You are on page 1of 10

A WATER QUALITY INDEX-

CRASHING THE PSYCHOLOGICAL BARRIER


Robert M. Brown
President, National Sanitation Foundation
Ann Arbor, Michigan 48106

Nina I. McClelland
Program Director, National Sanitation Foundation
Ann Arbor, Michigan 48106

Rolf A. Deininger
Associate Professor, School of Public Health
University of Michigan
Ann Arbor, Michigan 48104

Michael F. O'Connor
Research Associate, Public Systems Research Institute
University of Southern Calzfornia
Los Angeles, California 90007

COMMUNICATING EFFECTIVELY ABOUT THE QUALITY OF WATER


Use of this somewhat bombastic title stems from our desire to draw critical
attention to a process of decision making that in practice has been more of an
art than a science. At issue is the long standing need for a uniform method for
measuring water quality - a "yardstick" with units that are simple, stable,
consistent, and reproducible - and an unambiguous method for communicating
this information to everyone concerned.
Over the years, and even today, a decision regarding the "quality" of a lake
or stream is made as a judgmental determination by the "water expert." This
expert examines values obtained in the field or laboratory and, as a result of
these values, determines the suitability of a body of water for an intended use.
While his decision reflects a process of judgmental weighting and integration of
multiple parameter values, the end result does not lend itself to effective
communication.
In ancient days, before the unit of measure, the foot, was devised, the plea of
the community was for a consistent, uniform, dependable measuring stick so

173

W. A. Thomas (ed.), Indicators of Environmental Quality


© Plenum Press, New York 1972
174 BROWN, McCLELLAND, DEININGER, AND O'CONNOR

that when one bought a piece of cloth, a beam of wood, or a plot of land, he
could know with assurance what he was getting in return for his investment.
We submit that today the plea of the community is for a unit of measure by
which water quality can be judged as a commodity.
As requirements for water pollution abatement are being pursued and as
stated objectives for water quality attainment are being judged, there is urgent
need for those involved in decision making to be knowledgeably aware of the
quality status, and changes in such status, of a given surface water. Just who are
these decision makers? The general citizen - the taxpayer, persons elected to
political office at all levels of government, administrative and regulatory
authorities at all levels of government, leaders in business and industry, and
those concerned groups who prod the public and private conscience.
All of these decision makers need objective information on quality status
upon which to make value judgments. Popularly used descriptive terms like
"pristine," "clean," "dirty," "cloudy," "scummy," "stinky," "foul,"
"noxious," etc. are vague and uninformative at a time when pursuit of water
quality objectives calls for clear and precise understanding. "Dollars spent" and
"numbers of wastewater treatment plants constructed" are likewise inadequate
and improper for reporting benefits, or lack of benefits, from tax monies used
for pollution abatement programs. We can note with satisfaction that back in
1889 Lord Kelvin said, "When you can measure what you are speaking about,
and express it in numbers, you know something about it; but when you cannot
measure it, when you cannot express it in numbers, your knowledge is of meager
and unsatisfactory kind."
Thus, determination of water quality involves a process of value judgment.
There are several score parameters that may enter into assessment of water
quality. The quality determination is made by a knowledgeable person from an
array of figures - i.e., parameter values - which he evaluates.
One may ask, then, "Can such individual professional judgment processes be
factored and recombined into an empirical expression of water quality?" We
believe the answer to this question is an unqualified "yes!" However,
considerable reluctance on the part of many profeSSionals to accept the concept
of a general use index is clearly recognized. We respond to this attitude with a
challenge to the total professional community to help provide for public
understanding and critical scrutiny an objective appraisal of stream quality and
progress in attaining it.

WQI - CONCEPT AND OBJECTIVES

A mathematical expression generally described as a Water Quality Index


(WQI) was developed at the National Sanitation Foundation during 1970. The
composite influence of significant physical and chemical parameters is reflected
in the index. Application of the index offers a defined, understandable unit of
measure which responds to change in quality of water. The index method, by
A WATER QUALITY INDEX 175

virtue of its function to combine the results of change in parameter levels,


reflects a net quality value which can be observed and meaningfully interpreted.
The real significance of the net changes in quality is not in the magnitude of
change, per se, nor in the size of the numerical value, but in the fact that the
WQI does vary as a result of significant changes in the parameters entering into
its determination.
The index, then, provides an instrument or tool that appropriately
consolidates, and presents as a single number, the values of multiple parameters
selected to enter into the index formulation. With an understanding of the
responsiveness of the tool to changes in parameter values and its ability thereby
to reflect changes in water quality, the WQI can be viewed as a major resource in
communication.
Re-emphasizing, the basic objectives of WQI are two-fold:

1. Making available a tool for dependably treating water quality parameter data
and presenting it as a single numerical term; and
2. Promoting utilization of a process for effectively communicating water
quality conditions to all concerned.

