PB83-110957
Toward Instieai Water Quality Management
Illinois Univ. at Urbana-Champaign
Prepared for
Environmental Research Lab.
Athens, GA
Jun 82
-------
leadquarters ano Chemical Libraries
EPA Wes,i Bidg Roo^i 3340 PmJJ_n0957
iv/iailcode 34Q4T
1301 Constitution Ave NW EPA-boo/s-ez-ooa
Washington DC 20004 Juno 1982
202-566-0553
TOWARD INSTREAM WATER-QUALITY MANAGEMENT
by
C. Osteen, W. L. Seitz, and J. B. Stall
Contract Number 68-03-2597
Project Officer
Thomas E. Waddell
Technology Development and Applications Branch
Environmental Research Laboratory
Athens, Georgia 30613
ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH A'-iD DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ATHENS, GEORGIA 30613
amoucio sr
NATIONAL TECHNICAL
INFORMATION SERVICE
US CIPHSIMM OF COMKCffCt
SPRISGfttlD. VA IV.f.
-------
(.),§, Envirowrtsn'al Froiecfcen Ageing
f • >n Ui Information Resource
v: iiKur (3PM52)
fill Chestnut Street
FA 1910?
TECHNICAL KtPORT DATA
(I'lcnsc n'Oi! .'/uu/Hk nous <»t ific nc be lac umipL'iir.vi
i RljOmT K'O 2
£PA-500/5-82-003 0RD Report
3 ni'CIPIENT'S ACCESSION NO
PbTj ?, 1 1 0 9 5 7
"1 ; Li- AND SU5TP 1~'_E
Toward Iiistru/M Water Quali Ly Management
i, hm'omt oatc
June 1982
C PCHf-OHMKvG OK G A'V 12 AT J ON CODfc
7 AL^ HORiSI
C. Osteon, W.D. Seit/., and J.B. Stall
B P!:RI:ORMlNG ORGANIZATION REPORT NO j
•j rr^-oR fliNG Om-oAniza rii_i:j kahe wo Anyic-s
University of Lllinois at Urbaua-Champaign
Urbana IL 61801
10 PrlOGRAI.' ELLMENT NO
CARB1A
1 1 CONTRACT''GRANT NO
68-03-2597
'.2 S = ONSORir'G AGENCY NAME AND mODHEoS
Environmental Research Laboratory—Athens GA
Off^e of Research and Devel opmen t
U.S. Environmental Protection Agency
Athens GA 30613
13. TYPE GF" RCPORT AND PERIOD COVF.RliO
Pinal, 9/77-8/79
14 £r ONSORINU AGENCY CODE
LTA/600/01
15 Supplemental . notes
16. ABSTRACT
This report compares two approaches to the agricultural nonpoint source pollu-
tion control problem: source control and instream water quality management (ISWQM).
Source control is a strategy of controlling pollution loadings by using standards
sucii as soil Joss Limits or best management practices without relating them directly
to water quality goals. ISWQM is a strategy for determining water quality goals by
examining pollution effects and other considerations and developing a resource
management plan for achieving those goals. ISWQM relates Land management more
closely to water quality goals. The report discusses the information needs, insti-
tutional arrangements, and the strengths and weaknesses of ejeh approach,and suggest:
some intermediate alternatives that could be explored. Various physical, planning
and decision-making models that could be utilized under the two approaches are analy-
zed.
The analysis shows that the cource control approach to water quality manage-
ment is currently feasible. Institutions to implement such an approach on agri-
cultural lands already exist. Economic and physical models are readily available
for use under this approach.
17 key words and document analysis
j dcschii5 rons
li IDENTIFIERS,'OITN r.NDED TERMS
c COSm n I icM/l.220-: (9-" J)
-------
NOT ICC
This ciocumcnr lias been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
-------
FOREWORD
As environmental controls become more costly to implement and the
penalties of judgment errors become more severe, environmental quality
management requires more efficient management tools based on greater know-
ledge of the environment.il phenomena to be managed. As part of this
Laboratory's research on the occurrence, movement, transformation, impact -
and control of environmental contaminants, the Technology Development and
Applications Branch develops management ard engineering tools to help
oollution control officials achieve water quality goal s 'tfirou'gli'watershed "
management.
Agricultural sources contribute significantly to water pollution
problems in many areas of the United States., This report describes part
of a 2-year study in which the social, economic, legal r.nd institution?.!
issues involved in the management and control of pollutants from agricultural
nonpoint sources were examined.
David W. Duttweiler
Director
Environmental Research Laboratory
Alliens, Georgia
-------
ABSTRACT
This report; compares two approaches to the agricultural nonpoi;it
source pollution control problem- source control and instream water
quality management (ISWtM)• Source control is a strategy of controlling
pollution loadings by using standards such as soil loss limits or best
management practices (BMPs) without relating them directly to water
quality goals. ISWQM is a strategy of determining water quality goals by
examining pollution effec.s and ocher considerations and developing a
resource management; plan for achieving those goals. ISWQM relates land
management more closely to water quality goals. The report discusses the
informational needs, institutional arrangements, and the strengths and
weaknesses of each approach and suggests some intermediate alternatives
that could be explored. Various physical, planning and decision-making
models that could be utilized under the two approaches are analyzed.
The analysis shows that the source control approach to water quality
management is currently feasible. Institutions to implement such an
approach on agricultural lands already exist. Economic and physical
mode's are readily available for use under this approach. A number of
institutional changes, however, could be made to more effectively
implement this approach. ISWQM encounters more problems than source
control because of greater informational costs, technical problems and
political resistance. ISWQM has the capability, however, to define goals
and problems more accurately and plan resource management to control
pioblems more efficiently than source control..
This report was submitted in partial fulfillment of Contract No.
68-03-2597 by the University of Illinois at Urbana-Champaign. The report
covers the period September 1977 to August 1979.
i v
-------
CONTENTS
Page
Abstract iv
figures vi
1. Introduction 1
2. A Review of Agricultural Nonpoint Pollution Control
- Policies i............... i 3
3. Institutional Problems and Changes 4
4. Decision Models and information Constraints to Decision- . -
Making Under Source Control and ISWQM 5
Components of the Decision-System Model 6
Water-use Goals 6
Water Consumption 8
Demand for Recreation 8
Water Quality Criteria 10
Physical Processes 12
Models of Sediment Processes 18
Planning Models 21
The State of Sediment Research Modeling 28
Conclusions Concerning Physical Processes 31
Traits of Models 33
Summary of Advantages and Disadvantages of Planning
Models 33
Needed Research for the Land-Use Water-Quality
Linkage 35
Linking Models of Land Use to Models'of Physical
Proces-es 38
5. A Recommended Decision-Model S .ructure for the ISWQM
Approach 47
What Inf ^rmation Should the Model Provide 47
The Recommended i-todel Structure 48
6. Conclusion: Hw Feasible is ISWQM as. compared to Source
Control? b2
References 55
v
-------
FIGURES
Number Page
1. Names of Grain Sizes 14
2. Hydrographs of Flow and Suspended Sediment for a Storm During
August 1977, on the Vermilion River at Pontiac, Illinois. ... 16
3. Sediment Duration Curve for Water Year 1977 on Vermilion
River at Pontiac, Illinois 17
vi
-------
INTRODUCTION
This report compares two approaches to the agricultural nonpoint-scurce
pollution-control problem: source control and instream water-quality man-
agement (ISWQM). This report will view controlling sediment as a stream
water quality problem. However, it is recognized that there are other non-
point pollutants such as pesticides and fertilizers. The authors feel that
much of the discussion in this report will apply to non-sediment pollutants.*
Source control is a strategy of controlling pollution loadings by using
standards such as soil loss limits or best management practices (BMPs) with-
out relating them directly to water quality goals. ISWQM is a strategy of
determining water-quality goals by examining pollution effects and other
considerations and developing a resource management plan for achieving those
goals. ISWQM relates land management more closely to water quality goals.
Source control focuses on a more manageable piece of the overall problem be-
cause the techniques to be used are relatively well understood and the insti-
tutional framework is in place to implement such an approach. Nevertheless,
while the impacts of sjch a program on water quality may be significant and
positive, the precise impacts are simply not well known, and tne most effec-
tive source controls to improve water quality, may be hard to specify. ISWQM,
on the other hand, would capture a larger subset of problems, so the analy- •
tical problem would become much more difficult. In this report, we will
discuss the strengths and weaknesses of each approach and suggest some inter-
mediate alternatives that could be explored.-
Institutionally, ISWQM requires a close integration of nonpoint-source
pollution control with the management of water-uses and formulation of
water-quality goals. A land-management plan defining BMPs or effluent star-
dards would be related to water-quality goals for a stream and could be
changed if the water-quality goals are not met. Under source control, the
land-inanagement plan is typically applied without analysis of the impacts
on changes in water quality in a particular stream. The ISWQM planning pro-
cess requires an institutional structure such that agencies defining water
uses and managing water resources, agencies defining water-quality goals and
standards, and agencies developing land-management plans must work together
to relate land management to desired water quality. Under source controls, •
agencies planning land management could work independently of agencies de-
fining water uses and managing waterways.
* The authors are more familiar with sediment problems than other non-
point pollution problems. J. B. Stall is a hydrologist who has made a
career of studying sediment problems. All three authors participated on the
Erosion-Sedimentation Subcommittee of the Illinois Task Force on Agricul-
tural Nonpoint Sources of Pollution which advised the Illinois EPA in devel-
oping its statewide 208 plan.
-------
To be implemented effectively, ISWQM requires a reeoDacK metridnibm un
water quality that, in turn, requires water quality monitoring and periodic
reassessment of the effectiveness of pollution-control practices in achieving
water quality standards. Periodic reassessment of plans is also needed to
redefine goals and problems over time and to readjust plans n light of any
new knowledge of pollution-control practices and land-use/water-quality re-
lationships. Water-quality monitoring will also make possible advances in
the state of knowledge of land-use/water-quality relationships. A source
control approach would not involve this feedback mechanism.
There is also an intermediate approach between source control and ISWQM.
An initial source control plan might be developed that simply applies tech-
nical standards. The oerformanco of the plan is reassessed in light o4:
water-quality goals and changed where performance is inadequate. This ap-
proach will be considered to be an ISWQM approach because land management
plans and water-quality goals are being integrated in the planning process
ana a feedback mechanism exists to assess performance arid change^the plan.
Oregon's 208 basin-assessinent project appears to fit into this category
(Rickert and Beach, 1978).
For convenience, two different cases of ISWQM will be considered. With
the first approach, water quality goals or standards are precisely defined
and a land management plan is developed to meet the water quality goals. The
land management plan could then be "fine-tuned" over time to meet the water
ouality goals. With the second, simpler approach, priorities for pollutants
would be defined based on their expected impacts on water uses. Critical
sources of these pollutants would then be identified and land management
modifications would be concentrated in these areas. The first approach
would require a greater knowledge of the impacts of pollutants on water uses
and the relationship between land management and water quality.
As will be discussed later in the report, it appears likely that diffi-
culties in changing institutional structures, lack of information concerning
land use and water quality, and the costs of planning and implementation
will make ISWQM more costly and difficult to implement than source control.
It seems likely that an ISWQM approach of precisely defining water quality
goals for nonpoint pollutants and developing land management plans closely
linked to those goals will only be used in special circumstances. (RCWP
project areas could be such a circumstance.) Source control or the simpler •
ISWQM approach of identifying priorities for pollutants and critical areas
of pollution sources for management would be used in other circumstances.
For example, ISWQM might be applied to streams identified as having "criti-
cal" values such as esthetic or pristine characteristics to be preserved,
special uses to be managed (e.g., navigation), or severe pollution problems.
Source control could be applied to all other areas. Iowa has developed a
procedure to identify high-value waterways with severe problems on which to
concentrate management (National Association of Conservation Districts, 1978).
Another approach would be to apply source control to achieve long-term pro-
ductivity goals on all land and apply ISWQM later to address any remaining
water-quality problems.
2
-------
A REVIEW OF AGRICULTURAL NONPOINT
POLLUTION CONTROL POLICIES1
Two important laws define directions for agricultural nonpoint pollution
control: PL92-500 and PL95-217. Section 203 of PL92-500 required the states
to develop plans to control nonpoint sources of poT'ution. The governor of -
each state appointed agencies to develop statewide or areawide plans. EPA
regulations required that a statewide policy advisory committee consisting
of local elected officials, represenatatives of state and federal agencies,
and representatives of the genera-1 public. The USDA, U.S. Amiy Corps of-.
Engineers, and Department of interior were to be invited to oarticipate in ¦
area-wide policy groups. Public information and participation was required.
The current CPA guidelines to 208 plans require that priorities be developed
for pollutants and that critical source areas of those pollutant: on which
to concentrate control be identified. Hence, the EPA requires an ISWQM
approach. It is not clear how closely these guidelines will have to be
followed for a state's 208 plan to be acceptable. The plans must identify
anJ evaluate measures needed to produce a desired level of control. The
practices must be effective in reducing pollution problems. The plans must
also identify agencies and programs to implement the plans. Existing agen-
cies and programs should be used to the degree practicable with the estab-
lishment of new agencies to be avoided.
PL95-217 defines the Rural Clear Water Program (P.CWP) which will pro-
vide a limited amount of technical and financial assistance to implement
practices to control nonpoint pollution. Under the RCWP a number of demon-
stration areas or areas of special rural water quality problems will be
chosen. The USDA and USEPA will choose project areas from lists supplied
by the state governors. Technical and financial assistance will be given
to landowners and farm operators to install practices which reduce water
quality problems. At least 7152 of the owners or operators In "critical"
areas of pollution in a proposed project area must agree to participate if
assistance is to be granted. The Soil Conservation Service will take a
lead role in the PCVIP through soil and water conservation districts (SWCDs),
state soil and water conservation committees, or state water quality agen- .
cies. It must be emDhasized that the RCWP does not supply funds for the
entire nonpoint pollution program. Nor does it prohibit the states from
using other state programs or funds to implement pollution control practices.
There appears to be a great deal of leeway in defining agencies and
programs to implement 208 water quality plans. However, it seems likely
that state agricultural departments, extension services, and soil and water
conservatiOi, districts will fee involved with 203 agencies in implementing
water quality plans on agricultural lands in a great many states. Tradi-
tional soil and water conservation programs, with some modification, will
1 Information in this section was drawn from Groszyk, 1979, and Holmes,1979.
7
-------
probably play an important -o-le in providing technical and financial assis-
tance to agricultural landowners and fa mi operators who are implementing
nonpoint pollution control practices. In addition, the RCVJP, which is to
provide technical and financial assistance for special project areas, will
be implemented by the Soil Conservation Service. Hence, RCWP will reinforce
the role of traditional agricultural agencies.
In summary, agencies are to be designated by the governor of each state
to develop and update plans tc jontrcl agricultural nonpoint sources of
pollution on an areawide or statewide h3sis. A wide variety of agencies can
se^ve as 208 agencies. Local government officials, state and federal agen-
cies, and the general public are to be involved in the planning process. The
plans are to define priorities for pollutants and sources of pollutants.
Practices to impro\e water quality are to be identified in the plans. The
plans are to designate agencies and programs to implement the plans. In
addition, the RCWP defines a program of 'echnical and financial assistance
for applying practices ^n special project areas. It seems likely that&wra-
aitional state agricultural and soil and water conservation agencies'anfp'
soil and water co'nserva'tio!, districts wi 11-play- an important role . in-oaiRle-
menting 208 plans on agricultural land.
INSTITUTIONAL PROBLEMS AND CHANGES
A number of problems may result when SWCDs which have been traditionally
concerned with agricultural productivity become involved in programs of im-
plements o practices to improve water qualify. Some changes in the way
SWCDs have traditionally made decisions may result. Three problems could
arise to create difficulties for SWCDs. First, some people may be excluded
from decisions which they feel will affect their legitimate interests. Land
management decisions made by a SWCD may affect the use of a waterway outside
an SWCD. Also, in some states, only landowners can vote on decisions con-
cerning the application of land use regulations even though nonlandowners
tray feel they have a legitimate interest in how land is managed to control
water pollution. Second, decisions concerning land management, water use,
and water quality might not be coordinated closely enough under current
arrangements. In the past, for example, reservoirs have often been con-
structed without an erosion control plan to control siltation. The result-
ing loss of reservoir life could have been prevented by better coordination
of land management and water use decisions. Third, SWCD decisions might not
consider watershed level problems. For example, the land management de-
cisions of one SWCD might result in high quality water entering a stream
system. However, the land management decisions of a second SWCD might
result in much lower quality water entering the stream system. Coordinating
the decisions of the two SWCDs might result in a more efficient allocation
of resources to reduce i.onpcint pollution problems than allowing each SWCD
to act autonomously.
Addressing the interests of people living outside an SWCD, the problems
of coordinating land management and water use decisions, and watershed level
problems indicates that coordination of local land management decisions at
the scate or regional level is needed. Developing 208 plans provides an
4
-------
opportunity to address these problems by br.nging agencies and the general
public together on an area or statewide basis to define water quality and
land management plans to meet them. (Whether or not 2C3 plans will success-
fully address these problems is another question.)
SWCDs nave traditionally had much discretion in determining which
practices in a soil and water conservation program would be cost-shared and
where they would be located. Under a source control approach (such as a
program of meeting SC5 t-values), a traditional program of providing the
SWCDs with a list of practices and a source of funding and allowing them to
operate autonomously might work reasonably well. Under ISWQM, more coor-
dination of SWCDs1 land management decisions would .be needed to ensure that
.appropriate practices are applied to meet water quality goals. A 203 plan
would provide some degree of coordination of SWCDs' decisions. However-,- -
a 208 agency might have to monitor SWCD decisions closely and indicate
where changes in SWCD decisions should bo made to pursue the plan. The
criteria that SWCDs use to allocate funds and practices for purposes of
agricultural productivity might not fit a 20C plan well. A greater degree
of coordination would be needed when a land management plan is developed to
meet a set of water quality standards than under a simpler approach of
identifying critical source areas of high priority pollutants. The need for
precision in locating practices may be less important in the second case.
