feet. Trans-mountain diversions
export large quantities of this water to the South Platte
and Arknasas River Basins.
High mineral concentrations are characteristic of Colorado
River water as it crosses the mesa and plateau area of
western Colorado. These concentrations of dissolved solids,
sodium, sulfate, and chloride increase downstream. The
mineral concentrations are caused by the geological character
of the terrain over which the river flows.
The major tributaries to the Colorado River in Colorado are
the Blue River, Eagle River, Roaring Fork River, Gunnison
River and Uncompahgre River.
Figures D-5, D-6, D-7, and D-8 are used to graphically
characterize water quality of the main stem of the Colorado
River. Tables D-3 and D-4 contain the data used in char-
acterizing the Colorado River.
South Platte River Basin
The South Platte River begins in the front ranges of the
Rocky Moutains and flows in a northeast direction for 450
miles until it junctions with the North Platte River in
Nebraska. Its drainage area consists of 24,030 square
miles of which 19,022 are in Colorado. The basin lies
within the semi-arid plains region where the average pre-
cipitation is from 12 to 14 inches. The snow melt and
mountain showers constitute a large part of the flow.
Clear Creek, St. Vrain Creek, Little Thompson River, Big
Thompson River, and Cache la Poudre River are the main trib-
utaries. All of these drain mountain regions to the west
of the main stem and are perennial. They contribute heavily
to the regular flow of the main stem.
Irrigation constitutes the large use of South Platte
water. The basin is estimated to be 69 percent farm land,
most of which is fertile, irrigated crop land. Currently,
233
-------
$000
4000
4000
-------
U>
13-
12-
11.
10.
9.
8.
7-
6.
5-
4
3
2
1
O
Q
Stream Standard
10 20 -
Miles
,
en
-p
o
2
0)
in
4J
o
Q
-P
CO
rtS
U
I
23
O
I
nj
U
n) n3
•P g
•H O
Figure D-6. Mean Dissolved Oxygen for the Colorado Main Stream with Standard Deviation
and Stream Standard Illustrated.
-------
N3
U>
13 -
12 .
11 -
10 _
9 -
8
7 -
Q
O
ffl 6
Figure D-7. Mean Biochemical Oxygen Demand for the Colorado Main Stem with Standard
Deviation Illustrated.
-------
Figure D-8. Mean Total Dissolved Solids for the Colorado Main Stream with Standard
Deviation and Stream Standard Illustrated.
-------
Table D-3. Colorado Main Stem Stream Characterization Data.
Sta
No
45S
46P
47S
48S
49S
50P
Location
Hot Sulphur
Springs
Dotsero
New Castle
Cameo
Fruita
Loma
DO (mg/1)
M S D S S
7.5 1.9 4.0
7.2 2.3 6.0
8.2 2.0 6.0
7.9 2.5 5.0
7.4 2.1 5.0
7.3 2.4 5.0
BOD (mg/1)
M S D
1.6 0.7
2.1 2.1
1.3 0.5
2.4 1.6
2.3 1.3
3.4 2.1
PH
M S D S S
7.9 0.4 6.0-9.0
8.2 0.6 6.5-8.5
8.1 0.5 6.5-8.5
8.0 0.4 6.5-8.5
8.1 0.4 6.5-8.5
8.1 0.5 6.5-8.5
TDS (mg/1)
M S D S S
102 21 500
261 57 500
492 158 500
532 205 500
729 241 500
773 269 500
Flow (cfs)
M
272
2,082
3,888
3,831
5,692
KJ
OJ
00
-------
Table D-4. Colorado Tributary Stream Characterization Data.
