I EVALUATING ECONOMIC IMPACTS
OF PROGRAMS FOR CONTROL
OF SALINE IRRIGATION RETURN FLOWS
A CASE STUDY
OF THE GRAND VALLEY ,COLORADO
JUNE 1976

-------
(LA
U.S. EPA Region 8 Library
80C-L
999 18th St., Suite 500
Denver, CO 8020P-2466
Bhi/lL^AT/ZJC.
(EVALUATING ECONOMIC IMPACTS OF PROGRAMS FOR CONTROL
OF SALINE IRRIGATION RETURN FLOWS:
A CASE STUDY OF THE GRAND VALLEY, COLORADO
By
Kenneth L. Leathers and Robert A. Young
Department of Economics
Colorado State University
June 30, 1976
Project 68-01-2660
Project Officer
Mr. George Collins, Economist
Region VIII
Environmental Protection Agency
Denver, Colorado 80203


-------
ABSTRACT
Economic impacts of alternative on-farm water management programs
for controlling saline irrigation return flows are estimated. The study
focuses on the Grand Valley in west central Colorado, an area thought to
be representative of irrigation return flow problems in the Upper Colorado
River Basin. Direct economic impacts, in terms of increased cost or re-
duced incomes, are estimated with linear programming models of representa-
tive farm situations. A regional interindustry model was developed to
trace indirect economic impacts on related economic sectors in the three-
county local trade area. Since the hydro!ogic-geologic relationships
which govern salt pickup in the study reach are not entirely understood,
a correlation and regression analysis of water quality and quantity data
was performed. This analysis attempted to distinguish salt contributions
of natural origin from those due to irrigation, and further, to separate
irrigation contributions into components attributable to on-farm irri-
gation practices as compared to water distribution system losses.
It was found that, given plausible assumptions about the proportion
of irrigation water diversions which percolates into the saline aquifer
as drainage water, adjustments in on-farm water management practices
are a relatively inexpensive method for reducing salt loading. Other
nonstructural alternatives (such as land retirement) and structural
approaches (such as lining of canals to reduce seepage) appear to be
ii

-------
iii
considerably more expensive. Downstream benefits (salinity damages
avoided) are not judged to be large enough to warrant (on strict eco-
nomic efficiency grounds) undertaking any of the proposals other than
adjustment in on-farm water management.
The statistical analysis of sources of salt pickup lead to the
conclusion that natural forces account for considerably more salinity
in the study reach than has been heretofore assumed by the public agencies
responsible for salinity control. Further, there is some support for
the hypothesis that, of the irrigation-related salt loading, leakage
from the delivery system is the most significant source. This result,
if confirmed by subsequent investigation, has considerable significance
for salinity abatement policy in the Upper Colorado Basin.

-------
TABLE OF CONTENTS
ABSTRACT	ii
TABLE OF CONTENTS	iv
LIST OF ILLUSTRATIONS	vii
LIST OF TABLES	ix
ACKNOWLEDGEMENTS	xi
CONCLUSIONS	xii
RECOMMENDATIONS	xiv
SECTIONS
I INTRODUCTION
A.	Background	1
Salinity in the Colorado River Basin	1
Potential Means of Alleviation	5
B.	Objective and Plan of Study	6
C.	Order of Presentation	7
II THE GRAND VALLEY AS A CASE STUDY
A.	Overview	8
B.	Irrigated Agriculture	10
C.	Return Flow Salinity	11
Salt Pickup Mechanisms	12
Proposed Salinity Controls	17
D.	Justification	19
III RESEARCH PROCEDURES: CONCEPTS, METHODS AND ASSUMPTIONS
A.	Introduction	20
B.	Hypothesized Means of Reducing Salinity	21
On-Farm Control Options	21
Land Retirement	23
iv

-------
V
C.	Conceptual Framework for Estimating Program Costs	23
Representative Farms and Linear Programming	24
Interindustry Analysis	25
D.	Economic Benefits of Reduced Salt Discharge	26
IV THE ECONOMICS OF ON-FARM SALINITY CONTROL
A.	Introduction	29
B.	Simulating Farmer Response	30
Data Source and Sampling Method	30
The Programming Models and Assumptions	31
Modifications in Irrigation Practices	38
C.	Irrigation-Salinity Relationships	41
The Soil-Water-Crop Budget	41
Rate of Salt Pickup in Return Flows	45
D.	Results and Summary	54
Estimated Costs for Control of Deep Percolation	54
Estimated Costs of On-Farm Seepage Control	57
Summary and Implications	58
V THE ECONOMICS OF LAND RETIREMENT AS A MEANS OF CONTROLLING
SALINE IRRIGATION RETURN FLOWS
A.	Introduction	62
B.	Approach and Assumptions	64
Compulsory vs. Noncompulsory Programs	64
Partial vs. Complete Retirement	65
Valuation of Land and Water Resources	66
Analysis of Direct and Indirect Impacts	68
Accounting Stance	71
C.	Results and Discussion	72
Local Direct and Indirect Costs	72
Implications for the State of Colorado	76
D.	Summary of Cost-Effectiveness	77
VI RETURN FLOW SALINITY MECHANISMS: A REAPPRAISAL OF
ESTIMATES AND ASSUMPTIONS
A.	Introduction	81
B.	Review of Previous Research	82
Agricultural Pickup Mechanisms	83
Natural Pickup Mechanisms	87
Implications for Policy Decisions	91

-------
vi
C.	Methods of Analysis and Data Selection	93
The Techniques of Correlation and Regression	94
Data Sources and Evaluation	95
Stream Flow and Salinity	95
Natural Runoff	99
Irrigation Return Flows	101
The Empirical Hydro-Salinity Equations	108
D.	Results and Discussion	110
Part I: The Annual Data	110
Part II: The Monthly Data	128
Part III: The Quarterly Analysis	139
E.	Summary and Implications	144
Recommendations for Further Research	149
VII APPENDIX
Literature Cited
Supporting Information
150
155

-------
LIST OF ILLUSTRATIONS
FIGURE	PAGE
1.	The Colorado River Basin 		9
2.	Cross-Section of Geologic Formations in the Grand Valley ...	13
3.	The Grand Valley of Colorado	15
4.	Frequency Distribution of Sample Farm Size and Model
Acreage Parameters 	 32
5.	Harvested Acreages of Selected Crops Grown on Bureau of
Reclamation Project Lands (Garfield Gravity Division)
in the Grand Valley, 1948-1974 	 34
6.	Acreages of Selected Crops as a Percentage of Total Cropped
Acres for Model Farms	35
7.	The Grand Valley Trade Area	69
8.	Grand Valley Reach of the Colorado River 	 84
9.	Annual Estimates of Salt Pickup, Irrigation Water
Deliveries to Farms, Precipitation and Net Discharge
for the Grand Valley Reach, 1951-1972 	 Ill
10.	Annual Estimates of Irrigation Diversions, Distribution
System Waste Water, Canal and Lateral Seepage Losses, and
Crop Evapotranspiration for the Government Highline Canal,
1951-1972 	 112
11.	Data Plots of Net Salt Pickup With Precipitation and
Net Discharge: Annual Data 	 116
12.	Data Plots of Net Salt Pickup With Irrigation Diversions
and Farm Deliveries: Annual Data	117
13.	Data Plots of Net Salt Pickup With Canal Seepage and
Deep Percolation: Annual Data	118
14.	Monthly Mean Estimates of Net Salt Pickup, Precipitation
and Delivered Irrigation Water for the Government Highline
Canal, 1951-1972 	 129
15.	Monthly Mean Discharge and Salt Load of the Gunnison River,
and Precipitation at Three Locations in the Grand Valley
Reach, 1951-1972 	 130
Vit

-------
viii
List of Illustrations, continued
FIGURE	PAGE
16. Monthly Means of Seasonal On-Farm Deliveries, Crop
Consumptive Use, Field Tail Water, Deep Percolation,
Seepage, and Spillage for the Government Highline
Cana, 1951-1972 	 132

-------
LIST OF TABLES
TABLE	PAGE
1. Estimates of Additional Labor (Inconvenience) Cost
to Farmers for Shifting from Present to More Efficient
Irrigation Practices 	 40
2. Traditional Irrigation Practices and Soil Moisture
Relationships for Selected Crops in the Grand Valley 	 43
3.	Water Budget Summary for Traditional Irrigation
Practices in the Grand Valley: Annual Water Use and
Losses for Selected Crops 	 44
4.	Comparison of Total Water Budgets for Irrigation in
The Grand Valley: Aggregate Use and Losses 	 46
5.	Average Composition of Colorado River Water Diverted for
Irrigation and Predicted Ionic Equilibrium for Various
Leaching Fractions 	 48
6.	Ionic Analyses of Drainage Maters for Selected Drain
Outlets in the Grand Valley: Averaged Concentrations,
January and February, 1975 	 50
7.	Ionic Analyses of Ground Waters at Two Locations in
the Gravel-Cobble Aquifer and at One Location North
of the Aquifer: Averaged Concentrations 	 51
8.	Average Salt Concentrations for Sources of Surface and
Subsurface Irrigation Return Flows and Weighted Estimates
of Average Rates of Pickup 	 53
9.	Estimated Annual Direct Income, Crop Substitution and
Salinity Effects of Limiting Irrigation Water Delivered
to Farms	 55
10.	Estimated Annual Costs for Reducing On-Fartn Conveyance
System Seepage with Aluminum Pipe in the Grand Valley .... 59
11.	Community Income Reduction Under Two Options for
Retiring Irrigated Land in the Grand Valley Trade Area .... 73
12.	Summary of Annual Regional Costs and Cost Effectiveness:
Upper and Lower Bound Estimates for the Land Retirement
Options	 75
13.	Estimated Incremental Costs of Salt Removal for Proposed
Alternative Control Measures, Grand Valley, Colorado
(1975 dollars)	 78
ix

-------
X
List of Tables, continued
TABLE	PAGE
14.	Mean Salt Loads, Stream Flows, and TDS Concentrations
in the Study Reach: A Comparison of Three Time Intervals . . 98
15.	Spring and Fall Precipitation and Runoff at Badger Wash,
1951-1972 	 100
16.	Means, Standard Deviations and Coefficients of Variation
of the Regression Variables: Annual Data, 1951-1972 	 114
17.	Correlation Matrix of Selected Variables Used in
Regression: Annual Data 	 120
18.	Summary of Regression Results Relating Net Salt Pickup
to Natural Causes: Annual Data 	 122
19.	Summary of Regression Results Relating Net Salt Pickup
to Selected Measures of Irrigation Variables: Annual Data . . 125
20.	Summary of Regressions Results Combining the Natural and
Manmade Causes: Annual Data 	 127
21.	Coefficients of Variation for Selected Regression Variables:
Monthly Data	133
22.	Correlation Coefficients Between the Regression Variables
and Net Salt Pickup: Monthly Data	135
23.	Correlation Coefficients Relating the Influence of Time-
Lags in Subsurface Irrigation Return Flows on Salt Pickup:
Monthly Data	 136
24.	Summary of the Monthly Regression Analyses: Regression
Coefficients and Tests of Significance 	 138
25.	Means, Standard Deviations and Coefficients of Variation
for the Quarterly Data	140
26.	Summary of the Quarterly Correlation-Regression Results
and Tests of Significance 	 142
27. Empirical Estimates of Rates of Pickup and the Magnitude of
Total Salt Contributions from Natural and Manmade Sources . . 147

-------
ACKNOWLEDGMENTS
We are grateful to Dr. William Franklin, Department of Agronomy
(Colorado State University), William McClennagan, Bureau of Reclamation
(Grand Junction), and Dr. Wynn Walker, Department of Agricultural En-
gineering, for their assistance in modeling agronomic and physical rela-
tionships; to William Klapwyk, manager of the Grand Valley Water Users
Association, for his advice and assistance in data collection; to Charles
Palmer, U.S. Forest Service (Denver), for his help with the input-output
model; and to Roxanna Leathers, graduate student, Department of Economics
(Colorado State University), for her assistance with the statistical
analysis.
We also wish to acknowledge a number of our colleagues and pro-
fessional associates at Colorado State University who have raised ques-
tions and provided useful criticism along the way: Drs. Gordon Kruse
and Sterling Olsen (Agricultural Research Service, USDA), Dr. Norman
Evans (Environmental Resources Center), Dr. Raymond Anderson (Economic
Research Service, USDA), Professor Henry P. Caul field, Department of
Political Science, and Drs. Lee Gray, Anthony Prato, Lawrence Mack,
and Paul Huszar of the Department of Economics.
The authors would also like to thank Mrs. Denese Gekas, Mrs. Electa
Cameron, and Miss Elizabeth Young, for typing the drafts of this report.
Finally, the advice and patience of George Collins, Project Officer,
have been greatly appreciated.
xi

-------
CONCLUSIONS
Several policy instruments can be identified by which public agencies
might regulate saline irrigation return flows. These include direct
effluent standards, taxes on effluent, public subsidy of private sector
abatement activities, marketable effluent rights and direct public in-
vestment in abatement facilities. In the study area, salinity is pro-
duced by several hundred independent commercial farms. In such a case,
the monitoring and enforcement costs of direct approaches appear likely
to be exceedingly large. The analysis reported here assumes that the
policy instruments would be indirect ones, which regulate use of re-
sources used in the production process from which pollution emanates
(i.e., water and land).
On the basis of certain plausible assumptions with respect to the
proportion of applied irrigation water reaching the saline aquifer, a
physical-economic model was developed to measure costs of reducing saline
irrigation return flows. It was found that substantial reductions in
on-farm drainage and in associated salt pickup could be achieved at
relatively low cost (less than $3.00 per ton of salt removed).
Salt pickup associated with on-farm water conveyance systems is
somewhat more costly to avoid. Aluminum pipe to replace on-farm laterals
and head ditches can achieve salt reduction at about $8.00 per ton.
A more drastic approach would be permanent withdrawal of water sup-
plies and the associated lands from crop production. Allowing for in-
direct as well as direct effects leads to an estimate of between $20
xii

-------
xi i i
and $26 per ton of salt removed. (The preceding is a pure economic
efficiency measure, and does not include the considerable social and
political disruptions which would be the consequence of removing a sub-
stantial portion of the economics base of the region.) The estimated
cost of the proposed structural measures being planned under the Colo-
rade River Salinity Control Program is similar in magnitude, in excess
of $20 per ton of salt removed.
Although conceptual disagreements regarding measures of downstream
salinity abatement benefits are not resolved, the writers judge that
such benefits are not large enough to warrant undertaking of any of the
proposals other than adjustments in on-farm water management. Even this
step is questionable, since the analysis presented here omits considera-
tion of administrative and enforcement costs. Further, the burden of
such programs would fall almost entirely on irrigation water users,
although the efficacy of the programs is not fully proven.
A statistical analysis was performed to help clarify the origins
of salt loading in the study area. Correlation and regression techniques
were applied to annual, quarterly and monthly observations on salt pick-
up and natural phenomena (river flows, rainfall) and irrigation-related
variables (diversions, estimated seepage, etc.). The analysis suggests
that irrigation water diversions are considerably less important as a
source of salinity than has been previously assumed by the federal agen-
cies responsible for water quality monitoring and control. If the stat-
istical analysis is confirmed by further investigations, rather large
increases in the estimated cost per unit of salt removed (including the
estimates reported here) are implied.

-------
RECOMMENDATIONS
1.	The salinity control programs authorized in PL 93-320, title
II, for the Grand Valley should be postponed until uncertain-
ties as to amount and source of irrigation return flow salinity
are more clearly resolved.
2.	Further analysis to determine the source and magnitude of salt
pickup in the Grand Valley and elsewhere in the Upper Colorado
Basin should be undertaken at once. Such investigations should
include a more detailed and rigorous analysis of water quality
and quantity data, a review and interpretation of completed and
ongoing research by geologists, hydrologists and soil chemists
and continued experimental programs.
xiv

-------
Section I
INTRODUCTION
A. BACKGROUND
Managers of irrigated land in nearly all arid and semi-arid areas
of the world face problems in coping with salinity (dissolved solids) in
irrigation water [Moore, 1972]. Some dissolved solids are contained in
river waters from natural leaching of rocks and soils or from inflows of
saline water from underground sources. Examples of such contribution
include return flows from irrigation, urban or industrial withdrawals
which have picked up salts in addition to concentrating the salt load
already present, and concentration due to evaporation from storage
reservoirs.
Detrimental impacts are typically felt by water users from increasing
concentrations of dissolved solids. Such detriments include decreased
productivity and/or increased production costs for both agricultural and
industrial water users, and in household uses, lower palatability of
drinking water, reduced life of water pipes and appliances, and at higher
concentrations of some elements, adverse health effects.
Salinity in the Colorado River Basin.- The Colorado River Basin in
the southwestern United States presents a well-known example of the effects
of salinity on economic activity. The modification of the river's flows
has yielded benefits in the form of flood control, irrigation, electric
power, recreation and municipal and industrial water supply. However,
1

-------
2
the River picks up a relatively high salt load from natural sources, and
man's activities have served to further add to the natural salt load, as
well as to concentrate existing salts. The Lower Basin states, especially
California and Arizona, are the major withdrawers of Colorado River water
and are experiencing increasing difficulty with rising salinity. The
tail waters from the Colorado flow into Mexico. Farmers in that nation
also use the Colorado for irrigation and the Mexican government has been
quite critical of the quality of waters released from the United States
[Oyarzabal, 1976].
The Colorado River passes through seven states on the way from its
headwaters in the Colorado and Wyoming Rockies to its mouth in the Gulf
of California (in Mexico). The states of the Colorado River Basin (CRB)
include Wyoming, Utah and Colorado in the Upper Basin and Arizona, New
Mexico, Nevada, and California in the Lower Basin. The River's salinity
content ranges from an average of 50 parts per million (ppm) at its source
to over 800 parts per million at Imperial Dam, Arizona, the last major
U.S. diversion point before the water reaches Mexico. The salt load pass-
ing Lee Ferry, Arizona, the boundary between the Upper and Lower Basins,
between 1941 and 1966 averaged 8.2 million tons per year. As of 1970,
the annual salt load at Lee Ferry was around 8.5 million tons.
Sources.- The principal dissolved constituents in Colorado River
water are the cations calcium, magnesium and sodium and the anions sul-
fate, chloride and bicarbonate. The presence of these, and of small a-
mounts of other dissolved constituents, are commonly referred to as
salinity. Increasing concentrations of these dissolved mineral salts

-------
3
threaten to become a major economic problem for users of Colorado River
water.
Salinity increases result from two processes: salt loading and salt
concentrating. Salt loading increases the amount of salt for a given
amount of water and salt concentrating decreases the amount of water
for a given amount of salt. Salt loading is the addition to the river
system of mineral salts through natural and man-made sources. Salt con-
centrating is the rise in salinity through stream-flow depletions which
concentrate the salt burden in the river system into a lesser volume of
water.
Natural salinity sources contributing to salt loading may be further
subdivided into point sources, such as springs and wells, and diffuse
sources, such as high-salinity streams, percolation through highly saline
soils, and surface runoff from precipitation. Man-made sources include
return flows from irrigated lands and waste discharges from municipal
and industrial sources.
Effects.- The amount of salt in water directly influences the utility
of the water. Thus salinity levels greatly affect the types and amount
of water use. The extent of this relationship will become apparent after
an examination of the water uses in the Colorado River Basin.
Water uses are of two general types--withdrawal and instream uses.
Withdrawal uses involve removal of the water from the river system and
include irrigation, stock watering and municipal and industrial uses.
Instream (in situ) uses include power generation, water recreation, navi-
gation, groundwater recharge, fish and wildlife habitat, silt control,
and water quality control.

-------
4
Increases in salinity tend to raise costs of industrial water users
by increasing treatment, conditioning, and/or operating and maintenance
costs. For example, salinity can cause excessive water hardness and
speed corrosion of recirculating systems, resulting in higher costs for
industrial use. Too much salinity affects domestic users by affecting
water taste, hampering gardening, raising water softening costs, and
speeding pipeline corrosion. In addition, salinity shortens fabric life
and increases the consumption of detergents and soaps by domestic users.
Salinity may also affect the type and quantity of fish and other aquatic
life and thus adversely affects recreational values.
Increasing salinity in irrigation water imposes additional costs
on irrigators. The amount of extra water needed for leaching out salts
in the root zone rises with increased salinity of the applied irrigation
water. When extra water is not available, the producer is left with the
choice of suffering a drop in yield, retiring some of his land in produc-
tion to gain the extra water, or changing to more salt-tolerant, but
typically, less profitable crops. More expensive production practices
may also be required to assist in germination or to drain off excess
waters.
Salinity concentrations in the Upper Basin are not expected to ad-
versely affect agricultural crops in that Region in the foreseeable future,
except in poor drainage areas, such as those along the river in Grand
Valley. In the Lower Basin salinity concentrations are now approaching
critical levels because the types of crop grown in the Lower Region are
generally those which are more susceptible to salinity damage. The U.S.

-------
5
Environmental Protection Agency has recently calculated potential damage
from salinity in the Colorado River Drainage Basin. Total penalty costs,
defined as total marginal costs of increases in salinity concentrations
above 1960 base conditions, are the sum of direct penalty costs incurred
by water users and indirect penalty costs imposed on the economy of the
Region. For the Lower Basin and Southern California Service Area, total
penalty costs were projected to reach $25.4 million annually (in 1960
dollars) by 2010. Irrigated agriculture would account for $21.3 million
of this total. The Southern California Service Area would incur over
75 percent of the total $25.4 million penalty costs. (For an explanation
of how these estimates were calculated, see U.S. Environmental Protec-
tion Agency, "The Mineral Quality Problem in the Colorado River Basin,"
Appendices B and C [1971].)
Potential Means of Alleviation
There are numerous technically feasible salinity management pro-
grams for the Colorado River Basin [U.S. Bureau of Reclamation, 19723.
However, legal and institutional constraints and economic feasibility
tend to place limits on the practicality of many programs. Salinity
control measures resolve into two basic types: measures to augment flow
and measures to reduce salt input. Possible augmentation methods in-
clude: (1) importation of high-quality water, (2) weather modification
to increase precipitation of low-saline waters, (3) utilization of water
produced from geothermal energy resources, and (4) conservation of water
by reducing non-beneficial evapotranspiration through: phreatophyte
control, surface evaporation control, increasing efficiency to farm use,

-------
6
decreasing irrigated acreages, and control of future development. Salt
load reduction may take place by: (1) impoundment and evaporation of
major salt-contributing springs, wells, and other point sources; (2) con-
trolling salinity from irrigated lands by elimination of high salt-pro-
ducing areas from crop production, decreasing quantities of saline return
flows, avoiding contact between discharges and heavy salt-producing areas,
exporting or evaporating high salinity return flows, reducing groundwater
pumpage of high salinity water, or reducing seepage losses from conveyance
facilities; (3) discharging high-salinity municipal and industrial wastes
into evaporation ponds; and (4) constructing and operating desalinization
plants where feasible.
Examples of salinity management and control practices in an agricul-
tural area would be: irrigation scheduling or regulation of irrigation
frequencies, amounts, and methods; land preparation to promote efficient
surface irrigation practices; effective drainage practices; lining of
canals and conveyance systems; and perhaps the most drastic solution,
retirement of land from agricultural production.
The multiplicity of sources of salinity, the variety of potential
means for their abatement and the fact that the receptors of damage from
salinity pollution are located in different political jurisdictions than
the principal sources of pollution, implies that region-wide management
practices must be adopted.
B. OBJECTIVE AND PLAN OF STUDY
The primary objective of this study is to develop and apply a pro-
cedure for analyzing the economic consequences of implementing alternative

-------
7
technologies and institutional arrangements designed to control salinity
from irrigated agriculture. The Grand Valley, located on the Colorado
River near Grand Junction, Colorado, is selected as the case study area.
Specific objectives are:
1.	To identify alternative structural and nonstructural control
mechanisms which can influence stream salinity levels associated
with irrigated agriculture.
2.	To estimate and evaluate the direct costs of technically
feasible measures designed for on-farm salinity control.
3.	To estimate the direct and indirect economic impacts of
selected salinity control policies on the local economy.
A second major objective, which was undertaken late in the study,
is to statistically test competing hypotheses regarding the magnitudes
of salt pickup from irrigation as compared with natural causes in the
study area.
C. ORDER OF PRESENTATION
The balance of the report is organized into six sections. In section
II a brief description of the Grand Valley and justification for using
this area as a case study is given. We outline our theoretical frame-
work and analytical procedures in section III. The empirical analyses
are presented more or less in the order that they were performed: on-
farm water management (section IV), land retirement (section V), and
statistical analysis of salinity mechanisms (section VI). The appendix
contains the sources of literature consulted and supporting information.

-------
SECTION II
THE GRAND VALLEY AS A CASE STUDY
A. OVERVIEW
The Grand Valley is located in west central Colorado at the con-
fluence of the Gunnison and Colorado rivers in Mesa County (Figure 1).
Paralleling the Colorado River for about 30 miles, the Valley averages
seven miles in width and about 4400 feet in elevation. Summer weather
is characteristically hot and dry and the winters cold. Beginning in
April, the normal frost-free season averages about 190 days. With an
annual precipitation averaging 8 to 10 inches, irrigation is necessary
to maintain a viable commercial agriculture in the Valley.
Grand Junction, with a population of about 25,000, is the principal
commercial center in the Valley, and in Colorado's western slope. Ag-
riculture is an important source of employment and income to a local
population of about 60,000 people in Mesa County. However, in recent
years basic manufacturing and service industries have become the main-
stay for an otherwise traditional agricultural community. Approximately
56,000 acres of land are presently cultivated out of a total arable area
exceeding 100,000 acres. Urban and industrial expansion, service roads
and farmsteads, and idle and abandoned lands account for most of the
balance not farmed [Walker and Skogerboe, 1971].
The diversified agricultural industry in the Valley is comprised
of both livestock and crop production activities. Major crops grown
8

-------
9
^¦Grand Volley
COLORADO
Figure 1. The Colorado River Basin

-------
10
include corn, alfalfa, sugar beets, small grains and permanent pasture.
Slightly less than ten percent of the irrigated acreage is planted to
pome and deciduous orchards. Small acreages of vegetables and other
specialty crops such as turf grass are also grown in the area. The Grand
Valley has long been favored wintering area for cattle and sheep grazed
on high summer ranges to the east and south.
B. IRRIGATED AGRICULTURE
Following settlement in the early 1880's, irrigation companies were
organized to divert water for agricultural use. Many of the original
companies have since been consolidated, leaving five which presently
supply all the water diverted under original decrees: The Grand Valley
Irrigation Company (1882), the Grand Valley Water Users Association,
using water developed by the U.S. Bureau of Reclamation in 1916, the
Palisade and Mesa County Irrigation Districts (1890*s), and the Redlands
Water and Power Company [Skogerboe and Walker, 1972]. Because irrigable
acreages were typically overestimated within the newly-formed irrigation
districts, and due partially to a gradual decline in irrigated acreage
as a result of waterlogging, and more recently urbanization, Grand Valley
farmers have always had an abundant supply of water.
Early evaluations of irrigation efficiency, which were initiated in
response to immediate drainage problems in the lower-lying lands, docu-
mented Valley-wide efficiencies of 30 to 40 percent [U.S.D.A. and Colorado
Agricultural Experiment Station, 1957, and Decker, 1951]. However, the
threat of rising water tables and salinity problems encountered on water-
logged soils was not enough incentiye to offset wasteful use of very
low-cost project water. Average charges in 1975 were less than two dollars

-------
11
per acre foot. Average river diversions sometimes exceed 600,000 acre
feet annually, but only 175,000 acre feet are required to meet normal
crop consumptive use [Skogerboe et aK, 1974].
Soils vary throughout the Valley in surface textures ranging from
loam to fine silty clay but share a common parent material derived from
Mancose shale [Soil Conservation Service, 1955]. Being low in organic
matter, these soils are prone to nutrient leaching (especially nitrates)
and have restricted internal drainage at lower elevations. The prevailing
topographical slope of the Grand Valley ranges between 50 and 80 feet
per mile, which effectively limits irrigation methods to furrow and corru-
gation techniques [Bishop et al_., 1967].
C. RETURN FLOW SALINITY
U.S. Bureau of Reclamation [1974] analyses of river waters entering
and leaving a reach of the river which includes the Grand Valley indicate
that salt pickup on the average amounts to approximately 600,000 tons
per year. This reach, determined by the location of U.S. Geological
Survey sampling stations at Cameo, Colorado and Cisco, Utah, on the Colo-
rado River, and at Grand Junction, Colorado on the Gunnison River, in-
cludes a surface area of about 3,000 square miles of which less than
four percent are irrigated lands. Nonetheless, it has been concluded
that the overwhelming proportion of salt pickup (93 percent) in this reach
results from irrigation activities [Environmental Protection Agency,
1971].
Salt pickup as related to agricultural activity is defined as an
increase in the total salt load of water leaving an area after being
diverted for irrigation compared to the salt load of the water entering

-------
12
the area. The present analysis is based on an estimate of 460,000 tons
(per average year) of salt pickup attributable to irrigated agriculture
or over 8 tons per irrigated acre. An additional 140,000 tons is thought
to be contributed by natural sources such as surface runoff from sur-
rounding desert in the study reach.
Salt Pickup Mechanisms
Several physical explanations have been proposed to account for the
above pickup, but agreement concerning the relative importance of any
specific pickup mechanisms is lacking at the present time. Three basic
types of pickup mechanisms considered in this report are: (1) irriga-
tion waters which dissolve and transport salts fron subsurface sources,
(2)	irrigation waters which dissolve and transport surface salts, and
(3)	salts added to streams or water courses by surface runoff of precipi-
tation and snowmelt.
The salt ion composition of most streams and rivers is closely re-
lated to the geology of the watershed and the makeup of watercourses in
areas undisturbed by man. Diversion and use by man can impose marked
changes in water quality from that found under undisturbed conditions.
An understanding of the geology in the Grand Valley and surrounding area
is especially important with regard to salinity control and implementation
of specific control measures for irrigated agriculture.
Geological formations on the southern edge of the Grand Valley, the
Morrison, Summervi11e, Entrada, Kayenta, and Chinle, in descending order,
overlie granite, gneiss, and amphibolite at the Uncompahgre uplift (Figure
2). On the east and to the north of the Valley later geologic formations

-------
Figure 2. Cross-Section of Geologic Formations in the Grand Valley

-------
14
are exposed. In descending stratigraphic order, these formations are
basaltic lava, Green Riyer Wasatch, Plateau Valley, Mesaverde, and the
marine-deposited Mancos shale overlying the formations named above
[Knobel et al_., 1955]. The Mancos and Wasatch formations typically con-
tain large quantities of fossil salts. Mancos Shale is the dominant
formation in the Grand Valley itself, underlying 80 to 90 percent of
the area.
The Valley soils were formed residually from the Mancos Shale and
from alluvial materials deposited by the Colorado and Gunnison Rivers.
The alluvium covers thick beds of porous gravelly and cobbly sand (the
ancient river course) that extend from east of Grand Junction at the head
of the Valley to Fruita. The northern edge of the aquifer, depicted in
Figure 3, roughly parallels the Grand Valley Canal and underlies nearly
half of the Valley's irrigated acreage. This gravel-cobble layer con-
ducts water rapidly and a high water table occurs in the general area
contiguous to the River.
The primary cause of return flow salinity is thought to be the ex-
tremely saline aquifers (as high as 10,000 ppm) overlying the marine-
deposited Mancos Shale formation. Lenses of salt contained in the shale
are dissolved by water entering and coming into chemical equilibrium
with this formation before returning to the river channel. Seepage from
unlined canals and laterals flowing into and through the gypsiferous and
saline alluvium and Mancos Shale substrata has been cited as the primary
pickup mechanism. On the basis of canal seepage measurements, Kruse
et al_. [1975] reported that canal and lateral seepage alone could possibly

-------
LOCATION OF WELLS
A.	East of Grand Junction
B.	Fruita Area
C.	North of the Acquifer
Figure 3
Gvt. High line Canal
Grand Valley Canal
y-
x


Irrigated Area
Area Overlying the
Cobble Acquifer
Grand '.v
: a

• ci





Clifton.