The greater the responsibility that 208 agencies are to have in coordinating
the decisions of SWCDs, the more legal powers beyond the advisory role they
should have. A proposal to give a 208 agency power over the actions of an
SWCD would probably bring about political resistance from SWCDs. The
greater the activity proposed for the 203 agency, the greater the resistance
from SWCDs will probably be. Pollution control alte' n^J'ives are listed in
decreasing order of political resistance: 1) ISWQM - w- re a land manage-
ment plan is developed to a sot of precisely defined w.'.-r quality standards.
2) the simpler ISWQM approach of identifying pollutant priorities and criti-
cal source areas, and 3) source control.
To give nonlandowners within an SWCD the right to vote on land use
relations could require a change in many states' soil conservation district
Hws. Such a change could bring about strong political resistance from
landowners.
The institutional changes suggested here are not believed to bn radical.
Coordinating agencies which restrict th.-; powers of local governments have
been implemented in many cases (Bosselman and Callies, 1971). Voting rights
have changed. Hov.ever, political resistance to these chaiges, particularly
the movement away from local control, could make the implementation of an
ISWQM approach difficult but not impossible.
DECISION MODELS AND IN FORMAT JON CONSTRAINTS TO DECISION-
MAKING UNDER SOURCE CONTROL AND ISWQM
An agency-participating in the ISWQM process would ultimately be making
recommendations for land management. The process should inrlude evaluations
of (1) potentia" demands for water uses. (2) the capabilities of a particular
stream and others to supply different uses. (3) the opportunity costs of
-------
pollution-control strategies, and (4) the environmental and distributional
impacts of alternative strategies. A management plan of the watershed's
resources would be based on such an analysis ar.d would consist of a set of
appropriate uses for different sections of the watershed, water-quality
standards, and land-man?gement criteria. Mode-Is of the decision system
would help planners to study thfe impacts of pollution-control alternatives.
The purpose of this section is to discuss the -1imitations of available
methods and information sources and how they would restrict the modeling
effort and thus the planning process in the case of soil erosion and sedi-
mentation.
Components of the Decision-System Model
The.ideal decision system would consist of three components: (1) land
management, (2) physical processes', and""(3) "wate'r-iis'e"and'water-qual ity
criteria. The larid-managemenu component should model the production of
goods, land management practices (including tillage and conservation prac-
tices) and crop rotations, and the opportunity costs of different land-man-
agement plans. The physical component should model t'ie impacts of these
land-management alternatives on runoff, sheet-and-rill and gulley erosion,
the transport of sediment (and other pollutants) to streams, and the impacts
on stream processes such as channel erosion and deposition. Ideally, the
total amount of sediment moved, th(_ suspended portion, and the variability
of sediment loads over time at various locations in a stream, as well as
other water-quality parameters, should be estimated.
The vater-use and loatev-quality criteria component would (1) estimate
potential demands for uses of a stream and the benefits of providing those
uses (and, conversely-, the damages caused by not providing them) and (2)
include different relationships to estimate ivater-qual ity criteria for dif-
ferent uses. The analysis of alternatives ^ould be conducted to determine
the costs and benefits, the distribution of costs and benefits, and the
environmental impacts of pollution-control alternatives.
Ultimately, the minimum desirable modeling arrangement for ISWQM would
(1) relate land-management practices to the behavior of physical processes
and, in turn, estimate water-quality parameters and (2) estimate the costs
and benefits of the land-management practices. This type of model would
provide decision-makers with an idea of the water-quality ramifications of
different land-management plans. This information would be quite useful to
those in the political arene.
Water-Use Goals
The process of defining appropriate uses for a stream is largely politi-
cal and involves the balancing of interests. It can be carried out regard-
less of whether or not information sources to estimate uses of a stream are
available. People can voice their opinions through the political process
and present evidence concerning the uses of a stream.* Political influences
*The Illinois EPA is currently experimenting with classifying streams accord-
ing to uses and applying water-quality standards based on those uses. " "
c
-------
in the planning process can affect the determination of goals and their
priorities, alternative courses of action to be considered, trade-offs be-
tween alternatives, and the alternative finally chosen. As a result,
empirical analyses of demand and consumpLion are not necessary to determine
appropriate uses. The USCPA requires as a minimum that all existing uses of
a stream be protected (Iowa Law Review, 1977).
The role of empirical analysis of demand is to aid in estimating bene-
fits generated under the alternative stream uses. This analysis would also
be helpful in estimating values for stream uses considered in modeling ef-
forts where costs of land management and values of stream uses were consider-
ed simultaneously. Both demand functions and cost accounting imply a weight-
ing of-preferences (i.e., interpersonal comparisons) based upon the distribu-
tion of wealth and legal and administrative structures that define rights
and responsibilities in the market place (Samuels. 1972; Henderson and
Quandt, 1958; Alchian and Allen, 1967). This weighing of preferences in the
market might, in fact, be different rom that generated through a political
process. Resource allocations based upon demand or consumption analyses
would be preferable tc allocations based on the political process only if
the weighting of preferences implied by these demand or consumption functions
is "more desirable." The determination of the desired weights, however,
would ultimately be resolved in the political process.
The problems of data availability and cost and the inability to specify
units of measure for consumption arid values of uses present significant bar- ¦
riers to the estimation of consumption and demand functions.* It will be
cheaper and easier to estimate these functions if secondary data sources arc
available than if primary data must be collected, particularly if labor-in-
tensive survey research techniques must be used. ' Altho.igh secondary data
sources will generally be available to estimate water consumption by munici-
pal and industrial uses, secondary data sources for water consumption by
agricultural uses and for recreational uses often will not be available or
will be of poor quality. Also, it will generally be difficult to measure
consumption and values for most recreational uses. Because of these barriers
and the costs of overcoming them, it may be necessary to limit the sophisti-
cated analyses of demand or consumption to streams or lakes with particularly
important problems or critical values that will justify intensive management
efforts.**
* Although there may also be problems with the validity of the methods of
estimating these two types of functions, they will not be addressed here.
** The discussion of demand analyses will be limited to water consumption
and recreational uses. However, uses such as navigation and commercial
fishing could also be analyzed.
-------
Water Consumption
Water-consumption data from public or municipal water sources is readily
available. Municipalities can be contacted to determine annual withdrawals
and locations of withdrawals (Institute for Environmental Studies, 1977).
The costs of treating municipal water are also available, thus providing a
source of information to use in estimating the values or benefits of improv-
ing water quality (Jacobs and Timmons, 1974). Information on industrial with-
drawals can be obtained through regional EPA offices by reviewing NPDES per-
mits (Institute for Environmental Studies, 1977).
Estimating agricultural water consumption may be more difficult. Irri-
gation is generally the most important agricultural water use. Livestock
watering and household uses could also be important. Cooperative Extension
Service agents could be contacted to find out who irrigates, how much they
irrigate, and what crops they grow. Numbers of livestock in a watershed
could be a basis for estimates of livestock water consumption. Although
this information can be ohtained from the Census of Agriculture, it may be
outdated; in states such as Illinois, annual estimates of livestock raised
are maintained and could be used, but it would not be possible to determine
whether the water was from surface-water or groundwater sources (Institute
for Environmental Studies, 1977).
If estimates of consumption of water are needed that are more detailed
or accurate th3n could be obtained from secondary data sources, data may
have to be collected directly from farmers or from industrial managers. A
variety of sources such as Cooperative Extension Service agents, chambers of
commerce, and tax rolls could be used to define the populations to be samoled.
Demand for Recreation
Estimating demand for or participation in recreation activities will
probably be more difficult than estimating water consumption. Three pro-
blems appear with recreation. It will be difficult to (1) define and sample
user populations-for some dispersed activities, (2) define some uses, and
(3) develop measurement scales for some intangible values.*
Activities that are not site specific, such as hunting cr fishing, will
present problems because it will be difficult to identify participants. In
general, survey research techniques of some form will be necessary since
site-specific secondary data sources are generally not available. Unless
there are a few locations by which people must pass, such as park entrances,
it will be costly and time consuming to identify che users. In surveying
users, it would be desirable to study how participation and willingness to
pay would vary with success (e.g., fish caught) so that the value of clearer
water can be established.
Defining uses can also present difficulties for estimating recreation
*See Dwyer, Kelly, and Bowe.s (1977) for a discussion of techniques to
estimate recreation demand.
8
-------
demand or participation. An important problem in defining uses would arise
where seemingly closely related activities require significantly different
qualities of water. For example, different species of fish require different
water qualities. To the extent that fishermen prefer different species of
fish, there are different user populations. In fact, there may be large
contrasts among various types of fishing experiences and groups of fishermen.
" Defining measurement scales for uses can also present difficulties for
estimating recreation demand or participation. Measuring intangible values
such as esthetics or solitude is extremely difficult. For example, it will
be difficult to estimate the impacts of improving water quality on esthetic
values and then estimating the economic benefits. As a result, it is diffi-
cult to estimate the demand for such values or the benefits of providing
them.""
The problem of obtaining site-specific or watershed-specific informa-
tion suggests that an alternative approach be considered. Rather than at-
tempting to make location-specific estimates of demand for uses, particular-
ly recreational, it might be more practical to estimate demands for "ses
over a large geographic area using survey methods." Surveying such area
would be more practical than surveying a number of individual sites because
the users of specific sites would not have to be defined as populations to
be sampled. Priorities for uses could then be developed with political
inputs from the interested public.
Once the priorities for uses and quantities desired have been developed,
suitabilities of waterways for various uses would have to be evaluated. Each
use would require a scale of suitability that would allow comparisons of the
suitabilities of uses on all waterways. Criteria for suitability for a use ¦
could include the natural characteristics of the stream, its current charac-
teristics, and the estimated costs of reducing pollution to meet the cri-
teria for the use. It would be difficult,' however, to develop scales or
criteria for intangible values such as esthetics. Also, preferences for
particular sites would be difficult to account for. With input from the pub-
lic, an interdisciplinary team of scientists and resource managers could
assess the suitabilities of the waterway for different uses.
The next problem would be to allocate appropriate uses to streams. One
way would be to determine which waterways or portions of waterways meet the
criteria for a use and then allocate to them the highest priority for that
use. This allocation could meet some or ail of the quantities demanded for
compatible lower priority uses. This process would continue for each lc/ver
priority use in sequence of priority until the quantities demanded for each
use were met or until all available waterways were allocated. In this way,
demands for the highest priority uses would be met first and waterway uses
that were incompatible with each other would bp separated. One problem with
this approach to allocation, however, is that it would neglect decreases iri
priorities for a use as the number of waterways allocated to that use in-
creased. As a result, certain high-priority uses might be overallocated to
waterways. The process described of determining priorities for uses, suit-
abilities for waterways, and allocation of uses is similar to the multipur-...
pose planning process for the smallest units of the national forests
9
-------
(Thompson, Nygren, and Moore, 1976). A similar approach was used In develop-
ing the RARE II Draft Environmental Impact Statement (U.S. Forest Service,
1978).
In summary, the lack of secondary sources of data on consumption of or
participation in water uses is not an important barrier to the determination
of appropriate uses for a stream because goals for uses do not have to be
quantified. People can voice their opinions and show evidence concerning
the appropriate uses of a stream or watershed through the political process.
The lack of secondary-data sources and other data problems, however, includ-
ing difficulties of defining uses and user populations, could be costly bar-
riers to any planning effort where there is an attempt to ffuantify goals for
jses and trade-offs among uses. These barriers would also make it difficult
to include a quantified water-use component in a model of the decision
system. The amount of money that should be spent on' collecting"and analyz-
ing such data should be determined by the importance of the problems being
alleviated and the value of the attributes being protected. It seems likely
that sophisticated data collection and analysis should be limited to those
waterways with particularly important problems or critical attributes that
will require intensive management efforts.
Water Quality Criteria
The definition of water-quality criteria is closely related t.o the
definition of water uses because these criteria are statements of what water
quality is needed for a certain use. Water quality standards (goals) could
be based upon the criteria. Land management plans could then be developed
to reduce the quantity jf pollutants in the water to the standards.
Ultimately, water-quality standards would be determined along with
water uses in the political process. Standards could be arbitrarily set to
reduce the presence of undesirable or dangerous substances. It would be
desirable, however, to base standards upon the biological and ecological
impacts of those substances in order to improve stream quality and not to
allocate resources to such reductions if they will not improve desired uses.
Criteria for uses on which standards could be based have been developed that
could be applied in the ISWQM process. Complete standards have not been
developed, however, for a number of uses because the impacts of some sub-
stances are not well understood. This uncertainty is particularly a problem
for developing standards for individual fish species or biological communi-
ties.
The approach of the Lake and Stream Classification Study (Institute for
Environmental Studies, 1977) is to define a set of criteria for each broad
classification of use (e.g., public water supplies, fish and wildlife propa-
gation, recreation and esthetics, shellfish, livestock, irrigation, and dif-
ferent classes of industrial water use). The criteria are based on USEPA
water-quality criteria and state standards. When appropriate uses are de-
fined, the most restrictive criteria of the set definec for appropriate uses
define the stream criteria.
10
-------
lhe lack of information to determine criteria for certain pollutants,
and in some cases to identify substances as potential pollutants, makes it
difficult to develop a complete set of criteria. The criteria for some sub-
stances, however, are well documented. For example, 0.3 mg/1 of nitrogen
and 0.01 nig/1 of phosphorus are accepted by many limnologists as critical
concentrations for nuisance blooms of algae (Mackenthun, 1969; Task Force on
Agricultural Nonpoint Sources of Pollution, 1978). Also, water filters for
farmstead water can'filter out only so much sediment. Pressure sand filters
can be used when turbidity does not exceed 20 JTUs (Jackson turbidity units),
while in-line filter's will not work when turbidity exceeds 5 to 50 JTUs
(Oschwald, 1972).
The approach of defining criteria for broad categories of uses does not
account for variations in wo te^-qual ity needs within the categoryparticu-
larly for fish arid wildlife propagation. The Lake ana Stream Classification
Study, in its further research, has tried to define water quality and habi-
tat needs for fish within the state or Illinois. Rased on literature re-
views and interviews of experts, Horricks has shown that there is a lack of
published research concerning water quality and habitat needs for many
species.* kesearch seems to be concentrated on a small number of game spe-
cies, bass, and sur.fish (Sparks, 1978). Many of the parameters developed
for toxic chemicals and heavy metals are related to acute toxicities and are
not related to the lomj-tonn impacts on tha community or ecosystem (Sparks,
1978).
In particular, the effects of different concentrations of sediment are
poorly understood. Herricks used subjective judgments, based on the exper-
ience of Dr. Phillip Smith of the Illinois Natural History Survey, to esti-
mate the tolerance of species Co suspended solids. He also listed bottom
conditions desired by species. Sparks (1978) documents a number of studies
on the impacts of sedimentation on different species (Buck, 1956a; 1956b;
CI affey, 1955, Gammon, 1970, Meinstra, Damkut and Benson, 1969; Vinyard and '
O'Brien, 197C; and V/alien, 1951). Bass and sunfish are the predominant spe-
cies discussed. Sparks concludes that mean turbidity levels of less than
100 JTUs are needed to maintain desirable game species over time. Over this
mean level of turbidity, "rough fish" will eventually dominate the ecosystem.
With mean turbidity levels between 25 JTUs and 100 JTUs. however, the total
production of fish decreases. Sediment on the bottom of a stream or lake is
also important because habit?t changes change species composition. Analyses
are needed to determine the water-quality and habitat parameters for a large
nunber c" species and communities if water-quality standards are to be based
on desired fish species. This type of analysis probably will be time-con-
suming and expensive.
Stall recommended that a sediment standard for streams in 111!noi " and
other midwestern states be a geometric mean of 30 iny/l.** lie feels this
standard closely approximates 25 JTUs and is the effluent standard for
*Personal correspondence with Ldwa^d Derricks, Professor of Civil Engineer-
ing, University of Illinois at Urbana-Champaign, 1978.
"¦^Personal communication with John B. Stall, Consulting Research Hydrologist,
1601 S. Maple, Urbana, Illinois 61801.
11
-------
suspended solids from a point source in Illinois. Nevertheless, more worK to
determine a sediment-concentration standard is probably necessary.
\
One problei.i with defining a standard is that it often implies a critical
value above which a certain use is impossible and below which the use is un-
impaired. While it is conceivable that such a c?se may exist, there are com-
monly a broad range of concentrations over which a use could take place. The
quality or extent of use often decreases as the pollution concentration in-
creases. For example, Sparks (1978) points out that game fish are able to
survive in a range of mean turbidity levels, but fish populations and rates
of catch by fishermen might be greater at lower turbidity levels. In theory,
one would set an optimal turbidity standard by comparing the benefits and
costs of reducing turbidity to various levels.
Cairns (1968), writing specifically of sediment standards for fish spe-
cies, argues that ultimately stream standards should be based on stream con-
ditions rather than on arbitrary fixed standards. Variations in its ecologi-
cal conditions need to be recognized in determining the water quality stan-
dards for each stream. Cairns argues that this process requires three stages
in netting standards. First, a disregard of the ecosystem brings about
Juiages. Hopefully we have left this stage behind. Second, arbitrary stan-
dards are applied within jurisdictions after the damage caused by no re-
straints is seen. Third, regional variations in ecology are recognized,
resulting in more flexible standards. The time and costs required for re-
search to obtain the necessary information may make the third stage difficult
to attain.
There are good economic reasons for opposing uniform criteria. First,
even under "oristine conditions," some of the standards could conceivably be
exceeded ^n some streams. Second, resources allocated to reducing pollu-
tants in order to meet water quality standards might be misallocatrd to re-
ducing pollutants that do not limit the propagation of desired fish species
or other desirable uses of the waterway.
