EPA 815-R-00-024
GEOMETRIES AND CHARACTERISTICS OF
PUBLIC WATER SYSTEMS
REGION VI LIBRARY
U. S. eNVIRONfCNTAL PROTECTION
AGENCY
1445 ROSS AVENUE
DALLAS, TEXAS 7520?
December 2000
PREPARED FOR:
U.S. Environmental Protection Agency
Office of Ground Water and Drinking Water
1200 Pennsylvania Ave. NW, MS 4607
Washington, DC 20460
PREPARED BY:
Science Applications International Corporation (SAIC)
1710 Goodridge Drive
McLean, Virginia 22102
EPA Contract No. 68-C6-0059
SAIC Work Assignment No. 1-24
SAIC Project No. 01-0833-08-3562-000
AND
The Cadmus Group
1901 North Fort Myer Drive. Suite 900
Arlington,VA 22209
EPA Contract No. 68-C-99-206
Cadmus Group Work Assigment No 1-31
-------
Table of Contents
List of Exhibits iii
Glossary iv
List of Abbreviations vi
s
1: Introduction 1-1
2: Description of the Public Water Supply Universe 2-1
2.1 Existing Model 2-1
2.2 Rationale for Revising the Existing Model 2-2
2.3 Characterization of the Water Supply Universe 2-4
2.4 Population Served 2-4
2.5 Source Water 2-11
3: Community Water System Survey of 1995 3-1
3.1 Survey Overview 3-2
3.2 Statistical Design 3-2
3.3 Peer Review and Quality Assurance 3-4
4: Analysis of Population and Flow Relationships 4-1
V 4.1 Population and Flow as Critical Variables 4-1
">; 4.2 Analysis of Average Daily Flow versus Population 4-2
\ 4.2.1 Data Screening for Regression Analysis 4-2
r 4.2.2 Distribution Analysis of Regression Variables 4-3
1 4.2.3 Average Daily Flow and Population Regression 4-4
o^ 4.2.4 Adjustment of Sampling Weights for Item-Level Nonresponse 4-6
^ 4.3 Regression of Design Flow Versus Population 4-6
4,3.1 Data Screening 4-7
4.3.2 Maximum Daily Treatment Design Capacity and Population Regression .... 4-7
0s* 4.4 Regression Analysis of Different Categories of CWSs 4-9
^ 4.4.1 Regression Analysis for Different Strata 4-10
fij 4.4.2 Statistical Tests To Determine Differences in Regression Lines 4-11
r~ 4.5 Evaluations of Design-to-Average Flow Ratios 4-11
<~. 4.5.1 Design Flow Modification for Surface and Ground Water Systems 4-12
$vj 4.5.2 Design Flow Modification for Purchased Water Systems 4-13
\K 4.6 Summary of Population and Flow Relationships 4-14
^
^ 5: Analysis of Entry Point Configurations 5-1
^ 5.1 Data Cleaning 5-3
S^> 5.2 Number of Entry Points 5-3
5.3 Distribution of Flow Among Entry Points 5-5
5.4 Spatial Distribution of Entry Points 5-7
5.5 Well Depths 5-8
-------
6: Analysis of Treatment-in-Place 6-1
6.1 Rationale for Treatment-in-Place Analysis 6-1
6.2 Identification of Core Data Set 6-3
6.3 Summary of Results 6-3
7: Non-Community Water Systems 7-1
7.1 Overview of Non-Community Water System Population 7-1
7.2 Service Area Classification and Population Served 7-3
7.2.1 Approach 7-4
7.2.2 Characterization of General Service Area Classifications 7-4
7.2.3 Results 7-7
7.3 Average and Design Flows 7-11
7.4 Transient Versus Non-Transient Systems 7-14
Appendix A: Population Served for Public Water Systems A-1
Appendix B: Methodology for Quantifying the Cost Bias Caused by
Retail Population Categorization of Systems B-2
Appendix C: 1995 Community Water Systems Survey Questionnaire C-l
Appendix D: Weighted Regression Methodology D-l
Appendix E: Statistical Testing of Regressions for Different Categories of CWS E-l
Appendix F: Bootstrapping Results for Ground Water System Entry Points F-l
Appendix G: Distribution of Flow Among Entry Points G-l
Appendix H: Non-Community Water Systems Serving Greater than 10,000 People H-l
-------
List of Exhibits
2: Description of the Public Water Supply Universe 2-1
Exhibit 2.1 40 CFR §141.2 Definitions 2-1
Exhibit 2.2 Existing Model of Community Water System Characteristics 2-2
Exhibit 2.3 Community Water Systems (Number of Systems) 2-5
Exhibit 2.4 Nontransient Non-Community Water Systems (Number of Systems) 2-6
Exhibit 2.5 Transient Non-Community Water Systems (Number of Systems) 2-7
Exhibit 2.6 Representation of System Relationships '. 2-8
Exhibit 2.7 Comparison of System Categorization Schemes for Surface Water Systems 2-10
Exhibit 2.8 Comparison of System Categorization Schemes for Ground Water Systems 2-10
Exhibit 2.9 Analysis of Mixed Systems in SDWIS Non-Purchased Surface Water Systems 2-11
4: Analysis of Population and Flow Relationships 4-1
Exhibit 4.1. Population-Served to Number of Service Connections Ratios 4-3
Exhibit 4.2. Regression for Average Daily Flow 4-4
Exhibit 4.3. Regression for Average Daily Flow (Extreme Values Removed) 4-5
Exhibit 4.4. Regression Line for Design Flow (Extreme Values Removed) 4-8
Exhibit 4.5. Daily Production Flow Regression Equations for Subcategories of CWS 4-10
Exhibit 4.6. Design Flow Regression Equations for Subcategories of CWS 4-10
Exhibit 4.7. Comparison of Design-to-Average Flow Ratios 4-12
Exhibit 4.8. Recommended Approach for Estimating Design Flow for Surface and Ground
Water Systems 4-13
Exhibit 4.9. Estimated Design Flow for Purchased Water Systems 4-14
Exhibit 4.10 Population and Flow Relationships for Ground Water 4-15
Exhibit 4.11 Population and Flow Relationships for Surface Water 4-15
Exhibit 4.12 Population and Flow Relationships of Purchased Water Systems
Fed by Ground Water 4-16
Exhibit 4.13 Population and Flow Relationships for Purchased Water Systems
Fed by Ground Water 4-16
5: Analysis of Entry Point Configurations 5-1
Exhibit 5.1. Conceptual Diagram of Entry Point Configurations 5-2
6: Analysis of Treatment-in-Place 6-1
Exhibit 6.1. C WSS Water Treatment Codes (Question 18) 6-1
Exhibit 6.2. Treatment Code Groups 6-2
7: Non-Community Water Systems 7-1
Exhibit 7.1. Non-Community Water Systems by Water Source 7-1
Exhibit 7.2. Non-Community Water Systems by Population Range 7-2
Exhibit 7.3. SDWIS Service Area Classifications 7-3
Exhibit 7.4. Approach to NCWS Service Area Categorization 7-4
Exhibit 7.5. Random Sampling of Non-Community Water Systems with
Genera) Service Areas 7-5
Exhibit 7.6 Non-community Water System Service Areas 7-6
Exhibit 7.7 Non-community Water System Population Served by
Service Area Type 7-8
Exhibit 7.8. Explanation of Population Served for Non-Community Water Systems 7-10
Exhibit 7.9. Average Water Use Assumptions for Non-Community Water Systems
(gallons per person per day) 7-12
Exhibit 7.10 Non-community Water System Flow Rates by Service Area Type 7-13
Exhibit 7.11. Alternative Classification of Transient versus Non-Transient Non-Community
Water Systems 7-16
ill
-------
Glossary
Average Daily Flow - daily volume of water produced within a system or by a treatment plant, averaged
over 365 days (also called average daily production however).
Community Water System (CWS) - a public water system that has at least 15 service connections used
by year-round residents or regularly serves at least 25 year-round residents.
Design Flow - the design capacity, or the maximum amount of water per day that can be treated at the
treatment plant or within a system (also called design capacity)
Entry Point - Locations where finished water enters a distribution system or is sold.
Finished Water - water that has passed through a water treatment plant; all the treatment processes are
completed or "finished". The water is ready to be delivered to consumers.
Ground Water - water found below the surface of the land, usually in porous rock formations, without
significant occurrence of insects, microbes, or pathogens and without rapid shifts in water quality
parameters. Ground water is the source of water found in wells and springs.
Ground Water System - water system that gets a majority of its water from ground water.
Ground Water Under the Direct Influence (GWUDI) of Surface Water - any water beneath the
surface of the ground with significant occurrence of insects or other microorganisms, algae, or
large-diameter pathogens, or significant and relatively rapid shifts in water characteristics such
as turbidity, temperature, conductivity, or pH that closely correlate to climatological or surface
water conditions. Direct influence must be determined for individual sources in accordance with
criteria established by the State.
GWUDI System - water system that gets a majority of its water from GWUDI.
Maximum Daily Flow - the highest flow over one day measured within one year in a system or plant.
Non-Purchased Water System - a system that treats its own water for delivery to the public and does
not purchase water from other systems.
Non-Transient Non-Community Water System (NTNCWS) - a public water system that regularly
serves at least 25 of the same people more than 6 months per year that is also not a community
water system
Public Water System (PWS) - a system for the provision to the public of piped water for human
consumption if such a system has at least 15 service connections or regularly serves an average
of at least 25 individuals per day at least 60 days out of the year.
Purchased Water System - a system that purchases any amount of drinking water from another system
for distribution to its own customers.
Retail Population of a System- population receiving water treated and sold directly to them by that
system.
IV
-------
Source Water - The water used as the source for the water treatment plant's operations.
Surface Water - water that is open to the atmosphere and subject to surface runoff.
Surface Water System - water system that gets a majority of its water from surface water (CWSS
definition and used throughout the report).
Transient Non-Community Water System (TWS) - a non-community water system that does not
regularly serve at least 25 of the same persons more than 6 months per year.
Wholesale Population of a System - the population receiving water treated by a separate water system.
-------
List of Abbreviations
AWWARF - American Water Works Association Research Foundation
BAT - Best Available Technology
CWS - Community Water System
CWSS - Community Water Systems Survey
FRDS - Federal Reporting Data System (now known as SDWIS)
GIS - Geographic Information System
gpcd - gallons per capita per day
gpd - gallons per day
GWUDI - Ground Water Under the Direct Influence of surface water
MWDSC - Metropolitan Water District of Southern California
NCWS - Non-Community Water System
NTNCWS - Non-Transient Non-Community Water System
OGWDW - Office of Ground Water and Drinking Water
PWS - Public Water System
RIA - Regulatory Impact Analysis
SBREFA - Small Business Regulatory Fairness Act
SDWA - Safe Drinking Water Act
SDWIS - Safe Drinking Water Information System
TDP - Technology Design Panel
TWS - Transient Non-Community Water System
UMRA - Unfunded Mandates Reform Act
USGS - U.S. Geological Survey (dept)
VI
-------
1: Introduction
The U.S. Environmental Protection Agency (EPA) historically has analyzed the costs to public water
systems (PWSs) and their customers that stem from regulations pursuant to the Safe Drinking Water Act
(SDWA). Various regulatory reforms, particularly during the last five years, have also placed
considerable emphasis on evaluating the benefits and costs of regulation1. Consistent with this trend,
Section 103 of the SDWA Amendments of 1996 (codified in Section 1412(b) of the Act) mandates that
EPA perform benefit-cost analyses as part of the development process for all new drinking water
regulations.
In performing a Regulatory Impact Analysis (RIA) for any drinking water rule under development, EPA
must be able to highlight the impacts (i.e., benefits and costs) for typical affected parties, while also
capturing the more extreme situations. For example, analysis of new drinking water regulations could
address situations that range from campsites in remote national forests to the largest metropolitan areas in
the country, such as New York City and Los Angeles. Characterization of data for an RIA must include
information on the number of water systems of various types and sizes, average population served, and
average and maximum flows in a system. EPA uses these data in various ways to estimate national
benefits and costs. For example, costs of a proposed regulation are often estimated by establishing the
number of systems of a particular type affected by the rule (usually some proportion of the total systems)
and multiplying them by unit costs for implementing additional treatment technologies. To facilitate
benefit-cost analyses, system information must be organized into a manageable framework that should
inform rather than complicate, while provided adequate precision and accuracy for the necessary
evaluations
The purpose of this report is to present a basic set of significant PWS characteristics for use in future
Office of Ground Water and Drinking Water (OGWDW) rulemakings. As additional data are gathered
and analyzed, the characteristics established in this report can be revised. By describing and encouraging
discussion of the underlying data and models used to develop the characteristics, EPA hopes to facilitate
acceptance and use of a common set of inputs. A clearer consensus about the basic characteristics of
PWSs will help EPA and stakeholders focus attention on RIA results and decision making, rather than on
the basic characteristics of PWSs.2
The remainder of this document is organized as follows:
Chapter 2: An overview of the universe of PWSs in the United States and a discussion of
overarching issues in developing a framework for describing this universe, including
characterizing these systems in terms of population served and water source.
Chapter 3: A description of the source and quality of data used to analyze and develop baseline
profiles of PWSs.
'Regulatory benefit-cost analyses on various sectors of the economy are addressed by Executive Order
12866, the Unfunded Mandates Reform Act (UMRA), the Small Business Regulatory Fairness Act (SBREFA), and
the Paperwork Reduction Act (PRA).
" Contaminant occurrence and technology costs also are important determinants of the accuracy and
precision of an RIA. These issues are being addressed as part of other EPA initiatives to improve RIA data and
tools.
Geometries and Characteristics of 1-1 December 2000
Public Water Systems
-------
Chapter 4: EPA's model of community water system (CWS) flow rates, which uses regression
equations based on design and average flows and population served.
Chapter 5: EPA's analysis of entry point geometries for CWSs, including distribution of flow
among entry points.
Chapter 6: EPA's estimates of numbers and types of drinking water treatment technologies currently
in place for CWSs.
Chapter 7: EPA's estimates of the numbers and types of non-community public water systems
(NCWs) and typical system sizes (that is, flows and populations served).
Chapter 8: A list of references used in this document.
Geometries and Characteristics of 1-2 December 2000
Public Water Systems
-------
2: Description of the Public Water Supply Universe
The universe of PWSs comprises both community and non-community systems (see Exhibit 2.1) with a
wide range of characteristics in terms of water flow rates, size and composition of service population,
source water types, treatment configurations, types of ownership, etc. As mentioned in Chapter 1,
regulatory analysis requires a manageable framework for describing this universe. This chapter describes
EPA's existing model as well as the rationale for revising this model. An overview of the PWS universe
is then presented, including an inventory of the number of systems by population category, water source,
and ownership type. Finally, overarching issues are considered, including issues related to the
measurement of population served and source water type.
Exhibit 2.1
40 CFR §141.2 Definitions
Public water system (or PWS) means a system for the provision to the public of piped water for human
consumption, if such system has at least 15 service connections or regularly serves an average of at least 25
individuals daily at least 60 days out of the year. Such term includes (1) any collection, treatment, storage, and
distribution facilities under control of the operator of such system and used primarily in connection with such
system, and (2) any collection or pretreatment storage facilities not under such control which are used primarily
in connection with such system. A PWS is either a "community water system" or a "noncommunity water
system."
Community water system (or CWS) means a public water system that serves at least 15 service connections used
by year-round residents or regularly serves at least 25 year-round residents.
Non-community water system (or NCWS) means a public water system that is not a community water system.
Non-transient non-community water system (or NTNCWS) means a public water system that is not a
community water system and that regularly serves at least 25 of the same persons more than 6 months per year.
Transient non-community water system (or TWS) means a non-community water system that does not regularly
serve at least 25 of the same persons more than 6 months per year.
2.1 Existing Model
The existing model stratifies the drinking water universe into 12 population size categories. For each
population size category, the median population defines a typical drinking water system's size. Average
daily flow (volume of water produced per day) and design flow (design capacity), expressed in million
gallons per day for the corresponding population, are also included in the model. Exhibit 2.2 depicts the
existing model for community water systems and its 12 strata, median populations, and flows.
Geometries and Characteristics of 2-1 December 2000
Public Water Systems
-------
Exhibit 2.2
Existing Model of Community Water System Characteristics
Population Category
25 to 100
101 to 500
501 to 1,000
1,001 to 3,300
3,301 to 10,000
10,001 to 25,000
25,001 to 50,000
50,001 to 75,000
75,001 to 100,000
100,001 to 500,000
500,001 to 1,000,000
Greater than 1,000,000
Median Population
57
225
750
1,910
5,500
15,500
35,000
60,000
88,100
175,000
730,000
1,550,000
Average Flow (million
gallons per day)
0.0056
0.024
0.089
0.23
0.70
2.7
5.0
8.8
13.0
27.0
120.0
270.0
Design Flow (million :
gallons per day) •
0.024
0.087
0.27
0.65
1.8
4.8
11.0
18.0
26.0
51.0
210.0
430.0
The information in Exhibit 2.2 was used by regulatory analysts to estimate the cost for a typical system
for each of the 12 population size categories. Capital and operation and maintenance costs for the typical
system were based on design and average flows, respectively. Costs for the typical system were then
extrapolated to the national level for each population size category based on a relatively simple
probability decision tree. The sum of the costs for each population category yielded the national
compliance cost for a given regulatory scenario.
With the exception of system counts, a comparable model describing the flow and population
characteristics does not exist for non-community water systems.
2.2 Rationale for Revising the Existing Model
One of EPA's objectives is to expand the range of the existing model for drinking water treatment
profiles. The impetus for revising the existing model stems from Agency experience in promulgating
drinking water treatment regulations during the 1980s and early 1990s. EPA's OGWDW recognizes the
need to improve existing tools and models for regulatory impact analysis to more adequately describe
regulatory impacts. The need to develop improved methodologies came into focus with the advent of a
new regulatory climate in the 1990s; the passage of the Unfunded Mandates Reform Act (UMRA), the
Small Business Regulatory Fairness Act (SBREFA), and Executive Order 12866; the reauthorization of
the SDWA in 1996; and calls from stakeholders to improve process used to estimate benefits and costs
for upcoming drinking water regulations.
In 1996, EPA convened a Blue Ribbon Panel (the Panel) of experts from academia, the water treatment
industry, and State, local, and Federal governments to help critically evaluate various components of the
Geometries and Characteristics of
Public Water Systems
2-2
December 2000
-------
Agency's regulatory analysis approach. One of the components addressed by the Panel was the issue of
modeling PWS characteristics. Public comments and published reports suggested that EPA's existing
water system profile needed to be revised to accommodate more sophisticated analysis and to improve
their capability to assess impacts for various types of public water systems.
The Panel's recommendations that apply to improving the water industry profiles follow:
>• Examine the relationship between population served and flow. Examine whether
population truly relates to flow in a system. The Panel noted that the relationship does
hold up on average but variations across systems in per capita flow (as much as 10 to 1)
can be expected.
* Investigate the design-to-average flow ratios for small systems. The Panel believes that
these ratios may be high. For medium to large systems, a reasonable ratio is between 1
and 2.
>• EPA should incorporate greater diversity into the analysis so the results support a variety
of objectives and inquiries. The Panel proposed a classification scheme based on three
primary variables: system size, type of ownership, and source water.
* System size categories should account for the differences in technical, financial, and
managerial characteristics. At least five general categories are necessary to capture this
diversity. Very small, small, medium, large, and very large systems should be captured
in the scheme.
>• EPA should devote additional analysis to systems that have more than one treatment
plant or "entry point" into the distribution system.
In addition to strictly flow-based models, the Panel also suggested looking at other variables to develop
profiles. This work is being carried out under a separate assignment.
Recommendations by the American Water Works Association Research Foundation (AWWARF) also
focused on developing profiles based on source water characteristics. AWWARF concluded that average
daily and design flows are different for ground and surface water systems, and that the number of entry
points into a system affect compliance costs.
The Panel's and AWWARF's recommendations provided basic guidelines for developing revised models
to characterize PWSs. The existing model described above does not have the capacity necessary to
address some of the recommendations. For example, the data set on which the existing model is based
does not distinguish systems by ownership or by source water category, nor did it address characteristics
such as numbers of entry points per system. To address the various concerns and develop a revised
model of drinking water treatment profiles, EPA turned to the two most comprehensive sources of
information available on the full spectrum of drinking water systems: EPA's Safe Drinking Water
Information System (SDWIS), described below, and the Community Water Systems Survey (CWSS) of
1995, described in Chapter 3.
2.3 Characterization of the Water Supply Universe
SDWIS provides the most complete inventory of the U. S. water supply industry. It contains information
about public water systems and their violations of EPA regulations for safe drinking water. The
inventory is used not only for compliance tracking purposes, but also to assist in allocating grant monies
Geometries and Characteristics of 2-3 December 2000
Public Water Systems
-------
among the States. A considerable effort is expended to ensure that SDWIS accurately fulfills these
needs.3
Near the end of each calendar year, a snapshot of the SDWIS inventory is distributed to State drinking
water programs for verification of the number and types of systems, a process that customarily takes
several months. The inventory reflected in this report were derived from the December 1998 database.
Because both population figures and system counts change continuously, these figures should be
considered representative of a particular time, not a static universe. For example, those performing risk
analysis should consider that the number of private systems has been increasing over the years (Dysard,
1999), and some believe that the number of wholesale systems will increase as well. There has also been
a steady increase in per capita water use, which will affect system average and design flow data (Linsley
et. al, 1992). Notwithstanding, the figures presented in this report represent the broad universe of
populations and systems to be considered in the risk assessment. Information contained in SDWIS is
complete for the categories identified by the Blue Ribbon Panel as important for industry
subcategorization. Core verified data in the inventory include:
* Federal identification (ID) number
> Source water (ground, surface, and ground water under direct influence of surface water
(GWUDI)
> Ownership type (local government, private, mixed, etc.)
»• Regulatory classification (community, transient, etc.)
Further discussion of the characteristics of these data elements and how they are proposed to be used in
regulatory impact analyses is provided in Sections 2.4 and 2.5. Exhibits 2.3 through 2.5 present the
numbers of systems subcategorized into traditional analytical categories. The ownership classification is
limited to private versus public since this the only distinction shown to have appreciable effects on the
technology cost models (Chapter 4 addresses this issue). Systems are further categorized by source water
(surface, ground, or GWUDI) and by whether they are purchased or non-purchased systems. Population
figures for the various sizes of water systems are presented in Appendix A.
2.4 Population Served
PWSs serve commercial, industrial, and residential customers. While it is generally true that systems
distribute their treated water directly to their customers, there are cases where water is wholesaled to
another utility that subsequently distributes water to customers. Thus, commercial, industrial, and
residential customers can be part of the retail population of a system (i.e., they receive water directly
from that system) or they can be part of the wholesale population of a system (i.e., they receive water
from a second system that buys their water from the first system). Exhibit 2.6 shows the link between
source water, treatment plants, and residential and wholesale population served by a system. One of the
Jlt is important to note, however, that complete names and addresses are not available in some States. The
absence of this core information suggests the inventory may still include inactive systems.
Geometries and Characteristics of 2-4 December 2000
Public Water Systems
-------
vO
Os
Os Os
fN
vo vo
vo vo
oo Os
fN
VO
fN
(N
(N
VO
fN
VO
fN
vo
VO
o
fN
OS
Os
vo
8t *
vo
o
vS
vo"
i *
MS
OS
SO
VO
Os
Os
s
vo
VO
fN
fN
VO
Os
C/5
CJ
Geometries and Characteristics of
Public Water Systems
2-5
December 2000
-------
<»
(N
„*
rj
cT
so
00
ri
o
OS
V
•**
ce
U
Ol
1
s
Z
vO
SO
Os
Os
S3
Os
NO
Os
SO
Os
SO
r--
r-
e
a
E
E
o
U
o
Z
.-
"35
I
Is
o
2*
w
SO
Os
Os
sO
SO
Os
Os
a * SI
V)
(N
SO
tN
Os
Os'
Os
04
SO
(N
u
§
(N
SO
-------
i
3
>>
o
u
s
Z,
en
E
0)
«*
^1
c/2
s_
41
•*•»
'E
g
S
E
o
U
e
o
Z
C
W5
1
H
•
^1
•^
£
U
&
§*
»
w>
s
5§
J«
8-
9*
ffc
m»
^ .
•£*
£.
|i|
- Si
i: f 1
*- 21
§2|
«?. m
§• * Si
*•< wi
i
go
^ ** ^
>** **:
*•« 'C^
2& ^. *S^
~ ""^
*• *!
* *4"
S s s
**** ^^ -tfi
^ i
Wi A- 3C
Private
^j-
*—
o
o
o
o
(N
(^
w—
—
ON
c.
1
•a
01
Purcha
o
vT
0
o
o
fN
(N
OO
(N
S
,-_
!o
g;
oo
^
f-^
CJ
6
o
oo
ON
^?
O
o
r-
oo
fN
OO
l/~J
oo
_
?
JC
fN
—
f^l
r*^
V
£
Ol
u
a
^
/i
ON
r—
m
o
o
^
ON
ON
fV]
0
ON
fN
NO
fN
o.
ON
r-
r^-
o
o
o
m
(SJ
>.«
fN
en
fN
fN
ON
Tj-
~—
fN
Private
^j-
i — i
0
o
o
—
fN
O
^~
0
NO
fN
OO
NO
• M
^
3
Q.
•o
Ol
tt
Purcha
f*-i
TJ-
o
o
o
0
—
T-J-
-
fN
NO
—
rn
rn
_QJ
cc
'u
c.
•6
Ol
V)
Purcha
^>
fN
O
O
fN
fN
m
m
_
V)
O
oo
—
^«
01
6
1 —
oo
o
o
o
o
o
m
NO
fN
v5
fN
U
01
3
C
C
^
01
•o
C
^
V
fN
O
o
o
o
o
o
ON
rf
0
o.
VJ
O
o
o
o
o
fN
O
Tj-
*"•
ON
fN
Private
fN
O
O
0
o
o
o
o
fN
O
0
.«
.fl
0.
•o
01
Purcha
o
o
o
o
o
o
o
0
o
o
0
Ol
GQ
_>
e.
1
•o
Ol
V3
Purcha
rs
O
O
0
o
o
_
o
__
o
o
L.
Ol
o
VI
t--
VI
ON
O
ON
—
OO
r—
oo
r-
f^
r-
VO
o
fN
^j.
NO
ON"
—
S
Tj-
fN"
v^
ON
NO
^^
"o
f-
ON
ON
m
O
oo
oo
5
o
f^
^^
f^l
oo
NO
o
ON
fN
fN
O
fN
ON"
ON
|
O
NO
^
f—
o
—
V)
—
ON
OO
fN
oo
fN
_
2
NO"
—
£
—
oo"
oo
^«
NO
Private
Vl
ON
V
0
O
fN
fN
VI
fN
O
fl
fN
t-
r-
Q
vT
00
u
Ol
6
Geometries and Characteristics of
Public Water Systems
2-7
December 2000
-------
Exhibit 2.6
Representation of System Relationships
Wells
System A
Treatment
Plant
: ^7 Detail Pojiuiatibntbf System A "
' •""•'• "' '.• ? ' *" '*"•'.;
Distribution System
Treatment
Plant
River
Well
System B
j!^
(Equals Wholesale ^op^lation of System A)...
Distribution System
Geometries and Characteristics of
Public Water Systems
2-6
December 2000
-------
most complicated examples of these relationships is the Metropolitan Water District of Southern
California (MWDSC), which has been estimated to serve between 8 and 16 million people, all through
wholesale relationships (purchased water systems).
As noted earlier, the primary source of information on water treatment utilities is SDWIS, which tracks
population served by a system on a retail customer basis only. Comprehensive information on total (retail
plus wholesale) populations does not exist except for the largest water systems. This data reporting
approach is useful for many reasons, but, as can be seen by the MWDSC example above, any analysis
using these data must correctly interpret the results and explore the potential or degree of bias.
The current system categorization scheme within SDWIS introduces a bias into the cost analysis. Since
water systems are classified based on retail population served, systems that purchase their water from
another system and distribute it are considered stand-alone systems serving only their retail customers.
The overall result is that this classification process accounts for the total national retail population and
flows, but assumes more individual systems of smaller sizes treating their water than actually occurs.
Costs are then higher than would actually occur because no economies of scale are available (i.e., it is
less expensive on a unit cost basis to install a technology at larger plant than to install technologies at
smaller plants).
The next several paragraphs summarize an attempt to quantify the cost bias. The retail population served
by purchased water systems were allocated to nonpurchased systems and then theoretical costs were
generated using the two classification schemes. Detailed descriptions of the analyses and all associated
assumptions and calculations are included in Appendix B.
Step 1: Evaluate January, 2000 SDWIS data to determine the degree to which CWSs that purchase water
do so from systems of similar sizes or of different sizes. This analysis addressed only cases where
primary source of the buyer and seller is the same and did not include systems with cascading provider
relationships (i.e., where a seller provides water to a purchased system which in turn sells water to
another purchased system). The analysis evaluated surface water, ground water, and GWUDI systems
separately and looked only at four major size categories of systems (very smalls, smalls, mediums, and
larges). GWUDI systems are not included in subsequent analyses because they represent such a small
portion of the public water system universe.
Step 2: Allocate the populations for the purchased water systems to nonpurchased water systems to
estimate the effective shift to higher size categories. This was done by estimating the mean population
for the purchased water system and allocating it to the nonpurchased systems based on the percentages
developed in Step 1. The smallest impacts were observed when very small systems buy from any size
category (the new median population is not large enough to move a system into another size category),
and the largest impacts are seen when the large systems buy from any size category (no matter the size of
the seller, they all become large systems). Exhibits 2.7 and 2.8 summarize the results of this step for
surface and ground water systems, respectively. The first three rows summarize the total CWS universe
on a retail population basis. The final row presents the number of systems as they might exist if retail
and wholesale population were combined for modeling purposes.
Geometries and Characteristics of 2-9 December 2000
Public Water Systems
-------
Exhibit 2.7
Comparison of System Categorization Schemes for Surface Water Systems
System Typ*
Nonpurchased Systems (a)
Purchased Systems (b)
Total Systems . Retail Based
(a)+(b)
Total Systems, Retail +
Wholesale Based
Number ofSwrfsce Water CWSs by System Sfee Category*
25
to
im
382
483
865
198
101
to
500
696
1,185
1,881
362
501
to
1,000
411
719
1,130
461
''•• 1,001
to
3,300
1,086
1,282
2,368
1,217
3,301
to
10,000
958
828
1,786
1,074
10,801
to
50,800
913
603
1,516
1,080
50,001
to
100,080
178
89
267
210
100,001
to :
1,880,000
200
47
247
221
Greater
than
1,000,000
14
0
14
16
Totals ;
4,838
5,236
10,074
4,838
*Nonpurchased and purchased systems represent the sum of public and private systems for this analysis. Does not include
"other" ownership category.
Exhibit 2.8
Comparison of System Categorization Schemes for Ground Water Systems
System Type
Nonpurchased Systems (a)
Purchased Systems (b)
Total Systems , Retail Based
(a)+(b)
Revised Systems, Retail +
Wholesale Based
25:
to
180
13.438
285
13,723
13,401
1
tot
to
500
13,903
773
14,676
13,864
Smnbero
S&l
to
1,000
4.244
366
4,610
4,270
fGrottfld
WMM
to
3,300
5.316
355
5,671
5,349
Water C
3301
to
10,000
2.348
95
2,443
2,362
WSsby
10,001
to
50,000
1.180
35
1,215
1,181
System $
50,001
to
100,000
130
1
131
130
tee Cstegw
100,001
to
1, 000,000
57
4
61
59
ry*.
Greater
thai*
1,000,000
2
0
2
2
Totals
40.618
1,914
42,532
40,618
*Nonpurchased and purchased systems represent the sum of public and private systems for this analysis. Does not include
"other" ownership category.
Step 3: Estimate national costs of treatment using the two categorization schemes. Three technologies
were selected from the EPA Document, "Technologies and Costs for Control or Microbial Contaminents
and Disinfection Byproducts" (November 2000). Total annual cost per system (yearly operation and
maintenance costs and annualized capital cost) were estimated for each size category, and national costs
were calculated using the two system categorization schemes (retail based and retail+wholesale based).
Results from step three show that for surface water systems, national costs of treatment could be 22 to 45
percent higher than what might be incurred if retail+wholesale based system categorization is used. The
effects on cost estimates for ground water systems is much less (approximately 4 percent increase).
Despite this bias, EPA believes that the certainty afforded benefits estimates through the use of the
SDWIS inventory justifies its use.