The Grand Valley of Colorado

-------
16
account for total annual salt pickup. Over-irrigation and excessive
deep percolation losses through saline and gypsiferous substrata have
also been cited as a major salt pickup mechanism in conjunction with
seepage [Skogerboe et_. al_., 1974a and 1974b]. Skogerboe and others estimate
that about 45 percent of the total annual pickup is a result of deep
percolating water due to inefficient irrigation practices.
Another pickup mechanism possibly involved is the dissolution of salt
on soil surfaces in surface runoff water. Runoff containing suspended
solids from salt-bearing soils could result in salt pickup if erosive
furrow or rill irrigation took place. Since the majority of the land is
irrigated by rill or furrow methods in the Grand Valley, it is possible
that this pickup mechanism could be important. Although few studies of
this phenomenon have been undertaken, the available evidence suggests that
the quality of water leaving irrigated fields is about the same as the river
water at the point of diversion [Robinson and Franklin, 1973]. Initial
increases in ion concentrations found in the first few hours of irrigation
during the early part of the season were not considered to present a sig-
nificant change in overall water quality.
A third salt pickup mechanism, not related to agriculture but diffi-
cult to separate from irrigation effects is the dissolved salt contained
in natural runoff from rainfall and snowmelt. Since unmeasured subsurface
drainage water and surface runoff water from precipitation both flow into
the natural drainage ways and washes, it is difficult to assess the im-
portance of salt pickup resulting from precipitation in and around the
irrigated area.

-------
17
Proposed Control Measures
Until the initiation of the present study, research concentrated on
various structural control technologies including lining of conveyance
systems and on-farm drainage improvements. Although a program of sci-
entific irrigation scheduling designed to improve on-farm water manage-
ment has been under study since 1972, feasibility analyses have been
limited to a few selected farms with no detailed Valley-wide evaluations
being attempted. Other nonstructural control possibilities have had
little serious consideration.
A number of procedures for reducing salt loading in the Grand Valley
are under consideration by the U.S. Bureau of Reclamation [1976], the
federal agency responsible for implementing the Basin-wide salinity control
program [Public Law, 93-320, Title II]. One plan calls for the lining
of all main canals and laterals to prevent seepage. The published esti-
mates indicated an investment requirement of about $90 million to line
715 miles of canals and laterals. This proposal is expected to reduce
salt pickup by about 200,000 tons per year.
Several lengths of canals and laterals have been lined since 1970
in a demonstration area on the east side of the Valley [Skogerboe and
Walker, 1972]. These researchers estimate that 70 to 80 percent of total
seepage losses could be prevented by lining all canals and laterals (in-
cluding on-farm delivery ditches). However, the costs of such a program
appear to be quite high. Amortizing the investment over thirty years
at six percent yields an annual cost of $6.54 million, equivalent to
$32.72 per ton of salt removed.

-------
18
Inefficiency in on-farm water use, stemming from a combination of
abundant supply, low user costs, and problematic soil-topographic charac-
teristics, has encouraged interest in irrigation scheduling as a Basin-
wide possibility for salinity control. The results of irrigation schedul-
ing are presently inconclusive, but the program holds much promise as a
low-cost control measure [Skogerboe and Walker, 1973; and Anderson et al.,
1974]. Improved on-farm efficiencies may possibly have local benefits,
including increased yields and additional productivity on previously
waterlogged soils, in addition to reduction in downstream damages.
Research on the use of drainage technologies has emphasized inter-
ception of deep percolating water below the root zone before it has reached
chemical equilibrium in the saline aquifers. Because the deep, open
ditch-type drains in use since the early 19201s are largely ineffective
for this purpose, a drainage program would also have to include extensive
renovation of existing structures to be effective. Costs of field drain-
age and renovation also appear to be quite high, and resulting water
quality improvement uncertain. Additionally, drainage improvements with-
out improving on-farm water use efficiency would possibly make matters
worse than they are [Skogerboe et^ aK, 1974b].
Another option, currently under study by Walker [1976], involves
desalting highly saline aquifer and return flow waters which would be
intercepted before entering the river. Although expensive, this pro-
cedure is perhaps the only sure way (other than land retirement) that
significant amounts of salts can be removed from drainage waters.

-------
19
D. JUSTIFICATION
The Grand Valley is an interesting and challenging area to physical
and social scientists alike for investigating salinity abatement measures
for future implementation in the Valley and elsewhere. Because Mancos
shale is so widely distributed in the upper drainage of the Colorado
River Basin, the likelihood of other irrigated localities confronting
similar water quality situations suffices to justify extensive research
in the Grand Valley for immediate applications in other areas. Constrain-
ing factors have been encountered at all levels of study: in modeling
physical and hydrosalinity relationships, institutional inflexibilities
governing private water use and ownership, and identification of socio-
economic consequences for communities directly and indirectly affected.
Degraded irrigation return flows, by way of seepage and deep perco-
lation through saline soils and underlying geologic formations which re-
turn increased salt loads to the river system, make the Grand Valley one
of the most significant man-made sources in the entire river basin.
However, estimates of the specific mechanisms and sources of salt
pickup remain to be established. The various academic and federal agency
teams studying the problem have proposed conflicting estimates of the
proportion of salt pickup in the study area due to natural vs. man-made
causes, and within the man-made category, the proportions iue to seepage
from respectively, (a) the main canal system, (b) the on-farm distribution
systems, and (c) the fields themselves. Since the appropriate control
program may differ among these sources, it is important that these rela-
tionships be well understood. The estimates used in our analysis repre-
sent a compromise among the competing hypotheses and might not be endorsed
by any of the research groups.

-------
SECTION III
RESEARCH PROCEDURES: CONCEPTS, METHODS AND ASSUMPTIONS
A. INTRODUCTION
A complete economic analysis of a proposed pollution control policy
will attempt to measure (a) the benefits of pollution control (in terms
of willingness of receptors to pay to avoid the damages), (b) costs of
reducing or removing waste discharge, (c) the costs of monitoring and
enforcement of regulations, and (d) where relevant, indirect or secondary
costs and benefits. A feasible policy is one which the incremental
benefits exceed the incremental costs [Kneese and Schultze, 1975].
Further, the implications of proposed cost sharing or pricing arrange-
ments should be established.
Several policy instruments by which public agencies might regulate
salinity discharge can be identified. These include direct effluent
standards, taxes on effluent, public subsidy of private sector abatement
activities, marketable effluent rights, direct public investment in
abatement facilities, and indirect controls on use of resources comple-
mentary to the production process from which pollution emanates (i.e.,
water and land) [Peskin and Seskin, 1975].
Where salinity is produced by several hundred independent firms,
as is the case in the Grand Valley, the monitoring and enforcement
costs of most of the direct approaches (standards, taxes and effluent
permits) appear likely to become exceedingly large. Other analyses have
reported on costs of direct structural control methods [Skogerboe,
20

-------
21
et al_., 1974; and Utah State University 1975]. Hence the approach fol-
lowed in the studies reported here emphasizes indirect policy instruments
which regulate water or land use in irrigated crop production.
B. HYPOTHESIZED MEANS OF REDUCING SALINITY
Control of water supply seems to be an obvious approach to reducing
salinity in irrigation return flows. The prevailing water pricing pro-
cedures imply a negligible marginal price to the user and there is little
concern for limiting diversions to the amount specified by the water
right. Farmers now divert much in excess of even the most generous
estimates of evapotranspiration, leaching and carriage requirements.
These excess diversions lead, in turn, to potentially large deep perco-
lation and seepage losses. Hence, it is hypothesized that on-farm
sources of salt pickup could be reduced by constraining water supplies
thus encouraging farmers to adopt more efficient irrigation practices.
On-Farm Control Options
Three hypothesized practices which influence deep percolation are
examined in this study. The first has to do with modifying traditional
irrigation practices. Irrigators may, given factors such as size of
field, slope and row spacing, vary the rate of water applied per unit
area in the crop season by changing (a) the length of time water is
allowed to run in each furrow, (b) the size or number of siphon tubes
(so that the rate of application is changed), and/or (c) the interval
between irrigations. Previous research by agricultural engineers has
revealed that soil infiltration rates in the study area change dramati-
cally as the irrigation season progresses [Gilley, 1968; Skogerboe, 1973].
Early in the year, when soils are loose and open following pre-plant
tillage operations, they permit rapid infiltration of water. During this

-------
22
time, relatively large amounts of water can percolate below the root
zone if excess water is applied. This may occur when irrigation water
is allowed to run at a slow rate for long periods to achieve adequate
wetting in the lower end of fields. Later in the year, the soil tightens
up and the potential for deep percolation is reduced.
It has been proposed that a relatively simple adjustment of irri-
gation practices, namely, increasing the rate of discharge but reducing
the amount of time that water flows in each furrow, could still achieve
uniform wetting while effectively reducing deep percolation [Kruse, 1975].
This adjustment would be required only during the first two irrigations
of the season, since lower infiltration rates and higher evapotranspi-
ration losses during the hotter summer period lead to conditions where
significant deep percolation would not be experienced even with longer
sets.
A second on-farm management practice which could affect rates of
deep percolation is adjustment in cropping patterns. Since irrigation
crops typically vary as to planting dates, length of growing season and
evapotranspiration requirements, deep percolation is crop-specific. A
reduction in deep percolation per unit area could be achieved by substi-
tuting crops which are low contributors for those which are more of a
problem.
The third approach is to control seepage loss in the farm delivery
system. Most present conveyance laterals and head ditches in the Grand
Valley are unlined. By converting the on-farm conveyance system of
unlined ditches to solid, moveable aluminum pipe, as much as 95 percent
of on-farm seepage losses might be saved and used for irrigation.
Replacing head ditches with gated aluminum pipe may also improve the

-------
23
uniformity of water application (hence reduce deep percolation), but
no such empirical evidence is available to document this for Grant
Valley conditions.
Land Retirement
A more drastic approach to control of saline return flows in an
irrigated area would be permanent withdrawal of water supplies,[Howe and
Orr, 1974]. Because of the aridity of the study area, this would imply
the irrigated lands would be permanently retired from crop production.
Land retirement, either compulsory or non-compulsory, is a zero discharge
control option that could be implemented in conjunction with other less
costly structural and non-structural control measures.
Water supplies could be withdrawn in two ways. First, the least
productive soils could be taken out of production on a selective basis.
This option would have a minimum impact on the local economy, but would
not necessarily be less expensive in comparison to a second option,
namely retirement of an entire block of irrigated land and service canals
and laterals. This alternative could have a greater impact on the local
economy, since the soils included would have a higher average produc-
tivity. However, because seepage losses would be omitted under the
selective (or partial) retirement scheme, the amount of salts avoided
per acre or unit area would be considerably larger with the complete
retirement option.
C. CONCEPTUAL FRAMEWORK FOR ESTIMATING PROGRAM COSTS
Conventional theory of the firm and interindustry analysis comprise
the formal analytical approach and subsequent empirical analyses reported
in this study. The technique of linear programming is used to derive the

-------
24
cost-effectiveness of on-farm controls programs. Indirect impacts
associated with land retirement are assessed using a small region input-
output model of the impacted area.
Representative Farms and Linear Programming
Linear programming models of five representative farm situations
(each characterizing a different farm size category) provide the basis
for deriving estimates of the direct economic costs of on-farm salinity
controls. This approach is based upon the presupposition that farmer
resource allocation behavior can be appropriately represented in the
activity analysis-optimization format [Day, 1963]. The linear programming
algorithm utilized in the analysis solves a conventional short run land
and water allocation problem, subject to constraints on cropland, water
and acreages of specified crops. The objective function is net income
(defined as gross crop sales minus operating costs). Aggregated, the
models form a Valley-wide characterization of farm sizes, resource levels,
production technologies, cropping patterns and irrigation practices.
Under alternative price and operating assumptions, such models of rep-
resentative farms have proven useful in assessing production response
and decision behavior relating to a broad range of policy issues [Kelso,
et al., 1973].
Salinity levels are included in the model solutions via estimates of
deep percolation. The estimates are derived from a separate soil-water-
crop model which characterizes the hydro-salinity relationships for
surface irrigation techniques in the Valley. Each crop production acti-
vity incorporates a coefficient representing annual deep percolation per
acre. Our principal results assume that salt is picked up at the rate
of five tons of dissolved solids per acre foot of deep percolating water.

-------
25
This rate represents the average for the Valley and reflects a compro-
mise among conflicting estimates. However, since this parameter is
regarded as the least reliable in the analysis, we also consider alter-
native rates in evaluating the cost effectiveness of on-farm control
programs.
The model is solved to find the net income-maximizing situation
for each of a number of increasingly severe constraints on water supply.
Direct costs are computed in terms of net income losses due to hypothe-
sized imposition of water supply constraints. It is assumed that a
farmer, facing a constrained water supply, would seek modified irrigation
practices which would increase his efficiency of water use. Modifica-
tions in typical (current) irrigation practices are analyzed in the
model by introducing production processes with varying water supplies,
application rates, timing and methods.
Some of such practices would reduce deep percolation losses, hence
salt pickup contributed by that source. Other modifications may involve
substituting high water using crops for crops which require less water,
or adopting seepage control measures on the farm.
Interindustry Analysis
The adverse effects of some types of salinity control programs, for
instance land retirement, can be expected to involve indirect as well as
direct costs. Indirect or secondary effects are defined as the conse-
quences which stem from the original or direct effect. While small-
region changes in income are not synonymous with changes 1n national
income, a careful delineation of local consequences can be useful Infor-
mation for the public decision-making process.

-------
26
Techniques of secondary impact analysis, made possible with the use
of an input-output model of the U. S. economy, were formally introduced
by Leontief in 1936. Since that time the input-output approach has become
a popular tool for regional analysis. But like other useful analytical
tools the technique has been misused in that it was often applied in a
purely mechanical way [Richardson, 1972].
The indirect costs of land retirement are estimated using a small
region input-output model of the Grand Valley economy. The approach
followed 1n constructing and using the model draws upon some recent
extensions of the technique suggested by Harmston and Lund [1967]. These
economists have successfully applied the input-output framework to com-
munity (small) economic systems which, heretofore, have not received
the careful attention of regional analysts. Other modifications in
conventional methods are also used where deemed appropriate to the
purposes of our analysis. These are discussed briefly in section V
of this report, and in more detail in a supporting manuscript [Leathers,
1976a].
D. ECONOMIC BENEFITS OF REDUCED SALT DISCHARGE
We have not ourselves performed any studies of downstream benefits
of salinity control programs, but as with the upstream control program
costs, a number of questions of concept and method in the published
estimates remain unresolved. Some of the more important of these are
reviewed here. First, there is the question of secondary or indirect
economic benefits. Most economists agree that secondary benefits are
justified under a national accounting stance only if there can be identi-
fied a failure of market processes to adequately function in forward or
backward-linked sectors. Typical example of such failures include

-------
27
resources which might become unemployed or immobile unless salinity
levels are reduced. We cannot identify any such conditions which
would warrant incorporating secondary impacts into the economic evalua-
tion of salinity control. There undoubtedly would be pecuniary or
regional income distribution effects. Some of these would be negative,
e.g., the reduced income of household appliance and plumbing firms. Such
losses would probably be offset by gains in the regional economic sectors
which eventually profit by reduced consumer spending to mitigate salinity
damage, but the net effect is unknown.
A second issue involves the locus on the aggregate damage function
where marginal damages are computed. We disagree with the forecasts
which project salinity in the Lower Colorado River Basin to increase
in accordance with past trends. Since the damage function is non-linear,
marginal damages computed on the basis of projected high salinity levels
are much greater than estimates derived from current salinity levels.
Due to the fact that most salinity is from natural causes and since man-
made increments to salinity will be inhibited by environmental regulations,
we believe damage estimates will be most accurately derived by use of
roughly current levels of salinity.
A third issue rests on the need for employing the cost of the most
likely alternative for dealing with salinity as a measure of downstream
damages. Where damages can be feasibly mitigated at a cost less than the
damages avoided, the benefits are not greater than the cost of the alter-
native process. The southern California area has recently acquired
relatively large supplies of high quality Feather River water. Blending
such water with Colorado River water, or replacing the latter outright,

-------
28
represents perhaps a less expensive alternative than continuing to suffer
damages from more saline Colorado River supplies.
We interpret a report of the Environmental Protection Agency [1971]
to imply that direct downstream benefits (updated for 1975 prices at
current levels of salinity) are about $6 per ton of salt removed. At
this level of benefit, it is clear that the more expensive structural
control measures would not be economically justified. Further, the zero
discharge called for in recent U. S. water quality legislation could
only be achieved at economic costs several times larger than the down-
stream benefits, not to mention the significant social impacts on the
local community due to an abrupt cessation of production in an important
sector in the regional economy.
The Bureau of Reclamation [1976] has published an alternative esti-
mate of direct downstream benefits which amounts to $17 per ton of salt
removed. This estimate apparently does not consider the possibility of
using Feather River water as an alternative source in southern California,
and further, relies on higher projections of future salinity levels than
we consider appropriate.

-------
SECTION IV
THE ECONOMICS OF ON-FARM SALINITY CONTROL
A. INTRODUCTION
Management of irrigation water on the farm has received much at-
tention as a potential salinity abatement strategy in the Upper Colorado
River Basin. The attractiveness of on-farm controls stems from the
belief that such measures represent the least-cost means of removing
substantial quantities of salts while at the same time conserving irri-
gation water. However, the efficacy of this proposition has been only
partially examined thus far using sound empirical data. This section re-
ports the first such comprehensive study.
Our analysis focuses on three general managment options that farmers
could adopt to reduce return flow salinity. These include: (1) improving
irrigation application efficiency, and (2) crop substitution, both of
which could limit salts picked up via deep percolation from irrigated
fields; and (3) improving on-farm water handling with the use of aluminum
pipe conveyance systems, thus limiting seepage-induced salt pick-up.
The analysis and results are presented in four parts. In part one,
section B, we discuss the formulation of the representative farm models
used to characterize farmer response behavior. The assumptions and infor-
mational requirements which underlie salinity response effects are reported
in section C. In the last section, D, our major findings are reported. This
section also includes a discussion of policy implications and limitations of
the analysis.
29

-------
30
B. SIMULATING FARMER RESPONSE
Linear programming models were developed to simulate aggregate
(valley-wide) production response to various nonstructural salinity con-
trols that farmers could implement themselves. A disincentive, in the
form of a constraint on irrigation water supplied to farms, is presumed
to provide the impetus for adoption of more efficient irrigation prac-
tices thus influencing salts picked up in excess drainage waters.
Data Source and Sampling Method
Data for this analysis were collected by the authors in connection
with an earlier study. Personal interviews with farmers were conducted
in the Grand Valley during the summers of 1972 and 1973. The participants
were randomly selected from water user lists made available by the irri-
gation companies and associations which divert irrigation water from the
Colorado and Gunnison rivers. These lists typically include numerous
small users (one to three acres in size) as well as commercial farms.
Individuals with less than a 40 acre water appropriation (about 1 cfs)
were not included in the sampled population. Subsequently, from approxi-
mately 350 "commercial" farms (40 or more acres in size), 98 complete
interviews were secured, a sampling rate of about 28 percent.
The interview schedule was designed to obtain a wide array of infor-
mation: the numbers and sizes of farms; land tenure, planning and manage-
ment practices; resource inventories and production technology; crops and
livesotck grown, cropping patterns and cultural practices; prices paid
and received; and other data relevant to irrigated crop farming in the
study area. These data were used to delineate size parameters and resource
configurations of representative model farms.

-------
31
The Programming Models and Assumptions
Linear programming models of representative farms were developed from
detailed enterprise and resource budgets. A different model was used to
describe each representative farming situation (or general case). Descrip-
tive parameters of major importance include: farm size, enterprise alter-
natives, production levels, costs and returns, resource constraints,
amoung others. Typical constraints faced by fanners, including limits
on land, water, operating budgets, crop rotations, add realism to the
"optimized" solutions and to the policy implications derived from them.
Farm size. The fact that size can have an important influence on
operating efficiency and profit is well known. [Carter and Dean, 1961].
Whether or not Grand Valley farmers can harmlessly absorb the private
costs of salinity abatements, costs which can be imposed on them directly
or indirectly, depends in part upon their capacity to generate compen-
sating revenues. Larger, more efficient farms are generally in a better
position to meet additional obligations without materially affecting
their operations or output. Smaller, less efficient farmers may not be
so fortunate. The size distribution of farms, therefore, is an important
consideration in modeling farmer response to various salinity control
programs involving private costs.
On the basis of the sampled data, five model farm sizes were selected
as representative of the size distribution of farms in the Grand Valley.
These were farms of 40, 80, 130, 210, and 370 acres in size. A frequency
distribution of sampled farms is reported in Figure 4. More than half of
the farmers interviewed had operations of 120 cropped acres or less. Very
few had farms of more than 300 acres. A high percentage of the smaller
owner-operators work off the farm to support their families. These people

-------
NUMBER OF
FARMS
,0 MODEL MODEL
i n
> i ¦
9
8
7
6
5 h
4
3
2
I
MODEL
in
MODEL
32
MODEL
31
I
11
X
±
±
50 100 150 200 250 300 350
NET CROPPED ACRES PER FARM
400 450 500
CO
PO
Figure 4: Frequency Distribution of Sample Farm Size and Model
Acreage Parameters

-------
33
readily acknowledged that farming is not a primary source of family
income. Rather, it is viewed as a "consumption" activity or ... "a
nice environment for raising children". Judging from the number of
individuals (with established water rights) excluded from the sample
population because of the 4-0 acre cutoff, many families in the Grand
Valley (perhaps 2,000 or more) are considered small-scale "farmers".
Production Possibilities. A moderately-long growing season (about
200 days) permits the production of a wide range of field, forage,
vegetable and orchard crops in the Grand Valley. In terms of planted
acreages, the most significant crops are corn, small grains (including
wheat, feed and malting barley, and oats), sugar beets, permanent pasture
and alfalfa hay. As indicated in Figure 5, planted acreages of these
crops have remained fairly stable in recent years. Differences in crop
mix with respect to farm size are described in Figure 6.
Orchard and other speciality crop enterprises, which account for
about 15 percent of the total irrigated acreage, are quite diverse in
ownership, production techniques, size of operation, and productivity.
For this reason they are difficult to model with an acceptable degree of
accuracy, and therefore were not included among the enterprise alternatives
considered in the models. Further, no attempt was made to analyze in
detail the many kinds of livestock enterprises found in the Valley. These
are also very diverse in nature and would be difficult to model accurately.
Costs and Returns. Detailed enterprise cost and return budgets were
developed to provide the necessary technical and revenue coefficients for
the analyses. The unit budgets are model-specific, i.e. they are sensi-
tive to farm size. Costs of production were calculated from 1975 input
prices, and include all variable, fixed and overhead costs directly

-------
34
Figure 5. Harvested Acreages of Selected Crops Grown on Bureau of
Reclamation Project Lands (Garfield Gravity Division)
in the Grand Valley, 1948-1974
Acreage
8000
7000
6000
5000
4000
3000
2000
1000
0
1948 1951 1954 1957 1960 1963 1966 1969 1972 1974
Selected Years
A = Alfalfa
P = Pasture
SG = Smal1 grains
C = Corn
SB = Surgar beets
SOURCE:
Grand Valley Water Users Association project records
(selected annual production reports).Bureau of Recla-
mation project office, Grand Junction, Colorado.

-------
35
Figure 6: Acreages of Selected Crops as a Percentage of Total
Cropped Acres for Model Farms
PROPORTION OF SELECTED CROPS
TO TOTAL CROPPED ACREAGE
(PERCENT)
FARM MODEL

-------
36
chargeable to a particular crop enterprise. Fixed arid unallocatable
overhead costs were not included, hence net returns measure residual
income above operating costs.
Gross sales were estimated using average yields for each farm size
category and five year (1970-1974) average product prices. A price
"weighing" procedure was thought to more accurately reflect long-run
conditions for planning and evaluative purposes, since the price level
for farm products has been abnormally high in recent years. Tables 1
through 5 in the Appendix section of the report summarize net crop re-
turns and activity levels for each farm model. A detailed explanation
of the budgeting procedure and summary of representative farms in the
Grand Valley is reported elsewhere [Leathers, 1976b].
Resource and Other Constraints. Land available for irrigated crop
production in the Valley has changed very little since the 1940's. Prior
to that time significant acreages of land were abandoned. It has been
estimated that as much as 20 percent of the once-irrigable land is idle
due to high salinity and seepage problems resulting from poor drainage
[Robinson, 1969]. Any future decline in the present irrigated area of
approximately 56,000 acres would be primarily a consequence of suburban
growth near Grand Junction and other smaller towns located within the
Valley. Renovation of the drainage system (a proposed salinity control
measure) could eventually increase the present acreage base by as much as
15 percent. This potential increase in the land resource was not consi-
dered in our analysis, nor was the possibility of developing new lands
contiguous to present delivery systems. Following these assumptions the
constraint on available crop land was set at approximately 47,500 acres,
85 percent of current irrigated land in the Valley.