Given that information is lacking witn which to determine many water-
quality criteria, it will be difficult to vary water quality criteria for
individual streams by basing the criteiia on the streams' ecology, as sug-
gested by Cai-ns (1963). It woiTd, however, be possible to modify "arbi-
trary standards" to specific uses and stream conditions by using professional
judgments by experts such as ecologists to suggest variations in the criteria
for different user and streams would become subjective and, perhaps, prone-
to challenge. This process, however, could help to define the most appro-
priate pollutants to control in a watershed and, oerhaps, reduce the cost of
pollution control.
Physical Processes
The report will now focus upon the pb/sical processes involved in sedi-
mentation, an important component of nonpoint pollution. Models which have
been developed to describe these processes will then be discussed. An
12
-------
examination ot tne strengths and weaknesses of the models for an ISWQM
planning approach will be discussed.
' Nature of Sediment
The sediment that is carried from Midwestern corn-belt cropland down-
stream is predominantly fine-sized material. Figure 1 graphically relates
grain sizes to various particle-size classes (American Geophysical Union and
reported by the American Society of Civil Engineers, 1975). For 51 samples
of sediment deposited in nine typical reservoirs in Illinois draining water- ¦
sheds of 7.5 to 668 sq km (1.9 to 258 sq mi), the distribution of sediment
grain size i: as follows: sand, 1 percent; silt, 52 percent; clay, 45 per-
cent; and organic matter, 2 percent. Thus, the sediment depositing in
Illinois and presumably other midwestern reservoirs is predominantly clay
and silt with almost"no sand present.. "Eroded" soil that enters streains'ahd
reservoirs tends to be more finely grained, op the average, than the sur-
rounding surface soil (Swanson, Dedrick, and Weakly, 1969; Young and Mutch-
ler, 1969), but, soil particles may move as aggregates (Foster and Meyer,
1972). These fine-sized materials are probably carried mostly as washload
rathe." than bedload (the distinction between wash load and bed load is dis-
cussed in the section on streambed sediment transport) and thus are a func-
tion of the upland soil erosion. At any particular time the sediment load
in the stream contributed as wash load is a function of the rainfall and
runoff from the cropland.
Currently, most professionals still have doubt as to the effect of up-
land soil conservation in reducing downstream sediment yieid. There is some
evidence, however, that tne traditional soil conservation practices do reduce
sediment yield. These practices seem to reduce the silt and clay-sized
sediment particles reaching the stream. These particles are transported as
aggregates having an equivalent size to coarse silt particles. A study by
Stall (1962) describes 17 documented cases of soil conservation effective-
ness. The author located eight U.S. reservoirs that had been subjected to a
detailed measurement of sediment. Considerable soil-conservation measures
had been carried out on each of these watersheds. After a reasonable conser-
vation program had been applied to each watershed, the reservoirs were resur-
veyed to document the reduction in the sedimentation rate. Of these eight
reservoirs, a 28 percent reduction in reservoir sedimentation was shown in
Lake Crook at Paris, Texas; and a 73 percent reduction was shown in City
Reservoir No. 3 at Fairfield, Iowa.
The study by Stall (1962) also evaluated four watersheds in which sus-
pended sediment was reduced. These four watersheds were all in the Tennessee
Valley. In each case, forest cover had been increased, check dams built,
vegetable ccver increased, and improved pasture programs carried out. The
suspended sediment and streamflow were measured at stream gaging stations.
Reduction in suspended sediment over the years ranged from 33 percent at
Parker Branch Watershed to 90 percent at Whiterollow Watershed. This study
further describes the reduction in river turbidity that was accomplished by
a reforestation program in Georgia. There tree planting, improved forest
management, contour farming, grassing of waterways, a,id skid-trail stabiliza-
tion efforts were carried over a period of years. During this same era, the
13
-------
1000
100
GRAIN
SIZE,
'nn 1"1 i meters
10
1
i00r 0.1
GRAIN
SIZt' 10-0.01
nil crons
1 LO.001
t
BOULDERS
i
MEDIUM
SMALL
COBBLES
LARGE
SMALL
VERY COARSE
¦- • COARSE
GRAVEL
MtDIOi'i
FINE
VERY FII1E
VERY COARSE
COARTE
SAND
MEDIUM
nrJE
VERY FINE
COARSE
SILT
MEDIUM
FINE
VERY FINE
CLAV
COARSE
MEDIUM
1024
512
255
128
64
32
16
8
4
GRAIN
2 SIZE,
mi 'I I iii.eters
1
0.5
0.250
0.125
0.062
0.031
0.016
0.003
0.004
0.002
0.001
Fig. 1. Names of
i
Grain Sizes
•-1
-------
turbidity of river water was documented at five different water treatment
plants. At each of these plants the river turbidity was reduced. At Augus-
ta, Georgia, on the Savanna River, turbidity was reduced by 44 percent; and
at Macon, Georgia, on the Ocmulgee River, the river turbidity was reduced
by 89 percent.
Although the results described are from only 17 selected cases, it does
seem in each case that upland watershed conservation programs were effective
in reducing the sediment yield downstream. Nevertheless, further documented
studies are nteded.
Runoff and sediment concentration vary over time. Storm events can
result in overland flow of water which moves sediment overland and into
streams. Eetween :-tcrm events, runoff and sediment concentrations wiT de-
crease. The Vermilion Piver at Pontiac, Illinois, draining 1500 sq km
(579 sqmi) provides an example of how sediment and runoff vary over time in
a typ.cal midwestern watershed. The solid line in Figure 2 is the hydro-
graph of flow that occurred during August 1977. T,>e dashed line in Figure 2
graphs the suspended sediment carried down the river by this storm, Shown
in Figure 3 is a duration curve of suspended sediment in the Vermilion River
at Pontiac for the year 1977. As can be seen, the annual geometric mean
value is 102 mg/1. The median is :0 mg/1. For 60 percent of the days,., or
219 days per year, the suspended sediment in this stream exceeded the level
of 30 mg/1 considered by some to be detrimental to fish. During the rest of
the year, sediment concentration was below 30 mg/1.
Upla/id Erosion Processes
The universal soil loss equation (USLE), developed from measurements on
nationwide experimental upland plots, calculates the upland soil loss that
will leave the slope of the -farmland if a free outfall exists at the downhill
end of the slope. Under a condition of free outfall there are no obstruc-
tions to the flow of watur or sediment. Usually the water runoff "and the
soil particles that leave the slope enter an almost flat swale, depressional
area, strip of vegetation, or nonincised drainageway at the downhill end of
the slope. Often most of the runoff water is held back from the nearest in-
cised channel, or ditch, and the water infiltrates into the soil. The sus-
pended soil particles are mostly deposited at the end of the slope in the
same field from which they originated and are not delivered to the stream.
Several researchers have investigated "the process of deposition at the
end of a slope by examining sediment transport down a uniform slope and a
concave slope with the same loss in elevation over the same distance. Neib-
ling and Foster (1977) found that the amount of soil lost from a concave
slope that they tested was 23 percent of the soil lost from a uniform slope.
Your.g and Mutchler (1969) found that the sediment delivery from a concave
slope was 38 percent of that on a uniform slope. The reason for the lower
sediment delivery on the concave slope, of course, is that there was a large
deposition of soil on the lower portions of the concave slope where the
surface was becoming flatter.
A nonincised drainage way provides backwater onto the slope, creating a
15
-------
AUGUST. 1977
Fig. 2. Hydrographs nf flow and suspended sediment for a storm during
August, 1977, on the Vermilion River at Pontiac, Illinois.
16
-------
SUSPENDED SlDIKEW, nig/1
o
o
c>
o
«x>
o
rn CO
O
Oj
•H O
m
o
~ en
XI o
o
—
A
S"M
/
1
¦ /
! /
1
i
, /
J "
>
1 0"^
1 '
I i7'
1 --
1 r:!
J
f
m
O —
>
ih
/
U
1 '»
1 „
—< i r~ 1
O m .
•= o
(73 1 n '
|
102
/
^ | f" '1""
rn O [
1 rr»
j
/
J
CO
1 rn
/
T •—(~
to I
o 1 i-
/
|
-------
condition diffeient from the free outfall at the experiment station plot.
Nonincised drainage ways do not have distinct channel? and can be crossed by
a farm tractor and implement. Incised drainage ways are distinct channels
of such size, shape, and magnitude that they cannot be crossed by a farm im-
plement and tractor. Road ditches, natural ditches, or drainage ditcher are
incised drainage ways. If there is an incised drainage way at the downhill
end of a cultivated slope, more of the upland soil is delivered to the stream
than if there is a nonincised drainage way. Stall ana Bartelli (1S59) found
that the quantity of sediment delivered to a reservoir decreased as the total
length of nonincised channels in the watershed increased. The quantity of
sediment trapped in the watershed system increased as the length of nonin-
cised channels increased. Incised drainage ways trap very little sediment.
The presence cf vegetated areas can :lso have an impact upon the quanti-
ty of sediment delivered to a stream. If runoff water leaving a cultivated
slope and carrying soil particles is emptied onto a. grassed area, such as a
fencerow or pasture, soil particles will settle out in the grass. The flow
of water will move across such a grassed area to reach a di t:h, After a
period of time, and if the water runoff is sufficient, the flow will cut a
ditch or channel through the grass to create an incised channel or channels.
When such channels form, they deliver more soil from the upland to the stream
system, and Downstream sedimtnt delivery increases.
Tollner c.t al. (1976) simulated the effect of sediment trapped by grass
strips. They predicted that the efficiency of a grass strip in trapping soil
particles would vary with the size of the particles. The trap efficiency of
grass strips varied from 80 t° 100 percent for medium stand sand and from 1
to 5 percent for fine silt. For a more complete discussion of the impacts of
vegetation on stream processes, see Karr and Sch1o_>scr (1977).
To better estimate sediment delivery from n.idwestern cropland, studies
should be conducted co study runoff flows over typical soil types, with the
zone between the farmed field and the nearest incised drainage way being the
focus of research. Perhaps conditions in this overflow zone can be charac-
terized to explain and predict the variations of sediment delivery which have
been observed in Illinois to range from 10 to 50 percent of the upland soil
Irss.
Models of Sediment Processes
As a part of this study, a number of hydrologic and sediment models have
been examined to determine their suitability for use in sediment studies.
For a model of sediment movement to be valuable in the Midwest, it should be
capable of showing new fanning practices will affect the ¦tream sediment.
The model should also provide the temporal and spatial vat lability of the
stream sediment concentration.
The models discussed in this section will be classified into two groups:
planning models and research models. Planning models have been developed
enough and are economical enough to be used for resource planning. Research
models are developed for the purpose of furthering the state of knowledge of
sedimentation processes. These roncepts may be incorporated into planning
IB
-------
models in the future. It is felt, however, that these models have not been
adequately tested, arc too expensive, or require too much data to be classi-
fied as planning models.
l-'ddwer.t p.?l.ivsvy
Sediment delivery relates upland soil losses to the amount of sediment
delivered to some downstream point. Sediment found in the stream is alnost
in/anably a lesser amount than that calculated at the upland soil loss.
The Task Force on Agricultural Nonpoint Sources of Pollution (1978) states
that sediment delivery in Illinois streams varies from 10 to 50 percent of
the upland soil loss. The American Society of Civil Engineers (1975) in
their. handbook.Szdincnt &:$¦>'net?Ln.j provides_a graph showing sediment deliv-
ery. In is graph is based on a group of nationwide studies, and fh'e'same
curve is provided by the U.S. environmental Protection Agency (1975). This
curve is derived primarily from data from Texas, Oklahoma, Nebraska, Missis-
sippi, north Carolir.a, and Georgia. It is probable that in these regions
the soil is predominantly sandy. If so, this curve is not particularly
valid for the Midwest, exespt for areas where there are sandy soils.
A study ir. Illinois in 1977 compared the calculated upland soil loss
from the 086 sq km watershed nf Lake Springfield in central Illinois to th?
sediment deposited in the reservoir. It ".'as found that the sediment deliv-
ery was 25 percent. If the value provided by the American Society or Civil
Engineers (1975) were used, sediment delivery would be about 3 percent.
Consequently, this example shows that sediment delivery in Illinois, and
probably :n the Corn Beit, varies widely from the curve and table presently
used in the United States. "In cases where the sediment is transported pri-
marily as wash load, this- sediment j^ury?; does-not seem to be valid. Be-
cause a better on* erst.ar.ding of sedTwent"del ivory is badly needed, further
studies should be carried out to determine what factors govern sediment
delivery.
01 vca":h:.: O-sduncr.z ?r.iy:r,pcy.'V-
Streambeds ave carved by flowing water and are affected by the sediment
carried by the stream. The strcambed rraterial is usually much larger than
silt or clay and is usually comprised cf sand, gravel, cobbles, or boulders.
Einstein (1950) devised an equation to calc.^ate the transport of bed mater-
ial by a stream. He defined wash load as being that part cf the sediment
load th;jt consists of grain sizes finer than those'^synd irr the bed. As
can be seen in Figure 1 , the gravel and cobbles are much larger than the clay
and silt that comprise the wash load.
A stream system develops its hydraulics in accord with a pattern de-
scribed by Strahlc-r (ISo'i). Hydraulic characteristics are related to the
stream order. Stall and Yang (1970) shewed that for 12 river basins :n
various parts cf tnc- United States, the hydraulic characteristics of the
stream were related to th? flow and to the drainage area. Divinage area was
used as a substitute for stream order. The widtn, depth, cross-sectional
area, and velocity of the stream were shown to increase in a oownstream
direction alonq the stream, following (j consistent pattern termed iyidvcrilia
-------
geometry. Vang and Stall (1974) showed that virtually all important physical
stream processes are a function of the stream's unit stream power (USP),
which is a measure of the rsLe of energy expenditure of the stream. Yang
(1971a) showed that the tendency of the river to meander is governed by its
USP. He also showed as a result of its USP (1971b) that the stream has a
tendency to build riffles and pools along its bed. In addition, Yang and
Stall (1974) showed that the sediment carried by the stream is best explain-
ed by the USP concept. For two natural rivers the USP has been shown to be
superior to several other important suspended-sediment equations. Analysis
of the sediment carried at 17 stream-gaging stations showed it to be domin-
ated by the river's USP.
It was concluded by Stall and Yang (1970) that stream adjustments are
made-and -governed by the USP concept.. Ihere,.is .a hierarchy pf levels through
which the stream system can adjust itself to minimize the time rate of po-'
tential energy expenditure per unU weight of water, that is, the USP. This
hierarchy is as follows:
First level—The stream increases bed roughness by building bed forms
such as dunes and antidunes.
Second level—The stream creates bed undulati.ons that depart from the
uniform bed slope to form riffles and pools.
Third level--The stream attacks its banks to create meanders in the
course of the stream.
Fourth level—The stream flow carves a longitudinal profile that has a
slope in every reach adjusted to provide minimum stream
power throughout the stream system.
The energy forces within the river dominate the construction and adjust-
ment of the stream bed, including the transport of bed sediment by the stream
as bed load or suspended load. These processes are virtually independent of
the presence or absence of the wash load.
The movement of the wash load is thus superimposed on the movement of
the bed material; the movement of the stream's wash load stops only when the
stream flow stops. If upland soil losses are high, "the amount of sediment
delivered to the stream as wash load is high. If soil conservation measures
are applied to cropland and soil losses delivered to the stream are reduced
considerably, the wash load delivered to the stream is less. In both cases
the stream will carry downstream virtually all of the sediment delivered to
it. It seems that this reduction in washload will not 'nave an important
effect on the tendency of the stream to carry its bedload and to adjust its
bed.
The clay and silt-sized particles carried as wash load may be deposited
in pools in the stream during low flow periods. When high flows occur,
these particles may be scoured out and again become wash load to be moved
downstream Game fish usually prefer to lay eggs in a sand or gravel stream
bed. Clay or silt, carried by the stream as wash load and deposited in the
20
-------
stream bed, will cover the bed with mud, a much less desirable substrate or
bed material because the fish eggs are not protected in the crevices as in
che sand or gravel.The riffle reaches of the stream, however, will probably
have a high enough flow velocity; even during low-flow periods, so that the
stream bed will retain a sand, gravel, or cobble-sized bed material.
Planning Models
Stream Ti^ansport Equations
To understand movement of stream sediment tnroughout a basin, an evalua-
tion is made here of sediment-transport equations. As described earlier, the
concept of USP is a valid and dominant means for calculating and understand-
ing -the movement of sediment--in-streams. It is also possible _to..use the.Ein.-
stein equation and modified Einstein equations, or several other approaches.
These equations were al~ developed for sediment sizes much larger than the
wash load in Illinois and midwestern streams. This fact indicates that
streambed sediment is moved in a separate regime from upland soil loss or the
movement of the fine sediment that is carried by streams as wash load. Since
the detrimental impacts of bedload in the Midwest are considered relatively
minor in relation to those of wash load, they could be neglected in land-use/
water-quality models in that region without creating major errors.
Empirical Relations
A number of empirical models describing statistical correlations be-
tween upland factors and the amount of sediment produced have beer devised
for local regions. Empirical models providing physical input of the land
factors often use some version of the USLE, or the same type data as used in
the USLE.as input. When these factors are correlated with the downstream
sediment yield, a statistical correlation factor cr sediment-delivery rate
is often incorporated as a statistical coefficient. Williams (1975a) writes
such a relation as follows:
ce
S = 95(Q qp) K LS C P (Eq. 1)
where
S = sediment yield in tons
Q = volume of runoff in acre/feet
qp -- peak flow rate in cubic feet per second
K = soil erodibility factor
LS = slope length and degree factors
P = erosion-control practice factor
This model was developed from 11 watersheds of about 7 hectares (3 acres),
about five watersheds of 1.5 sq km (1.6 sq mi), and one watershed of 16 sq
km (6.8 sq mi). These basins were located in Texas and Nebraska. This model
has been evaluated and described by Mulkey and Falco (1977).