Sta
No
52P
53P
54P
55P
56S
Tributary
Eagle River
(at Gypsum)
Roaring Fork
(at Mouth)
Gunnison
(S.E. of G. Junct)
Uncompahgre
(Tributary to
Gunnison River
at Delta)
Gunnison
(near Delta)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
7.60
1.31
8.09
501.26
561
8.02
1.39
8.23
366.71
1,367
7.95
1.36
8.28
721.41
2,558
7.78
1.97
8.20
1451.86
274
t
7.94
1.31
8.26
376.57
S D
2.25
.56
.53
209.14
2.48
.59
.58
95.42
1.91
.43
.45
322.15
2.02
.97
.54
496.87
1.90
.46
.59
186-69
S S
6 ppm
6.5-8.5
Agric
6 ppm
6.5-8.5
<500 mg/1
5 ppm
6.5-8.5
<560 mg/1
4 ppm
6.0-9.0
<500 mg/1
5 ppm
6.5-8.5
<500 mg/1
239
-------
Table D-4. (Continued)
Sta
No
57S
58S
59S
87S
74S
Tributary
Gunnison
(west of Gunnison)
Taylor River
(Tributary to
Gunnison River
at Almont)
Tomichi Creek
(Tributary to
Gunnison River
at Gunnison)
Rock Creek
(nea-r McCoy)
Gore Creek
(at mouth)
(Tributary to
Eagle River)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
pH
' TDS
Flow
DO
BOD
pH
TDS
Flow
M
8.60
1.44
8.00
140.79
775
8.19
1.20
7.95
86.13
338
7.67
1.63
7.85
177.75
166
9.97
2.00
7.50
230.00
—
7.87
3.17
8.28
144.16
-
S D
2.31
.76
.46
21.59
2.01
.97
.47
22.73
2.41
.71
.38
41.37
.40
.36
14.53
2.28
4.85
.65
108.39
S S
6 ppm
6.5-8 .5
<500 mg/1
6 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
Agric
240
-------
Table D-4. (Continued)
Sta
No
75S
77S
76S
83S
79S
78S
Tributary
Gore Creek
(Tributary to
Eagle River at
Bighorn Subdiv)
Ten Mile Creek
(at Kokomo)
(Tributary to
Blue River)
Eagle River
(at Aron Bridge)
Eagle River
(near Pando)
Uncompahgre
(at Ridgway)
(Tributary to
Gunnison River)
East River
(Tributary of Taylo
River at Confl.
with Taylor)
Parameter
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
C BOD
pH
TDS
M
7.90
1.56
7.88
61.19
27.3
9.07
.70
7.40
213.83
-
8.28
2.66
7.79
287.60
-
8.52
2.84
7.41
86.57
51.0
7.87
1.06
7.91
448.23
155
7.89
.97
8.01
174.95
S D
2.12
1.46
.46
22.63
1.80
.17
.31
172.07
1.63
2.95
.52
207.67
1.42
4.02
.37
40.90
1.65
.40
.40
154.03
2.21
.85
.46
24.77
S S
6 ppm
6.5-8.5
Agric
6 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
Agric
6 ppm
615-8.5
<5bO mg/1
4 ppm
6.0-9.0
<500 mg/1
6 ppm
6.5-8.5
<500 mg/1
Flow 341
241
-------
municipal and industrial water needs are competing with
agriculture for the water and the competition grows as the
urbanization of the basin continues at a very:rapid rate.
Figures D-9, D-10, D-ll, and D-12 graphically describe the
water quality of the South Platte main stem. Figure D-9
is a composite, while Figures D-10, D-ll, and D-12 illus-
trate variations in the river. Tables D-5 and D-6 give
the water quality data for the main stem and tributaries,
respectively.
Other River Basins
The three river basins discussed so far are the major streams
with respect to flow and population. The rivers mentioned
here are sizeable, but they do not have large drainage
basins or high concentrations of human activity.
The Dolores River is located in the southwestern part of
Colorado and is a tributary to the Colorado River junction-
ing with the main stem outside of Colorado. The Dolores
River drainage area contains little population. Stream
characterization data for the Dolores River is in Table D-7.
The Green Yampa Basin is in northwestern Colorado and these
waters also reach the Colorado River outside the state.
The Green River just barely dips into Colorado as it makes
its way down from the north. The Yampa River drainage area
lies almost completely in Colorado and in junctions with
the Green, while the Green River is in Colorado. Table D-8
contains the available water quality data for the Green-
Yampa Basin.
The North Platte River rises in the north central portion
of Colorado and flows into Wyoming. The river drains .he
North Park area of Colorado where there is little popula-
tion. Table D-9 contains the existing water quality char-
acterization data for the North Platte River.'-
The Republican River Basin is located on the high plains
in the eastern central part of Colorado. Bonny Reservoir
regulates the flow of the river which has had an average
discharge of 24.8 cfs since 1951. The Republican River
joins the Kansas River "in Kansas. Table D-10' contains the
characterization data for this river.
The headwaters of the Rio Grande River are located-in the
south central part of Colorado and the flow yields 380,000
acre-feet or about 8 percent of'the total for the state.