-------
37
As pointed out in Section II, water availability is not an important
constraint on crop production. With the possible exception of peak con-
sumptive use periods during July and August (when farmers in the Bureau
of Reclamation project area may face limited water rationing) water supplies
are well in excess of normal irrigation requirements. Since sufficient
supplies are generally available during season (April to October), it was
not necessary to distribute water supplies in a monthly or multiperiod
format.
The base constraint used in the model is approximately 4.5 acre
feet per acre on an annual basis. From this initial solution, water
supplies are reduced in five percent increments of the original supply.
This forces optimal responses to an increasingly severe constraint on
alternative uses of water on the representative farms. Because of high
groundwater salinity in the Valley, it is unlikely that pump water will
be substituted for surface water as supplies are reduced. However, this
possibility was not studied.
Acreage restrictions were placed on each crop in an attempt to main-
tain an aggregate (Valley-wide) cropping pattern reasonably close to the
present crop mix. The cropping pattern in 1973 was used as the base year
for setting the limits on allowable shifts in crop acreage. A minimum
acreage base was established for feed grains and forages since these crops
are low income earners in relation to others. Maximums and minimums
were set for corn and sugar beets. The limits used (which are shown in
Tables 1 through 5, Appendix) allow substantial shifting in crop mix to
occur in response to limited water supplies. But a degree of realism is
maintained by allowing acreage changes to go only so far, since "other"

-------
38
factors besides profit and efficiency criteria determine the Valley
cropping pattern, but nonetheless cannot be included in models of deci-
sion behavior.
Modifications in Irrigation Practices
Irrigation is one of the most time-consuming tasks on Grand Valley
crop farms. Many of the smaller farms, which account for a significant
portion of total irrigated acreages, are managed by owner-operators who
maintain part-time or full-time jobs off the farm. They accomplish farm
work after normal working hours and on weekends. Accordingly, irrigation
settings are usually changed at 24 or 48 hour intervals, i.e. once a day
either before or after work. Because irrigations are worked around other
farm tasks, larger farms typically follow the same practice. As noted
earlier, low cost for incremental water deliveries (about $2.00 per
acre foot) and non-enforcement of water rights provide a situation in
which there is little incentive for using water efficiently.
The additional cost of reducing the length of an irrigation set
(reducing the amount of water applied and the time of water intake)
involves no real resource costs. The tasks involved in changing sets
must be performed in any case. However, an "inconvenience cost" is in-
curred, since, for row crops, the operator would have to set up and moni-
tor his water twice each day at twelve hour intervals rather than once
per day as is typical at present. There is no particular empirical basis
upon which to assign an accounting price to this apparent inconvenience.
Our crude approximation consisted of using a charge for the relevant
irrigations amounting to twice the normal irrigation wage to reflect an
"overtime" labor requirements. This figure, twice the prevailing irrigation

-------
39
wage, is also close to the average off-farm wage in the area, and might
be construed as an opportunity cost for those employed off the farm.
A number of assumptions regarding present irrigation labor require-
ments were made to obtain an estimate of the additional labor (or conve-
nience) required to make the necessary modifications. These assumptions,
in the form of labor input coefficients for a given irrigation set, under
specified operations and conditions, are summarized in Table 1.
First, a reasonable labor input requirement for each irrigation sit-
uation was assumed (i.e., for a particular crop or row spacing siphon
discharge rates per furrow, water supply at the farm headgate, and irriga-
tion timing during the season.) The labor input was specified on the
basis of two compenents: (1) setup time, which included moving canvases
or portable dams, starting the siphon tubes, and adjusting siphon flow;
and (2) monitor time, which includes clearing clogged furrows, managing
field tailwater, and periodic checks of the field. A further refinement
was made to allow for variations due to seasonal conditions. The first
two irrigations in the spring typically require more attention (monitor
time) than those that follow primarily because of differences in soil
characteristics and cultural practices. The assumed labor coefficients
reported in Part A are not experimental but are thought to be reasonable
and valid for the purpose at hand.
Irrigation set times were reduced from 24 to 12 hours for corn, small
grains, and sugar beets and from 48 to 24 hours for permanent pasture and
alfalfa. The inconvenience cost was determined by multiplying the addi-
tional labor time by a charge twice the normal wage of $2.25 per hour.
A similar (though weaker) argument is made for changing alfalfa and
pasture irrigation from two to one day sets. The procedure allows for

-------
40
TABLE 1. ESTIMATES OF ADDITIONAL LABOR (INCONVENIENCE) COST TO FARMERS
FOR SHIFTING FROM PRESENT TO MORE EFFICIENT IRRIGATION PRACTICES
DISCHARGE RATE PER FURROW^

ITEM

8 gpm
4gpm
A.
LABOR REQUIRED PER SET:






	 Hours -


Set up time

1.5
2.0

Monitor time:




First two irrigations (ea.)

3.0
4.0

Later irrigations (ea.)

1.5
2.0

Total labor time:




First two irrigations (ea.)

4.5
6.0

Later irrigations (ea.)

3.0
4.0
B.
ACRES IRRIGATED PER SET:—^






	 Acres -


Furrow spacing:




24 inches

2.06
4.11

30 inches

2.57
5.12
C.
HOURS OF LABOR PER ACRE






	 Hours/Acre


First two irrigations (ea.)




24 inch spacing

2.18
1.46

30 inch spacing

1.75
1.17

Later irrigations (ea.)

1.46


24 inch spacing

.97

30 inch spacing

1.17
.78
D.
ANNUAL IRRIGATION LABOR COST\—





Present
Modified Additional

Crop
Practices
Practices
Cost



-- $ per acre 	


Corn
18.44
22.38
3.94

Small grains
14.92
18.86
3.94

Sugar beets
22.14
27.05
4.91

Permanent pasture
16.68
20.62
3.94

Alfalfa
16.68
20.62
3.94
a/ Assumes water supplied at the farm headgate at a rate of 1 cfs per
40 acres net of on-farm conveyance losses.
b/ Number of acres capable of being irrigated with 1 cfs per 40 acres
water supply.
c/ Labor requirements per set (part A) * acres irrigated per set
(part B).
d/ All cost estimates based on $2.25 per hour wage rate.

-------
41
increasing irrigation labor costs to farmers as a result of adopting a
more efficient irrigation practice, while the actual labor time on a per
unit basis (the total irrigation labor input per acre per year) remains
the same. Annual irrigation labor costs with and without this modifica-
tion are compared in Part D. These costs of adopting the more efficient
irrigation practice are, to emphasize, not real resource costs, and may
be an over-estimate of the actual inconvenience to the farmer.
C. IRRIGATION-SALINITY RELATIONSHIPS
To identify the consequences of modifying irrigation practices on
deep percolation (and subsequent salt pickup), a substantial amount of
information about soil-water-crop relationships is required. Furthermore,
this information, if used in an analytical model to deduce policy impli-
cations, must be both reliable and representative of actual conditions
and practices.
A simplified soil-water-crop budget was developed to trace the
effects of various salinity control programs on water utilization, plant
growth and salinity in the Grand Valley [Leathers and Franklin, 1975].
The budgeting model focuses on deep percolation and seepage mechanisms
and attempts to estimate empirically the rate of salt pickup in irrigation
return flows. The results of this investigation and the policy impli-
cations are summarized here.
The Soil-Water-Crop Budget
The Grand Valley is characterized by considerable heterogeneity in
soil conditions, topography, operating procedures of irrigation companies
and districts, and farm production practices. Accordingly, the data

-------
42
summarized herein represents a careful selection of available information
thought to be broadly representative of the Valley as a whole.
Typical irrigation practices followed by farmers and resulting impli-
cations for seasonal soil moisture depletion and crop evapotranspiration
are described in Table 2. These observations do not differ markedly from
similar data reported by others [Bureau of Reclamation, 1974]. In Table 3
the results of the budgeting model are summarized. The most critical
budgeting assumption is the seasonal variation in water intake rates.
Two seasonal conditions, early and mid-late season, are assumed to
represent the change in water infiltration as the irrigation season
progresses. A more precise approach would call for a specified rate
for each sub-period (e.g., at each irrigation) since the actual change
is more gradual [see Skogerboe, et al_., 1973 and 1974; and Gilley, 1968].
Nonetheless, our results pose some interesting policy implications for
control of on-farm salinity.
First, it is apparent that all crops do not contribute to deep
percolation uniformly. Annual crops generate a deep percolation percen-
tage (or leaching fraction) of 20 percent or more with small grains having
the highest at 28 percent. Perennials contribute significantly less to
deep percolation, since these crops have high evapotranspiration demands
and are more likely to be stressed rather than overirrigated (see Table 2).
But if stress should occur early in the growing period—enough to suffi-
ciently reduce plant populations—the likelihood of greater deep percola-
tion occurring later in the season (and in subsequent years) would be much
higher.
Second, farm irrigation efficiency and deep percolation are not always
closely related. That is, reduced deep percolation does not necessarily
follow from improved irrigation efficiency. Because of the seasonal nature

-------
43
Table 2. Traditional Irrigation Practices and Soil Moisture Relationships
for Selected Crops in the Grand Valley^/
CROP
ITEM
Corn
Small
Grains
Sugar
Beets
Pasture
Alfalfa
Number of Irrigations
8
6
9
7
7
Number of Pre-irrigations
1
—
—
—
—
Cumulative ET (inches)
26.5
20.9
31.6
30.12
32.16
Mean ET per Irrigation—^
3.35
3.45
3.48
4.30
4.59
Root Zone Storage (inches)
9.2
9.2
9.2
9.2
14.0
Soil Moisture Depletion
at Each Irrigation^/
38%
38%
38%
47%
33%
Irrigation Interval (days)-''





Longest
Shortest
42
9
27
9
34
12
31
19
25
18
Length of Irrigation Set
(Hours)





First two irrigations
All others
24
24
24
24
24
24
48
24
48
24
a/ Assumes a normal growing season.
b/ Cumulative ET * number of irrigations per season.
c/ Mean ET per irrigation * root zone storage capacity.
d/ Number of days between irrigations where cumulative ET reduces the
percent available soil moisture to the indicated levels.

-------
44
Table 3. Water Budget Summary for Traditional Irrigation Practices in the
Grand Valley: Annual Water Use and Losses for Selected Crops



CROP


ITEM
Corn
Smal 1
Grains
Sugar
Beets
Pasture
Alfalfa



Annual AF/A


Water Applied—^
3.74
2.98
5.28
4.80
4.80
Root Zone Additions
2.94
2.40
3.27
3.06
3.06
Crop Consumptive Use
2.21
1.74
2.63
2.51
2.68
Deep Percolation
.73
.66
.64
.55
.38
Field Tail Water
.80
.58
2.10
1.74
1.74
Farm Irrigation Efficiency—^
59%
58%
50%
52%
56%
c /
Leaching Fraction-
25%
ro
00
20%
18%
12%
a/ "Net" of on-farm delivery losses,
b/ Defined as crop consumptive use * water applied.
c/ Defined as deep percolation t root zone additions.

-------
45
or infiltration rates for Grand Valley soils, in all but about the first
two irrigations water applied in excess of the amounts indicated for a
24 hour set would lower irrigation efficiency (by increasing tailwater
loss) but not measurably alter deep percolation. Therefore, salinity
controls which involve improving irrigation efficiency need only concen-
trate on the first part of the irrigation season and should be crop-
specific.
Farm irrigation efficiency studies conducted in the Valley during
the 19501s by the Bureau [1957] and more recently by Skogerboe and Walker
[1973], estimated overall efficiency at 40 percent. This does not agree
with the results of our aggregated water budget, which is compared with
Skogerboes' in Table 4. Although some cropped acreage is omitted in the
computations (approximately 8,500 acres of orchards and speciality crops),
our findings differ most significantly on the estimate of deep percolation.
Our own estimate of water use and deep percolation are used for analysis
purposes, since these are believed to be a middle ground for available
estimates of these parameters.
Rate of Salt Pickup in Return Flows
In the Grand Valley seepage and deep percolation waters do not return
to the river as direct, separable flows but mix with saline waters con-
tained in gravel-cobble acquifers. This mixing of return flows with
acquifer water is supported by the observation that surface return flow
in open drains (typically very low rates of flow) couldn't by itself
account for the total salt pickup [Kruse, 1975]. Both seepage from canals
and deep percolation water create a hydrostatic pressure gradient, forcing
some saline water out of the gravel-cobble acquifer into the river. For

-------
Table 4. Comparison of Total Water Budgets for Irrigation in the Grand Valley:
Aggregate Use and Losses
TOTAL VALLEY BUDGET—^	CROPS-ONLY BUDGET	
ITEM AF AF/A Percent	AF AF/A Percent
annual summary
Water Applied
373,000
6.2
100
214,380
4.4
100
Root Zone Additions
211,000
3.5
57
147,146
3.0
69
Crop Consumptive Use
150,000
2.5
40
126,310
2.5
59
Deep Percolation
61,000
1.0
16
20,836
.42
9
Field Tail Water
162,000
2.7
43
67,234
1.4
31
Total Acreage

60,800


47,480^

aj Adapted from Skogerboe, et a]_., [1974].
b/ Excludes orchard, truck, and other speciality crops.

-------
47
this reason it is difficult to assess the relative importance of canal
and lateral seepage and deep percolation because there is no easy way to
measure and differentiate the return flows as two different mechanisms.
A further complication stems from the practice of farmers to run "tail"
water into the open drains and natural washes during the irrigation season
thus diluting the concentration of drainage water.
The ionic constituents and ion ratios of drainage, acquifer, ground,
and river water entering and leaving the Grand Valley area were compared
in an attempt to derive empirical estimates of salt pickup. The ion
contents were used to calculate an average salt pickup rate per acre-foot
of deep percolation water. To obtain an estimate of the average ionic
composition of diverted river water, various data from the U.S.G.S.
sampling station located at Cameo were used [U. S. Geological Survey,
1966-1970]. The period of analysis was April through September, 1966-
1970.
Based on these inflow data, theoretical equilibrium compositions of
soil water were calculated to compare results for different assumed
leaching fractions. The predicted equilibrium values, reported in Table 5,
were obtained from a computer program developed by Rai and Franklin [1973].
The model compares results with a nongypsiferous, calcareous soil type, and
calculates the drainage water composition if no 'fossil" or soild phase
salts are dissolved from the subsoil. According to salt balance theory,
different fractions of irrigation water (leaching or deep percolating water)
applied to growing crops result in different ion concentrations in soil
water at the bottom of the root zone. Assuming that the soils contained
lime (CaCOg) but not gypsum or other soluble salts associated with the
shale substrata, the drainage waters should be comparable to those shown.

-------
48
Table 5. Average Composition of Colorado River Water Diverted for Irrigation
and Predicted Ionic Equilibrium Compositions for Various Leaching
Fractions
Diverted
River Water^-' 	Percent Leaching Fraction
Constituent (average)	30	25	20	10

me/1
- - - -
equilibrium values
in me/1—^
- - -
Ca
2.71
3.96
4.10
4.47
6.18
9.43
Mg
1.27
4.23
5.08
6.35
12.70
25.40
Na
3.10
10.33
12.40
15.50
31.00
62.00
K
.07
.23
.28
.35
.70
1.40
hco3
2.39
2.74
2.84
2.89
2.90
3.03
so4
2.00
6.67
8.00
10.00
20.00
40.00
CL+N03
2.76
9.20
11.04
13.80
27.60
55.20
CaC03
.02
5.42
6.74
9.09
20.92
24.77
Average TDS/AF = .7 tons
a/ Quality of water at Cameo, Colorado, or approximate point of diversion
for irrigation.
b/ Obtained from a computer program developed by Rai and Franklin [1973].
me/1 = milliequivalents per liter.

-------
49
The data illustrate that only CaCo^ would be expected to precipitate.
As irrigation becomes more efficient, (i.e., the leaching fraction is
reduced) Mg, Na, SO^, and CI would increase to relatively high value.
Estimates of the quality of drainage water was facilitated with the
use of water samples taken from natural drains in early 1975 (January
and February). Detailed analyses of the samples were conducted by
Agricultural Research Service personnel [Kruse, et^ al_., 1975]. Average
ion concentrations for the Little Salt, Big Salt, Adobe, Persigo, and
Hunter drain outlets are summarized in Table 6. The ionic composition
is reasonably consistent among the different drains, although average
TDb varies between 3.1 and 4.6 tons per acre foot (T/AF). Using a
value of 0.7 T/AF to approximate the TDS of diverted irrigation water,
the average net salt pickup for all the drains amounts to about 3.2 T/AF.
This value might be representative of salt pickup in deep percolation
and seepage waters, not associated with the gravel acquifer, but which
are intercepted by the drainage system. Also, since TDS concentrations
of drainage waters are highest in the winter months, the estimate of
3.2 T/AF is probably a maximum value.
A similar procedure was used to estimate ground water salinity.
Ground water analyses of 11 wells located in an area east of Grand
Junction to Bethel Corner, of four wells near Fruita (all in the gravel
acquifer), and of 25 wells located north of the acquifer are summarized
in Table 7. The approximate location of these sampling sites is shown
in Figure 3, section II.
The acquifer waters near Fruita apparently contain about one-half
the TDS of those to the east of Grand Junction. This illustrates the
uncertainty in estimating an average rate of salt pickup in that salt

-------
Table 6. Ionic Analyses of Drainage Waters for Selected Drain Outlets in the Grand Valley:
Averaged Concentrations, January and February, 1975
3
EC x 10 	Ionic Constituent in me/1	 TDS in
Wash or Drain
mmhos/cm
Ca
Mg
Na
K
hco3
S04
CL
no3
Tons/AF
Little Salt
3.71
16.5
12.8
20.4
.2
7.2
40.6
6.5
1.0
4.4
Big Salt
3.07
15.9
12.8
12.0
.3
7.1
29.7
4.1
0.5
3.1
Adobe
3.90
17.7
16.7
20.4
.3
7.3
40.9
6.6
1.0
4.6
Persi go
3.12
234.2
8.0
11.1
.2
4.4
37.2
3.6
0.3
3.2
Hunter
3.67
19.8
18.1
14.6

6.5
41.9
5.6
0.7
4.4
Average
3.49
18.8
13.9
15.7
.3
6.5
38.1
5.3
0.7
3.9

-------
Table 7. Ionic Analyses of Ground Waters at Two Locations in the Gravel-Cobble Acquifer and
at One Location North of the Acquifer: Averaged Concentrations
3
EC x 10		Ionic Constituent in me/1	 TDS in
Location	mmhos/cm	Ca Mg Na K HCO^ SO^ CL NO^ Tons/AF
Acquifer Waters:
Grand Junction Area	6.03	23.3 39.4 32.7 .4 8.5 74.6 11.0 1.2	8.5
(11 wells)
Fruita Area	3.88	18.0 13.5 24.5 J- 4J5 50.6 7.0 .3	4.6
(4 wells)
Average	4.96	20.6 26.4 28.6 .4 6.6 62.6 9.0 .8	6.6
Waters North of the Acquifer:
(25 wells)	4.92	18.1 19.1 51.9 .4 7.9 68.0 9.6 1.9	5.9
Overall Average	4.94	19.4 22.8 40.2 .4 7.2 65.3 9.3 1.3	6.2
tn

-------
concentrations within and outside of the gravel acquifer are noticeably
different. However, when the acquifer concentrations are averaged
together the resulting concentration is close to that of the ground
water. Total dissolved solids for averaged acquifer waters is 6.6 T/AF
compared to the average ground water concentration of 5.9 T/AF.
To obtain a salt pickup rate which is representative of "mixed"
return flows (i.e., some combination of surface drainage and acquifer
waters) it is necessary to weight the relative concentrations by the
appropriate magnitudes of flow. Subsurface return flows (deep percola-
tion and seepage water) were assumed to comprise about 80 percent of the
total return flow from irrigation, and water intercepted by surface
drains about 20 percent. This proportion reflects a concensus among
researchers based on limited measurements [01 sen, 1975]. Three estimates
of the average (Valley-wide) salt pickup rate, based on this 80:20
weighting scheme, are reported in Table 8.
A pickup rate of 5.36 T/AF reflects a combination of acquifer and
surface drainage waters. This rate might be representative of the
irrigated acreage situated over the acquifer. Ground and surface waters
together yield a pickup rate of 4.89 T/AF, possibly representative of
irrigated acreage located north of the acquifer. Our choice of an average
salt pickup rate was the third option, 5.04 T/AF, a combination of all
return flow sources. This rate represents an average for the Grand Valley
and reflects a compromise among different rates of salt pickup associated
with the various pickup mechanisms. A more detailed description of the
data and procedures used is also contained in the Leathers-Franklin
manuscript [1975].

-------
53
Table 8. Average Salt Concentrations for Sources of Surface and Subsurface
Irrigation Return Flows and Weighted Estimates of Average Rates
of Pickup
Item
RETURN FLOW SOURCE AND QUALITY
Surface
Drains
Acquifer
Ground
Wa ter
Combined ,
Subsurface-
Mean Values
Less: TDS Inflow
Net TDS Outlfow
3.9
.7
3.2
TDS in Tons/AF
6.6	5.9
.7	.7
5.9	5.2
6.2
.7
5.5
Weighting Scheme:
20
20
20
Percent of
Total Return Flows
80
80
80
Average
Pickup
Rate(T/AF)
5.36
4.80
5.04
a/ Simple average of acquifer and ground water concentrations.

-------
A note of caution in using this figure should be mentioned. This
estimate of 5 T/AF is based on calculations which assume that the total
annual salt pickup is attributable only to irrigated agriculture. Signi-
ficant contributions of salt from natural runoff (if they prove important)
will cause the control effectiveness of on-farm measures to be over-
estimated.
D. RESULTS AND SUMMARY
it was hypothesized that improved efficiency in on-farm water use,
stimulated by constraining water deliveries to farms, would reduce salinity
in irrigation return flows. The direct costs of these farmers response
strategies, namely, improved irrigation techniques, crop substitutions, and
use of aluminum pipe, were derived with linear programing models of
representative farms. A soil-water-crop budget was used to measure the
appropriate water use and salinity coefficients with which to evaluate
cost effectiveness. In this section, the results of our analyses and
policy implications are summarized.
Estimated Costs for Control of Deep Percolation
Aggregated results of our analysis dealing with control of deep
percolation are summarized in Table 9. Row [1] portrays our estimates
of existing conditions, while row [2] is the initial or benchmark solution
of the five programming models in which irrigation water is nonlimiting.
Subsequent rows represent the predicted effects of successive 5 percent
reductions in available water supply on salt discharge, net crop income
and crop acreage. The estimate of salt discharge is based on the rate of
5 tons per acre foot of deep percolation water.

-------
Table 9. Estimated Annual Direct Income, Crop Substitution and Salinity Effects of Limiting
Irrigation Water Delivered to Farms
Irrigation Salt Discharge Total Incremental	Cropping Pattern of Model Farms
Water .in Irrigation. .Net Crop Income Loss of	Small Sugar	Total
Deliveries-'Return Flows— Income Salt Removal£/ Corn Grains Beets Pasture Alfalfa Acreage

A. INITIAL
CONDITIONS^7







214,745
146,540
5,962,301

13,893
6,460
3,582
13,328
13,895
51,158
201,234
147,813
7,915,280

23,897
2,730
7,635
2,896
10,315
47,473

B. INCREMENTAL REDUCTION IN
IRRIGATION WATER SUPPLY—^







2.04






191,173
131,594
7,882,176

23,897
2,730
7,635
2,896
10,315
47,473



1.88






181,111
112,125
7,845,633

23,897
2,730
7,635
2,896
10,315
47,473



1.65






171,049
86,601
7,803,431

23,897
2,730
7,635
2,896
10,315
47,473


1.69






160,987
51,635
7,744,495

23,864
3,038
7,635
2,896
10,040
47,473



18.88






150,926
30,370
4,343,006

19,413
10,245
7,635
2,896
7,284
47,473


85.31






140,864
18,205
6,305,168

13,677
19,858
7,617
2,896
3,425
47,473



406.26





130,802
15,525
5,216,403

7,370
28,350
4,944
2,896
2,913
47,473



293.36






120,740
7,211
2,777,371

7,370
31,092
3,202
—
2,913
44,577
a/ Water 1s measured as deliveries to farm headgates.
b/ Based on an assumed pickup rate of 5 tons of salts per acre foot of deep percolation water.
cf Defined as the change in income per unit change in salts removed.
d/ Row [1] represents the assumed present conditions. The estimates apply only to irrigated crops,
since specialty and orchard enterprises are not included in the farm models. Row [2] is the base solution
of the L.P. model with no constraints on water supply.
ej Successive Increments are withheld at five percent of the initial supply (part A, [2].)

-------
56
In general, our results indicate that substantial reductions in
this particular portion of salt discharge are possible at relatively
low cost. The first twenty percent reduction of water supplies is
absorbed inexpensively by more efficient irrigation techniques, yet
estimated salt discharge is reduced by 96 thousand tons over this range.
About 62 percent of the estimated on-farm initial salt discharge appear
to be avoidable at costs to farmers of less than $2.00 per ton. These
results are consistent with similar studies reported by Anderson, et al.,
[1974]. However, once water use is reduced to minimum evaporative and
leaching requirements, savings take place only by crop substitution and
actual cessation of crop production, at much greater cost.
The optimal (least cost) reponse of farmers to limits imposed on
delivered water would be first to adopt the more efficient irrigation
practice, namely to experience the inconvenience of having to irrigate
twice in 24 hours rather than once for the first two irrigations of each
crop. The alternative strategy of substituting crops, i.e., crops
which contribute less to deep percolation for those which are more of
a problem, becomes an efficient solution only following the complete
adoption of the more efficient irrigation process for all of the crop
activities. It was noted in Table 3 that corn contributes more to deep
percolation than alfalfa: .16 acre feet and .02 acre feet, respectively.
Using model III as an example (Table 3, Appendix), if substitution takes
place (alfalfa for corn), the net change in income per acre, $89 minus
$134 or -$45, divided by the net exchange of salt pickup (.16 minus .02
acre feet of deep percolation times five tons per acre foot), or .7 tons,
gives a cost of $64.29 per ton. Although this is comparatively high,
substituting lower value crops for corn would generate even higher costs.

-------
57
Conversely, substituting sugar beets for corn would yield negative direct
costs of removing some salts. Institutional and market constraints on
the expansion of crop production (especially sugar beets), however,
limits this type of crop tradeoff to a minimum in this analysis.
Reduction of salt loading attributed to crop substitution begins
at the level of about 165,000 acre feet of water supply and ends at
about 130,000 acre feet, or 3.3 and 2.7 acre feet per acre, respectively.
Beyond a 2.7 acre feet per acre water supply, salts can be removed only
by retiring some of the irrigated acreage base.
A sensitivity analysis of ftiese costs with respect to assumed salt
pickup rates and amounts of deep percolation suggest that the above
estimates should be regarded with extreme caution. For example, if the
amount of deep percolation per average acre in the Valley is reduced
by half (from 5 to 2.5 inches) and the rate of salt pickup is reduced
by half (from 5 to 2.5 tons per acre foot), the costs of improved irri-
gation practices will increase 4-fold, from $2.00 to about $8.00 per ton
removed. Therefore, until the physical relationships are well established,
the above conclusions should be regarded as preliminary estimates.
Estimated Costs of On-Farm Seepage Control
A second category of salt pickup associated with on-farm irrigation
water management practices stems from seepage 1n the farm water conveyance
system. On-farm conveyance losses are estimated to account for about the
same amount of salt as the field losses just discussed, 150,000 tons per
year. Aluminum pipe to replace laterals and gated aluminum pipe for head
ditches represent a means by which ninety percent of present on-farm
delivery seepage can be avoided.

-------
58
Losses from the farm conveyance system (farm laterals and head
ditches) could be influenced by constraining water deliveries, or
encouraged by a subsidy program. If faced with a limited water supply
it is quite likely that farmers would adopt such a strategy before they
would engage in crop substitution. We estimate the cost of a 90 percent
reduction in this portion of salt pickup can be achieved at an average
cost of less than $8 per ton. The procedure and results supporting this
conclusion are described in Table 10. Possible additional benefits
which may accrue to farmers as a result of adopting gated pipe, e.g.,
increased yields through better control of irrigation water, and/or
irrigation labor savings have not been included here since experience
with this system in the Grand Valley is rather limited.
However, these estimates should also be regarded as provisional
for the same reasons cited above. The cost effectiveness of on-farm
control measures (or for that matter, any control) is more sensitive to
the physical assumptions about salt removal than to assumptions of
economics.
Summary and Implications
We have reported in this section the economic impacts of several
alternative methods for reducing saline return flows from irrigated lands
in the Grand Valley. Our analysis suggest that adjustments in on-farm
water management practices are a relatively inexpensive methods of dealing
with the problem. Other proposed structural measures appear to be consid-
erably more expensive. However, several specific limitations should be
recognized in interpreting the results of this analysis.