Empirical methods of relating sediment yield to watershed factors show -
some promise and seem to have value. The method described by Williams -
21
-------
(1975b) is primarily a statistical correlation and does not simulate physical
processes. The effects of upland farming practices on sediment yield cannot
be defined objectively. The transferability of results from one place to
another is not po~siblo. The internal coefficients and exponents need to be
determined by the statistical analysis of considerable observed data in any
region in which they are used. The Williams (1975b) relation, for example,
was developed for sandy soils in Texas and Oklahoma. While the relationship
may be valid for areas with sandy soils, it probably would not be valid in
the Midwest, where sediments are fine in size.
Williams (1978) describes a sediment-graph model based on an instan-
taneous-unit sediment graph as an addition to this sediment-yield model.
The sediment is routed by a time-delay scheme described by Williams (1975a)
to" provi'de' the travel time of^sediment" particles to-a- do\ istream-location.
The sediment-graph model was derived for agricultural watersheds in the Texas
blackland prairies and may provide useful results for areas with similar
soiIs.
LAUDRUN
LAilDRUN is a model developed at Marquette University in Milwaukee and
described by liovotny and Chin (1976). LANDRUN has two submodels linked
together to - in sequentially. One predicts watershed runoff, and the other
predicts watershed sediment yield. The LANDRiJN model has been used by
Sharpe and Berkowitz (1979), linked with a linear-programming model to
evaluate the economics of farm management. The results provide a straight-
forward analysis of the Kewaskum-North watershed of 172 hectares (425 acres).
LANDRUcalculates the hydrologic budget with Equation 2 as follows:
Q = P - 1 - DS (Eq. 2)
where
Q - runoff
P = rainfall
I = infiltration
DS = depression storage available
Infiltration is calculated using an equation devised by Holtan (1961).
Seasonal values for the infiltration function are provided for various
months of the year and for various crops.
LiNDRUN generates sediment yield in a manner similar to the USLl but
using the modified Equation 2 as follows:
SL = E LS K C P DR (Eq. 3)
where
SL = soil loss in tons per acre per simulation period (can be expressed
in metric units)
E = rainfall runoff energy factor
where LS, K, C, and P are the same as for the Williams empirical method (Eq.
2) and Equation 1 and where DR - a delivery ratio factor. The rainfall
runoff factor .E addresses -the fact that both raiafall and runoff can initiate
erosion, and it is calculated with Equation 4 as follows:
22
-------
E = RA + RU
(Eq.4)
where
E = rainfall runoff energy factor
RA = rainfall energy factor
RU = runoi'f energy factor
The factor RA is the R factor taken directly from the USLE and is the kinetic
energy of the storm multiplied by the maximum 30-minu;e intensify. The fac-
tor RU is the storm runoff volume multiplied by the peak runoff rate raised
to the one-third power. The delivery ratio, DR shown in Equation 4, is made
up of two delivery ratio factors calculated using Equation" 5:
The'firs't"factor, DR , assumes that the' amount of's'ofl 'Toss delivered is'"'
related to the fraction of rainfall that infiltrates as shown in Equation 6:
DRa = delivery-rate factor for rainfall
I = infiltration
P = rainfall
It can be seen that DRa is essentially a percent runoff factor.
The second factor, DR. , is also a seuiment delivery and is the tradi-
tional sediment delivery ratio incorporated into the LANDRUN model. It is
derived from observations from the watershed being modeled. For the Kawas-
kum-North basin, the delivery ratio, DR., of 10 percent yielded the best
correspondence between the model and observed values.
The LANDRUN model uses predominantly physical relations. It has the
one critical coefficient for sediment delivery DR. , which is determined by
local observations. The input needed for the LANDRUN model is virtually
equivalent to that needed for the USLE. The example described by Sharpe and
Berkowitz (1979) was for a basin of only 172 hectares. Its use on larger
basins would probably be limited by the amount of input data required. The
LANDRUN model has the advantage that the effect of upland farming practices
on downstream sediment yield is based on physical predictable factors. This
mod;! appears tc be rather good for sediment-yield studies.
The ANSWERS model was developed by engineers at Purdue University for
use in the Black Creek demonstration project in Indiana sponsored by the
U.S. Environmental Projection Agency. The results of this project, includ-
ing a description of the ANSWERS model, are provided by Lake (1977). The
model is based on strongly supported physical relations, and it requires
input similar to that needed for the USLE. Data from a contour map of the
basin are also required. ANSWERS is a distributed model that provides a good
representation of the spatial variability of land forms, land use, and hydro-
lcgic processes. It is also an event model, which means that it primarily -
DR = DP. x DRb
(Eq. 5)
(Eq. 6)
where
ANSWERS
23
-------
moaels the results of one storm. It is possible, however, that the model
could be used to estimate temporal variability of runoff and sediment by the
use of a group of design storms that coul,d be developed. ANSWERS is a deter-
ministic model based on the idea that at every point within a basin there is
a functional relation between the rate of surface runoff and the rainfall
intensity, infiltration, topography, soil type, and otner upstream factors.
Further, these surface runoff rates can be used to model transport-related
functions such as soil and chemical movement within the basin. The soil
particles are detached by overland Flow when the shear stress at the surface
is sufficient to overcome the particles' gravitational and cohesive forces.
Inter-rill erosion is described by' equations published by Foster, Meyer,
and Onstad (1977). The Holtan equation is used for infiltration.
In the ANSWERS model, the detachment of soil particles by raindrop im-
pact is calculated using the relationship described by Meyer and Wischmeier
(1969) and given as follows in Equation 7:
DR = .027 C K Ai I2 (Cq. 7)
where
DR = rainfall impact detachment rate, kg per min
C = cropping management factor, same as in USLE (Eq. 1)
K = soil erodibility factor, same as in USLE (Eq. 1)
A.j = area increment, sq meters
I = rainfall intensity, millimeters per minute
The detachment of soil particles by overland flow in the ANSWERS model is
calculated as described by Meyer and Wischmeier (1969) and modified by
Foster (1976), and given as follows in Equation 8:
DF = .018 C K A. S Q (En. 3)
where
DF = overland flow detachment rate, kg per min
S = slope steepness, or degree
Q = flow rate per unit width, sq meters per min
The ANSWERS model uses an expression for particle transport in overland flow
that is based generally on a paper by Valin (1963) and is given as follows
in Equation 9:
T = 146 s/q~ (Eq. 9)
if Q is eaual or less than .74 sq meters
and T = 14,600 S Q2 (Eq. 10)
If Q is greater than .74 sq meters
where
T = potential transport rate of sediment, in kg per min, per meter
In a final report on the Black Creek project by Lake (1977), the ANS-
WERS model was calibrated for a small basin of 714 hectares (1,765 acres or
24
-------
.75 sq mi). For one storm the rainfall was 6<1 millimeters. The calculated
runoff was within 19 percent of trie guaged amount, and the total sediment
yield was within 13 percent of the observed amount. The report provided a
runoff hydrograph and a sediment graph that seem valid and reasonable. The
ANSWERS model also provided a map showing isolimos of sediment loss for this
71/1 hectare basin. The ANSWERS model was thus used to evaluate the effects
of contouring and minimum tillage on selected regions within this basin. It
was illustrated that by controlling erosion on just two small parts of this
basin the downsti ^am sediment yield was reduced significantly.
The example provided in the Black Creek project was for a relatively
small basin of only 714 hectares. The present program, however, can handle
a maximum of 2,000 watershed elements, which means that it could be applied
to a basin of about 81 sq km (31.2 sq mi).
A P.'-.'
The ARM model is the agricultural runoff management model. It has been
devised by the Hydrocomp Inc. company under the sponsorship of the U.S.
Environmental Protection Agency. This model has been described in important
publications by Donigian and Crawford (1976) and by Donigian et ol. (1977).
The ARM model contains soil-loss functions that are different from most
others but thi11 appear to be valid. It is based on the continuous simula-
tion of flow originally conceived as tfie Stanford watershed model and later
developed by the Hydrc-comp Company as their hydrologic simulation program-
ming, iiSP. The HSP has been used widely throughout the United States in
solving many hydrologic problems. Accompanying the basic hydrologic model
are models of pesticide absorption-desorption, pesticide degradation, and
nutrient transformations. In the ARM model, only the sheet-and-ril1 erosion
is included. The detachnient of soil fines and silt and clay fractions (by
raindrop impact) and the transport of soil fines by overland flow are calcu-
lated.
In the ARM model detach-nent of soil fines is calculated usin Equation
11 as follows:
RER(t) = (1 - COVER(T)) * SMPF * KRER * PP.(t)JRER (Eq. 11)
where
RER(t) = soil fines detached during time interval t,
in tons per acre
COVER (T) = fraction of vegetal cover as a function of time, T,
within the growing season
SMPF = crop-management factor, same as P in the USLE (Eq. 1)
KRER = detachment coefficient for soil properties
PR(t) = rainfall during the time interval, inches
JRER = exponent for soil detachment
* = multiplication
In the ARM model the -crane port of soil fines is calculated using Equation 12:
SER(t) = K5ER * 0VQ(t)JSER
if SER(l) is equal or less than SRER(t)
2b
-------
= SRER(t)
if SER(t) is greater than SRER(t)
(Eq. ro
where
SER(t) = transport of fines by overland flow, tons per acre
KSER = coefficient of transport
OVQ(t) = overland flow occurring during the time interval tv in inches
JSER = reservoir of soil fines at the beginning of time interval t,
In the operation of the model the soil fines detachment RER during eaoh 15
minute time interval is calculated by using Equation 11 and added to the
total fines storage SRER. Next, the total transport capacity of the ove>land
flow, SER, is determined using Equation 12. Sediment is assumed to be trans-
ported at capacity if sufficient fines are available. Otherwise the amcunt
of fines in transport is limited by the fines storage SRER. The sediment
loss to the waterway in a time interval is calculated using the fraction of
overland flow that reaches the stream, as in Equation 13:
where
ERSN(t) = sediment loss to the stream during the time interval t,
in tons per acre
F = fraction of overland flow reaching the stream during the
time interval t
Land covered by growing crops and crop residue has an impact on sediment
loss. It is allowed for in the ARM model using the land-cover variable in
Equation 11 denoted as COVER(T)» which represents the fraction of the land
surface effectively protected from the kinetic energy and detachment capa-
bility of rainfall. Monthly values of this cover are specified. Judicious
use of the land-cover function allows the simulation of various land-surface
conditions for different practices. The effect of tillage operations is to
increase the mass of soil fines available for transport.
As reported by Donigian et al. (1977), ARM can represent runoff, sedi-
ment, pesticide, and nutrient loads from small agricultural watersheds in
Georgia and Michigan. The calculated sediment and nutrient levels have
agreed fairly well with the observed levels. The model shows that the loss
of nitrates and soluble pesticides is highly related to the simulated
interflow component of water ri'Noff. That is, a change in the amount of
surface runoff and an increase in the amount of water infiltrating into the
earth provides a change in the nitrate and pesticide constituent. It is
reported by Donigian or. al. (1977) that watersheds greater than 2 to 5 sq km
(0.8 to 1.9 sq mi) are approaching the upper limit of applicability of the
ARM model, because this model provides for continuous simulation of flow,
sediment, nitrogen, phosphorus, and such things as oxygen, it is probably
the best of those evaluated for the purpose- of depicting sediment and water
quality. It reveals secondary effects like the change that conservation
practices will bring about in subsurface flow K'd the resultinq change in
the nutrient budget.
The input data needed to operate the ARM model are considerable. It
in tons ner acre
ERSN(t) = SER(t) * F
(Eq. 13)
26
-------
requires relatively large computer storage and large amounts of computer
time. In addition, the maximum basin size of 5 sq km is probably a limiting
factor. These problems mean that the ARM model will probably riot be adopted
for basinwide water-quality planning.
The 1079 SEA-AR 5edir;*ent Planning Model
The U.S. Environmental Protection Agency (1975; 1976) published Volume 1
and Volume 2 of the report Control of Water Pollution from Cropland. Volume
1 was a handbook and Volume 2 was an overview of how sediment moves from
•cropland into streams and how pollutants similarly move. These two volumes
summarized the available information on this topic from the Agricultural
Research Service (ARS) of the U.S. Department of Agriculture. In 1978, the
agency was renamed the Science and Education Administration—Agricultural
Research, SEA-AR.
During 1978 the SEA-AR took a new initiative to provide a new field-
1 scale sediment planning model. The immediate purpose of this effort is to
estimate the movement of pollutants from agricultural lands to streams and
to assess the effectiveness of current and future land-use practices on this
movement. The scope of this effort has been to compile a state-of-the-art
combination of known results from agricultural experiment stations in the
United States to make a readily usable system, on a field-size scale, to
quantify the effects of best management practices (Bf'Ps) to reduce pollutant
movement.
This SEA-AR sediment model is to be applicable at the field size, that
is, perhaps 20 to 40 acres. The input to the model is to be simple, straight-
forward information of the type readily available. The model is conceived
to be a compilation of unit practices and information that would have some
similarity to the USLE. The model is to characterize the results of exist-
ing agricultural physical conditions on croplands and existing cropland prac-
tices to predict soil loss and sediment movement as well as the movement of
pollutants off of a field. The model is to contain: a group of hydrology
functions, a group of erosion and sedimentation functions, and a set of
chemistry functions. The operation of all of these functions is to provide
an understanding of the movement of sediment and pollutants from the field
slopes into the streams leaving a field-sized areas. A working group of 50
to 75 research scientists of the SEA-AR have been assigned to provide the
tiodel functions within these three principal areas of the model.
The input needed for the hydrology model is either daily or hourly
rainfall. Its output will divide the rainfdll into two categories daily:
runoff and infiltration into the earth.
The chemistry model will require an input as to the nutrients in the
watershed soil. It will utilize the results jf the hydrology-and-erosion
model and will provide estimates of the nutrients that leave the field in
the water runoff, that leave the field with or attached to the sediment, and
fhat leach through the soil to the groundwater. The chemistry model will
also provide for the nutrients utilised in plant growth, which amount is sub-
tracted from the nutrients available for leaching into the groundwater. This
27
-------
model will require input as to the properties of various pesticides as to
whether they are normally transported in solution or attached t.c sediment.
Also.tc be provided is an enrichment ratio for pesticide-transported pesti-
cides that, are preferentially transported by clay and organic matter.
The ?rosion~cediment functions in thns model will be based on the physi-
cal laws governing the detachment, transport, and deposition of soil parti-
cles as these factors occur on the watershed field. The required informa-
tion will largely be those factors required to sclve the USLE. In addition
to that information, data will be required on the primary size distribution
and the specific gravity of the particles eroded from the soil. It is felt
that this type of information can readily be obtained. Thc- sediment-yield
model will provide the amount and concentration of sediment moved from each
element of the field during each stonn. It will also provide tne total
sediment yield.
Because the eros ion-sedimentation mode"1 is based on the physical laws
governing the detachment, transport, and deoosition of sediment particles,
it will not be required to estimate what is usually called acdirneni delivery.
The detachment, transport, and deposition functions are calculated for ele-
ments of the field-size watershed based or existing farming practices. It
is also readily possible to calculate the same detachment, transoort, and
deposition amounts of the sediment as a result of particular soil conserva-
tion measures. This factor will make this model useful in evaluating upland
soil conservation.
The 5EA-AR sediment planning model will be useful to generate peak
amounts of flow, sediment, pesticides, and nutrients and to evaluate acute
pollutijn problems. It will also simulate the general daily runoff and per-
colation losses to use in the evaluation of chronic pollution problems.
The selected research personnel within the SEA--AR have been focused
during the past year on the development of this state-of-the-art sediment
planning model. In addition, the group is to approach the development of a
sediment planning model for a watershed or basin-size area that would model
areas much larger than the 1979 field-scale mv-.el.
The State of Sediment Research Modeling
and Foatav (137?)
NeiMing and Foster (1977) describe sediment yield frem overland flo.v
processes. This study derives and provides considerable theoretical support
for equations to represent the physical processes involved in sediment de-
tachment, transport, and routing. The authors report that eroded soil parti-
cles and that most overland Flow of aggregates and sand upon entering the
stream is transported as bed load. As a result, they use the Yalin sediment
transport equation and calibrate- it for the agricultural field conditions.
Thpy develop a soil-ciodibility graph similar to the one described by Wisch-
meier, Johnson and Cross (1971). This nomograph was developed for 40 differ-
ent bare Indiana soils.
?8
-------
The authors develop an equation to provide the basic relationship be-
tween transport capacity, runoff flow energy, and slope length and steepness.
The sediment routing algorithm described by Meyer and Wischmeier (1569) is
ucilizeu. An example analysis is made for bare soils of a uniform slope and
a composite slope showing the sediment availability, transport, and deposi-'
tion for various parts cf the slope.
Li, Simons, and Shiao (1977)
Li, Simons, and Shiao (1977) describe and provide ar elaborate research
mathematical model describing the entire soil detachment and transport pro-
cess. A silvicultural application of this model is discussed in Li, Eggert,
and Simons (1979). In this model, the interception and infiltration losses
are subtracted from the actual rainfail. The remaining water becomes runoff.
The..runoff wave is calculated using a kinematic wave model that makes the
modci costly to use. This process has been described elsewhere'by Li", "Simons,
and Stevens (1975). This routing procedure is used to compute overland flow
and channel flow. The overland flow computations include the effects of
rainfall on the flow resistance. For shallow flow the impact of raindrops
causes energy loss in the flow in addition to that caused by the rigid
boundary of the channel. The validity of this routing procedure has been
calibrated and confirmed on a small paved parking lot at Johns-Hopkins Uni-
versity in Baltimore.
The Li, Simons, and Shiao (1977) sediment model calculates the sedimc-nt
supply as that caused by soil detachment by raindrop impact and surface run-
off. The sediment load is broken into two main categories (bed material and
wash load) and into a specific number of size fractions. The transporting
rate for each sediment size is divided into the bed load transport rate and
the suspended-load transport rate. The Meyer-Peter bed load transport equa-
tion is used although the authors state that any suitable bed load transport
equation might be used. The suspended load is calculated as a function of ,
the sediment concentration at any point above the bed.