METHODOLOGY FOR DEVELOPMENT OF WQI


In formulating the WQI, a modification of the opinion research technique
known as the DELPHI process was used. A panel of 74 persons with expertise in
water quality management responded to a series of four questionnaires,
described in detail in a previous publication. 1 Parameter selection, quality
curves, and the weighted significance of each parameter in the index were
determined by the experts. The resulting index formulation is an additive model,
expressed as follows:

n
WQI =.l; w·q·
1=1 1 1

where: WQI is a number between 0 and 100; qi' the quality of the ith parameter,
a number between 0 and 100; Wi' the unit weight of the ith parameter, a number
between 0 and 1; and n, the number of parameters.
In responding to the questionnaires, the experts were asked to make
judgments with respect to overall quality. As a result, eleven parameters were
proposed for the index:

dissolved oxygen temperature


fecal coliforms turbidity
pH total solids
5-day BOD toxic elements
nitrate pesticides
phosphate
176 BROWN, McCLELLAND, DEININGER, AND O'CONNOR

In the WQI equation, however, n = 9. Toxic elements and pesticides are


classes of substances rather than single parameters and, therefore, cannot be
treated as simple additive terms. In these two instances, there will be maximum
limits established for each substance within the two classes. When the level of
any toxic element or pesticide exceeds its maximum limit, the index will
automatically go to zero.
In developing quality curves for the WQI, respondents were asked to assign
values for the variation in level of water quality produced by different strengths
of each of the nine selected parameters. This was accomplished by utilizing a
series of graphs. Levels of "water quality" from 0 to 100 were shown on the
ordinate; various levels (or strengths) of the particular parameter were arranged
on the abscissa. Resulting curves for dissolved oxygen and turbidity are shown in
Fig. 1. In each of the graphs, the arithmetic average of all responses is
represented by a solid line; 80% confidence limits are shown by dotted lines.

80
~
I-
::i
ct
::>
o
60
,
ffi 40 /
~
~
20

o ~~£-~--~--~--~--~--~
o 20 40 60 80 100 120 140
DISSOLVED OXYGEN SATURATION ("0)

100
---------- -- --.. _-
~
I- -... ............
::i
ct 60
::>
0
Q:
ILl
40
~
~
20

w ~ 40 ~ W ro ~ ~ ~
TURBIDITY (UNITS)
Fig. 1. Arithmetic means (solid lines) and 80% confidence limits (dashed lines) for two
water quality parameters.
A WATER QUALITY INDEX 177

Convergence was good for some parameters, as shown by the narrow confidence
band for dissolved oxygen. However, it varied appreciably for other parameters,
as indicated by the wide band for turbidity.

STUDY OF GENERAL VERSUS USE INDICES

In exchanging information with our experts, a number of them continued to


be identified with the "use concept hangup." These persons found it extremely
difficult, if not impossible, to consider quality of a body of water except in
relation to its intended use. Scaling of quality for general understanding is
therefore not regarded as relevant by them. Since such single use concepts may
not be summarily rejected or ignored, considerable attention has been given to
exploring the meaningfulness of single use indices as an alternative to the general
purpose index.
A number of disadvantages are perceived with use-related indices: para-
meters, weights, and scales will vary for each of the large number of uses; much
more data will be required to support the additional parameters to be measured;
greater expense will be incurred; and communication processes with the public
will become commensurately more complex.
Despite these drawbacks - effort, expense, and loss of simplicity - the
relative merit of use-related indices versus an index of overall quality was
addressed by O'Connor in a doctoral dissertation. 2 He interviewed in person
representative members of the panel of experts originally contacted by Brown et
al. l
One of O'Connor's basic purposes was to probe the methods by which water
experts selected parameters to enter into quality determinations, assigned
weights (drew curves) for judging variation in the individual parameter values,
and rated the parameters relatively in terms of importance. He sought to
ascertain whether, in expert decision -making, failure to assign and use parameter
weights may result from the complexity of sheer numbers of items to be judged.
Might some form of mathematical or statistical appraisal of multiple values be
more a.dvantageous than utilizing overall judgments of a large number of
dimensions?
O'Connor found that a truly crucial factor lies in identification of the critical
parameters which would enter into an index determination. Selection of
parameters was ascertained to be of greater importance than determining the
shape of the quality curves for designated parameters. Accepting the critical
nature of parameter selection, the following factors were identified as having
importance:
1. Solid consensus among water evaluation experts as to those parameters that
generally and consistently relate to determination of water quality, i.e., those
that would be regularly required in routine examination of waters in streams
and lakes;
2. Parameters for which standard methods are available for routine field
sampling and for field or laboratory examination of samples;
178 BROWN, McCLELLAND, DEININGER, AND O'CONNOR