-------
59
TABLE 10.ESTIMATED ANNUAL COSTS FOR REDUCING ON-FARM CONVEYANCE
SYSTEM SEEPAGE WITH ALUMINUM PIPE IN THE GRAND VALLEY
ITEM
PER
CROPPED ACRE
FOR
56,000 ACRES
A. MATERIALS REQUIREMENTS

Unlined laterals and ditches replaced—
aluminum pipe and structures:
8" pipe (laterals)
6" gated pipe (head ditches)
Concrete turnouts (one per 12 acres)
B. OWNERSHIP AND OPERATING COSTS
New
4- b/
cost:-
8" solid
6" gated
Turnouts
pipe ($2.18/ft.)
pipe ($1.78/ft.)
($114.96 installed)
Initial investment
c /
Annualized costs
Depreciation
Interest, insurance, and taxes
Repairs and maintenance
Total
C. CONTROL EFFECTIVENESS
Salt pickup avoided with 90% efficiency
Average annual cost per ton
51.2 feet
17.05 feet
34.15 feet
$37.14
58.65
9.58
$ 105.37
$ 9.48
5.26
4.21
$ 18.95
2.41 tons
$ 7.86
543 miles
180.8 miles
362 miles
4,667
$5,900,664
$1,061,368
135,000 tons
a/ These estimates are "net" of approximately 33 feet of unlined
laterals per cropped acre (350 miles). Although situated within the
confines of farms, these laterals are owned and maintained by the various
irrigation districts.
b/ Pipe cost estimates (in 1975 prices) were obtained from local
dealers. 100% replacement of all unlined laterals, and 50% replacement
of all irrigation head ditches owned by farmers is assumed (see footnote
a/. The number of concrete turnouts (from laterals to farm head ditches)
reflects an average field size of 12 acres in the Valley.
c/ Calculated using conventional farm budgeting techniques assuming
a useful life of 10 years and an interest rate of 6 percent.

-------
60
First, neither the amount of drainage water associated with specified
irrigation practices nor the rate of salt pickup per unit of drainage
water are well established. In fact, considerable disagreement is found
on these points among hydrology and soils specialists. A better under-
standing of the relative salt contribution of field percolation losses
versus seepage from conveyance systems is also necessary before any
definitive assessment of nonstructural controls can be achieved. Simi-
larly, the relative contribution of the various crops needs to be estab-
lished more precisely before the crop substitution alternative is rejected.
Second, it may not be possible to increase irrigation efficiency to
the degree assumed without some sacrifice in crop yield. A small decrement
in yield can have a significant impact on net returns to farmers. Similarly,
unstable farm prices could make reliable estimation of program costs and
benefits still more uncertain.
Third, for purposes of this analysis, we have assumed that farmers
could in some way be costlessly persuaded to adopt more efficient irri-
gation practices. To assure uniform adoption of such practices would no
doubt require a considerable regulatory and monitoring jeffort, which
would be both expensive and unpleasant for all concerned. We have not
dealt with the regulatory and social costs of imposing water quality
standards (or constraints on water supplies) in this sort of situation
where the effluent of individual irrigators is not identifiable. Present
water distribution policies in the area and Colorado water law do not
provide any incentive for reducing return flows, and relatively drastic
penalties might be required to accomplish this. We have analyzed the
technical and economic feasibility of the approach, but the political-
administrative procedures for implementation remain to be specified and

-------
61
evaluated. The structural measures may be expensive, but they would be
relatively straightforward to implement.
We also wish to comment on the incidence (that is, which groups
bear the costs and benefits) of water-use efficiency measures to reduce
salinity. The type of water use controls we have hypothesized have their
impacts directly on the water-user's pocketbook. This is in direct
contrast to the structural measures (canal lining, desalination, etc.)
which are slated to be financed from federal appropriations. Our cal-
culations indicate that up to about one-half of the estimated Grand
Valley salt discharge could be avoided by more efficient water use on
the farms themselves. However, the reduced income necessary to accom-
plish this would total over $1 mi 11 ion annually. Although this amount
would not destroy the agricultural economy of the Valley, it represents
a significant part of the return to the land and managerial resources.
It seems, therefore, appropriate to give careful consideration to some
arrangement for cost sharing.
Finally, we have not considered the questions involving use of water
resources that might be released from agriculture as a result of imple-
menting such programs. If alternative beneficial uses exist for "new"
supplies (which incidently, could include bringing new lands under irri-
gation in the Grand Valley [Bureau of Reclamation, 1957]), salinity
control costs may be partially mitigated by new developments or expansion
of the present economy.

-------
SECTION V
THE ECONOMICS OF LAND RETIREMENT AS A MEANS OF
CONTROLLING SALINE IRRIGATION RETURN FLOWS
A. INTRODUCTION
Much of the research effort by physical and social scientists con-
cerned with controlling rising levels of salinity in the Colorado River
Basin has focused on structural technologies. These means typically
require extensive technical and material input often leading to sub-
stantial public and private investments. This chapter reports an
evaluation of land retirement as a salinity abatement option for im-
plementation in the Grand Valley. Land retirement is one nonstruc-
tural control option that deserves careful empirical study since
acreage reduction is the only technically feasible method which can
achieve the zero discharge objective proposed in the 1972 federal
water quality legislation (Public Law 92-500).
To determine whether land retirement can be feasible on economic
efficiency grounds, two sources of information are required: (1) the
direct and indirect costs of removing land from irrigation, and (2) the
benefits or incremental reduction in damages that would occur as a
consequence. Direct costs should accurately reflect the incomes fore-
gone from farming of the retired irrigated lands. Included in the in-
direct costs should be the net effects of costs and benefits issuing
from resource reallocation, social transition, impacts on environmental
amenities, and other consequences in the affected region. Program
62

-------
63
benefits may also involve direct and indirect effects: direct benefits
represent the increment of technological externality removed or penalty
cost avoided, and indirect benefits (or costs) are impacts which "stem
from" the direct effects. Land retirement is said to be economically
feasible only if long run net benefits can be demonstrated, i.e., if
incremental benefits exceed incremental costs over the life of the pro-
gram.
An interindustry (input-output) model serves as the underlying
structure for this analysis. The input-output model is an analytical
accounting technique commonly used in the evaluation of "total"
economic impacts of exogenous (or outside-induced) change in an econ-
omy. Because of the interdependence among industries in a well developed
economy (which may include small or large regions), secondary (or
indirect) impacts are often thought to be just as important as the
primary (or direct) impacts of an induced change. For this reason the
basic approach adopted in this study is indirect impact analysis.
This section is organized into four subsections. In Part B our
approach to the problem and the basic assumptions which underlie our
analyses are discussed in some detail. A discussion of results is
presented in Part C. These results pertain to both local effects
in the Grand Valley area and to the impact on the State's economy.
The last part, Subsection D, is devoted to an overview of the chapter with
emphasis placed on the limitations of the study and some implications
for more definitive research on the topic.

-------
64
B. APPROACH AND ASSUMPTIONS
Land retirement mechanisms might include one or more of a number
of options, and can be either voluntary or involuntary depending on
the level of public acceptance and participation in the program. The
objective is to discontinue irrigation on selected acreage, thus elim-
inating all future salt loading from these sources. Specific program
options evaluated here involve a permanent withdrawal of water supplies.
Compulsory vs. Noncompulsory Programs
Withholding irrigation water from previously cultivated acreage in
the Grand Valley might be accomplished on a voluntary basis by State
purchase of existing water rights from willing sellers [Trelease, I960].
Because of the Grand Valley's arid climate, this would mean that farm-
land is taken out of production altogether, eventually returning to
desert. State purchase of privately-held water rights from legal con-
demnation proceedings would constitute a compulsory mechanism [Gross,
1965]. Both would, in effect, amount to land purchase since desert-
grazing land is of nominal value by comparison.
If a voluntary land retirement program is to be viable, this im-
plies that some farmers in the Valley would be willing to sell their
farms (or a portion of their acreage) at a price that exceeds the
present value of their long-run, capitalized earnings. In the case of
"marginal" farms, this would mean that some individuals would be willing
to trade the present value of farming (in the long run) as a "con-
sumption activity" for an alternative activity made available by the
purchase offer. Under an involuntary program, it would probably be
necessary that the offer price exceed the present market value of
representative irrigated acreage. The program would have an added

-------
65
flexibility if willing sellers under a voluntary retirement scheme
could select the marginal acreages on their farms (perhaps difficult
areas to irrigate where water losses are high and productivity low)
for purchase by the State -- analogous to the soil bank program
[Public Law 89-321]. Under such an option the costs of the program
might be reduced appreciably.
Partial vs. Complete Retirement
Two different program options are examined. The first option
(which we label Option I) represents a partial retirement scheme.
Specific areas of irrigated land which are markedly less productive
(since they tend to have high salt content and/or serious natural
drainage problems that hinder plant growth) are selected for retire-
ment. Chief among the soil groups that fit this description are
those silty clays and clay loams derived residually from Mancos Shale
IKnobel, 1955], Although these soils typically exhibit poor yields
as compared with the remainder of the area, conventional irrigation
practices are normally followed resulting in deep percolation and
seepage losses equivalent to the area averages. Approximately 8,600
acres in the Grand Valley (as well as 10,200 acres in the neighboring
Uncompaghre Valley) fall into this category. Together they represent
15 percent of the area's irrigated lands and 8 percent of the areas'
crop output. Since these areas of relatively unproductive soils are
not contiguous in large blocks, retirement of such lands would control
deep percolation from fields and seepage from farm laterals and head
ditches, but would not account for seepage losses in the main distri-
bution system.

-------
66
A different strategy (Option II) considers the effect of re-
tiring an entire irrigation district. All canals and laterals con-
trolled as an integrated unit and the acreage (both poor and productive)
they service would be withdrawn from production. With this option,
land retirement implies the inclusion of canal and lateral seepage
losses which would be excluded under the first option. Accordingly,
we compute the costs of retirement on the basis of two assumed rates
of annual salt reduction per acre: 5.4 tons under Option I, and
8.2 tons under Option II. A sensitivity of program costs to these
estimates is also reported. The Government Highline Canal, a Bureau
of Reclamation project operated by the Grand Valley Water Users Asso-
ciation and which serves approximately 20,500 acres of irrigated crops,
was chosen to illustrate the impacts of Option II. It is assumed that
this district is representative of the Valley as a whole in terms of
both productivity and salt pickup, so that results could be generalized
to a full retirement program.
Valuation of Land and Water Resources
The selection of a method for valuing irrigated lands to be retired
is dependent upon ownership and use. Capitalizing future earnings of
a marginal farm, situated in an area where the fair market value of
comparable land is well in excess of the capitalized value, is certainly
not a workable approach under a voluntary program. Income capital-
ization is a relatively precise method for valuing a productive resource,
but its applications are limited in some circumstances. As a general
rule market forces have a more direct influence on property values.
Speculative market values reflect alternative uses as well as
present use, and, accordingly, market values tend to run higher than

-------
67
capitalized values for the same resource. But in cases where land
has only one use (its present use), capitalized and fair market values
converge. This phenomonen is usually observed in traditional agri-
cultural areas such as the midwest, and is likely true in the Grand
Valley. Larger, more efficient farms typically establish land prices
through marginal additions to land holdings thereby setting a general
price level which prevails for other farmers in the immediate area
[Tweeten, 1971]. For this reason the income approach was used in this
analysis to approximate the market value for land and water resources.
Income losses from a land retirement scheme can be measured by
the foregone net income from land and immobile improvements (mainly
the irrigation distribution systems). The estimated income residual
allocated to these resources will vary with farm size and soil pro-
ductivity, among other factors. Our procedure for determining the
residual return to land and immobile improvements is based on tra-
ditional crop share rents in the study area. The landowner typically
receives one-third of the gross crop sales, but is responsible for a
similar proportion of fertilizer expenses as well as for all real
estate taxes and water charges. Gross sales on the low productivity
lands (Option I) were estimated at $177 per acre (1975 price levels),
while the estimated average of all lands was $349 per acre [Leathers,
1976b]. Computing the landowner's share (one-third) and deducting fer-
tilizer, tax and water charges yield annual net returns foregone of
$39 for Option I lands and $90 for Option II. Capitalized at 6 percent
over an infinite time horizon these rates of return translate to present
values of $650 and $1,500 per acre respectively. The current asking
price of productive irrigated land in the Grand Valley ranges upward
from $1,500 per acre [Mesa County Assessor's Office, 1976].

-------
68
The market value for less productive acreage is not known. At these
prices, the costs of retiring cropland per ton of salts avoided, in
annual terms, would be $7.22 under Option I ($39 r 5.4 tons per acre)
and $10.98 under Option II ($90 f 8.2 tons per acre).
Analysis of Direct and Indirect Impacts
We estimate the direct and indirect impacts of the two options by
detailed analysis of the community (small region) economic structure
of the Grand Valley Trade Area (GVTA). A revised (scaled-down) input-
output model of the upper mainstem subregion of the Colorado River
Basin, reported by Udis and others [1973], was used to characterize
economic interactions and levels of income and production for the three-
county area of Delta, Montrose, and Mesa, which together define the West-
Central Colorado GVTA (Figure 7). The base year of the model is 1970.
A total of seven agricultural production sectors and 12 non-agricultural
industries form the endogenous processing portion of this model. Local
governments and households were also considered endogenous. State and
federal governments, inventories, capital investment, and import-export
accounts comprise the final demands or exogenous sectors. Sector defi-
nitions and classifications and an interpretation of the model are
reported by Leathers [1976b]. The conventional input-output tables —
interindustry transactions, technical production coefficients, and direct
and indirect coefficients — are reproduced in the appendix of this
report, tables 6, 7, and 8.

-------
COLORADO
(Q
C
S
a>
CD
cn
s
0>
3
a.
o>
<<
-j
ai
Q_
a>
5»
~S
to
a>
Grand Valley Trade Area
Irrigated Farming Land
CT>

-------
70
A number of modifications in conventional interindustry analysis
techniques were necessary to satisfy the purposes of our study. First,
given a small region (or community) orientation, our results consider
only community impacts "as if" interregional (non-local) impacts, which
result from interactions with the GVTA, didn't matter. This normally
restrictive assumption is not a severe one here since trade flows be-
tween the GVTA and the rest of the state and regional economy are rel-
atively insignificant for the field crop commodities in question.
Second, we define the household sector as after-tax return to
locally invested capital, resident wages and profits. Accordingly,
the adverse effects of crop acreage reductions are estimated internally
"net" of state and federal taxes. Thus, our measure of income is, in
effect, community value added rather than the more conventional form
which measures before-tax (Keynesian) income effects.
Third, we omit the effects of forward linkages in estimating the
direct and indirect impacts. The fed livestock and food processing
sectors in the GVTA do not rely wholly upon locally grown produce;
significant amounts of feed grains are imported by local feeders and
grain handlers while some locally grown grains and forage are exported
to other regions. Hence, sufficient imports to the GVTA to maintain
current levels of output in the forward-linked industries is assumed.
The sectorial income multipliers presented here are adjusted for back-
ward linkages only, and relate to per dollar changes in output rather
than per dollar changes in final demand (as conceptualized by Martin
and Carter [1962 ]). We use the "adjusted" income multipliers to est-
imate adverse effects since it is the effect of a reduction in production

-------
71
on community income we wish to measure rather than the effect of a
reduction in final demand. The scaled-down multipliers, which ranged
from 90 to 99 percent of their unadjusted values, measure the change
in regional wages, profits and returns to local capital per dollar
change in crop revenue.
An interindustry model for the State of Colorado, recently reported
by Gray, ^t [1975], was used as a means of assessing the implications
of land retirement in the Grand Valley on the State's economy. Forward
linkages were also omitted in our use of this model to perform the in-
direct impact analysis.
Accounting Stance
We consider each land retirement option under each of two assump-
tions about accounting stance and resource mobility. This procedure
permits the establishment of reasonable upper and lower bounds on the
social costs of the program. An upper bound estimate of land retire-
ment cost is obtained by computing the direct and indirect income effects
via a regional accounting stance. Under this assumption capital, labor
and entrepreneurial resources in both the agricultural production sectors
and the related processing sectors are assumed to have only a limited
opportunity for reemployment in productive activity elsewhere in the
region. This limited resource mobility is reflected in an assumption
that only 30 percent of such displaced persons and capital resources
will be reemployed within a reasonably short time. This approach fol-
lows that of Kelso, et al [1974].
A national accounting stance and an assumption of instant and cost-
less mobility of non-land resources provides a lower bound estimate on

-------
72
land retirement costs. This perspective implies that all labor and
mobile capital resources employed in irrigated crop production (plus
those affected in forward or backward linked sectors) can immediately
be reemployed elsewhere in the national economy without loss of pro-
ductivity. In essence, a national accounting perspective does not
view secondary effects as legitimate economic consequences unless a
full employment condition exists in the national economy.
C. RESULTS AND DISCUSSION
The findings of our analysis are reported in two subsections which
follow: the first is devoted to a discussion of the local impact of re-
tiring irrigated land in the GVTA, and the second considers the effect
of reduced crop production on the Colorado economy.
Local Direct and Indirect Costs
The adverse effects of reduced crop production in the GVTA are sum-
marized in Table 11. Annual direct and indirect costs, measured in terms
of reduced community income, are estimated for the retirement Options I
and II under a regional accounting stance. These costs are derived from
sectorial income multipliers (shown in Appendix Tables 7 and 8) which are
net of forward linkages. In this adjusted form, the multipliers measure
direct, indirect (and induced) changes in household income per dollar
change in output of the impacted crop sectors.
The forage and feed crops sector includes corn grain and silage, perma-
nent pasture and alfalfa; sugar beets and small grains are handled in
the food and field crop sector; and orchard and vegetable enterprises are
represented in the fruit and specialty crop sector of the model. Since

-------
TABLE 11. COMMUNITY INCOME REDUCTION UNDER TWO OPTIONS FOR RETIRING IRRIGATED LAND IN THE GRAND VALLEY TRADE AREA
Gross Sectorial Income Community Income-Reduction
Impacted	Total Income . 	Multipliers—''	 	in $1000^-'	
Sectors
Acreage
Per Acre-'
Direct Indirect Total
Direct
Indirect
Total
Nonrecoverable
OPTION I:
RETIREMENT OF SELECTED LOW PRODUCTIVITY LANDS




Forage and Feed Crops
6,050
$158.40
.4020 .4530 .8550
$385
434
819
574
Food and Field Crops
2,550
223.60
.5185 .4273 .9458
296
243
539
377
Totals
8,600
($177 ave.)

681
677
1,358
951
Annual Nonrecoverable Income
Loss Per Acre: $110.23





OPTION II:
RETIREMENT OF ONE COMPLETE
IRRIGATION SYSTEM (AVERAGE PRODUCTIVITY)


Forage and Feed Crops
15,720
$275.40
.4020 .4530 .8550
$1,740
1,961
3,701
2,591
Food and Field Crops
4,510
537.96
.5185 .4273 .9458
1,258
1,037
2,295
1,606
Fruit and Specialty Crops
270
1,292.00
.3234 .5176 .8410
113
181
294
205
Totals
20,500
($349. ave.)

$3,111
3,179
6,290
4,402
Annual Nonrecoverable Income
Loss Per Acre: $214.77





a/ Sector gross Income per acre (valued at 1975 prices) reflects a weighting of Individual crops according to
their proportion of the total acreage Included In the sector aggregation.
b/ These multipliers, net of forward linkages, measure the change In household income (returns to local capital
Investment, wages and profits) per dollar of change In sector output (in this case gross revenue per acre).
c/ Column entries are foundby multiplying the product of sector acreage and weighted gross income by the ap-
propriate multiplier. Nonrecoverable income loss 1s total reduced community income less a recoverability factor of
30 percent.

-------
74
orchard and other high value crops are rarely grown on poorer soils, the
fruit and specialty crop sector is not considered under the first option.
Assuming a recoverability of displaced resources at 30 percent, a cut-
back in production involving 8,600 acres of poor quality soils in the
Grand Valley would generate a net loss in community income of $951,000
annually. Under Option II, which retires 20,500 acres of average produc-
tivity, the loss is $4,402,000. On an annual per acre basis these income
losses are approximately $110 and $215, respectively.
These costs are compared with our lower bound estimates (based on a
national accounting stance) in terms of total program costs (part A) and
cost effectiveness (part B) in Table 12 . Total program costs reflect re-
duced income plus an additional charge for implementation and adminis-
tration of the program. We estimate these charges at 10 percent of
annual income losses. Since the amount of salts avoided by retiring
selected irrigated croplands is not known with certainty, we "bracket"
the provisional estimates for salt pickup per acre to demonstrate the
sensitivity of this parameter to the cost effectiveness of the program.
In general, the incremental costs of salt removal (in $ per ton),
using the provisional estimates of salt pickup, appear to be competitive
with other more expensive controls such as canal lining. However, the
cost effectiveness of the program is quite sensitive to assumptions re-
garding estimates of salts removed and accounting stance. Accordingly,
it is important that these assumptions are considered very carefully in
comparing alternative salinity control programs. For example, the cost
of partial retirement (Option I lands) vary from $6.13 per ton to $14.30
per ton depending on the assumed rate of salt pickup per acre, and more

-------
75
TABLE 12. SUMMARY OF ANNUAL REGIONAL COSTS AND COST EFFECTIVENESS: UPPER
AND LOWER BOUND ESTIMATES FOR THE TWO LAND RETIREMENT OPTIONS
Measures	Estimated Annual Costs (1975 Dollars)
and	Lower Bound-'' Upper Bound—'f
Options	Per Acre Total Per Acre Total
A. PROGRAM COSTS-/
Partial Land Retirement, Option I:
Income Reduction	39.00 335,400 110.23 951,000
Total Program Costs	42.90 368,940 121.25 1,046,100
Complete Land Retirement, Option II:
Income Reduction	90.00 1,845,000 214.77 4,402,000
Total Program Costs	99.00 2,029,500 236.25 4,842,200
B. COST EFFECTIVENESS—^
Salts Removed in
Dollars Per Tons
Tons Per Acre	Lower Bound	Upper Bound
Partial Land Retirement, Option I:
3	14.30	40.42
5.4	7.94	22.45
7	6.13	17.32
Complete Land Retirement,	Option II:
5	19.80	47.25
8.2	12.07	28.81
11	9.00	21.48
a/ Assumes a national accounting stance and perfect mobility of dis-
placed resources; hence the costs reflect compensation payments to par-
ticipating farmers as determined by income capitalization of long run
earnings to irrigated farming.
b/ Assumes a regional (or local) accounting stance, 30 percent re-
coverability of displaced resources, and the absence of forward linkages.
Costs are expressed in terms of reduced community income in response to
the direct, indirect and induced effects of reduced production.
cj Total program costs reflect additional charges for implementation
and administration which are assumed at 10 percent of reduced income.
d/ Estimated annual costs per acre divided by the indicated rates
of salts removed in tons per acre.

-------
76
than double if a regional accounting stance is assumed. We believe the
regional accounting stance, however, provides a fairly generous upper
bound on total program costs.
Implications for the State of Colorado
The impact of reduced production of irrigated crops in the Grand
Valley on the State's economy was evaluated with the use of the State
input-output developed by Gray and others [1975]. This model is also
constructed on a 1970 data base. As with the GVTA model, the income
multiplier was adjusted for forward linkage effects. Further, we as-
sume 30 percent recoverability of persons and resources displaced by
the program, and a 10 percent additional cost to account for the expense
of implementation and administration. Given these procedures, our
estimates of State impacts are directly comparable with the upper bound
"local" impacts reported in Table 12.
The income reduction effect of retiring cropland in the Grand Valley
(Options I and II) is apparently not as severe on the State economy as
it might be for the economy of the small region. The reason for this
is attributed to the difference in size of the adjusted income multi-
pliers. Referring again to Table 11, the GVTA sectorial multipliers
for the impacted crop sectors were shown to have a magnitude of about
.8 or .9. Under similar assumptions, the State multiplier for irrigated
agriculture (in which case all irrigated crops are lumped together) is
.4304 or about half that of the small region multipliers. Accordingly,
the program costs to the State would be about half the upper bound costs
reported in part A, Table 12.
The smaller multiplier effect at the State level is attributed to

-------
77
some of the differences observed between the two regions. Irrigated
agriculture plays a more prominant role in the economic structure of the
GVTA as compared to the State economy. Also, because orchard enter-
prises and small farms tend to be more labor intensive, smaller farms
and specialty crop enterprises common to the GVTA are not typical of the
State as a whole. However, a careful study of this particular aspect
of the problem was considered to be outside the scope of our analysis.
A detailed analysis of State consequences would probably require less
aggregation of the agricultural sectors, but more importantly, would
involve a careful specification of alternative cost-sharing schemes
between the State, the impacted area, and the federal government.
D. SUMMARY OF COST-EFFECTIVENESS
The purpose of this chapter has been to estimate the economic
consequences and cost effectiveness of land retirement as a means of
reducing irrigation-induced salinity in the Grand Valley. Land re-
tirement is obviously a drastic salinity control measure. Nonetheless,
a cut-back in irrigated farming in many areas of the Upper Colorado
River Basin is the only sure way that significant amounts of salts can
be removed from the river system. As the Upper Basin States continue
to develop their respective rights to Colorado river water, hence un-
avoidably add to the rivers' salt load, each State is responsible for
enacting procedures for removal of salts in order to insure that the
1972 water quality standards (which apply at the State boundaries) are
not exceeded [U. S. Environmental Protection Agency, 1975]. In effect
this trade-off between present and future uses may well involve land
retirement schemes if alternative control measures prove ineffectual
or become relatively more expensive in the future.

-------
78
Table 13. ESTIMATED INCREMENTAL COSTS OF SALT REMOVAL FOR PROPOSED
ALTERNATIVE CONTROL MEASURES, GRAND VALLEY, COLORADO (1975 dollars)
RANGE IN ESTIMATES
Control Measure
Annual Salt Discharge
Avoided
(tons)
Average
Cost
($ per ton)
Modified Irrigation Practices—^
Aluminum Pipes for Orr-Farm . ,
Conveyance and Irrigation—
c/
Lining of Main Canals-
and Laterals
Drainage Modifications
Desalting Saline Ground Waters—^
Partial Land Retirement, Option 1-^
Lower Bound
Upper Bound
Complete Land Retirement, Option 11-^
Lower Bound
Upper Bound
50,000 - 100,000
75,000 - 150,000
200,000
3.42 - 1.71
14.15 - 7.07
32.72
unknown
25,800 - 60,200
25,800 - 60,200
102,500 - 225,500
102,500 - 225,500
14.30
40.42
6.13
17.32
19.80 - 9.00
47.25 - 21.48
a/ Based on an annual water delivery of 215,000 acre feet, Table 9
(section IV).
bj Refer to table 10, section IV.
c/ Adapted from [U. S. Bureau of Reclamation, 1976]
d/ Adapted from [Walker, 1976]
e/ Refer to table 12.

-------
79
With the aid of a small-region input-output model of the Grand
Valley Trade Area (GVTA), local community impacts were estimated for
two types of retirement programs. The options considered were: (1)
withdrawal of water supplies from 8,600 acres of less-productive soils
in the Valley — a "partial" retirement scheme; and (2) the phasing out
of an entire irrigation district of about 20,500 acres of average pro-
ductivity soils -- a "complete" retirement scheme. Under both Options
withdrawal of irrigation water means termination of all crop production
on these acreages. This would also hold true for much of the irrigated
lands in the Upper Basin outside Colorado. A second input-output model,
representing the economy of Colorado, was used to estimate the probable
impacts on the State of reduced crop production in the Grand Valley.
The results of our analysis, given a number of assumptions which
provide upper and lower bounds on our estimates, show land retirement
to be competitive with some of the more expensive structural controls
under consideration in the Grand Valley and elsewhere (Table 13). These
include desalting, drainage improvements and canal lining, the costs of
which appear to be above $30 per ton of salts removed. Depending upon
the nature of assumptions used to derive the cost effectiveness of land
retirement, we have shown that the incremental cost vary over a wide
range — from less then $7 per ton to over $40 per ton. However, owing
to the way in which the upper bound estimates are calculated, we feel
they represent a generous approximation of the full economic costs of
the program.
In closing, we wish to stress the limitations of our analysis,
particularly with regard to the precision of our knowledge of the
sources and mechanisms of salt pickup in the Grand Valley and of those
"social" costs of the program which cannot be evaluated through the market

-------
mechanism. Further, we have not considered monitoring and enforcement
costs directly, except to the extent that our estimates were inflated
by 10 percent to approximate the overall costs of administering the pro-
gram. This assumption may prove to be an oversimplification of a very
difficult and costly problem for the enforcing public agency.
Although conceptual disagreement concerning measure of downstream
benefits (damages avoided) are not resolved, it is our judgment that such
benefits are not large enough to warrant undertaking of any of the proposals
other than adjustment in on-farm water management. We are dubious about
even this alternative because of uncertainties involved in measuring the
hydro-salinity relationships. Where the market system is being by-passed
to serve some presumably over-riding public interest, it is desirable to
include in the public programs some incentives to help assure an effi-
cient allocation of resources. In this case, we would suggest that at
least a half share of the burden of improving water quality be borne
by the water user groups in the Lower Basin. If such a plan were adopted
we would be more certain of that region's real need for reduced salinity.

