DEVELOPMENT OF A GEOMORPHIC MODEL
TO PREDICT EROSION OF PRE-DAM COLORADO RIVER TERRACES
CONTAINING ARCHAEOLOGICAL RESOURCES
BOR Cooperative Agreement No. 1425-98-FC-40-22610

Prepared for
GRAND CANYON MONITORING AND RESEARCH CENTER
2255 North Gemini Drive, Room 341
Flagstaff, Arizona 86001
(520) 556-7094

Edited and compiled by
Kate S. Thompson and Andre R. Potochnik
Prepared by
Kate S. Thompson
Andre R. Potochnik
Ron Ryel
Gary O'Brien
Lynn A. Neal
Technical Editor
Jean H. Ballagh
SWCA, INC., ENVIRONMENTAL CONSULTANTS
114 North San Francisco Street, Suite 100
Flagstaff, Arizona 86001
(520) 774-5500

SWCA Project No. 2362-1383
SWCA Cultural Resources Report No. 99-257
February 18, 2000

TABLE OF CONTENTS
List of Figures ..................................................................................................................................v
List of Plates .................................................................................................................................. vi
List of Tables ................................................................................................................................ vii
Acknowledgments.......................................................................................................................... ix
Abstract .......................................................................................................................................... xi
1. INTRODUCTION by Kate S. Thompson and Andre R. Potochnik ........................................1
STUDY OBJECTIVES........................................................................................................3
Significance of Work ...............................................................................................3
Specific Objectives and Hypotheses........................................................................3
Value of Anticipated Results ...................................................................................4
BACKGROUND .................................................................................................................4
Previous Archaeological and Geomorphic Studies..................................................4
Rationale for This Study ..........................................................................................7
2. METHODS by Kate S. Thompson and Andre R. Potochnik..................................................11
SELECTION OF STUDY SITES......................................................................................11
TESTING THE HYPOTHESES........................................................................................11
The Null Hypothesis ..............................................................................................13
The Climate-Variation Hypothesis ........................................................................16
The Base-Level Hypothesis ...................................................................................21
DEVELOPMENT OF A PREDICTIVE GEOMORPHIC MODEL .................................22
The Small-Catchment Geomorphic System in the Grand Canyon ........................23
Developing the Matrix of Geomorphic Conditions ...............................................25
Intrinsic Factors of Catchments .................................................................25
Extrinsic Factors of Catchments ................................................................27
STEPS IN ASSEMBLING THE PREDICTIVE MODEL................................................28
Statistical Model Approach....................................................................................28
Mathematical Model Approach .............................................................................30
Model 1 ......................................................................................................31
Model 2 ......................................................................................................32
DEVELOPMENT OF THE CLIMATE PARAMETER FOR THE GEOMORPHIC
MODEL .............................................................................................................................35

iv

Table of Contents

3. RESULTS AND DISCUSSION by Kate S. Thompson, Andre R. Potochnik,
Ron Ryel, and Lynn A. Neal....................................................................................................39
HYPOTHESIS TESTING .................................................................................................39
Null Hypothesis .....................................................................................................39
Climate-Variation Hypothesis ...............................................................................50
Base-Level Hypothesis ..........................................................................................53
THE GEOMORPHIC MODEL .........................................................................................93
Geomorphic Settings..............................................................................................93
Alluvial Fan Setting ...................................................................................93
Tributary Plain Setting...............................................................................94
Talus Slope Setting ....................................................................................95
Debris Lobe Setting ...................................................................................95
Dune Field Setting .....................................................................................96
The Process-Based Conceptual Model ..................................................................96
The Process of Channel Initiation..............................................................97
Diffusion Capacity of the Terrace "Sponge" .............................................99
The Predictive Geomorphic Model......................................................................100
Model 1 Results ...................................................................................................102
Model 2 Results ...................................................................................................104
Integrating a Climatic Parameter into the Model.................................................109
4. CONCLUSIONS AND RECOMMENDATIONS by Kate S. Thompson and
Andre R. Potochnik..........................................................................................................114
RESULTS AND CONCLUSIONS OF THE CURRENT STUDY.................................114
Hypotheses...........................................................................................................114
Predictive Mathematical Model and Mitigation Measures..................................115
RECOMMENDATIONS FOR FUTURE WORK ..........................................................119
REFERENCES...........................................................................................................................121
APPENDIXES
A.
B.
C.
D.
E.

DATA SHEETS (2 pages)
SITE DESCRIPTIONS AND VULNERABILITY RATINGS (36 pages)
SUMMARY OF CATCHMENT DATA USED IN MODEL (8 pages)
AIR PHOTOS OF GRAND CANYON CATCHMENT AREAS (44 pages)
AIR PHOTOS OF CATARACT CANYON CATCHMENT AREAS (3 pages)

Table of Contents

v

List of Figures
1.1.
1.2.

Areas showing concentrations of archaeological sites within geomorphic reaches ............5
Schematic cross section of river terrace sequence with various geomorphic features ........6

2.1.