The soil-detachment function includes an accommodation of surfece
changes on the soil as a result of armoring. The sediment size Dg^* is used
to estimate the thickness of the armored layer.
This sediment model is generally calibrated ori a bare area of mine
spoil in Colorado about which considerable data are available on particle
size, slope length, and degree as well as the size distribution of the eroded
material. The calculated and simulated hydrographs and sediment graphs
reasonably compare.
This model is a valuable research tool. The understanding developed
with it will undoubtedly be of value in future sediment studies.
Simons et at. (1977)
Simons et al, (1977) describe a simple procedure for estimating soil
* 84% of the sediment particles will have a larger diameter than Dg^.
29
-------
erosion. For this model the interception, in 'iltration, runoff routing, and
sediment transport capacity are largely similar to the above-described model
of Li, Simons, and Shiao (1977). This procedure is more simplified, however,
in that, instead of routing all of the sediment load as a large number of
particle-size fractions, the single representative particle size Dr„ is
utilized. The authors show that the simplification of the model provide:
reasonable general results. The Li, Simons and Sniao (1977) model requires
computer time 10 to 15 times greater than this model by Simons, al.
Savings in computer time are enormous, and the results are more simplified
and limited.
Smith (1978)
Smith (1976) describes a distributed watershed erosion and sedimentation
model. In the initial ctep in this model, the incoming rainfall is divided
into the infiltration and runoff. The point-rainfall detachment r?te is
• calculated, as is the detachment rate of sediment-by-the hydraul-ics-of the -¦
flow. The model employs the (JSP relation described by Yang and Stall (1974)
as a sediment transport function. The ether inputs are generally the physi-
cal inputs required by the USLE.
This model is an experimental research one calibrated for storms on a
small watershed in Arizona. Although it is not a readily available engineer-
ing planning tool, its results are promising. It provides a time-distri-
buted output of runnff and sediment from a small upland watershed. -
, Onstadj Piest, and Saxton (1976)
Onstad, Piesc and Saxton (1976) provide a sophisticated research model.
Soil detachment is described as the detachment by rainfall energy and by
flowing water. The transport capacity depends on the slope degree, length,
and physical factors. The irput to this research model is largely that re-
quired for the USLE.
This model was calibrated for research watersheds near Treynor in
southwest Iowa. On one watershed, 62 rainfall events were evaluated, and on
the other watershed 48 rainfall events were evaluated. The model predicts
sediment yieldr. from single storms. The individual storms tested produced
sediment yields of 0.01 to nearly 50 tons per acre, and the model predicted
storm sediment quite accurately.
Woolhiser and Reruzrd (1978)
It is possi^e to calculate the sediment yield at any point along a
stream system as a complex residual of many stochastic processes. In the
paper, "Stochastic Aspects of Watershed Sediment Yield," Woolhiser and
Renard (1978) have given a description of several stochastic models. Many
components of the sediment-yield processes are best described by their proba-
bilistic structure because randomness is the very essence of the individual
processes. The most obvious source of randomness in sediment yield from
upland areas is rainfall. Other sources include factors such as the raindrop
;size, wind effects, soil-infiltration rates, and erodibility._„ Sediment-yield
30
-------
models based on stochastics range Prom detailed physically-based models with
distribution functions obtained by sampling from Montecarlo simulations to
empirical, event-based models for which distribution functions of sediment"
yield are obtained analytically.
Conclusions Concerning Physical Processes
Fillings
It appears new that there is no one model availatle for planning the
control of agricultural nonpoint.-source pollution on a basin approaching
the 13,000 sq krr: (5,000 sq mi) of one of the example cited, the Sangamon
Riv^r Basin. Because the LANDRUN, ANSWERS, and ARM models all seem to
simulate physical processes, it would seem best to use one of these models
for various representa ive areas wi thuva total -basi n-and - then generalize-,
the results using judgment based on watershed characteristics. As reported
here, the LANDRUN model was effective when' used on a 1.7 sq kin (0.G5 sq mi)
area. The ANSWERS model has been used on an area up to 81 sq km (31 sq mi),
and the ARM model has a maximum size limit of approximately 5 sq km (1.9 sq
mi). All three of these models seem to accurately reflect the physical pro-
cesses, the ARM model being the most sophisticated, the ANSWERS model being
oistributed arid an event model, and the LANDRUN model being a lumped model
that can be used to provide monthly or annual sediment yields.
'iffact of 'Basin Si:-.e
In 1979 there is no existing sediment modeling procedure-that can accom-
modate the sediment problems of an area the size of the United States, of
the Water Resource Regions, or of the principal river- basins. For these
types of sediment problems, the sediment loads of the waterways are estimated
from water samples. Various sources of sediment such as upland erosion or
strearnbank erosion are then studied to estimate their contribution to the
total problem. Sediment sources causing severe problems can be pointed out
in this way. A study of this kind has be'-n carried out in the eastern half
oi the United States by the Soil Conservation Service (1977). This"rcethod
has also boon used on a number of large river basins such as the lower Colo-
rado (Johns, 1976) and the upper Colorado (Workman and Keith, 1975), on the
Eel and Mad Rivers in California (California Department of Natural Resources,
1970), and on the upper Mississippi (U.S. Army Corps of Engineers, 1970). The
method also had to be applied to small subbasins of the upper Mississippi of
259 sq km (100 sq mi).
Existing sediment models, however, could be \ ~ed for sediment movement
out of small watersheds within a larger basin. To provide a meaningful
understanding of the sediment movement at downstream locations within the
basin by generalizing such results, more must be' kno m about sediment deliv-
ery in amount and time. The amount of sediment delivery is now predic cl by
methods that are generally valid for sand-sized sediment, for the si'i'-v.d-
clay-sized sediment of the midwest Corn Belt, rio existing method to predict
a delivery ratio without an empirical measurement is valid.
31
-------
To estimate the time of delivery of the sediment from a small watershed
to a downstream location, it is necessary to perform a hydraulic routiny of
the sediment from the upstream watershed to the downstream location. This
routing can probably be accomplished largely by a procedure that merely pro-
vides the time delay required for the sediment-laden water from the upstream
watershed to flow down to the downstream location. Williams (1978) has de-
scribed a simple time-delay routing procedure for translating a sediment
graph from an upstream to a downstream location. This procedure would be
needed if there were a concern about sediment duration curves for such pur-
poses as estimating impacts on fish. Further research is needed to provide
better understanding of sediment routing. It also appears difficult to
model the deposition of wash load in pools in the stream.
Time VaricbLLity
To understand sediment movement, it is necessary to know the time distri-
bution of the sediment load of a stream. This information is particularly
important for studying impacts on fish. When and how often does the sediment
move? This time sampling is provided in different ways.
In the ARM model, an actual record of hourly rainfall for a long time
period of perhaps 20 years is used. This rainfall is input to the model,
and hourly or daily output is simulated for the basin modele-.. This simula-
tion is continuous and gives the best possible time sampling because it
models all the events that have occurred and provides the predicted water
runoff and sediment runoff.
These results can be plotted for a particular event to show the hydro-
graph and sediment graph for that event. These graphs would look similar to
the ones in Figure 2. These same results from the ARM model Caii be analyzed
to provide a suspended-sediment duration curve similar to that shown in
Figure 3. Using ARM, the upland farming practices cculd be altered to reduce
soi! losses, and the entire 20 years cculd be resiniulated. Those results
could be analyzed to give a different, lower suspended-sediment duration
curve. The reduction in the days per year that suspended sediment concen-
trations were predicted to exceed a particular level, as in Figure 3, could
be compared with the costs of the upland farming practices thet caused the
reduction.
The LANDRUN model does not provide a hydrograph, a sediment graph, or a
suspended-sediment duration curve. As a result, it is not particularly use-
ful for studying the impacts of changing land-use patterns or. fish habitat.
It provides only the soil loss from a watershed in fons per hectare per year.
As used by Sharpe and Berkowitz (1979), however, some time sampling was pro-
vided. They developed a 20-year record of the rainfall energy factor R show-
ing its value each year. These values were ranked and assigned probabilities
within the 20-year period. The LANDRUN model was njn usii g the R factor ex-
pected at recurrence intervals of about 20, 10, 5, and 2 years. The annual
soil loss for the watershed for each recurrence interval was calculated.
A graph of annual soil loss versus return period was presented that is a
time sampling. Upland farming practices were assumed, and different sets of
32
-------
soil losses were calculated and olotted. The benefits of the reduced soi"1
losses cojld thus be cor..pared with the costs of the farming practices.
The ANSI-OS model is an event model that provides a hydrograpn of run-
off and suspended sediment tor an actual or a design storm. It does not pro-
vide a duration curve of flow or suspended sediment. To develop a suspended-
sediment duration curve, one would have to develop a set of design stonns
and assign them recurrence intervals, which could best be done by analyzing
a recorded rainfall record for a period of 20 years or more to get the occur-
rence and timing for each year of the rainfall events to provide the amount
of the rainfall energy factor R. This analysis would be similar to the one
made by Sharpe and Berkowiti (1979), which provided time sampling for the
LANDRY model. Such a record of R throughout each year of the 20-year period
cculd be used with the AKSWCR5 model to provide the time sampling.
Traits of Models
Below art- summary statements about the advantages and disadvantages of
some of the models and methods studied here.
An advantage of ull the models is that their physical input is similar
to the input factors of the 'JSLE. Thus these factors can be readily altered
by assuming that upland fanning practices .ire changed. This change in fann-
ing practices is chen reflected through every model to show the effect on
sediment yield,
A disadvantage of all the models is that they are all cmrently limited
to very small watersheds ana provide the sediment yield at the outflow of
small hydro logic units only. Moreover, no model provides any nnysical basis
for modeling the '-outing of sediment delivered from the Outlet of" this small
basin to a point .awnstream.
The timing of sediment delivery from an upstream basin to a downstream
povit is not now provided in any model reviewed here except the Williams
empirical model. It could probably be provided, however, by a simple time-
delay sediment routing. The temporary deposition of wash load in stream
pools is not accounted for, and no methods appear to be available for model -
i ng.
Summary r.f Advantages and D^advantages of Planning Models
Advantages:
Wei; based physically
Has sophisticated hydrology
Simulates all components of hydrology
Models nutrients and const- -vative pollutants
Time series output.
Is being tested on a number of data sets not used in developing
the model
33
-------
Disadvantages:
Has maximum size basin of 5 sq km
Requires as input hourly rainfall data on the basin for a long time
period
Uses large amounts of computer time and is thus relatively costly
Is a lumped parameter model
ANSWERS
Advantages:
Is well based physically
Is a distributed model; solves :hejnany _parts_of_the,basin..
Models flow and suspended sediment
Currently has a maximum size basin of about 81 sq km
Requires moderate computer time and cost
Can provide results for various sized sediments
Has been tested on data sets not used in developing the model
Disadvantages:
Is an event model
Needs a 20-year record of rainfall factor R to provide time sampling
Requires considerable .land input, data
Considers only sediment and not other pollutants at this point in time
lAi.'DRUH
Advantac^s:
Is well based physically
Meeds simple input
Provides annual soil loss in tons per hectare per year
Requires moderate computer time and cost
Di sadvantages:
Is a lumped model; solves the basin as one unit
Needs a 20-year record of rainfall factor R to provide time sampling
Has been demonstrated on only a 1.7 sq km basin
Considers only sediment
Does not provide results for various-sized sediments
Bnpii'ical 1-letkods
Advantages:
Need simple nput
Require moderate computer time and cost
Have already been developed for Texas-Oklahoma sandy soils
34
-------
Hisadvantages:
Are statistically based; not physically
Have not proven valid for silt and clay soils of the Midwest
Provide no time sampling
Often consider only sediment
Valid only for the data set used to develop the model (Do iiot provide
results for various-sized sediment.) ........
SEA-AR
Advantages:
Is well based physically ;
Has-.sophisticated hydrology „
Provides continuous simulation
Models nutrients and pesticides
Requires simple information tc predict erosion
Can provide results for various-sized sediments
Ha: been tested in data sets not used in developing the model
Disadvantages.
Has current field-size model of 20 tc 40 acres
Requires hourly rainfall data
Is a lumped parameter model
Needed Research for the Land-Use Water-Quality Linkage
A number of recommendations for future research will be made in light of
the examination of strengths and weaknesses of the various models.
Pcunage to Water Quality
1. Studies should be made of water pollutants such as fertilizers and
pesticides to determine the extent to which they are carried into
water while attached to sediment panicles.
2. Studies should be instituted to model the ecologic stream environ-
ment to define the role of the streamhed material and the sediment
carried by the stream in the total use of the stream environment by
aquatic life. For example, a typical stream reach in the humid part
of the United States could be studied. . The typical food chain,
aquatic population, and hydraulic and hydrologic factors such as
depth, velocity, and bed material would be assurr.ed. The carbon bud-
get would be modeled to explore the importance of organic matter
nutrients and sediment.
Soil Losses
3. Continuing research is needed to develop better understanding of the
soil erodibility factor K for U.S. soils. K faccors are especially
35
-------
needed for disturbed land such as strip-mine spoil and subsoils.
4. Studies should be made to determine more accurately the rainfall
factor R for mountainous areas of the United States where rainfall
is highly variable at different altitudes and is affected greatly
by orographic effects of mountains.
5. Further exploration is needed as to the variability of drop size in
rainfall in the United States and the effect of drop size on the
rainfall factor R in the USLE.
Stream Sedvnent
6. Some of the planning models described in this report need to be
tested to demonstrate their strengths, capabilities, and limita-
tions. The ANSWERS and the LANDRUN models'"sliould"'be""tested"on a
number of watersheds to check their sediment predictions.
7. The ARM model, although it is complex, should be further tested and
developed. It accomodates budgets of nutrients and other pollu-
tant matei ials not available in ANSWERS or LANDRUN.
8. When the 1979 SEA-AR field-scale sediment planning model becomes
available, it should be exercised, calibrated, and used to evaluate
stream sediment as well as nutrient budgets in various locations.
9. It would be highly fruitful to carry out applied research to cali-
brate some of the complex research models described herein. De-
tailed data from existing watersheds should be assembled. The
models of Neibling and Fuster (1977); of Li, Simon, and Shiao
(1977); and of Smith (1976) should be used to predict stream sedi-
ment. The results would begin to show the quality of our current
understanding of the soil detachment, transport, and deposition
processes that govern our prediction of stream sediment. This re-
search would also help to focus attention on the processes for
which we need better understanding and on the scope of data needs
required for better planning models.
10. Research is needed to study the particle-size characteristics of
upland soils and stream sediments. What are the particle-size
distributions actually found in streams? To what extent do soil
and sediments move as aggregates? What are the physical and chemi-
cal properties that govern the formation and stability of aggre-
gates?
11. Studies are needed of the chemical and physical relations among soil
particles, aggregates, organic matter, various nutrients such as N
and P, and other chemical constituents. Involved would be the
chemistry and physical attributes of the water, clays, colloids,
and chemical elements. Other recommendations are discussed by ¦¦
Robinson and Renard (1977).
36
-------
Upland Farming Practices
12. Studies are needed to document whether upland soil conservation
practices reduce sediment in downstream streams or lakes. It
would be valuable tc compile some case studies to show whether up-
stream soil-conservation practices reduce sediment yield down-
stream. Stall made such a study (1S62), but much more documenta-
tion is desirable. Projects should be located where extensive
soil-conservaticn measures have been install'd on a watershed and
downstream sediment has been monitored. After the project, the
impact on sediment yield should be analyzed.
13. An elaborate, broad-based, well-designed research effort should be
initiated to use a sophisticated sediment-research mo:iel, such a*:
one of those described in item 9, *o calculate upland soil loss
and resulting stream sediment for a-se-lected"specified set-of •
existing farming conditions. A field-size area of perhaps 40
acres would be used. Various upland conservation practices such
as use of contours, terraces, grassed waterways, and jinimum til-
lage or combinations of these practices would then be modeled and
their effect on stream sediment would be modeled and evaluated.*
Linking Upstream and Do-jnstrcam Physical Processes
14. In spite of the limitations on existing planning models, it would
be useful to model the upstream-downstream relationship between
land use and water quality by linking a planning model such as
ANSWERS or LANDRUN to a stream-processes model which routes runoff
and sediment (for example, Williams, 1975a, routing model).
Land Resource Areas within each Water Resource Region
15. Research should be carried out to evaluate the effectiveness of
nonincised channels in trapping eroded soil. This one parameter,
the length of nonincised channels in the watershed, holds promise
of explaining the differences between known upland soil losses and
stream sediment. Actual watersheds could be selected where upland •
soil losses could be calculated and where downstream sediment
yield has been measured over a period of years in a reservoir or
by a sediment-gauging station. The length or density of nonincised
channels could be determined by field mapping.
16. Research is needed to develop a much better understanding of the
flow of sediment-laden water through grass. Theoretical calcula-
tions could be made of flow conditions. A study could be made of
the performance and problems of real grassed waterways that have
been functioning for several years. What are the design criteria
and what is the expected performance of a qreenbelt area estab-
lished along a stream? How does the greenbelt affect the movement
of upland soil loss into the strtam?
*lhe Ub"EDA is currently undertaking such a project with the Iowa Field
Evaluation Project.
37
-------
17. Unit stream power should be studied throughout a stream system in
order to evaluate the means through which the stream adjusts its
energy expenditure. This energy system governs transport of sedi-
ment and erosion or deposition in the stream system. Understand-
ing this energy system could reveal differences iri the energy
regimen between stream:, which difference could absolutely govern
the downstream delivery of sediment. For basin-scale watersheds,
say 2590 sq km (1000 sq mi), the USP might provide the linkage
between upstreai and downstream sediment. About 6 to 12 basin-size
stream systems could be evaluated and compared by their USP. This
power would be related to upstream soil loss calculated by the USLE,
and downstream sediment could be measured at a stream gauging
station.