-------
SECTION VI
RETURN FLOW SALINITY MECHANISMS:
A REAPPRAISAL OF ESTIMATES AND ASSUMPTIONS
A. INTRODUCTION
In the preceding sections, which dealt with the costs of alternative
salinity control programs, it was shown that cost-effectiveness (the rel-
ative marginal costs) is very sensitive to assumptions about the quantity
of salts removed by such programs. This is most notably demonstrated by
the fact that structural and land retirement measures are "borderline",
even under generous assumptions, in terms of national (or regional) ef-
ficiency benefits. Since these programs can only be implemented at very
high costs to taxpayers, it seems useful at this point to reconsider
some of the assumptions upon which the proposed programs are predicated;
and, for the most part, this involves reevaluating the "physical" rather
than the economic aspects of the problem.
In this section, the fundamental propositions and empirical evidence
which supports the need for public intervention in the Grand Valley for
the purpose of controlling salinity are questioned. Beginning with the
early investigations in the Valley, the results of significant "macro"
and "micro-level" efforts are compared and their implications discussed
(Part B). An alternative research methodology, formulated in Part C,
proposes that least-squares regression analysis of historical time series
data is an appropriate approach with which to test the competing hypo-
theses, and to suggest areas of research which require more critical
examination. The results of the regression analyses are reported in
Part D, and the implications and limitations of the study are sum-
maried in Part E.
81

-------
82
B. REVIEW OF PREVIOUS RESEARCH
It is well known that Increased water salinity (dissolved solids)
is attributable to both, natural and manmade causes, but the extent to
which each contributes to the total salinity load in a particular sit-
uation is often difficult to discern. This difficulty arises because
of the complex, variable nature of many different processes which occur
simultaneously within the hydro-geologic-salinity system. Unfortunately
only the manmade causes are amenable to control in many actual situations.
Therefore, a valid economic assessment of alternative control measures
for agricultural and other manmade sources depends in large part upon
our ability to separate them from natural processes.
The first comprehensive study of this problem was conducted during
the mid sixties [Iorns et al_., 1965]. In that study detailed analyses
of separate subbasins within the Colorado River Basin were made to iden-
tify the sources of salt loading and to determine the relative magnitude
of each contribution. Input-output budgeting techniques were used to
estimate average monthly changes in flows and salt loads covering a
period from 1914 to 1957. A second study, using a similar methodology
but more comprehensive in scope, was concluded shortly thereafter [En-
vironmental Protection Agency, 1971J - This study was based on monthly
Input-output data collected during a one-year period, June 1965 to May
1966. Because of limited time and resources, most of the agency's
monitoring effort was concentrated in the upper basin. Further, no
attempt was made to distinguish between salt "loading" and "concentrating"
effects,
The results of the EPA study strongly implicate irrigated agriculture
as a major manmade contributor to the River's salinity. The EPA report

-------
83
concluded that salt loading as a consequence of irrigation accounted for
37 percent of the total salt load in the upper basin. Natural runoff
was the largest source, accounting for 52 percent of the total load, while
natural point sources (.highly mineralized springs) and municipal-industrial
sources explained the balance (9 and 2 percent, respectively). The Grand
Valley area and the Gunnison River subbasin, both in Western Colorado,
were found to be particularly significant agricultural sources representing
50 percent of the upper basin's agricultural contribution.
In the Grand Valley reach, an area extending from Cameo to below
Fruita (see figure 8), the average annual salt pickup was estimated at
700,000 tons. Of this, EPA estimated that 93 percent was the consequence
of irrigation return flow. Less than 5 percent of the total was attri-
buted to natural runoff. The conclusions of Iorns, et al_. [1965] were
substantially the same with regard to proportion from irrigation, al-
though the average pickup in the reach was estimated at about 500,000
tons. Because of the broad orientation of both studies, no detailed an-
alyses of specific pickup mechanisms for either agricultural or natural
runoff contributions were attempted.
Agricultural Pickup Mechanisms
Return flows from an irrigated area, defined as that portion of
water diverted for irrigation which eventually returns to the stream,
tend to increase the total dissolved solids load carried by that stream.
Often included in the dissolved solids load are quantities of nitrates,
phosphates and nondegradable chemicals which are used on the irrigated
lands. Since conventional treatment methods are normally ineffectual in
removing these constituents from return flows, agricultural use may im-
pose serious problems for other downstream users.

-------
Precipitotion Only
H Precip 8 Temp
4) Temperature Only
(v Prectp , Temp , S Evaporetitft
w Stream Flow
^ Water Quality 8 Flow
0 Water Quolity
Irrigated Lond
Source: Hyattj et. il [1970]
Q Badger Wash Area
CO
Figure 8. Grand Valley Reach of the Colorado River

-------
85
The quantity of water that ultimately returns to the stream system
is influenced by a number of factors; canal and lateral seepage, soil
water holding capacity and infiltration characteristics, precipitation,
cropping pattern evapotranspiration rates, and irrigation practices.
The quality of return flows are also affected by many of these same
factors. In general, the quality of return flows depend upon the time
and the extent to which the water is in contact with salt bearing sur-
faces. Waters which receive minimal exposure to soil surfaces will
pick up comparatively less salts than waters exposed to large soil sur-
face areas for long periods of time. Similarily, waters which penetrate
below the root zone of growing crops have greater opportunity to inter-
face ground waters below irrigated lands which frequently contain high
dissolved mineral concentrations. This problem is particularly acute
in the Grand Valley, and is believed to be fairly common elsewhere in
the upper basin [EPA, 1971]. These phenomena, coupled with the concen-
trating effect of evapotranspiration, can produce concentrations of
salts in the soil solution up to 10 times that of the irrigation water
at the point of diversion [Bishop and Peterson, 1969].
Under normal circumstances if irrigation diversions are high in
relation to evapotranspiration requirements (a condition which typi-
cally exists early and late in the growing season), rates of return
flow are high, and salinity concentration in the subsurface components
of return flow (deep percolation and seepage) are usually low. This
is largely due to the high proportion of surface runoff in the total re-
turn flow and the relatively short period of time surface waters are in
contact with soils. On the other hand as evapotranspiration require-
ments rise relative to stream diversions (normally mid-summer conditions),

-------
86
salinity concentrations in return flow tend to increase. Most authorities
believe, however, that the total rate of salt outflow from an irrigated
system is greater under higher irrigation applications than for low ap-
plication rates £Hyatt, 1970].
Skogerboe and others [1974] have conducted extensive investigations
of return flow salinity in the Grand Valley as a followup to the earlier
studies. They applied a similar input-output methodology, but incorpo-
rated detailed submodels of specific pickup mechanisms emphasizing the
components of return flow waters. Using monitoring instruments to mea-
sure surface and groundwater flows and salinity concentrations at a field
site east of Grand Junction, they were able to calibrate a hydrosalinity
budgeting model which they compared to the total Valley input-output data
for the 1968 water year. Their results indicate that seepage losses in
main canals and laterals can account for about 55 percent of the total
salt picked up in the Grand Valley reach. The balance (45 percent) was
attributed to deep percolation losses that occur as a result of ineffi-
cient irrigation practices. Surface return flows (field tail water and
canal spillage) were not found to be significant sources, and natural
runoff was not accounted for in their model.
The mechanism by which subsurface return flow waters (deep perco-
lation and seepage losses) pick up salt is thought to be explained by a
displacement process. Deep percolating waters recharge the aquifers
which underlie protions of the irrigated lands, and the hydrostatic head
created displaces some saline ground water (as high as 10,000 ppm) out
of the aquifer and into the river £Skogerboe and Walker, 1972]. As a
means to control salt pickup in the Valley these researchers have studied
improved irrigation practices, canal and lateral lining, and drainage

-------
87
systems — all designed to reduce or remove deep percolating waters
before coming in contact with the water table.
The results of experiments conducted by the Agricultural Research
Service as a site west of Grand Junction are not consistent with those
of Skogerboe and others [Kruse, 1975]. These researchers have found that
conveyance system seepage losses alone could account for nearly all of
the salt pickup in the study reach. Kruse and his associates found
relatively impermeable soils and low infiltration rates. Their study of
irrigation practices therefore concludes that deep percolation losses
from irrigation are apparently quite small, hence this mechanism of
salt pickup may not be as important as earlier hypothesized. Accordingly,
the control measure most likely to be effective would be canal lining,
since measures to improve on-farm irrigation efficiency would probably
not yield significant reductions in salinity.
Natural Pickup Mechanisms
Water quality of the upper Colroado River Basin is profoundly in-
fluenced by the geology of that area. Large portions of the area are
underlain by the Mancos shale of Late Cretaceous age and by other ancient
formations with similar physical and chemical properties [Iorns et al.,
1964, 65], Distinctive landforms have resulted from erosion of these
formations, and as a result rock and residual materials have been carried
away leaving an extensive network of natural washes, plateaus, steep
canyons, and sparse vegetation. The contact of water with the saline geo-
logic formations, either from natural precipitation or from irrigation,
causes a serious degredation of water quality in streams receiving runoff
or drainage waters.

-------
88
In the higher reaches of the watershed where much of the basin's
water supply originates, a continuous interchange between surface waters
(snow melt and overland runoff) and groundwaters occurs. During this
process, rock and residual materials react with the water and impart to
it chemical constitutents which are characteristic of the watershed geol-
ogy. When waters of differing constituents are mixed chemcial reactions
normally take place. These reactions can cause either a loss in con-
stituents via precipitation or an increase as new salts are dissolved.
Similarily, when an aquifer receives a direct recharge, the outflow of
the aquifer is indicative of the chemical makeup of the alluvium through
which the water has passed.
Within the subbasins of the upper Colorado River Basin influent
phenomenon typically exist at the higher elevations while effluent flows
predominate in the lower reaches of most tributaries [Iorns, et aj[., 1964].
An influent stream is one which adds to a ground water system whereas an
effluent one intersects the water table and allows groundwaters to surface.
Most perennial streams are effluent over a portion of their length, and
the existance of both conditions in a single reach is common [Linsley,
et al., 1958].
Waters passing through alluvium as influent flow eventually return
to the mainstream channel. Within a particular subbasin the amount of
interchange between surface and groundwaters may be influenced by water
levels in the channel itself. This specific mechanism, namely the "inter-
change hypothesis" as Hyatt has chosen to call it, has not received care-
ful study in the Grand Valley subhastn, but may be a significant salt pick-
up mechanism iHyatt et^aj_., 1970J.

-------
89
During periods of high stream flow (which normally occur during the
runoff season in late spring, but may also occur after severe thunder
showers in the late summer) some increase in interchange waters would be
expected. Since dissolved solids content of groundwater is usually
higher than that of surface waters, a mixing of the two will generally
cause an increase in the salt load of surface waters. The quality of
surface waters in the upper Colorado River Basin exhibit (over a period
of years) wide variations, and because groundwaters of this area tend to
be quite salty, it seems plausible the interchange hypothesis can account
for significant increases in salt loads at the outlet of a subbasin.
Studies by the U. S. Bureau of Reclamation [1967] indicate that along
certain reaches of the Colorado River there are large increases in the
salt load which cannot be attributable to the agricultural system.
Increased salt loads can be expected to be particularly significant
in mildly sloping valleys underlain by large, permeable alluvia. The
fact that these same features also characterize irrigated agricultural
areas apparently has fostered the belief that any incremental salt loads
added to the stream which traverses such valleys are largely the result
of irrigation (e.g., "for the Grand Valley area the unbudgeted increase
in dissolved-solids load...at least 440,000 tons per year...may be at-
tributable to irrigation." (Iorns et al_., [1964 ] ). Because it is a
natural phenomenon, Hyatt suggests that the interchange hypothesis ex-
plains salinity increases which occur even during periods of the year
when return flows from agricultural lands are insignificant.

-------
90
Hyatt and others [1970] constructed an analog simulation model of
the upper basin to test a number of alternative hypothesis about natural
and manmade salt pickup mechanisms. Essentially the same watershed areas
were studied as described in the EPA and Iorns reports. Specific pickup
mechanisms were analyzed on the basis of historical hydrosalinity re-
lationships estimated from available secondary data on stream discharge,
diversions, climatic conditions, and water quality. With few exceptions
the results indicated significant interchange processes, especially for
subbasins with irrigated lands. The computed mass rate of salt flow
leaving each subbasin, before interchange was introduced, was found to
be much less than the measured rate of salt outflow, often from one-half
to one-tenth of the actual values recorded at U.S.G.S. gaging stations.
The implication of their findings is that the earlier studies over-
estimated agricultural contributions and underestimated natural ones.
For the Grand Valley area and the Gunnison river subbasin, they estimate
that salt loading by natural processes (including interchange) possibly
exceeds agricultural sources in magnitude of salt pickup.
Most authorities have discounted the possibility that surface run-
off from snowmelt and rainfall is an important salt pickup mechanism in
the Grand Valley reach. Consequently, surface runoff mechanisms in the
area have received little study, and their contribution to salt loading
is unclear. At a research stte (Badger Wash) located northwest of
Grand Junction, researchers with the U. S. Geological Survey have studied
salt loading from surface runoff as part of a more general investigation
of erosion processes [Schumm and Lusbey, 1963]. Schumm found that season-
al changes in soil characteristics (the lithosols) are great enough to
affect both the rates of erosion and the erosion process itself. Based

-------
91
upon analyses of seven years of storm data, runoff was much greater
in late summer and fall for storms of comparable inetnsity, and the
average runoff and the ratio of runoff to precipitation were consis-
tently larger for all sizes of storms during this period. Unfortunately
the amounts of dissolved solids carried away by runoff water was not
carefully documented in these early experiments. But as a result of
these findings, it is reasonable to hypothesize that the majority of
salt loading which occurs from surface runoff likely takes place during
the late summer and fall months of the year.
Hadley [1976], reporting on more recent results from studies at
Badger Wash, estimates natural runoff at 2 to 3 times that reported by
the Environmental Protection Agency [1971] and Iorns et^ al_. [1964] for
this reach of the river. Soil Conservation Service estimates, based on
recent analyses of sediment samples taken above the irrigated lands at
the base of the Bookcliffs (see figure 1) and farther west, concur with
Hadley's conclusions [Elkin, 1976] . Elkin estimates the average an-
nual salt yield at 80,500 tons for a runoff area of approximately 1,370
square miles, or 59 tons per square mile of surface area compared to
18 tons by EPA estimates. Clearly, these divergent conclusions warrant
further study of this aspect of the problem.
Implications for Policy Decisions
In most scientific disciplines the results of research tend to raise
more questions than answers. The divergent opinions which surround the
question of how to control salinity in the Grand Valley probably result
from the complex nature of the problem rather than from faulty analyses.
However, it may be useful to summarize the methodological procedures dis-
cussed abcvein terms of an alternative approach, and to consider briefly

-------
92
some of the policy implications which obtain.
The simulation approach, which has been used extensively in the
Grand Valley and elsewhere in the Colorado River Basin, is essentially
a detailed input-output budgeting procedure. Structural coefficients
within these models represent an aggregation of the many separate pro-
cesses which occur in the hydro-geologic-salinity system and are typi-
cally estimated from historical data when it is available. The estimated
models are then "calibrated" to a base year or other reference period.
Calibration involves comparing the model outputs with the reference
period data and adjusting the parameters until a suitable correlation
is obtained. Once calibrated the model is said to be representative of
the physical system being studied, i.e., for the reference period. In
the case of hydro-salinity systems the simulation approach should be re-
garded as "partial" analysis since it is normally used to represent the
base year or reference period and not the period of years which may com-
prise the data base. Thus, the appropriateness of the model for de-
scribing salt pickup mechanisms and magnitudes of salts added to a water
course is uniquely tied to the reference year. Given cyclic and/or
stochastic phenomena, a reference year selected at random may be quite
unrepresentative as a predictor of historical or future trends. Perhaps
this explains some of the discrepancies in research findings.
Least-squares regression is an alternative analytical technique to
the simulation (budgeting) approach. Regression techniques, sometimes
used for estimating the structural parameters of simulation models, are
specifically designed to perform statistical tests on the functional
relationships among specified variables. An advantage of regression
analysis is that time-trend data (including stochastic processes)can be

-------
93
carefully evaluated. In most instances this is not done with conventional
input-output budgeting techniques. In fact, this study is believed to
be the first attempt to use regression methods and time series analyses
to describe salt pickup mechanisms in the Colorado River Basin.
The divergent research findings reported by many of the state
and federal agencies and private commissions having become involved
with the salinity problem pose an interesting dilemma for policy makers.
On the one hand there is a sense that action on the problem is urgent,
while on the other hand, they face a high degree of uncertainty as to cost
effectiveness in choosing among different control measures.
C. METHODS OF ANALYSIS AND DATA SELECTION
Many important problems in the social, biological and physical
sciences involve measurements in time. In fact, most information we have
about the real world only exists in the form of time series records of
events or "variables". Controlled experiments involving the study of
these variables are expensive, time-consuming, and in some cases, im-
possible. For this reason, the place of statistical analysis in scien-
tific research is well established.
Statistics provide a means of measuring the elements involved in
relationships and in examining the relationships themselves, but it does
not itself furnish an explanation of the phenomenon under examination.
The effort to form hypotheses as a mathematical model of actual processes,
and to reduce the variables under study to succinct numerical statements,
forces the investigator to approach his problem more clearly and logically.
However, statistical analysis is not a substitute for careful thinking;.
it is, as best stated by Ezekiel [1959], "	an aid which may make that
thought and skill even more productive of worthwhile results".

-------
94
The Techniques of Correlation and Regression
In any system in which variable quantities change, it is often neces-
sary to examine the effects that some variables exert on others in order
to gain insights into the functioning of the system. This is especially
important in the study of natural systems where actual functional relation-
ships are too complicated to understand or to describe in complete detail.
The statistical techniques of correlation and regression are useful
methods whereby complicated natural processes can be reduced to simple
mathematical equations based on the testing of hypothesized relationships
between appropriate variables [Draper and Smith, 1966],
Correlation and regression analyses are a means to understanding
the nature of association between two (or more)variables; correlation
is a measure of linear relation and indicates whether two variables are
positively or negatively related, whereas regression is applied to spe-
cific functional forms among two or more variables that can be either
linear or nonlinear. Although neither technique claims to identify cause
and effect relationships, in regression analysis hypotheses are set up
to confirm whether or not variable X has an "influence" on variable Y.
In multiple regression, the influence of more than one independent vari-
able (X's) on the dependent variable (Y) can be jointly evaluated.
Important test statistics in regression analysis relate to (1) the sig-
nificance of the estimated equation parameters (regression coefficients),
which facilitate tests of the strength of the hypothesized associations;
and (2) the significance of regression equations themselves (the pro-
portion of total variation in the dependent variable associated with the
variation in the independent variable(s) in a specified functional re-
lationship) .

-------
95
Data Sources and Evaluation
The data base for this analysis was compiled largely from secondary
sources. Information selected for careful study was limited to the natural
runoff and irrigation sources of salinity; industrial and municipal con-
tributions and natural point sources were not examined independently due to
time and resource limitations. A time series of monthly data, covering a
22-year period (1951 through 1972) was obtained on stream flows, total
dissolved solids concentrations, the disposition of waters diverted for
irrigation, cropping patterns, precipitation and evapotranspiration in the
study area. Although much longer records are available for some of these
data, the period of study chosen represents a consistent set of actual ob-
servations, thus avoiding the necessity for extrapolation or use of syn-
thetic data.
Stream Flow and Salinity
The only permanent, systematic record of water flow and quality
available for the reach is maintained by the U. S. Geological Survey [1973].
Measurements of surface flow are recorded on a daily basis at each perm-
nent gaging station whereas complete water quality analyses are typically
performed two or three times each month. The water quality analysis in-
cludes measurement of specific conductance, temperature, turbidity, alka-
linity, and differentiates specific ionic concentrations or nonconservative
constituents. The water quality parameter selected as the dependent
variable in this study was total dissolved solids (TDS) in tons per unit
time.
The salt load carried by inflow and outflow waters was estimated by
multiplying TDS concentrations (tons per acre foot) and flows (acre feet).
Tons per acre foot indicates the dry weight of dissolved solids in one acre

-------
96
foot of water, and is computed by multiplying the concentration in milli-
grams per liter by 0.00136. Since water quality samples were taken most
consistently on or near the 20t(i of each month, the sample concentration
nearest the 20th was used in computing the average monthly salt load at
each station location. Monthly net salt pickup for the reach (S^) was
calculated by subtracting the computed values at each inflow station from
the reach outlet station value (_equation 1). The annual estimates of net
salt pickup for the reach (S^.) are the sum of the monthly estimates, and
therefore reflect a weighted average of monthly flow rates and TDS con-
centrations (equation [2]).
sit - csu - (sc+ysd^t
12
[1]
[2]
where: S = net salt load in tons
i = month of year, i = 1, 2, 	12
t = year in times series, t = 1, 2, 	T
Location of sampling stations
u = Colorado River near Cisco, Utah (reach outlet)
c = Colorado River at Cameo, Colorado (inflow)
g = Gunnison River at Grand Junction, Colorado (inflow)
d = Dolores River near Cisco, Utah (inflow)
An equivalent procedure was used to derive net river discharge
i.e., the net gain or loss in flow for the study reach.

-------
97
Due to rather large variations in flow and concentration of salts
which can occur over a period of 30 days, the above procedure could yield
inaccurate measurements of monthly salt pickup. For example, if the TDS
concentration on the 20th of the month is 30 percent below the "true"
weighted average, our estimate of the salt load will also be 30 percent
below its true value. Similarily, an error made in estimating the salt
load at one of the four station locations will cause an error in computing
the net pickup for the reach. Errors in measurement are indicated by a
number of "outlyers" in the data, because a fairly small sample size
was used (22 years). This presents problems for estimating statistical
parameters. The implications of this in interpreting the results of the
analysis is discussed in detail in a later section.
Hydrologic systems are characterized by stochastic processes (in-
cluding possibly cyclic behavior), hence a significant bias could be in-
troduced by the particular interval of observations selected for analysis.
As an illustration, the mean flow and salt load for the Grand Valley reach,
computed from three different intervals of time, are compared in Table 14.
These computations suggest stream flows have apparently declined during
the entire period while the net average salt load rose in the second in-
terval and fell in the third. The magnitude of mean salt loads for two of
the three intervals is substantially lower than the generally accepted long
term average of 600,000 tons [U. S. Bureau of Reclamation, 1975], The fact
that the mean values are different suggests the possibility that time trends
or cycles exist and should be considered in a careful analysis of salinity
cause and effect.

-------
TABLE 14 MEAN SALT LOADS, STREAM FLOWS, AND TDS CONCENTRATIONS IN THE STUDY REACH: A COMPARISON OF THREE
TIME INTERVALS


Annual Mean Values
for Three
Periods of
Water
Flow and Quality-^

Location of
sampling stations
in the study reach

1914-1957-/

1943-1960^


1951-1972-7

Flow
Cone.
TDS
Flow
Cone.
TDS
Flow
Cone.
TDS
Colo. River, Cameo
2,998
.53
1,578
2,859
.54
1,537
2,656
.57
1,525
Plateau Creek, Cameo
170
.39
66
165
.41
68
(159)
(.43)
(68)
Gunnison River,
Grand Junction
1,884
.81
1,519
1,705
.84
1,430
1,594
.87
1,383
Dolores River, Cisco
681
.68
460
(581)
.83
491
480
.96
462
Colo. River, Cisco
5,534
.74
4,120
5,030
.82
4,133
4,546
.86
3,922
Net Change-^
-199
—
497
-280
—
607
-343
—
484
a/ Flows are measured in 1000 acre feet, concentrations in tons TDS per acre foot, and salt loads (TDS)
in 1000 tons.
b/ Adapted from Iorns, et al_ [1965],
c/ U. S. Department of Agriculture [1965]. The number in parenthesis was not given in the report cited,
but was supplied by the author.
d/ Summarized from published records, U. S. Geological Survey [1973]. The estimates in parentheses are
extrapolations made by the author from incomplete historical records.	^
CO
e/ Cisco data less the sum of the other stations.

-------
99
Natural Runoff
Estimating the salt loading effects of naturaUdiffuse surface
sources is complicated by the fact that natural washes and intermittent
streams which transport salts to the main stream channel are ungaged
and only periodically sampled. In the absence of direct observation,
precipitation data was used as an indirect measure of surface runoff
(orsalt pickup). Records of monthly rainfall were obtained for three
different locations within the study area [U. S. Department of Commerce,
1972]. Three weather stations, located at Grand Junction, Fruita, and
near Cisco, were used in order to account for the possible effects of
the spatial distribution in rainfall patterns.
The relationship between monthly or seasonal rainfall and rates of
surface runoff is fairly well grounded on empirical evidence [Musgrave,
1947; Thomas and Benson, 1972; and Flaxman, 1972]. Although the actual
process of salt pickup takes place on a day-to-day basis, depending upon
(1) the frequency of storms, (2) rainfall amount and intensity, (3) sur-
face vegetation and slope, and (4) soil salinity, infiltration, antecedent
moisture, etc., Schumm [1963] has shown that the magnitude of runoff is
closely associated with variations in seasonal precipitation. The data
summarized in Table 15 illustrate the frequency with which storms of dif-
ferent sizes occur, and the seasonal effects on the precipitation-runoff
ratio for the Badger Wash area(which can be located in figure 1). Both
storm frequencies and precipitation-runoff ratios are clearly greater
during the fall months, and according to our hypothesis, the same should
be true of salt pickup from this source.
An alternative measure of natural runoff is net river discharge since
ungaged overland flow must obviously be reflected in this variable. If

-------
100
TABLE 15. SPRING AND FALL PRECIPITATION AND RUNOFF AT BADGER WASH,
1954 - 1961^/
Average Runoff-
Range of Number of Runoff Precipitation
Precipitation	Storms	per Storm		Ratio
per Storm
Spring Fall
Spring Fall
Spring
Fall
-(Inches)-



—(Inches)
—

.10-.20
13
23
.001
.008
.007
.050
.21-.30
4
11
0
.039
0
.145
.31-.40
3
11
.007
.063
.022
.185
.41-.50
3
7
.020
.093
.043
.198
.51-.60
1
2
.120
.180
.207
.322
CT>
1
O
1
4
.004
.290
.006
.440
.71-1.00
0
2
0
.380
0
.400
1.01-1.40
0
1
0
.470
0
.350
All storms
25
61
.004
.079
.015
.232
Source: Schumm, S. A. and G. C. Lusby. "Seasonal Variation of
Infiltration Capacity and Runoff on Hi 11 si opes in Western
Colorado," Journal of Geophysical Research, Vol. 68,
No. 12, June, 1963.
a/ Spring and Fall precipitation periods include the months
of April through June and August through October,
respectively.

-------
101
the gaged stream flows are reasonably accurate for the study reach, then
net discharge (Ft) will also act as a proxy for salt pickup in surface
runoff.
Irrigation Return Flows
Irrigation was first practiced in the Grand Valley in the early 1880's,
and by the late 19201s land under irrigation probably exceeded 75,000
acres. Since that time certain portions of the Valley have been retired
from production as a consequence of problems encountered with waterlogging
and a buildup of surface salts [Robinson, 1972]. At present approximately
56,000 acres are irrigated.
Colorado River water is diverted for irrigation by four irrigation
entities: the Grand Valley Irrigation Company, the Grand Valley Water
Users Association (a federal project), the Palisade Irrigation District
and the Mesa County Irrigation District. A small amount of water is di-
verted from the Gunnison River by the Redlands Light and Power Company
for irrigation south of the Colorado River (figure 8). For an average
water year total Colorado River diversions have been estimated at 550,000
acre feet, nearly 10 acre feet per irrigated acre in the Valley [Skogerboe
et al., 1974]. About half of this amount is actually delivered to farms,
and of that delivered only about 3 acre feet per acre is typically needed
to satisfy crop evapotranspiration and leaching requirements.
Because of the hypothesized significant effect of irrigation on the
salinity flow system of the Grand Valley reach, return flow processes
were defined as accurately as possible. Data sources and procedures used
to measure the principal components, both subsurface (deep percolation and
seepage) and surface (waste and field tail water), are described in the
subsections which follow.