3. Parameters that will consistently be influenced by introduction of water


pollution factors or by removal of such factors;
4. Parameters that are independently variable, that are not redundant indicators
of the same pollution factors, and that do not parallel the indications given
by another parameter;
5. Parameters that, from the above four standpoints and the additional
considerations of expense and practicable management, will constitute the
ingredients for the least common denominator of water quality for all water
requirements.

Using procedures developed in the study of multidimensional utility analysis,


O'Connor developed two water quality indices: (1) for a surface body of raw
water to be used to sustain a jish and wildlife (FAWL) population and (2) for a
water source to be treated and used as a public water supply (PWS). The goal of
this study was to develop valid indices for two quality-demanding uses of water
and to determine the significance of the difference in numbers assigned samples
of water with their application. By correlational analysis, the values assigned
with F AWL and PWS were compared with those for the overall WQI. Results are
shown in Table 1. It is apparent from Table 1 that the WQI correlated better
with FAWL and PWS than the two use-related indices did with each other. Thus,
O'Connor concluded, "It appears that the WQI is a sort of mean approximation
to the PWS and FAWL indices."

Table 1. Correlations among three water quality indices

Index Correlation Correlation Sample


identification with F AWL with PWS set no.a

WQI .879 .684 1


FAWL 1.000 .670 1
WQI .744 .854 2
FAWL 1.000 .667 2
WQI .792 .652 3
FAWL 1.000 .501 3
WQI .863 .860 4
FAWL 1.000 .733 4

aSets 1 and 2 are from hypothetical data; sets 3 and 4


are from real data.

It is interesting to note that four parameters are common to each of the


indices: dissolved oxygen, pH, nitrates, and turbidity. Five-day biochemical
oxygen demand (BOD) and fecal coliforms in the WQI are not included in
FAWL; conversely, phenols and ammonia in FAWL are not used in the WQI.
However, n = 9 in both WQI and FAWL, and O'Connor concluded that "the
WQI can be fairly well described as a FAWL index with fecal coli forms added."
A WATER QUALITY INDEX 179

Table 2. Index parameters for three water quality indices

WQI FAWL PWS


Dissolved oxygen Dissolved oxygen Dissolved oxygen
Nitrate Nitrate Nitrate
Turbidity Turbidity Turbidity
pH pH pH
Temperature Temperature Fluoride
Phosphate Phosphate Hardness
Fecal coliform s Ammonia Fecal coliforms
BOD (5 day) Phenols Phenols
Total solids Dissolved solids Dissolved solids
Chloride
Alkalinity
Color
Sulfate

Thirteen parameters are included in the PWS index, the higher number reflecting
economic and aesthetic considerations which are not pertinent to WQI or F AWL
(Table 2).
Deininger and Maciunas 3 extended the study of Brown et al./ further
querying members of the WQI panel, in developing a specific water quality index
for public water supplies. Query and analysis of responses followed the processes
of the earlier work. Similarities of response were noted with those defined by
O'Connor. 2
Analysis leading to index computation (for public water supplies) was made
on data derived from the parameters selected for the WQI (in the report of
Brown et al.) and on ll-parameter and 13-parameter groups derived from the
further querying of the expert panel. Index formulation was accomplished by
use of the arithmetic equation of Brown et al. l as well as by use of a specially
devised geometric equation. Experience with the geometrically derived index
provided additional perspective with regard to the ranging of values on the index
scale.
It is interesting to note the conclusions of Deininger and Maciunas 3 after
their explorations had been completed:

"1. It is possible to develop a water quality index for surface waters designated
for the specific use of public water supply ...
2. The comparisons show that this index developed with a specific use-
orientation does not seem to rate water quality levels in a manner markedly
different from the rating made by a general, non-specific use-oriented index.
Thus, it is possible to argue that water of a certain quality retains that
relative quality rating regardless of the use for which it is being considered.
Hence, waters of different streams can certainly be compared with regard to
changes in quality levels, using one uniformly applied rating scheme.
3. Instead of developing a number of indices for the many water uses, it
appears to be more meaningful to further develop and refme a sensitive and
general water quality index."
180 BROWN, McCLELLAND, DEININGER, AND O'CONNOR

This further study is encouraging and strongly supportive, both with respect
to the responsiveness of an appropriately devised index to changes in parameter
values and to the sensitivity of a general water quality index in projecting
variations in quality irrespective of use orientation.