.Linking Models of Land Use, to Models of Physical_ Processes
There are two basic approaches to linking models of land management and
physical processes. The first approach is to incorporate land-management and
physical-processes models into one large mathematical model that determines
land-management patterns and water-quality parameters simultaneously. The
second approach is to keep the land-management and physical-processes models
separate. The land-management model would generate a land-management pattern
and a set of parameters or coefficients that would be input to the physical-
processes model, which would then be run to determine water-quality paramet-
ers. Through an iterative process, the desired '
-------
explored because it is usually cheaper and quicker to develop alternatives
once a model has been built than to develop plans without the aid of models.
Thus, a wider range of possibilities could be explored end more detailed
information on tne economic impacts of the alternatives could be estimated
if costs, prices-, and economic relationships are included. The models could
then be helpful iri finding cost-effective alternatives to improve water quali-
ty. Professional judgment, however, could be applied to eliminate or change
alternatives that appear to be infeasible because of factors not included in
the model. Professionals could use land-management plans developed by mathe-
matical models as aids for the development of actual plans to" be recommended.
A land-management model should predict crop production and agricultural
land management, including crop rotations, conservation practices, and til-
lage practices, fertilizers, pesticides, costs, and revenues. Constraints
on crcp production or land management could be included to achieve confcrm-
ity to existing patterns. Ideally the model should allow for consideration
of structural and any other practices available' to control runoff, chemical
losses, and all forms of erosion including sheet-and-rili , gulley, and
itreamtiank erosion as well as changes in land use. Management of rural,
"onagncultural land such a*- woodlots and unfarmed fields should also be in-
cluded. Such management, nowever, might be less important because, in many
cases, those fields would not contain significant sources of pollution.
O'site damages (or benefits) resulting from (the control of) soil loss and
p ant-nutrient losses should also be estimated. A land-management model
should be able to generate input for the physical model used. It should also
be able to accommodate policies that might be used to control pollution,
po icies such as cost sharing, taxes, and restrictions on pollution loading
anc land-management practices. In addition, problems such as noncompliance
that are 1 ikely to be encountered in implementing a "prograi.i ought to be con-
sidered when one estimates .(hat water quality would result when_ a program
is mplementec. Sensitivity analysis could be used to study the impact of
uncertain parameters of water-quality relations on land-management plans
with both approaches to link ir.g land-management and physical models.
Several general approaches to modeling a land-management component
include: linear-optimization models, nonlinear-optiinizaticn models, and
nonop -.imizing or simulation models.
Lineiii'-Qpi itrizolion Mode Is
Linear programming is traditionally applied to agricultural land-manage-
ment problems, including some applications to water-quality problems. Many
line;." programs are implemented at the voters lied level (Alt, Mi ranowski, and
Heady, 1979; Jacobs and Timmons, 1974; Kasal, 1976; Sharpp and Berkowitz,
1979; nn-i White and Partenheimer, 1979). Others are implemented at the
nationa" or subnational level (Osteon and Soitz, 1978; Taylor and Frohberg,
1977; ana Wade and Heady, 1977). When one considers programs implemented at
the national or large subn- tional levels, he should take into account the
impacts of programs on crrp prices, whereas this provision is not necessary
at the small watershed le °1. At the watershed level, farm or regional in-
come could be maximized, assuming constant prices, subject to constraints
that certain levels of crop production, land management, and levels of or.site
39
-------
pollution loadings (or water quality under certain formulations) be met.
Given the state of the art of modeling physical processes, currently one
could model only very small watersheds when linking models of land manage-
ment and the more sophisticated physical-processes models. Hence, constant
prices for commodities could be assumed. One must keep in mind, however,
that large-scale changes in land management, particularly in crop production,
can have impacts on crop prices and farm income. In addition, planners might
want to study the impacts of different nonmonetary priorities for outputs of
land-management plans. In such a case, goal programming, a form of linear
programming, would be an appropriate technique. Tt woulu require an objec-
tive function with units of production weighted to meet production goals,
thus minimizing weighted deviations from production goals. As a . esult,
production goals would not have to be met (Field, 1973). Goal-programming
algorithms are available. Impacts on farm income cf alternative land-manage-
ment plans could be summed in an accojnting row. Goal programming and
profit- or income-maximization formulations could be built into the same
1 inear.-programming tabl :?au.
The greit advantage of linear programming is that algorithms are widely
available ar.d knowledge of the technique is vjidespread. A major problem,
however, is that the nonadditive pattern effects of land management cannot
be considered (Patterson, 19/2). Even with the same acreages of land manage-
ment practices on a farm or field, differing the spatial allocation could
impact runoff, erosion, or other variables such as wildlife habitat.
Linear programming would not be able to account for such interactions. If
these patterns were known it would be possible to constrain the model to re-
quire these land-management patterns while allowing the model to allocate
other land-mar,ageinent practices.
Honlineav Opt-i miration Models and Simulation
There are two alternative approaches to linear programming, economic,
land-management models: non-linear programming and simulation (non-opti-
mizing models). No applications of these two approaches to water quality
management have been found by the authors. Nonlinear optimization techniques
would be useful where tn?re are interactions between land management practices
or nonlinear water quality relationships could be stated. It is conceivable
that the applications of such models to water quality problems could be
difficult and expensive to solve.
Sirrulaticn Modclc
Simulation models could also be used for economic models of land manage-
ment. It is difficult to generalize about simulation models because they
can take so many different forms. A simple simulation model could develop
budgets, crop p.oduction, and water quality impacts of land management plans.
They have not been applied to nonpoint-source pillution-cor.trol problems to
any great extent. The value of simulation models for nonpoint-source
pollution-control problems will depend upon the quality of the economic re-
lationships and decision rules included in the model. A well-constructed
simulation model might be able to overcome the limitations of the basic ;
assumptions of linear programming and be able to predict landowner responses
40
-------
to various incentive programs more accurately. Simulation models may also
be cheaper to run than linear-programming models, so more land-management
alternatives could be explored. The basic disadvantage is that developing
such a model could be quite expensive. It is, however, an alternative ap-
proach to linear programming for land-management modeling and is worth con-
sidering in the future.
The rlsthod of Linking Land Mcviagement bo Va.tevsh.ed Models
In this section of the report, the method of linking land management
models to watershed models such as empirical methods, LANDRUN, ANSWERS, ARM,
and SEA-AR is investigated. The situations where models of land management
and of physical processes can be incorporated into ore mathematical model
that simultaneously determines land-management plans and water-quality para-
meters will be identified. Situations in which the models of land manage-
ment and physical processes must be run separately will also be identified.
The"sediment-related water-qua!ity parameters that will be considered in'
this section will be the total quantity of sediment moved into the stream
and the duration curve of suspended sediment. The total quantity of sediment
can then be related to damages associated with the silting in of reservoirs
and drainage ditches and the covering of spawning beds of aquatic animals.
The duration curve of suspended sediment can, thus, in the latter case, be
related to reductions in fish populations.
The emphasis of this section will be on linking physical-processes
models to linear-programming land-management models. With linear-program-
ming models, it would be advantageous to include directly into the linear-
programming matrix constraints on sediment delivered or suspended sediment
because one could relate a land-management pattern directly to water-quality
goals rather than going through a trial-and-error process to find a satisfac-
tory plan. The question of whether physical-processes models should be in-
corporated into one large model with a simulation, or nonoptimizing model of
land management is not particularly interesting because it is unlikely that
the information necessary to accomplish the task is available. The simula-
tion model will not develop an "optimal" solution of land management in re-
sponse to water-quality constraints. Hence, a trial-and-error process of
some sort is probably required. Whether a land management simulation model
and a physical-processes model should be run toqether or separately is pri-
marily a matter of convenience.
Only two physical-processes models considered in this report can be
incorporated directly into a linear-programming matrix: the empirical
models developed by Williams (1975b) and by Simons, et at. (1977). When a
runoff model which assumes that runoff is not changed by changes in land
management is used, the hydrograph of each design storm would not change
when the pattern of land management changed. Sediment delivered, however,
would change because the crap-practice factors of the USL-E would change.
Total sediment delivered and the sediment duration curve could be modified
with each land-management pattern without including the hydrologic model
into the linear program.
_The modeling procedure would begin by making a baseline run of the
41
-------
land-management linear-programming model wiln no constraints on sediment to
determine a land-management pattern and a set of economic parameiars for
comparison with later runs of the model. Coefficients for the empirical
model would be derived from the r.aseline land-management pattern. The physi-
cal-processes model would be run fo>~ each.design storm determining total
sediment moved,-the storm hydrograph, ".i.J the storm sediment graph (Williams,
1978). From the "eries of hydrographs and sediment graphs derived for each
storm, a" suspended-sediment duration curve can be defined for the series of
design storms. This procedure requires the assumption that the sediment
moved during a storm is primarily washload transported in a suspended state.
This assumption appears to be legitimate for the Midwestern agricultural
reg ion.
The farm-management model could then be optimized, subject to constraints
on total sediment, to study the impacts on land management and costs. The
"optimal" land management to meeu water-quality goals would be determined by
minimizing'opportunity cost as defined'by'the object i've" :unction: The sedi-
ment duration curve can be modified for each land-management pattern without
including the hydrologic model in the linear-programming matrix. Only the
C and P factors would change Hence, the sediment duration curve would be
rotated around the intercept of the time axis, but the basic shape would not
change. The constraint on total sediment moved could easily be included in
the model with the following inequality:
7: y: A. . X. Y (Eq. 14)
t j i j i j a
where
Y. = goal for tota1 sediment delivered over a certain period of time
to a point in the watershed. (Y can also be an average sediment
yield per unit of time.)
X., -- acreage of land practice i (a combination of rotation and tillage
and conservation practices) on soil j.
A.j = sediment delivered per acre with practice i on soil j.
Y. . C ¦ ¦ . F. .
b 11.1
x . cb. . P. .b
TJ "IJ
C.j = USLE Topping factor for practice l on soil j
P-j = USLE practice factor for practice i on soil j
= cropping factor under baseline conditions For soil j
F'bj -- practice factor under baseline conditions for soil j
Y| = sediment yield under baseline conditions (could also be an
average per unit of time)
4;;
-------
X = acreage of entire watershed above the point in the watershed where
the goal is set
Constraints on total sediment delivered for a number of points in the
waterway could be included in the model, if desired.
Constraints on suspended sediment could also be included in the model.
The goal would be stated such that suspended sediment would not exceed a
given concentration more than a certain percent of the time. The constraint
would affect all area upstream and would take the following form:
E ^ B.. X. . 1 Zan (En. 15)
i J ij iJ '
where:
Z - = suspended sediment goal; suspended sediment would not exceed this
quantity n percent of the time
acre n percent of the time with practice
B — = suspended spJiment per
i on soil type j
^ = *bn • Ci,rPi,i
x rb pk
' i J ' i J
Z. = suspended sediment under baseline conditions that are exceeded
bn r , r , .
n percent of time
Accounting rows with coefficients similar to B-. in Fquation 15 could also
' J
be included to compute the new sediment concentration for a number of differ-
ent duration probabilities. In this way, the suspended-sediment duration
curve for the new land-management pattern could be constructed. Constraints
on suspended sediment could be applied in this same way at different loca-
tions in the waterway. It is likely that only one of the locations would
actually restrict the solution since all locations will be closely related.
There are a number of situations where combining physical-processes
models and linear-programming land-management models into one large linear
program will be difficult, if not impossible. If the hydrologic relation-
ships of the physical-processes model are affected by land management so
that the volume of runoff, the peak flow, or the shape of the hydrograph is
changed, the land-management 1inear-programming model and the physical-pro-
cesses model will generally have to be run separately. Under these circum-
stances, interactions among variables determined in the model will violate
the basic linear-programming assumption of additivity and independence of
variables. The cost of running a physical model arid its size may also make
it difficult to combine a physical-processes model and a land-management
1inear-programming into one large linear-programming model.
When the model assumes that volume of runoff, the peak flow, or the
43
-------
shape of the hydrograph is affected by land management, the runoff and sedi-
mentation models would have to be made a part of the 1inear-programming
matrix. A simple modification of the baseline runoff and sedimentation
conditions using the previously described methods could not be made. Both
sediment and runoff variables would have to be determined in the linear-
programming matrix, which would create problems for LA'IDRUN, ARM, ANSWERS,
the Williams empirical model (when runoff is affected by land management),
and, probably SEA-AR. In all of these cases, the basic linear-programming
assumption of additivity and independence of variables would be violated.
With the Williams empirical method and LANDRUN, the runoff factor would have
to be determined within the 1inear-programming matrix. The runoff factor is
determined by the product of peak flow and total runoff. The product is
input to an exponential function which, in turn, would be multiplied by the
acreage of a given practice. This series of operations violates thx- assump-
tions of linear programming because variables determined within the model
canrot. be multiplied together or taken to the nth power. Hence, LANDRUN
and the Williams empirical method (when runoff is affected by land management)
cannot be incorporated into a linear-programming "matrix.
- Similar problems occur with ARM and ANSWERS. With the ARM model, total
runoff would be determined within the linear-programming matrix and would be
input to an exponential function. Also, sediment transported by overland
flow is multiplied by the fraction of runoff reaching the stream. Both of
these variables would have to be determined within the linear-programming
model. These computations would violate the assumption of additivity of
variables. With the ANSWERS model, volume of flow c-nters a relationship
that determines flow detachment of sediment. In this relationship, the
cropping factor C of the USLE is multiplied by the flow rate. Unless each
land unit in the ANSWERS model is restricted to one cropping practice, the
acreage of each practice applied would be needed to compute a weighted aver-
age of the C factor. Hence, the acreage of each practice and flow, both
determined in the linear-programming matrix, would be multiplied together.
The assumption of additivity and independence of variables would be violated.
If each land unit of the ANSWERS model were restricted to one practice,
integer programming would be necessary for the land-management model. The
specifics of the SEA-AR model have not been released as of this writing. It
seems likely, however, that this model would also violate the assumptions of.
additivity and independence of variables because its hydrology c;id sedimen-
tation relationships are similar to those in ANSWERS.
These five approaches could not be included in a large 1 inear-program-
ming matrix not only because they are theoretically inconsistent with linear
programming uuc also for a number of practical reasons. Both ARM and SEA-AR
are continuous simulation moaels. The physical-processes model would have
to simulate the entire time period for each iteration of the 1 inear-program-
ming model, which would be a very expensive operation because linear pro-
gramming can iterate many times. Running ARM, in particular, is rather
expensive, so running it for every iteration of a linear-programming model
would be impractical. Also, the number of time intervals that would have to
be considered by the ARM model would require so many rows in a linear-pro-
gramming model that the cost would be prohibitive. ANSWERS, LANDRUN, and .
the Williams empirical model when runoff is affected by land management
-------
would require that the hydrology model be run for every design storm for
every iteration of the 1inear-programming model, which would be extremely
expensive.
Because the land-management lHnear program and the more sophisticated
physical-processes models must be run separately, the next problem is to
determine the type of linkage between them. A set of land-management alter-
natives would first be developed by the land-management model. Each land-
management plan would define a set of coefficients for the physical-proces-
ses model, which would then be run .j estimate the pollution loading para-
meters associated with the plan. An important consideration is that the
land-management model be able to provide the coefficients used in the physi-
cal-processes model as a printed output. USLE factors weighted by the area
of practice could be computed for the watershed or a portio.. of the water-
shed in accounting rows and provided as printed output. In fact, any compu-
tation that does violate the assumption of linear programming could be
.performed in accounting rows to.provide coefficients for '¦he physical-pro-
cesses model. All of the physical-processes models considered as planning
models in ~'his report use USLE-type input.
The first step in this iterative procedure would be to estimate current
land management practices. The physical processes model would be run to
estimate the pollution loading parameters associated with the current land
management pattern. The impacts of land management alternatives could be
compared with these baseline conditions. The next step would be to develop
a set of alternatives to the current pattern of land management. Various
constraints corresponding to alternative erosion- and sedimentation-control
policies as well as constraints for other land-management goals would ...
included in the land-marragement model to develop a set of alternatives for
further consideration. Constraints with respect to erosion and sedimentation
would include: (1) restrictions on soil loss per acre or per larger aieas
as computed by the USLE, (2) restrictions on sediment yield as computed by
the USLE modified by a delivery >-atio, or (3) restrictions on sediment yield
on suspended-sediment concentration as computed by the Williams empirical
method when runoff is not affected by land management. The method used to
constrain suspended sediment wo-jld be the method discussed earlier in this
section (Eq. 15). The land-management model could be constrained to differ-
ent levels of sediment yield or concentration to develop different land-
management plans. Other policies such as restrictions on land-management
practices, cost sharing, subsidies, or soil-loss .taxes could also be usc-d to
develop land-management aIternatives. Once this initial set of alternatives
was developed, those with "unacceptable" impacts could be eliminated. Cri-
teria for eliminating alternatives could include: (1) the unacceptabi1ity
of certain policy tools to the public, (2) the lack of available resources
to implement alternatives, and (3) the unacceptable concentration of an im-
pact on a certain group of people.
Once a set of land management alternatives has been developed, land
management coefficients for the physical-processes model would be obtained
from the land-management model. Pollution-loadiny parameters fcr each alter-
native would then be cievelopec. Those results could then be compared to the
pollution goals developed for the watershed. If the results of a particular
45
-------
alternative did not meet the water-quality goals yet seemed to be promising
enough for further consideration, the alternative would have to be modified
and rerun, until it met the water-quality goals. Professional judgment could
be used to concentrate changes in land management to those areas which would
bring about the greatest changes in water quality in order to minimize the
trial-and-error aspects of this process. New constraints could also be
applied to the land-management model to modify alternatives. When a number
of alternative land-management plans had been developed to meet alternative
water-quality goals, the impacts on water quality, land management, and
economics, would be displayed for the interested public to investigate.
This type of iterative approach to defining and choosing alternatives
might not find the global optimum for the system. It might, however, sug-
gest significant improvements over tu,c current situation by allowing a choice
of implementable alternatives. Ultimately, the water-qua!ity goals and a
land-management plan would be chosen after public scrutiny and discussion
between resource-management professionals and the interested public. - This
approach would "be 'orTented toward finding a politically acceptable approach
that would improve water quality rather than defining some global-optimum
solution.