-------
102
Canal Diversions. The time distribution and quantities of water di-
verted from the river and conveyed to the irrigated land area is funda-
mental to this analysis, since estimates of conveyence losses and irri-
gation waste water are derived from these statistics. Data on major canal
diversions is available through the office of the State Engineer, but in
consultation on the matter it was learned that in many cases only partial
records are available for all diversions on a monthly basis. Accordingly,
the approach followed in describing the history of diversions for the
Grand Valley was to use available records for selected canals in the area
thought to be representative of the total system.
The Government Highline Canal, managed by the Bureau of Reclamation,
was selected as the "representative" canal system in the Valley. This
canal services about 38 percent of the Valley's irrigated acreage and
handles about 36 percent of total river diversions [Skogerboe, et al_, 1974].
The data were obtained from annual summary reports prepared by the project
office personnel in Grand Junction [Klapwyk, 1974]. Since the records
maintained on this canal are historically complete and believed to be the
most accurate of the four main canal systems, it was not necessary to ex-
trapolate or extend the data to cover missing observations.
Seepage from Canals and Laterals. Numerous experimental observations
are available on rates of seepage and the amount of seepage loss for most
lateral and canal supply systems in the Valley [Kruse, 1975], but little
or no information is available on the time distribution of monthly or an-
nual seepage loss. The only source of data (known to the author) upon
which limited inferences can be made is an annual published report of
monthly water distribution statistics [Bureau of Reclamation, 1974].

-------
103
Because of some discrepancies found between the published version and the
actual field reports submitted to the regional office, the project office
field records were selected for use iKlapwyk, 19743.
In these annual summaries estimates of seepage loss, in acre feet per
month, are computed using a simple accounting procedure: river diversions
(RD-t) less gaged deliveries to the lateral system (LD^) and gaged spill-
age losses (Wc^t) yield unexplained residual water quantities (SLc^t)
which are attributed to canal seepage losses (equation3). Lateral seep-
age loss (SLl.jt) is estimated by subtracting from the lateral supply the
amount delivered to farms (FD^) and the amount spilled (Wl^) to maintain
an appropriate system flow (equation 4).
[3] SLCU = RD.t - (LD ~ Wc)n
PI SL11t=LD1t" (FD + W1>1t
[5] SLdn = SLc,t ~ SLln
wheret SLdt =j=, SLc1t SLl1t
(i and t previously defined)
Undoubtedly some of this residual can be explained by measurement
error, but there is no empirical grounds for scaling the estimates down
for this reason. Thus, the estimates of delivery system seepage loss
(SLd^) used for purposes of statistical analysis were obtained by sum-
ming canal and lateral "residual" losses (equation 5).
A major shortcoming of this approach is that the residual estimates
are not supported by (not directly comparable with) existing empirical
data on seepage rates for the Government Highline canal and service

-------
104
laterals. On an annual (average) basis, estimates derived by the resid-
ual method appears to be 20 to 40 percent above what the best available
experimental data would imply for this particular delivery system [Kruse,
1975]. In terms of the total canal-lateral system in the Valley,the
estimates are probably on the low side.
However, for the purposes of analysis of historical relationships,
this type of estimate is the appropriate one. The monthly data indicate
a close similarity between the seasonal distribution of seepage loss and
canal diversions (a relationship expected a-posteriori). Therefore, the
chosen measure of seepage loss is perhaps more appropriately an "index",
reflecting the intertemporal effects (both seasonal and annual) heretofore
untested as an explanatory variable in subsurface return flow and quality.
On-Farm Seepage and Deep Percolation. Salt pickup from subsurface
sources directly attributable to irrigation practices has not been
measured for the Grand Valley as a total unit area. As discussed in
section B, sound empirical data are available on a site-specific basis,
but no conclusive evidence is offered to dispell the significant incon-
sistencies which pertain to a valley-wide characterization of on-farm
pickup mechanisms.
Conceptually, the disposition of irrigation water delivered to farms
is separable by component of use: on-farm conveyance seepage (SLf), crop
water requirements (ETw), rootzone water storage (RZ), deep percolation
(DP), and field tail water (TW). The annual on-farm water budget used
to analyze these potential sources of salt pickup is represented by
equation [6]. Subsurface water losses (DP and SLf) were estimated as a
residual since secondary time series data only exists for (or is applicable
to) those variables on the right-hand side in equation [7].

-------
105
n
[6]	FD. = z (SLf + ETw + RZ + DP + TW)it
t i = l
[7]	(DP + SLf)t = FDt - (ETw + TW + RZ)t
Information on the monthly irrigation water deliveries to farms
(FD^) were obtained from historical records maintained by the Bureau
of Reclamation [Klapwyk, 1974]. Unlike other supply canals which are
operated on a "continuous flow" basis, the Government Highline Canal
is under the "call system". Water is delivered at farm headgates on
demand, and the delivered flows are recorded by project personnel
(ditch riders) for the purpose of billing users and, in the event of
water-short periods, to ration scarce supplies. Water costs to farmers
served by the other canals are "fixed" at the beginning of the ir-
rigation season and do not vary with the amounts diverted. Also, these
canals carry flows in sufficient quantity to offset the need for ration-
ing during periods of peak demands or shortages. For these reasons
there might be some question as to whether the Government Highline Canal
is "representative" of the other canals in terms of actual water deliveries
to farms. However, owing to the low cost of water generally, and the
fact that deliveries, even under periods of rationing (which are not
common), are normally more than sufficient to satisfy crop requirements,
it is not likely that differences between the canal systems will be very
large.
The crop water requirment, ETw in equation [7] was estimated with the
use of the standard Jensen equation [Jensen and Haise, 1963]. Data for
the computations was obtained from the Bureau of Reclamation's project
office located at Grand Junction [Wiscombe, 1975]; and [Walker,1975]. The

-------
106
procedure, outlined in equations 8 through 10, builds upon individual
crop demands and incorporates the changing crop mix which occurs over
time.
[8]	(ETp, ¦ Kfj)t
[9]	ETwjt = (A • ET)Jt
m
[10] ETw. = E ETw.
t j=l J
where:
j = crop, j = 1, 2, 	m; i,t previously defined
ETp = potential evapotranspiration (inches per acre)
K = crop growth stage coefficient (% of ETp)
ET = actual evapotranspiration requirement (inches per acre)
A = crop acreage
ETw = total irrigation water requirement, (ETw t 12inches = acre feet)
The aggregate estimate of total crop water demand is seasonally weighted
to reflect historical growing conditions and cropping patterns which
apply to the Government Highline Canal system covering the period of
this analysis.
This procedure provides a fairly precise measure of the potential
water requirement, but it does not purport to accurately measure the
timing or amounts of water actually applied to growing crops. Other variable
which play an important role in plant water requirements, for example com-
mercial fertilizer or new crop varieties, also influence water use but
could not be incorporated into the analysis. The measured variable ETw
is, therefore, an index of crop water demand representing the seasonal
changes in irrigation water delivered to farms in response to changes in
atmospheric conditions and the growth stages of a particular mix of crops.

-------
107
The aforementioned question of "representativeness" of the canal
system is especially critical here, since water requirements vary mea-
surably between types of crops. If the acreage and mix of crops served
by the canal system selected for study is not representative of past
trends for the Valley as a whole, then the time distribution of seasonal
irrigation water deliveries obtained from the above procedure will cause
a misspecification of salt pickup via deep percolation and seepage on
farms equation 1 . Although scant evidence is available upon which to
verify crop mix and acreage being the same for the canal and the Valley,
the available data suggests that the two were reasonably similar during
the period of study (1951-1972). However, this may not be true prior to
the mid-forties when sizable acreages of lowlying lands (situated below
the Highline Canal and contiguous to the river) were being taken out of
production due to a rising water table [Bureau of Reclamation, 1956].
Estimates for tail water loss (TW) and rootzone water storage (RZ)
used in the model equation 7 were adapted from the results of irri-
gation efficiency studies in the Valley[Walker, 1976; and Bureau of
Reclamation, 1956]. Annual tail water losses were computed with the use
of a fixed proportion (P^) or percentage of monthly delivered supplies
(FD^), the proportion of runoff varying with the seasonal characteristics
of the irrigated soils:
[11] TW+ = s (P • FD)
r i=i
Essentially P reflects the rate of water applied in excess of the
potential rate at which water can infiltrate the soil in a given period
of time. By this method tail water loss, computed as a proportion of
water delivered to the farm, is estimated for each month and varies

-------
108
directly with the amounts of delivered water. It was assumed that the
monthly values of P are the same each year, i.e., P. is a constant. This
is perhaps an oversimplification but the limited evidence available tends
to support this presumption [Walker, 1976],
Monthly depletion of water from the rooting zone of the soil pro-
file was defined as either crop consumptive use or deep percolation.
Waters entering the rootzone in excess of consumptive use during the ir-
rigation season (April - October) were counted as percolation losses.
Further, the soil moisture stored in the rootzone at the end of one
season was assumed to carry over to the following season, thus natural
moisture accumulation in the intervening period (November - March) was
considered to be about equal to evaporative losses during the same period.
On the basis of these assumptions, rootzone water storage (RZ) in equation
[7] was omitted from the computations.
This summarizes the sources and nature of the empirical data used
to test some of the important hypotheses discussed in section B. To
conclude this section the dependent and explanatory variables which form
the basis for the statistical tests are reviewed.
The Empirical Hydro-Salinit.y Equations
The dependent variable or quantity to be explained is defined as
net salt pickup (S) in the study reach. The magnitude of salt pickup
at any point in time is said to be the result of agricultural (Sa) and
natural (Sn) sources. Other sources of salt loading, municipal-industrial
contributions (denoted Su in equation 12), were not studied directly.
[12] S-t = (Sa + Sn + Su)^ (from [1])
We distinguish between these conceptually, but in reality the separate
quantities are unknowns.

-------
109
A further distinction was made between surface and subsurface mech-
anisms. This applies to both natural and agricultural sources, but only
the agricultural processes are quantifiable with existing historical
data. The surface contribution from agriculture (Sa) is accounted for
by administrative spillage in the delivery system (Wd) and tail water
(TW) from irrigated fields:
[13]	San = (Wd + Tw).t	(from [3], [4], and [6])
Subsurface salts (Sa_) are picked up in the seepage water of the delivery
system (SLd), from water that percolates below the rootzone of irrigated
crops (DP), and from head ditch seepage on the farms (SLf):
[14]	Sa.t = [SLf + DP].t + SLd (from [5] and [7])
The explanatory variable used to account for natural salt loading
within the reach is recorded rainfall. The amount of rain (R) is used
as an index in an attempt to measure indirectly both surface runoff and
ground water interchange mechanisms:
[15]	SnH = f (Rit)
These equations essentially comprise the conceptual model of the
salinity-flow system for the study reach. It should be noted that the
explanatory variables are all expressed in terms of water quantities
while the dependent variable is expressed as quantities of salts. In
this form the estimated regression coefficients will be interpreted as
"rates of salt pickup" or tons of salts per acre foot of return flow
waters. Since the manmade and natural portions of salt pickup are un-
knowns, least-squares estimation of the structural coefficients is based
upon the total salt load (S in equation 12). Accordingly, estimates of
Sa or Sn cannot be obtained directly from the statistical estimation
procedure, but must be inferred from the explanatory power of the in-
dependent variables included in the equation.

-------
110
D. RESULTS AND DISCUSSION
The results or findings are reported in three phases in accordance
with the procedures followed in analyzing the data. In the first phase,
annual data were used to examine the historical relationships between the
various explanatory variables (developed in the previous section) and water
quality for the period of study, 1951-1972. The second phase of the analysis
focuses on the monthly data. An attempt was made to discern the nature of
specific pickup mechanisms as they may be manifest in a monthly or seasonal
context. Special attention was paid to testing of the hypothesized "sea-
sonal" contributions of natural runoff and irrigation return flow to salt
loading. The possibility of time lags in irrigation return flows was also
addressed in this phase. The monthly data were aggregated into "quarters"
of the year in the third phase of the study to further explore the question
of seasonal effects as well as to corroborate these findings with the monthly
and annual analyses.
Part I: The Annual Data
The long-term relationships between the natural and manmade variables
and salt pickup are compared in a series of data plots. In figures 9 and 10
each variable is plotted with respect to time, and in figures 11 through 13,
selected explanatory variables representing possible natural and irrigation
contributions are plotted against net salt pickup.
The time plots yield some interesting implications. First, annual net
salt pickup in the reach has apparently been declining during the period of
study while irrigation diversions and water deliveries show a notable in-
creasing trend. This observation, if true, would be contradictory to the

-------
Ill
figure 9. Annual Estimates of Salt Pickup, Irrigation Water Deliveries
to Farms, Precipitation and Net Discharge for the Grand Valley
Reach, 1951-1972.

-------
230
220
210
200
190
180
70
65
60
55
50
45
40
35
75
70
65
60
55
50
45
35
34
33
32
31
30
29
28
112
951 53 55 57 59 61 63 65 67 69
Annual Estimates of Irrigation Diversions, Distribution System
Waste Water, Canal and Lateral Seepage Losses, and Crop Evapotre
spiration for the Government Highline Canal, 1951-1972.

-------
113
hypothesis that salt pickup is caused by excessive irrigation losses. Since
crop consumptive use has remained about the same during the period, irriga-
tion return flows (as reflected in the spillage, seepage and deep percola-
tion variables) show some increase with time. One variable which moves
most closely with net salt pickup is net river discharge. This apparent
relationship might be explained by increased consumptive withdrawals for
municipal or industrial uses but no attempt was made to confirm or refute
this conjecture.
A second general observation has to do with annual variability. It is
apparent that in certain years large shifts in some of the variables are
coincident. For example, the years 1957, 1961, and 1965, periods of sharp
increase in salt pickup, are periods of exceptionally high annual precipi-
tation. If the precipitation variable is an indicator of natural runoff as
hypothesized, then this would tend to support the contention that natural
mechanisms could be a significant source of salt loading. As expected,
extremes in annual precipitation are also reflected in net discharge. The
inverse relationship between extremes in precipitation and irrigation is
fairly obvious, since salt pickup and irrigation are almost mirror images of
each other.
Sample means, standard deviations and coefficients of variation for the
regression variables are reported in table 16. The coefficient of variation
(the standard deviation divided by the mean) for net discharge and deep
percolation are exceptionally large due to the inclusion of positive and
negative values in the range of observation. Deep percolation has the
highest coefficient of variation among the irrigation variables. Intuition
would suggest that seepage and canal flows should be highly related, and
based upon the sample statistics, canal divisions, seepage and deliveries

-------
TABLE 16. MEANS, STANDARD DEVIATIONS AND COEFFICIENTS OF VARIATION OF THE REGRESSION VARIABLES:
ANNUAL DATA, 1951-72
Statistical Parameter:
Regression
Variables
Unit of
Measure
Mean
Standard
Deviation
Coefficient of
Variation^/
Net Salt Pickup
1000 TONS
541.14
184.36
.34
Net Discharge
1000 AF
180.67
129.91
.72
Precipitation:




Grand Junction
inches
8.02
2.71
.34
Fruita
inches
9.05
3.03
.33
Cisco
i nches
7.98
3.06
.38
Irrigation Diversions
1000 AF
216.41
24.61
.11
Seepage Losses
1000 AF
62.26
7.72
.12
Spillage
1000 AF
54.61
12.04
.22
Farm Deliveries
AF/A
4.64
.51
.11
Deep Percolation
inches/A
4.39
4.56
1.04
Crop ET
inches/A
32.05
2.10
.07
^/standard Deviation * Mean

-------
115
to farms all have a coefficient of variation of about the same magni-
tude.
The annual data relating each of the more important independent
variables to net salt pickup are plotted in figures 11 through 13. These
plots, although reflecting extreme variability in the data, suggest
that some relationships do exist. However, the "fit" of the paired
variables to a straight-line function is generally quite poor.
Figure 11 compares two independent variables which are indicative of
natural influences (precipitation and net river discharge). Both
demonstrate a positive relationship with net salt pickup, although the
influence of precipitation is perhaps less clear. Increases in net
discharge (denoted by smaller negative values) are hypothesized to
reflect the influence of natural runoff, but in reality they also
reflect irrigation diversions. Hence, this variable is a composite of
many potential effects and therefore its interpretation as an explana-
tory variable of natural phenomena may be misleading.
River diversions and water deliveries to farms, compared in figure
12 both relate inversely to salt pickup. That is, in years of high
diversions of water or deliveries to farms the salt pickup in the reach
is low, and vice versa. This empirical relationship is in juxtaposition
to the contention that increased irrigation efficiency will automati-
cally lead to improved water quality, in the sense that reduced system
and on-farm losses will permit less water to be diverted, thus avoiding
some salt pickup. The data plots presented in figure 13, relating esti-
mated subsurface return flow variables to salt pickup, are less clear.
Seepage is positively correlated with salt pickup over the range of
data, but may be more appropriately plotted in two functions: one

-------
4 5 6 7 8 9 10 11 12 13 14 15 16
+50
-100 -200 -300 -400 -500 "600
Figure 11. Data Plots of Net Salt Pickup with Precipitation and Net Discharge: Annual Data

-------
Net
Salt
ckup
1000
Tons
900
800
700
600
500
400
300
200
100
12. Data Plots of Net Salt Pickup with Irrigation Diversions and Farm Deliveries: Annual Data

-------
Net
Salt
ckup
1000
Tons
900
800
700
600
500
400
300
200
10G
13. Data Plots of Net Salt Pickup with Canal Seepage and Deep Percolation: Annual Data

-------
119
negatively sloped for salt pickup values above the mean, and one posi-
tively sloped for values of pickup below the mean. However, this kind
of function is incongruous with existing theories of how the hydro-
salinity system functions in the Grand Valley. The plot of on-farm
losses (deep percolation and seepage) with salt pickup is similar to
that found for farm deliveries, indicating an inverse relation.
The appropriate set of variables to be regressed on net salt
pickup was determined by an examination of the intercorrelation between
pairs of variables. If two or more explanatory variables are highly
correlated, it is difficult to separate their respective effects on the
dependent variable. In such cases only one variable was included in
the regression equation. In multiple regression, a high interdependence
between the explanatory variables will lead to biased estimates of the
regression coefficients—a problem known as multicollinearity. The
correlation matrix is presented in table 17.
The correlation coefficient is a measure of the degree of linear
association between two variables. A negative coefficient implies an
inverse relationship while a positive coefficient indicates a movement
of the two variables in the same direction over the range of sampled
data. An examination of the correlation coefficients reveals, for ex-
ample, that the three precipitation variables are highly intercorrelated.
Accordingly, only one of these, the Grand Junction data, was used in
the regression analysis. Farm deliveries and deep percolation are also
highly correlated, and therefore these variables could not be considered
jointly in a multiple regression equation.
The specific functional form of a quantitative relationship 1s
usually determined empirically, but consideration should always be
given to the theoretical implications of doing so. The easiest functional

-------
TABLE 17. CORRELATION MATRIX OF SELECTED VARIABLES USED IN REGRESSION: ANNUAL DATA^
Variable
Net River
Discharge
Precipitation
Grand Jet. Cisco
Fruita
FD
Irri<
SLd. ¦
nation
(SLf+DP)
Time
Time
-.54
.01
.01
.09
.69
.01
.77

Salt Load:








Colorado R. at Cameo
.95
.26
.47
.28



-.01
Gunnison R. at Grand Jet.
.95
.56
.61
.60



.15
Dolores R. at Cisco
.90
.49
.46
.50



.09
Colorado R. at Cisco
.96
.56
.64
. 56



-.12
Net Salt Pickup
.64
.36
.32
.29
-.72
.50
-.74
-.65
Net River Discharge

.33
.15
.29
-.64
-.33
-.51

Precipitation:








Grand Jet.


.87
.98
-.59
-.27
-.32

Cisco



.86
-.52
-.11
-.25

Fruita




-.53
-.16
-.23

Farm Deliveries (FD)





-.50
.92

Seepage (SLd)






-.58

—^Coefficients for the two-variable case only.

-------
121
form to explain and estimate is a linear equation. In some cases an
exponential or logarithmic function better describes the curvature of an
empirical relationship. Some of the criteria that can be used to choose
among the many possible forms for regression equations include simplicity,
theory, predictive power, etc. However, the procedure followed in this
study placed more emphasis on the estimated equations themselves: (1)
the goodness-of-fit (to rely on the significance of R2 as shown by the F
statistic); (2) the significance of regression coefficients (as shown by
the t statistic); and (3) the analysis of residual patterns (a test of
appropriateness in equation form based on time-trend in the error term).
With few exceptions, no firm justification was found, on the basis of these
three considerations, to warrant the use of other-than linear equation
forms.
The series of regression equations, formulated to test the explana-
tory power and significance of the hypothesized causal factors relating
to annual net salt pickup, are reported in tables 18 through 20. In table
18 the equations which account for the natural effects are summarized.
Considered by itself annual precipitation is not a significant deter-
minant of salt pickup (equation 18.1). The F statistic indicates that the
total variation in salt pickup associated with precipitation, a measure
O
expressed by the adjusted coefficient of determination (R ), is not sig-
nificantly different from zero. R is the normal coefficient of determina-
o
tion (R ) "adjusted" for the degrees of freedom in each regression equation.
Secondly, the t test shows that the intercept parameter (a) is sig-
nificantly different from zero. The regression coefficients (6, c, etc.)
define the slope of the regression line, and are interpreted as the change
in the dependent variable that is associated with a given (or specified)

-------
TABLE 18. SUMMARY OF REGRESSION RESULTS RELATING NET SALT PICKUP TO NATURAL CAUSES: ANNUAL DATA
	Test Statistics^/	
Regression Coefficient(s) 2 } value
Equation Intercept	(Standard Errors)	R F Ratio abed	d.w.
1	339.4 + 25.15 R	.14 3.01 2.77 1.74	1.34
(122.4)	(14.49)
2	1478.9 + 26.27 R - 18.52 T	.53 9.96 4.75 2.38 3.83	.99
(311.2) (11.04) (4.83)
3	724.8 +	1.017 F	.51 20.02 14.47 4.48	2.07
(50.1)	(.23)
4	1092.0 + .809 F - 6.46 T	.52 4.59 3.13 2.01 1.06	2.07
(349.2)	(.40) (6.1)
5	633.6 + .947 F + 9.81 R	.53 7.48 5.30 3.90 .84	1.41
(119.5)	(.24) (11.65)
6	607.7 + .65 F + 12.72 R + .038 (FR).53 6.51 4.23 .73 .87 .35 1.87
(143.7) (.89) (14.62) (.11)
^Critical values at the 5 percent level	of significance:
F (n-i, n?): (1,20) = 4.35 t (df):	(20) 2.09 d.w. (n, k'): (22,1) = 1.12>d.w.>l.31
1	(2,19) = 3.52	(19) 2.09	(22,2) = 1.04>d.w.>1.42
3,18) = 3.16	(18) 2.10	(22,3) = .95>d.w.>l.54
4,17) = 2.96	(17) 2.11

-------
123
change in the independent variable(s). The intercept, a, is the constant
term in the equation, or the point on the vertical axis (which describes
the range in observation values of the dependent variable) through which
the regression line passes. Lines of regression between net salt pickup
and the separate independent variables are shown with the data plots des-
cribed earlier (figures 11 through 13). But as other statistical tests
will demonstrate, one-on-one relationships can be misleading in terms of
identifying meaningful empirical relationships.
Third, the Durbin-Watson statistic (d.w.), which is a measure of
"serial" correlation (i.e., the degree of correlation between the dependent
variable and the error term), indicates the regression equation might be
misspecified (i.e., an additional variable or variables should be in-
cluded in the equation). Thus, when time (T) is included in the equation
(shown in 18.2), the explanatory power of the equation is substantially
enhanced. Improvements are also noted in the level of significance of
the regression coefficients (£, £, and c are now significant), and by the
absence of some serial correlation.
In equation 18.3 net river discharge is shown to be a significant
factor in explaining net salt pickup. Since time and net discharge are
intercorrelated (table 17), the addition of the time variable does not
improve the equation (18.4). In equation 18.5 net discharge and precipi-
tation are considered together. In this case the addition of precipita-
tion reduces the level of significance of the other coefficients and
introduces some serial correlation into the equation. Use of the inter-
actions term FR, computed by multiplying variable F by R, is one method
of reducing regression bias caused by serial correlation. However, it
is shown in equation 18.6 that, although Including the interactions term

-------
124
increases the d.w. statistic, the significance of the regression coeffi-
cients is also reduced, indicating that the estimated parameters in equation
18.5 may have an upward bias. But, the fact that the interaction term is
not significant would suggest that the degree of regression bias is not
large enough to warrant including the term in the equation.
The evaluative approach described above was used to carefully exam-
ine each regression equation reported in subsequent tables. In the remain-
der of this chapter only the more general observations and findings of
the analyses are discussed. Interpretations as to the meaning of specific
regression coefficients, for example, the change in salt pickup attributed
to a specified change in seepage water, will be taken up in the summary
portion of this section.
The results of the analysis of manmade effects are summarized in
table 19. Perhaps the most important observation to be made here is that
all but the seepage variable (SLd) have negative signs. The time variable
did not materially improve any of these regression equations. A new
variable, total irrigation return flows, was computed by subtracting
crop evapotranspiration from river diversions. Since surface return flows
(Wd and IV) were not tested directly as were the other variables, this
variable (RD - ET) is used to represent the combined effect of both surface
and subsurface sources. The influence of total irrigation return flow on
net salt pickup was not significant. In general, these results establish
a statistically significant but inverse relationship between the subsur-
face components and the net pickup of salts.
A second approach is to consider the joint influence of natural and
manmade variables in the same equation. In this form the empirical equation
is more realistic in terms of the total water-salt relations for the study

-------
TABLE 19. SUMMARY OF REGRESSION RESULTS RELATING NET SALT PICKUP TO SELECTED MEASURES OF
IRRIGATION VARIABLES: ANNUAL DATA
Test Statist!csA/


Regression Coefficient(s)
(Standard Errors)



t value

Equation
Intercept
R2
F Ratio
a
a
b
A /S
c d
d. w.
1
1754.9 -
(271.1)
261.80 FD
(58.14)
.50
21.88
6.47
4.50

.88
2
1628.7 -
(372.5)
162.30 FD - 5.16 T
(86.80) (6.64)
.28
4.25
4.37
1.87
.78
.93
3
282.4 +
(108.8)
4.54 SLd
(1.80)
.21
6.35
2.60
2.52

1.20
4
1294.4 +
(430.4)
2.47 SLd - 14.42 T
(1.82) (5.97)
.37
6.89
3.01
1.35
2.41
1.53
5
673.3 -
(38.6)
30.13 (SLf + DP)
(6.19)
.54
23.73
17.42
4.87

1.35
6
831.9 -
(184.4)
13.02 (SLf + DP) - 22.04 T .36
(8.99) (16.20)
3.75
4.51
1.45
1.36
1.19
7
611.7 +
(125.6)
.90 SLd - 27.82 (SLf +
(1.74) (7.77)
DP).51
7.85
4.87
.52
3.58
1.24
8
889.0 -
(255.9)
3.75 SLd - 50.51 (SLf +
(4.13) (19.88)
DP).53
6.98
3.47
.91
2.54 1.24
1.49


+ .44 [SLd (SLf + DP]
(.36)





9
1211.4 -
(348.4)
87.03 (RD - ET)
(44.32)
.13
3.86
3.53
1.96

1.07
10
1644.1 +
(350.9)
21.73 (RD - ET) - 20.49 T
(58.85) (8.25)
.31
3.53
4.69
.37
2.48
1.30
a/ Refer to table 18, footnote [a], for critical values.

-------
126
reach, and by including both types of effects, their separate importance
is more clearly ascertained. These results, summarized in table 20, are
generally inclusive. Because of the interaction between rainfall and
irrigation, the rainfall variable is not significant in any of the equations
pairing rainfall with the return flow variables. The flow variable was
not used here because of problems with multicollinearity.
In summary, the analysis of annual data demonstrates the difficulty
of isolating causal factors which explain annual variations in water
quality. The relationship between precipitation (or net river flow) and
irrigation activity in the study reach does not allow a statistically
significant separation of their individual effects. However, in the two
variable cases the importance of natural or manmade sources is identifiable
and significant, but in this form the regression equations are not consis-
tent with the conceptual hydrosalinity model. Regression parameters for
natural sources exhibit the appropriate sign, but variables associating
salinity to irrigation have the wrong sign to confirm a positive impact
from this source.
An important objective of the study was to evaluate the effects of
time on changes in salt pickup. In some cases including a time variable
enhanced the explanatory power of the regression equation. During the years
covered by the analysis, annual salt pickup in the study reach declined
significantly with time. This means that the combined hydrosalinity
processes during the period functioned in such a way that they contributed
to a lessening of net salt pickup. And since it was shown that irrigation
return flows tended to increase during the same period, this finding is
at least a cause to question further the deleterious impact irrigation is
believed to have on the salinity problem in the Grand Valley area.