The drainage area of the river consists of the San Luis
242
-------
5000
4000
3000
O
Q
N>
-P*
10
3000
2000
1000
2000
1000
MILES
10 20 30 4*0
FLOW(cfs) - *
PH
TDS(mg/l) - O
OOdng/l) - •
BOD(mg/|) .
Figure D-9. South Platte Main Stem Water Quality Characterization with
Proposed Sampling Point Locations.
-------
o
Q
NJ
13
12
11
10
9
8
7
6
5
4
2
1
0 10 20 30
I 1 1 1 .p
Miles to
Stream Standard
Figure D-10. Mean Dissolved Oxygen for the South Platte Main Stream with Standard
Deviation and Stream Standard Illustrated.
-------
KJ
JN
(Jl
13
12 J
11
-io -I
9
8 J
7
6
5
4
3
2
1
0 10 20 30
Miles
Q
8
1 •
ith Platte,
1
Littleton.
1
1
s
0
w
M
i
0)
u>
S4
a)
t4
_____
u
td
N
•H
(«
m
Cn
H
3
£!
to
0
iH
3
ID
3
o
w
Figure D-ll. Mean Biochemical Oxygen Demand for the South Platte Main Stream with
Standard Deviation Illustrated.
-------
to
-p-
4500
4000 -
3500
3000
2500 -
2000
1500 -
Figure D-12. Mean Total Dissolved Solids for the South Platte Main Stream with
Standard Deviation and Stream Standard Illustrated.
-------
Table D-5. South Platte Main Stem Stream Characterization Data.
Sta
No
20
21
22
23
24
25
Location
Julesburg
Balzac
Kersey
Henderson
Littleton
South Platte
DO (mg/1)
M S D S S
7.7 1.7 4.0
8.2 2.0 4.0
6.6 1.8 4.0
5.2 1.6 4.0
9.2 0.3 6.0
9.8 1.7 6.0
BOD (mg/1)
M S D
4.2 3.4
3.5 2.3
9.6 4.9
13.8 9.4
2.2 0.7
1.9 0.2
pH
M S D S S
8.1 0.4 6.0-9.0
8.1 0.5 6.0-9.0
7.9 0.4 6.0-9.0
7.7 0.4 6.0-9.0
7.7 0.2 6.5-8.5
7.6 0.2 6.5-8.5
TDS (mg/1)
M S D S S
1438 282 <500
1287 296 <500
1045 290 <500
584 221 <500
218 54 <500
126 47 <500
Flow (cf
M
458
359
130
328
217
474
-------
Table D-6. South Platte Tributary Stream Characterization Data.
Sta
NO
26S
27P
28P
29S
30S
Tributary
Cacha La Poudre
(above Ft. Collins)
Cacha La Poudre
(near Greeley)
Big Thompson
(near Mouth)
St. Vrain
(near Mouth)
St. Vrain
(at Weld-Boulder
County Line)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
M
7.72
1.78
8.29
69.31
282
6.22
26.99
7.93
1421.26.
96.0
6.38
17.21
8.00
1673.89
71.9
6.44
5.88
7.94
936.45
196
4.00
14.27
8.1
864.71
-
S D
1.84
.99
.54
34.10
1.73
22.38
.47
192.24
2.43
21.18
.46
451.18
1.45
3.26
.38
337.48
1.46
5.79
0.32
138.27
S S
6 ppm
6.5-8.5
<500 mg/1
3
5.0-9.0
Agric
4 ppm
6.0-9.0
<500 mg/1
4 ppm
6.0-9.0
<500 mg/1
4 ppm
6.0-9.0
<500 mg/1
248
-------
Table D-6. (Continued)
Sta
No
3 IP
32S
33P
34P
89P
Tributary
St. Vrain
(below Longmont)
Left Hand Creek
(near Niwot)
(Tributary to
St. Vrain River)
Boulder Creek
(at Boulder-Weld
County Line)
(Tributary to
St. Vrain River)
Clear Creek
(near Mouth)
Clear Creek
(at Wheat Ridge)
Parameter
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
6.62
11.51
8.28
673.54
-
7.80
1.57
7.90
632.00
-
7.91
7.38
8.23
358.07
90.6
(Boulder)
7.60
9.93
7.99
412.83
90.9
8.95
15.00
7.45
243.00
-
S D
1.17
4.79
.55
414.49
.40
2.30
2.79
.53
180.39
2.05
3.89
.45
179.46
-
-
-
S S
4 ppm
6.0-9.0
<500 mg/1
4 ppm
6.0-9.0
<500 mg/1
5 ppm
6.5-8.5
<500 mg/1
4 ppm
6.0-9.0
<500 mg/1
4 ppm
6.0-9.0
<500 mg/1
249
-------
Table 0-6. (Continued)
Sta
No
35P
36P
Tributary
Clear Creek
(above Golden)
Bear Creek
(at Jeff.-Arap.