General climatic zones of the southwestern Colorado Plateau based on physiography and
elevation.............................................................................................................................18
Location of weather stations from which 64-year daily precipitation data were
compiled.............................................................................................................................19
Schematic illustration of the five geomorphic settings along the Colorado River ............24
Schematic illustration showing process of channel integration with the river ..................26
Flow chart showing inter-relationships of geomorphic factors and model
development.......................................................................................................................34

2.2.
2.3.
2.4.
2.5.
3.1.
3.2.
3.3.
3.4.
3.5.
3.6.
3.7.
3.8.
3.9.
3.10.
3.11.
3.12.
3.13.
3.14.
3.15.
3.16.
3.17.
3.18.
3.19.
3.20.

Degree of arroyo development since 1965 based on air photo analysis at 23
catchments..........................................................................................................................40
Schematic cross section of river terrace sequence, Cataract Canyon ................................46
Diagnostic flaked stone tools from the Rapid 4 site ..........................................................47
Relationships of monsoon precipitation events between equal periods in pre-dam
time and post-dam time......................................................................................................52
Historical Grand Canyon photos showing elevation of sand bars below archaeological
sites 55
Repeat photos showing change at "old Unkar camp" ........................................................57
Repeat photos showing change at Granite Park recirculation zone...................................61
Repeat photos showing effects of annual flood at Cross Canyon recirculation zone........63
Repeat photos showing effects of annual flood on Rapid 12 channel margin bar.............65
Repeat photos of Teapot Canyon area showing cycle of periodic deposition from
annual flood and subsequent gully erosion from monsoon season....................................67
Repeat photos showing gully infilling at Brown Betty camp, Cataract Canyon ...............69
Cut-and-fill map for the Rapid 4 study area showing changes in topography following
the 1999 annual flood.........................................................................................................71
Topography of the Rapid 4 study area following the 1999 annual flood ..........................73
Cross sections at four Rapid 4 locations showing elevation before and after the
1999 annual flood ..............................................................................................................75
Cut-and-fill map for the Cross Canyon study area showing changes in topography
following the 1999 annual flood ........................................................................................77
Topography of the Cross Canyon study area following the 1999 annual flood ................79
Cross sections at four Cross Canyon locations showing elevation before and after the
1999 annual flood ..............................................................................................................81
Cut-and-fill map for the Rapid 12 study area showing changes in topography following
the 1999 annual flood.........................................................................................................83
Topography of the Rapid 12 study area following the 1999 annual flood ........................85
Cross sections at four Rapid 12 locations showing elevation before and after the
1999 annual flood ..............................................................................................................87

vi

Table of Contents

3.21.
3.22.
3.23.
3.24.
3.25.
3.26.
3.27.
3.28.
3.29.
3.30.
3.31.
3.22.

Percent of surveyed sites (n=119) that support or negate the current base-level
hypothesis ..........................................................................................................................90
Number of sites containing 1996, 1983, or pda flood deposits .........................................91
Percent of catchments that exhibit varying degrees of eolian activity...............................92
Processes causing stream integration with river ................................................................98
Processes that resist erosion on terraces by increasing diffusivity ..................................101
Relation of two runoff parameters, peak flow (Qp) and total runoff volume (Q),
as applied to the uppermost terrace at each site...............................................................103
Vulnerability of uppermost terrace versus gully depth ratio using Model 1 ...................103
Vulnerability of uppermost terrace versus gully depth ratio using Model 2 ...................105
Frequency distribution showing vulnerability ranges of all measured terraces...............106
Relationship of terrace vulnerability (all terraces) to gully depth ratio ...........................107
Mean vulnerability of uppermost terraces grouped by geomorphic reach and
geomorphic setting...........................................................................................................108
Mean annual precipitations for monsoon season, in millimeters.....................................110

List of Plates in Pocket, Inside Back Cover
1.
2.
3.
4.
5.
6.

Key
Nankoweap Alluvial Fan Geomorphic Setting
Pallisades Tributary Plain Geomorphic Setting
Upper Unkar or Furnace Flats Talus Slope Geomorphic Setting
122 Mile Canyon Debris Lobe (Fan Lobe) Geomorphic Setting
Lower Tanner Dune Field Geomorphic Setting

Table of Contents

vii

List of Tables
2.1.
2.2.
2.3.
2.4.
2.5.
2.6.
2.7.
2.8.
3.1.
3.2.
3.3.
3.4.
3.5.
3.6.
4.1.
4.2.

Cultural Areas in Grand Canyon Containing Study Catchments.......................................12
Photographic Assessment of Arroyo Development in Grand Canyon ..............................14
Comparison of Physical Characteristics of Cataract Canyon and the Furnace Flats
Reach of Grand Canyon.....................................................................................................15
Summary of Data Used for Investigating Precipitation Changes Before and After
Emplacement of Glen Canyon Dam ..................................................................................20
Scores Assigned to the Presence of Specific Geologic Units or Geomorphic Processes ..22
Point Scale for Parameters within the Catchment..............................................................29
Infiltration Rates of Pre-dam, 1983, and Eolian Sand Deposits ........................................33
Summary of Precipitation Data from River-Level Stations for Thirty-two Years
Following Emplacement of Glen Canyon Dam.................................................................37
Historical Peak Flow Events in Cataract Canyon ..............................................................41
Study Sites in Cataract Canyon with Data Indicating River Terrace Sequence ................42
Gully Density of Three Paired Areas, Grand Canyon and Cataract Canyon.....................49
Comparison of Degree of Gully Incision, Grand Canyon and Cataract Canyon...............50
Volume of Sand Lost or Gained at Three Topographically Surveyed Terrace
Catchment Study Areas in Cataract Canyon......................................................................60
Climate Parameter for the Predictive Model ...................................................................112
Catchments in Grand Canyon that Do Not Support the Restorative Base-Level
Hypothesis........................................................................................................................114
Ranking of Highest-Risk Sites Based on Vulnerability Rating of Top Terrace ..............117