FIELD EV ALUA nON

Further evaluation of the effectiveness of the WQI is currently underway in a


field sampling program, with the National Sanitation Foundation serving as the
coordinating agency. The Michigan Water Resources Commission, Grand River
(Michigan) Watershed Council, Maryland Department of Water Resources,
Pennsylvania Department of Environmental Resources, California Department of
Water Resources, Tennessee Valley Authority, U. S. Army Environmental
Hygiene Agency (Edgewood Arsenal), Larimer County (Colorado) Health
Department, and the University of Notre Dame Civil Engineering Department
are participating in this program. Use problems that may be attributable to

100
-, \ LEGEND:
90
\, - - ARITHMETIC MEAN
---80% CONFIDENCE
LIMITS
80
\\

70
\,
\
,
>- 60
!::

,
...J \

,
<t
:;)
0 50 \
0::
w
~ 40
• ...
~
"- " ..........
I I
"-..
\
30
I

,I I

\
20

I \

\
10 I

0
-5 0 5 10 15 20
DEPARTURE FROM EQUILIBRIUM TEMPERATURE (OC)

Fig. 2. Arithmetic mean (solid line) and 80% confidence limits (dashed lines) of change
in water quality with departure from equilibrium temperature ee).
A WATER QUALITY INDEX 181

sampling, choice of parameters, and application of the additive model are being
evaluated.
This study began in May and is expected to proceed for a time sufficiently
long for effects of seasonal variation to be apparent (6 to 12 months). The initial
goal is to evaluate the WQI with no modification of its current form, i.e, "total
coliform density" cannot be substituted for "fecal coliforms," etc. Even before
the study began, the subject of temperature as a parameter in the index had to
be addressed. In order to develop a meaningful quality curve, the experts were
asked to consider the effect that degrees of departure from equilibrium would
have on water quality. Thus, the average curve developed for this parameter
ranged from -5 to +15°C departure from equilibrium (Fig. 2). This somewhat
idealistic approach was quite acceptable for developing the WQI, but the need
for defining "equilibrium" was immediate when faced with a field sampling
program. By general consensus of the participants, it was agreed that two
temperatures would be recorded for every field sampling station - one at the
station for measuring degrees of departure and a second, at some point upstream
where any effects of heated or cooled discharge were known to be absent, to be
called "equilibrium." No problems have been known to result from using this
approach.

FURTHER NEEDS
Methods for presenting the finalized WQI to the professional community and
the public are currently being reviewed. There is need for innovative educational
techniques to help communicate water quality information to the public. Use of
a color-coded spectrum for visually depicting use-related applications of the WQI
could materially assist in achieVing public understanding.

SUMMARY
Determination of water quality is generally regarded as a decision-making
process requiring expert judgment. Today this process varies, as different experts
apply different weights to various parameters upon which individual determina-
tions of quality are made.
Extended exploration has been made of the need for, and practicability of, a
general-use water-quality index that would provide a uniform method for (1)
reflecting the quality of water and (2) communicating quality status, and
changes in status, to the public. While there is marked reluctance or
unwillingness on the part of some water experts to accept a procedure that is not
specific use-oriented, significant contra-indications to the feasibility and objec-
tivity of the proposed general WQI have not been encountered during trial uses
under field conditions. It would, therefore, seem that obstacles or barriers to the
acceptance and use of WQI may be considered psychological rather than
technical. We propose to crash that barrier.
182 BROWN, McCLELLAND, DEININGER, AND O'CONNOR

REFERENCES

1. Brown, R. M., McClelland, N. I., Deininger, R. A., and Tozer, R. G. A Water


Quality Index - Do We Dare?, Water and Sewage Works (Oct. 1970); also
presented at National Symposium on Data and Instrumentation for Water
Quality Management, July 1970.
2. O'Connor, M. F. The Application of Multi-Attribute Scaling Procedures to
the Development of Indices of Water Quality. Ph.D. dissertation, University
of Michigan, 1971.
3. Deininger, R. A., and Maciunas, J. J. A Water Quality Index for Public Water
Supplies. Report of a research study, University of Michigan, 1971.

You might also like