County Line)
Parametei
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
6.71
2.40
7.49
160.95
227
7.52
2.15
7.82
203.67
34.9
S D
3.85
1.01
.34
70.64
3.04
.65
.32
119.27
S S
6 ppm
6.5-8.5
<500 mg/1
5 ppm
6.5-8.5
Agric
250
-------
Table D-7. Dolores River Stream Characterization Data.
Sta
No
84S
61S
4
85S
SOS
Tributary
San Miguel
(at Confl. with
Dolores)
(Tributary to
Dolores)
Dolores
(at Gateway)
Dolores
(at Bedrock)
Dolores
(near Dolores)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
7.18
5.44
7.94
655.40
339
(Naturita)
7.83
8.57
7.97
1759.96
-
7.10
.45
7.94
3328.83
-
7.85
.92
8.04
229.23
430
S D
1.80
1.47
.33
295.26
2.02
13.59
.38
1092.74
1-45
.21
.27
2705.83
1.77
.59
.56
67.23
S S
4 ppm
6.0-9.0
<500 mg/1
5 ppm
6.5-8.5
<500 mg/1
5 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
<500 mg/1
251
-------
Table D-8. Green-Yampa Rivers Stream Characterization Data.
Sta
No
88S
38S
39S
40S
41S
Tributary
Yampa
(above Oak Creek
Confl.)
Yampa
(at Milner)
Yampa
(near Maybell)
Yampa
(below Little
Snake River)
Little Snake
(above Lily)
(Tributary to
Yampa River)
Parameter
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
•
-------
Table D-9. North Platte River Stream Characterization Data.
Sta
No
37S
Tributary
North Platte
(below Cowdrey)
Parameter
DO
BOD
PH
TDS
Flow
M
6.82
1.17
7.89
190.52
429
S D
1.53
.73
.39
45.29
S S
6 ppm
6.5-8,5
<500 mg/1
253
-------
Table D-1D. Republican River Stream Characterization Data.
Sta
NO
70S
Tributary
Republican
(below Bonny Res . )
Parameter
DO
BOD
pH
TDS
Flow
M
7.69
2.11
8.29
317.84
24.8
S D
2.20
.78
.49
34.93
S S
5 ppm
6.5-8.5
Agric
254
-------
Valley, which is 7.9 percent of the land area of Colorado.
The valley, however, only contains 2 percent of Colorado's
total population. Nearly 52 percent of the area is federal
land serving as National Forests and Monuments. Agricul-
ture dominates the economy and is the major user of the Rio
Grande water. Table D-ll contains the characterization data
for the Rio Grande River.
The San Juan River is located in the southwestern part of
the state and its tributaries drain the southern part of
the San Juan Mountains. The San Juan leaves Colorado early
in its journey to the Colorado River, and consequently, its
tributaries must cross state boundaries. The San Juan barely
cuts the Colorado corner of the "Four Corners" area before
it leaves Colorado for good. Characterization data in con-
tained in Table D-12.
The White River is located in the west central area of
Colorado, and it meets the Green River in Utah. The popu-
lation in the White Basin is very small, but the oil shale
deposits might create problems as they are mined in this
and other areas of Colorado. Characterization data for the
White River is seen in Table D-13.
255
-------
Table D-ll. Rio Grande River Basin Stream Characterization Data,
Sta
No
18S
19S
Tributary
Rio Grande
(East of Manassa)
Rio Grande
(at Alamosa)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
7.53
1.66
7.92
250.41
603
7.25
2.20
7.83
148.72
254
S D
1.86
.46
.41
102.87
1.77
2.37
.13
37.42
S S
6 ppm
6.5-8.5
Agric
6 ppm
6.5-8.5
Agric
256
-------
Table D-12. San Juan River Stream Characterization Data.