ACKNOWLEDGMENTS
A number of people contributed in a variety of ways toward the successful completion of this
report, many of them in more than one capacity. Steve Carothers and Rich Valdez helped get us
started with the initial research design. Rich Durnan advised us on database development.
Duane Hubbard and Diane Boyer directed us toward important historical photo archives. Helen
Wadsworth, GCMRC Librarian, provided us with a plethora of air photos from GCMRC files.
Bob Webb, Jack Schmidt, and Ruth Lambert advised us on hypothesis testing. Lisa Leap,
Richard Hereford, and Stephanie Yard acted as our technical advisors throughout the length of
the project.
We thank those who made the effort to review and comment on the draft final report: Rich
Hereford, Ruth Lambert, Lisa Leap, Stephanie Yard, and nine anonymous reviewers. Their
prodigious comments made the final report a better product, while we take full responsibility for
the ideas and results contained therein.
Field-support staff were essential for their advice and assistance on the ground. Lisa Leap, of
the Grand Canyon River Corridor Monitoring Project, directed us to our areas of study in Grand
Canyon. Bob Webb, John Weisheit, and Steve "Teaberry" Young provided help in locating field
areas in Cataract Canyon. Steve Wiele consulted with us on river dynamics and sediment
transport at several of our type geomorphic settings. Mark Gonzales and Steve Lamphear trained
us to conduct field survey work in Grand Canyon and Cataract Canyon. Steve Lamphear was
particularly helpful in producing the topographic maps of the type geomorphic settings. Chris
Brod, Rebecca Seago, and John Douglass were instrumental in producing the survey results.
And, of course, we could not have carried out this project without our sterling field assistants,
who worked hard to successfully complete fieldwork on our river trips: Laurie Lewis, John
Sterling, Kathy Durham, Bill Hatcher, Rich Durnan, H. J. Schmidt, Ally Martinez, Lynn Neal,
Lissa Bartlett, Robert O'Brien, Rebecca Seago, Matt Johnson, Helen Fairley, and Billy
Ellwanger. Finally, our greatest appreciation goes to Gary O'Brien, who helped tremendously in
getting us down the river and counseled us through all our crises of data collection.

ABSTRACT
Most of the cultural history of the Colorado River corridor in Grand Canyon is preserved in
high river terraces below Glen Canyon Dam that are no longer flooded by the dam-controlled river.
Apparent increased rates of gully erosion and loss of cultural sites in these terraces have been
attributed to base-level effects caused by dam operations and/or climatic variation. Two avenues
of investigation are pursued here. First, we address in the form of hypotheses three questions
concerning gully erosion of archaeological terraces: (1) Has the amount of erosion increased in the
post-dam period? (2) Can an apparent increase in erosion be attributed to climatic variation? (3)
Does the presence and/or operation of the dam contribute to an apparent increase in erosion?
Second, we develop a conceptual geomorphic model that describes the variety of processes that
operate in the small-catchment system and provide a mathematical model that predicts relative
vulnerability of terraces to gully erosion at various small catchments.
We followed several lines of inquiry to address these issues. We selected a sample set of 119
small catchments in Grand Canyon for field investigations and data collection. We classified these
catchments into five distinctive geomorphic settings and examined them for process-response
relations that affect gully development. Air photo sequences from 1965 to the present provided
evidence to determine the extent of gully development in Grand Canyon since the dam was
completed in 1963. Selection of a control reach in Cataract Canyon with an ecosystem that is
analogous to that of the pre-dam river in Grand Canyon allowed us to investigate small-catchment
and river processes as they may have interacted in the pre-dam setting.
A review of the work of previous researchers shows that a climatic pulse of particularly high
warm-season and winter rainfall has occurred since 1978 but that such perturbations are not unique
in the climatic record (Hereford and Webb 1992; Webb, Griffiths, and Melis 2000). Similar pulses
occurred during the first three decades of the century and for a short period during the early 1940s.
Results of our analysis and examination of climate for the three decades before and three decades
after dam construction show that, although local areas vary from one another, warm-season
precipitation patterns across the southern Colorado Plateau have changed little. Air photo analysis
in Grand Canyon and investigations in Cataract Canyon show that gully erosion has increased in
Grand Canyon during the past two decades in comparison with probable conditions in the several
previous decades. It can be assumed that periods of high precipitation earlier in the century also
formed extensive gullies, yet analysis of 1965 air photos shows little evidence for this, indicating
that alluvial deposition in the pre-dam period was probably responsible for restoring sand to
gullies. Therefore, the recent period of high precipitation may not be the sole cause of accelerated
gully degradation observed in the past two decades.
The "base-level hypothesis" of Hereford and others (1993) states that the interruption of high
sediment flux and large annual floods by the presence of Glen Canyon Dam in Grand Canyon has
changed dynamic equilibrium conditions at the mouths of gullies and arroyos, resulting in
accelerated rates of gully erosion. Our work in the control section of Cataract Canyon and in
Grand Canyon shows that this hypothesis needs refinement. The relative absence and small size of
gullies in Cataract Canyon suggests that conditions are intrinsically different from those now
present in Grand Canyon, despite geomorphic and climatic similarity between the two reaches.