Geometries and Characteristics of
Public Water Systems
2-10
December 2000
-------
2.5 Source Water
Another area of interest to the Blue Ribbon Panel and others relates to distinguishing systems on a source
water basis. While most small water systems have one source only, it is not unusual for larger water
systems to have multiple sources of water supply. These source waters may include ground water,
surface water, and ground water under the direct influence of surface water (GWUDI). Most systems,
however, have predominately one source water type. Many past regulatory estimates have not modeled
mixed systems separately. To the extent that occurrence profiles differ in ground versus surface water, or
when regulations only impact one type, accounting for the numbers of these mixed systems is important.
SDWIS defines any water system with a continuous input of surface water as a surface water system.
This is the case even if 99 percent of the water is of ground water origin. The CWSS, however,
distinguishes and groups systems based on the predominant source water type, or the source water(s) that
supply more than 50 percent of water for the entire system. The CWSS not only categorizes the entire
system by source water type, but further categorizes each entry point into the distribution system by
source water type and treatment.
Extrapolating from the CWSS information, Exhibit 2.8 presents an estimate of the number of non-
purchased mixed systems that SDWIS classifies as surface water systems. Exhibit 2.8 also summarizes
the number of systems out of the total inventory that would require regrouping if SDWIS had used
predominant source type as the classification scheme. As shown in the exhibit, it is estimated that 1,069,
or 21 percent, of non-purchased surface water systems in SDWIS have some ground water source flow,
and 435, or 8 percent, of the non-purchased surface water systems get the majority of flow from ground
water sources. The CWSS did not provide enough information to perform a similar analysis for
purchased water systems. Analysts performing inventory subcategorizations by source need to carefully
consider these numbers when performing regulatory impact analyses.
Exhibit 2.9
Analysis of Mixed Systems in SDWIS Non-Purchased Surface Water Systems
>
System Type
(1) Total Ground Water
Systems (SDWIS)
(2) Total Non-Purchased
Surface Water Systems
(SDWIS)
(3) Number (and %) of
Surface Water Systems
with Ground Water
Component*
(4) Number (and %) of
Surface Water Systems
with Majority Flow from
Ground Water Sources*
: Population Category
l*ss
than
in
14,391
599
22
(3 7%)
22
(3 7%)
100
to
500
15,070
853
123
(14%)
82
(9.6%)
501
to
t,ooo
4,739
473
33
(7 0%)
0
1,00 1
to
3,390
5,726
1,179
225
(19%)
69
(5 9%)
3,301
to
10,000
2,489
1.008
264
(26%)
125
(12%)
WMl
to
50,000
1,282
934
249
(27%)
93
(10%)
50,001
to
106,000
139
180
68
(38%)
16
(8 9%)
100,001
to
1,000,000
70
200
80
(40%)
28
(14%)
Greater
than
1,000,000
2
14
5
(35.7)
0
Total
43,908
5.440
1,069
(21%)
435
(8.4%)
Extrapolated from CWSS Data
Geometries and Characteristics of
Public Water Systems
2-11
December 2000
-------
This page intentionally left blank
-------
3: Community Water System Survey of 1995
As discussed in Chapter 2, EPA used two primary data sources in this study: SDWIS and the 1995
CWSS. SDWIS was used to develop the detailed inventory of public water systems presented in Chapter
2. The CWSS provided the data necessary to analyze the characteristics of public water systems:
population and flow relationships, number and types of treatments-in-place, and entry points into the
distribution system. The CWSS is described in detail below, along with the reasoning behind its
selection as a primary data source.
The EPA OGWDW periodically conducts surveys of the financial and operating characteristics of
community water systems. The most recent of these is the 1995 CWSS. The purpose of the 1995 CWSS
was to collect information that would do the following:
>• Help EPA and States develop and implement proposals for reauthorizing the Safe
Drinking Water Act (SDWA)
*• Facilitate water system capacity/development
>• Help determine the need for and design of Best Available Technology (BAT) programs
> Support economic and financial analyses of new and revised regulations
* Help EPA identify, evaluate, and provide guidance for best management practices
Because the purpose of the CWSS is closely aligned with the purpose of this analysis, the survey
provides an excellent source of information for subsequent chapters of this report. The following
specific characteristics of the CWSS make it a useful data source:
>• The survey collected the type of data required for this analysis (e.g., population, flow,
treatment technologies)
•• The sampling method allowed generally accepted statistical protocols
*• The survey was specifically designed to capture systems with differing sizes, types of
ownerships, and water sources
* The survey incorporated extensive peer review of its design and the results were
subjected to extensive quality-assurance procedures
> Specific validation efforts focused on the data critical to this analysis (specifically,
population, flow, and treatment facility information)
>• The survey data are readily available and amenable to the additional screening required
for this analysis
The sections below provide additional background on the CWSS, its statistical design, and the quality-
assurance efforts incorporated in the survey. These sections provide further details on the characteristics
discussed above and illustrate the survey's usefulness for this analysis. Much of the discussion below is
Geometries and Characteristics of 3-1 December 2000
Public Water Systems
-------
based on information provided in the EPA report "Community Water System Survey: Volume II:
Detailed Survey Result Tables and Methodology" (January 1997).
3.1 Survey Overview
EPA began the 1995 CWSS in the fall of 1994. The survey included two phases: a telephone screening
survey (Phase I) and a substantive mail survey (Phase II). Phase I was targeted toward a sample of 5,856
water systems from the more than 57,000 systems identified in the Federal Reporting Data System
(FRDS), EPA's registry of public water systems (now known as SDWIS). The purpose of Phase I was to
identify water systems eligible for the Phase II mail survey and the appropriate contacts for the Phase II
survey. Phase I also collected basic data on system size, ownership, and water sources to verify the
preliminary information from FRDS and contribute to the design (i.e., set the "sampling frame") of Phase
II. In the Phase I screening, conducted from November to December 1994, 4,729 eligible community
water systems were identified.
Based on the Phase I findings, 3,681 systems were selected to receive the Phase II mail survey. Phase II
involved collecting a variety of substantive operating and financial data, including information about the
following:
* Production and storage
»• Distribution
> Operator training
> Water sources and treatment
* Source water protection
* Revenues and expenses
*• Assets, liabilities, and debt
*• Capital investment
Of particular importance for this analysis, information on population served, drinking water flow rates
(average daily flow, peak daily flow, and maximum daily treatment design flow), treatment systems in
place, and number and location of entry points into the distribution system was collected during Phase II.
Appendix C presents the questionnaire used in Phase II. The mail survey was conducted from June 1995
through March 1996. A total of 1,980 systems (approximately 50 percent of those surveyed) responded.
Although not every system responded to every question, the majority of systems provided the data crucial
to this analysis (e.g., population, flow). Thus, the CWSS provides a substantial database for use in this
analysis.
3.2 Statistical Design
This section describes those elements of the CWSS's statistical design that are significant for the
purposes of this analysis. The EPA report, "Community Water System Survey: Volume II: Detailed
Survey Result Tables and Methodology," contains a more complete and detailed description of the
survey design.
To ensure that the results would capture a range of system sizes, types of ownership, and water sources,
the CWSS utilized a stratified sample design. A stratified sample is appropriate when subpopulations
within the larger population are expected to differ from one another in meaningful ways. In stratified
sampling, the population is first divided into subpopulations called strata and random samples are
Geometries and Characteristics of 3-2 December 2000
Public Water Systems
-------
selected from within each stratum. This technique ensures adequate statistical representation for each
subpopulation.
Phase I of the CWSS defined strata based on eight categories of size, two categories of ownership, and
two categories of source water, for a total of 32 strata. Responses to Phase I resulted in identifying
inaccuracies in the initial placement of systems into strata. That is, the Phase I responses showed that the
size, ownership, or source water categorization of some systems was different from that expected based
on the initial FRDS data. This reclassification of systems, or "stratum migration," required the use of a
more complex stratification scheme in Phase II in order to obtain optimum sampling rates; 37 strata were
used during Phase II.
Interpreting stratified sampling results at the full population level requires the use of sample weights.
The importance of an individual system's survey response depends on how much of the stratum it
represents. For example, if one samples 100 systems out of a stratum with a total population of 200
systems, the base weight of each sampled system is 200/100 = 2. These base weights must be adjusted to
account for nonresponses during sampling. When some systems in a stratum do not respond, the
proportion of the stratum represented by each respondent changes.
For the CWSS, base weights were adjusted weights for nonresponses in both Phase I and Phase II.
Further adjustments were made to account for the new strata introduced by stratum migration after Phase
I. Weights were adjusted to account for aggregation in the responses (i.e., some respondents submitted
combined responses for multiple systems). Finally, weights were "trimmed" for some systems, with
extreme weights to reduce variation and increase the precision of sampling estimates. The EPA report,
"Community Water System Survey: Volume II: Detailed Survey Result Tables and Methodology,"
describes each of these adjustments. All of these adjustments resulted in a final weight for each survey
response, which was reported along with the survey results.4
In addition to characterizing a stratified population, another of the survey's design objectives was to
achieve a minimum statistical confidence level. Specifically, the number of samples taken from within
each stratum had to be sufficient to obtain estimates with an error not exceeding 10 percent at the 95
percent confidence level. That is, if 50 percent of sampled systems in a stratum reported a certain
characteristic, EPA could be 95 percent confident that between 40 and 60 percent of the full population
of the stratum have that characteristic. Because of stratum migration and nonresponses, the CWSS did
not quite achieve this confidence level for all strata. However, most strata did meet this confidence level
and the maximum error did not exceed ±15 percent for any stratum.
One additional relevant characteristic of the CWSS design is the actual sampling strategy employed.
Within each stratum, candidate systems were sorted by EPA region and, within each region, by
population served. The CWSS then used systematic equal probability sampling to select the surveyed
systems. This approach ensured geographic representation of the systems sampled and increased the
probability that a range of population sizes within a stratum was represented.
All these elements of the survey's design are relevant for the purposes of this analysis. The sampling
design allows characterization of systems with different sizes, ownerships, and source water types, while
ensuring geographic representation. The sampling weights facilitate the use of data in modeling. The
achievement of reasonable reliability imparts confidence in estimates based on CWSS data.
4 The final weights were further adjusted for item-level nonresponse. The process and rationale for
making these adjustments is discussed in Chapter 4.
Geometries and Characteristics of 3-3 December 2000
Public Water Systems
-------
3.3 Peer Review and Quality Assurance
Prior to its implementation, the CWSS was the subject of peer review and testing. Draft versions of the
survey questionnaire were peer reviewed by representatives of the National Rural Water Association, the
American Water Works Association, and the National Drinking Water Advisory Council; by a consultant
from the Government Finance Group; and by an independent consultant specializing in the operational
characteristics of drinking water systems. The questionnaire was pretested with nine water systems in
Maryland and Delaware. Following the pretest, the full survey process (the sampling routine, Phase I
telephone screening, and Phase II mail survey) was pilot tested with 81 systems. As a result of this
review and testing, the survey designers made improvements to the sampling plan and to telephone
interviewer training. There also were changes in the terminology, content, and structure of both the
Phase I and Phase II surveys. These changes increased the likelihood that respondents correctly
interpreted the survey questions which enhanced the validity of the results.
The CWSS also incorporated extensive quality assurance (QA) procedures during and after both Phase I
and Phase II. QA during Phase I included automated online response checks and periodic staff review of
accumulated results. After completion of Phase I, the results were reviewed against FRDS data. QA
during Phase II included supervision and spot checking during mailing preparation, pre-data-entry
editing, independent double-key entry to minimize data entry errors, and automated data range checks
during entry. At the end of Phase II, a final automated data validation effort included statistical
evaluation, cross-tabulation checks of related variables, and internal logic checks. The purpose of this
validation was to verify consistency and reasonableness and to guide expert review of individual
responses. The automated validation examined 500 of the 600 survey variables, including all those used
in this analysis. Problems resolved by this process included order-of-magnitude reporting errors, such as
the use of gallons instead of million gallons for some questions. At the conclusion of the validation
process, a few extreme data outliers (approximately one quarter of 1 percent of the data points) were
excluded from the results.
In addition to the in-process QA and automated validation, the CWSS also included manual validation
and expert review of responses to eight critical survey questions. These included several questions of
importance to this analysis: sources of water (including the data from which average daily flows were
derived), population served, and treatment facility information. The manual validation process was used
to review answers to these questions for completeness and internal consistency with answers to other
questions. When problems were found, the reviewers attempted to derive an answer using responses to
other questions, estimate an answer using best professional judgment, or contact the respondent for
clarification. Examples of corrected problems include incorrect units and mathematical errors.
The extensive review, testing, and QA incorporated in the CWSS allow increased confidence in the
validity of the survey results. The additional manual validation of the data elements used in this study
provides further assurance of a realistic basis for further analysis.
Geometries and Characteristics of 3-4 December 2000
Public Water Systems
-------
4: Analysis of Population and Flow Relationships
This chapter presents a model for defining CWS size categories for regulatory analysis. The model is
based on a regression analysis of the relationship between flow and population. Section 4.1 presents the
rationale for selecting flow and population as the variables for performing the regression. Sections 4.2
and 4.3 present the analysis of average daily flow and maximum daily treatment capacity (design flow)
for CWSs.
4.1 Population and Flow as Critical Variables
Population and flow (average daily and design), are two key variables in the development of regulatory
impact analyses. A system's average daily and design flows are driving factors in estimating potential
operating and capital expenditures. The corresponding population served identifies the number of people
who will derive benefits from compliance. Further, population served, when coupled with exposure
information, provides a basis for estimating household benefits and costs of regulatory alternatives.
EPA studies5 and the published literature point to population and flow variables as basic defining
features of CWSs. Thus, an analysis of population and flow is a logical starting point for modeling the
characteristics of drinking water systems for regulatory benefit and cost analysis. There are more
sophisticated means of forecasting urban water use than basic flow and population equations; however,
considering the water characteristics of interest described above, the single variable approach for
predicting flow based on population will meet regulatory analysis needs. Also, other variables do not
readily lend themselves to developing a model for estimating regulatory benefits and costs. For example,
variables linked to a system's water sales could be used, but revenue information is more difficult to
obtain and not as easily linked to the amount of water to be treated.
Acceptance of population and flow as critical variables still leaves an important issue to be resolved. As
discussed in Chapter 2, EPA tracks systems based on retail population served. This criterion also was
used in the stratification of the CWSS (i.e., respondents were asked to report the number of people served
by their systems and number of residential connections). Treatment expenditures, however, relate to total
water treated. Consequently, if EPA used retail population with total flow estimates, double counting
would result. In particular, the wholesale portion of flow would be repeat-counted when costs are
estimated for purchased water systems. While one could eliminate purchased systems from the
cost/benefit analysis to avoid this double counting, data are not available on wholesale customers served.
Even if double counting were avoided, inclusion of wholesale flows without including the associated
population would bias the flow model by distorting the population/flow relationship. Household costs
would be overestimated.
For all of the aforementioned reasons, this analysis adjusts system flows to remove sales for resale from
wholesale systems prior to regression analysis. The net effect of this approach is to "assign" these flows
to the purchased water systems. This assignment makes it possible to use the SDWIS inventory directly
for compliance cost estimation. The disadvantages of this modeling approach are twofold:
5 Cummins, Michael D. 1987. Analysis of Flow Data. Report prepared for EPA, Office of Drinking
Water. Octobers.
Geometries and Characteristics of 4-1 December, 2000
Public Water Systems
-------
>• Costs and benefits to the largest systems may be underestimated. Because half of the 65
largest utilities wholesale to smaller utilities, the associated costs and benefits will
appear in lower size categories.
* Total national costs will be overestimated. Due to economies of scale, unit treatment
costs generally vary inversely to system size. Wholesale flows from larger systems that
reflect economies of scale are assigned to smaller purchased water systems that
otherwise do not capture these economies of scale.
The impact of the first disadvantage will be addressed by not applying these models to the largest
systems; rather, EPA will attempt specific estimation based on actual configurations for these systems.
As for the second disadvantage, the affect of this bias has been quantified in Chapter 2.
4.2 Analysis of Average Daily Flow versus Population
A regression analysis of average daily flow and a system's corresponding retail population was
performed. A regression analysis examines the nature and strength of the relationship between a variable
of interest (known as a dependent variable) and one or more other variables (known as independent or
explanatory variables) that are believed to affect the first variable. In this case, average daily flow is the
dependent variable and retail population is the single independent variable.
Based on previous studies, a water system's flow is known a priori to be strongly dependent on
population served. Therefore, the purpose of this regression analysis is to (1) confirm the strength of this
relationship using the statistically sound CWSS data set, and (2) develop a model or models describing
the relationship between population and flow.6
4.2.1 Data Screening for Regression Analysis
Prior to regression analysis, the CWSS database of 1,980 respondents was screened for missing data
points. The initial screening procedure entailed a two-step process: applying a formula for average daily
flow followed by the elimination of nonresponses and zero values. In the first step, adjustments were
made to average daily flow reported in Question 4 of the CWSS questionnaire (see Appendix C).
Question 4 reports total annual surface water, ground water, and purchased water flow.
The reported flow in Question 4 represents the total flow of a system. This may include wholesale flow,
which is the portion of total average daily flow the system sells to other water utilities. To obtain retail
flow, total flow reported in Question 4 was adjusted by subtracting wholesale flow as described below.
Question 29 of the CWSS questionnaire reports a facility's total production flow divided into various
uses (see Appendix C). These uses include water delivered to residential, commercial, industrial,
6 In the CWSS, flow is reported at the system-level and at the entry point level. An entry point is typically
a treatment plant but can also be any location where potential treatment could be installed. Systems (particularly
large ones) often can have more than one entry point. Analysis of flow and population relationships can be
performed at either the system or entry point level. Since population is reported in the CWSS only at the system
level, the regression analysis was performed at the system level. Analysis of entry point characteristics are
presented in Chapter 5.
Geometries and Characteristics of 4-2 December, 2000
Public Water Systems
-------
wholesale, local, and other customers. In theory, the flow reported in Question 4 should match the sum
of the flows reported in Questions 29 and 32 (losses and water supplied free to municipal uses). Where
significant discrepancies existed, responses were omitted. As flows had been subjected to the highest
level of QA in conducting the survey, it was believed that other categories would be, at best, no more
reliable.
The following formula was used to adjust the dependent variable (i.e., average daily flow):
Averaae Dailv Flow = (Question 4 Annual Flow - Question 29 Annual Wholesale Flow)
365 days
This formula was applied to all 1,980 respondents in the CWSS database. Following the adoption of this
convention, the data in Questions 4 and 11 were screened. Of the 1,980 respondents, 184 systems did not
report adequate flow information in Question 4, and 44 did not report a population or the number of
service connections in Question 11. Nine of the largest systems also were eliminated from the data set
since these systems will be analyzed on a site-specific basis.7 An additional nine systems were
eliminated based on either population or flow-reporting discrepancies. Ninety-nine systems did not
report a population but reported the number of service connections. Populations were imputed for these
systems by using a population-served to number of service connection ratio. For the eight population
categories in CWSS, population served to number of service connection ratios shown in Exhibit 4.1 were
used to make this adjustment.
Exhibit 4.1. Population-Served to Number of Service Connections Ratios
System Size
(Population
Served)*
Ratio
Less
than
100
2.3
iOi
t«
500
2.4
501
to
1,000
2.6
1,601
to
3,300
2.9
3,361
to
10,000
3.0
10,001
to
50,000
3.7
50,001
to
100,000
3.8
100,001
to
500,000
5.3
* CWSS population categories
Data source: 1995 CWSS. Connections include residential, commercial, and industrial connections.
These eliminations resulted in a data set consisting of 1,734 records with paired responses for population
and total average daily flow, weighted for item-level non-response.
4.2.2 Distribution Analysis of Regression Variables
The distributions of the average daily flow and population data were evaluated using stem-and-leaf plots,
box plots, normal probability plots, coefficients of variation, and the Shapiro-Wilk test statistic. This
evaluation showed that it would not be reasonable to assume the data were normally distributed (e.g.,
plots did not appear normal in shape and the hypothesis of normality was rejected using the Shapiro-Wilk
statistic at the 5 percent significance level). When the data were transformed to a natural logarithm scale,
however, the same evaluation showed that it would be acceptable to assume flow and population were
Only 65 systems in the United States serve a total population greater than 500,000. Because they are few,
but have a potentially large impact (large systems serve about 20 percent of the population), EPA plans to perform
regulatory analyses for these systems individually.
Geometries and Characteristics of
Public Water Systems
4-3
December, 2000
-------
log-normally distributed (e.g., plots displayed normality and the hypothesis of log-normality was not
rejected at the 5 percent significance level). Based on this result, both the population and flow variables
were transformed to a natural logarithm scale (i.e., a log-transformed regression model was used).
4.2.3 Average Daily Flow and Population Regression
Exhibit 4.2 presents the regression for average daily flow versus retail population. The data show a very
good correlation, as indicated by the r value (0.90). While the fit is excellent, a limited number of
systems lie a considerable distance from the regression. A more detailed analysis of these points was
undertaken to determine whether the systems were outliers representing extreme isolated cases, or should
be included in developing nationally representative models.
Two approaches were considered for incorporating the variability of each system while eliminating those
that represented extreme cases. First, capping flow at the minimum and maximum per capita per day
values across the United States from U.S. Geological Survey (USGS) data,8 which would have resulted in
eliminating any system with flows less than 109 gallons per capita per day (gpcd) or greater than 344
gpcd, was considered.9 Comparison of flows reported in the CWSS with these values indicated that
trimming the data set based on the USGS information would have resulted in eliminating about 20
percent of the data. This would have dramatically reduced the sample size. Furthermore, the available
literature and engineering judgment indicated that a variability in flow rates much wider than that
suggested by the USGS data would be reasonable (Bauman et. al, 1998). That is, only a small fraction of
systems have flows that would be considered truly extreme.
The second approach consisted of evaluating data points outside the 95 percent confidence band to
determine if they represent typical CWSs. Based on the literature and engineering judgment, systems in
this extreme range would encompass only facilities with very "low" or "high" flows. Also, these data
points, as observed in Exhibit 4.2, lie a significant distance from the vast majority of data points in the
regression. This approach was selected to maintain the representativeness of the data set while reducing
the effect on regression analysis from extreme values. 16 of 1,734 systems that reported average daily
flow outside the 95 percent confidence band were evaluated for representativeness.
8 U.S. Geological Survey. 1993. Estimated Use of Water in the United States in 1990. Circular 1081.
9 Note that these USGS values reflect "public supply" and include commercial and industrial uses.
Because some small CWSs serve domestic uses only, a more reasonable lower bound might be that for
domestic use only. The lowest per capita domestic use rate reported by USGS was 23 gpcd. Even using
this lower bound,
Geometries and Characteristics of 4-4 December, 2000
Public Water Systems
-------
Exhibit 4.2. Regression for Average Daily Flow
I
LN (Flow) = 1 0415 x LN (Pop) - 2 5288
R = 090
2 4
Source 1995CWSS
Of these 16 systems, 10 had very low flows, ranging from 0 to about 9 gpcd, and 6 systems had very high
flows, ranging from about 1,000 to 23,000 gpcd. Using their specific survey responses, it was possible to
draw conclusions about whether some of these 16 systems actually represented CWSs. Three of the low-
flow systems wholesaled all or nearly all of their flow and reported a population that appeared to
represent client systems, not retail, customers served. Two of the other low-flow systems were mobile
home parks and may have represented seasonal use. Other low flow systems may include populations
using wells. Of the high-flow systems, one was an abbey, one appeared to be a movie studio, one was an
irrigation system, one used most of its flow for agriculture, and one used most of its flow for commercial
and industrial uses. Therefore, based on the available information, the very high-flow and very low-flow
systems that were removed appear not to represent typical CWSs and were eliminated from the data set.
Exhibit 4.3 depicts the resulting regression after the removal of the extreme values. The r value increases
from 0.901 to 0.971 after the implementation rule. Both r values demonstrate excellent
correlation but the latter number is probably more representative of the universe of CWSs without biases
from systems producing extremely high or low volumes of water.
Note that, after data screening and elimination of extreme values, 1,718 systems remained with valid data
for average daily flow and population. To improve the accuracy of subsequent analyses of this smaller
data set, adjustments were made to the CWSS sample weights, as described below.
Geometries and Characteristics of
Public Water Systems
4-5
December, 2000
-------
Exhibit 4.3. Regression for Average Daily Flow (Extreme Values Removed)
14
S
&
5
3 4
LN (Flow) = 1 0698 x LN (Pop) - 2 7193.".
R =0 97
024
Source 1995CWSS
LN (Population)
4.2.4 Adjustment of Sampling Weights for Item-Level Nonresponse
The phenomenon of a system not responding to a question (e.g., not reporting a flow) is known as item-
level nonresponse (to distinguish it from primary sample unit nonresponse, which means a system did not
respond to the survey at all). When some systems in a stratum do not respond (or their responses are
excluded), the proportion of the stratum represented by each respondent changes. In these instances, the
accuracy of the analysis can be improved by adjusting samples weights.
After the initial regression analysis presented above, the adjustment for item-level nonresponse was
performed for the remaining 1,718 records. The adjustment consisted of further modification of the
sampling weights provided with the CWSS results, following the same approach used to account for
primary sample unit nonresponse (see Section 3.1.2). Subsequent analyses of average daily flow in this
report (e.g., in Section 4.4) used the newly weighted data. Although in this case the adjustment resulted
in only minor differences in results, the adjustment for item-level nonresponse was made to maintain the
validity of the survey design. For details on item-level nonresponse and the technique used to perform
regression analysis using weighted data, see Appendix D.
4.3 Regression of Design Flow Versus Population
Question 5b of the CWSS reports the maximum daily treatment capacity for the system, which is based
on a variety of engineering, planning, and design considerations. These include peak daily flow, peak
hourly flow, fire-fighting requirements, and population-growth estimates. Maximum daily treatment
capacity (design flow) determines the total amount of treatment facilities that may be necessary and is
used to estimate capital costs of complying with drinking water regulations. Therefore, a regression
Geometries and Characteristics of
Public Water Systems
4-6
December, 2000
-------
analysis was performed with design flow as the dependent variable and population as the independent
variable.
Note that peak daily flow also was reported in the 1995 CWSS. Peak daily flow also can be used to
estimate capital costs. Initial results for peak daily flow were presented to the Technology Design Panel
(TOP) at a workshop in November 1997. The TDP included government (local, State, Federal) and
industry representatives with expertise in the technical and regulatory aspects of the water treatment
business. The TDP recommended that design flow was more appropriate than peak daily flow for
regulatory analysis. In addition, systems responding to the CWSS generally reported design flows that
were greater than peak daily flows, further indicating that design flow is more representative of the
maximum required treatment capacity. Therefore, model systems were based on design flow (peak daily
flow was not analyzed further).
4.3.1 Data Screening
The analysis of population and design flow began with the 1,718 records remaining after the analysis of
daily production flow. Like daily production flow, the design flow was manipulated to link it directly to
the retail population reported in Question 11. The method used to adjust the design flow differed from
the method for daily production flow as follows. An initial review of the design to daily production flow
ratios showed that some systems had a design to daily production flow ratio less than one (reported
delivery exceeds capacity). Since this is not a desirable design or operating condition, it was inferred
that some respondents (i.e., systems categorized as primarily ground or surface water users) included
purchased water flow in Question 4 (daily production flow), but not in Question 5b (design flow). This
provides one explanation for the low design-to-average ratios. Based on this inference, the following
adjustment was made to design flow reported in Question 5B:
Design Flow =
Question 5b Design Flow
+
Question Ac Purchased Flow
- Question 29 Wholesale Flow
Wholesale flow was subtracted from the total to link the retail population to retail design flow.
Following this adjustment, it was found that 239 systems did not have useable design flows (design flows
either were not reported by the respondent or were negative following the adjustments). Therefore,
paired design flow and population data for the remaining 1,479 systems were used for the regression
analysis.
4.3.2 Maximum Daily Treatment Design Capacity and Population Regression
The regression line for design flow and population (Exhibit 4.4) indicates a strong correlation, with an r
value of 0.90, comparable to that for daily production flow (Exhibit 4.2). While there are fewer paired
data points (1,480) for design flow than for daily production flow, population and design flow continue
to show an excellent correlation. There are more systems that reported atypical design flows than daily
production flows. As noted previously, this could be the result of design flow being based on variables
other than population. An example of a variable affecting design flow is seasonal demand (e.g., for
irrigation). Atypical design flows are similar to the extreme flows observed for daily production flows in
that they are outside "reasonable" variances used to define baseline characteristics for national-level cost
and benefit analysis. To minimize the impact of these atypical or extreme values, systems that reported
Geometries and Characteristics of 4-7 December, 2000
Public Water Systems
-------
design flows outside a 95 percent confidence band around the regression line were set aside. As in the
previous regression analysis, these extreme values were analyzed separately.
Of the 48 atypical systems, 26 had very low design flows, ranging from less than 1 to about 19 gpcd, and
22 systems had very high design flows, ranging from about 1,500 to 65,000 gpcd. Using the specific
survey responses, it was possible to confirm that some of these design flows were atypical of CWSs. For
example, the low design flow systems include three mobile home parks, a university, and a country club.
Of the high design flow systems, three were mobile home parks and one was an apartment complex.
Two more of the systems with high design flows supplied the majority of their water to commercial or
industrial customers. Another system supplied all its flow to municipal buildings or parks. Finally, three
other systems with high design flows indicated that seasonal demand was more important than or equally
as important as current peak needs in determining design flow, suggesting these systems have a high
demand for irrigation or have fluctuating seasonal populations (e.g., resort areas). Therefore, based on
the available information, these atypical systems appeared not to represent CWSs, and were removed
from the analysis. This resulted in a final data set consisting of 1,431 records for subsequent
subcategorization (see following sections).
Exhibit 4.4 shows the regression line after the atypical values were removed. Following the elimination
of the 48 systems, sample weights were regenerated for the remaining systems using the protocol
described in Section 4.2.4 to account for item-level non-responses.
Exhibit 4.4. Regression Line for Design Flow (Extreme Values Removed)
140
Ln(Flow) = 1.0367 x LN (Pop) -1.6835
R = 0.94
60 80 100
LN (Population Served)
Geometries and Characteristics of
Public Water Systems
4-8
December, 2000
-------
4.4 Regression Analysis of Different Categories of CWSs
The regression analyses of daily production and design flows presented in the previous sections confirm
the strong relationship between population and flow. These analyses were based on pooled data that did
not distinguish between systems based on water source or ownership.
For regulatory analysis purposes, a model that distinguishes systems by ownership and water source is
advantageous in terms of precision and accuracy. For example, a given regulation might apply
differently to ground water systems than to surface water systems (e.g., a specific contaminant might be
expected to be present only in ground water). The treatment configurations of ground water and surface
water systems also vary. Surface water systems have fewer intakes and entry points, while ground water
systems can have large networks of wells. Purchased water systems often do not have their own treatment
facilities, but buy all treated water from another system. Finally, upcoming regulations address ground
and surface water systems separately. Considering these factors, regulatory analyses often must be able
to address costs and benefits by water source category.
In addition to source water type, stratifying systems by ownership category may also be advantageous for
regulatory analysis purposes. For example, costs for labor and capital can differ for public and private
systems. It also was suspected that the additional oversight provided by public utility commissions could
affect typical system capacity. For these reasons, it was deemed appropriate to examine systems by
ownership category to determine whether subcategorization of flow models was necessary.
Systems were categorized by ownership (public or private) and by source (ground water, surface water,
or purchased water), resulting in the following six strata.10"
* Public surface water systems
>• Private surface water systems
>• Public ground water systems
>• Private ground water systems
* Public purchased water systems
> Private purchased water systems
A regression line for each source water type was generated. Then, separate regression lines for each of
the six classifications were generated. The lines were tested statistically to determine if the lines are
different to provide a statistical basis in addition to regulatory analysis needs.
4.4.1 Regression Analysis for Different Strata
Exhibits 4.5 and 4.6 present the regression equations (weighted for item-level nonresponse) for different
CWS categories for daily production flow and design flow, respectively. Based on their r values, all the
10 The CWSS also classifies systems as ancillary. These systems produce water as a secondary activity to
their primary business function (for example, a paper mill that supplies potable water to its workers or sells it to the
public). These systems typically serve small populations (less than 500) and for all practical purposes function like
a private utility. Accordingly, they were combined with other private water systems.
" As discussed in Section 2.5, approximately 20 percent of surface water systems have some ground water
flow. For purposes of this report, these systems were categorized based on the source accounting for the majority of
their flow.