-------
TABLE 20. SUMMARY OF REGRESSION RESULTS COMBINING THE	NATURAL AND MANMADE CAUSES: ANNUAL DATA
Test Statistics^/
Regression Coefficient(s) _2	—-	* value		
Equation Intercept	(Standard Errors)	R	F Ratio abed	d.w._
1	977.1 + 22.67 R - 80.21 (RD - ET) .20	3.48 2.73 1.66 1.88	.70
(357.7)	(13.67) (42.61)
2	1879.1 - 5.49 R - 279.09 FD .47	9.77 4.51 .40 3.80	.90
(416.4)	(13.72) (73.55)
3	593.9 + 8.83 R - 28.17 (SLf + DP) .52	11.68 5.51 .748 4.21	1.33
(107.8)	(11.26) (6.70)
4	204.2 + 26.30 R + 2.024 SLf .05	1.51 .54 1.74 .38	.62
(376.5)	(15.13) (5.32)
^Refer to table 18 footnote [a], for critical values.

-------
Part II: The Monthly Data
The monthly data are perhaps more appropriate to the study of salt
pickup processes. Rainfall and irrigation events are more "seasonal"
in nature, and thus the influence of these factors will not always show
up distinctly in the annual data. Plots of the sample means, which pertain
to the 1951-1972 period, are depicted in figures 14 and 15. In figure 14 the
monthly distribution of mean net salt pickup, farm deliveries and precipi-
tation (recorded at Grand Junction) are compared. The implications sug-
gested by these data are several. First, the monthly distribution of salt
pickup appears to follow a definite pattern possibly reflecting both the
influence of irrigation and precipitation. In June the function registers
a sudden increase and later in the year, about October, the function
peaks out. The sharp increase about mid-year is fairly consistent from
year to year, and thus represents a unique event to attempt to explain.
Second, it is noted that the monthly distribution of farm deliveries
follows nearly the same general form as salt pickup. The initial jump
in water use occurs one month prior to the sudden increase in net pickup,
and might be attributed to the general practice of preirrigation in the
Valley. The month of highest irrigation diversion (and use) typically
occurs in July, and this could coincide with peak salt pickup later in
the year if a sufficient time lag is allowed for return flows to reach
the river.
Lastly, the precipitation-runoff ratios discussed in table 15 indicate
that the period of most significant runoff can be expected to occur late
in the summer (August through October) when monthly precipitation is the
largest and of greatest intensity. The peak salt load does appear to
coincide with peak precipitation, especially in the month of October. A
128

-------
129
.1
J FMAMJJASOND
Figure 14. Monthly Mean Estimates of Net Salt Pickup, Precipitation and
Delivered Irrigation Water for the Government Highline Canal,
1951-1972.

-------
130
Figure 15.
Monthly Mean Discharge and Salt Load of the Gunnison River, and
Precipitation at Three Locations in the Grand Valley Reach,
1951-1972.

-------
131
better indication of the possible importance of runoff sources can be seen
in figure 15. Here, the monthly mean distribution of salt loads carried
by the Gunnison River, calculated for the same period of analysis, is
described along with its seasonal mean flow and area precipitation data.
The Gunnison River Basin is quite similar to the study reach in that
much of the nonirrigated drainage area is composed of soils derived
residually from Mancos shale. The relationship between salt load and
flow is apparently highly related. Also, the variations in salt load and
precipitation in the months of July, August, September and October are
indicative of the natural runoff effect. The noticeable increase in flow
in October, if it can be attributed to overland runoff, will further
support this hypothesis.
Other components of irrigation return flow are plotted in figure 16.
These data show the relationship between the monthly means for farm deli-
veries, crop consumptive use (ET water), field tail water, deep percola-
tion (including on-farm seepage losses), canal and lateral seepage, and
administrative spills. It is important to observe that the deep percola-
tion variable is largest in May, but for the balance of the irrigation
season its magnitude is quite small. Nonetheless, deep percolation has
been considered to be a significant contributor of salinity in the Grand
Valley; and for this reason, year-to-year variations in magnitude are
perhaps more significant than its "mean" value.
The extent of variation observed in the monthly regression variables
is summarized in table 21. In contrast to the annual data, monthly varia-
tions are more extreme as indicated by many coefficients of variation
exceeding a value of l. Months with high coefficients are apparently
"transition" periods, most notably the beginning and ending of the

-------
L3Z
M
Figure 16. Monthly Means of Seasonal On-Farm Deliveries, Crop Consumptive Use,
Field Tail Water, Deep Percolation, Seepage, and Spillage for the
Government Highline Canal, 1951-1972.

-------
TABLE 21. COEFFICIENTS OF VARIATION FOR SELECTED REGRESSION VARIABLES: MONTHLY DATA
Coefficients of Variation
Month
Net Salt
Pickup
Net River
Discharge
Precipitation
(Grand Jet.)
Irriqation Variables
FD SLd (SLf + DP)
January
.62
1.50
.84



February
.59
1.23
.74



March
1.14
1.67
.83



April
.72
.79
.72
.63
.43
16.32
May
1.37
.55
.74
.14
.12
.30
June
.74
.76
1.06
.14
.12
15.40
July
.84
.37
.81
.09
.11
.78
August
.92
.52
.79
.17
.12
2.75
September
.57
.96
.97
.23
.18
1.85
October
.73
4.87
.82
.54
.30
2.09
November
.44
1.31
.59



December
.65
4.32
.74




-------
134
irrigation season, the summer storm period, and spring snowmelt.
Table 22 reports the correlation coefficients of the monthly variables.
Correlations between the net discharge variable and salt pickup are
generally higher during the first and last few months of the year, the
period of normal low flow. However, the correlations also show higher
positive coefficients during two of the summer storm months, August and
September. For the rainfall data, the largest coefficients also occur
during this period which is consistent with the plotted data in figure 14.
Farm deliveries and deep percolation are inversely correlated with salt
pickup in all but the early part of the irrigation season. On the other
hand, the seepage variable is positively correlated in all but the last
two months. The time variable indicates a negative relationship generally
during nonirrigation months but a positive (although weak) relation during
the months of April, May, July and August. The intercorrelation between
the independent variables is not shown in table22, but was taken account
of in selecting variables for the regression runs.
A second set of correlation coefficients is reported in table 23.
These coefficients provide some insight into the possibility of lagged
irrigation return flows. If a lag does exist, then the correlation of
irrigation variables directly with salt pickup (table22) would not show
this. A careful examination of these data, however, does not strongly
support the existence of return flow lags. Correlation coefficients
relating to seepage show some increase when lagged one, two and even three
months, but in general the difference does not appear large enough to
materially affect the significance of the regression coefficients. Farm

-------
TABLE 22. CORRELATION COEFFICIENTS BETWEEN THE REGRESSION VARIABLES AND NET SALT PICKUP:
MONTHLY DATA®/
Net Salt
Net River
Precipitation


Irrigation

Pickup
Discharge
Grand Jet.
Fruita
Cisco
FD
SLd
(SLf + DP)
Time
January
.79
.29
.23
.23



-.79
February
.52
.16
.17
.18



-.38
March
.50
.22
.17
.13



-.44
April
.17
.43
.43
.31
-.30
.12
-.29
.17
May
.19
CO
o
•
1
-.12
.13
.10
.50
.02
.42
June
.46
-.18
-.33
-.10
.11
.37
-.12
-.03
July
.01
.42
.51
.53
-.36
.30
-.30
.08
August
.64
.48
.41
.54
-.49
.48
-.36
.10
September
.58
.59
.66
.57
-.65
-.08
-.55
-.01
October
.31
.47
.44
.63
-.29
-.30
-.24
-.24
November
.78
.18
.25
.27



-.42
December
.79
.11
.04
-.01



-.46
Annual
.64
.36
.29
.32
-.72
.50
-.74
-.65
^Coefficients for the two-variable case only.

-------
TABLE 23. CORRELATION COEFFICIENTS RELATING THE INFLUENCE OF TIME-LAGS IN SUBSURFACE IRRIGATION
RETURN FLOWS ON SALT PICKUP: MONTHLY DATA—'
	Lagged Subsurface Irrigation Return Flows (in months)	
Net Salt	Canal and Lateral Seepage (SLd)	Deep Percolation (SLf + DP)
Pickup
No lag
1
2
3
4
No lag
1
2
3
4
April
.12




-.29




May
.50
.56



.02
-.17



June
.37
.40
.33


-.12
.20
-.22


July
.30
.32
.34
.13

-.30
-.43
-.30
-.36

August
.48
.52
.47
.50
.41
-.36
-.21
-.23
.06
-.20
September
-.08
.20
.14
.18
.17
-.55
.11
-.20
.17
.07
October
-.30
-.31
-.25
-.30
-.19
-.24
-.17
-.14
.04
.18
November

.23
.18
.23
.21

.19
-.16
-.55
-.36
December


.20
.18
.17


.18
-.15
-.62
January



.09
.04



-.14
-.28
February




-.11




.04
—^Coefficients for the two-variable case only.

-------
137
deliveries are not reported here since the results are similar to those
for deep percolation (due to a high intercorrelation between these two
variables).
It is not necessary to report the complete regression results as
the estimated equations are not in themselves particularly meaningful.
Rather, the 6 coefficient and level of significance of each estimated
parameterare reported in table 24 as a convenient way to summarize the
rather large volume of results. The conclusions to be drawn from the
statistical tests have important implications in terms of the hypothesized
"seasonal" effects. First, the only months in which precipitation is
significant as an explanatory variable are the high rainfall months of
August and October. July and September are significant at the 6 and 9
percent levels, respectively. These results are offered as a confirma-
tion of the runoff hypothesis. Secondly, the influence of irrigation on
net salt pickup is less significant. The only variable that is both
statistically significant and has the appropriate sign is deep percola-
tion, and this occurs only in May, the month of excessive irriqation
losses attributed to preirrigation practices in the Valley.
Estimates of the magnitude of salts contributed by precipitation or
deep percolation are not ascertainable from the results given in this
form. Relative contributions of salts by source will be discussed in
section E. A series of regression runs testing the lagged return flow
variables (i.e., time lags of one to four months after each month in the
irrigation season) did not yield any consistent results. For this reason,
a quarterly analysis of the data was undertaken as a means to further
explore this question.

-------
138
TABLE 24. SUMMARY OF THE MONTHLY REGRESSION ANALYSES: REGRESSION
COEFFICIENTS AND TESTS OF SIGNIFICANCE!/
Regression Coefficients (and Significance^/)
Net Salt
Net
Precipi-
tation
(Grand
Jet.)
Irrigation Variables
FD
SLd (SLf + DP)
January
1.92
(.01)
2.09
(.83)



February
.92
(.01)
.25
(.98)*



March
1.82
(.02)
6.45
(.66)



Apri 1
.18
(.46)
24.36
(.58)
-30.41
(.18)
-1.08
(.69)
-2.78
(.80)
May
.28
(.39)
-2.37
(.91)*
38.19*
(.64)
5.93
(.23)*
.70
(.01)*
June
.60
(.03)
-8.87
(.55)
40.24
(.61)
3.16
(.63)
-5.96
(.59)
July
.02
(.97)*
40.81
(.06)
-113.02
(.10)
-6.60
(.05)
-9.03
(.18)
August
1.11
(.01)
18.91
(.05)
-103.10
(.02)
-2.01
(.64)*
-11.29
(.11)*
September
1.11
(.01)
17.20
(.09)
-199.92
(.01)
-8.13
(.14)*
-29.49
(.01)
October
.74
(.16)*
33.09
(.01)
-184.60
(.19)
-6.25
(.36)
-21.98
(.30)
November
1.60
(.01)
.14
(.99)*



December
1.48
(.01)
12.27
(.26)*



^Coefficient for the two-variable case only.
^Calculated t values are used, where .02 d®Jotes a
ficient is significant at the 2 percent level. The notatio [
significant serial correlation at the 5 percent level.

-------
139
Part III: The Quarterly Analysis
Due to the extreme variability of the monthly data, it was conjec-
tured that monthly variables might not be appropriate for testing salt
loading effects caused by irrigation. If the incidence of irrigation in
the spring of each year is highly influenced by early season weather
conditions, perhaps the quantity of return flow water would be more con-
sistent from year to year if measured over a period of months. Thus, the
aggregated data were thought to be a better form in which to test the
plausibility of lagged effects.
The monthly data were aggregated into four three-month quarters,
roughly corresponding to the four seasons—spring, summer, fall and winter.
This is not entirely consistent with a normal irrigation cycle of seven
months in the Grand Valley, but the spring and summer quarters—April -
June, and July-September, respectively—account for a major proportion of
annual irrigation water use (refer to figure 16). The spring quarter en-
compasses preirrigation and early season irrigation while the summer
quarter essentially represents the period of peak water use. What little
irrigation that typically occurs in October was not considered important
enough not to omit it from the analysis. Alternative formulations of the
aggregated data may in fact be better than "seasonal" quarters, but time
and resource constraints did not permit a more detailed study of these
considerations.
The means, standard derivations and coefficients of variation for the
quarterly variables, summarized tn table 25, provide some new insight into
the nature of the problem. Net salt pickup is shown to be higher in the
last two quarters of the year and less variable 1n comparison with the
first two quarters. Net discharge is lowest in the spring quarter. This

-------
140
TABLE 25. MEANS, STANDARD DEVIATIONS AND COEFFICIENTS OF VARIATION
FOR THE QUARTERLY DATA
Variable
Mean
Standard
Deviation
Coefficient
of Variation
Net Salt Pickup:
January-March
April-June
July-September
October-December
Farm Deliveries:
April-June
July-September
Seepage:
Apri1-June
July-September
Deep Percolation (SLf + DP)
Apri1-June
July-September
118.48
114.19
126.52
180.90
Net Discharge:
January-March	25.43
Apri1-June	-116.24
July-September	- 99.86
October-December	9.90
Precipitation (Grand Junction):
January-March	1.72
Apri1-June	1.87
July-September	2.25
October-December	2.19
2.04
2.44
25.40
31.57
4.62
.03
72.14
83.05
68.09
95.17
30.99
52.33
43.32
48.08
.80
1.01
1.02
1.10
.33
.26
3.33
3.51
3.32
1.84
.61
.73
.54
.53
1.22
.45
.43
4.86
,47
.54
,45
,50
16
12
13
,11
.72
61.33

-------
141
seems contrary to the fact that crop consumptive use is highest in the
third quarter, and thus should be reflected by net discharge being the
lowest at this time. However, this is consistent with the hypothesis of a
time lag in irrigation return flow (about three months according to
Hyatt, [1970]), and it possibly accounts for the increased salt pickup
which seems to occur in the last two quarters of the year. This infer-
ence is bolstered by relatively high seepage and deep percolation in the
spring in contrast to farm deliveries which follow closely the pattern
of consumptive use (figure 16). As anticipated, the quarterly precipita-
tion data now show substantially less variation.
The results of the regression analyses are summarized in table 26.
In part A, the correlation coefficients are given; in part B, the esti-
mated b parameters and significance levels are shown; and in part C, the
regression runs of selected lagged variables are presented. Consistent
with the annual data, the time variable is negatively correlated with salt
pickup, and among the three irrigation variables only seepage is posi-
tively correlated with salts. The quarterly data do improve the signifi-
cance level of the regression coefficient for seepage; in this case both
the spring and summer parameters have appropriate signs and are significant
(the summer quarter parameter at 6 percent). Rainfall is not a particu-
larly meaningful variable in any quarter, whereas the flow variable is most
significant during the fall and winter, typically the low flow period of
the year.
In terms of a three-month lag, the only period in which significant
irrigation return flows are indicated to have a positive effect on salt
pickup occurs in the spring. Seepage loss in April through June has a
significant positive impact on net salt loading in the following quarter

-------
TABLE 26. SUMMARY OF THE QUARTERLY CORRELATION-REGRESSION RESULTS AND TESTS OF SIGNIFICANCE—^
Net Salt
Pickup
Net
Discharge
Precipitation
(Grand Jet.)
FD
Irri gati on-Vari ables
SLd
(SLf + DP) Time
A. Correlation Coefficients
January-March
April-June
July-September
October-December
B.
.67
.25



-.76
.37
.13
-.58
.35
-.65
-.14,
.42
.34
-.12
.48
-.13
-.17
.67
.19



-.36
s (and
level of significance)




1.08
22.23




(.01)
(.28)*




.59
10.69
51.56
5.16
-6.12

(.10)*
(.57)*
(.37)*
(.02)
(.28)

.67
21.98
87.13
2.80
-12.39

(.06)*
(.14)
(.14)
(.06)
(.14)

1.34
16.62




(.01)
(.40)*




January-March
April-June
July-September
October-Decembe r
Regression Coefficients (and significance level) for a three-month lag in Return Flows
April-June	4.61
(.55)*
July-September	-5.14	-7.27	3.70
(.66)	(.46) (.05)
October-December	16.10 -166.56 -2.95
(.36)	(.04) (.19)
January-March
(.14)
.55
(.92)*
-1.84
(.42)*
^Refer to footnotes [a] and [b], table 24.

-------
143
(July-September). In the same period deep percolation has the appro-
priate sign but is not significant. The relatively high salt pickup
which occurs in the last quarter of the year is not significantly related
to any single variable in the preceding period (except farm deliveries,
in which case the sign is negative).
In summary the quarterly analyses indicate some improvement over the
monthly data in that the effects of the irrigation variables appear to
be more accurately measured. On the basis of the results, the seepage
component of irrigation return flow is the only variable which is signifi-
cantly related to salt pickup. The regression coefficient relating summer
quarter salts to spring quarter seepage shows improved statistical signifi-
cance over the nonlagged regression, and therefore is an indication (not a
confirmation) of the presence of a lag in subsurface return flows. Whether
the appropriate lag is two, three or four months is a question that might
be answered by further study of the aggregation scheme.

-------
144
E. SUMMARY AND IMPLICATIONS
This chapter has reported an investigation of the sources of net
increase in dissolved solids which occurs each year in the Grand Valley
reach of the Colorado River. Earlier studies of this reach concentrated
on the identification of salt pickup mechanisms,emphasizing agricultural
causes which were believed to be responsible for nearly all the addition
to the rivers' salt load as it passed through the irrigated area. Con-
sequently, natural pickup mechanisms, which are known to contribute sig-
nificantly to the salt burden of streams in other regions of the Basin,
have not received the careful attention of researchers in the Grand
Valley.
The purpose of this inquiry was to focus on questions relating to
the incidence of salt pickup which have not been adequately considered
(or answered) in the previous investigations, but questions, nonetheless,
that have a definite bearing on the economics of proposed control pro-
grams. Two such questions of major importance are: (1) What is the
long term significance of salt pickup in the study reach (its magnitude,
trends, etc.,) in the absence of controls?, and (2) What are the likely
consequences, in terms of salt load reductions, that can be credited to
the various strategies, both direct and indirect, for controlling sa-
linity? The economic implications of adopting alternative controls will
not be meaningful or relevant to public policy unless these and related
questions are answered with some degree of confidence.
The analysis was formulated with the principal objective in mind of
testing hypotheses regarding (1) the nature of agricultural contributions
developed in the earlier work, and (2) the natural contributions which

-------
145
have been essentially overlooked in the previous investigations. The
statistical technique of least-squares regression was used to examine
the available historical data relating to the hydro-geologic-salinity
relationship of the study area. Data from one of the larger irrigation
systems, the Government Highline Canal, accounting for approximately 40
percent of total Valley diversions and irrigated acreage, were employed
to represent the impact of agricultural sources. The potential salt
pickup effects due to natural causes were evaluated with the aid of
historical precipitation records.
The findings of this analysis are not conclusive, but the results
raise a number of important policy implications. First, the absolute
volume of dissolved solids added to the river varies considerably from
year to year. During the period 1951-1972, the mean annual salt pick-
up in the study reach was about 540,000 tons. If an allowance is made
for the normal contribution of salts from the major ungaged (perennial)
tributaries in the reach, notably the Little Dolores River and Plateau
Creek, the mean is nearer to 440,000 tons, or 160,000 below the generally
accepted benchmark figure.
Second, the analysis of annual data indicates that net salt pickup
in the Grand Valley reach decreased during the study period. This de-
creasing trend was found to be statistically significant but inversely
related to irrigation water diversions and use. The simple fact that
salt load and river discharge is highly correlated suggests that natural
processes, whether they be explained by precipitation runoff or ground
water interchange phenomena, have apparently been greatly underestimated.
Although a precise estimate of the relative magnitudes of these sources
will require a more detailed analysis, the significant associations found
between precipitation and net pickup strongly suggest that the natural

-------
146
sources are responsible for more than just the 5 percent of the total
salt pickup in the reach estimated by the generally accepted authorities.
Third, on the basis of statistical tests, deep percolation on farms
was not found to be a very important source of salt pickup in irrigation
return flows. This finding is especially meaningful since on-farm controls
are believed to be the least-cost abatement strategy in the Upper Colo-
rado River basin. However, due to the nature of the data and assumptions
required to perform tests of this hypothesis, this conclusion should not
be construed as conclusive, but warrants furthermore careful examination.
Finally, the results implicate canal and lateral seepage as the ap-
parent cause of salt pickup via irrigation return flows. Regression re-
sults,using quarterly data, is the only case, however, where seepage was
found to be statistically significant and "positively" associated with
salt pickup. The magnitude of seepage-induced salts remains to be iden-
ti fied.
It was not the purpose of this study to estimate empirically the
actual amounts of salt loading by source. Regression analysis can gen-
erate such estimates, but this would require additional information.
Nonetheless, the regression coefficient can be interpreted as a "rate
of salt pickup" — the change in net salts per unit change in the ex-
planatory variable. In this form, the regression results are comparable
to the earlier field investigations which have reported experimental es-
timates of pickup rates by type of pickup mechanism. Selected results of
the regression runs are summarized in Table 27.
Net salt pickup associated with precipitation in a normal year (i.e.,
when precipitation is evaluated at its mean value), given the effects of
time, is shown (in Part A) to range from 122,000 to 299,000 tons. Con-
verting these estimates to tons per square mile of surface area in the

-------
TABLE 27. EMPIRICAL ESTIMATES OF RATES OF PICKUP AND THE MAGNITUDE OF TOTAL SALT CONTRIBUTIONS
FROM NATURAL AND MANMADE SOURCES
Explanatory ,
Variable (s)^
Significance
Level of b
Rate of Net,,
Salt Pickup-'
(Tons/Unit)
Range in
Values of
Predicted .
Salt Pickup
Low
Est. Mean High
	(1000 Tons)- - -
Est.
A. ANNUAL RELATIONSHIPS
Precipitation
Precipitation
Precipitation
Precipitation
Seepage
Seepage|Time
Time A,
(RD-ET)S/
Seepage
B. QUARTERLY RELATIONSHIPS
Seepage (Apr.-June)
Seepage (July-Sept.)
Seepage (July-Sept.)
(3 mo. lag)
.10
25,150/Inch
+
14,490
85
202
318
.04
26,270/Inch
+
11,040
122
211
299
.12
22,670 Inch
+
13,670
72
182
291
.10
26,300/Inch
+
15,130
90
211
332
.03
4.54/AF
+
1.80
171
283
395
.19
2.47/AF
+
1.82
40
154
267
.02
.06
.05
5.16/AF
2.80/AF
3.70/AF
+
+
1.95
1.40
1.78
82
44
61
131
84
117
181
133
173
a/ Results given for two and three variable cases. The slash (|) can be interpreted as
	"given the influence of time, etc., evaluated at its mean."
b/ Defined as the change in net salt pickup per unit change in the explanatory variable
(the regression coefficient ± its standard error).
c/ Found by multiplying the mean of the variable by its regression coefficient. The standard
error provides a high and low range on the estimate (a confidence interval of roughly 67
percent about the mean).
d/ (RD-ET) is a proxy for total irrigation return flows.

-------
148
reach, this amounts to 40.6 and 99.6 tons per square mile , respectively.
As pointed out in section B, this range brackets Elkins' [1976] estimate
of the surface runoff contribution of about 58 tons per square mile. The
estimated rate of salt pickup attributed to seepage ranges from about 2.5
to 4.5 tons per acre foot of seepage water on an annual basis. Using the
quarterly relationships, part B, the pickup rate appears slightly higher
for early season conditions and for a three month lag in the late season.
These rates of pickup for seepage waters are not much different from those
found by Kruse [1975] and Olsen [1976].
These inferences, drawn from the interpretation of the regression co-
efficients, tend to support the general validity of the conceptual model
and the analytical approach used in this study. However, in closing,
certain limitations of the results should be reviewed. One critical as-
sumption in the analysis is that a single irrigation district or contig-
uous canal and lateral system can adequately reflect the cropping patterns,
water use, seepage losses, etc., for the Grand Valley as a whole. Suf-
ficient evidence to substantiate the validity of this assumption is not
available. Secondly, because sound empirical data is generally unavail-
able except for the time period used in the analysis, there is a question
as to the long term representativeness of the sampled data. Lastly, the
conceptualization and formulations of the problem for analysis were, out
of necessity, nonrigorous and somewhat simplistic.

-------
149
Recommendations for Further Research
The findings reported in this chapter are not conclusive. In view
of the announced time schedule of the programs for control of irrigation
return flows in the Colorado River Basin, it is important that the rel-
ative importance of natural versus manmade sources of salinity be more
clearly delineated. Two lines of inquiry to this end are noted below.
A first step would be to apply the same type of statistical analysis
as was reported here to the Gunnison-Uncompaghre basin in western Colo-
rado. Data from the gaging station at the mouth of the Gunnison River
(at Grand Junction, Colorado), would provide the necessary information
on salt loading. The area is quite similar to the Grand Valley in
geology, climate and agricultural practices. The Uncompaghre and the
Grand Vallies together are thought to contribute a majority of salinity
from irrigation in the Upper Basin. Corroboration that natural sources
of salinity are more important than has been thought previously, as we
would anticipate from some of the relevant hydrologic data, would have
important implications for the salinity control program.
A second recommended extension of our study would be to analyze
daily observations. Since some natural sources (i.e., surface runoff)
have more rapid effects on water quality than return flow phenomena,
testing this hypothesis on the basis of available daily observations
might uncover relationships which are masked when longer time periods
are employed.
Finally, since we can claim neither statistics nor watershed hy-
drology as primary fields of expertise, such studies would probably be
more productive if such specialists were drawn into the study design.

-------
Section VII
APPENDIX
LITERATURE CITED
Anderson, J. C., R. J. Hanks, L. G. King, S. W. Childs, and J. R.
Cannon (1974). "An Evaluation of Farm Practices as a Means to Control
the Water Quality of Return Flow," Report No. 19, Agricultural Experiment
Station, Utah State University, Logan, Utah.
Bishop, A. A., M. E. Jensen, and W. A. Hall (1967). "Surface Irriga-
tion Systems," in Irrigation of Agricultural Lands, edited by Hagan, Haise
and Edminster, No. 11 in the "Agronomy" series, American Society of Agronomy,
Madison, Wisconsin.
	 , and H. B. Peterson (1969). "The Characteristics of
Pollution Problems of Irrigation Return Flow." Contract No. 14-12-408,
Federal Water Pollution Administration, U. S. Department of Interior.
Carter, H. 0. and G. W. Dean (1961). "Cost-Size Relationships for
Cash-Crop Farms in a Highly Commercial Agriculture," Journal of Farm Economics.
Vol. 43, May.
Day, L. M. (1963). "Use of Representative Firms in Studies of Inter-
regional Competition and Production Response," Journal of Farm Economics,
Vol. 45, No. 5.
Decker, R. S. (1951). "Progress Report on Drainage Project," Lower
Grand Valley Soil Conservation District, Mesa County, Colorado.
Draper, N. R. and H. Smith (1966). Applied Regression Analysis.
(First edition). John Wiley and Sons, Inc., New York.
El kin, A. D. (1976). "Grand Valley Salinity Study: Investigations
of Sediment and Salt Yields in Diffuse Source Areas, Mesa County, Colorado."
Review draft submitted to the State Conservation Engineer, Soil Conservation
Service, Denver.
Ezekiel, M. and K. A. Fox (1965). Methods of Correlation and Regression
Analysis (Fourth edition). John Wiley and Sons, Inc., New York.
Gray, S. L., J. R. McKean, J. Weber, and E. Sparling (1975). "Use of
the Colorado Input-Output Model to Analyze the Impacts of Alternative Economic
Scenarios on Output, Income, and Water Use and Employment in Colorado's
Economy." A Report to the Colorado Energy Research Institute, available
through the Department of Economics, Colorado State University, Fort Collins.
Gilley, J. R. (1968). "Intake Functions and Border Irrigation," M.S.
Thesis, Department of Agricultural Engineering, Colorado State University,
Fort Collins, Colorado.
150

-------
151
Gross, A. D. (1965). "Condemnation of Water Rights for Preferred
Uses—Replacement of Prior Appropriations?" Willamette Law Journal,
3(4): 263-283.
Hadley, R. F. (1976). Geological Survey, U.S. Department of Interior,
Denver Federal Center, Denver, Colorado (Personal communication).
Harmston, F. K. and R. E. Lund (1967). Application of an Input-Output
Framework to a Community Economic System. University of Missouri Studies,
Volume XLII, the University of Missouri Press, Columbia, Missouri.
Howe, C. W. and D. Y. Orr (1974). "Effects of Agricultural Acreage
Reduction on Water Availability and Salinity in the Upper Colorado River
Basin," Water Resources Research, V. 10, No. 5.
Hyatt, M. L., J. P. Riley, M. L. McKee, and E. K. Israel sen (1970).
"Computer Simulation of the Hydrologic-Salinity Flow System Within the Upper
Colorado River Basin." Utah Water Research Laboratory, College of
Engineering, Utah State University, Logan, Utah.
Iorns, W. F., C. H. Hembree, and G. L. Oakland (1964,65). "Water
Resources of the Upper Colorado River Basin." General report and appendix,
U.S. Geological Survey Professional Papers 441 and 442.
Jensen, M. E. and H. R. Haise (1963). "Estimating Evapotranspiration
from Solar Radiation." In annual conference proceedings, American Society
of Civil Engineers, Irrigation and Drainage Division.
Kelso, M. M., W. E. Martin, and L. E. Mack (1973). Water Supplies
and Economic Growth in an Arid Environment: An Arizona Case Study, The
University of Arizona Press, Tucson, Arizona.
Klapwyk, W. (1974). Manager, Grand Valley Water Users Association,
Grand Junction (Personal communication).
Kneese, A. V. and C. W. Schultze (1975). Pollution, Prices, and Public
Policy. Brookings Institute, Washington, D.C.
Kruse, E. G. (1975). "Alleviation of Salt Load in Irrigation Water
Return Flow of the Upper Colorado River Basin." Annual Progress Report,
Agricultural Research Service, U. S. Department of Agriculture, and U.S.
Salinity Laboratory, Riverside, California.
Leathers, K. L. and W. T. Franklin (1975). "Irrigation Water Infiltra-
tion and Salt Pickup Mechanisms: Grand Valley, Colorado," unpublished manu-
script, Department of Economics, Colorado State University, Fort Collins,
Colorado.
		(1976a). "The Economics of Managing Saline Irrigation
Return mows in the Upper Colorado River Basin: A Case Study of Grand Valley,
Colorado," Ph.D. dissertation 1n preparation, Department of Economics,
Colorado State University, Fort Collins, Colorado.