xii

Abstract

It is apparent that before emplacement of the dam gully degradation in terraces was restored by
periodic alluvial deposition from river floods, but perhaps even more important is the redistribution
of flood sands onto higher terraces by wind. Thus, we propose the term "restorative base-level
hypothesis" to emphasize the dynamic equilibrium between gully erosion and renewed deposition,
a process that remains active in Cataract Canyon but is disrupted in Grand Canyon by the presence
and operation of the dam.
We developed type geomorphic settings to develop a conceptual process model for the diverse
small-catchment geomorphic system in Grand Canyon. Research findings explain how streams are
able to cross broad, flat terraces given a rainfall event and how they become progressively more
integrated with the river. The primary channelization processes are ponding and overflow, alluvial
fan progradation, and infiltration and piping, all of which contribute to nickpoint migration. An
understanding of these processes was essential to building the geomorphic model.
The predictive mathematical model quantifies erosional vulnerability by applying a
hypothetical rainfall event of 25 mm/hour onto a catchment above a "pristine" terrace sequence.
The principal driving factor for erosion is basin area. The principal resisting factor for erosion is
terrace diffusion capacity, which is a function of terrace sand cross-sectional area and infiltration
capacity. Several important modifying factors are applied to the basic model to determine relative
vulnerability of each terrace to gully erosion. Vulnerability of the top terrace at each catchment is
plotted against the measured amount of gully erosion in that terrace, providing a base line against
which progressive changes in gully depth can be easily monitored in the future.
Field studies and research show that: (1) gully erosion of terraces has been severe during the
past 20 years in Grand Canyon due to unusually high precipitation; and (2) sediment deprivation
coupled with the lack of large annual floods has caused a reduction in restorative (depositional)
factors. Continued measurement and documentation of geomorphic processes in catchments,
particularly at type geomorphic settings, will further refine and verify the predictability of the
model. We conclude that beach-habitat-building flows are essential for initiating natural
restorative processes and that one of the most important processes in gully mitigation may be
eolian reworking of newly deposited flood sands onto higher terraces. Prior to the construction of
Glen Canyon Dam, gully-deepening and river/wind depositional processes were in dynamic
equilibrium, allowing the preservation of ancient cultural sites for the past several thousand years.

CHAPTER 1
INTRODUCTION
Kate S. Thompson and Andre R. Potochnik
The study reported here was commissioned by the Cultural Resources Program of the Grand
Canyon Monitoring and Research Center (GCMRC), the technical arm of the Adaptive
Management Program that assesses the effects of the operations of Glen Canyon Dam on
downstream resources, to address the processes of gully erosion in archaeologically rich river
terraces along the Colorado River in Grand Canyon. Previous workers have proposed that erosion
of these sand terraces has been accelerated since emplacement of Glen Canyon Dam and that the
dam and its operations are partly responsible for this change (Fairley et al. 1994; Hereford et al.
1993; Lucchitta and Leopold 1999). Others have proposed that decadal-scale climate change may
be responsible for the apparent rapid erosion of these terraces (Hereford and Webb 1992; J.
Schmidt, personal communication 1998; R. H. Webb, personal communication 1998). Lastly,
there remains the possibility that the apparent acceleration of erosion of cultural features is simply
continuing at pre-dam rates and processes and that the dam has no influence on the geomorphic
system.
Whatever the cause, valuable archaeological resources are being lost to erosion, a matter of
regulatory, legislative, and ethical concern to government agencies, tribal groups, archaeological
researchers, preservationists, and the general public. In particular, the resource management
entities of Grand Canyon National Park (GRCA), the Bureau of Reclamation (BOR), the Arizona
State Historic Preservation Office (SHPO), the Advisory Council on Historic Preservation, the
GCMRC, and six Native American tribes (the Hopi Tribe, the Hualapai Nation, the Kaibab Band of
Paiute Indians, the Paiute Indian Tribe of Utah for the Shivwits Band of Paiutes, the Navajo
Nation, and the Pueblo of Zuni) are mandated to preserve and mitigate impacts to these cultural
resources. Meeting this mandate calls for a thorough understanding of the geomorphic system, the
causes of erosion, and the development of a model that can predict the relative vulnerability of
cultural sites to gully erosion.
We investigated how the presence or operations of Glen Canyon Dam may be directly or
indirectly impacting archaeological and other cultural properties along the Colorado River corridor
within Grand Canyon. A cultural property is any prehistoric or historic material or feature that
represents human habitation or use along the river. The objectives of this study were threefold: (1)
to test hypotheses that may explain the apparent increased erosion rates at cultural sites on river
terraces; (2) to develop a predictive geomorphic model useful as a tool for long-term monitoring,
preservation, and mitigation of these resources; and (3) to identify the most threatened sites so that
GRCA managers can prioritize remedial action needs. We were able to compare model predictions
with on-site field evaluations to test the validity of our model. Furthermore, we discuss how
monitors can use this model as a practical predictive tool for anticipating the vulnerability of sites
to incipient degradation.