Geometries and Characteristics of 4-9 December, 2000
Public Water Systems
-------
lines continue to display a strong correlation. This confirms that the strong relationship remains for all of
the classifications.
Exhibit 4.5. Daily Production Flow Regression Equations for Subcategories of CWS
CWS Category
Ground (Public)
Ground (Private)
Surface (Public)
Surface (Private)
Purchased (Public)
Purchased (Private)
Regression Equation ;
(Daily Production Bow)
Y = 0.08575 X ' °5839
Y = 0.06670 X ' °6284
Y = 0. 14004 X099703
Y = 0.09036 X ' °3338
Y = 0.04692 X ' 10189
Y = 0.05004 X108339
95% Confidence Interval*
±1.965 [0.000651 + (!n(X)-8.049)2 /75 12.73] 1/2
±1.965 [0.000973 + (ln(X)-6.914)2 74746.7 1] 1/2
±1.969 [0.001004 + (ln(X)-9.334)2 /4858.16] 1/2
±1.976 [0.003628 + (ln(X)-7.620)2 /1 795.38] 1/2
±1.970 [0.001584 + (ln(X)-7.954)2 /3251.88] l/2
±1.972 [0.001335 + (ln(X)-6.873)2 /2388.03] 1/2
* Due to the statistical complexity involved in the calculating weighted confidence intervals,
confidence intervals shown are those for the corresponding unweighted regression results. Weighted
confidence intervals would be very similar.
Notes: Y = daily production flow (thousand gallons per day); X = population served. Regression
equations are weighted for item-level nonresponse.
Exhibit 4.6. Design Flow Regression Equations for Subcategories of CWS
CWS Category
Ground (Public)
Ground (Private)
Surface (Public)
Surface (Private)
Purchased (Public)
Purchased (Private)
Regression Equation
(Design Flow)
Y = 0.54992 X09553S
Y = 0.41682X096078
Y = 0.59028 X094573
Y = 0.35674 X °96188
95% Confidence Interval*
±1.967 [0.001384 + (ln(X)-8.335)2 73439.07] 1/2
±1.968 [0.002070 + (ln(X)-7.415)2 72194.66] l/2
±1.970 [0.001 326 + (ln(X)-9.38 1)2 73540.72] 1/2
±1.979 [0.004480 + (ln(X)-7.948)2 /1 39 1.46] m
See Section 4.5
See Section 4.5
* Due to the statistical complexity involved in the calculating weighted confidence intervals,
confidence intervals shown are those for the corresponding unweighted regression results. Weighted
confidence intervals would be very similar.
Notes: Y = design flow (thousand gallons per day); X = population served. Regression equations are
weighted for item-level nonresponse.
4.4.2 Statistical Tests To Determine Differences in Regression Lines
As noted above, the regression lines were tested to provide a statistical basis for subcategorizing the flow
and population model by ownership. First, for each source category, a regression analysis was performed
to relate population and daily production flow assuming no differences between systems based on
ownership. Then, a "dummy variable" was created to account for ownership category. A dummy
variable is an artificial measure created to describe a qualitative factor (in this case, a categorization) that
Geometries and Characteristics of
Public Water Systems
4-10
December, 2000
-------
cannot be measured numerically. The dummy variable used here was set equal to 1 for public systems
and 0 for private systems. Additional regressions were performed incorporating the dummy variable into
the original regression equation in various ways. The effect of ownership category
was then assessed by comparing the results of these regressions and the original regression using
statistical tests.
For each source category, these statistical tests concluded that regression models for public and private
systems are not identical. That is, F-tests comparing the original model to one incorporating the dummy
variable in both the slope and intercept term found differences between the two models at the 1 percent
significance level for ground and surface water systems and at the 10 percent significance level for
purchased water systems. Further tests examined the nature of these statistical differences. While these
tests did not demonstrate that ownership has a significant effect on the slope of the regression line, they
did conclude that ownership significantly affects the intercept (i.e., causes a parallel shift, up or down, of
the flow). That is, F-tests comparing the original model to one incorporating the dummy variable into the
intercept term found differences at the 1 percent significance level for ground and surface water systems
and at the 5 percent significance level for purchased water systems. These results were corroborated by
comparing the least square means of private and public systems within each water source category. This
second set of tests showed a statistical difference between ownership categories at the 5 percent
significance level for all three source water categories. The statistical calculations are provided in
Appendix D.
Once the statistical tests for daily production flow supported subcategorization, the same source and
ownership stratification was extended to design flows. Inconsistent subcategories among daily
production flow and design flow could result in a incongruous (and inconvenient) subcategorization
scheme.
4.5 Evaluations of Design-to-Average Flow Ratios
While the regression results for both design and average (daily production) flows are quite strong, it is
also important to consider the relationship between the two flows to confirm the reasonableness of the
full design capacity, which is necessary in water systems to ensure adequate service during peak demand
periods, for emergency flows, and for seasonal demand. EPA consulted the Denver Technology Design
Panel on the subject of full design capacity needs.
The Panel recommended that all water systems should have a minimum of 100 percent design capacity
(meaning the ratio of design flow to average flow should be at least 2). Further, it was believed that more
design capacity would be required by smaller systems. The Panel, however, did not reach agreement on a
specific minimum design capacity for these systems.
Design-to-average flow ratios were evaluated by source water type (ground, surface, purchased) for the
range of flows being modeled. In general, design-to-average ratios decrease with increasing system size
because of differences in treatment plant and distribution system configuration and water demand.
Smaller systems typically use less storage and experience sharper peaks in demand. Accordingly, they
are expected to have higher design-to-average ratios (to meet sharper peaking factors) than larger
systems. Exhibit 4.7 shows design-to-average ratio plots for the six categories of CWS. Both public and
private ground water systems have design-to-average ratios ranging from about 1.5 to 4.3. Surface water
systems have lower design-to-average ratios than ground water systems. Also, public surface water
systems have higher design-to-average ratios than private surface water systems. This difference may
stem from public surface water systems adding extra capacity based on a longer planning horizon, or
Geometries and Characteristics of 4-11 December, 2000
Public Water Systems
-------
from private systems having more frequent meter replacement and better control of unaccounted for
water.
Exhibit 4.7. Comparison of Design-to-Average Flow Ratios
Public Ground Water
Private Ground Water "'••5*
Public Surface Wafer
Private Surface Water -.
3
Public Purchased Water
Private Purchased Water
10
100
1,000
Population
10,000
100,000
1,000,000
This analysis of design-to-average flow ratios suggests that some modifications to the design flow and
population model derived in the previous section are necessary for regulatory analysis purposes. The
reasoning behind and details of these modifications are discussed in the following sections.
4.5.1 Design Flow Modification for Surface and Ground Water Systems
The Denver Technology Design Panel suggested to EPA that, for regulatory analysis purposes, a
minimum design-to-average ratio of 2 should be used for large systems. The panel indicated it was
unlikely that systems would install new capacity below this ratio. The population where the regression
equations produce design-to-average flow ratios less than 2 varies by CWS type. The equations in
Exhibit 4.8, therefore, should be used for estimating design flow for the ranges of population shown. The
equations in Exhibit 4.8 were developed using the following approach:
(1) For populations for which the design-to-average flow ratio is greater than 2, use the
design flow equation resulting from the regression analysis in Section 4.4.
(2) For populations for which the design-to-average flow ratio is less than 2, the daily
production flow equation (from the regression analysis in Section 4.4) was modified to
produced a design flow twice as large.
Geometries and Characteristics of
Public Water Systems
4-12
December, 2000
-------
Exhibit 4.8. Recommended Approach for Estimating Design Flow for Surface and Ground Water
Systems
Category
Public Ground
Private Ground
Public Surface
Private Surface
Recommended Approach to Estimate Design Flow
X<100,000. Y = 0.54992 X095538
X>100,000. Y = 0.17150X105839
X <90,000. Y = 0.4 1 682 X ° 96078
X >90,000. Y = 0. 1 3340 X ' °6284
For all populations. Y = 0.59028 X °94573
X<20,000. Y = 0.35674 X096188
X >20,000. Y = 0. 1 8072 X ' °3338
Note: Y = design flow (thousand gallons per day); X = population served
4.5.2 Design Flow Modification for Purchased Water Systems
As described in Section 4.2, systems are categorized by retail flow and population. This is done by
removing the wholesale portion of flow and assigning it to purchased water systems. It is assumed that
purchased water systems would bear the full cost of the design flow of their parent water system and
would exhibit similar design-to-average flow ratios. Preliminary regression analyses of the design flows
reported by purchased water systems, however, showed design-to-average flow ratios less than 1.5 for all
populations, as shown in Exhibit 4.7. The design flow equations were therefore modified for Exhibit 4.8
using the following approach:
(1) Starting with the population of the purchased water system, calculate the daily
production flow for the system using the equations developed in Section 4.4.
(2) To estimate a design flow with a similar design to average flow ratio as the parent water
system, back-calculate a "virtual" population for an appropriate parent water system
(e.g., a private ground water system) using the flow from step 1 above and the average
daily flow equation for the parent water system. That is, for a system purchasing water
from a private ground water system, substitute the purchased average daily flow into the
private ground water daily flow equation and solve for a corresponding "virtual"
population.
(3) Use this "virtual" population, calculate a theoretical design flow for the purchased water
system using the equations presented in Exhibit 4.6. That is, for a system purchasing
water from a private ground water system, use the "virtual" population derived in Step 2
in the equation in Exhibit 4.6 to estimate design flow.
Geometries and Characteristics of 4-13 December, 2000
Public Wafer Systems
-------
Exhibit 4.9. Estimated Design Flow for Purchased Water Systems
Corresponding $q«rcg
Water Category
Public Ground
Private Ground
Public Surface
Private Surface
Theoretical Estimated Design Flow for Purchased
Water System
X< 109,000. Y = 0.3 191 X09946
X > 109,000. Y = 0.09384 X ' 10189
X< 94,000. Y = 0.3215X09794
X > 94,000. Y = 0.10008 X ' °8339
For all populations. Y = 0.2092 X ' °452
X<21,000. Y = 0.2058 X10084
X>21,000. Y = 0.10008X108339
Note: Y = theoretical design flow (thousand gallons per day); X = purchased water system
population served
4.6 Summary of Population and Flow Relationships
Exhibits 4.5 through 4.9 detail the equations derived in this analysis for average daily flow and design
flow for the various categories of CWSs. Exhibits 4.10 through 4.13 present the results of this analysis
graphically. Using the model developed here, as shown in these figures, derived design flows are always
greater (at least 2 times) than derived average daily flows. On the logarithmic scale presented here, flows
for private systems are generally (but not always) lower than those for public systems. In the Exhibits,
differences between private and public systems may appear small, but these differences are statistically
significant, as discussed in Section 4.4.2. For example, a ground water system serving 100 people would
be projected to have a design flow of 45,000 gallons per day if it were a public system versus a design
flow of 35,000 gallons per day if it were a private system.
Geometries and Characteristics of
Public Water Systems
4-14
December, 2000
-------
Exhibit 4.10 Population and Flow Relationships for Ground Water
1000000
100000
10000
1000
§ 100
10
Design Flow (Public systems)
Design Flow (Private systems)
'Daily Produc8onFloW(Publicsysfems)
Daily Production Flow (Private Systems)
0.1
10
100
1000
Population
10000
100000
1000000
1000000
100000
10000
1000
100
10
Exhibit 4.11 Population and Flow Relationships for Surface Water
Design Flow (Public systems)
Design Flow (Private systems)
Daily Production Flow (Public systems)
Daily Production Flow (Private Systems)
01
10
100
1000
Population
10000
100000
1000000
Geometries and Characteristics of
Public Water Systems
4-15
December, 2000
-------
Exhibit 4.12 Population and Flow Relationships of Purchased Water Systems Fed by Ground Water
1000000
100000
10000
1000
100
10
Design Flow (Public systems)
Design Flow (Private systems)
Daily Production Flow~(Public systems)
Daily Production Flow (Private Systems)
0.1
10
100
1000
Population
10000
100000
1000000
Exhibit 4.13 Population and Flow Relationships for Purchased Water Systems Fed by Ground Water
1000000 f-
1 ——•— Design Flow (Public systems)
• Design Flow (Pnvate systems)
Daily Production Flow (Public systems)
Daily Production Flow (Private Systems)
I
100000
10000
1000
100
10
0.1
10
100
1000
Population
10000
100000
1000000
Geometries and Characteristics of
Public Water Systems
4-16
December, 2000
-------
5: Analysis of Entry Point Configurations
Entry points are locations where untreated water, treated water, or purchased water enter the distribution
system network. A public water system may just have one entry point supplying all of its drinking water,
or multiple entry points from different types of sources. In a system with more than one entry point, one
may provide a majority of the flow to the distribution system. Generally, larger systems have more
complex configurations. Also, configurations can be more complex for ground water systems because
individual treatment plants can be supplied by networks of wells. Ground water systems also can have
untreated wells (or networks of untreated wells) connected to the distribution system. Exhibit 5.1
presents examples of surface and groundwater entry point configurations.
In Chapter 4, a design and average flow model was developed at the system level. This chapter expands
on that model by providing the information necessary to address systems based on the number of entry
points. For regulatory impact analysis purposes, it may be relevant to distinguish systems with multiple
entry points. For example, consider a ground water system with multiple wells. If a regulated
contaminant affects only one ground water well, then treatment would be required only for the entry
point using that well. An accurate estimate of compliance costs for the system would consider treatment
only at that entry point. Also, the regulatory impact analysis may need to consider differences in
economies of scale between systems that treat all water in a common facility versus separate facilities at
separate entry points.
For systems with multiple entry points, other aspects of their configurations beyond the number of entry
points also may be relevant for regulatory analysis purposes. Because estimates of compliance cost
typically are based on flow rates, the distribution of flow across entry points may be relevant. For ground
water systems, impacts may be limited to individual wells, depending on their depth. For example,
immobile contaminants may affect shallow wells only.
The purpose of this chapter is to analyze the characteristics of entry points for ground and surface water
systems. Specifically, this chapter analyzes the following:
The numbers of entry points for ground and surface water systems l2
The distribution of flows among entry points
The spatial distribution of entry points
The depths of ground water wells.
12 Ownership distinctions (i.e., public and private) are not made for ground and surface water systems
because disaggregation at this level would severely limit the number of data points available for analysis of entry
points.
Geometries and Characteristics of 5-1 December, 2000
Public Water Systems
-------
Exhibits.!. Conceptual Diagram of Entry Point Configurations
H TREATMENT PLANT
DISTRIBUTION SYSTEM
A surface water PWS with two treatment plants
Wells supplying
water to treatment plant
Untreated wells
supplying water directly
to distribution system
Deep
Well
. Deep .
Well : : Shallow
: Well :
Deep
Well
\ \
\ \
DISTRIBUTION SYSTEM
A ground water PWS with a treatment plant and untreated wells B
5.1 Data Cleaning
Data for entry points are reported in Questions 18 and 20 of the CWSS (Appendix C). Each row in
Question 18 represents a treatment facility. Question 18 of the CWSS questionnaire reports latitudes and
Geometries and Characteristics of
Public Water Systems
5-2
December, 2000
-------
longitudes for each treatment plant, number of wells treated (if it is a ground water plant), range of depth,
flows, and treatments provided. Question 20 provides information for each well or surface water intake
not receiving treatment. Question 20 reports the aquifer or surface water source name and type (ground
or surface), location of each well and its depth, and flow.
For ground water systems, all of the wells reported in Question 18 are connected to treatment plants,
while the wells reported in Question 20 are connected directly to the system's distribution network. For
surface water systems, Question 18 reports intakes connected through treatment plants, while Question
20 reports untreated intakes connected directly to the distribution network. Thus, the sum of the rows in
Questions 18 and 20 corresponds to the number of entry points for each system.
Questions 18 and 20, which were not validation fields in the CWSS, were completed less frequently than
the validated fields (e.g., Question 4). Any system that did not report well and/or flow information for an
entry point in Questions 18 and 20 was eliminated. Purchased water systems also were eliminated from
the data set, since their configurations are not representative of those of stand-alone systems. The data
cleaning process resulted in two data sets comprising 840 ground water systems with 2,249 entry points
based on Questions 18 and 20 and 376 surface water systems with 476 entry points based on Questions
18 and 20. These data sets were used to analyze entry point characteristics.
5.2 Number of Entry Points
For each of the eight CWSS population size categories described in Chapter 3, Exhibit 5.1 shows the
frequency with which the sampled ground and surface water systems have multiple entry points. The
data in Exhibit 5.1 generally follow the expected trend. That is, the smaller systems tend to have a single
entry point and larger systems tend to have multiple entry points. Often, smaller systems can meet the
demand with a single source.
For ground water systems, Exhibit 5.1 indicates that the majority of systems sampled in the two smallest
population categories have a single entry point. About one-third of the systems in the 501 to 1,000
category, which are considered small systems, have more than one entry point. In the next two
population categories, the percentages of systems with more than one entry point are 42 percent and 54
percent, respectively. The use of multiple entry points increases as systems get larger. The reliance on
multiple entry points by a significant number of small systems is relevant for regulatory impact analyses
because of issues discussed above (e.g., situations in which contaminant impacts are entry point specific,
rather than system wide).
The pattern for surface water systems is different. Multiple entry points become an issue for systems
serving more than 3,300 persons. Furthermore, with respect to the number of entry points, surface
systems did not report as great a variety of configurations as ground water systems. Even for the large
population categories, the majority of surface water systems reported one or two entry points, with a
maximum of six reported. By comparison, groundwater systems reported values ranging from 1 to more
than 30 entry points. To further characterize the number of entry points for systems of various
population sizes, EPA performed additional statistical analyses as presented below.
Geometries and Characteristics of 5-3 December, 2000
Public Water Systems
-------
Exhibit 5.2. Frequency of Multiple Entry Points
Population
Category
Less than 100
101 to 500
501 to 1,000
1,001 to 3,300
3,301 to 10,000
10,001 to 50,000
50,001 to 100,000
100,001 to 1,000,000
Ground Water Systems
Percent
with One
Entry
Point
86.9%
80.5%
66.9%
58.1%
46.4%
33.0%
26.5%
24.2%
Percent with
Multiple
Entry Points
13.1%
19.5%
33.1%
41.9%
53.6%
67.0%
73.5%
75.8%
Maximum
Number
of Entry
Points
4
11
4
13
23
18
37
31
Surface Water Systems
Percent
with One
; Entry
: Point
100.0%
97.7%
100.0%
98.1%
89.8%
91.2%
59.2%
45.2%
Percent with \
Multiple
Entry Points
0.0%
2.3%
0.0%
1.9%
10.2%
8.8%
40.8%
54.8%
Maximum
Number
of Entry
Points
1
2
1
2
4
2
6
4
Ground water systems even small systems, reported a wide range of configurations with respect to
number of entry points. Mean and percentile values for the number of entry points were calculated for
ground water systems in each population category. Because the number of systems samples in each
population category was relatively small (compared to those in previous chaptersO, these statistics were
generated using a computer-intensive statistical procedure13. Using the bootstrap estimates, Exhibit 5.2
characterizes the mean number of entry-points and the percentile distribution of the number of entry
points for ground water systems in each population category. The percentile data in Exhibit 5.2 are the
"typical" (i.e., bootstrap mean) number of entry points for systems in the xth percentile. For example,
these data may be interpreted as follows: The typical number of entry points for systems in the 75th
percentile of all systems serving 100 to 500 people is one. Appendix F presents a more detailed
breakdown of the bootstrapping results.
Because surface water systems did not report as great variation in the number of entry points, similar
detailed characterization was not necessary. As discussed above, even for the large population
categories, the majority of surface water systems reported one or two entry points. Recently collected
data from Information Collection Request for large surface water systems supports this estimate.
b This procedure, known as "Bootstrapping", allows statistical estimates to be generated from smaller
sample sizes with nonnormal distributions without the need for extensive assumptions. The bootstrap method draws
a large number of random samples (in this case 10,000) with replacements and calculates the statistics of interest for
each sample. Item nonresponse factors and adjusted weights, as discussed in previous chapters, were used in the
bootstrap analysis.
Geometries and Characteristics of
Public Water Systems
5-4
December, 2000
-------
Exhibit 5.3. Percentile Distribution of Number of Entry Points for Ground Water Systems
Popnlation
Category
Less than 100
101 to 500
501 to 1,000
1,001 to 3,300
3,301 to 10,000
10,001 to 50,000
50,001 to 100,000
100,000 to 1,000,000
NiiW ber of Entry Paints*
Mte»n
1
1
2
2
2
4
6
9
tOth
Percentile
1
1
1
1
1
1
I
1
;2Sth
Perceatiie
1
1
1
1
1
1
1
1
50th
Percentile
1
l
l
1
2
3
4
5
75th
P«rC*nfite
1
1
2
2
3
5
8
15
90th
PercenfiJe
2
2
3
4
4
8
17
24
* Bootstrap value, rounded to the nearest integer.
5.3 Distribution of Flow Among Entry Points
For systems with multiple entry points, this section evaluates the distribution of the total system flow
among the entry points. The purpose of this evaluation was to determine if flows are distributed evenly
for systems with multiple entry points or if one of the entry points accounts for a majority of the system's
total flow.
The distribution of flow among entry points can be significant for regulatory analysis purposes. For
example, in a case where a single entry point is affected for a system having three entry points,
compliance costs would be estimated based on flow for the affected entry point. The costs would be
greater if the affected entry point accounted for a majority of the flow, as opposed to an equal share (one-
third) of the flow.
Systems reporting multiple entry points were examined further for the distribution of flow across entry
points. The ratios of entry point flow may differ during peak production, however the CWSS does not
provide data on peak flow from individual water sources. Therefore for each system, the percentage of
total flow accounted for by each entry point was calculated by comparing the entry point's average daily
flow to the system's total average daily flow. Entry points were ordered according to their percent
contribution to flow (i.e., first entry point = the largest, etc.). Systems were grouped by water source and
number of entry points.14 For each group, arithmetic mean percentages were estimated for each entry
point in the ordered set of entry points (i.e., mean percent for the largest entry points, mean percent for
14
Further subcategorization of systems by population category resulted in too few samples in each group to
generate statistics with a high degree of confidence. Examination of the available data did not support the
conclusion that there are significant differences between population categories in distribution of flow across entry
points. Thus, systems were grouped across all population categories.
Geometries and Characteristics of
Public Water Systems
December, 2000
-------
the second largest entry points, etc.). Exhibits 5.3 and 5.4 report the mean distribution of flow across
entry points for ground and surface water systems, respectively.
Exhibit 5.4. Distribution of Flow by Entry Point (Ground Water Systems)
Number df
Entry;
Points (EP)
2
3
4
5
6
Number of
Sample
Systems
136
67
42
25
12
Percent of Total Flow at...
Largest
EP
64.7
50.5
40.5
36.7
42.3
2nd
Largest
EP
35.3
30.4
26.7
24.4
18.1
3rd ;
Largest
EP
-
19.1
19.0
16.0
15.4
4*h
Largest
EP
-
-
13.8
13.0
10.1
5th
Largest
EP
-
-
-
9.9
8.3
$th
Largest
EP
-
-
-
-
5.9
Data are shown only for systems having up to 5 entry points. Similar data was generated for systems with up to
37 entry points and is presented in Appendix G.
Exhibit 5.5. Distribution of Flow by Entry Point (Surface Water Systems)
Ntttt&feT
-------
Exhibit 5.4 shows similar data for surface water systems. Flow for surface water systems is distributed in
almost the same proportions as ground water systems for a comparable number of entry points. For
example, for two entry points, the split for surface and ground water systems are about the same for the
largest entry points. For six entry points, the split is comparable. Unlike ground water systems, there are
fewer data points (sample systems) and the maximum reported number of entry points is six (compared
with up to 37 for ground water systems).
5.4 Spatial Distribution of Entry Points
As discussed in the introduction, when contamination is localized, only a single entry point in a system
may be affected. Which entry points are affected by the presence of contamination in the environment
depends on their proximity to the contaminated area. In some cases, the number of entry points affected,
and therefore compliance costs, may depend on the distance between entry points. In addition, systems
facing localized contamination may have the option of shutting down affected sources and drawing water
from unaffected (hydrologically separate) entry points. Provided the distant entry points have sufficient
capacity and transmission costs are reasonable, this practice can serve as a less costly alternative to
installing treatment.
Questions 18 and 20 provide latitude and longitude data for the location of each entry point. For systems
with two entry points, a preliminary analysis was conducted by entering these data into a Geographic
Information System (GIS), which was then used to calculate the distance between entry points. A limited
number of data points were available for this analysis. Twenty-six surface water and 56 ground water
systems with two entry points provided latitude and longitude data.16
Calculated distances between entry points ranged from 0 meters to 67 kilometers for ground water
systems and from about 2 kilometers to 106 kilometers for surface water systems. The upper bounds of
both of these ranges appear anomalous and may be the result of inaccurate latitude and longitude data.
The majority of distances were between a few hundred meters and 9 kilometers for ground water systems
and between 2 kilometers and 30 kilometers for surface water systems. There were too few data points to
examine differences between population categories.
The GIS data were also used to analyze the nine surface water systems with more than two entry points
that provided latitude and longitude data. For each of these systems, the entry point coordinates were
used to define a polygon and the distance from each entry point to the polygon's centroid was calculated.
This analysis yielded a range of calculated distances similar to that above, but also revealed a number of
anomalous data points. (For example, based on the entry point coordinates, one system appeared to span
an area nearly the size of Pennsylvania.) Due to data and other limitations about the accuracy of the
latitude and longitude data, a similar analysis was not conducted for the 68 ground water systems with
more than two entry points that provided latitude and longitude data.
Thus, preliminary analyses suggest a wide variation in spatial configurations for entry points. They
confirm that distant entry points can exist for both ground and surface water systems; however, the
available data are too limited to accurately quantify differences between systems. Furthermore,
regulatory impacts are likely to be highly system-specific, depending not only on distance between entry
points, but also on the nature of the contamination and available supply alternatives.
16 Latitude and longitude coordinates for a few entry points appeared to be transposed; these data points
were corrected prior to analysis. Entry points with coordinates outside the United States were deleted.
Geometries and Characteristics of 5-7 December, 2000
Public Water Systems
-------
5.5 Well Depths
Referring back to Exhibit 5.1, ground water systems with multiple wells can draw water from various
depths. In some cases, an immobile ground water contaminant may affect only shallow wells. Thus, the
number of entry points affected, and, therefore, compliance costs, may depend on well depth. In
addition, when immobile contaminants are present, systems may have the option of shutting down
shallow wells and drawing water from deeper, unaffected wells. Provided the deep wells have sufficient
capacity, this practice can serve as a less costly alternative to installing treatment.
Questions 18 and 20 provide well depth information. In Question 20, depths are reported for individual
untreated wells. In Question 18, the system may report multiple wells connected to each treatment
facility. Where multiple wells are connected to a treatment facility, the depth data are in the form of a
range of depths for the wells connected, as opposed to a depth for each individual well. Three different
approaches were used to convert the Question 18 ranges to point estimates so that these data could be
examined along with the Question 20 individual depths. The three approaches, respectively, used the
minimum of each range, the maximum of each range, and the midpoint of each range as point estimates
of depth. When the Question 18 and Question 20 data were combined using these approaches, a
relatively large number of data points were available for analysis (2,249 entry points for 840 ground
water systems).
Differences in well depths among population categories were examined by calculating the mean depth of
all entry points in each category using each of the three approaches to estimating depth. Exhibit 5.5
shows the results of this analysis. Based on the category means, entry points in the smallest population
category appear to have much shallower minimum, midpoint, and maximum depths. Systems in the
largest population category appear to have deeper well depths. Larger systems also may have greater
ability to switch to deeper wells in response to contamination.
The presence of very deep wells in some systems, however, suggests that, faced with immobile
contaminants affecting shallow wells, these systems may be able to switch to deep, uncontaminated
wells. Examining the range of variation of well depths reported by individual systems (i.e., comparing
the minimum well depth reported by a system to the maximum well depth reported by the same system)
supports this hypothesis. Exhibit 5.6 shows, by population category, the frequency with which systems
reported various ranges of well depths. This analysis shows that the majority of systems in the smaller
population categories do not have large depth ranges. That is, wells in most of these systems tend to be
at similar depths. Therefore, these systems are not likely (based on depth alone) to have the option of
switching wells to avoid contaminants.
Geometries and Characteristics of 5-8 December, 2000
Public Water Systems
-------
Exhibit 5.6. Mean Entry Point Depths by Population Category
Population
Category
Less than 100
101 to 500
501 to 1,000
1,001 to 3,300
3,301 to 10,000
10,001 to 50,000
50,001 to 100,000
100,000 to 1,000,000
Number of
Entry Points
116
220
170
256
335
452
387
313
Number of
Systems ;
99
159
115
136
140
109
49
33
Mean of Entry Point,.
(depth in feet)
Minimum
Depth
111
272
254
378
386
351
275
459
Midpoint
Depth
120
300
281
408
410
382
309
589
Maximum
Depth
128
328
307
438
435
414
344
718
Exhibit 5.7. Frequency of Variation of Well Depths by Population Category
Population
Category
Less than 100
101 to 500
501 to 1,000
1,001 to 3,300
3,301 to 10,000
10,001 to 50,000
50,001 to 100,000
100,000 to 1,000,000
Percent of Systems with Variation of Well Depths (in feet)...
Less than
50
75.0%
64.6%
65.4%
50.0%
37.8%
28.2%
17.0%
16.1%
58 to
100
15.0%
12.5%
8.7%
9.5%
14.2%
9.7%
8.5%
12.9%
100 to
206
10.0%
13.2%
10.6%
15.1%
20.5%
14.6%
14.9%
12.9%
200 to
500
0.0%
6.9%
10.6%
17.5%
18.9%
27.2%
27.7%
29.0%
500 to
1,000
0.0%
2.8%
5.8%
5.6%
5.5%
14.6%
21.3%
9.7%
Greater
than 1,000
0.0%
0.0%
0.0%
2.4%
3.1%
5.8%
10.6%
19.4%
Geometries and Characteristics of
Public Water Svstems
5-9
December, 2000
-------
This page intentionally left blank
-------
6: Analysis of Treatment-in-Place
This chapter presents the approach and results of an analysis of treatment-in-place (existing treatment
technologies) at existing CWSs. Section 6.1 presents the rationale for the analysis. Section 6.2 provides
a discussion of the data used for the analysis, including the data cleaning performed. A summary of the
results of the analysis is provided in Section 6.3.
6.1 Rationale for Treatment-in-Place Analysis
To estimate potential costs incurred by public water systems as a result of future revisions to drinking
water regulations, EPA needs to know the types of treatment currently in place at existing water systems.
Having that information allows EPA to more accurately estimate costs associated with treatment plant
modifications and upgrades that would be necessary for compliance.
To model treatment-in-place, EPA used responses to two questions in the CWSS, namely, Questions 18
and 20. Question 18 requested basic information for the treatment facilities within each CWS.
Specifically, responses to Question 18 identify the source of raw water treated (i.e., ground or surface),
daily maximum and average flows, and the type of treatment provided at each plant within the CWS.
Question 20 identifies untreated wells and surface water intakes, along with flows from these entry
points.
As presented in the CWSS questionnaire, Question 18 requested that respondents identify treatment
technologies using specific water treatment codes (as presented here in Exhibit 6.1). Many existing
technologies, such as ozone and chlorine dioxide were in limited use at the time of the survey.
Consequently, there would be a high degree of uncertainty in estimates based on only a handful of
responses. To simplify the treatment-in-place model and reduce uncertainties in the data, subcategories of
treatment technologies that address common treatment issues (i.e., water treatment codes) were grouped
into classes considered significant for future EPA rulemakings. For example, a facility using chlorine and
another using chlorine dioxide are both identified as performing "disinfection." Similarly, a facility that
performs coagulation and flocculation using aluminum salts is grouped together with a facility that uses
polymers in the coagulation and flocculation process. The combined headings of these groupings are
presented in Exhibit 6.2.
A number of treatment plants identified a treatment technology as "other." For these facilities, EPA
reviewed CWSS questionnaire submissions to determine whether these data represented any major
treatment category that should be characterized in the model system evaluation. This analysis did not
identify any additional treatment categories that should be included in the analysis.