-------
152
	 (1976b). "Costs and Returns to Crop Farming in the
Grand Valley of Western Colorado." Manuscript in preparation, Department
of Economics, Colorado State University, Fort Collins.
Linsley, R. K., M. A. Kohler, and J. L. H. Paulhus (1958). Hydrology
for Engineers. McGraw-Hill, Inc., New York.
Martin, W. E. and H. 0. Carter. "A California Interindustry Analysis
Emphasizing Agriculture," Giannini Foundation Research Report No. 250,
University of California, Berkeley, 1962.
Mesa County Assessor's Office (1976). Grand Junction, Colorado
(personal communication).
Moore, C. V. (1972). "On the Necessary and Sufficient Conditions for
a Long-Term Irrigated Agriculture," Water Resources Bulletin, Vol. 8,
No. 4, August.
Olsen, S. R. (1976). Research Leader, Soil Fertility and Management.
Agricultural Research Service, U. S. Department of Agriculture, Colorado
State University, Fort Collins (personal communication).
Oyarzabal-Tamargo, Francisco (1976). "Economic Impact of Saline
Irrigation Water, Mexicali Valley, Mexico." Ph.D. dissertation, Department
of Economics, Colorado State University, Fort Collins, Colorado.
Peskin, H. M. and E. P. Seskin (1975). Cost Benefit Analysis and Water
Pollution Policy. The Urban Institute, Washington. D.C.
Public Law 89-321 (1965). Food and Agriculture Act of 1965, section
602, 79 Stat. 1206, as amended 7 U.S.C.A. 1838.
Rai, D. and W. T. Franklin (1973). "Program for Computing Equilibrium
Solution in CaC03 and CaS04 Systems from Irrigation Water Compositions."
CUSUSWASH Water Management Technical Report No. 29, Colorado State University,
Fort Collins.
Richardson, H. W. (1972). Input-Output and Regional Economics. Red-
wood Press Limited, Trowbridge, Wiltshire, England.
Robinson, C. W. (1969). "Reclamation of Saline-Sodic Soils in the
Upper Colorado River Basin." Bulletin 535-S, Colorado State University
Experiment Station, Fort Collins, Colorado.
and W. T. Franklin (1973). "A Survey of Salt Content
of Irrigation Runoff," Colorado State University Experiment Station,
PR 73-58, Fort Collins.
Schumm, S. A. and G. C. Lusbey (1963). "Seasonal Variation of Infiltration
Capacity and Runoff on Hillslopes in Western Colorado." Journal of Geo-
physical Research, Vol. 68, No. 12.

-------
153
Skogerboe, G. V. and W. R. Walker (1972). "Evaluation of Canal Lining
for Salinity Control in Grand Valley." Environmental Protection Technology
Services, EPA-R2-72-047, Office of Research and Monitoring, U.S. Environ-
mental Protection Agency, Washington, D.C.
	 , W. R. Walker, J. H. Taylor, and R. Bennett (1974a).
"Evaluation of Irrigation Scheduling for Salinity Control in Grand Valley,"
EPA-660-2-74-052 Series, Office of Research and Development, U.S. Environ-
mental Protection Agency, Washington, D.C.
, W. R. Walker, R. S. Bennett, J. E. Ayars, and J. H.
Taylor (1974b). "Evaluation of Drainage for Salinity Control in Grand
Valley," EPA-660-2-74-084 Series, Office of Research and Development, U.S.
Environmental Protection Agency, Washington, D.C.
Trelease, F. J. (1960). Severance of Water Rights from Wyoming Lands,
Wyoming Legislature Resource Committee, Report No. 2, Cheyenne, Wyoming.
Tweeten, L. (1971). Foundations of Farm Policy. University of
Nebraska Press, Lincoln, Nebraska.
Udis, B. et al. (1973). "The Interrelationship of Economic Development and
EnvironmentaTT)uality in the Upper Colorado River Basin: An Interindustry
Analysis." Report to U.S. Department of Commerce, Economic Development
Administration, University of Colorado, Boulder.
U.S. Bureau of Reclamation, Department of Interior (1967). "Quality
of Water of Colorado River Basin." Progress Report No. 3, Salt Lake City,
Utah.
			 (1972). "Colorado
River Water Quality Improvement Program," Denver, Colorado.
		 	(1973). "Quality
of Water, Colorado River Basin," Progress Report No. 6, Washington, D.C.
			(1974a). "Colorado
River Water Quality Improvement Program," Status Report, Denver, Colorado.
	^			(1974b). "Irrigation
Management Services—Annual Report: Grand Valley, 1973." Western Projects
Office, Grand Junction, Colorado (unpublished).
		_______ (1975). "Colorado
River Improvement Program, Title II: Salinity ControlAct." Progress
Report, P.L. 93-320, Denver, Colorado.
U.S. Department of Agriculture and Colorado Agricultural Experiment
Station (1957). "Soil, Water, and Crop Management Studies 1n the Upper
Colorado River Basin," Annual Research Report, Colorado State University,
Fort Coll ins.

-------
154
and the Colorado Water Conservation
Board (1965). "Water and Related Land Resources, Colorado River Basin
in Colorado." Denver, Colorado.
	, Economic Research Service (1975).
"Demand and Price Situation," Washington, D.C.
U.S. Department of Commerce (Selected years, 1951-1972). "Climato-
logical Data: Colorado Annual Summaries." National Oceanic and At-
mospheric Administration, Environmental Data Service, National Climatic
Center, Asheville, North Carolina.
U.S. Environmental Protection Agency, Regions 8 and 9 (1971). The
Mineral Water Quality Problem in the Colorado River Basin, 5 volumes
(Summary report and appendices A,B,C,D), San Francisco, California.
U.S. Geological Survey (Selected years, 1952-1973). "Water Resources
Data for Colorado, . . . and Utah, Parts I and II." Surface and water
quality records, Denver, Colorado, and Salt Lake City, Utah. U.S.
Government Printing Office, Washington, D.C.
Utah State University, Water Research Laboratory (1975). "Colorado
River Regional Assessment Study, Parts 1-4," prepared for the National
Commission on Water Quality, Logan, Utah, June 1975 (review draft).
Varble, G. L. (1969). "Grand Valley Trade Area, 1964." M.S. thesis,
Department of Economics, Colorado State University, Fort Collins, Colorado.
Walker, W. R. and G. V. Skogerboe (1971). "Agricultural Land Use
in the Grand Valley," Department of Agricultural Engineering, Colorado
State University, Fort Collins, Colorado.
Walker, W. R. (1976). Department of Agricultural Engineering,
Colorado State University, Fort Collins. (Personal communication.)
Wiscombe, E. (1975). Projects Manager, Upper Colorado Region,
Western Colorado Projects Office, U.S. Bureau of Reclamation, Grand
Junction, Colorado (personal communication).
Young, R. A. et al. (1975). "Economic and Institutional Analysis
of Colorado Water Quality Management," OWRR Project No. 13-042-Colorado,
Environmental Resources Center, Colorado State University, Fort Collins,
Colorado (Completion Report Series, No. 61).

-------
Table 1. LINEAR PROGRAMMING TABLEAU FOR CROP PRODUCTION MODEL I: 40 ACRE FARMS
Item
Unit
Crop Production Activities and Irrigation Processes
Corn Small Grains Sugar Beets Perm. Pasture Alfalfa
5	B" 	A		
B
TT
B
B
Constraint Levels
Maximum Minimum
Net Revenue $/Acres
Irrigable Land	Acres
Crop Acreage
Corn	Acres
Small Grains	Acres
Sugar Beets	Acres
Perm. Pasture	Acres
Alfalfa	Acres
Irrigation Water AF/A
Deep Percolation AF/A
107 103
1.00 1.00
1.00 1.00
3.74 3.02
.73 .16
9 5 48 44
1.00 1.00 1.00 1.00 = 2,434
- 730
1.00 1.00
4.80 3.36
.55 .19
1.00 1.00- 2,054
4.80 3.36-10,909
.38 .02
> 0
289

-------
Table 2. LINEAR PROGRAMMING TABLEAU FOR CROP PRODUCTION MODEL II: 80 ACRE FARMS
Item
Unit
Crop Production Activities and Irrigation Processes	
Corn Small Grains Sugar Beets Perm. Pasture Alfalfa
~R B" A B ~X B ~S B A B
Constraint Levels
Maximum Minimum
Net Revenue
Irrigable Land
Crop Acreage
Com
Small Grains
Sugar Beets
$/Acres 113 109
Acres 1.00 1.00
Acres 1.00 1.00
Acres
Acres
Perm. Pasture	Acres
Alfalfa	Acres
Irrigation Water	AF/A
Deep Percolation	AF/A
3.74 3.02
.73 .16
256 251
1 .00 1.00
12
50 46
1.00 1.00 1.00 1.00 1.00 1.00 = 4,932
<	2,601
<
1.00 1.00
5.28 4.32 4.80 3.36
.64 .07 .55 .19
f 408
<
1.00 1.00 - 3,114
4.80 3.36 - 21 ,112
.38 .02 - 0
425

-------
Table 3. LINEAR PROGRAMMING TABLEAU FOR CROP PRODUCTION MODEL III: 130 ACRE FARMS
Item
Unit
Crop Production Activities and Irrigation Processes	
Corn Small Grains Sugar Beets Perm. Pasture Alfalfa
"A B" A B "T	A B A B
Constraint Levels
Maximum Minimum
Net Revenue $/Acres 138
Irrigable Land	Acres 1.00
Crop Acreage
Corn	Acres
Small Grains	Acres
Sugar Beets	Acres
Perm. Pasture	Acres
Alfalfa	Acres
Irrigation Water	AF/A 3.74
Deep Percolation	AF/A .73
134 80 76
1.00 1.00 1.00
1.00 1.00
1.00 1.00
3.02 2.98 2.31
.16 .66 .06
294
1.00
5.28
.64
289
1.00
1.00 1.00
4.32
.07
21 17
1.00 1.00
1.00 1.00
4.80 3.36
.55 .19
93 89
1.00 1.00=	16,624
-	8,048
<
<	696
<
1.00 1.00 -	9>040
4.80 3.36 -	69,307
.38 .02	- 0
1,259
1,261

-------
Table 4. LINEAR PROGRAMMING TABLEAU FOR CROP PRODUCTION MODEL IV: 210 ACRE FARMS
Item
Unit
Crop Production Activities and Irrigation Processes	
Corn Small Grains Sugar Beets Perm. Pasture Alfalfa
~K B A B A B A B A B
Constraint Levels
Maximum Minimum
Net Revenue
Irrigable Land
Crop Acreage
Corn
Small Grains
Sugar Beets
$/Acres 171 167
Perm. Pasture	Acres
Alfalfa	Acres
Irrigation Water	AF/A
Deep Percolation	AF/A
92
88
413
408
35
31
Acres 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Acres 1.00 1.00
Acres
Acres
.73
1.00 1.00
1.00 1.00
.16 .66 .06
.64
.07
1.00 1.00
3.74 3.02 2.98 2.31 5.28 4.32 4.80 3.36
•55 .19
116 112
1.00 1.00 =	9,523
f	4,248
<
<	2,218
<
1.00 1.00 -	3,946
4.80 3.36 -41 ,063
.38 .02	- 0
664
626

-------
Table 5. LINEAR PROGRAMMING TABLEAU FOR CROP PRODUCTION MODEL V: 370 ACRE FARMS
Item
Unit
Crop Production Activities and Irrigation Processes	
Corn Small Grains Sugar Beets Perm. Pasture Alfalfa
"A B A B A B A B A B
Constraint Levels
Maximum Minimum
Net Revenue
Irrigable Land
Crop Acreage
Com
Small Grains
Sugar Beets
$/Acres 162 158 91 87
Acres 1.00 1.00 1.00 1.00
Acres 1.00 1.00
Acres
Acres
Perm. Pasture	Acres
Alfalfa	Acres
Irrigation Water	AF/A
Deep Percolation	AF/A
1.00 1.00
3.74 3.02 2.98 2.31
.73 .16 .66 .06
455 450 33 29 115 111
1.00 1.00 1.00 1.00 1.00 1.00-13,960
- 8,270
<
1.00 1.00
5.28
.64
4.32
.07
1.00 1.00
4.80 3.36
.55 .19
- 4,313
<
1.00 1.00 - 5,048
4.80 3.35 -58,843
.38 .02 - 0
807
295

-------
Table 6. MONETARY TRANSACTIONS FOR THE GRAND VALLEY TRADE AREA, 1970
PRODUCTION AND PROCESSING SECTORS
Sector Identification
LOCAL f 1NA. OEMANCS
. Wlit LIVESTOCK
rcc i;vfvoct
. MIRY
. FORAGE AND fEED CROPS
FOOD AND fULD CROPS
fRUlT A NO SPCCIAUTt CROPS
OTHER AGRICULTURE
fORESTRT
MiNlNC
FOOD PROCESSING
MANUFACTURING
WHOitSAit TKADi
8£7A|L TRADE
AGRICULTURAL services
OTHER SERVICES
TRANSPORTATION
UTILITIES
CONSTRUCTION
nwNCt
LOCAL 60VE»MMENT
HOUSEHOLDS
78
15 30
2560 17
19453 1467
201 0
1101
1483 31 6
45639 *6S0
2390
3381
3404
2*34
m
4636
1995
4 064 0
1668
1740
1406
2711C
2347
WO
1007
21 30
(780
1 5?8
12000
592R
544C
492' 0
?34
16210
7140
12700
7 970
33'00
1 7038
31520
T01*L
GROSS
X'TPUT
2623
S4S
;ea
6*70
1293!
3360
44R1
23100
i.5?be
290! 3
53S6
25 7 S
435,69
5917
215'6
8€',C
244bJ
100
37527
187S9
2^39
7?5C
11 I'j6
30045
5«6
33S8
6 >43
2879
11 £291
7203
2"050
vac
36576
i 36
' "321
6C181
i7o«;
24716
7232
e:?6
5603
1S5C
5899
138273
23846
52307
10*97
91714
5"29
74107
48118
39403
9P235
7 1 967
58252
4' 344 0
CEPSECJATION ALLOWANCES
nrvfWTORi' omfTioM
STA'E I FEDERAL GOVERNMENTS
IMPORTS
TOTAL IMPJTS
3495
3799 73 3*4 S12 1116 *W
39452 4622 3479 723? 827< 5603
U214 597
3100 960
1890 2230
2335
1433
4219
405 42963 2862 25651
S899 133666 23846 52307
1120	3378
3950	' 058?
1450	1884
8310	1 1601
30697	91714
4380
12650
1660
1174?
3910
7590
39403 98235
93188
61740
51820
4i?Q74
40750
S3S75
16399 1 CI 222
1442-7 37822?
666541 1797456
cn
o

-------
Table 7. DIRECT INPUT COEFFICIENTS (Based on Table 6)
PftQOUCTlON AND MOCCSS1NC UCTOftS
lOCAl FINAL DtKANDS
OUTSJDf F J MAI D£H»«)S

uoiur
IfVfcntOry
Stite-Federjl

mxal
OEMANOS
MOSS
OUTPUT


z
3
4
5
1

9
9
10
11
12
13
14
15
16
17
19
19
20
21

4 -i-JLi-
Goverrwfm



1. RANK IJVESTOCI
GSM
• J 68 3




.1097


.1519



.0010






.0030
.0016

.0104
CB70
.0451
.0219
2. HO LIVESTOCK









0292














-0167
.0061
.0033
3. OAIftT
.003*
.0140




¦ 1072


.1014










.0001


.0002
.oois
.0008
.0019
«. rtmu wo fir cws
.0)26
.>427
¦ 0204
.0007


¦ 0221


¦ 0094



.OOM






.0001


.0012
. 0098
.0051
.0040
S. FOOO AND FIEIO CMTS

0083







.0943










.0002


.0062
-01 56
.0092
.0046
*¦ rauiT mo srfCiALiTi c*o»s
.0034

.0197






.0192


.0001







.OOM



.0126
0061
.0031
'. OUCt MAIClHTURf


.0003
.0007
.0004

.0009


.0397










.0009



-0021
0010
.0011
FQitSTRV
.0002









.0549









¦ 0003



.0069
0043
.0033
». mint








.1234
.0001
.0)90
.0009
. 000}

.0009

.0*09
.0140
.0004
.0079
.0037
.0450
.046)
.35 70
1353
1730
.0769
10. roco MOCCSSINS

.1247
.013?



.09*9


.0051


.0291
.0031
-0122




.0021
¦ 0290

.0226
-0003
0194
.0108
.0133
II. MMtlTACTURIW
.0106
.0012
.0121
-0S12
.0731
.0192
¦ 02JC
-0205
.0139
.0123
.0139
.0010
.0349
.0109
.OOM
1U7
.006*
.0219
-004]
¦ 0392
.0143
.0029
QUO
.0111
0670
.0409
.0291
11. WOUSAU TWOC
.0099
.OOM
.0109
¦0099
.0120
.0077
.0092
.0099
.0099
.0097
.0023
.002Q
.0121
.0093
.0099
.0113
.0021
.0035
.0012
.0081
.0204
.QMS
-1192
-0022
0267
.0256
.0171
U. *nAIl TMX
.0243
oon
.0)42
.0994
.OMS
.0299
01S9
.0190
.0029
.0097
.0099
.0041
.007)
.0107
.0099
0229
.0037
.0090
.0053
.0031
.1192
.0116
-1770
.0013
0759
.0534
.0510
i«. AotiaanjKM. sctqfjcts
.0011
.0127
.1049
.1037
.0727
.4914
.0239













.0009



0003
.0002
.0029
is. otncr unices
.0014
.0019
.0009
.0119
.0109
.0034
.0092
.0020
.00)4
-0099
.0093
.0109
.0191
.0079
.030*
.0347
.01»
.0074
.0119
.0344
¦ 0392
.0007
.0231
.0(17
.1165
.0701
.0412
H. TtAttSMKTATIOM
.OOM
.0700
.owo
.0227
.0194
.0009
.0219

.0999
.0019
.0272
.0993
.0371
.0030
.00(9
.0091
.0020
.0130
.0008
.0292
-0173
.0009

.0010
.0594
.0296
.0269
1?- ITTHITICS
.0000
.0012
.0199
.0390
.0092
.0049
.0123

.0192
.0134
.0199
.0114
.0320
.0274
.0599
.003*
.0932
.0039
-0172
.0173
.0307
.0003
.1224
.0043
0088
-0)55
.0219
X. CORSnUCTIOK








.0001
.0009
.0049
.0099
.0101

¦ 0190
.0041
.011)
-2135
.0028
.0399
.0193
.3375
.2355
.0167
-0226
.0918
.0547
i«. fimjcc
.0900
.0130
.0239
.0349
.0099
.0094
.0139
.0219
.00*9
.0030
.0093
.Ot 30
.090%
.0149
.0240
.0122
.0111
.0050
.0234
.1194
.0910
.07)0

.0249
.0349
.0259
.0400
20. LOUt tOVOMCMT
.0*49
.009
.0940
.0309
.0499
.0399
.0200
.0293
.0107
.0133
.0209
.0199
.021)
.0191
.0190
.0192
.0790
.0014
.0095
.0209
.0412


.1294
.0090
.0371
.0324
21. MUSCKlLK
.4*31
.2490
.3091
.4020
.4199
.3234
.2910
.9391
.3301
.1990
.2290
.3379
.4491
.1722
.39S9
3784
.3097
.2335
.7349
.4509
.0792
.01)1

.3101
.04 5?
1096
.2300
TOTAL AMtt
.7949
•70
.74*4
.7791
.ta?z
.9910
.7932
.9341
.9972
.7212
3999
.9199
.9904
.9413
.9499
.9302
.9293
.9991
•120

.4963
.4447
.4055
.9212
.7649
.7599
.6906
KPftEClATlO* AU0MCC9
COM
.0117
.1449
.004
.OStO
.0332
.0)79
.0309
.0939
.0290
.0149
.OMS
.03(9
-0990
.0991
.0*12
.09*2
.0131
.0313







.0227
jurwrotr oenmoM
.01"







.0232
.0403
.0274
.1207
.1194

.1707

.1929
.1324


.2261





.0300
SW7I t fCOCMt flWDWKvn
.0129
000?
.0049
.0997
.0090
.0023
.0012
.2*9)
.014)
.Ot>9
.0007
.0172
.0209
.009*
.0094
.0349
.0299
.0025
¦ 0959
-007)
)496


.0678
.0130
.0246
.059)
twain
.0*93
.0191
.10t«
.vm
.1391
.0835
.19*7
.0997
.3219
.1200
490*
.7707
.12*7
.392*
-2193
.2441
.1993
.2839
.0900
.2263
.1250
.5533
. 1945
0110
.2221
.2165
.2104
1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
1.0000 >0000 1.0000

-------
Table 8. DIRECT AND INDIRECT TRANSACTIONS PER DOLLAR OF FINAL DEMAND (AND OUTPUT)

Sector Identification
1
2
3
4
5
6
7
8
PRODUCTION AND
9 10
PROCESSING SECTORS
11 12 13
14
15
16
17
18
19
20
21
1.
RANGE LIVESTOCK
1.1007
.2156
.0155
.0086
.0092
.0087
.1461
.008.1
.0060
.1853
.0044
.0060
.0126
.0077
.0085
.0068
.0058
.0050
.0106
.0090
.0136
2.
FED LIVESTOCK
.0009
1.0041
.0018
.0008
.0008
.0008
.0034
.0007
.0006
.0263
.0004
.0006
.00)4
.0006
.0009
.0006
.0005
.0004
.0010
.0008
.0012
3.
DAIRY
.0080
.0327
1.0081
.0037
.0039
.0036
.1225
.003-4
.0025
.1145
-0018
.0025
.0062
.0029
.0040
.0029
.0024
.0021
.0044
.0039
.0057
4.
FORAGE AND FEED CROPS
.0369
.3742
.0233
1.0024
.0016
.0041
.0322
.001 0
.0007
.0280
.0005
.0007
.0017
.0076
.0011
.0008
.0007
.0006
.0013
.0011
.0016
5.
FOOD AND FIELD CROPS
.0031
.0222
.0062
.0028
1.0030
.0028
.0115
.0027
.0020
.0883
.0014
.0020
.0048
.0023
.0031
.0022
.0019
.0016
.0034
.0030
.0044
6.
FRUIT AND SPECIALITY CROPS
.0059
.0093
.0206
.0020
.0022
1.0020
.0092
.001 9
.0014
.0511
.0010
.0014
.0035
.0016
.0021
.0016
.0014
.0012
.0025
.0021
.0032
7.
OTHER AGRICULTURE
.0022
.0074
.0037
.0026
.0024
.0019
1.0064
.001 9
.0014
.0412
.0010
.0014
.0028
.0015
.0019
.0015
.0013
.0011
.0024
.0020
.0031
fl.
FORESTRY
.0032
.0039
.0038
.0051
.0064
.0036
.0038
1.0029
.0026
.0033
.0569
.0019
.0041
.0031
.0017
.0080
. 001 9
.0030
.0025
.0043
.0027
9.
MINING
.0093
.0101
.Q094
.0108
.0102
.0089
.0087
.007 7
1.1475
.0083
.0236
.0070
.0103
.0074
.0096
.0087
.0552
.0267
.0107
.0195
.0115
10.
FOOD PROCESSING
.0343
.1623
.0721
.0314
.0336
.0310
.1340
.0297
.0219
1.0434
.0159
.0219
.0554
.0254
.0355
.0251
.0211
.0181
.0385
.0335
..04 90
11.
MANUFACTURING
.0484
.0644
.0648
.0882
.1098
.0608
.0635
.047 3
.0435
.0547 1
.0337
.0308
.0693
.0519
.0271
.1406
.0304
.0506
.0393
.0730
.0404
12.
WHOLESALE TRADE
.0364
.0326
.0342
.0307
.0380
.0335
.0312
.028=1
.0235
.0315
.0148
1.0189
.0341
.0255
.0260
.0302
.0184
.0181
.0290
.0318
.0350
13.
RETAIL TRADE
.1521
.1604
.1384
.1761
.1586
.1370
.1281
. 1264
.0844
.1185
.0648
.0857
1.1147
.0910
.0927
.1129
.08 07
.0731
.1460
.1185
.1800
14.
AGRICULTURAL SERVICES
.0171
.0633
.1183
.1061
.0752
.4539
.0465
. 002*1
.0016
.0476
.0011
.0016
.0034
1.0024
.0022
.0018
. 001 5
.0013
.0028
.0023
.0035
15.
OTHER SERVICES
.0547
.0608
.0489-
.0602
.0620
.0516
.0518
.0460
.0390
.0498
.0308
.0469
.0642
.0410
1.0666
.0742
.04 5 7
.0377
.0685
.0846
.0706
16.
TRANSPORTATION
.0457
.0678
.0902
.0562
.0521
.0322
.0604
.028.1
.0918
.0406
.0450
.1090
.0678
.0251
.0273
1.0648
.0268
.0369
. 0350
.0608
.0423
17.
UTILITIES
.0506
.0682
.0623
.0857
.0551
.0592
.0589
.0407
.0500
.0570
.0403
.0434
.0761
.0601
.0936
.0419
1 .0876
.0310
.0702
.0656
.0644
18.
CONSTRUCTION
.0367
.0360
.0328
.0329
.0352
.0312
.0301
.0300
.0240
.0293
.0236
.0305
.0441
.0230
.0380
.0318
.04 02
1.3577
.0417
.0819
.0471
19.
FINANCE
.1554
.1451
.1235
.1348
.1139
.1073
.1144
.113-2
.0742
.1003
.0587
.1127
.1432
.0839
.0975
.0907
.0848
.0676
1.1400
.2172
.1459
20.
LOCAL GOVERNMENT
.1299
.0962
.1496
.0957
.1050
.0971
.0929
.074 7
.0507
.0894
.0506
.0541
.0746
.0573
.0622
.0595
.1157
.0328
.0717
1.0762
.0779
21 .
HOUSEHOLDS
.9623
.98 99
.7851
.8571
.9486
.8427
.7992
.8642
.6428
.7486
.4611
.6380
.8407
.6377
.6657
.7099
.6130
.5286
1 .1339
.9126
1.4569

HOUSEHOLDS (ADJUSTED)
.8924
.9858
.7788
.8550
.9458
.8410
.7941
.861 7
.5602
.7175
.4461
.6262
.7542
.6362
.6241
.6667
.5636
.38 9 3
.9947
.8480
1.0000
TOTAL MULTIPLIER
2.9137 3.6266 2.8127 2.7941 2.8270 2.9761 2.9548 2.4609 2.3118 2.9567 1.9314 2.2169 2.6350 2.1591 2.2674 2.4168 2.2312 2.2953 2.8552 2.8038 2.2599

-------