2

Chapter 1

The National Park Service (NPS) mandate under Section 110 of the National Historic
Preservation Act (NHPA) requires preservation of the integrity of these resources for the public,
with serious consideration given to Native American concerns for cultural resource management
and preservation. To facilitate compliance with the BOR’s responsibilities under Section 106 of
NHPA, a Programmatic Agreement (PA) was signed in 1994 by various federal and state agencies
and the six Native American groups listed above. The Grand Canyon Protection Act of 1992
(Section 1802) specifies that Glen Canyon Dam "shall be operated in such a manner as to protect,
mitigate adverse impacts to, and improve the values for which Grand Canyon National Park and
Glen Canyon National Recreation Area were established, including, but not limited to, natural and
cultural resources and visitor use."
Research conducted under Glen Canyon Environmental Studies Phase II during the early 1990s
documented degradation of numerous sites within the old high-water zone (OHWZ) of the pre-dam
period (Carothers and Brown 1991; Fairley et al. 1994). This finding prompted Hereford and
colleagues (1993) to formulate the "base-level hypothesis": that site degradation is fundamentally
caused by rainfall-induced erosion of the pre-dam terraces and is accelerated by degradation of
sand bars fronting these terraces. These sand bars, located at the "toes" of the culturally rich predam terraces, served as buttresses in the pre-dam period, slowing downcutting by small tributary
streams that drain onto the higher terraces.
Before the dam, annual sediment-laden floods periodically renewed these lower terraces and
presumably infilled incipient gullies. The ten-year flood of about 140,000 cubic feet per second
(cfs) (O’Connor et al. 1994) maintained a higher local base level of sand bars and thus minimized
the rate of downcutting by gullies and arroyos (Hereford et al. 1993; Hereford 1996). Many
streams that once drained onto a higher terrace level have recently breached the lower terraces and
now drain to the Colorado River as river-based streams (Leap et al. 2000). Erosion and nonrenewal of sediment in the river corridor have disrupted the balance between gully erosion and
annual deposition/infilling of river sediments in those catchments. The drop in local "effective"
base level for these streams results in deepening and widening of existing gullies (Higgins, Hill,
and Lehre 1990; Schumm 1977), with consequent loss of cultural resources contained within the
higher terraces (Hereford et al. 1993).
The draft Historic Preservation Plan (HPP) (Bureau of Reclamation [BOR] et al. 1997) targets
the need for a refined understanding of erosional processes and in particular calls for "a predictive
model of geomorphic processes related to archaeological site erosion." The current study focused
on testing three hypotheses of accelerated gully erosion in the post-dam period, creation of a
predictive geomorphic model to identify site vulnerability to gully erosion related to selected
archaeological sites, and prioritizing existing sites in a risk-hierarchy system.

Introduction 3
STUDY OBJECTIVES
Significance of Work
Essential to protecting cultural resources along the Colorado River is the understanding of
geomorphic processes that are directly or indirectly related to the operations of the dam. In our
research we sought a more accurate understanding of pre-dam depositional and erosional events as
a basis for determining whether specific erosional processes are related to dam operations. As
stated in the draft HPP, this effort will help close gaps in the Cultural Resource Monitoring
Program (CRMP) database by integrating archaeological findings with the geomorphic model,
thereby assisting in preservation of cultural sites along the river corridor.
Specific Objectives and Hypotheses
The specific objectives of the study were:
1. To evaluate data from previous geomorphic work on river- and terrace-based arroyos,
archaeological monitoring, and sedimentation and climate studies, incorporating relevant
research into three primary sets of hypotheses.
2. To test the null hypothesis (H0): that terrace erosion rates and gully erosion rates on the
terraces have remained unchanged from pre-dam time to post-dam time.
(a) Report on river-level air photo analysis to determine timing for the inception of the
largest gullies in pre-dam terraces
(b) Compare amount of gully erosion on terraces of a control section of river (Cataract
Canyon) that emulates the alluvial history and the pre-dam geomorphic setting in Grand
Canyon
3. To determine if H0 is falsifiable based on research findings.
4. To test the alternate hypothesis (H1): that gully erosion on alluvial terraces associated with
archaeology has increased from pre-dam time to post-dam time due to variation in
precipitation patterns.
(a) Evaluate previous research on decadal-scale precipitation variations that may be
responsible for increased erosion rates.
(b) Investigate variations in monsoon-season rainfall at equivalent time periods before and
after closure of Glen Canyon Dam
5. To test the alternate parallel hypothesis of Hereford et al. 1993 (H2): that gully erosion on
alluvial terraces associated with archaeology has increased from pre-dam time to post-dam
time due to base-level effects.