Geometries and Characteristics of 6-1 December, 2000
Public Water Systems
-------
Exhibit 6.1. CWSS Water Treatment Codes (Question 18)
Code
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Treatment
Raw water storage
Presedimentation
Aeration
PRE-D1S1NFECTION/OXIDATION
Chlorine
Chlorine dioxide
Chlorammes
Ozone
Potassium Permanganate
Pre-dismfection/oxidation combinations
Lime/Soda ash softening
Recarbonation with carbon dioxide
IRON AND MANGANESE REMOVAL:
Green sand filtration
Chemical oxidation filtration
Aeration filtration
FLOCCULATION/COAGULAT1ON
Aluminum salts
Iron salts
pH adjustment
Activated silica
Clays
Polymers
Other flocculation/coagulation
Flocculation/coagulation combinations
FILTRATION:
Slow sand
Rapid sand
Dual/Multi-media
Diatomaceous earth
Code
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Treatment
Reverse osmosis
Pressure filtration
Other filtration
Filtration combinations
ORGANICS REMOVAL:
GAC adsorption post contactors
GAC adsorption filter adsorbers
PAC addition
Ion exchange
Air stripping
Orgamcs removal combinations
POST-DISINFECTION:
Chlorine
Chlorine dioxide
Chloramines
Post-disinfection combinations
Fluondation
Hypochlorination
CORROSION CONTROL.
pH adjustment
Alkalinity adjustment
Corrosion inhibitors
Corrosion control
OTHER TREATMENTS NOT ELSEWHERE
CLASSIFED
Other treatment
Exhibit 6.2. Treatment Code Groups
Disinfection:
Aeration:
Oxidation (Fe/Mn):
Ion Exchange:
Reverse Osmosis:
GAC:
The numbers refer to the
04, 05, 06, 07, 08, 09,
37, 38, 39, 40, 42
03, 14,35
12, 13
34
27
31,32
treatment category numbers in
PAC:
Filtration:
Coagulation/Flocculation:
Lime/Soda Ash Softening:
Recarbonation:
33
23,24,25,
30
15, 16, 19,
10
11
26, 28, 29,
20,21,22
the CWSS as presented in Exhibit 6.1.
Geometries and Characteristics of
Public Water Systems
6-2
December, 2000
-------
6.2 Identification of Core Data Set
This analysis used the same data set as the entry point analysis in Chapter 5 (i.e., purchased water
systems and nonrespondents to Questions 18 and 20 were not included). To address facilities that failed
to adequately respond to Questions 18 and 20, an item-level nonresponse weighting factor, as discussed
in Chapter 4, was incorporated into the analysis. The incorporation of item-level nonresponse weighting
allows the results for the systems that adequately responded to be representative of the full survey
population. This approach assumes that non-respondents are not statistically different from respondents
in terms of treatments implemented. Review of responses to core validated questions, financial, and
other elements of the questionnaire suggest this is reasonable.17
Additionally, a facility with multiple "hits" in one category would count as one treatment technology
only. For example, a facility reporting water treatment codes of 23 (slow sand filtration), 24 (rapid sand
filtration), and 29 (other filtration) counts as only one facility using filtration in the tabulation of
treatments-in-place.
6.3 Summary of Results
To be consistent with the models of population and flow, the treatment-in-place analysis results are
presented by source water type. The results are also presented by population size category. As discussed
above, the analysis used sample weights adjusted for item-level non-response.
Exhibits 6.3 and 6.4 (for ground water and surface water systems, respectively) present the frequency, by
population category, with which systems have treatments in each of the categories defined for this
analysis. As such, any one treatment facility within a system that reported a specific treatment classifies
the system as having that treatment technology. Systems with two treatment plants at one system or
systems with two treatment technologies within the same treatment plant (e.g., pre- and post-disinfection)
are counted only once.
Exhibit 6.5 presents the frequency, by population category, with which systems have no treatment, one
type of treatment, or multiple methods of treatment. Note that, in this table, both of the following cases
would be counted as a system with two methods of treatment: (1) a system with two treatment plants,
each providing a different type of treatment for a different entry point, and (2) a system with two
methods of treatment for a single entry point.
17 Furthermore, EPA's analysis of information from the AWWA Waterstat database for systems serving
more than 50,000 people resulted in treatment frequencies similar to those here. The similarity of these results
confirms that the sample used here is unbiased for system size categories where a comparison is possible.
Geometries and Characteristics of 6-3 December, 2000
Public Water Systems
-------
Exhibit 6.3. Percent of Ground Water Systems with Treatment
Treatment
Category
Disinfection
Aeration
Oxidation
Ion Exchange
Reverse Osmosis
GAC
PAC
Filtration
Coagulation/
Flocculation
Lime/Soda Ash
Softening
Recarbonation
Population Category
Less :
than 100
52.8%
1.5%
3.2%
0.7%
0%
0%
0%
11.8%
1.5%
2.1%
0%
ioi
to
500
77.9%
6.3%
6.6%
1.6%
1.2%
0.5%
0%
8.0%
5.4%
3.7%
0.5%
501
to
1,000
84.0%
17.1%
9.4%
3.8%
0%
0%
0%
15.9%
4.2%
4.1%
0%
1,001
to
3,388
79.7%
19.9%
4.2%
1 .9%
0.9%
0.4%
0%
14.9%
3.4%
5.2%
1.1%
3,301
to
10,800
86.8%
29.7%
10.9%
4.6%
1.2%
0%
0.2%
29.5%
8.1%
7.0%
3.0%
JOJ80I
to
50,000 :
96.5%
33.0%
9.3%
3.3%
0.7%
6.7%
0.3%
29.6%
15.1%
12.2%
6.1%
58,001
to
100,000
86.3%
49.1%
18.6%
1.2%
1.2%
7.5%
0%
50.3%
24.2%
17.4%
7.5%
More
than
100,000
96.4%
44.1%
5.4%
0%
0%
9.0%
1.8%
51.4%
25.2%
32.4%
10.8%
Note: Percentages shown are weighted for item-level nonresponse.
Exhibit 6.4. Percent of Surface Water Systems with Treatment
Treatment
Category
Disinfection
Aeration
Oxidation
Ion Exchange
Reverse Osmosis
GAC
PAC
Filtration
Coagulation/
Flocculation
Lime/Soda Ash
Softening
Recarbonation
Population Category
Less
than 100
92.8%
0%
0%
0%
0%
3.9%
0%
78.5%
27.5%
3.9%
0%
101
to
500
94.1%
0%
2.0%
0%
0%
4.3%
2.0%
71.2%
52.6%
8.1%
0%
501
to
1,000
100.0%
1.4%
7.2%
0%
0%
1 .4%
3.0%
79.3%
70.2%
20.5%
0%
1,001
to
3,300
100.0%
5.5%
5.8%
0%
0%
2.3%
4.6%
81.7%
78.5%
17.5%
0%
3,301
to
10,000
96.0%
8.5%
7.7%
0%
0%
4.7%
18.6%
86.5%
95.4%
10.8%
0%
10,001
to
50,000
98.0%
3.5%
10.5%
0%
0%
10.2%
24.6%
96.3%
94.5%
6.9%
0%
50,001
to
100,000
100.0%
10.3%
5.7%
0%
0%
14.9%
34.2%
88.0%
93.7%
5.7%
1.1%
M<»re
than
100,000
100.0%
14.3%
4.6%
0%
0%
11.2%
45.9%
93.4%
99.5%
5.1%
5.1%
Note: Percentages shown are weighted for item-level nonresponse.
Geometries and Characteristics of
Public Water Svstems
6-4
December, 2000
-------
Exhibit 6.5. Percent of Systems with Multiple Categories of Treatment
Number «f
Treatment
Categories '
Population Category .
Less
than 106
161
to
S06
SOI
to
1,006
1,001
to
3300
3301
to
10,060
10,00!
to
SMX>0
50,001
to ;
100,006
Mor*
thau
106,000
Ground Water Systems
No treatment
1 treatment
2 treatments
3 treatments
4 treatments
5 treatments
6 treatments
42.5%
43.9%
12.3%
0.4%
1.0%
0%
0%
19.0%
63.3%
9.4%
5.8%
0.6%
1.4%
0.6%
16.0%
57.4%
7.0%
12.3%
6.6%
0.8%
0%
18.4%
55.8%
10.7%
9.4%
3.2%
1.4%
1.1%
13.1%
45.1%
8.3%
19.6%
9.4%
4.2%
0.3%
0.9%
52.6%
12.9%
11.8%
13.0%
6.2%
2.6%
11.2%
22.4%
14.9%
11.2%
25.5%
13.7%
1.2%
0%
28.8%
18.0%
18.9%
19.8%
10.8%
3.6%
Surface Water Systems
No treatment
1 treatment
2 treatments
3 treatments
4 treatments
5 treatments
6 treatments
7.2%
14.3%
43.3%
35.2%
0%
0%
0%
5.9%
23.0%
1 7.5%
40.3%
1 1 .4%
2.0%
0%
0%
15.9%
18.9%
31.7%
33.6%
0%
0%
0%
7.5%
14.0%
60.8%
10.5%
7.3%
0%
0%
1.9%
12.1%
49.6%
28.7%
7.7%
0%
2.0%
0.7%
60.6%
22.6%
13.6%
13.6%
0.5%
0%
4.0%
2.3%
43.4%
40.0%
6.9%
3.4%
0%
0.5%
4.6%
35.7%
40.3%
12.2%
6.6%
Note: Percentages shown are weighted for item-level nonresponse.
Exhibit 6.6 presents similar data for groundwater systems at the individual entry point level. That is, the
table describes the frequency with which an individual entry point is untreated or has a treatment plant
providing one or more categories of treatment. Because the majority of surface water systems have only
one entry point, percentages at the entry point level are nearly the same as those presented in Exhibit 6.5.
Therefore, Exhibit 6.6 does not include data for surface water systems.
The data shown in Exhibits 6.3 through 6.6 suggest some patterns. Large systems appear to use more
treatment methods and more advanced treatment methods than small systems. Also, surface water
systems seem more likely to be treated than ground water systems. Additional evaluation of these data
are necessary to incorporate these findings into analytical models.
Geometries and Characteristics of
Public Water Systems
6-5
December, 2000
-------
Exhibit 6.6. Percent of Entry Points in Ground Water Systems with Multiple Methods of Treatment
Number of
Treatement
Categories
No treatment
1 treatment
2 treatments
3 treatments
4 treatments
5 treatments
6 treatments
Population Category :
Less
thai*
im
50.4%
38.0%
10.4%
0.3%
0.8%
0%
0%
101
to
§00
24.8%
61.3%
7.8%
4.3%
0.4%
1.0%
0.4%
501
tO :
1,000
24.9%
57.2%
4.5%
8.6%
4.2%
0.5%
0%
1,001
to
3,360
26.9%
56.2%
7.2%
6.6%
1.8%
0.8%
0.6%
3^01
to
10,000
26.4%
50.2%
7.6%
9.7%
4.1%
1 .9%
0.1%
10,001
to
SG»OOO
7.0%
70.1%
11.4%
5.7%
3.9%
1.2%
0%
50,001
to
100,000
39.6%
38.0%
10.5%
4.5%
4.4%
2.9%
0.2%
More j
than :
100,006:
17.4%
52.1%
18.6%
6.4%
1.6%
3.9%
0%
Note: Percentages shown are weighted for item-level nonresponse.
Geometries and Characteristics of
Public Water Systems
6-6
December, 2000
-------
7: Non-Community Water Systems
As discussed in Chapter 2, EPA's Safe Drinking Water Information System (SDWIS) database includes
information on non-community water systems (NCWSs), as well as the community water systems
(CWSs) that are the focus of previous chapters of this report. According to SDWIS, CWSs serve more
than 90 percent of the total public water system population. Therefore, in previous drinking water
regulations, EPA has used CWS flows to model NCWS flows. NCWS flows, however, are generally
substantially lower than typical CWS flows. Furthermore, while NCWSs make up a small percentage of
the population served, these systems actually comprise two-thirds of the total number of public water
systems regulated under the SDWA.
The goal of this chapter is to provide an improved characterization of the NCWS universe. This
quantitative discussion of NCWS model systems is based on data extracted from SDWIS in November
1997. Data limitations constrain the analysis of NCWSs. NCWSs are modeled separately from CWS
because of inherent differences between the two types of systems and the lack of national NCWS survey
data (i.e., comparable to the CWSS). The modeling approach presented herein for NCWSs uses SDWIS
data and relies on various references for typical water consumption patterns for various types of
NCWSs. Though not addressed here, system differences attributable to regional setting, variations in
exposure routes related to system type, system residence times, water storage capabilities, and existing
treatment profiles may be considered in subsequent efforts.
7.1 Overview of Non-Community Water System Population
SDWIS identifies 115,948 NCWSs in the United States (Source: 1997 SDWIS frozen database18). These
NCWSs represent more than 67 percent of the total number of public water systems. Water sources for
these systems include ground water, surface water (including ground water under the direct influence of
surface water), and purchased water. Exhibit 7.1 presents a breakdown by water source of the number of
NCWSs as reported in SDWIS. Based on SDWIS data, NCWSs serve more than 25-million people.
NCWSs that serve less than 10,000 persons per system serve more than 15-million people of this total,
with about 97 percent using ground water. As discussed below, systems that reported serving more than
10,000 people are treated separately in this analysis (see Appendix H).
Exhibit 7.1. Non-Community Water Systems by Water Source
Water Source
Ground Water
Surface Water
Purchased Water
Other
Total
Number of Systems
112,214
2,119
1,613
2
115,948
18,
' The number of NCWSs reported in this chapter differ from those reported in Chapter 2. which reflect 1998 SDWIS
data. Updated 1998 estimates \vere not available for use in the Chapter 7 analysis.
Geometries and Characteristics of
Public Water Systems
7-1
December, 2000
-------
Exhibit 7.2 provides summary statistics for all NCWSs in SDWIS, sorted by water source and mean
population served. Approximately 95 percent of the NCWSs serve less than 500 people, with about only
0.3 percent of NCWSs serving more than 3,300 persons a day. The remainder of this evaluation does not
separate systems into population categories except to differentiate large NCWSs serving more than
10,000 persons a day. According to SDWIS, these large systems account for less than 0.1 percent of the
total number of systems, but serve nearly one third of the total NCWS population (these systems are
identified in Appendix H). Large NCWSs warrant separate evaluation because SDWIS data for some of
these systems may be in error (e.g., systems reporting yearly population served rather than daily).
The remainder of this chapter focuses on the 112,214 ground water systems and 2,109 surface water
systems serving fewer than 10,000 persons. Purchased water systems are excluded from discussion as
these are effectively either CWS or NCWS customers.
Exhibit 7.2. Non-Community Water Systems by Population Range
Population Range
Less than 100
101-500
501-1,000
1,001-3,300
3,301-10,000
10,001-50,000
50,001-100,000
Greater than 100,000
Subtotal
Ground Water Systems
81,483
25,411
3,747
1,276
201
74
12
10
112,214
(112,118)*
Surface Water Systems
1,145
654
154
114
42
7
3
0
2,119
(2,109)*
Total Systems
82,628
26,065
3,901
1,390
243
81
15
10
114,333
(114,227)*
* Numbers in parentheses indicate values excluding all systems serving more than 10.000 persons.
Note: Totals shown exclude purchased water systems, systems using ground water under the direct influence of surface water,
and systems reporting source water type as "other."
A key characteristic for NCWSs is the distinction between transient and non-transient systems. The
distinction is an important one, since regulations for chronic contaminants are not applied to transient
water systems. A typical non-transient system may supply drinking water to employees (e.g.,
manufacturing facilities) or extended-stay residents (e.g., nursing homes), while a typical transient
system may supply drinking water to service areas with short term and variable (i.e., transient)
populations (e.g., amusement parks and restaurants). The SDWIS inventory on which this chapter is
based initially categorized 95,858 systems as transient and 20,090 as non-transient (94,389 transient and
19,766 non-transient excluding purchased water systems, ground water systems under the direct
influence of surface water, systems with source water specified as "other," and systems serving more
than 10,000 people).
The distinction between transient and non-transient systems can be unclear. For instance, SDWIS
classifies churches as both transient and non-transient systems. While the population served can vary
(i.e., it varies for certain days of the week and throughout the year), most churches serve the same
Geometries and Characteristics of
Public Water Systems
7-2
December, 2000
-------
individuals on a year-round basis, suggesting a classification as non-transient. However, well over 90
percent are classified as transient in SDWIS. Likewise, 10 percent of schools and 20 percent of daycare
centers are classified as transient systems.
Given the "grey areas" between transient and non-transient systems, individual systems might be
miscategorized in SDWIS. SDWIS ideally reflects the basis for distinguishing applicability of
regulations; however, controversy over classifications could arise with regulations that require significant
capital expenditures for compliance. Accordingly, this chapter presents the breakdown of transient
versus non-transient systems based on the initial SDWIS inventory (Section 7.2) as well as a breakdown
based on best professional judgment (Section 7.4). The analyst should carefully consider uses to be
made of these data in determining which set to apply.
7.2 Service Area Classification and Population Served
SDWIS characterizes NCWSs by type of service and population served, among other variables.19 In
SDWIS, each NCWS is characterized by up to six service area type codes (see Exhibit 7.3). To develop
a simple model of NCWSs, one service area type was assigned to each system (or multi-use systems were
grouped, where appropriate). For purposes of the evaluation, these service area types are split into two
categories: "specific" (those that narrowly define a population served, such as "daycare center") and
"general" (those that lack a usable designation, such as "other area").
A review of SDWIS data found that approximately two-thirds of the NCWSs are codified as general
service areas. A brief review of those systems suggests they are quite different from those with specific
designations. To better define this segment of the NCWS population, an in depth evaluation of these
facilities was performed. This multistep evaluation is described in Section 7.2.1.
Exhibit 7.3. SDWIS Service Area Classifications
Specific Service Area Classifications: General Service Area Classifications:
- daycare center - industrial/agricultural
- highway rest area - institution
- hotel/motel - no service area
- interstate carrier - other area
- medical facility - other non-transient area
- mobile home park - other residential area
- restaurant - other transient area
- school - recreation area
- service station - residential area
- summer camp
- wholesaler
19
Characterizing the service area is important for NCWSs because service area directly impacts the exposure scenario.
For instance, assumptions of daily water consumption for residents is inapplicable to restaurant customers.
Geometries and Characteristics of 7-3 December, 2000
Public Water Systems
-------
7.2.1 General Approach
In general, modeling NCWSs requires categorizing all systems into specific service areas. Systems with
one specific service area were simply categorized as presented in SDWIS. Systems with a general
service area or more than one service area required more detailed examination. Specifically, systems
with at lest one general service area were categorized into four distinct service area categories, depending
on the service area types reported for each individual NCWS. A summary of the categorization process
is presented in Exhibit 7.4.
Exhibit 7.4. Approach to NCWS Service Area Categorization
Category
Number
1
2
3
4
Number of
Specific
Service Areas i
1
2 or more
0
0
Number of
General
Service Areas
any number
any number
1
2 or more
Final
Categorization Criteria
Categorized solely on specific service area as reported in
SDWIS (e.g., "day care center")
Given "Mixed known" category and subcategorized separately
according to the specific service area reported in SDWIS; e.g.,
"mixed known with daycare center"
Recategorized using best professional judgement by name of
system (process described in detail in Section 7.2.2)
Given a "Mixed unknown" category for further categorization
using best professional judgement by name of system (using the
same process as category 3)
Categories 1 and 2 were sorted and represented by a specific service area. Categories 3 and 4 represent
the body of NCWSs reporting only general service areas. Further analysis to sort these systems by
specific service area is summarized in the next section.
7.2.2 Characterization of General Service Area Classifications
To further characterize the makeup of the NCWSs reporting only general service areas, a random sample
of NCWSs was collected from each general service area classification. The sample size was based on
achieving a 95 percent probability that no service area representing more than 0.5 percent of the NCWS
population would be missed in the process. As depicted in Exhibit 7.5, a total of 1,152 NCWSs were
selected from the universe of 76,179 NCWSs identified with general service area types. The systems
were further subcategorized according to their initial categorization in SDWIS as "transient" or "non-
transient."
Geometries and Characteristics of
Public Water Systems
7-4
December, 2000
-------
Exhibit 7.5. Random Sampling of Non-Community Water Systems with General Service Areas
General Service Area
< Classification
Industrial/Agricultural
Institution
No Service Area
Other Area
Other Non-Transient Area
Other Residential Area
Other Transient Area
Recreation Area
Residential Area
Mixed Unknown
Total
Transtettf*
Sampling Frame
1,166
492
22,000
3,539
687
274
17,939
16,383
631
1,027
64,138
Sample Size
9
9
200
26
9
9
161
141
9
9
582
N&a-Traastent*
Sampling Frame
4,428
374
3,650
552
1,802
793
627
221
123
171
12,041
Sample Size
226
14
180
21
80
7
21
7
7
7
570
* As initially categorized in SDWIS. See Section 7.4 for further discussion of transient versus non-transient categorization.
Specific to any Delphi approach, the goal of this effort was to make forecasts that systematically use
insights and assessments of selected specialists. A service area type was assigned to each of the 1,152
general NCWSs using the consensus of best professional judgment from four reviewers based on the
system name as provided in SDWIS. A total of 59 service area types were identified as a result of this
process, including the 11 SDWIS specific types. A list of the 59 service area types is provided in Exhibit
7.6
The sampled NCWSs were scaled-up based on the ratio of the number of systems in the sampling frame
to the number of systems in the random sample. For example, facilities coded in the "transient" and
"other area" category were scaled up using a 3,539/26 scaling factor.
Geometries and Characteristics of
Public Water Systems
7-5
December, 2000
-------
Exhibit 7.6 Non-communitv Water System Service Areas
Existing SDWIS Specific Categories
Code Service Area Type ISIC Codel
(DC) Daycare Center [8351]
(HRA) Highway Rest Area [NO SIC]
(HM) Hotel/Motel [70] (includes rooming and boarding houses.
lodges, and resorts)
(1C) Interstate Carrier (includes truck stops, bus and railroad
terminals, airports, couriers, postal service)
(MF) Medical Facility [80]
(MHP) Mobile Home Park [6515]
(R) Restaurant [581]
(S) School [82] (includes colleges, vocational schools, dance
studios, and universities or places of higher learning)
(SS) Service Station [5541 and 75]
(SC) Summer Camp (include basketball camps, baseball
camps etc ) [7032]
(WPP) Water Wholesaler or Producer (include washetenas)
Categories Identified in Sampling of General Service Areas
Code Service Area Type ISIC Codel
(AG) Agricultural [01,02, and 07]
(AP) Air Park [4581]
(B) Bowling Centers [7933]
( C) Construction [ 15, 16, and 17]
(CH) Churches [866]
(CRV) Campground or RV Parks [7033]
(FD) Fire Departments [9224]
(FP) Federal Parks [9512]
(FS) Forest Service [9512]
(GC) Golf and Country Clubs [7992]
(L) Laundries, Including Industrial Laundries [721]
(LIB) Libraries [8231]
(LFL) Landfill [4953]
(M) Mining [10, 12. 13, and 14]
(MAMU) Amusement Parks (includes Fairgrounds and Water
Parks) [7996]
(MB) Military Bases [9711]
(MFCC) Industrial and Commercial Machinery and Computer
Equipment [35]
(MFCE) Electronic and Electrical Equipment and Components,
Except Computers [36]
(MFCH) Chemicals and Allied Products [28]
(MFF) Furniture and Fixtures [25]
(MFI) Miscellaneous Mfg Industries [39]
(MFLL) Leather and Leather Products [31 ]
(MFM) Fabricated Metal Products. Not Transportation. [34]
(MFO) Food and Kindred Products [20]
(MFP) Paper and Allied Products [26]
(MFPE) Petroleum Refining and Related Industries [29]
(MFPM) Primary Metal Industries [33]
(MFPR) Printing, Publishing, and Allied Industries [27]
(MFRU) Rubber and Miscellaneous Plastics Products [30]
(MFSC) Stone. Clay. Glass, and Concrete Products [32]
(MFT) Tobacco Products [21]
(MFTE) Transportation Equipment [37]
(MFTX) Textile Mill Products [22]
(MFTX) Apparel and Other Finished Products [23]
(MFW) Lumber and Wood Products, except furniture [24]
(MFWW) Measuring. Analyzing, and Controlling Instruments.
Photo. Medical and Optical Goods, Watches and Clocks
[38]
(MLC) Migrant Labor Camps [0761 ]
(MREC) Miscellaneous Recreation Services [799] (excluding
amusement parks)
(MU) Museums [84]
(NH) Nursing Homes [805]
(OP) Office Parks [6512]
(PRI) Prisons [9223]
(RCC) Racing, including track operation [7948]
(RET) Retailers (Non-food related) [53 and 55]
(RETF) Retailers (Grocery Stores. Fruit/Vegetable Markets, Meat
and Fish Markets, Dairy Products, Bakeries, etc.) [54]
(SP) State Parks [9512]
(UT) Non-Water Utilities (includes power plants, natural gas,
electric companies) [491,492]
(ZG) Zoological Gardens [84] (e.g.. arboretums)
Geometries and Characteristics of
Public Water Systems
7-6
December, 2000
-------
7.2.3 Results
Based on the categorization approach presented in Sections 7.2.1 and 7.2.2, combined estimates were
developed for the number of NCWSs, the average population served, and the total population served for
each identified service classification. Exhibit 7.7 presents these results. Exhibit 7.8 identifies the
estimated population served for each service classification (e.g., for restaurants, the average population
served is represented as the number of customers daily).
The totals derived after application of the categorization approach are slightly different than those
presented previously in Exhibit 7.2. These differences, however, are relatively insignificant (114,227
systems originally identified in SDWIS versus 114,726 systems after categorization, or a 0.44 percent
difference). These differences are a result of the approach used to characterize the systems with general
service classifications and subsequent rounding.20
This evaluation does not summarize NCWSs by population category since, as noted in Section 7.1, more
than 95 percent of all NCWSs serve fewer than 500 persons a day, with nearly 99 percent serving 1,000
persons or fewer. From a regulatory impact standpoint, additional stratifications would have little impact
on the accuracy of models. Rather, the diversity of ownership is the primary variable of interest.
A separate evaluation of large systems (i.e., those serving more than 10,000 persons) was performed to
identify the types of systems represented. These systems were categorized based on system name,
similar to the approach used for identifying the smaller systems. Specifically, each of the 106 systems
was assigned a service classification based on best professional judgment. A list of these largest
NCWSs, with the assigned code, is provided as Appendix H. Of the systems serving more than 10,000
persons, approximately two-thirds are State parks, with highway rest areas, miscellaneous amusement
parks, and campgrounds accounting for most of the rest. Many of the systems reporting a daily
population served of greater than 10,000 appear to be reporting errors. Many of the populations appear
to be monthly, yearly, or peak daily figures. For example, campgrounds reporting populations of 12,000
persons or more and highway rest areas serving over 60,000 people per day. Based on a best-
professional-judgment evaluation of the NCWSs reporting service populations greater than 10,000,
approximately two-thirds appear to be incorrectly recorded. As such, the percentage of systems serving
greater than 10,000 is likely even smaller than presented here.
For the same reasons, similar small differences will exist between the population served totals shown here and
similar figures derived directly from SDWIS (see Appendix A).
Geometries and Characteristics of 7-7 December, 2000
Public Water Systems
-------
*
o.
H
^
Q,
.«
i
•8
t
1/3
e
'•5
Q.
O
fi-
s
sU
tt>
u
•fa*
*•*
e
3
E
o
U
e
o
I-;
^-
£
X
U
S
£
4(4
C
£
jjf
E
§
^
•**
4>
•35
f
§
* 4?
||
S =
•S 0
**
J£ *£
II
JE*
C
si
o ^
H &
*
Sf®
i-s
•*¥ -o
B.
fe.
•P -{ft
J&I
1*
.
<5
s 2
£ 3
H a
£
e
{)£. «•.
C 5
1 »
s
U*
-«
= Si
Z;
Service Area Type
00
r-
g
0
_-
K
so
—
r-
o
00
m
fH
o
VI
o
0
(N
| Daycare Centers
£
rsi
fN
S
(N
__
O
SO
O
J
^
Ts
SO
S
S
—
m
~~
VI
ca
(U
CO
ou
if
rs
ID
O
o
r-
s
00
s
^
s
ID
r*"i
m
?
oo
r-
rs
m
oo
| Hotels/Motels
^
-
(N
r-i
£
fs
00
^
m
o
oo
s
rs
m
—
so
rs
ON
OO
O
^
r--
S
*D
O
rs
™~
t/3
ra
C-
E
o
^
o
oo
0
^~
rs
S
o
oo
ID
rs
00
rs
ID
S
O
1 '
"s-t
ON
ON
ID
rs
oo
rs
Tj-
ID
| Restaurants
rs
ID
—
S
P;
ON
ID
—
!2
o
oo
•D
^t
00
1/1
£
o
s
^
ON
| Schools
rs
oo
r*l
O
r-
r*"i
£
—
SO
Os
SO
—
00
sO
V,
-I2
| Campgrounds/RV Pa
£
SO
S
rs
00
o
Tf
oo
ON
.^
oo
r-
rs
oo
m
—
m
1 Fire Departments
o
sO
o
OO
r-~
Os
rsi
VI
SO
so
5!
5
f.
NO
1
•a
u,
o
rs
ON
ON
ID
OO
ON
"fr
rs
0
0
o
NO
m
o
ID
rs
r^
o
0
u.
NO
rs
oo
o
oo
NO
rs
NO
T*~
—
O
^
• — '
NO
o
ID
rs
oo
o
rs
m
rs
-C
| Golf and Country CU
'
oo
r--
JN
^3-
f»>
^
r^
c->
o
o
Landfills
m
o
m
-
o
o
r-i
m
o
—
—
| Libraries
f
rl
Os
r-
^~
—
m
—
0
0
o
on
c
E
'i
o
—
r^j
O
NO
rs
T)-
NO
NO
OO
u
•D
on
o
oo
oo
NO
r^
SO
| Amusement Parks
r4
o
(N
Os
(N
v>
rsi
i/i
r-
V)
Os
' '
S
VI
rsi
rsi
00
VI
r^
rsi
VI
S
SO
Os
(N
VI
m
2
| Misc Recreation Serv
^
oo
rsi
VI
O
(N
VI
(N
O
O
0
oo
rs*
VI
o
rs|
r^
Museums
Os
o
o
o
OS
r<>
o
—
l->
o
o
| Nursing Homes
rs
fi
rs
NO
rs
ON
rs
rs
•D
ON
rs
NO
ID
ON
0
ON
O
O
NO
r-
~~
if
O
Geometries and Characteristics of
Public Water Systems
7-8
December, 2000
-------
?
3
_=
's
O
u
0)
Q.
t
i
1U
_u
^
^
£
1
L.
^^
g«
"s
Q.
O
0.
E
CA
>T»
u
•2
•2
^
•4-t
c
3
E
E
o
r \
U
'
c
o
Z
t~:
^Uj
in
X
S~
**
e
Stmr
U :
'1
!E
g
C
2
w.
; *
*3± ^C
i« a.
: j2
= 3
MI'S
£•18
>
"X.
• s
^5 ^*"
,s> ""3
§"
- B.
: C
£ J
< »
^
| |
fit
g s
£
"5 *
||
Z*3
41
«!
^
ll
!
•03
o
„
fN
O
fN
OO
_^
SO
O
— "
fN
O
fN
OO
™
sO
O
O
O
C/5
C
.i
o
00
0
=
o
o
o
o
o
oo'
Vi
o
0
NO
_*
U
£
00
c
•5
3
C c
'"", o
00 *-
.S 2
ra o.
^ O
«n
0
^j-
o
m
r-
-
O
_'
O
ON
ft
OO
fN
rs
oo
"~~
o
oo'
»n
^
ON
rs
m
tn
U.
DO
3
,a
1
o
o
•y
o
o
o_
rs'
o
0
o
o
o
o_
fN
O
rs
o
o
o
Machinery
DO
C
3
,C3
3
C
re
r^
2
fN
rs
i —
S
rs
rs
ON
I —
O
o
o
Electrical
60
n
i_
3
,«
3
1
^
OO
rs
o
oo
oo
rs
o
ON
OO
~
O
O
o
: Chemicals
OO
c
3
,a
3
C3
fN
rs
rv)
Tf-
rs
VO
£
„
fN
sb
o
fN"
•n
fN
o
: Furniture
00
c
3
•—
3
1
^
r'N
T^-
OO
rs
NO
f)
ON
O
rs"
NO
V~i
r-
rs
NO
I/I
O
NO
r-
ON
oo'
^
fN
r**i
r^
Miscellaneous
00
c
a
,cs
c
1
o
1*^
fN
00
NO
VI
s
oo
tn
ON
NO
"^
8
m
m'
o
o
"TO
Fabricated Met
00
c
3
tra
3
c
TO
O
NO
V)
00
-
rs
o
NO
o
NO
00
rs
o
NO
o
o
o
"O
o.