Sta
No
63S
62S
64S
65S
66S
Tributary
San Juan River
(near State Line)
McElmo Creek
(west of Cortez)
Mancos River
(3 mi. north of
State Line)
La Plata River
(north of La Plata)
Animas River
(near Bondad)
Parameter
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
M
7.61
2.00
8.07
465.14
-
7.31
2.84
8.14
2982.04
43.1
7.78
1.49
8.19
1550.65
50.7
7.81
1.53
8.07
1594.61
45.3
8.25
1.48
8.22
326.09
878
S D
2.39
.97
0.42
157.63
1.74
1.57
.32j
1036.85
2.04
.85
.44
548.42
2.20
.74
.41
1108.73
2.03
.70
.46
103.09
,s s
5 ppm
6.5-8.5
-
6 ppm
6.5-8.5
-
6 ppm
6.5-8.5
<500 mg
257
-------
Table Er-12. (Continued)
Sta
No
81S
82S
67S
68S
69S
Tributary
Animas River
(above Durango)
Animas River
(near Silverton)
Los Pinos
(near La Boca)
San Juan
(above Navajo Res.)
Piedra River
(N.E. of Arboles
County)
Parameter
DO
BOD
pH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
DO
BOD
pH
TDS
Flow
DO
BOD
PH
TDS
Flow
M
8.30
1.83
7.76
184.08
846
6.66
.27
7.71
288.80
104
8.01
1.78
8.29
152.83
190
8.17
1.66
8.03
215.17
636
8.39
1.67
7.80
223.79
327
S D
1.86
3.46
.58
63.31
1.50
.06
0.40
146.18
2.05
.75
.44
44.90
1.86
.99
.35
76.26
1.80
.77
1.13
87.80
S S
6 ppm
6.5-8.5
<500 mg/1
3 ppm
5.0-9.0
-
6 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
<500 mg/1
6 ppm
6.5-8.5
-
253
-------
Table D-B. White River Stream Characterization Data.
Sta
NO
43S
44S
Tributary
White River
(at Meeker)
White River
(at Rangely)
Parameter
DO
BOD
PH
TDS
Flow
DO
BOD
PH
TDS
Flow
M
8.92
1.66
8.28
353.20
621
7.64
1.79
8.25
466.60
S D
1.83
.93
.54
99.32
1.89
.73
.45
145.67
S S
5 ppm
6.5-8.5
<500 mg/1
5 ppm
6.5-8.5
Agric
«U.S. GOVERNMENT PRINTING OFFICE:1973 514-156/366 1-3
259
-------
1
V
b
Accession Number
V
2
Subject Field & Group
07B
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
e, « n vj i
10
Title
Data Acquisition Systems in Water Quality Management
Authors)
Robert C. Ward
1 A Project Designation
16090FUO
21 Note
22
Citation
EPA Research Series EPA-R5-73-014, May 1973
23
Descriptors (Starred First)
Data acquisition, sampling, information retrieval, network design, cost comparisons,
monitoring operations.
25
Identifiers (Starred First)
Water Quality management, data collection.
27
Abstract
The role of routine water quality surveillance was investigated, including a
delineation of the objectives a state water quality program based upon the state and
federal laws. Seven specific objectives are listed under the two general objectives
of prevention and abatement: planning, research, aid programs, technical assistance,
regulation, enforcement, and data collection, processing, and dissemination. Each
objective was broken down into the general activities required for its accomplishment
and the data needed for each activity was identified.
A survey of systems for grab sampling, automatic monitoring, and remote sensing
was performed, each data acquisition technique being analyzed for capabilities,
reliability, and cost.
A procedure was developed for designing a state water quality surveillance
program responsive to objectives. The design procedure has two major aspects: (1)
determination of the agency's control strategy, and (2) characterization of streams
according to use and quality conditions. The optimum grab sampling network is then
designed and possibilities of substititing automatic monitoring and remote sensing at
various points in the network are explored.
The design procedure was applied to the State of Colorado and a water quality
monitoring system was developed for the Water Pollution Control Division of the Colorado
_Department of Health. Financial and manpower constraints were considered in the design.
Abstractor
institution
WR:I02 (REV. JULY 1969)
VVRSI C
SEND. WITH COPY OF DOCUMENT, TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON, D.C. 20240
* GPO: 1 970-389=930
------- |