4

Chapter 1
(a) Report on rebuilding of sand bars that buttress upper terraces and infill gullies in both
Grand Canyon and the control section of Cataract Canyon
(b) Report on process of gully erosion and integration with the river at a large number of
archaeological sites in Grand Canyon
6. To develop a geomorphic model that explains the apparent increased rates of gully erosion
and more clearly defines our understanding of how gully erosion is likely to impact
archaeological resources in the future.
7. To develop a predictive geomorphic model that ranks each site as to its vulnerability to
erosion.
8. To identify for remedial action the sites where resources are most threatened.
9. To evaluate and discuss how the project results can be used in the monitoring program to
aid in protecting the most-threatened resources.
Value of Anticipated Results

To test the three hypotheses, we examined the concepts outlined by Hereford et al. (1993), who
suggest that erosion of culturally rich pre-dam river terraces is a direct response (1) to rainfall and
(2) to base-level change, a consequence of the presence and operation of the dam. We apply this
hypothesis at each study site and test, as best as resources allow, the null hypothesis: that gully
(arroyo) erosion has not increased in post-dam time. Other investigators have suggested that baselevel factors are generally over-emphasized and that other drainage integration mechanisms or
climatic patterns are often more important (Lucchitta 1991; J. Schmidt and R. H. Webb, personal
communication 1998). Nonetheless, Hereford and colleagues have produced a series of papers
(Hereford et al. 1993; Hereford et al. 1996) and maps (Hereford 1996; Hereford, Burke, and
Thompson 1998, 2000a, 2000b) describing the stratigraphy and geomorphic context of the pre-dam
alluvial terraces and have set the stage for further important investigations. Our study expands on
the types of geomorphic processes that operate in this small-catchment, ephemeral-stream
geomorphic system and discusses their relative importance. This report also provides cultural
resource managers and constituencies with a practical predictive tool for anticipating the
vulnerability of sites to incipient degradation.
BACKGROUND
Previous Archaeological and Geomorphic Studies
Early surveys of archaeological sites along the river were conducted by Taylor (1958), Euler
and Taylor (1966), Jett (1968), and Schwartz, Chapman, and Kepp (1980). Archaeological survey

Introduction 5
efforts intensified in the early 1980s, when extraordinarily high river flows, combined with
increased human visitation, began to profoundly impact sites along the river. In 1984 Jones (1985)
conducted an excavation of five sites and recognized that little was known of the stratigraphic
context and geomorphology of these sites. Because of concerns over degradation of cultural
resources, the first comprehensive documentation of archaeological sites was performed
throughout the length of Grand Canyon (255 river miles) at or below the 300,000 cfs stage height
(Fairley et al. 1994). Between Lees Ferry and Separation Canyon, 475 prehistoric and historic sites
were catalogued, 265 of which are on or in Colorado River alluvium. Most of these alluvial sites
are in the broader geomorphic reaches described by Howard and Dolan (1981) and Schmidt and
Graf (1990), such as "Furnace Flats" and "Western Grand Canyon" (Figure 1.1).

Figure 1.1. Areas showing concentrations of archaeological sites within geomorphic reaches of
Grand Canyon. Cataract Canyon is the control area of study.

The geomorphology associated with archaeological sites along the Colorado River corridor
was investigated by Hereford (1996), Hereford et al. (1993), Hereford, Burke, and Thompson
(1998), Lucchitta (1991), Lucchitta et al. (1995a), Lucchitta et al. (1995b), and Lucchitta and
Leopold (1999). These researchers identified periods of deposition and erosion from Quaternary
times to the present by mapping the great variety of these depositional sequences along the river
corridor. Detailed mapping of specific sites has provided an understanding of geomorphic
processes affecting these deposits, which has, in turn, allowed the formulation of a hypothesis that
attempts to explain the accelerated erosion of associated archaeological sites. Stratigraphic profiles
developed by Hereford and others (1993, 1996) (synthesized in Figure 1.2) illustrate a sequence of
river flood sands interbedded with slopewash or tributary deposits. Interestingly, these deposits
contain cut-and-fill features that suggest a dynamic process of channel erosion and subsequent
infilling by river flood sand in the pre-dam era. Post-dam alluvium typically contains only channel
cuts, suggesting net erosion since emplacement of the dam. While erosion to the archaeological
terrace was evident in pre-dam time, it was generally healed periodically with an influx of river or
eolian sediment. Hereford and others (1993) therefore deduced that erosion of pre-dam alluvial
terraces is an indirect result of dam operations, but they have been criticized for not establishing a
control site for comparison outside the Grand Canyon where the river is largely without dams.