TO
Q.
00
3
.TO
3
C
TO
m
«n
00
VI
m
r^
VJ
O
»T1
oo
m
m
m
v%
O
O
O
O
Petroleum
00
c:
3
TO ^
11
S &.
00
rs
NO
rs
r**>
m
ON
1^
OO
r^-
rs
NO
rs
fi
ON
O
O
o
c/]
TO
S
TO
E
oil
c
3
,TO
3
TO
O
^f
0
rs
o
rs
0
^
o
o
fN
O
rs
o
o
o
00
c
00
c
3
(2
3
TO
O
fN
O
O
o
8
rs
o
o
o
o
o
8
Rubber & Plast
00
=
3
,TO
3
C
TO
_
rs
ON
_
ON
ON
(N
rs
NO
ON
rs
(N
oo
~
£
fN
Vi
fN
__
~
y,
o
TO
O
C
C
55
00
c
3
,ra
3
c
1
O
o
VI
—
v»
r-
o
fN
8
Vl
—
Vt
r-
o
fN
O
O
o
TJ
Tobacco Produ
00
c
3
O
c
cc
o
§
__
fN
O
0
oo
o
„_
fN
0
o
o
o
Transportation
00
c
3
f«
3
C
CO
o
ON
Vj
*3>
0
oo
1
•*
m
o
oo
o
o
o
Textiles
00
c
3
,C3
C
TO
»n
O
00
s
fN
8
Vl
Vt
00
o
oo
Vt
r-
fN"
VI
fN
•—
—
T3
Lumber & Woe
00
c
3
,co
3
C
CO
SO
Vl
fN
fN
V)
oo
so
fN
SO
v»
00
so
fN
ON
5
o
o
V)
v»
fN
O
fN
fN
CA
B
U
<
D
f_>
E
r°
C
s
o
c
c
r-
2
r-
o
NO
rt
—
r^
o
o
o
VI
3
s
^
—
00
"""
oo
Vl
V
^
C/]
1
4.
u
'•S-
^
X
fN
S
^f"
vf
p;
fN
r--
^r
—
00
VI
*n
fN
v>"
sO
fN
a
rX
S
__'
^
o"
o
~™
o.
o
0
c
CS
<—
CO
OJ
:ms serving I
eft
^^
c/i
L.
0
U
*
Geometries and Characteristics of
Public Water Systems
7-9
December, 2000
-------
Exhibit 7.8. Explanation of Population Served for Non-Community Water Systems
Service Area Type
Daycare Centers
Highway Rest Areas
Hotels/Motels
Interstate Carriers
Medical Facilities
Mobile Home Parks
Restaurants
Schools
Service Stations
Summer Camps
Water Wholesalers
Agricultural Products/Services
Airparks
Bowling Centers
Construction
Churches
Campgrounds/RV Parks
Fire Department
Federal Parks
Forest Service
Golf and Country Clubs
Landfills
Libraries
Mining
Amusement Parks
Military Bases
Migrant Labor Camps
Miscellaneous Recreation Services
Museums
Nursing Homes
Office Parks
Prisons
Racing, including Track Operations
Retailers (Non-Food Related)
Retailers (Food Related)
State Parks
Non-Water Utilities
Zoological Gardens
Manufacturing
Unknown Service Areas
Mixed Service Areas
Population Served Represents
Daily occupancy and employees
Daily visitors
Daily occupancy and employees
Employees and/or daily passengers
Patients and employees
Daily residents
Daily customers and employees
Students and employees
Daily customers
Daily campers
Daily customers
Employees
Daily visitors and employees
Daily customers and employees
Daily workers
Average congregation
Daily visitors
Population protected
Daily visitors
Daily visitors and/or employees
Daily patrons and employees
Employees
Employees
Employees
Daily visitors and employees
Personnel
Daily occupancy
Daily visitors and employees
Daily visitors and employees
Occupants and employees
Employees
Inmates and employees
Daily visitors and employees
Daily customers and employees
Daily customers and employees
Daily visitors
Employees
Daily visitors and employees
Employees
Unknown
Depends on types represented
Geometries and Characteristics of
Public Water Systems
7-10
December, 2000
-------
7.3 Average and Design Flows
Estimates of the average water-use rate, in gallons per person per day for each of the 59 service
classifications, were developed using a variety of literature sources. Where water-use rates were not
identified in the literature for a given service classification, best professional judgment was used to
estimate a usage for similar facilities. Exhibit 7.9 provides a summary of these average water-use rates,
including the basis for best professional judgment determinations. Based on these average water
consumption rates and the estimated population served for each service area type shown in Exhibit 7.7,
average daily system flows within each service classification were estimated. Exhibit 7.10 provides these
results. Similar to CWSs, average daily flows are an important input in estimating operation and
maintenance costs for regulatory analysis purposes.
Design flows for NCWS are also an important input in estimating capital costs for regulatory analysis
purposes. Design flows for NCWSs can be estimated based on design-to-average flow ratios. Design-to-
average flow ratios for NCWSs may differ from those for CWSs. Some NCWS may have greater ratios
than CWSs because storage is typically not available and systems are designed to accommodate larger
variations in demand. However, some NCWSs may have lower design-to-average ratios because their
demand changes little from day to day. In general the design-to-average flow ratios for ground water
CWSs are thought to be a reasonable approximation of ratios for NCWSs of similar size.
The following approach was used to estimate design flows for NCWSs that have similar design-to-
average flow ratios to ground water CWSs:
(1) Use the average daily flow per system for each NCWS service area type (Exhibit 7.10) to back-
calculate an equivalent, or "virtual" population using the CWS average daily flow equation for
public ground water systems (Chapter 4).
(2) Use the equivalent, or "virtual" population produced in step 1 to estimate the design flow for
each NCWS service are type using the CWS design flow regression equation for public ground
water systems (Chapter 4).
The results of these calculations are presented in Exhibit 7.10
Geometries and Characteristics of 7-11 December, 2000
Public Water Systems
-------
Exhibit 7.9. Average Water Use Assumptions for Non-Community Water Systems
(gallons per person per day)
Specific Service Area Type Water Use
Daycare Centers 15
Highway Rest Areas 5
Hotels/Motels 65
Interstate Carriers 5
Medical Facilities 100
Mobile Home Parks 100
Restaurants 8.5
Schools 25
Service Stations 10
Summer Camps 42.5
Water Wholesalers 100
General Service Area Type Water Use
Agricultural Products/Services 100
Airparks 4
Bowling Centers 3
Construction 3
Churches 10
Campgrounds/RV Parks 45
Fire Departments 100
Federal Parks 10
Forest Service 5
Golf/Country Clubs 25
Landfills 25
Libraries 15
Migrant Labor Camps 50
Military Bases 100
Mines 25
Miscellaneous Amusement Parks 20
Miscellaneous Recreation Areas 5
Museums 10
Nursing Homes 100
Office Parks 15
Prisons and jails 100
Race Tracks 5
Retailers (excluding food) 10
Retailers (food) 8.5
State Parks 7.5
Utilities 25
Zoological Gardens 25
Manufacturing (Food and Kindred Products) 35
All Other Manufacturing Categories 25
(3)
Source with "assumption"
(1) "Daycamp"
(1)
(1)
(1)
Best Professional Judgment
(4)
(3)
(3)
(1)
(1)
(4)
Source with "assumption"
Best Professional Judgment
(3)
(1) "Movie theater "
Best Professional Judgment
(1) "Picnic with Toilet Facilities "
(1)
(4)
($)" Picnic with Toilet Facilities"
(1) "Campsite no toilet, bath, or shower'
(1) "Dav workers "
(2)
(3) "Construction workers"
(4) "Residential"
(1) "Day worker"
(3) "Picnic with toilet, shower, etc. "
(3) "Theater"
(2) "Departmentstore"
(4) "Residential"
(2) "Office"
(1) "Institution"
(1) "Fairgrounds"
(2) "Departmentstore"
(3) "Restaurant"
(1) "Picnic with toilet facility"
(1) "Day workers "
(1) "Day workers "
Best Professional Judgment
(1)
Sources:
1
2
3
4
Salvato. Joseph A. Environmental Engineering and Sanitation, 4th Edition
Metcalf& Eddy. 1991. Wastewater Engineering Treatment, Disposal, and Reuse.
Corbitt. Robert A. 1990. Standard Handbook of Environmental Engineering.
Community Water System Survey (CWSS), 1997.
Geometries and Characteristics of
Public Water Systems
7-12
December, 2000
-------
Exhibit 7.10 Non-community Water System Flow Rates by Service Area Type
Service Area Type
Daycare Centers
Highway Rest Areas
Hotels/Motels
Interstate Carriers
Medical Facilities
Mobile Home Parks
Restaurants
Schools
Service Stations
Summer Camps
Water Wholesalers
Agricultural Products/Services
Airparks
Bowling Centers
Construction
Churches
Campgrounds/RV Parks
Fire Departments
Federal Parks
Forest Service
Golf and Country Clubs
Landfills
Libraries
Mines
Miscellaneous Amusement
Parks
Military Bases
Migrant Labor Camps
Miscellaneous Recreation Areas
Museums
Nursing Homes
Office Parks
Prisons
Racing, including Track
Operations
Retailers (Non-Food Related)
Retailers (Food Related)
State Parks
Zoological Gardens
Manufacturing: Food
Manufacturing: Machinery
Manufacturing: Electronic
Equipment
Manufacturing: Chemicals
Manufacturing: Furniture &
Fixtures
Manufacturing: Miscellaneous
Average Flow p«r
Capita p*r Day
-------
Service Area Type
Manufacturing: Fabricated
Metal
Manufacturing: Paper & Allied
Manufacturing: Petroleum
Refining
Manufacturing: Primary Metals
Manufacturing: Printing
Manufacturing: Rubber &
Plastics
Manufacturing: Stone, Clay,
Glass, etc
Manufacturing: Tobacco
Products
Manufacturing: Transportation
Equip.
Manufacturing: Textiles
Manufacturing: Lumber &
Wood
Unknowns
Average Flow per
Capita per Day (gpd)
25
25
25
25
25
25
25
25
25
25
25
25
Average Daily Flow
per System (gpd )
2.044
6,025
8,575
8.316
5,000
1.250
3.485
1.875
675
10.174
1,553
2.125
Design Flow
(gpd)
9.999
28,079
39.338
38,202
23,497
6,249
16.642
9,205
3,469
46,317
7,688
10.375
Design/Average
Ratio
4.9
4.7
4.6
4.6
4.7
5.0
4.8
4.9
5.1
4.6
5.0
5.0
7.4 Transient Versus Non-Transient Systems
As discussed earlier in this chapter, some question exists regarding the accuracy of the SDWIS
subcategorizations of NCWS. The sampling procedure described in Section 7.2 could exacerbate such
miscategorizations. For example, restaurants are considered transient systems. If, during the service
classification sampling, one restaurant was selected that was miscategorized as non-transient, this would
lead to a final estimate reflecting a much larger number of non-transient restaurants. In fact, the final
estimate resulting from the service area sampling described in Section 7.2 included a number of systems
that appeared to be miscategorized (e.g., it included some non-transient restaurants, transient
manufacturing facilities, etc.).
While the categorization may be technically correct (e.g., restaurant could have more than 25
employees), an alternative breakdown of transient versus non-transient systems based on service class is
offered to illustrate the potential difference between the existing SDWIS classifications and reality.
Service area types (e.g., restaurants, service stations) whose populations are variable (e.g., representing
customers, visitors, or guests) were classified as transient. Service classes (e.g., schools, manufacturing
facilities) whose populations are consistent (e.g., representing employees or residents) were classified as
non-transient. Some systems reasonably could be either transient or non-transient. The breakdown of
transient versus non-transient for these systems was not changed. These system types included the
following:
»• Interstate Carriers: includes truck stops and bus and railroad terminals where the
primary water users would be transient (e.g., passengers), but also includes freight depots
and postal service operations where the primary water users would be employees (non-
transient)
Geometries and Characteristics of
Public Water Systems
7-14
December, 2000
-------
*• Hotels: usually transient, but includes boarding houses in which the population might
more appropriately be categorized as non-transient
*• Medical Facilities: includes some extended stay facilities (e.g., nursing homes) that are
non-transient
>• Mobile Home Parks: includes some with seasonal populations (transient) and some that
are more similar to CWSs (non-transient)
>• Agricultural Products and Services: includes facilities more similar to retail food
operations (transient) and facilities more similar to farms or food manufacturers (non-
transient)
> Airparks: similar to interstate carriers
>• Forest Service: includes areas that are primarily recreational (transient) and areas in
lumber production where the primary users would be employees (non-transient)
Exhibit 7.11 summarizes the revised estimate of transient versus non-transient systems. This estimate
makes the breakdown of transient versus non-transient systems consistent with types of service classes
estimated for the population of NCWSs.
Geometries and Characteristics of 7-15 December, 2000
Public Water Systems
-------
w
E
C/3
L.
Qj
«
immun
o
u
c
o
e
1
u
1
s
o
Z
3
£
Q>
^
-^
e
VI
C
a
H
o
e
_o
assificat
U
1
W
^
**
!c
[5
s
o
("•
c
I
u>
O
yr
i
1
J-
».«
II
* ^
a*
o.
X S
&l
>. *j^
<£
^
•O Tfi
flS fi
ft
.2
^J 15
t- a.
^
c
||
•*^
».
•^ ft
| J»
Z.
1
s^i
H o.
^
»l
2 »
«» 3
*» O.
"S »,
.fa S
|I
Service Area Type
w
fe
*r
o*
w
^
si
ft!
Si
CX5
fS
n
^3"
fN
fN
fi
SO
fN
—^
O
O
o
r-
fN
fN
I
sO
rs
_
| Highway Rest Areas
r-
00
o
Tt^
fN"
ON
0
CO
vC
rs
o
o
0
r-
oo
O
fN
ON
O
00
Vi
rs
1 Restaurants
rs
oo
00
f*"l
o
NO
—
r*-T
O
O
O
rs
oo
oo
r«1
r~-
o
p-
r*-j
| Service Stations
R
•^f
r-
fS
%
—
NO'
o
o
o
r*-i
P
«^i
fN
VI
NO
| Summer Camps
S
00
O
1
_'
o
o
o
Tf
o
r-
oo
I
so
—
| Water Wholesalers
0
r-
rN
p^
-
f*^
o
o
o
o
r-
fN
o
r-
m
| Bowl ing Centers
o
00
•r,
NO
.S
~-
-
NO
•/%
O
o
o
o
oo
oo
NO
NO
i
vC
W3
•^
ra
| Campgrounds/RV P
iTi
^~
*t
—
«
NO
O
o
o
i
s
o
r-
NO
| Federal Parks
P
NO
fN
0
oo
SO
fN"
O
O
o
rs
m
VI
fN
00
o
00
sO
fN
C/l
-C
3
1 Golf and Country C!
0
m
r^
-
—
o
o
o
o
m
f>
o
-
1 Libraries
o
o
—
S
fN
3
!--
0
O
O
o
o
V*
Jn
m
o
S
| Amusement Parks
o;
ON
fN
Tf
£
NO
O
0
o
s
rN
sO
V)
so
C/l
| Migrant Labor Cam]
-------
T3
3
C
'S
O
U
*
i
d>
vj
VI
L.
•w
R
e
3
E
S
U
E
O
z
VI
E
2
e
o
VI
4*
01
VI
E
£
H
o
E
O
•*j
03
U
(^
v>
V}
u
>
4J
^
^
x:
X
S
s-
»
>5
£
sa
£
«
+•*
S*
8
r-
§
« 'S
3 'S
K.S.
ft.
'„
II
If :i.
•£
* «
=>?»
£ :
o
-f &
E** o.
G?
JS>*
ijjs
1U
*.
£ 1
E S,
«C *
§
1*
si.
«N
B
?** -5
13
* S.
*?
O _«5
$t s?
K
1
i
I
Ua
f~
3
•iif*
03
O
J^
lu
u.
M
1
57
ft
:±
SO
OO
—
-
g
o
__
SO
SO
00
•~~
_
g
o
—
o
o
0
Daycare Centers
0
in
Vt
—
m"
S
£
Os
O
NO
m'
Os
r-
Os
O
o
o
Schools
(^
r-l
m
S
S
r-
r^
8!
o
O
O
Construction
(N
o
m
m
—
nt
oo
«
•~
o
~
_'
_
oo
—
o
o
o
Churches
SO
"Ti
sO
£
R
§
"^
Vi
R
o
o
o
Fire Departments
Pf
H
ii.
z.
ca
~
gj
H-
O)
ryi
^
E—
m
—
m
so
O
r-
sO
SO
OO
O
oo
^f
o
__
m
m
oo
r-
rsj
oo'
Hotels/Motels
oo
•*T
sO
(N
m
-,
L.
0.
u-
o
o
o
=
P
o
i=
o
C/l
C/5
0>
en
>^
C/5
X)
o
"O
I
•o
(L)
Q.
^
E
trt
^
Q.
CL>
n.
o
u
o
o
o_
r
CO
c
CD
Covers systems s
*
Geometries and Characteristics of
Public Water Systems
7-17
December, 2000
-------
This page intentionally left blank
-------
8: References
58 FR 51735. Executive Order 12866. "Regulatory Planning and Review." October 4, 1993.
Aschengrau, A, S. Zierler and A. Cohen. 1993. "Quality of Community Drinking Water and the
Occurrence of Late Adverse Pregnancy Outcomes." Arch. Environ Health. 48:105-113.
Bauman, Duane D., et. al. Urban Water Demand Management and Planning, McGraw-Hill, New York,
NY, 1998.
Corbitt, Robert A. Standard Handbook of Environmental Engineering. McGraw-Hill, New York, NY,
1990.
Cummins, Michael D. 1987. "Analysis of Flow Data." Report prepared for EPA, ODW, October 5th 199X
Dysard, J.A. 1999. "Trends in Privatization." Journal of the American Water Works Association. 91:44-
47.
Linsley et. al. Water-Resources Engineering. 4th edition. Irwin McGraw-Hill, Inc., Boston, MA. 1992.
Salvato, Joseph A. Environmental Engineering and Sanitation. 4lh Edition. John Wiley & Sons, New
York, NY, 1992.
Tchobanoglous and Burton. Wastewater Engineering: Treatment. Disposal, and Reuse. 3rd Edition.
McGraw-Hill, Inc., New York, NY. 1991.
U.S. Environmental Protection Agency. Community Water System Survey. Office of Science and
Technology, Office of Water, Washington, D.C., 1997.
Viessman and Hammer. Water Supply and Pollution Control. 6th edition. Addison-Wesley Longman, Inc.,
Menlo Park, California. 1998.
Geometries and Characteristics of 8-1 November 20, 2000
Public Water Systems
-------
This page intentionally left blank
-------
Appendix A: Population Served for Public Water Systems
-------
This page intentionally left blank
-------
(«
£
0*
C/5
yT*
u
>
^
^i
."S
c
3
E
E =
u =
w ^
•o *
E *
(/) Q
C "O
O C
£ es
^" u
° 3
cj O
= ™
E
3
z
Ml
2
i^l
in
SO
00
(N
p~
00
fN
fN
O
SO
p-
•"
fN
§
o
so"
Ox
[ —
Ox
fN
0
fN
O
00
m
in
TJ-
f~-
o
in
P^
rxj
oo
fN
fN
'
u
2
3
C.
rchased-
3
CL.
in
in
co
I
I
1
r*"i
10
=
fN
in
so
Ox
^j-
p~
in
in
p~
fN
j —
rxi
in
so
2
es
'u
C.
rchased-
3
a.
m
00
OO
,
,
i
o
0
oo
„_
^^
o
in
„_
Ox
P^
—
P~
^_
^j-
fN
P~
m
so
u
6
^
^ —
T—
co"
t^
0
^^
0
fN"
fN
Ox
O
in
fN
fN
XO
r<"i
oo
SO
in
(N
Ox
fN
fN
^
p^
Ox
m"
SO
OO
Ox
in
t —
^
oo
fN
O
SO
*
0)
1
u
3
en
in
CM
co_
fsj"
CM
so
f^
r*">
o"
m
fN"
O
(^1
^j-
m
fN
p~
"n
so
oo
SO
Ox
in
—
fN
OO
fN
0
SO
^_
00
o
Ox
t —
ro
0
m
m
rN
£
a.
CD
T—
CD"
•""
XO
O^
^r
in
— "
^-
fN
m
*•«
m
in
Ox
in
00
SO
Ox
rf
in
fN
TT
—
SO
O
fN
Ox
OO
SO
OO
in
CN
00
m
2
es
£
^.
CM
(N
CD"
i
rx)
(N
oo
Ox
(N
p^
[^
SO
T^-
m
Ox
fN
^_
fN
Ox
m
Ox
fN
Ox
m
t —
(•<•)
0
^.
SO
—
0
2
3
CL
rchased-
3
a.
co
o
oo"
1
o
oo
fN
O
so
SO
fN
^3"
_^
P^
in
r^
—
in
fN
,
f^
p^
in"
oo
o
Ox
r^
p^
Ox
SO
fN
fN
SO
~
"S
"J^
c.
rchased-
£
OO
f^.
co_
^— "
,
,
o
o
0
SO
in
^j-
r*">
fN
CN"
fN
„_
o
SO
^^
oo
oo
m
in
t —
^
[^
fN
oo
m
1
,
6
co
o
OO
CO"
,
o
o
0
o"
•_
fN
o
o
0
o
so
OO
—
in
fN
fN
[^
r*")
m
_
in
oo
rxj
m
t —
oo
T^-
fN
,
SO
1
uence
c
s
^
nder SW
^
^
O
^_
in
co"
,
o
0
o
o"
fN
0
o
o
o
SO
f")
fy-j
SO
fN
SO
m
m
^
m
oo
Pi
t —
^
oo
fN
SO
SO
1
3
3
CM
OO
,
,
1
i
in
oo
0
SO
0
so
Ox
—
P--
fN
fN
O
SO
1
"es
£
CO
CD
CM_
CM"
,
,
1
0
oo
r*~l
2
m
^^
•^3-
m
m
so
SO
SO
00
f)
Cxi
,
1
u
2
3
C.
rchased-
3
a.
0
CM
OO
,
i
'
i
i
i
0
o
Ox
1
_
fl
1
"5
*tn
o.
rchased-
0.
10
OO
00
1
1
'
o
fN
o"
,
o
oo
>n
fN
—
so
SO
SO
in
SO
t
Ol
6
m
^>
co_
^-"
Ox
Ox
o"
0
rxf
^
Ox
oo"
^^
fN
O
in
P-"
so
0
o
fN
_
ri
oo
in
^
Ox
C»
—
p~-
m
m
fN
^_
SO
2
v>
E
t/3
M
<
t
o
1
Ox
1 o
?: tn
O O
C/) £-
Geometries and Characteristics of
Public Water Systems
A-l
December, 2000
-------
1
ts
1/5
•2J
1
£*
*5
•3
E
E
o
u
S
o
Z
••• Z
g Cfl
2 U
1 ^
•n "O
5 c
> ej
u U
Cfl (J
I S
1 0
DC
^
fN
— "
v->
O
^_
oo
fN
v-t
V~i
oo
|
S
ON
i
i
OO
O
C^}
™~™
fN
SO
*O
so
fN
— "
?
r-
oo
fN
m
"^
r-
£
ft
>
'Z
Cu
oo
oo
fN"
i
i
i
r-
so
fN
so
fN
O
ON
^^T
O
Tf
fN"
SO
t^
ON
fN
O
SO
'
Z
3
Q.
ot
ffl
a
j=
u
u
a.
r~-
i
i
,
O
0
o
\s~i
"~
i
,
oo
SO
SO
<"•>
fN
t^
m
i
9i
1
"£
Q.
•o
01
f/1
9}
JS
U
u
a.
ON
r-
•
1
1
00
r-
SO
so
—
O
rt
t~
rf
fN
fN
O
ro
f
6
,__,
0
1
ON
O
K
^j.
O
°i
ON
O
O
""i.
ON"
fN
2
Tf
V^
0
OO
—
fN
SO
r-
00
SO
fN
fN
in
ON
O>
CQ
cs
'u
a.
-o
O)
M
R
f
|
d!
SO
1
1
1
1
o
o
V)
CO
^_
fN
OO
—
SO
t —
f)
so
fN
,5.
m
^
Ol
O
0
00
1
1
1
o
o
f-
•n
^
ro
•n
—
—
t —
oo
0
fN
SO
^
oo
u
S
3
C
C
i
%f\
•o
e
^
^
U
•n
^
i
i
i
i
t
—
t —
SO
SO
fN
u->
fN
i
—
3
a.
,
5
.
,
'
o
o
*/"T
t
1
^J.
^
ON
OO
W
OS
CL
,
•
,
1
'
,
1
,
,
•
Z
3
a.
•D
Ol
VI
R
.C
U
1-
3
e.
,
-
,
1
1
,
•
,
,
•
v
"5
*£
Q.
•i
w
cs
f
u
L.
e.
r-
i
i
i
i
i
r-
•n
— "
'
1
1
•
Ol
O
oo
o
r-i
i
ON
o
K
TJ-
0
(N
ON
OO
ON"
fN
SO
fS|^
*o"
^ —
I/-J
so
— "
fN
t—
^
fN
fO
>n
•o
E
•w
C«
•<
4
.
^ <
Geometries and Characteristics of
Public Water Supplies
A-2
December 1, 2000
-------
V)
E
"5
>*>
ft
1
•to*
'e
3
E
E
o
C
ZQ.
Ma
e S,
V ft
*"* f^
& CC
C bd
2 z
E— ^
4> *o
> E
4) *
c y
o ~
i o
a. ft
i_
V
E
3
41
DO
U
o
09
^
jg
f
«
W
•O
^S-
W-*
1
&
<>
al
WM
:
I
5
|.
T
*%! 9
•4& ;W
•C& ^&
""^ w?
* ^S
t^^ jn
1
1*** CS
m M
tf R
1 ^b
*** O
St ^5>.
~\5. 39-
f^ ¥**!
^ '
S §
55
^ «
*ft ^
•**«
^ ^M.
*K S
. •»» W)
*!
?
^5^
v
s +^*
S iff?
I1
o
p-
1
^^
SO
oo'
SO
fN
0
o
p
ON
tN
fN
ON
SO
in
O
r-
^.
SO
fN
fN
fN
in
**
*^
round Water
U
oo
fN
00
oo
p-'
p-
_— 1
t^
•n
oo
£
o
oo
p^
fN
O
SO
vt
oo
ft
oo
ON
OO
p-
p-
fN
ft
^
*-*
1
SO
fN
i
Ov
ON
ON
ON
O
0
O
ON'
SO
SO
SO
fN
fN
SO
P~
iri
so
0
0
1^1
f^
_
fN
SO
v-i
Private
^
fN
i
i
,
'
^
OO
oo
in
o
ON
m
p-
p~
ON
_«
m
^
^t
o
tN
«
A
3
C.
Purchased-
ON
f^»
i
,
1
O
O
0
^
ON
fN
0
m
P?
_^
fN
tN
tN
Tt
"~ '
O
CS
"^
c.
Purchased-
00
00
1
1
£
o
so'
SO
00
IN
SO
oo
SO
f)
p^
•n
fN
fN
SO
ft
in
L.
6
00
_~
1
J^
•t^1
t""-
oT
r^
^~
(N
«0
^O
l/-j
^
f~-
ON
fl
^>
fN
in
'^-
SO
urface Water
c«
ON
Vl
— '
,
0
o
^
m
t —
fN
OO
CO
OO
W~l
t^
oo
ft
p-
SO
tN
VO
VO
3
0.
O
p-
,
,
•
o
o
r;^-
fN
oo
oo
p~
t^
tN
fN
0
«n
SO
Private
SO
SO
1
,
g;
^j1
o'
tN
•n
tN
*n
so
oo
m
00
oo
p-
._
ON
tN
m
ft
"^
»
Z
3
a.
Purchased-
•n
i— i
1
1
1
O
T^-
o
p-
in
fN
oo
so
fN
P-
(N
ON
ft
SO
4)
S3
'C
a.
Purchased-
o
tN
i
i
O
fN
—
tN
O
O
2
fN
fN
•n
oo
O
•n
ON
p-
in
^
fN
^.
TJ-
fN
u
-------
This page intentionally left blank
-------
Appendix B: Methodology for Quantifying the Cost Bias Caused by
Retail Population Categorization of Systems
-------
This page intentionally left blank
-------
Appendix B
Methodology for Quantifying the Cost Bias Caused by
Retail Population Categorization of Systems
This Appendix presents data sources, assumptions, and overall methodology for estimating the impacts
of the retail categorization of water systems on national costs. Section B.I describes the methodology for
determining the percentages of purchased water systems (buyers) buying water from Nonpurchased
systems (sellers) of various sizes. Section B.2 describes how the population served of the purchased
water systems was added to the non-purchased systems and how this changed the size categorization of
the nonpurchased systems (shifting them to larger size categories). Section B.3 shows how a national
cost estimate was developed for the retail-based system classification (purchased and nonpurchased
systems) and for the retail+wholesale-based classification system (nonpurchased systems only, adjusted
to include purchased system populations).
B.I Methodology for Determining Percentages of Purchased Systems (Buyers) Buying from
Different Sizes of Nonpurchased Systems (Sellers)
B.1.1 Population Size Categories for Analysis
Four population size categories were used for this analysis. They were as follows:
500 & under
501 to 10,000
10,001 to 100,000
Over 100,000
B.I.2 Obtain purchaser/seller data from SDWIS/FED
Data for all community water systems (CWSs) (active CWSs in the current inventory) and CWSs that
purchase water were obtained from the SDWIS/FED frozen database from January 2000. In order to use
all available data, all water purchase records were retrieved. These data were downloaded from the EPA
mainframe and imported into dBase IV databases for analysis .
B.1.3 Classification of Primary Source in SDWIS/FED
A water system may have one or more sources that are used regularly or on a more limited basis (e.g.,
emergency or seasonal). The regularly used sources are considered permanent sources, while the others
are nonpermanent sources. Many States indicate this type of usage when providing source data; however,
in some cases, these data are not submitted since it is not a required reporting element. In addition, until
recently, if a system had multiple sources, only one source of water was required to be reported. This
policy has been changed; however, actual implementation of this will take time. Some underreporting of
source data is common.
Geometries and Characteristics of B-l December 2000
Public Water Systems
-------
If the State does not provide a value for primary source, SDWIS/FED calculates primary source based on
a review of existing source data that have been provided by the State. All permanent sources are
reviewed first. Primary source is determined based on the presence of at least one source record using the
following hierarchy in descending order:
• Surface Water (SW)
• Purchased Surface Water
• Ground Water Under the Direct Influence of Surface Water (GWUDI)
• Purchased GWUDI
• Ground Water (GW)
• Purchased Ground Water
For example, if you have a system with two permanent sources, one GWUDI and one Purchased Surface
Water, the primary source would be Purchased Surface Water. If there are no permanent sources,
SDWIS/FED will review all available sources and assign primary source using the hierarchy listed
above..
B. 1.4 Analysis of SDWIS Data
a. Review of Primary Source Data for Nonpurchased Sellers and Purchased Buyers
Water purchase records were included in this analysis, based on review of the primary source of
both the seller and purchaser. Since availability (e.g., permanent, seasonal) was not required to
be reported, these data are not present for all purchased systems. Since primary source is
computed by SDWIS/FED, all systems should have a primary source.
If the Primary Source for the
Purchased BUYER is....
Purchased Surface Water
Purchased GWUDI
Purchased Ground Water
the Primary Source of the
Nonpurchased SELLER is...
Surface Water
GWUDI
Ground Water
A water system's primary source is calculated based on review of its source data, as discussed
previously. If a CWS purchases water from another system on a nonpermanent basis, and it has
permanent sources, the permanent sources would be used to calculate primary source. If a water
system has more than one source, and the sources are of various types, primary source is
assigned based on a descending hierarchy of source types. If a system had a permanent ground
water source and an emergency purchased surface water source, it would be considered to have a
primary source of ground water. In cases where the primary source of the purchaser did not
reflect the primary source of the seller (e.g., purchaser had a primary source of ground water,
while seller was surface water), it was assumed that the source was not permanent, and therefore
not included in this analysis.