Finally, we will briefly summarize the advantages and disadvantages of
the two approaches to linking a land-management 1inear-prograrmiing model to
a physical-processes model. The main advantage -to being able to incorporate
these two models into one linear-programming matrix is that a land-manage-
ment plan can be developed in direct response to water-quality goals without
going through an extended trial-and-error process. Because of the simplicity
of the physical-processes model required, this approach would probably be
inexpensive. The main disadvantage, however, is tho: the simp] hydrologic
models and the empirical methods of determining sedimentation are considered
to be less realistic than the more sophisticated models, particularly ARM
anJ ANSWERS. The main advantage of running the land-management model and
the physical-processes model separately is that the more sophisticated
physical-processes models could be used. Hence, the predictions of water-
quality parameters associated with a land-management plan might be more
accurate and realistic. The disadvantages of this approach are that a fairly
time-consuming, trial-and-error process might be required. The time involved
and the cost of running the more sophisticated physical-processes model would
make this approach relatively expensive.
The main trade-off between running the land-management and physical-
processes models as one model or separately .s accuracy and realism versus
cost and convenience. Running the models separately may produce more
accurate results but is costly and inconvenient, while running the models
together is less accurate but also cne^p and convenient. The nc-ed for accu-
racy and realism in the ISWQM planning process appears to mandate the ap-
proach of running a physical-processes model and a land-management model
separately. This conclusion, however, is not firm. Both approaches to
linking land-management and physical-processes models really need to be im-
plemented on a number of watersheds and compared to each watershed's actual
behavior. The differences in accuracy and realism of results and the differ-
ences in costs need to be determined¦to better define the trade-offs between
46
-------
the two approaches. A more realistic assessment of the trade-offs is needed
to enable one to make a better choice between the two approaches.
A RECOMMENDED DECISION-MODEL STRUCTURE FOR THE ISWQM APPROACH
* What Information Snouid the Model Provide?
The model of the decision system should provide information to help
make decisions about what water-quality goals to choose and how to manage
resources to pursue those goals. The model of the decision system needs to
provide information useful for making comparisons among alternative courses
of action with respect to such criteria as physical and biological feasi-
bility, economic efficiency, equity, political and cultural acceptability,
•anc!-operational' or administrative practical ity..( CI, aviso",. 19751... A discus-
sion of some of the different types of information a decision model should
provide and some of the assumptions that could be used follows:
1. Water-quality parameters estimated by the model need to be related
to water-quality goals or pollution damages. Total sediment moved
cjn fill reservoirs and channels and, hence, can have impacts on
drainage- water-supply capacity, recreation, commercial navigation,
and fish habitat. Suspended sediment can have impacts on fish pop-
ulations and species. Alternative water-qua! ity goals should be
developed and then considered by the decision model, which will
develop land-management alternatives to me?t eacn set of water-
ouality goals. rhe benefits and costs of achieving each alterna-
tive set of water-quality goals should be estimated and then dis-
played for the consideration of professionals and the interested
public. Planners could thus choose waLer-use and water-qua!ity
goals and a land-management plan while remaining aware of the sug-
gestions and preferences of the public.
2. The model should allow for the consideration of a wide variety of
crop-production anc! pollution-control practices. While the first
priority in model development should be crop production, the model
should also consider other agricultural land uses such as grazing
or forage, forest-crop production (such as trees and nuts) and
wildlife management.. The aqronomic crop-production portion of the
model should oe able to consider crop rotation, tillage practices,
conservation practices, and such off-the-field erosion- and runoff-
control practices as field borders, grassed waterways, and green-
belts along streams. Whether or not practices such as field bor-
ders, grassed waterways, and greenbelts can be considered depends
on the state of development of physical processes models.
3. The model, ideally, should be able to define a number of economic
variables useful to analysing economic efficiency and equity, such
as the- production of crops and other market and nonmarkst outputs,
net income, and the capital and operating costs of pollution-control
practices. The objective of the decision model should be to deter-
mine the opportunity costs and equity implications of pursuing
>5 7
-------
di.'Ferent courses of action. Because of the restrictions on the
size of watersheds of the physical-processes models, the impacts of
changes iii ciop production on crop prices can be ignored. The
equity analysis should consider the income impacts on different
groups of people (such as the owners of different types of land)
and the impacts of changes in the production of nonmarket commodi-
ties on the people who consume them. These efficiency and equity
impacts will affect the cultural acceptibility and operational
practicality of an alternative.
4. The decision model should be able to estimate both total sediment
moved to a point (or points) in a stream and the concentration of
suspended sediment. The modal should also be able to determine the
impacts of nonpoint-source pollution-control practices on these two
parameters by considering their impacts on soil detachment, runoff,
and the delivery of soil to a waterway. ¦ The impacts of incised ana
nonincised channels, concavity of slope, and vegetative strips on
sediment delivery to streams need to be considered in the future
development of physical processes models.
In the Midwest, wash load and bed load move in separate regimes
within streams. Hence, they can be treated separately in modeling
stream processes. Bed load, in the Midwest, appears to be of minor
significance in causing sediment damages, so it can be left out of
the model without causing a large error in calculating sediment
damages. As a result the model can concentrate on wash load, which
is, to a large extent, controlled by upland processes. In the Mid-
west, upland soil particles are fine and move primarily as wash
load. Wash load can be deposited in pools during low flow periods
and scoured out in storms, thus complicating stream processes.
Currently this process of deposition and scouring probably cannot
be modeled in a land-management water-quality model. In addition,
sedimentation and runoff processes should be related to the move-
ment of other pollutants such as nutrients and pesticides.
The Recommended Model Structure
Early in any ISWQM planning study, alternative water-use and water-
quality goals, alternative approaches to managing nonpoint-source pollution,
criteria to be addressed in a benefit-cost analysis, and criteria for elim-
inating alternatives at later stages in the study need to be defined. With '
an actual planning study, public opinion and professional judgment need to
be integrated in making these determinations. Kith a research study con-
ducted to test methods or examine issues, however, public involvement is not
necessary. Quantitative models may be useful in addressing water-use and
water-quality goals. Quantitative methods are not applicable for choosing
criteria or determining alternative policy approaches for consideration.
Estimating demand functions for all potential uses of a stream would be
a useful way to address water-use and water-quality goals from an economic
point of view. Estimating demand functions is not necessary because defining
and choosing among alternative goals are ultimately political. Rather, a
48
-------
series of alternative goals could be developed for consideration by inte-
grating public opinion and professional judgment without using demand func-
tions. Demand functions would be useful because they could estimate the
number of different water uses demanded under different situations and the
water-use benefits received (or damages incurred) under alternative water-
quality goals. The advantages of demand functions must be compared with the
costs of estimation when deciding whether to use them. The use of demand
functions would seem to be most appropriate on waterways where there are
complex trade-offs among uses and difficult choices to make. Under such
situations, good information about the implications of choices would be very
desirable.
If the decision is made to develop demand functions for uses, different
approaches may be required for different uses. Estimating demand functions
for transportation uses,'recreation'occurring on-easily definable sites, and-
some consumptive uses of water, such as municipal or industrial uses, will
be easier than estimating demand Function for other uses. The necessary
data for these uses can be obtained without much trouble. It will be more
difficult and expensive, however, to develop demand functions for recreation
activities that are not confined to a specific site. Fishing, for example, -
can occur throughout a watershed, making it difficult to define the popula-
tion to be sampled. In such situations, demand could be estimated over a
large area. The quantity of demand could then be allocated to individual
waterways, recognizing priorities of uses and the suitability of waterways
for uses. The value determined for the use over the large area could be used
to estimate benefits and damages on the individual waterway. It will proba-
bly be oifficult to estimate demand functions for such intangible values as
esthetics because of the difficulties involved in measuring changes in
esthetics and relating them to monetary values. Hence, it may be more rea-
sonable to indicate a goal to provide a more or less esthetically pleasing
experience under an alternative or indicate that an alternative will provide
a more or less esthetically pleasing experience as compared to some other
alternative.
Once alternative sets of water-use goals are determined for considera-
tion, water-quality criteria must be defined. Water-quality criteria car. be
used to determine water-quality standards or to help measure benefits and
costs associated with an alternative.' Water-quality criteria for a number of
water uses have been defined and could be used in ISWQM planning studies.
The list of water-quality criteria is incomplete, however. For example,
criteria for suspended-sediment concencration have not been defined for most
uses. Criteria often have not been developed for different species of fish.
The latter are important because changes in water quality can change total
fish production and species composition, thus affecting the use of the stream.
More research is needed concerning the effects of water-quality parameters
on uses to expand the list of criteria. This research could be expensive
and could take a long time. In the meantime, the existing list of water-
quality criteria should be used whenever possible. It may be necessary to
use professional judgment, some of which might not be supported with pub-
lished research, to modify existing criteria, to suggest new criteria, or to
identify factors.limiting the improvement of a stream for different uses.
r:-iuon thd <;ubioctive nature of this approach to defining criteria, such
49
-------
criteria should be subjected to professional debate so that their strengths
and weakr,esses will be more fully understood. Particularly controversial
criteria could be removed or changed.
The approach to linking models of land management and physical processes
is to run them separately so that the more sophisticated, physical-processes
models can be used. The recommended land-management model consists of a farm-
management 1 inear program fcr a small watershed. (A simulation model could
be a reasonable substitute.) The model should estimate crop production,
crop rotation, tillage and conservation practices, soil loss, and any other
land-management and pollution-loading parameters needed for the physical-
processes model used. The model should also estimate profits for the water-
shed and different classes of land to estimate the opDortunity cost and dis-
tributional impacts of different parameters. The objective function for the
.model .should be the maximum of.farm income.with crop prices, held constant
because of the small area being considered. Constraints oh land management
and crop production could be applied to the model to simulate current condi-
ti ons.
The la:,d-management model should be constructed so that it can handle
the policy approaches develoDed in the planning study. Such policy approach-
es could include restrictions on soil loss, agricultural chemical applica-
tions, or land-management practices Other policy approaches include ^axes
or subsidies on soil loss, agricultural chemical application, or l.ind-manage-
aient practices. The model could include relationships to determine total
sediment moved and the sediment grapn at a point in the stream using the
Williams empirical method assuming that runoff does not change with land
management. These relationships could be used to provide the first-cut
approximation of how land should be managed to meet goals for sediment con-
trol. This approach would help to reduce the number of runs in the itera-
tive process associated with running physical-processes and land-management
models separately. Once the agricultural portion of the model has been con-
structed, management of rural land not in agronomic crops could be modeled.
Sine? SWCDs will probably play an important role in nonpoint-sourcc pollution
control, the interrelationships between nonpoint-source pollution control
and other nc.icrop production-related objectives such as timber and fiber pro-
duction and wildlife habitat should be studied.
ANSWERS, ARM, and the SEA-AR model would all be good choices for the
physical-processes model. They are all fairly sophisticated but have reason-
able information requirements. If the planning study is to be primarily
concerned with sediment problems, ANSWERS would probably be the best choice
because of its ti-atment of sediment, processes. Also, because it is a dis-
tributed model and is based upon a grid system to account for land variations,
it is better suited for looking at the impacts of changing land-management
patterns than the other point models. ANSWERS also handles larger areas
than the other models, but it is still a small watershed model. The ARM and
SEA-AR models are bettor choices if ^he planning study is to be concerned
with agricultural chemicals as well as sediment problems. ARM and SEA-AR
contain relationships to predict concentrations of agricultural chemicals
while ANSWERS does not. The ANSWERS model would have to be modified to
include relationships to predict the behavior of agricultural chemicals
50
-------
before it would be useful in studies where the emphasis would be on nutrient
and/or pesticide problems.
Once a modeling system has been developed, a land-management alterna-
tive must be evaluated for each alternative set of goals. When the physical-
processes and land-management models are run separately, land-management pat-
terns cannot be developed by constraining thr? modeling system with water-
quality goals. Hence, an iterative process would be needed to develop a
land-management pattern. This iterative process might not find the global-
optimum solution to maximize farm income subject to constraints. It may be
a useful method, however, to develop a set of alternatives for further con-
sideration.
The iterative process of evaluating alternatives begins by running the
"land-management model nth no environmental constraints and then running the
physical-processes model to estimate water quality under baseline conditions.
Plans to control nonpoint-source pollution are developed by constraining the
land-management model or adjusting the model coefficients in accordance with
an alternative water quality goal. Such changes in the land-management model
could include financial incentives for certain practices or restrictions on
soil loss, fertilizer application, or sediment concentration as computed by .
the Williams empirical method when assuming that runoff does not change with
land management. The land-management model is then run subject to the con-
straints or changes in coefficients. The results of the land-management
model define the parameters needed to run the physical-processes model,
which estimates water-quality parameters.
If the estimated water-quality parameters should diverge from the water-
quality goals by more than some predetermined amount, the land-management
plan would have to be changed and the model rerun. If the estimated water-
quality parameters should excoed water-quality goals, the land-management
model could be changed by increasing financial incentives, making the environ-
mental restrictions more restrictive, or constraining the model according to
some external change in the land-management pattern. If the estimated water-
quality parameters were less than the water-quality goals, the land-manage-
ment model could be changed to relax financial incentives or environmental
restrictions. Both the land-management and physical-processes models would
be rerun to determine a new land-management pattern and estimate water-
quality parameters. This iterative process would be continued until water
quality goals were met. This process could also be used to develop a series
of more and more restrictive alternatives of the same policy approach.
As stated earlier, criteria to reject alternatives could be developed.
If, in the process of developing an alternative, it appeared that the alter-
native should be rejected, the process of developing it could be stopped to
save time and money. Criteria for rejection of an alternative could include'
an excessive opportunity cost or undesirable distributional impact.
Once the development of alternative goals and land-management plans was
completed, a benefit-cost analysis of each alternative would be needed so
that the strengths and weaknesses of all alternatives could be compared. . .
Criteria for comparison could include economic efficiency, equity, political
51
-------
and cultural acceptability, and operational practicality (Clawson, 1975).
One alternative set of goals and a land-management pattern should be chosen
for implementation. Both professional judgment and public opinion should be
considered when a choice of alternative is made. The alternative finally
chosen could be one of those designated for consideration or could be a new
one developed in light of the analysis of alternatives.
CONCLUSION: HOW FEASIBLE IS ISWQM AS COMPARED TO SOURCE CONTROL?
The source-control approach to water quality management is currently
feasible, an example being the soi1-conservation program. Economic and
physical models are readily available for use under this approach. Insti-
tutions to implement such an approach on agricultural lands already exist.
Some changes in existing institutions, specifically SWCDs, would likely be
desirable: (1) A role in planning land management should be provided for
people living outside of the "boundaries of an SWCD who have an interest in
the impact of the management of land within the SWCD on water quality. (2)
Nonlandowners should be allowed to vote in elections concerning land-use
regulations. (3) Urban residents should be brought into the planning pro-
cess. These changes in institutions, however, could require changes in state
soil and conservation laws, a process that could be time-consuming and could
encounter political resistance.
ISWQM encounters more proteins than source control because the costs
and technical problems of implementation and pol:tical resistance to imple-
mentation would; be greater. It would not, however, be impossible to imple-
ment ISWQM.
Institutional changes are needed for ISWQM in addition to those outlin-
ed for source control. A closer integration of land management and water-
use and water-quality goals is needed for ISWQM than for source control.
Agencies managing waterways, defining water uses, defining water-quality
goals, and developing land-management plans need to work more closely toge-
ther in the ISWQM planning process than under source control. It would be
desirable, but not necessary, to clarify .he roles of agencies involved in
the water-resources planning and management process to eliminate overlapping
responsibilities. In addition, the plans of local governments located
within the same watershed need to be coordinated so that their goals and
land-mar.'n^ment plans do net conflict. As part of the ISWQM planning pro-
cess, there should be a feedback mechanism to evaluate the progress of the
plan in terms of meeting goals and to periodically reassess Che goals and
plans.
The creation of nonlocal coordinating agencies, such as 208 planning
agencies, would provide the changes needed in institutions and the planning
process for ISWQM. These agencies will require some legal powers to coordin-
ate the activities of SWCDs. Such agencies could make the implementation
of ISWQM more difficult than source control, which already has in place in-
stitutions to implement the program. There would probably be political re-
sistance to giving a coordinating agency a role-in local decision making.
The more power proposed for the coordinating agency to develop and implement
-plans, the greater the political resistance would probably be. Another
52
-------
problem would be the cost of funding the coordinating agency. The imple-
mentation of ISWQM would thus more difficult than source control because
of these institutional problems.
There are a number of technical difficulties for developing land and •
water-resource plans under ISWQM. While defining water use goals is a poli-
tical process and demand functions would not have to be estimated, demand
functions would be useful in estimating the quantities of uses desired and
the Denofits (damages foregone) associated with different management alter-
natives. 11" demand functions are estimated, the necessary costs and time
involved in estimation would have to be incurred.
Lists of water-auality criteria have already been developed and could
be used to define wrter-quality goals for uses. The lists are incomplete,
however.... Criteria have not been developed for sediment and other pollu-"
tants. Seme criteria refer ;.nly tc more general uses when more 'specific '
breakdowns mignt be desirable. For example, a criterion might be developed
for wildlife and fish propagation, but a criterion for a specific fish
species or equatic ecosystem might be desirable but unavailable. Research
to estimate values for these criteria would be time-consuming and expensive.
Waiting for more complete lists of water-quality criteria to be developed
wou^d preclude implementation of ISWQM in the foreseeable future. The alter-
native would be to use professional judgment tc modify the values of exist-
ing criteria, suggest values for criteria for uses and pollutants for which
no values av,e currently defined, or to identify factors limiting thermovi-
sion of a use without defining a r.u.nerical value for criteria. This judg- -
ment could serve to identify the most important water-quality problems in
meeting water-use goals. Such an identification would reduce the wasteful
allocation of resources to control nonpoint-source Dollucants that are not
limiting factors to providing desired water uses. Values for.criteria that
are developed using professional judanent could evolve into an accepted
standard through further research or experience that indicates how changes
in water-quality parameters affoct uses.