Figure 1.2. Schematic cross section of river terrace sequence with various geomorphic features
(after Hereford et al. 1993, 1996).
Hereford and others (1993, 1996) described and synthesized the late Holocene alluvial
terrace stratigraphy in four discrete study areas of the Furnace Flats reach of Grand Canyon. In
these areas, pre-dam alluvial terraces comprise an extensive geomorphic system that interfaces
with two adjacent geomorphic systems: the tributary alluvial fan system (Hereford et al. 1996;
Webb, Pringle, and Rink 1989) and the river channel system (Rubin, Schmidt, and Moore 1990;
Schmidt 1990; Schmidt and Graf 1990) (Figure 1.2). Hereford, Thompson, and Burke (1997,
1998) developed methods for dating Late Holocene debris fans by integrating archaeological and
radiocarbon techniques from adjacent fluvial terraces. In addition, they published detailed geologic
maps of five major tributary fan complexes: Lees Ferry (Hereford, Burke, and Thompson 2000a),

Nankoweap Creek (Hereford, Burke, and Thompson 1998), Palisades Creek (Hereford 1996), and
Lower Tanner Canyon and Granite Park (Hereford, Burke, and Thompson 2000b) that illustrate the
late Holocene stratigraphy.
Researchers (Hereford et al. 1993; Lucchitta et al. 1995b) have also described various
geomorphic processes affecting the pre-dam alluvial terraces, particularly the effects of local
ephemeral stream networks that cross the terraces. Hereford and others (1993) distinguished
terrace-based streams from river-based streams, depending on whether the stream terminates on a
pre-dam alluvial terrace or is integrated with the river or a major tributary to the river. They noted
that river-based streams had cut deep arroyos in the sandy pre-dam terraces, exposing and eroding
numerous archaeological sites. These researchers also suggested the possible effects of local
climatic variations (Hereford and Webb 1992) that have been eroding pre-dam alluvial terraces
containing most of the prehistoric and historic cultural sites. However, all concur that Glen
Canyon Dam has exacerbated a possible natural cycle of erosion by further decreasing the sediment
supply to the river.
Rationale for This Study
Hereford and others (1993, 1996) pointed out the need to understand cause-and-effect
relationships between the pre-dam alluvial terrace system and post-dam water flows and associated
deposits. Before regulation of the Colorado River, most small ephemeral streams probably drained
to a higher local base level, represented by the pre-dam terraces. If any streams drained to the
river, they were infilled with eolian sand or river sand from a large flood, as evidenced by cut-andfill structures identified in pre-dam terraces such as that recorded at Locality 1 of the Palisades
Creek area (Hereford 1996). Under pre-dam conditions, sand and silt deposited in the mouths of
small tributary channels during annual floods would infill previously eroded headcuts, thus
preventing them from migrating up-channel. Because regulated flows have lowered the level at
which river sand bars are deposited, small streams are regrading to a lower "effective" base level.
The concept of base-level change invoking gullying or arroyo-cutting is a well-known concept that
has been employed with equilibrium of streams (Gilbert 1879; Leopold and Bull 1979). With local
intense rainfall as the driving force, channels are presently extending headward and widening as
they adjust to the new post-dam base level of sand-bar building. Once streams reach this
configuration, erosion of upper terraces is greatly accelerated, rendering remedial efforts difficult,
if not fruitless.
If the base-level hypothesis is correct, the size and characteristics of channel-margin sand
bars along the river are of critical importance in determining extent of erosion of cultural sites. We
would expect that areas with persistent sand bars would maintain the base level of streams. Gullies
can be infilled or buried, and wind can rework sediment, often mounding it onto the next higher
terrace (Hereford 1996; Hereford, Burke, and Thompson 1998). A model for sediment transport
and deposition by the post-dam river was produced by Rubin, Schmidt, and Moore (1990), Schmidt
(1990), and Schmidt and Graf (1990). The continuing monitoring work of Kaplinski et al. (1993)
evaluates changing conditions of sand bars along the river under different flow regimes. Modeling
of the 1996 flood deposited sands by Wiele (1997) and Yeatts (1996, 1998) helps to determine the
utility of a controlled flood as a management tool. These studies are extremely important in
predicting the longevity of the pre-dam terraces in the post-dam regime.
The changing conditions of large tributary alluvial fans, which lie adjacent to the sandy

alluvial pre-dam terraces, will determine whether some of the small ephemeral streams become
integrated with the river via the fans. As cited in several reports (BOR et al. 1997; Leap et al.
1998; Neal 2000), a predictive model is needed for properly assessing erosional risk to
archaeological sites. Therefore, any geomorphic model that attempts to predict erosion of the predam alluvial terrace system will need to investigate the important effects of base-level control from
both channel margin bars and alluvial fans.
To date, no formal geomorphic model ranking vulnerability of archaeological sites to
erosion has been fully developed. Recorded observations by River Corridor Monitoring Program
(RCMP) archaeologists working in the Grand Canyon (Leap et al. 2000) and in Glen Canyon
provide a record of erosion since 1992. While these data provide good qualitative documentation
of change, there are deficiencies in quantifying the amount and extent of erosional change within a
site, from site to site, and from year to year (Neal 2000). These deficiencies have been directly
related to the lack of a geomorphic model to guide monitors in documenting and interpreting
change over time at impacted archaeological sites.
The RCMP monitoring efforts have recorded by fiscal year (October 1–September 30) both
natural and visitor-related impacts to 285 sites within GRCA from 1992 to 1999. Since 1994, 55%
of all monitored sites in the Grand Canyon have been impacted by some kind of channelization
process. During FY 1998 alone, 37% of all recorded physical impacts to these sites were caused
by active gully erosion; incipient thalweg development (keyed as "surface erosion" in RCMP
reports) was noted in 25% of all observations. Changes in streams have become significant, as
river-based streams are draining across archaeological features at 25% of monitored sites.
Unfortunately, we could not determine from the Grand Canyon RCMP data when most of these
streams breached the lowest alluvial deposits. Furthermore, many of these streams have become
river-based since the completion of the dam (Leap et al. 2000 ), probably developing after 1976
(Hereford et al. 1993), infilling during the controlled flood in 1996 (Yeatts 1996), and then
becoming river-based again.
An effort to prioritize sites for remedial action in FY 1999 (Leap et al. 1998) was recently
undertaken to attempt to identify highest-risk sites. Neal (2000) could not determine precisely how
the ranking process was developed or carried out, as either no consistent descriptive history of a
given site is provided, or there are discrepancies in data or recommendations from year to year.
We were able to determine that prioritization was based on present-day conditions, whether erosion
appears "active," and whether archaeological materials and features are being exposed. The RCMP
coding system appears to be the key to mitigating adverse effects to actively eroding sites.
However, if we can help to refine RCMP's methods so that recorded data are more objective and
quantifiable, justification for either preserving or excavating these sites would be strengthened.
Since the Grand Canyon/Glen Canyon river corridor archaeological monitoring program could
very well set the precedent for other NPS archaeological monitoring programs nationwide, it is
critical to refine monitoring methods, decrease subjectivity, increase quantifiable parameters, and
maintain consistency in annual data collection and management.