Geometries and Characteristics of
Public Water Systems
B-2
December 2000
-------
If a system had multiple types of sources, e.g., a system that purchased water had a primary
source of surface water, but actually had one surface source and one purchased ground source, it
was assumed that it would be adequately represented as a nonpurchased system for Information
Collection Rule purposes, and was not included in this analysis.
b. Non-CWS Sellers
Some CWSs purchased water from systems that were not CWSs. For the purposes of this
analysis, only sellers and purchasers that were CWSs were included.
c. Orphan Purchasers
Purchasers without Sellers were omitted, since this analysis focused on sellers.
d. Cascading Provider Relationships
A Nonpurchased Seller (System A) may provide water to a purchased system (System B) that
does not have any other sources of water. This purchased system (System B) may then in turn
provide water to another purchased system (System C) that does not have any other sources of
water. For the purposes of this analysis, purchased systems (i.e., those systems with a primary
source of purchased GW, purchased GWUDI, or purchased SW) were assigned to nonpurchased
sellers (i.e., those systems with a primary source of nonpurchased GW, GWUDI, or SW). In a
few cases, a purchased system (e.g., System C) could not be associated with a nonpurchased
system (e.g., System A) due to a lack of information. If a purchased system (e.g., System C)
could not be associated with a nonpurchased system (e.g., System A), it was not included in this
analysis.
B.I.5 Developing Percentages
a. Numbers of Buyers by Seller and Buyer Primary Source and Population Category
After review and analysis of water purchase records as outlined above, purchasers and sellers
were assigned to the four population categories. Based on nonpurchased seller primary source
and population category, numbers of unique buyers (i.e., systems with purchased primary source)
by buyer primary source and population category were counted.
b. Numbers of CWSs by Primary Source and Population Category
To obtain a baseline for the percentages of systems that meet certain characteristics, the
inventory of CWSs was subdivided by primary source and the four population categories
outlined above.
Geometries and Characteristics of B-3 December 2000
Public Water Systems
-------
c.
Percent of Buyers that Purchase Water from Various System Sizes
The table Percent of Buyers that Purchase Water From Various System Sizes was developed by
dividing the numbers of buyers in 5a by the number of CWSs in 5b above.
B1.6 Results
Results for surface water, ground water, and GWUDI systems are presented in Tables Bl though B3,
respectively. Percentages in bold font indicate the percent of systems that buy water from sellers of
similar sizes. The following examples show how these results can be interpreted.
> Table Bl (purchased surface water systems): For very small purchased systems (serving
500 persons or less) that purchase surface water from nonpurchased surface water
systems, 8 percent buy from other very small systems, 40 percent buy water from small
systems, 34 percent buy water from medium systems, and 18 percent buy water from
large systems.
*• Table B2 (purchased ground water systems): For small purchased ground water systems
(serving 501 to 10,000 persons) that buy water from nonpurchased ground water
systems, 9 percent buy from very small systems, 54 percent buy from small systems, 30
percent buy from medium systems, and 7 percent buy from large systems.
Table Bl. Percent of Buyers that Purchase Water from Various System Sizes
Surface Water Systems
System Sfee of
Purchased Buyer
(Population Served)
Very Small
(500 & under)
Small
(501 - 10,000)
Medium
(10,001 - 100,000)
Large
(Over 100,000)
Percent of Buyers by $i»e Categories of Sellers
(Population Served of the Seller)
Very Small
(50fl& under)
8%
12%
23%
43%
Small
(S01-:iO,000)
40%
28%
6%
2%
Medium
(KMXU-KKMWQ)
34%
40%
21%
4%
Large
(QverHMMIOty
18%
20%
50%
51%
Total
100%
100%
100%
100%
Geometries and Characteristics of
Public Water Systems
B-4
December 2000
-------
Table B2. Percent of Buyers that Purchase Water from Various System Sizes
Ground Water Systems
System Size of
Purchased Buyer
(Population Served)
Very Small
(500 & under)
Small
(501 - 10,000)
Medium
(10,001 - 100,000)
Large
(Over 100,000)
Percent of Buyers by Size Category of Sellers
(Population Served of the Seller)
Very Small
(S0)
63%
54%
0%
0%
Medium
(10,001 -100,009)
21%
30%
46%
0%
Large
(Over 100,000)
4%
7%
50%
50%
Total
100%
100%
100%
100%
B.I.7 Conclusions
Purchased surface water systems tend to buy water from systems that are in the next larger size category.
Many (40%) of the very small purchased surface water systems buy their water from small nonpurchased
surface systems. Many (40%) small purchased surface systems buy their water from medium
nonpurchased surface water systems. Half (50%) of the medium purchased surface systems buy their
water from large nonpurchased surface systems. Over half (51%) of the large purchased surface systems
buy their water from large nonpurchased surface systems.
Ground water systems tend to buy water from systems that are similar in size or in the next larger size
category. Many (63%) of the very small purchased ground water systems buy water from small sellers.
Over half (54%) of the small purchased systems also buy water from small sellers. Most of the medium
purchased ground systems buy water from either medium (46%) or large (50%) sellers. Half (50%) of the
large purchased ground systems buy water from large nonpurchased ground water sellers.
Since GWUDI systems make up a very small portion of the total universe of water systems, they were not
considered in assessing the impacts of retail-based system categorization on developing national costs.
Geometries and Characteristics of
Public Water Systems
B-5
December 2000
-------
B.2 Methodology for Allocating Population of Purchased Water Systems to Nonpurchased
Water Systems to Represent Combined (Wholesale and Retail) Systems
Attributing the retail population of purchased water systems (water buyers) to the systems that sell water
would effectively shift nonpurchased systems into higher size categories. This section shows how the
population served by purchased water systems is allocated to nonpurchased water systems and what
appearance the resultant shift might take. The analysis proceeds along four steps:
(1) Aggregate SDWIS system count data into four general categories. Calculate the mean
population served for each general category using SDWIS data.
(2) Using percentages in Tables Bl and B2, determine the number of purchased systems
(buyers) buying from sizes of nonpurchased systems (sellers).
(3) Determine shift in nonpurchased systems by adding the mean population served of the
allocated purchased systems to nonpurchased systems' mean population on a one plant to
one plant basis.
(4) Reallocate systems from general population size categories into specific SDWIS
population categories.
Tables B3 and B4 show the number of systems by their wholesale relationship and the mean population
served for four general population categories for surface and ground water systems, respectively. The
mean population is the weighted average of population per system as given for specific SDWIS
population categories.
Table B3. Number of Nonpurchased and Purchased Systems
Surface Water
System S&e i
(fopulation Served) j
Number of
Nonpurchased
Systems
(Sellers)
Number of
Purchased Systems
(Buyers)
Mean Population
Served per System*
500* refer
CVefy SmdllS
1078
1668
211
m - HMftft
(Swsall)
2455
2829
3,054
J WJ » iQftBQO
{Nfeliuilj)
1091
692
30,073
: Ov»rI«M»0
CLaff^l 1
214
47
343,487
* Based on December 1998 SDWIS data, average includes purchased and nonpurchased system information
Geometries and Characteristics of
Public Water Systems
B-6
December 2000
-------
Table B4. Number of Nonpurchased and Purchased Water Systems
Ground Water
System Size ;
(!*0patjttwin Served
Number of
Nonpurchased
Systems (Sellers)
Number of
Purchased Systems
(Buyers)
Mean Population
Served per System*
(V&y Steal!)
27341
1058
158
(Sraai)
11908
816
2,191
{Mediant}
1310
36
24,973
Ov«r 109,000
(Large)
59
4
276,684
: Based on December 1998 SDWIS data, average includes purchased and nonpurchased system information
Step 2 in this analysis is to determine the number of purchased systems buying from different sizes of
nonpurchased systems. Tables Bl and B2 showed the percentage of buyers buying from different sizes of
sellers. These percentages were multiplied by the total numbers of systems shown in B3 and B4 to
calculate the numbers buying from each size category. For example, very small surface water buyers buy
8 percent of their water from very small surface water sellers, so 8 percent of the very small surface
water buyers were assigned to the very small seller category. Tables B5 and B6 summarize results for
surface and ground water systems, respectively.
Table B5. Number of Purchased Systems Buying from Different Sizes of Sellers
Surface Water
Siaia^e Water
Buyer $&e
Very Small
Small
Medium
Large
\ , Mlw Si#J
Very Small
134
339
159
20
Small
667
792
42
1
Medium
567
1132
145
2
Large
300
566
346
24
Total Systems
1668
2829
692
47
Geometries and Characteristics of
Public Water Systems
B-7
December 2000
-------
Table B6. Number of Purchased Systems Buying from Different Sizes of Sellers
Ground Water
€mttnd Water
Buyer Size
Very Small
Small
Medium
Large
StlterSize •• .
Very Small
127
73
1
2
Small
667
441
0
0
Medium
222
245
17
0
Large
42
57
18
2
Total Systems
1058
816
36
4
For very small surface water purchased systems, 134 bought their water from very small surface water
nonpurchased systems, while 667 bought from small systems, 567 bought from medium systems, and
300 bought from large systems. This accounts for all 1668 very small surface water purchased systems.
Step 3 involves estimating shifts in size category of the nonpurchased sellers when purchased systems
are linked to nonpurchased systems (wholesale and retail population categorization). In order to simplify
the analysis, it is assumed that:
• All plants within a given general size category have the same population as the mean population.
• The relationship between sellers and buyers is one to one.
To estimate shift, the mean population of the buyer is added to the mean population of the seller. Shifts
basically occur when the buyer size is larger than the seller size. For example, the addition of a very
small system's mean population to a seller would not increase the seller's mean population enough to
move it to a higher size category. However, adding a small system's mean population to a very small
seller would shift that seller to a higher size category (the small category). Tables B7 and B8 show the
shifts as positive or negative changes to the number of nonpurchased systems. Note that the net change in
numbers of nonpurchased sellers is zero.
Geometries and Characteristics of
Public Water Systems
B-8
December 2000
-------
Table B7. Changes in Number of Nonpurchased Seller
Surface Water
Buyer Size
Very Small
Small
Medium
Large
TOTAL CHANGE IN
SELLER SIZE
Very Small
0
-339
-159
-20
-518
Small
0
+339
-42
-1
+296
Medium
0
0
+201
-2
+199
Large
0
0
0
+23
+23
Net Change
0
0
0
0
0
Table B8. Changes in Number of Nonpurchased Systems
Ground Water
Buyer Size
Very Small
Small
Medium
Large
TOTAL CHANGE IN
SELLER SIZE
Very Small
0
-73
-1
-2
-76
Small
0
+73
0
0
+73
Medium
0
0
+ 1
0
+1
Large
0
0
0
+2
+2
Net Change
0
0
0
0
0
Tables B9 and B10 were then derived by adding or subtracting the total changes to the nonpurchased
system from tables B3 and B4. These now represent a gross estimate of the re-categorization of systems
if they were categorized according to retail and wholesale population.
Geometries and Characteristics of
Public Water Systems
B-9
December 2000
-------
Table B9. New Surface Water System Totals
Retail + Wholesale Population Categorization
System
- 100,000
{Metal}
Number of
Nonpurchased
Systems
560
2751
1290
237
Table BIO. New Ground Water System Totals
Retail + Wholesale Population Categorization
System Size
580 & ta«te»
50* - l
-------
Table Bll. Nonpurchased Systems Sorted by Specific SDWIS Size Category
Surface Water
System Typ*
(Surface Water)
Nonpurchased Systems (a)
Total in General Category (b)
Specific Category's Percent
ofTotal(a)/(b) = (c)
New Total in
General Category (d)*
New Nonpurchased Systems
(After Allocation) (c) x (d)
Number of Satiate Water CWSs- by Population Category* :
580 or Less
25
to
100
382
101
to
500
696
1,078
35.4%
64.6%
560
198
362
SOt to 10,000
50-t
to
1,000
411
1,001
to
&300
1.086
3^01
tO
19,000
958
2.455
16.7%
44.2%
39.0%
2,751
461
1.216
1.074
14,691 to tOOK
10,001
W
50,000
913
50,001
10
100,000
178
1.091
83.4%
16.3%
1.290
1.080
210
>100,«00
100,001
to
1,000,000
200
Greater
thai*
i, 000,000
14
214
93.5%
6.5%
237
221
16
Totals
4,838
4.838
4,838
4,838
* from Table B9
Table B12. Nonpurchased Systems Sorted by Specific SDWIS Size Category
Ground Water
System Type
(Ground Water)
Nonpurchased Systems (a)
Total in General Category (b)
Specific Category's Percent
ofTotal(a)/(b) = (c)
New Total in
General Category (d)*
New Nonpurchased Systems
(After Allocation) (c) x (d)
Number of Surface Water CWSs by Population Category*
500 At Les&
25
to
190
13.438
101
to
500
13.903
27.341
49.1%
50.9%
27.265
13.401
13.864
501 to 10,000
501
f»
1400
4.244
1,001
to
3,300
5,316
3,301
to
30,080
2.348
11.908
35.6%
44.6%
19.7%
11.981
4.270
5.349
2.362
10,001 to IOOK
10,001
to
50»000
1.180
50,001
to
100,000
130
1,310
83.4%
16.3%
1.311
1.181
130
>1 09,000
100,001
to
1,000,000
57
Greater
than
1,000,000
2
59
93.5%
6.5%
61
59
2
Totals
40.618
40,618
40,618
40.618
from Table BIO
Geometries and Characteristics of
Public Water Systems
B-ll
December 2000
-------
The results of this analysis are summarized in Tables B13 and B14. Changing the population
classification from retail based to retail plus wholesale based results in a reduction from 10,074 surface
water systems to 4,838, and shifts the relative distribution of surface water systems toward higher
population categories. There is a proportional rise in systems per category for all systems with 1,001 or
more persons per system, with a proportional decrease in systems with fewer than 1,001 persons per
system. For ground water, the change in classification scheme does not have nearly as great an effect,
reducing the total from 42,532 to 40,618.
Table B13. Comparison of System Categorization Schemes
Surface Water
Sy&tem Type
Nonpurchased Systems (a)
Purchased Systems (b)
Total Systems . Retail Based
(a)+(b)
Revised Systems. Retail +
Wholesale Based
Number *f S« rtaet Water C WSs by System Size Category*
25
to
10O
382
483
865
198
101
to
500
696
1.185
1,881
362
501
to
1,000
411
719
1,130
461
1,00!
to
3,300
1,086
1.282
2,368
1,217
3,301
to
10,000
958
828
1,786
1,074
1O,001
to
50,000
913
603
1,516
1,080
50,001
to
100,000
178
89
267
210
100,061
to
1,000,000
200
47
247
221
Greater
titan
1,000,000
14
0
14
16
Totals
4,838
5.236
10,074
4,838
*Nonpurchased and purchased systems represent the sum of public and private systems for this analysis. Does not include
"other" ownership category.
Table B14. Comparison of System Categorization Schemes
Ground Water
System Type
Nonpurchased Systems (a)
Purchased Systems (b)
Total Systems . Retail Based
(a)+(b)
Revised Systems. Retail +
Wholesale Based
Number of Ground Water CWSs by System Size Category*
25
to
100
13.438
285
13,723
13,401
101
to
500
13,903
773
14,676
13,864
501
to
1,000
4,244
366
4,610
4,270
1,001
to
J,300
5.316
355
5,671
5,349
3^01
to
10,000
2,348
95
2,443
2,362
10,001
to
50,D00
1.180
35
1,215
1,181
56,001
to
100,000
130
1
131
130
100,001
to
1,000,000
57
4
61
59
Greater
than
1,000,000
2
0
2
2
Totals
40.618
1.914
42,532
40,618
*Nonpurchased and purchased systems represent the sum of public and private systems for this analysis. Does not include
"other" ownership category.
Geometries and Characteristics of
Public Water Systems
B-12
December 2000
-------
B3. National Cost Estimates Using Retail-Based and Retail+Wholesale-Based System
Classification
For a given technology, operations and maintenance (O&M) expenses and capital costs generally
increase on a per gallon basis as system size decreases. In other words, economies of scale increase for
larger systems. The purpose of this section is to quantify the effects of retail system categorization by
comparing gross national costs based on the retail system categorization in SDWIS to those based on
retail+wholesale categorization as presented in Tables B13 and B14.
The cost bias introduced by categorizing systems by retail population (i.e., creating a larger number of
small systems) depends on the slope of the costs equations for a particular technology. In other words,
the bias depends on how the technology costs per gallon change with system size. For this exercise, the
cost bias was quantified using treatment technologies from the EPA Document, "Technologies and Costs
for Control or Microbial Contaminents and Disinfection Byproducts" (November 2000). Three treatment
technologies were selected for the analysis:
>• Microfiltration/ultrafiltration (MF/UF)
> Ultraviolet light (UV) disinfection
*• Conversion to chloramines for secondary disinfection (CLM)
The first two technologies reduce the level of microbial contaminants, particularly Cryptosporidium, in
surface water. Converting to chloramines for secondary disinfection is an inexpensive way to reduce the
formation of disinfection byproducts. Of the technologies shown, MF/UF is the most expensive
technology at all size categories, followed by UV disinfection and then conversion to chloramines.
Total annual costs were calculated on a per system basis for each of the three technologies. This was
done by annualizing capital costs assuming a 3 percent discount rate over 20 years and adding the
annualized capital to O&M costs. Then national costs were estimated by multiplying the number of
systems in each size category by the total annual cost per system using both system categorization
schemes (retail based and retail+wholesale-based). The percent increase in costs caused by retail-based
size categorization was estimated as the difference between the two national estimates divided by the
retail+wholesale based national estimate.
Tables B15 and B16 summarize the results. National costs were calculated assuming all systems
implemented each technology, although the percent increase from retail+wholesale categorization to
retail categorization is independent of the number of systems selecting the technologies (as long as the
same proportions are used for each size category). These results show that, for surface water systems,
total annual treatment costs using the retail-based system information from SDWIS can be 22 to 45
percent higher than actual cost incurred. The effect on treatment cost estimates for ground water systems
is much less (4 percent higher than what systems would actually incur for the three technologies
evaluated).
Geometries and Characteristics of B-13 December 2000
Public Water Systems
-------
Table B15.
Percent Increase in Cost From Retail+Wholesale-Based to Retail-Based System Categorization
Surface Water Systems
Technology
MF/UF
UV
CLM
Total National Cost ($ Million)*
Retail+Wholesale Based
System Classification
$4,490
$377
$64
Retail Based System
Classification
$3,656
$295
$44
Percent Increase
23%
28%
45%
* Based on 100 percent of systems installing the technology. Actual percent selecting the technology did not affect the percent
increase from retail+wholesale-based classification to retail-based classification.
Table B16
Percent Increase in Cost From Retail+Wholesale-Based to Retail-Based System Categorization
Ground Water Systems
Technology
MF/UF
UV
CLM
Total National Cost ($ Million)*
Retail+Wholesale Based
System Classification
$3,818
$379
$100
Retail Based System
Classification
$3,668
$364
$96
Percent Increase
4%
4%
4%
* Based on 100 percent of systems installing the technology. Actual percent selecting the technology did not affect the percent
increase from retail+wholesale-based classification to retail-based classification.
Geometries and Characteristics of
Public Water Systems
B-14
December 2000
-------
Appendix C: 1995 Community Water Systems Survey Questionnaire
-------
This page intentionally left blank
-------
OMB No.:
Expinx 7/31/97
United States
Environmental Protection Agency
SURVEY OF PUBLIC
COMMUNITY WATER SYSTEMS
ll/M/M
-------
Please return this questionnaire in the enclosed postage-paid envelope
or mail to:
EPA Community Water Systems Survey
1650 Research Boulevard
Room GA 45
Rockville, MD 20850-9973
The following questionnaire is estimated to require 45 minutes to in hour to complete.
This includes time for reviewing instruction*, gathering and reporting the requested data, and reviewing the questionnaire.
Send comment! regarding the burden estimate or any other aspect of this survey, indicating suggestions for reducing this burden, to:
Chief, Information Policy Branch, 2136 • US. Environmental Protection Agency • 401M Street, S.W. • Washington, DC 20460. and
Desk Officer for EPA • Office of Information and Regulatory Affairs •• Office of Management and Budget • Washington, DC 20503.
-------
Please respond about:
If you have any questions about the survey or how to
complete the questionnaire, please call:
Please return your completed questionnaire in the
enclosed postage-paid envelope by March 10, 1995.
-------
1994 Community Water Systems Survey:
Public Systems Questionnaire
GENERAL INSTRUCTIONS
This questionnaire asks three preliminary questions and then is divided into two major parts:
PART I - OPERATING CHARACTERISTICS (Questions 4-27); and
PART II - FINANCIAL CHARACTERISTICS (Questions 28-40).
Please complete the questionnaire as follows:
• In Question 1, provide the best contact person for each part (I and II);
• In Question 2, indicate the latest full-year reporting periods for which your operating information,
and financial information are available;
• In Part I of the questionnaire, use the period indicated in Question 2(A) to report 'last year's*
operating data; and in Part II, use the period indicated in Question 2(B) to report "last year's*
financial data;
• In Part II of the questionnaire, record dollar amounts as whole dollars;
. Please record your answers for the questionnaire by filling in the blank(s)
or circling the appropriate number(s) for each item; and
• Make a copy of the completed questionnaire for your records before sealing it in the enclosed
envelope.
1. Please provide the name, title and telephone number of the most knowledgeable person to
contact for information on:
T
(A) PART I - OPERATING CHARACTERISTICS:
Name: Title:
Tel. No. ( ) - Fax No..
T
(B) PART II - FINANCIAL CHARACTERISTICS
(Write "SAME" if same as above)
Name: Title:
Tel. No. ( ) - Fax No.
-------
2. Please specify the end date of the most recent 12-month reporting period for which your
drinking water system can provide operating and financial information.
T
Can be reported
for the
12 months ending
(A) Operating information / /
(B) Financial information / /
3.
Please indicate, by circling the appropriate numbers in columns A, B, and C, whether the
organizations or people listed below provide your drinking water system with:
(A) Information on drinking water requirements and guidance;
(B) Operator training; and
(C) Technical assistance.
(Circle all numbers that apply for each information source)
INFORMATION SOURCE
T
(A)
Source providing
Information
on drinking
water requirements
and guidance
1. State Department of Natural Resources, state
Health Department, or state EPA 1
2. Other state government departments or
extension services 1
3. U.S. Environmental Protection Agency 1
4. Other federal agencies or extension services
(e.g., FmHA, Rural Development Administration) 1
s. County government 1
6. Local government 1
7. State rural water associations 1
8. Other associations 1
9. Rural community assistance program 1
10. Contracted engineering services 1
11. Citizen volunteers 1
12. Electronic bulletin boards 1
i
13. Technical publications 1
u. Radio or television 1
I is. Local newspapers 1
16. Federal register 1
17. Other (Please specify) 1
! 18. 1
T
(B)
Source
providing
operator
training
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
V
(C)
Source
providing
technical
assistance
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
-------
PART I - OPERATING CHARACTERISTICS
t
I
PRODUCTION AND STORAGE
4. For each type of water source listed below, please indicate which ones you use and:
(A) the number of gallons (in millions of gallons) produced in the last year (i.e., the amount
of water going into the distribution system); and
(B) the number of water intake points with disinfection.
ft
k
5.
6.
WATER SOURCE
Do you obtain water
from this source?
Ground water ....
Surface water . . . .
Water purchased from
other systems . . .
YES
1
1
NO
2
2
If YES, enter the number of:
(A) (B)
Gallon* Number of
produced In Intakt
th» lut y«ar point* wtth
(In millions) disinfection
What was your system's peak daily production of non-purchased drinking water during the past
year, and what is the system's maximum daily treatment design capacity?
T
Gallons
per day
(A) Peak daily production
(B) Maximum daily treatment design capacity
You reported your system's maximum daily treatment design capacity in Part B of Question 5.
There are several possible factors that may have resulted in a maximum capacity of this size.
Some possibilities are listed below. Please circle the number to the right of each factor that
indicates how important that factor was in determining your system's maximum design capacity.
Factor Determining Maximum Design Capacity
1. Current peak needs (beyond average daily flow)
2. Seasonal demand (e.g., irrigation)
3. Emergency flows (e.g., fire, drought)
4. Expected growth
s. Limited choice in package plant sizea
6. Other (Please specify)
How Important was this factor?
Very Important __ __ Not Important at all
2
2
2
2
2
2
3
3
3
3
3
3
4
4
4
4
4
4
5
5
5
5
5
5
k
k
k
k
k
I
I
a
i
i
i
I
-------
Do you have treated water storage?
T
1 Yes
2 No —»
Go to Question 9
8. Please indicate whether you have the following types of treated water storage listed below; and
if so, for each type of storage:
(A) how many tanks do you have;
(B) what is their storage capacity (in millions of gallons); and
(C) do you disinfect water in these tanks after storage?
TYPE OF TREATED
WATER STORAGE
GROUND LEVEL OR SUB-SURFACE STORAGE
Natural materials (e.g., wood, earth):
Does your water
system have this
type of treated water
storage?
YES NO
If YES, complete the following:
W (B) (C)
Total Dlalntact
•tonga capacity altar
Numbar (In mllHona of ttoraga?
of tank* gallona) YES NO
1. Uncovered
2. Covered .
Synthetic materials
(e.g., steel, concrete):
3. Uncovered . . . .
4. Covered.
ELEVATED STORAGE
Natural materials (e.g., wood):
5. Uncovered
e. Covered
Synthetic materials
(e.g., steel):
7. Uncovered . . ,
8. Covered . . . ,
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
-------
vnm&t 4
1
1
1
9. Please indicate the types of pipe used in your distribution system. For each type of pipe, what
is the number of: »
(A) miles (or feet) of existing pipe; ™
(B) miles (or feet) of pipe replaced in the last year;
(C) water main repairs in the last year; and k-
(D) months between flushes for that type of pipe. B
!,
Doesyo
distribut
systen
haveth
type ol
pipe?
TYPE OF PIPE YES r
IRON:
1. w/ Cement Lining 1
2. w/o Cement Lining 1
ASBESTOS CEMENT:
3. w/ Vinyl 1
4. w/o Vinyl 1
s. PVC 1
e. Other plastic 1
7. Other (Please specify)
1
If YES, enter the number of:
ur t.
on (A) (B) (C) (D) B
1 MilMor MilMor Walw If.
* f**t Mil** f**t(*p*clfy Mil** main
(ap*cify or which) of or repair* In Month* __
which) f**t? pip* replaced f**t? th*l**1 b*tw**n ^R
of«tl*t- (Circle Inth* (Circle y*arfor flu*h**for 0[
4O ing pip* MorF) la*t y**r MorF) typ* of pip* typ* of pip*
MM
2 F F
M M fe
2 f f «P
M M |p
2 F F
2 F F BBBB!
m
M M
2 F F -
M M •
2 F F ^
M M j^
2 F F SB
0. How many miles (or feet) of new pipe (for expansion purposes) have you installed in the last 5 mj§
years? Enter response for either miles or feet, but not both.
(If zero, enter "0")
T T -P
MILES OF NEW PIPE OR FEET OF NEW PIPE
•
0
-------
11.
How many people and connections does your system currently serve with piped drinking water,
and how many did it serve 5 years ago?
(Please estimate if you dont know the exact number.)
NOTE:
If your system serves a population that changes on a seasonal basis (for example, a winter
or summer resort area), please indicate the highest seasonal number of people served or
active connections.
(A)
(B)
T
Currently
T
5 years ago
PEOPLE SERVED WITH PIPED DRINKING WATER
ACTIVE CONNECTIONS WITH PIPED
DRINKING WATER
12. What are the ZIP codes of your service area? (If your system's service area covers more than
10 ZIP codes, record only the first 3 digits of the ZIP code(s), i.e., all ZIP codes covered by the
same 3 starting digits can be summarized as one ZIP code by recording the first 3 digits.)
OB.
FIRST 3 DIGITS OF ZIP CODES (For systems whose service area covers more than 10 ZIP codes)
XX XX XX XX XX
OPERATOR
13. Do you have any drinking water treatment plant operators currently employed by your system?
T
1 Yes
2 No -
Go to Question 15
-------
14. Please indicate whether the treatment plant operators you employ have attained any of the
training level categories listed below. Provide the number of operators and average operator
work week (in hours) for each applicable training level category:
TRAINING LEVEL CATEGORY
STATE CERTIFIED (i.e., with state-approved
certified training for drinking water)
- Full time operator(s) [Definition:
Works at least 35 hours a week]
- Part time operators) who also operate other
drinking water plants (e.g., "circuit riders") .
• Other part time, state certified
operators
TRAINED THROUGH A NATIONAL OR STATE
PROGRAM, BUT NOT STATE CERTIFIED
- Full time operators) (see definition above). .
• Part time operators) who also operate other
drinking water plants (e.g., "circuit riders") .
- Other part time, trained operators
OTHER TRAINING LEVEL
(e.g., on-the-job training)
- Full time operator(s) (see definition above). .
• Part time operators) who also operate other
drinking water plants (e.g., "circuit riders") .
- Other part time operators not
classified above
Do you employ
drinking water
treatment
operator* who
have attained this
training lave)
category?
YES NO
If "YES"
Average hour* per week
per operator
Other
How many Drinking
operator* Treatment Water
do you have? Dutto Dutie*
(Number) (Hra) (Hra)
2
2
2
2
2
2
2
2
B
1
R
R
I
I
R
i
WATER SOURCES Af«J TREATMENT
15. Is your water system interconnected to another system that you can use for emergency
purposes (e.g., hot summers)?
V
1 Yes
2 No
P
I
I
1
-------
*».-•• •
-
16. If your primary source of drinking water became permanently unusable due to contamination,
please indicate whether or not you would adopt any of the solutions listed below:
SOLUTION
1. Draw more heavily upon other sources on the present system. .
2. Draw upon another system to which you are now connected . .
3. Draw upon alternative sources (e.g., hook up to another system)
4. Implement a water management plan (e.g., rationing)
s. Drill new well(s).
6. Curtail service
7. Other (Please specify)
If primary water
sources became
unusable, would you
adopt this solution?
YES NO
2
2
2
2
2
2
2
17. If you are currently interconnected to your long term alternate water source, check the box
indicated and go to Question 18.
V
Q
Currently interconnected —>
Go to Question 18
If you have no long term alternate water source(s), check the box indicated and go to Question 18.
No long term alternate water source(s) —»
Go to Question 18
What is the name of your |gng term alternate water source(s) and how many miles is it from the
nearest connection point on your current system?
Nam* of long Mm
alternate water aourc«(s)
1.
2.
3.
T
Diatanc*
from
ayatem
(to nearest mile)
V
If distance is under
one mite, please
estimate distance
in feet
-------
I I
I I
w « « w ^r 5 *•
•5 *
<0
UJ
a
o
o
III
I
£
£
X
o
to
IU
§r>- a o> o *- CM
«- t- «- CM CM CM
O
00
-------
19. Are there places in your distribution system other than those reported in your answer to
question 8C (storage) or question 18 (treatment facilities) where you boost disinfectant
residuals?
T
1 Yes
2 No
20. Please supply the following information for each well or surface water intake not receiving
treatment (include only sources that were active in 1994). If you have more than five wells and
intakes, please check here Q. (Record the information about the additional wells and intakes
on a photocopy of this page or use a blank sheet of your own.)
T
AQUIFER OR
SURFACE WATER SOURCE
SOURCE NAME TYPE (enter
(e.g. Ogalala Aquifer G for Ground
or Ohio River ) or S for Surface)
1.
2.
3.
4.
LOCATION OF WELL OR INTAKE:
(Enter latitude and longitude from
local plat map or permit) Average
Latitude Longitude flow
(Degrees Min. Sec.) (Degree* Mm. Sec.) (Gal/day)
If well,
please list:
Potential Well
ftft^, rf ••*•»>
now civjjui
(GaVOay) (teet)
SOURCE WATER PROTECTION
21. Does your drinking water system participate in a source-water or wellhead protection program?
V
1 Yes
2 No —>
Go to Question 25
22. Please indicate whether or not the following measures are being adopted in your source-water
or wellhead protection program:
MEASURE
Is this measure
adopted in your
source-water or
wellhead protection
program?
YES NO
i.
2.
3.
4.
Education on land use impacts 1
Ownership of a watershed 1
Zoning or land use controls 1
Best Management Practices
(such as run-off controls, fertilizer scheduling,
less toxic road maintenance materials) . . . .
Other (Ptease specify)
2
2
2
2
2
10
-------
I
23. Who leads or manages this program? I
(Circle only one number)
T F~
1 Local government
2 Regional authority (e.g., Section 208 Agency) fr
3 State agency
4 Other (Please specify)
24. How is the management area delineated?
(Circle all numbers that apply and fill in the blanks if 3, 4 or 5 is circled)
T
1 By watershed boundaries '
s
2 By aquifer boundaries
3 By a fixed radius around well of feet r
4 By a fixed distance from a surface water body of feet
25.
1
2
3
4
5
6
7
8
9.