There are no particular problems for ..he implementation of ISWQM in
the area of land-management modeling. Land-.nanaqement models have been
developed '-^at could be npplicci to a ISWQM study, but more realistic models
would be desirable. There is also a need to develoD a better understanding
of the impacts of agricultural land management on noncrop production values.
Modeling the land-management/woter-qiM iity relationships is die biggest
technical problem in applying the ISWQM. A number of planning models of
physical processes can estimate the movement of water, sc^ls, and agricul-
tural chemicals' from a farmer's field to a strain. They can be used to es-
timate total ^ediment delivered, sediment concentration, and agriculcural
chemical concent!ations in the water. All of the physical-processes models
have strengths and weaknesses. The more sophisticated models require an
iterative procedure to develop land-management plans when they are linked
to a land-managemmt model. Stream processes models that can be linked to
these physical-processes models are in early stages of development. In
particular the behavicr of washload and its impact on suspended sediment ii
poorly understood. All of the more sophisticated models are restricted to
53
-------
small watersheds. For larger watersheds, the general approach has been to
determine1 water-quality problems and sources of the problem to concentrate
on. As an alternative, small watersheds within a largo watershed'could be
sampled for development of land-management i^lans. Guidelines determined in
these lond-management plans could be applied to other parts of the large
watershed having similar characteristics. As a result of these technical
problems, applying ISWQM to a large watershed could be costly and difficult;
These significant technical problems for the application of ISWQM are
the important areas for further research. More work needs to be directed
toward studying the impacts of resource management on natural processes af-
fecting water quality and toward the development of models of these process-
es. The study and modeling of the impact of stream processes on water qual-
ity is also desirable. Model development of stream and physical processes
should f'"rst concentrate" on small watersheds and then oe extended to larger
watersheds. Research is needed to estimate the water-quality needs for
various uses to aid in defining water-quality goals and in estimating bene-
fits and damages. Finally, research should be directed to estimating the
damages and benefits associated with water-quality goals.
Technical difficulties and analytical costs will be greater for an
ISWQM approach where a land management plan is developed to meet a set of
precisely defined water quality standards than for a simpler ISWQM approach
of identifying pollutant priorities and critical source areas. The more
sophisticated ISWQM approach will require more costly techniques and, in
some cases, more costly data collection. It may also be difficult to find
techniques to closely relate land management to water quality parameters.
In summary, ISWQM is a feasible approach to agricultural nonpoint-source
pollution control. ISWQM will be more expensive and difficult to apply than
source control. ISWQM has the capability, however, to define goals and pro-
blems more accurately and plan resource management to control problems more
efficiently than source control. If ISWQM is to be applied, the greater
analytical and administrative expense and difficulty of applying ISWQM as
compared to source control will have to be justified by the more efficient
allocation of resources to alleviate water-quality problems than would occur
under source control. However, because of relatively high expenses and
difficulties of application, a sophisticated ISWQM approach of developing
a land management plan to meet a precisely defined set of water quality goals
seems most appropriate for waterways with complex management problems or
critical values to protect. Other waterways could be managed with a simple
ISWQM approach of defining priorities for pollutants and critical source
areas cr with a source control approach. In some cases, a source control
approach might solve water quality problems.
54
-------
REFERENCES CITED
Alchian, A. A., and W. R. Allen. 1S67. University economics. 2nd ed.
- Belmont, CA: Wadsworth, Inc.
Alt, K. F.; J. A. Miranowski; and E. 0. Heady. 1979. Soci?, costs and
effectiveness of alternative nonpoint pollution control practices. In
Best management practices for agriculture and oilvicultus ?¦, ed. R. C.
.Loehr, D.A. Haith, M. F. Walter, an'! C. S. Martin. Ann Arbor, MI:
Ann Arbor Science Publishers, Inc.
American Society of Civil Engineers. 1975 Sedimentation enqineerina. New
York, N. V.
Bosselmart, F., and D. Callies. 1971. The quiet revolution in land use
control. Washington, D.C.: Council on Environmental Quality.
Bowles, S., and H. Gintis. 19/8. The invisible fisx: Have capitalism and
democracy reached a parting of the ways? In Proceedings of the 90th
annual meeting of the American Economic Association, 28-30 December,
1977, at New York.
Buck, D. H. 1956a. Effects of turbidity on fish ar.d fishing. Report no.
56, Norman, OK: Oklahoma Fisheries Research Laboratory.
1956a. Effects of turbidity on fish and fishing. In Trans-
actions of the 21st North furerican Wildlife Conference, ed. J. B. Tre-
fether, pp. 249-61. Washington, D.C.: Wildlife Management Institute.
Cairns, J. R. 1968. Suspended solids standards for the protection of
aquatic organisms. In Proceedings of the 13th annual Purdue Industrial
Waste Conference. Engineering extension series No. 129, pp. 12-27.
Wect Lafayette, IN: Purdue University.
California Department of Water Resources. 1970. Sediment yield and land
treatment, Eel and Mad River Basins, Appendix No. 1, in cooperation with
USDA Soil Conservation Service and Forest Service. Sacramento, CA.
Claffey, F. J. 1955. The productivity of Oklahoma waters with special
reference to the relationships between turbidities and soil, light pene-
tration, and the population of plankton. Ph.D. dissertation. Stillwater,
OK: Oklahoma Agricultural and Mechanical College, Graduate School.
Clawson, M. 1975. Forests for whom and for what? Baltimore, MD: Johns
Hopkins University Press.
55
-------
Donigian, A. S., Jr., and N. H. Crav-iford. 1976. 'Modeling peat 'aides and
nutrients on agricultural lands. EPA-600/2-/6-043. Athens, GA: U.S.
Environmental Protection Agency.
Donigian, A. S., Jr.; D. C. Beyerlein, H. H. Davis, Jr.; arid N. H. Crawford.
197/. Agricultural resea.reh m.anagement, AH'-!, model varc-ion II: refine-
ment and testing. EPA-600/3-77-098. Athens, GA: U.S. Environmental Pro
tection Agency.
Dwyer, J. F.; J. R. Kelly; and M. D. Bowes. 1^77. Iroroved procedures for
valuation of vh.e conirib'ition of recreation to national economic develop-
ment. Research report no. 128 (UILU-WRC-77-0128). Urbana, IL: Univer-
sity of Illinois, Water Resources Center.
Einstein, H. A. 1950. The bed load function for sediment transport in open
channel flo-js. Technical bulletin 1026. Washington. O.C.: U.S. Soil
Conservation Service.
Field, D. B. 1973. Goal programming for forest management. Forest Science
19:125-36.
Foster, G. P.., and L, D. Meyer. 1972. Transport of soil particles by shal-
low flow. Tran.sac tions ottie Society of Agricultural Engineers. 15(1):
99-100.
Foster, G. R., L. D. Meyer, and C. A. Onstad. 1977. An erosion equation
derived from basic erosion principles. In Transactions of the American
Society of Agricultural Er.gineers 20(4):678-82.
Gammon, J. R. 1 970. The effect of inorganic sediment on strcc*-' biota.
Water Pollution Control Research Series 18050 DWC. Washington, D.C.:
U.S. Government Printing Office, U.S. Environmental Protection Agency.
Groszyk, W. S. 1979. flonpoint source p^llut^on control strategy. In
Best management practices for agriculture and silviculture, ed. R. C.
loehr, D. A. Haith, M. F. Walte1", and C. S. Martin. Ann Arbor, MI:
Ann Arbor Science Publishers, Inc.
Hoimstra, W. N.; D. K. Damkut; and N. G. Benson. 1969. Some effects of
silt turbidity cm behavior of juvenile largemouth bass a-'.d green sun fish.
Technical paper no. 20. Washington, D.C.: U.S. Dept. of the Interior,
Bureaj of Sport Fisheries and Wildlife.
Henderson, J. M. , and R. E. Quandt. 1950. I'.icroeconomic theory. New York:
McGraw-Hill.
Holmes, Beatrice H. 1979. Institutional oases j or coiitrol oj n.onooirit
source pollution under tne Clean '¦¦¦'crier Act--'..rith emphasis on agricultural
nonpoint sources. U.S. Environmental Protection Agency, Office of Waste
Management. WII-554, V.'as'ningtori, O.C.
-------
Holton, N. H. 1961. "A concept for infiltration estimates in watershed
engineering," USDA, Agricultural Research Service, ARS-41-51. October.
Institute for Environmental Studies. 1977. Feasibility of a systematic
approach to water quality management in Illinois. IIEQ-20.083, I1EQ
document No. 77/35. Urbana, IL: University of Illinois.
Iowa Law Review. 1977. A procedural framework for implementing nonpoint
source water pollution control in Iowa. Iowa Law Review 63:184-229.
Jacobs, J. B., anc! Timmons, J. F. 1974. An economic analysis oF agricul-
tural land use practices to control water quality. Am. J. Aq. Econ.
56:791-98.
Johns, E. L. 1976. Sediment problems in the Mohave Valley—a case history.
In Proceedings of the Zrd Federal Inter-agency Sedimentation Conference,
Washington, D.C.: Water Resources Council. ~
Karr, J., and Schlosser, I. 1977. Impact of near-stream vegetation and
stream rnorplwlogy or. water quality and stream biota. EPA-500/3-77-097.
Athens, 6A: U.S. Environmental Protection Agency, Environmental Research
Laboratory.
Kasal , J. 1976. Trade-offs between fawn income and selected environmental
indicators. Technical bulletin No. 1550. Washington, D.C.: U.S. Depart-
ment of Agriculture, Economic Research Service.
Krcusz, N. G. P. 1972. Intergovernmental ar^rangerents for water use regula-
tion in Illinois. Bulletin 741. Urbana, IL: University of Illinois,
College of Agriculture, Agricultural Experiment Station.
Lake, J. 1977. Envirorsnertal impact of land use on water quality. Final
Report on the Black Creek Project. EPA-905/9-77-007-B. Chicago, IL:
Environmental Protection Agency.
Li, R. M.; K. G. Eggert; and D. B. Simons. 1979. Simulation of stormwater
runoff and sediment yield for assessing the impact of silvicultural
practices. In Best management practices for agriculture and silviculture,
ed. P.. C. Loehr, D. A. Haith, M. F. Walter, and C. S. Martin. Ann Arbor,
Mich.: Ann Arbor Science Publishers, Inc.
Li, R. M.; R. K. Simon:; and L. Y. Shiao. 1977. Mathematical modeling of
on-site soil erosion. In Proceedings International Symposium on Urban
Hydrology, Hydraulics, and Sediment Control. Lexington, KY: University
of Kentucky.
Li, R. M.; D. B. Simons; and M. B. Stevens. 1975. Nonlinear kinematic wave
approximation for water routing. Water resources research 11(2):245-52.
Mackenthun, K. M. 1969. The practice of water pollution bi-ology. Washing-
- ton, D. C.: Government Printing Office, U.S. Dept. of the Interior,
Federal Water Pollution Control Administration.
57
-------
Meyer, L. D., and W. H. Wischmeier. 1969. Mathematical simulation of the
processes of soil erosion by water. Transactions of the American Society
of Agricultural Engineers 12(6):754-58-
Mulkey, L. A., and J. W. Falco. 1977. Sedimentation and erosion control
implications for water quality management. In Proceedings of the National
Symposium on Soil Erosion and Sedimentation by Water. St. Joseph, MI:
"American Society of Agricultural Engineers.
National Association of Conservation Districts. 1978. Selecting priority
areas for implementing nonpoint source control measures. Honpoint notes
on 208 implementation, Note No. 17. Washington, D.C.
Neiblin^, H., and G. R. Foster. 1977. Estimating deposition and sedi-
ment yield from overland flow processes. In Proceedings International
Symposium on Urban Hydrology, Hydraulics, and Sediment Control, pp. 75-86.
Lexington, KY: University of Kentucky.
Novotny, V., and M. A. Chin. 1976. Nonpoint overland pollution transport
- model. Iri IM bRUll ue^rc manual. Milwaukee, WI: Marquette University,
Department of Civil Engineering.
Onstad, C. A., R. F. Piest, K. E. Saxton. 1976. Watershed erosion model
validation for southwest Iowa. In Proceedings 3rd Federal Interagency
Sedimentation Conference, pp. 1-34. Washington, D.C.: i.'ater Resources
Council.
Oschwald, W. R. 1972. Sediment water interactions. J. Environ, quality
1:360-65.
Osteen, C, and W. D. Seitz. 1973. Regional economic impacts of policies to
control erosion and sedimentation in Illinois and other corn-belt states.
Am. J. kg. Econ. 60(3):510—17.
Patterson, R. L. 1972. Application of integer linear programming to prob-
lems of land use allocation. MichU-Su-72-213. Ann Arbor, HI: Univer-
sity of Michigan, Sea Grant Program.
Rickert, D. A., and G. L. Beach. 1978. Assessing impacts of land management
activities on erosion-related nonpoint source problems. J. Water Poll.
Con. Fed. 50(11):2439-45.
Robinson, A. R., and K. G. Renard. 1977. ARS research on sediment transport
yield, and properties. Research report no. 256- Tucson, AZ: Southwest
Rangeland Watershed Research Center.
Samuels, W. 1972. Costs arid power. Unpublished paper. East Lansing, M.I:
Michigan State University, Department of Economics.
Sharpe, B. M. H.,.and S. J. Berkowitz. 1979. Economic, institutional, and
water quality considerations in the analysis of sediment control alterna-
tives: A case study. In Best management practices, Loehr, R.C. ,
-------
[). Haith, M. r. Walter, and C. S. Martin. Ann Arbor: MI: Ann Amor
icience Publishers, Inc.
Simons, D. S3. , A. .J. Reese, R. M. Li, and T. J. Ward. 1977. A simple method
for estimating sediment yield in soil erosion: Production and control.
In ?r:>rradinnv of a t'cbional Conferance on Soil F/rocion. Ankeny, I A:
- Soil Conservation Society.
Smith, P. E. 1976. Held test of a distributed watershed erosion and sedi-
mentation model. In Soil mosion: Prediction arxl control, pp. 201-209.
Ankeny, IA: Soil Conservation Society.
Soil Conservation Service. 1977. Cropland croci^:.. Washington, Q.C.: U.S.
Department of Agriculture.
Sparks,' R. 1978. Effect.; of oadirnenCov. aquatic life.' Pa,:er prepared for
the subcommittee on Soil Erosion and Sedimentation of the Illinois Task
Force on Agricultural fionpeint Sources of Pollution. Havana, II: Illi-
nois Natural History Survey, River Research Laboratory.
Vinyard, G. L., and VI. 0. O'Brien. 197G. Effects of light and turbidity on
the reactive distance of bl u eg ill {leponis macrochirus). J. Fioh.r.ries
tljij&arch Board of Cctiada. 33:2845-49.
Wade, J. G., and E. 0. Heady. 1977. Conttolling nonpoint sediment sources
with cropland management: A national assessment, f.rn. J. Ac;. Fcon.
59:13-35.
Wallen, E. I. 1951. ?i:e direct effect of turbidity or-, fiishac. Series 2.
No. 43, Bulletin Mo - 48. Stillwater, OK: Oklahoma Agricultural and
Mechanical College.
White, G. fS., and Partenheimer. E. J. 1979. The economic implication of
erosion and sedimentation control plans for selected Pennsylvania dairy
farms. In 0. A. Haith, M. F. Walter, and C. S. Martin, n^nt Umiapcr.cr.t
r-rocviccc, Loenr, P. C. , D. A. Haith, M. F. Walter, C. S. Martin (cds.)
Ann Arbo": Ann Arbor Science Publishers, Inc.
Williams, J. P.. 1975a. Sediment routing for agricultural watershed. l-Jaicr
rcvow-coa tmilctir.. 1 1 (5): 965-74 .
Williams, J. R. 1975b. Sediment-yield production with universal equation
using run-off energy factor. In Pro'j^.diuq:; of the fJcdvr.ent iic'l.d Work-
shop, Prevent aid vr<:opeciivn tachnoloo-y for predicting z&lhr.cr.i. yields
and ccurcpa, ARS-S-40, pp. 244-52. Oxford, MS: USDA Sedimentation
Laboratory.
Williams, J. R. 19.78. A sediment graph mod~l based on an instantaneous
unit sediment graph. Water rozcurcco rcr.earch. (14(4}:659-64.
Wischmeier, W. H., C.. B. Johnson, and B. V. Cross. 1971. A soil credibility
nomograph for farmland and construction sites. Journal of soil and water
conservation 26(5):189-93.
r 9
-------
Woolhiser, D. ft., and K. G. Renard. 1978. Stochastic aspects of sediment
yield. In Verification of physical, models in hydraulic er.gi>iecvi>ig. hew
York, N.Y.: American Society of Civil Engineers.
Workman, J. P., and J. E. Keith, 1975. Economics of soil treatments in the
upper Colorado. In Watershed management, pp. 591-96. New York, N.Y.:
• American Society of Civil Engineers.
Yalin, Y. S. 1963. An expression for bed load transportation. Jourrsjl of
the Hudraulics Division of American Dociaty of Civil Engineers. 89(HY3):
221-250.
Yang, C. T. 1971a. On river meanders. .Journal of hydrology 13(3):231-53.
Yang, C. T. 1971b. Formation of riffles and pools. Water resources
research 7(6): 1567-7/l.
Yang, .C.-.I^.,-and-.iL_.Bw-S-ta.il 1974. Unit stream power—for sed-Lv.ent tyccns-- -
port in natural rivers. Research report No. 88 (UILU-WRC-74-0QS8),
Lirbana, IL: University of Illinois, Water Resources Center.
Young, R. A., and C. K. Mutch!er. 1969. Effect of slope shape on erosion
and runoff. In Transactions of the American Society of Agricultur'al
Engineers !2(2):231-33.
60
-------
------- |