10
11
12
13
14.
Please indicate if any of the potential sources of contamination listed below exist
of your water supply intakes:
^
Does this potential so
exist within 2 miles o
POTENTIAL SOURCE OF CONTAMINATION YES
Industrial or manufacturing facilities 1
Agricultural runoff . . 1
Animal feed lots . ... 1
Urban runoff . . . . . 1
Sewage discharge • 1
Hazardous waste site 1
Solid waste disposal 1
Nitrates 1
Pesticides, rodenticides, fungicides
(e.g., mixing or storage facilities) 1
Mining, oil, or gas activities 1
Petroleum products (e.g., auto repair shops) 1
Solvents (e.g., dry cleaners) 1
Septic systems or other sewage discharges . 1
Other (Please soecifv) 1
1
within 2 miles
1
r
urce of contamination 1
f your water supply? •
NO t
2
2
*
f
2
2 f
2 *
2
2 I
1
2
2 1
2
2 I
2 1
2
11
-------
26. Who performs laboratory analysis on your drinking water?
LAB ANALYSIS PROVIDER
Does this provider perform your lab analysis for ._
M«UU/ Ottwr
Inorganics?
YES
The state 1
A private firm 1
In-house employees 1
Other (Spedfv) 1
NO
2
2
2
2
Microbial*?
YES
1
1
1
1
NO
2
2
2
2
VOC«*7
YES NO
1 2
1 2
1 2
1 2
Organic*?
YES NO
1 2
1 2
1 2
1 2
•VOCs»yote(00 organic confounds (e.g., carton
totrachtorid*. tenzirw, TWMs, 0(Cj
27. How do you pay for your laboratory analysis?
PAYMENT METHOD
Do you use this payment method?
YES NO
Direct payment for tests to state or private lab 1
Included as part of state permit 1
Don't pay 1
Other (Please specify) 1
2
2
2
2
PART H - FINANCIAL INFORMATION
REVENUES AND EXPENSES
28. Are your financial reports or income and expense statements for your drinking water system
completed in accordance to Generally Accepted Accounting Principles (GAAP)?
(Circle one number)
T
1 Yes
2 No
3 Don't have separate income and expense
statements for our drinking water system
4 Don't know
12
-------
I
I
To simplify your task of providing financial information, please follow the guidelines
below when filling out the remainder of the questionnaire.
PROVIDING ESTIMATES:
The following questions ask for information on drinking water supply operations, exclusive of other
activities with other types of operations. Where possible, please provide exact information from
your system's records. Otherwise provide your best estimate of financial information that is
applicable to your drinking water system only.
ROUNDING:
Please record your dollar amounts to the nearest dollar. DO NOT record fractional
dollars (i.e., dollars and cents).
I
I
I
I
i
I
I
i
I
i
13
-------
29. During the last year [as defined in your response to Question 2(6)] what were your drinking
water system's revenues from water sales for each of the following customer categories:
(If zero, enter '0')
T T
WATER SALES Water Sales Gallons delivered
CUSTOMER CATEGORIES Revenues (in millions)
1. Residential customers $_
2. Commercial customers $_
3. Industrial customers $_
4. Wholesale customers (i.e., those
who redistribute your water
to other users) $_
s. Local municipal government $_
6. Other government customers $_
7. Agricultural customers $_
8. Other (Specify $_
9. TOTAL $
30. Please indicate your drinking water system's revenues during the last year from the other water-
related revenue sources listed below.
(If zero, enter VJ')
T
WATER RELATED REVENUE SOURCE
(EXCLUDING WATER SALES) Revenues
1. Connection fees $
2. Inspection fees $
3. Developer fees ' $
4. Other fees $
s. General fund revenues (operating transfers in) $
6. Interest earnings (on water fund, etc.) $
7. Fines/penalties $
Please specify other water system
revenues (not elsewhere reported)
8. $.
9. $_
14
-------
31. For each customer category listed below, please identify your drinking water system's billing
structure, indicate the-year and percent of the two most recent rate increases, and provide the
number of metered and unmetered active connections.
(If zero, enter "0")
T T T
r
1.
2.
3.
4.
5.
6.
7.
8.
CUSTOMER CATEGORY
Residential customers
Commercial customers
Industrial customers
Wholesale customers (i.e., those
who
redistribute your water to other users)
Local municipal government
Other government customers . .
Agricultural customers
Other (Specify!
Note: The total of all metered
and unmel
Billing
structure
(Circle all
code(s)
from Box 2
that apply)
1234567
1234567
1234567
1234567
1234567
1 234567
1234567
1234567
tered connectlc
Year and percent of
two most recent rate
increases
YR. % YR. %
Number
of active
connections
Metered/Unmetered
/
/
/
/
/
/
/
/
ns should be the same as the
current active connections reported In question 11 (B).
BOX 2 • BILLING STRUCTURE
Metered Charges
CODE Billing Structure
1 Uniform rate
2 Declining block rate
3 Increasing block rate
4 Peak period rate
(e.g., seasonal)
Unmetered Charges
CODE Billing Structure
5 Separate flat rate for water
6 Combined flat rate for water and other services
(e.g., rental fees, association fees, pad fees)
Other Type of Charges
CODE
7
Billing st
Other (Sp
ructure
ec/M
I
I
I
I
I
I
t
I
I
15
I
I
1
I
I
-------
32. How many gallons (or dollar equivalents) of uncompensated usage did your water system have
in the last year for each of the usage categories listed below:
UNCOMPENSATED USAGE CATEGORY
1. Free service to municipal buildings and parks.
2. Fire protection, street cleaning,
hydrant flushing
3. Leaks, breaks, failed meters
4. Uncoltocted bills
5. Other (Specify)
Uncompensated usage
(Enter either millions of gallons
Q[ dollar equivalent, if gallons unknown)
million gals, or $
..million gals, or $_
.million gals, or $_
.million gals, or $_
.million gals, or $.
The next question is intended to account for all of your drinking water expenses. Please list
your
=*> Routine operating expenses in Part A;
=*• Capital-related expenses (including interest or
principal repayment) in Part B; and
«* Other expenses in Part C.
16
-------
I
33A. Please enter the routine operating expenses of your drinking water system in the last year, according
to the operating expense categories listed below:
PART A
OPERATING EXPENSES
Last year's expenses
17. TOTAL ALL OPERATING EXPENSES $
B. Please enter the amount of debt service expenditures for your drinking water system in the last year.
PART B
DEBT SERVICE EXPENDITURES
C. Please enter the amount of other expenses (excluding operating and debt service expenses reported
in Parts A and 8) for your drinking water system in the last year.
PARTC
OTHER EXPENSES
I
DIRECT COMPENSATION (wages, salaries, bonuses, etc
1. Managers ..................... $ _
2. Operators ..................... $ _ I
3. Others ....................... $ _ •
4. Benefits (hearth & insurance premiums, FICA,
FUTA, and pension contributions) $ I
"IERGY COSTS:
Electricity $
Other energy (gas, oil, etc.) $ I
ENERGY COSTS:
5. Electricity
6.
CHEMICALS:
7. Disinfectants ................... $ _ L
8. Precipitant chemicals ............... $ _ |
9. Other chemicals .................. $ _
10. Materials and supplies ................ $ _ ^ k
11. Outside analytical lab services ............ $ _ I
12. Other outside contractor services ........... $ _
13. Depreciation expenses ................ $ _ |
u. Water purchase expense "
G raw water Q treated water ........ $ _
15. Payments in lieu of taxes or other •
cash transfers out .................. $ _ •
16. Other operating expenses (general and administra-
tive expenses not reported elsewhere) ........ $ _ k
I
is. Interest payments $ •
19. Principal payments $ "
20. Other debt service
expenditures (Specify) $ • I
21. TOTAL ALL DEBT SERVICE
EXPENDITURES $
I
I
22. Capital improvements (e.g., _
expansion, new treatment) $ •
23. Advance contributions to sinking funds $
24. Other (Specify) $ _
25. TOTAL OTHER EXPENSES $ |
26. TOTAL ALL EXPENSES (FROM PARTS A-C). . . . $
I
-------
1
" I ASSETS, UAKirnES,:DE8tf I
34. Please provide the following information on your drinking water system's total assets and
• liabilities, outstanding debt, and total capital reserve fund.
1 *
1 1. TOTAL ASSETS i.
2. TOTAL LIABILITIES $_
I TOTAL DEBT OUTSTANDING:
DIRECT NET DEBT (see definition belowf.
1 3. Due within 5 years $_
4. Longer than 5 years 5L
I s. Revenue Bond Debt J_
1 6. All Other Debt $_
7. TOTAL CAPITAL RESERVE FUND S
Amount at end
of last year
1
{DEFINITION:
Direct Net Debt • Gross direct debt (owed directly by a jurisdiction) less debt that is self-supporting
(revenue bonds) and double-barreled bonds (general obligation bonds secured by earmarked revenues
I which flow outside the general fund).
1 P ,„;„•,,„ •'•••••" ',,n;£:ii£ii'i™:l|
I 35. Have you paid for major capital improvements, repairs or expansion since January 1 , 1 987?
T
11 Yea
2 No — > Go to Question 37
I
I
I
!
I
-------
36. What sources of funds did you use to pay for these major capital improvements, repairs, or
expansion?
;•
-------
I
I
I
I
I
I
I
I
38. Have your bonds ever been rated by a rating service?
T
1 Yes
2 No
Go to Question 39C
T
RATING SERVICE Rating
I 39A. What was your system's latest bond rating?
I
• Moody's /_/__/_y__/ (e.g., Baal)
Standard and Poor's /__/__/__/__/ (e.g., BBB-t-)
Other (Specify).
39B. What was the year of your system's latest bond rating?
T
19
39C. What was the type of bond that was last issued by your system?
I (Circle one number)
T
I 1 Revenue or industrial development bond
2 General obligation bond
3 Other (Specify)
I
40. Please enter any additional comments (optional):
THANK YOU FOR COMPLETING THIS QUESTIONNAIRE. YOUR
TIME AND EFFORT ARE GREATLY APPRECIATED.
MAILING INSTRUCTIONS ARE INSIDE THE FRONT COVER.
20
-------
Appendix D: Weighted Regression Methodology
-------
This page intentionally left blank
-------
Appendix D
Community Water Systems Survey - Item Non-response Adjustment
Introduction
In support of developing methodology to calculate national estimates from CWS accounting for item
level non-response and to develop models, namely regression, and incorporating the probability design of
the CWS, SA1C reviewed six data files, the Community Water System Survey Database Documentation:
Codebooks and Supporting Documents, January, 31,1997, and the Community Water System Survey
Volume II: Detailed Survey Result Tables and Methodology Report published in January 1997 report.
Based on the review of the CWS Survey documentation it was evident that there were several iterations
and adjustments incorporated into the "final" weight calculations. These included aggregation,
classification, and trimming, to name a few. To ensure adequate understanding of the weights and
documentation associated with the CWSS, SAIC generated several estimates documented in the Results
Tables reported in the Volume II document cited above. In reviewing these estimates, SAIC verified that
estimates reflect responses to individual questions and no imputation was performed for those systems
that did not provide a response to each question of interest. A cursory review of all of the tables reported
do not reveal many national total estimates. For the tables identified reporting national totals, all of the
1980 survey respondents were able to be classified into the categories of interest. Therefore, no item non-
response adjustment "directly" was required. It appears that all other results are presented as proportions
or means calculated using responses. Thus, it is implied that the non-respondents demonstrate, on
average, similar characteristics as the respondents and therefore the estimates are interpreted as reflecting
the status of the entire national population being considered. This approach is consistent with standard
statistical practice.
To generate national estimates of totals or incorporate the probability design into modeling efforts, item
level non-response should be taken into account. The ramifications of not addressing item level non-
response, typically, is an under estimation of the population characteristics. The general approach
adopted for addressing item level non-response is to calculate the average characteristic and/or the
proportion of the population having the characteristic of interest using the respondents, then project the
results to the entire sampled population. Or, as demonstrated in the CWS documentation in addressing
primary sample unit (PSU) non-response, a non-response adjustment factor is calculated and the weights
scaled to account for the item non-response. The underlying assumption of each of these approaches is
that the non-respondents, on average, behave similarly to the respondents.
As mentioned, in application there are two ways to incorporate item non-response into estimated totals or
modeling efforts. One is, for each question, calculate an item non-response adjustment factor. Then
multiply the weight by the adjustment factor. Using the adjusted weights, national estimates are
calculated as the sum of the product of the adjusted weights and the item of interest. Another way is to
calculate the average or proportion based on the item respondents. Then project the national average or
national proportion to the population by multiplying the estimate
Geometries and Characteristics of D-I December 2000
Public Water Systems
-------
and the total number of systems in the population with the characteristic of interest. This second
approach is not necessarily appropriate for modeling efforts.
SAIC was tasked to model flows' (daily production, peak design) with the population served
incorporating into the model the probability design. Appendix A presents a summary of the information
reported in the CWS survey databases by stratum, final weight, number of survey respondents, proportion
of those sampled within each stratum with each weight, the item non-response adjustment factor for the
questions of interest1, and the adjusted weight incorporating the non-response.
1.1. Methodology
The following documents the methodology adopted to perform a weighted regression for Average Daily
production, peak, and design flows for ground, surface, and purchased water systems, by ownership
(public or private) from the CWSS. Note that the weight used is an adjusted weight rather than the
original weight.
Step 1 : Adjusted weights for each respondent was calculated using the final weights in the CWSS
survey. Average daily production flow for each system was calculated as follows :
Average daily production flow = Sum of Ground, Surface, and Purchased water flows ( as
reported in Question 4 of the CWSS survey) less the wholesale
flow as reported in Question 29, 4B of the survey.
Systems with zero or missing average daily flows were considered as non -respondents. 16 data points
were considered as outliers and were excluded. Out of 1980 systems, data for 1718 systems were thus
considered. Adjusted weights were calculated to account for the non-respondents.
For example, if n systems (each with a weight of w) out of x systems for a particular weight category had
a positive daily production flow, then the adjusted weights for each system in that category was
computed as w* (x /n) where (x/n) is the item level non-response factor.
Step 2 : Probability of selection was computed for each system as :
Probability for any system = Adjusted weight for the system / sum of adjusted weights
Step 3: Calculate the intercept and the slope estimators2. For a sampling design in which the inclusion
probability for the ith unit is n, ,for i = 1, ,N, an unbiased Horvitz-Thompson estimator'uy = (£ y, / n,
) / N where the summation is over v distinct units in the sample. With an auxiliary variable x, the
Horvitz-Thompson estimator based on the sample x- values, ]ix= (Z x, / TT, ) / N is an unbiased estimator
of the population mean of the x values.
1 Daily production flow = sum of Ground (Q4AA), Surface (Q4BA), and Purchased (Q4CA) flows in
kgal/day minus Q29_4B (wholesale flow) in kgal/day. If daily production flow is greater than 0, then it is
considered as a response. Otherwise, it is considered a non-response.
" Thompson, Steven. Sampling. Wiley-Interscience Publication. 1992. p. 82
Geometries and Characteristics of D-2 December 2000
Public Water Systems
-------
A generalized regression estimator, which is approximately (or asymptotically) unbiased for the
population mean under the given design, is |tc = uy+ B(ux - ftx)
where Bis a weighted regression slope estimator, based on the inclusion probabilities, given by
v X, v V v X, 2
XV 7 = 1 71, / = ! 7T, v v- , = l 7T
v 1 1=1 n, v 1
I— ' I —
/ = ! 71, , = 17C
An appropriate expression for the mean square error or variance of uc is obtained by using
i v -A-Bx
N ,=i 7t.
By defining the new variable y', = (y, - A - Bx, ) , the Horvitz - Thompson variance formula may be used,
giving
N 1 ~7C ^' 7C ~ 71 71
7I;7I
An estimator of this variance is obtained using y = y, - A- Bx,, in the Horvitz - Thompson variance
estimation formula, where A is a weighted regression estimator given by
v V - v X
v 1
I —
, = 131
1.2 Results
Using the methodology documented in Section 1.1, the results for Ground, Surface, and Purchased Water
systems were generated. The equations are listed in Table 1 documenting the linear regression models
based on the weighted and unweighted CWS survey response information.
Geometries and Characteristics of D-3 December 2000
Public Water Systems
-------
Table 1. Modeled Average Daily Production, Peak, and Design Flows
for Ground, Surface, and Purchased Water Systems
by Ownership (public or private)
Modeled Flow
Type of System
Models
Daily Production
Ground
Unweighted
Public & Private -
ln(daily production flow)
Public -
2.7721 + 1.0809*ln(pop)
2.5235+ 1.0629*ln(pop)
ln(daily production flow) = - 2.8316 + 1.0744*ln(pop)
ln(daily production flow)
Private
Weighted :
Public & Private -
ln(daily production flow)
Public
2.74444+ 1.07652*in(pop)
2.45627 + 1.05839*ln(pop)
ln(daily production flow) = - 2.70748 + 1.06284*ln(pop)
ln(daily production flow)
Private -
Surface
Unweighted
Public & Private -
ln(daily production flow)
Public -
2.2657+ 1.0278*ln(pop)
2.0607+ 1.0123*ln(pop)
ln(daily production flow) = - 2.3122 + 1.0201*ln(pop)
ln(daily production flow)
Private
Weighted :
Public & Private -
ln(daily production flow)
Public
2.24998+ 1.02058*ln(pop)
1.96581 + 0.99703*ln(pop)
ln(dai!y production flow) = -2.40391 + 1.03338*ln(pop)
ln(daily production flow)
Private -
Geometries and Characteristics of
Public Water Systems
D-4
December 2000
-------
Table 1. Modeled Average Daily Production, Peak, and Design Flows
for Ground, Surface, and Purchased Water Systems
by Ownership (public or private)
Modeled Flow
Type of System
Models
Purchased
Unweighted
Public & Private -
ln(daily production flow)
Public
2.9031 + 1.0758*ln(pop)
2.8548+ 1.0758*ln(pop)
ln(daily production flow) = - 2.8221 + 1.0551*ln(pop)
ln(daily production flow)
Private -
Weighted :
Public & Private -
ln(daily production flow)
Public -
3.09118+ 1.10117*ln(pop)
3.05937+ 1.10189* In(pop)
ln(daily production flow) = - 2.99497 + 1.08339*In(pop)
ln(daily production flow)
Private -
Daily Peak
Ground
Unweighted
Public & Private -
ln(daily peak flow) = -2.3231+ 1.0857*ln(pop)
Public -_
ln(daily peak flow) = -1.9338+ 1.0553*ln(pop)
Private -_
ln(daily peak flow) = - 2.4798 + 1.0890*ln(pop)
Weighted :
Public & Private -
ln(daily peak flow) = -2.26049+ 1.07599*!n(pop)
Public -_
ln(daily peak flow) = -1.75484 + 1.03750*ln(pop)
Private ;_
ln(daily peak flow) = - 2.39872 + 1.07971*ln(pop)
Geometries and Characteristics of
Public Water Systems
D-5
December 2000
-------
Table 1. Modeled Average Daily Production, Peak, and Design Flows
for Ground, Surface, and Purchased Water Systems
by Ownership (public or private)
Modeled Flow
Type of System
Models
Surface
Unweighted
Public & Private -
ln(daily peak flow)
Public -
1.6510+ 1.0133*ln(pop)
1.2907+ 0.9854* In(pop)
ln(daily peak flow) = - 1.8215 + 1.0116*ln(pop)
In(daily peak flow)
Private -
Weighted :
Public & Private -
ln(daily peak flow)
Public -
1.66512+ 1.00856*ln(pop)
1.24242+ 0.97694*ln(pop)
ln(daily peak flow) = - 1.79922 + 1.01119*ln(pop)
ln(daily peak flow)
Private
Purchase
Unweighted
Public & Private -
ln(daily peak flow)
Public -
2.9590+ 1.1009*!n(pop)
2.9033+ 1.1011*ln(pop)
ln(daily peak flow) = - 2.8565 + 1.0753*ln(pop)
ln(daily peak flow)
Private
Weighted :
Public & Private -
ln(daily peak flow)
Public
3.16121 + 1.13044*ln(pop)
3.10593+ l.I3085*ln(pop)
ln(daily peak flow) = -3.01829+ 1.10391*ln(pop)
ln(daily peak flow)
Private -
Geometries and Characteristics of
Public Water Systems
D-6
December 2000
-------
Table 1. Modeled Average Daily Production, Peak, and Design Flows
for Ground, Surface, and Purchased Water Systems
by Ownership (public or private)
Modeled Flow
Type of System
Models
Daily Design
Ground
Unweighted
Public & Private -
ln(daily design flow)
Public
0.7810+ 0.967 l*ln(pop)
0.6245 + 0.9593*ln(pop)
ln(daily design flow) = - 0.7989 + 0.9544*ln(pop)
ln(daily design flow)
Private -
Weighted :
Public & Private -
ln(daily design flow)
Public -
0.92536 + 0.97708*ln(pop)
0.59798 + 0.95538*ln(pop)
ln(daily design flow) = - 0.87509 + 0.96078*In(pop)
ln(daily design flow)
Private -
Surface
Unweighted
Public & Private -
ln(daily design flow)
Public -
0.7027 +0.9549* In(pop)
0.3744 + 0.9326*ln(pop)
ln(daily design flow) = - 0.7637 + 0.9337*ln(pop)
!n(daily design flow)
Private -
Weighted :
Public & Private -
ln(daily design flow)
Public
0.99503 + 0.97757*ln(pop)
0.52715+ 0.94573*ln(pop)
ln(daily design flow) = - 1.03074 + 0.96188*ln(pop)
ln(daily design flow)
Private -
Geometries and Characteristics of
Public Water Systems
D-7
December 2000
-------
Table 1. Modeled Average Daily Production, Peak, and Design Flows
for Ground, Surface, and Purchased Water Systems
by Ownership (public or private)
Modeled Flow
Type of System
Models
Purchased
Unweighted
Public & Private -
ln(daily design flow) = - 2.9196 + 1.1052*ln(pop)
Public ;_
ln(daily design flow) = - 2.8445 + 1.1028*ln(pop)
Private -_
ln(daily design flow) = - 2.8610 + 1.0863*ln(pop)
Weighted :
Public & Private -
ln(daily design flow) = -3.11425 + 1.13024*ln(pop)
Public -_
ln(daily design flow) = - 3.06460 + 1.12951*ln(pop)
Private ;_
ln(daily design flow) = - 3.02199 + 1.11256*In(pop)
Geometries and Characteristics of
Public Water Systems
D-8
December 2000
-------
US EPA / ICR - Drinking Water (CWSS) Phase II . 13-17 21AUG98 M \PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Count Item
Primary Total tt Within Non-Resp
Pop. Water of PSU Est. CWSS each % of # of Factor Adjusted
Categories Ownership Source Respondents Tot # Final Weight Weight Total Responses {Q4 & Q29-4B) Weight
< = 100 PUBLIC GROUND 37 899 4 2 5 41 1 20 8.0
5 2 5 41 1 2.0 10 0
13 1 2.70 1 1.0 13.0
17 1 2.70 1 10 17.0
19 23 62.16 17 14 25.7
38 1 2 70 1 1.0 38.0
51 4 10 81 4 10 51 0
56 2 5.41 2 1.0 56.0
60 1 2 70 1 10 60 0
<=100 PUBLIC SURFACE 11 56 4 5 45 45 2 25 10.0
6 6 54 55 1 6.0 36 0
<«100 PUBLIC PURCHASED 12 127 5 6 50 00 6 10 50
12 1 8 33 1 1.0 12.0
14 2 16.67 2 10 14 0
19 3 25 00 3 10 19.0
<=100 PRIVATE GROUND 60 5307 5 11 67 1 10 50
78 7 11 67 3 23 182.0
82 4 6.67 3 1.3 109 3
92 46 76.67 33 14 128 2
98 2 3.33 1 2.0 196.0
<=100 PRIVATE SURFACE 24 204 8 20 83 33 14 14 11.4
11 4 16.67 2 2.0 22 0
<«100 PRIVATE PURCHASED 13 63 4 3 23.08 3 1.0 4 0
5 9 69 23 9 10 50
6 1 7 69 1 10 6.0
<«100 ANCILLARY GROUND 62 7008 24 1 1 61 1 1.0 24.0
56 4 6 45 4 1.0 56 0
63 1 1 61 1 1.0 63 0
72 1 1 61 1 10 72.0
105 5 8 06 2 25 262 5
122 50 80 65 35 14 174 3
<=100 ANCILLARY SURFACE 17 90 5 15 88 24 11 1.4 6.8
6 1 5 88
9 1 5 88 1 10 9.0
< = 100 ANCILLARY PURCHASED 9 45 5 9 100 00 S 1.0 5.0
101 - 500 PUBLIC GROUND 76 3935 4 1 1.32
19 3 3 95 3 10 19 0
35 3 3 95 2 15 52.5
38 3 3.95 3 10 38 0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of D-9 December 2000
Public Water Systems
-------
US EPA / ICR - Drinking Water (CWSS) Phase II : 13.17 21AUG98 M-\PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Pop
Categories
Primary
Water
Ownership Source
Count Item
Total # Within Non-Resp
of PSU Est. CWSS each % of # of Factor
Respondents Tot # Final Weight Weight Total Responses (Q4 & Q29-4B)
PUBLIC
GROUND
76
3935
101 - 500
101 - 500
101 - 500
39
40
45
47
49
50
51
54
56
60
70
3
4
5
11
12
14
19
34
47
3
4
5
10
11
12
13
14
17
31
47
50
56
4
5
11
18
19
21
32
45
73
77
78
82
1
1
2
1
1
2
4
1
42
9
2
1
2
4
2
8
13
1
1
1
1
2
3
2
3
9
1
22
1
1
1
1
5
1
1
1
2
1
2
1
1
3
5
6
34
1.
1.
2
1
1 .
2.
5
1
55
11
2
3.
6.
12.
6
24.
39
3
3.
3.
1
3
5
3
5
17
1
42
1
1
1
1
9
1
1
1
3
1
3
1
1
4
7
9
51
.32
.32
.63
32
.32
.63
.26
32
.26
.84
.63
03
.06
.12
06
24
39
03
.03
.03
.92
85
.77
.85
.77
31
.92
.31
92
.92
92
.92
.62
52
52
52
.03
52
03
52
52
.55
58
.09
52
1
1
2
1
1
2
3
1
39
5
2
1
1
3
1
4
10
1
1
2
3
2
3
9
1
22
1
1
1
1
5
1
1
2
1
2
1
1
3
4
4
29
1 0
1 0
1.0
1.0
1.0
1 0
1.3
1 0
1.1
1.8
1 0
1.0
2.0
1.3
2.0
2 0
1.3
Adjusted
Weight
39.0
40.0
45.0
47.0
49.0
50 0
68.0
54.0
60 3
108 0
70 0
3.0
8.0
6.7
22.0
24 0
18.2
1 .
1 .
1
1
1 .
1.
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
.0
.0
0
0
.0
.0
0
0
.0
.0
.0
0
0
.0
0
0
.0
.0
.0
0
0
.3
5
.2
3.
4
5
10
11.
12.
13
14
17.
31.
47.
50
56
4
5
18
19
21
32
45
73
96
117
96
0
0
0
0
0
0
0
0
0
0
0
0
0
.0
0
.0
0
0
0
.0
.0
.3
0
1
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-IO
December 2000
-------
US EPA / ICE - Drinking Water (CWSS) Phase II • 13-17 21AUG98 M \PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Count Item
Primary Total # Within Non-Resp
Pop. Water of PSU Est. CWSS each % of tt of Factor
Categories Ownership Source Respondents Tot # Final Weight Weight Total Responses (04 & Q29-4B)
101 - 500
101 - 500
101 - 500
101 - 500
PRIVATE GROUND
PRIVATE SURFACE
PRIVATE PURCHASED
ANCILLARY GROUND
4808
101 - 500
101 - 500
ANCILLARY SURFACE
ANCILLARY PURCHASED
501 - 1,000 PUBLIC GROUND
501 - 1,000 PUBLIC SURFACE
92
98
5
6
7
14
19
24
7
8
9
14
23
5
6
49
53
56
58
63
105
122
7
a
10
5
6
4
17
20
35
37
40
45
46
50
51
56
60
5
7
5
3
2
20
2
1
1
3
29
8
2
1
8
2
2
1
1
51
1
5
5
4
2
13
2
8
1
1
1
1
50
1
5
2
5
4
2
3
1
4
1
7
4
6
68
6
3
3
10
60
16
4
2
16
2
2
1
1
70
1
6
6
5
11
76
11
88
11
1
1
1
65.
1
6
2
6
5
2
3
1
13.
3.
.58
.55
90
.97
90
45
.45
34
42
67
.17
08
.67
.78
78
39
39
.83
39
94
.94
.56
76
47
.76
.89
11
.32
32
32
.79
32
58
63
58
26
63
95
.32
.79
.45
4
3
2
11
1
1
2
29
8
2
1
8
2
2
1
1
40
1
3
4
2
1
7
2
8
1
1
1
47
5
2
5
4
2
2
1
4
1
1 3
1.0
1 0
1 8
2.0
1 0
1.0
1 0
1.0
1.0
1 0
1 0
1 0
1.0
1 0
1.3
1.0
1 7
1.3
2 0
2.0
1 9
1 0
1 0
1 0
Adjusted
Weight
115 0
98.0
5.0
10 9
14 0
14.0
36.0
7.0
8 0
9.0
14 0
23.0
5 0
6.0
49 0
53.0
71.4
58 0
105.0
131 3
244.0
14.0
14 9
10.0
5 0
6.0
1 0
1 1
1 0
1.0
1 0
1 0
1 0
1.5
1 0
1 0
1 0
20
37
40
45
46
SO
51
84
60
5.
7 .
.0
2
.0
0
.0
.0
0
.0
0
.0
.0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-ll
December 2000
-------
US EPA / ICR - Drinking Water (CWSS) Phase II 13-17 21AUG98 M:\PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Pop.
Categories
501 - 1,000
501 - 1,000
501 - 1,000
501 - 1,000
501 - 1,000
Ownership
PUBLIC
PUBLIC
PRIVATE
PRIVATE
PRIVATE
Primary Total #
Water of PSU Est . CWSS
Source Respondents Tot tt Final weight
SURFACE 29 341 10
14
19
20
35
PURCHASED 46 791 3
5
7
10
11
12
14
20
21
23
35
41
46
56
GROUND 61 1892 4
5
18
19
20
21
32
39
50
73
74
77
82
SURFACE 18 88 4
5
8
9
10
PURCHASED 41 311 4
5
6
9
10
11
13
Count
Within
each
Weight
15
5
1
2
1
1
3
2
18
2
1
3
2
1
6
3
1
1
2
2
4
1
3
2
32
3
2
1
2
2
3
4
14
1
1
1
1
22
4
1
4
1
2
1
%
of
Total
51.
17
3.
6.
3
2
6
4
39
4
2
6.
4.
2.
13.
6.
2
2
4
3
6
1
4
3
52
4
3
1
3
3
4
6
77
5
5
5
5
53
9
2
9
2
4
2
72
24
45
.90
45
17
52
35
.13
35
.17
.52
.35
. 17
.04
.52
.17
.17
.35
28
56
64
92
28
46
92
28
64
28
28
.92
.56
.78
56
.56
56
.56
66
76
44
76
44
88
44
t* of
Responses
13
4
1
1
1
1
3
2
18
2
1
3
2
1
6
3
1
1
2
2
4
1
3
2
26
3
1
1
2
2
3
4
12
1
1
1
22
4
1
4
1
2
1
Item
Non-Resp
Factor
(04 i. Q29-
1.
1
1.
2.
1.
1
1
1
1
1
1.
1
1
1.
1
1
1
1 .
1
1
1.
1
1.
1
1
1
2.
1
1
1.
1.
1
1
1
1
1.
1
1.
1
1
1
1
1
4B)
2
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
Adjusted
Weight
11.5
17 5
19.0
40.0
35.0
3 0
5.0
7 0
10.0
11.0
12.0
14.0
20.0
21.0
23.0
35 0
41.0
46 0
56.0
4 0
5.0
18.0
19.0
20.0
25.8
32.0
78.0
50 0
73.0
74 0
77 0
82 0
4 7
5 0
8 0
9 0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-12
December 2000
-------
US EPA / ICR - Drinking Water (CWSS) Phase II : 13 17 21AUG98 M \PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Count Item
Primary Total ft within Kon-Resp
Pop Water of PSU Est. CWSS each % of tt of Factor Adjusted
Categories Ownership Source Respondents Tot ft Final Weight Weight Total Responses (Q4 & Q29-4B) weight
501 - 1,000 PRIVATE PURCHASED 41 311 14 2 4 88 2 1.0 14 0
22 4 9 76 4 10 22 0
501 - 1,000 ANCILLARY GROUND 4 217 49 1 25.00
56 3 75.00 2 15 84 .0
501 - 1,000 ANCILLARY SURFACE 1 55 1 100 00 1 10 50
1,001 - 3,300 PUBLIC GROUND 79 3515 23 2 2 53 2 10 23 0
10 35 0
1.3 52 0
10 40 0
10 41.0
1.0 45.0
10 46 0
1.0 47 0
10 53 0
10 54 0
1,001 - 3,300 PUBLIC SURFACE 46 984 4 12 17 1 10 4.0
10 50
2.0 20.0
10 18 0
1 0 20.0
1.1 26 2
10 46 0
1,001 - 3,300 PUBLIC PURCHASED 39 734 4 2 5 13 2 10 40
10 50
1.0 10 0
1.0 14 0
1 0 18.0
1.0 20 0
10 21 0
1 0 23.0
10 28 0
1 0 46.0
10 47 0
1,001 - 3,300 PRIVATE GROUND 62 1215 4 4 6.45 4 1.0 4 0
20 10 0
10 10 0
10 14 0
1.0 15 0
1 0 16.0
11 19 1
10 19 0
13 25 0
23
35
39
40
41
45
46
47
53
54
4
5
10
18
20
23
46
4
5
10
14
18
20
21
23
28
46
47
4
5
10
14
15
16
18
19
20
2
3
4
4
1
2
57
4
1
1
1
1
2
5
3
33
1
2
3
5
4
1
5
1
15
1
1
1
4
2
2
1
1
1
34
4
5
2
3
5
5
1
2
72
5
1
1
2
2
4
10
6
71
2
5
7
12
10
2
12
2
38
2
2.
2
6
3
3
1
1
1.
54.
6
8
53
.80
.06
.06
.27
.53
15
06
27
.27
17
.17
35
.87
.52
.74
17
13
.69
82
26
.56
82
56
.46
56
.56
.56
.45
23
23
61
61
.61
.84
.45
06
2
3
3
4
1
2
57
4
1
1
1
1
1
5
3
29
1
2
3
5
4
1
5
1
15
1
1
1
4
1
2
1
1
1
32
4
4
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of D-13 December 2000
Public Water Systems
-------
US EPA / ICR - Drinking Water (CUSS) Phase II : 13.17 21AUG98 M:\PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Pop.
Categories
1,001 - 3,300
1,001 - 3,300
1,001 - 3,300
Primary
Water
Ownership Source
Count Item
Total # Within Non-Resp
of PSU Est. CWSS each % of (t of Factor
Respondents Tot # Final Weight Weight Total Responses (Q4 & Q29-4E)
PRIVATE
GROUND
62
1215
1,001 - 3,300
3,301 - 10,000
ANCILLARY GROUND 1
PUBLIC GROUND 74
46
1717
3,301 - 10,000
21
32
43
51
82
3
4
5
6
8
19
3
4
5
7
8
10
17
18
20
22
24
2
1
3
1
1
2
1
3
2
12
1
2
4
1
2
13
2
1
5
1
2
1
3
1
4
1
1
9
4
14
9
57
4
5
11
2
5
38
5
2
14
2
5
2
23
61
.84
.61
.61
.52
.76
29
52
.14
76
.88
.76
.94
88
24
.88
94
71
94
88
94
2
1
3
1
1
1
1
1
10
1
2
4
1
2
13
2
1
5
1
2
1
6
10
16
17
18
20
21
22
23
24
45
46
5
15
18
19
20
21
23
1
2
1
1
2
2
1
1
58
1
3
1
1
5
23
2
2
2
3
1
2
1
1
2
2.
1
1
78
1
4
1
2
13
60
5
5
5
7
35
70
35
.35
.70
.70
35
35
38
.35
.05
35
63
16
53
26
.26
.26
89
1
2
1
2
2
1
1
56
1
3
1
5
23
2
2
2
3
1.0
1.0
1 0
1.0
1.0
1.0
3.0
2 0
1 2
1 0
1.0
1.0
1.0
Adjusted
Weight
21.0
32.0
43.0
51.0
82.0
4.0
15.0
12.0
9.6
19.0
3 0
4 0
5.0
7 0
8 0
10.0
17.0
18.0
20 0
22 0
24.0
46.0
1.0
1.0
1 0
1 0
1.0
1 0
1.0
1 0
1.0
1.0
1 0
1 0
1 0
1.0
1.0
1.0
1 0
6.
10
16.
18.
20.
21
22.
23.
24
45
46.
15
18
19
20
21
23
.0
0
.0
.0
0
0
.0
e
0
0
.0
.0
0
0
.0
.0
.0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-14
December 2000
-------
US EPA / ICR - Drinking Water (CWSS) Phase II : 13 17 21AUG98 M:\PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented strata by Final Weight
Pop.
Categories
3,301 - 10,000
Primary
Water
Ownership Source
PUBLIC PURCHASE
Count Item
Total # Within Non-Resp
of PSU Est CWSS each % of tt of Factor
Respondents Tot # Final Weight Weight Total Responses (04 & 029-4B)
38
791
3,301 - 10,000
3,301 - 10,000
3.301 - 10,000
10,001 - 50,000
PRIVATE
11
15
17
18
21
22
23
28
42
45
2
3
4
5
6
8
13
14
16
18
21
33
43
2
3
5
8
9
17
3
4
7
8
14
19
2
4
15
16
17
20
22
23
44
1
1
1
20
1
1
10
1
1
1
5
4
46
1
1
2
1
1
4
9
3
1
1
2
17
1
2
1
3
14
5
2
4
1
1
1
1
3
2
49
3
3
3
1
2
2
2
52
2
2
26
2
2
2
6
5
58
1
1
2
1
1
5
11
3
1
1
7
65
3
7
3
11
51
18
7
14
3
3
1
1
4
3.
74.
4 .
4
4
1 .
63
.63
63
63
.63
.63
32
.63
.63
.63
.33
.06
23
.27
.27
.53
27
27
.06
39
80
.27
.27
69
.38
85
69
.85
.54
85
.52
.41
81
70
70
.52
.52
55
03
.24
.55
55
55
.52
1
1
1
20
1
1
10
1
1
1
5
4
45
1
1
2
1
4
8
3
1
1
2
15
1
2
1
3
14
5
2
4
1
1
1
1
3
2
47
3
3
3
1
ir
>9-4B)
1 0
1 0
1 0
1.0
1.0
1 0
1 0
1.0
1 0
1 0
1.0
1 0
1 0
1.0
1 0
1 0
1 0
1 0
1.1
1 0
1 0
1 0
1.0
1.1
1.0
1 0
1.0
1 0
1.0
1 0
1 0
1 0
1.0
1 0
1 0
1.0
1.0
1 0
1.0
1.0
1 0
1 0
1 0
Adjust'
5d
Weight
11
15
17
18
21
22
23
28
42
45
2
3
4
5
6
8
13
16
20
21
33.
43
2.
3.
5.
8.
9
17.
3 .
4.
7
8.
14
19
2
4
15
16.
17
20
22
23
44
0
.0
.0
.0
.0
.0
.0
0
.0
0
0
0
. 1
0
0
0
0
,0
3
.0
.0
0
.0
.4
.0
.0
.0
.0
.0
.0
.0
.0
0
0
0
0
0
0
7
0
0
0
0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-15
December 2000
-------
US EPA / ICR - Drinking Water (CWSS) Phase II 13 17 21AUG98 M:\PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Pop.
Categories
10,001 - 50,000
Primary
Water
Ownership Source
Count Item
Total # Within Non-Resp
of PSU Est. CWSS each % of # of Factor
Respondents Tot M Final Weight Weight Total Responses (Q4 & Q29-4B)
PUBLIC
SURFACE
57
827
10,001 - 50,000
10,001 - 50,000 PRIVATE GROUND
10,001 - 50,000 PRIVATE SURFACE
10,001 - 50,000 PRIVATE PURCHASED 10
50,001 - 100,000 PUBLIC GROUND
50,001 - 100,000 PUBLIC SURFACE
50,001 - 100,000 PUBLIC PURCHASED 22
3
11
14
15
16
17
18
21
3
11
15
17
18
2
3
4
10
15
19
2
3
4
2
3
4
2
3
10
12
17
1
2
3
5
6
11
15
16
1
2
3
4
1
2
40
4
2
2
2
1
1
14
2
2
1
12
36
1
1
1
1
21
3
2
6
2
33
3
4
1
1
4
1
34
2
1
1
1
1
1
7
9
7
1
3
70.
7
3
3
3
5
5
70
10
10
1
23.
69
1
1
1
4
84
12
20
60
20
78
7
9
2
2.
8
2
75
4
2
2
2
2
4
31
40
02
75
51
.18
.02
51
51
.51
.00
.00
00
.00
00
.92
.08
.23
92
92
92
.00
00
00
.00
00
.00
.57
14
52
38
.38
89
22
56
.44
.22
.22
22
22
55
.82
91
4
1
1
38
4
2
1
2
1
1
14
2
2
1
11
34
1
1
1
21
3
2
6
2
32
3
4
1
1
4
1
34
2
1
1
1
1
7
9
>r
19-4B)
1.0
1.0
2 0
1 1
1 0
1.0
2.0
1 0
1 0
1.0
1.0
1.0
1 0
1.0
1.1
1.1
1 0
1 0
1.0
1 0
1 0
1.0
1 0
1 0
1 0
1 0
1.0
1 0
1.0
1 0
1.0
1 0
1 0
1 0
1 0
Adjusted
Weight
3 0
11 0
28.0
15.8
16.0
17.0
36 0
21 0
3 0
11 0
15.0
17.0
18.0
2 0
3.3
4.2
10 0
15.0
19 0
3.0
4.0
2 0
3.0
4.0
2 1
3 0
10 0
12.0
17.0
1 0
2 0
3 0
5 0
6 0
11.0
1.0
1.0
1 0
1.0
2 0
3.0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-16
December 2000
-------
US EPA / ICR - Drinking Water (CWSS) Phase II : 13-17 21AUG98 M-\PW\ICR-1\REQ33\R33AK11
Appendix C Summary of CWSFINAL Documented Strata by Final Weight
Count Item
Non-Resp
Factor
1.0
1.0
1 0
1.1
1.0
1.0
1.0
1 0
1 0
1 0
1.0
1 1
1 1
1.0
1 0
1.0
1.2
1 0
1 0
1.0
1 0
1 0
1.0
1 0
1.0
1 0
Primary Total tf
Pop Water of PSU Est.
Categories Ownership Source Respondents Tot #
0,001 - 100,000 PUBLIC PURCHASED 22 101
0,001 - 100,000 PRIVATE GROUND 11 23
0,001 - 100,000 PRIVATE SURFACE 15 33
0,001 - 100,000 PRIVATE PURCHASED 4 9
00,001 - 1,000,000 PUBLIC GROUND 31 86
.00,001 - 1,000,000 PUBLIC SURFACE 51 185
100,001 - 1,000,000 PUBLIC PURCHASED 23 74
100,001 - 1,000,000 PRIVATE GROUND 9 18
100,001 - 1,000,000 PRIVATE SURFACE 7 15
cwss
Final Weight
10
11
17
2
3
2
5
2
3
1
2
3
1
2
3
5
6
10
11
1
2
3
6
2
2
3
Within
each
Weight
2
2
1
10
1
14
1
3
1
3
1
27
23
1
5
1
18
1
2
5
3
9
6
9
6
1
%
Of
Total
9
9
4
90
9
93
6
75
25
9
3
87
45
1
9
1.
35
1
3
21
13
39.
26
100
85
14
.09
.09
55
.91
.09
.33
67
.00
.00
68
23
10
10
.96
80
.96
29
.96
92
74
04
.13
09
00
71
29
ft 01
Respon!
2
2
1
9
1
14
1
3
1
3
1
25
21
1
5
1
IS
1
2
5
3
9
6
9
6
1
100,001 - 1,000,000 PRIVATE
Adjusted
Weight
10 0
11 0
17 0
2.2
3.0
2.0
5.0
2 .0
3 .0
1.0
2.0
3.2
1.1
2.0
3.0
5.0
7 2
10 0
11.0
1 0
2 0
3 .0
6.0
2 0
2 0
3 0
Item Non-response Factor computed for Daily Production Flow
Geometries and Characteristics of
Public Water Systems
D-17
December 2000
-------
This page intentionally left blank
-------
Appendix E: Statistical Testing of Regressions for Different Categories of CWS
-------
This page intentionally left blank
-------
Appendix E
Statistical Methodology for Comparing two Straight Lines using a
Single Regression Model
Introduction
Regression equations for systems using Ground, Surface, and Purchased water were computed.
The equations are as follows :
A. Ground Water Sources
Both Public and Private :
Log(Avg. Flow)=-2.7685 + 1.0803*Log(service population) + Err
Public :
Log(Avg. Flow)=-2.5164 + 1.0619*Log(service population) + Err
Private :
Log(Avg. Flow)=-2.8316 + 1.0744*Log(service population) + Err
B. Surface Water Sources
Both Public and Private :
Log(Avg. Flow)=-2.2741 + 1.0289*Log(service population) + Err
Public :
Log(Avg. Flow)=-2.0856 + 1.0153*Log(service population) + Err
Private :
Log(Avg. Flow)=-2.3098 + ] .0198*Log(service population) + Err
C. Purchased Water Sources
Both Public and Private :
Log(Avg. Flow)=-2.9031 + 1.0758*Log( service population) + Err
Public :
Log(Avg. Flow)=-2.8548 + 1.0758*Log(service population) + Err
Private :
Log(Avg. Flow)=-2.8221 + 1.055l*Log(service population) + Err
In response to client's request for further analysis, tests were done to find whether there were significant
differences between private and public systems in each category (i.e., Ground, Surface, and Purchased).
Geometries and Characteristics of E-l December 2000
Public Water Systems
-------
Methodology: Comparisons were done using two different methods.
Method 1
Two new variables were created as follows:
1. A dummy variable (DUMMY) such that DUMMY=1 for public systems and 0 for Private systems
2. A variable (PROD) indicating the product of DUMMY and Log(service population)
Regression analysis was done for systems using Ground, Surface, and Purchased water using the model
log(Avg. Flow)=BetaO + Betal*Log(service population) + Beta2*DUMMY
+ BetaS* Prod + Err
The equations were tested for the following :
Parallelism - whether for any category, there is statistical evidence to beleieving that the
lines are parallel. The null hypothesis will be Beta3=0.
Equal Intercepts - whether the lines have equal intercepts. The null hypothesis will be
Beta2=0.
Coincidence - whether the lines are identical or not. The null hypothesis will be
Beta2=Beta3=0.
Method 2
Least Square Means (LSMEANS) were computed for public and private facilities within each category
(i.e. Ground, Surface, and Purchased). The LSMEANS were then compared.
Results:
Regression analysis using the full model yielded the following equations :
A. Ground Water Sources
Log(Avg. Flow)=-2.831569 + 1.074404*Log(service population) + 0.315131*DUMMY
- 0.012493*log(service population)*DUMMY + Err
B. Surface Water Sources
Log(Avg. Flow)=-2.309820 + 1.019778*Log(service population) + 0.224172*DUMMY
- 0.004478*log(service population)*DUMMY + Err
C. Purchased Water Sources
Log(Avg. Flow)=-2.822093 + 1.055108*Log( service population) - 0.032712*DUMMY
+ 0.020654*log(service population)*DUMMY + Err
Geometries and Characteristics of E-2 December 2000
Public Water Systems
-------
It should be noted that by replacing DUMMY with 1 and 0 for Public and Private Systems respectively
for each category (i.e. Ground, Surface, and Purchased), one gets the same set of equations for public and
private systems described in the introductory paragraph.
Summary of Test Results for Ground Water Systems
The results of the tests are summarized as follows:
Source
Regression (LOGPOP)
Residual
Regression (LOGPOP,
DUMMY)
Residual
Regression
(LOGPOP, DUMMY,
DUMMY*LOGPOP)
Residual
Degrees of
Freedom
1
876
2
874
o
3
873
Sum of Squares
5290.48132
320.14664
5300.57648
310.05147
5300.74142
309.88654
Mean Square
5290.48132
0.36546
2650.28824
0.35434
1766.91381
0.35456
F-Statistic
14476.059
7479.410
4983.381
Statistic for test of parallelism
F(LOGPOP*DUMMY | DUMMY,LOGPOP) =
regression SSffull model) - regression SS (LOGPOP.DUMMY)
MS Residual (full model)
= 5300.74142-5300.57648
0.35456
= 0.4651963 with 1 and 874 degrees of freedom
(P = 0.49539)
Hence we conclude that we have no statistical basis for believing that the two lines are not parallel.
Statisitic for test of equal intercepts
F(DUMMY|/LOGPOP) = regression SSCLOGPOP. DUMMY) - regression SS (LOGPOP)
MS Residual (full model)
= 5300.57648-5290.48132
0.35456
= 28.4723 with 1 and 874 degrees of freedom
(P< 0.0001)
Hence we conclude that there is strong statistical evidence that the intercepts are different.
Statisitic for test of coincidence
F(LOGPOP*DUMMY, DUMMY|LOGPOP)
= (regression SS(full model) - regression SS (LOGPOP)V2
MS Residual (full model)
= (5300.74142 - 5290.48132V2
0.35456
Geometries and Characteristics of
Public Water Systems
E-3
December 2000
-------
= 14.467 with 2 and 874 degrees of freedom
(P< 0.001)
Hence we conclude that we have strong evidence that the lines are not identical.
Summary of Test Results for Surface Water Systems
The results are summarized as follows:
Source
Regression (LOGPOP)
Residual
Regression (LOGPOP,
DUMMY)
Residual
Regression
(LOGPOP, DUMMY,
DUMMY*LOGPOP)
Residual
Degrees of
Freedom
1
408
2
407
3
406
Sum of Squares
2864.55216
146.28277
2867.48157
143.35336
2867.49329
143.34164
Mean Square
2864.55216
0.35854
1433.74079
0.35222
955.83110
0.35306
F-Statistic
7989.576
4070.588
2707.290
Statistic for test of parallelism
F(LOGPOP*DUMMY| DUMMY,LOGPOP) =
regression SS(full model) - regression SS (LOGPOP.DUMMY)
MS Residual (full model)
= 2867.49329-2867.48157
0.35306
= 0.03319549 with 1 and 406 degrees of freedom
(P = 0.85552)
Hence we conclude that we have no statistical basis for believing that the two lines are not parallel.
Statisitic for test of equal intercepts
F(DUMMY|LOGPOP) = regression SS(LOGPOP. DUMMY) - regression SS (LOGPQP)
MS Residual (full model)
= 2867.48157-2867.55216
0.35306
= 8.2972 with 1 and 406 degrees of freedom
(P =0.0041811)
Hence we conclude that there is statistical evidence that the intercepts are different.
Statisitic for test of coincidence
F(LOGPOP*DUMMY, DUMMYJLOGPOP)
= (regression SS(full model) - regression SS (LOGPOPW2
Geometries and Characteristics of
Public Water Systems
E-4
December 2000
-------
MS Residual (full model)
= (2867.49329 -2864. 55216V2
0.35306
= 4.16512 with 2 and 406 degrees of freedom
(P =0.01612)
Hence we conclude that we have evidence that the lines are not identical.
Summary of Test Results for Purchased Water Systems
The results are summarized as follows :
Source
Regression (LOG POP)
Residual
Regression (LOGPOP,
DUMMY)
Residual
Regression
(LOGPOP, DUMMY,
DUMMY*LOGPOP)
Residual
Degrees of
Freedom
1
437
2
436
3
435
Sum of Squares
2322.29146
145.62492
2323.67027
144.24612
2323.84732
144.06907
Mean Square
2322.29146
0.33324
1161.83513
0.33084
774.61577
0.33119
F-Statistic
6968.871
3511.776
2338.863
Statistic for test of parallelism
F(LOGPOP*DUMMY| DUMMY,LOGPOP) =
regression SS(full model) • regression SS (LOGPOP.DUMMY)
MS Residual (full model)
= 2323.84732 - 2323.67027
0.33119
= 0.53458 with 1 and 435 degrees of freedom
(P = 0.46508)
Hence we conclude that we have no statistical basis for believing that the two lines are not parallel.
Statisitic for test of equal intercepts
F(DUMMY|LOGPOP) = regression SS(LOGPOP. DUMMY) - regression SS (LOGPOP)
MS Residual (full model)
= 2323.67027-2322.29146
0.33119
= 4.1632 with 1 and 435 degrees of freedom
(P =0.041915)
Hence we conclude that there is statistical evidence that the intercepts are different.
Statisitic for test of coincidence
F(LOGPOP*DUMMY, DUMMY|LOGPOP)
Geometries and Characteristics of
Public Water Systems
E-5
December 2000
-------
= (regression SSffull model") - regression SS (LOGPOP)V2
MS Residua] (full model)
= (2323.84732 -2322. 29146V2
0.33119
= 2.34889 with 2 and 435 degrees of freedom
(P =0.096685)
Hence we conclude that we have evidence that the lines are not identical at 0.10 significance.
Summary of Results of LSMEANS Analysis
The Results of comparison of the LSMEANS are summarized as follows:
Type of Water Source
Ground
Surface
Purchased
Ownership
Private
Public
Private
Public
Public
Private
LSMEAN
(log)
5.5326429
5.7708019
5.4086391
5.5416025
5.6512160
5.7721848
T for null hypothesis
LSMEAN(Private)=LSMEAN (Public)
- 5.8623
-2.32112
- 1.95662
Pr>|T|
0.0001
0.0204
0.0506
Summary : We conclude that regression lines for private and public systems are statistically
different (at less than 0.01 significance, except for purchased where the significance is less than
O.JO) within each category (i.e. Ground, Surface, and Purchased). The results from regression
analysis are corroborated by comparing the least square means of Private and Public systems
within each category (i.e. Ground, Surface, and Purchased).
Geometries and Characteristics of
Public Water Systems
E-6
December 2000
-------
Appendix F: Bootstrapping Results for Ground Water System Entry Points
-------
This page intentionally left blank
-------
-
4-1 0) O O
—( 0)
a D
E r-t
C C C C C
U U U O U U U
c c c c c
c c c a c c c
CO) O
2 *>
-------
§
t!
0'
a-
.
c
QJ
OJ
Ou
x:
CJl
.
c
QJ
U
OJ
Cl
CTi
OJ
OJ —i
C J-J C
V-. ^ QJ
"*-i ii 0) DJ
o u a.
C J^
s S SSS
....
c c: c c
ffff
S N S S
01
C
OJ
OJ
C CU
(1) 0
0)
c
OJ
o
a.
o
QJ QJ 0)
C C C
o) ai 01
01 O> OJ
£ S S
0) QJ
c c
a! a!
S S
c
01 M
O U
c
QJ 4-> 4->
E Cfl V5
(U
4-1
c
OJ
a.
u-)
QJ
OJ
-C
0
OJ
c
QJ
t)
01
a,
r-j
0)
c
01
u
QJ
.c
o
0)
c
0)
u
re
-C -4
O OJ
M- E;
o
Q
? -0
-------
T! U ~
C C Ct,
D Q) it
Q T3 ^
T5 L —.
c c a,
3 QJ t
-O
§
QJ d) Q; OJ
a> tu CL a>
u u u u
QJ QJ CJ QJ
OJ 1) QJ
U U U
CJ QJ O
-C JT .C
OJ 14 QJ
O UI QJ
C
OJ 0!
U U
QJ QJ
QJ QJ 0)
O U U
QJ Q) QJ
a, a, a.
QJ QJ QJ
u u u
0) CD QJ
c a. cu cu
c c c
(U QJ QJ
U U U
QJ QJ OJ
a, a. a,
c
QJ
U
QJ
a.
c
QJ ^
O W
J U U
u
-------
0) O 0
Q. <_> 0
*-t 0) O O
—I
-------
Appendix G: Distribution of Flow Among Entry Points
-------
This page intentionally left blank
-------
U
a en
cc u
-C
C
-------
G 4-*
E-
-------
'
IQ
-------
This page intentionally left blank
-------
Appendix H: Non-Community Water Systems Serving Greater than 10,000 People
-------
This page intentionally left blank
-------
Appendix H
NCWS SERVING A POPULATION GREATER THAN 10,000
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
PWS Name
LAKE SOLANO CAMPGROUND
USNPS KATMAI BROOKS CAMP
WATER'S EDGE CAMPGROUND
WHITE PINES CAMPSITES
CLEARWATER TWP FIRE DEPT
NAVAJO NATIONAL MONUMENT
TOPSMEAD STATE FOREST-CH
PACHAUG STATE FOREST * 1
SAHARA HOTEL
FLYING J TRAVEL PLAZA
1-90 REST AREA ff 16
CAL-TRANS WESTLEY ROADSIDE W
BUFFALO 1 90 EAST AUTOTRUCK C
TEJON RANCH GRAPEVINE WATER
1-95 REST AREA
M D O.T-GRAYL1NG REST 1&2.R403
TOUTLE RIVER REST AREA NB/SB
CALIFORNIA STATE FAIR
KERN COUNTY FAIRGROUNDS
PIKE FAIR
BULL RUN SPECIAL EVENTS CENTER
GORGE AMPHITHEATRE
SIX FLAGS OVER MID-AM
MCCLELLAN AFB - MAIN BASE
KERN MEDICAL CENTER WATER
EPIC ENTERPRISES. INC
STAMFORD MUSEUM-W1
STAMFORD MUSEUM-W2
INVERNESS W&SD (OFFICE PARK)
CHRIS GREENE LAKE
NH 1NTL SPEEDWAY
FACTORY OUTLET CENTER
UNIVERSITY OF CALIFORNIA-DAVIS
SCOTTSDALE COMMUNITY COL
DEM HORSENECK BEACH CAMPGROUND
HOP BROOK LAKE REC AREA
SQUANTZ POND STATE PARK-MW
SQUANTZ POND STATE PARK-SPRING
TALCOTT MTN STATE PARK
HADDAM MEADOWS STATE PARK
PENWOOD STATE PARK
CHATFIELD HOLLOW STATE PARK-MW
HOPEVILLE POND STATE PARK-W2
KENT FALLS STATE PARK
WADSWORTH FALLS STATE PARK-BW
Assigned
Code
CRV
CRV
CRV
CRV
FD
FP
FS
FS
HM
HRA
HRA
HRA
HRA
HRA
HRA
HRA
HRA
MAMU
MAMU
MAMU
MAMU
MAMU
MAMU
MB
MF
MFI
MU
MU
OP
R
RCC
RET
S
s
sc
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
Type*
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
NTNCWS
NTNCWS
NTNCWS
TWS
TWS
TWS
NTNCWS
TWS
TWS
NTNCWS
NTNCWS
NTNCWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
Source
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
Pop Served
200.000
13.045
12.000
12.000
201.504
40.022
39.000
11.150
30.000
75.000
60.000
31.256
25.000
24.099
20.000
12.000
10.971
125.000
50.000
50.000
22.000
13.720
12.000
17.600
19.400
100.000
55.000
55.000
1 1 .00 1
10.500
20.000
11.000
23.898
11.200
15.000
.350.000
200.000
200.000
119.000
117.000
105.000
100.000
96.000
86.000
79.000
|
Service Area Types
REC
REC
NSA
NSA
OTA
NSA
NSA
NSA
OTA
HRA
NSA
HRA
NSA
HRA
HRA
OTA
OA
REC
OTA
NSA
OA
REC
ONTA
ONTA
MED
OA
NSA
NSA
ONTA
EAT
REC
OTA
S
S
sc
NSA
NSA
NSA
NSA
NSA
NSA
NSA
NSA
NSA
NSA
REC
OA
RES
ss
_
Geometries and Characteristics of
Public Water Svstems
H-l
December 2000
-------
No.
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
PWS Name
LORIMAR PARK
LIME KILN CSP
MT TOM STATE PARK-MW
INDIAN WELL STATE PARK-NW
INDIAN WELL STATE PARK-SW
BRANBURY STATE PARK
SAND BAR STATE PK
SQUANTZ POND STATE PARK-CLW
MASHAMOQUET BROOK STATE PARK-PW
PICACHO STATE PARK
NORTH HARTLAND LAKE
TOWNSHEND LAKE
STOUGHTON POIND
WINHALL BROOK REC AREA
LAKE WARRAMAUG STATE PARK-BW
LAKE WARRAMAUG STATE PARK-CW
MACEDONIA BROOK STATE PARK-CS4I
MACEDONIA BROOK STATE PARK-CSI7
JONES BEACH STATE PARK
PUTNAM MEMORIAL STATE PARK
CARMI LAKE STATE PARK
SHAFTSBURY STATE PARK
SOUTHFORD FALLS STATE PARK-SV
FORT SHANTOCK STATE PARK-BW
CDPR FOUR RIVERS DIST-BASALT
CDPR FOUR RIVERS D1ST-SAN LUIS CREEK
BURR POND STATE PARK-CW
HANGING ROCK STATE PARK
LAKE ST. CATHERINE STATE
ELMORE STATE PARK
MILLERTON SRA MEADOW TANK HILL
KETTLETOWN STATE PARK-BW
KETTLETOWN STATE PARK-CW
CRANBERRY LAKE
INDIAN ROCK RESERVE-ORCHARD HSE
CDPR-MCCONNELL
GILLETTE CASTLE STATE PARK-MPW
HOUSATONIC MEADOWS STATE PARK-MS
ANTELOPE VALLEY POPPY RES STATE PAR
JAMAICA STATE PARK
JOHN M1NETTO STATE PARK-UPW
JOHN MINETTO STATE PARK-BW
HOPEVILLE POND STATE PARK-W3
WOODFORD STATE PARK
GROTON STILLWATER
MT. PHILO STATE PARK
BURTON ISLAND STATE PARK
DEVIL'S HOPYARD STATE PARK-CW
DEVIL'S HOPYARD STATE PARK-PAW
Assigned
Code
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
Type*
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
Source
GW
SW
GW
GW
GW
SW
SW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
SW
SW
GW
GW
GW
GW
SW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
GW
SW
GW
GW
Pop Served
75.000
75.000
60,000
55.000
55.000
53.000
51.240
50.000
50.000
50.000
50.000
50.000
50.000
50.000
45.000
45.000
43.000
43.000
40.000
38.000
38.000
32.700
32.000
31,000
30.210
30.140
30,000
30.000
30.000
29.000
26.344
25.000
25.000
25.000
25.000
20.499
20.000
20.000
20.000
19.000
18.500
18.500
18.000
18.000
18.000
16.000
16.000
15.000
15.000
Service Area Types
REC
REC
NSA
NSA
NSA
REC
REC
NSA
NSA
REC
REC
REC
REC
REC
NSA
NSA
NSA
NSA
NSA
NSA
REC
REC
NSA
NSA
REC
REC
NSA
REC
REC
REC
REC
NSA
NSA
NSA
NSA
REC
NSA
NSA
REC
REC
NSA
NSA
NSA
REC
REC
REC
REC
NSA
NSA
Geometries and Characteristics of
Public Water Systems
H-2
December 2000
-------
No.
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
PWS Name
TAYLOR BROOK STATE PARK-ULW
TAYLOR BROOK STATE PARK-LLW
DEM HORSENECK BEACH ST. RESERVATIO
LAKE WELCH BEACH W.S
UNDERHILL REC. AREA
HAYSTACK MTN STATE PARK
GROTON BOULDER BEACH
R1CKETTS GLEN STATE PARK
MOLLY STARK STATE PK
ASCUTNEY STATE PARK
OUECHEE GORGE STATE PARK
SMUGGLERS NOTCH ST PARK
W1LGUS STATE PARK
NORTH COUNTRY SPRING WATER
PONTIAC LAKE LAUNDRY
Assigned
Code
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
WPP
WPP
Type*
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
TWS
Source
GW
GW
GW
SW
GW
GW
SW
GW
GW
GW
GW
GW
GW
SW
GW
Pop Served
15,000
15.000
15.000
15.000
13.000
12.700
12,600
12.000
12.000
11.480
11.126
11.000
10.494
3.000.000
165.163
Service Area Types
NSA
NSA
REC
NSA
REC
NSA
REC
EAT
REC
REC
OA
REC
OA
NSA
OTA
REC
* Abbreviations:
TWS = Transient Non-Community Water System
NTNCWS = Non-Transient Non-Commumtv Water Svstein
Geometries and Characteristics of
Public Water Systems
H-3
December 2000
-------
This page intentionally left blank
------- |