CHAPTER 2
METHODS
Kate S. Thompson, Andre R. Potochnik, Gary O’Brien, and Ron Ryel
This study pursued two separate lines of investigation that address erosion of cultural sites
in sand terraces along the Colorado River in Grand Canyon National Park (GRCA). In the first
part of the study we investigated three sets of hypotheses that have been developed by previous
researchers; in the second part we sought to developed a predictive geomorphic model to identify
archaeological areas where risk of gully erosion is high. Both parts of the study required a
selection of study sites within those geomorphic reaches that contain high concentrations of
archaeological sites. The study sites were identified in coordination with the River Corridor
Management Program (RCMP) to ensure that catchment selections are closely related to nearby
archaeology.
SELECTION OF STUDY SITES
Four reaches of Grand Canyon were found to contain most of the cultural sites that are on
high sand terraces: Marble Canyon, Furnace Flats, the Aisles, and Western Grand Canyon (Figure
1.1). Within each of these reaches, catchments were identified that drain through culturally rich
areas (Table 2.1). The main criterion for selection was that channels extend at least as far as the
lower mesquite terrace (lmt) that was overtopped by the 300,000 cubic feet per second (cfs) flood
of 1884. For the most part we also selected terrace-based streams so that they could be rated in the
predictive geomorphic model and incorporated into future monitoring of cultural resources.
Several catchments were chosen to represent end points in the rating system so that the predictive
model could be calibrated. The highest-risk sites, such as Axehandle Cove, have the largest active
arroyos, and the lowest risk-sites, such as those in the dune field at Nankoweap, have stable, swalelike channels.
Field work at these sites was conducted following procedures that were highly sensitive to
cultural resources and to fragile desert soils. The concept of using the "minimum tool" to minimize
the amount of human intrusion is important for preventing the creation of research-related trails
and unnecessary human-induced erosion to archaeological sites.
TESTING THE HYPOTHESES
Three hypotheses describing the relationship between the emplacement of Glen Canyon
Dam and erosion at archaeological sites along the river corridor were tested. The null hypothesis
was examined by Hereford and others (1993), Lucchitta (1991), and Fairley and others (1993).
The two alternate hypotheses tested were proposed by Hereford and Webb (1992) (referred to in
this report as the climate-variation hypothesis) and by Hereford and colleagues (1993) (referred to
in this report as the base-level hypothesis).

Table 2.1. Cultural Areas in Grand Canyon Containing Study Catchments
Reach
Marble Canyon

Furnace Flats

Aisles

Western Grand

Total

Area

River Mile/Side

Paria Cove
Axehandle Cove
Ten Mile Rock
Soap Creek
Willy’s Grave
Little Nankoweap
Main Nankoweap
Kwagunt Canyon
60 Mile Canyon
Lava Canyon
Palisades Creek
Espejo Creek
Comanche Creek
Tanner Canyon
Basalt Canyon
Lower Tanner
Upper Unkar
Old Unkar Camp
Lower Unkar
122 Mile Canyon
Owl Eyes
Fishtail Canyon
140 Mile
Saddle Horse
Old Heli Pad
186 Mile
190 Mile
194 Mile
196 Mile
201 Mile
202 Mile
Indian Canyon
Arroyo Grande
Granite Park
Below Granite Park
Fall Canyon
36

10R
20L
10 0 R
11 2 R
44 8 L
51 7-52 0 L
52 2-53 3 R
56 1 R
59 7 R
65 3 L
65 4-65 6 L
66 9-67 0 L
67 1-67 8 L
68 9-69 0 L
69 3 R
69 6-70 3 L
71 5-71 6 R
72 2 R
73 2 L
121 9-122 0 R
134 5-134 6 L
139 0 R
139 7 L
176 3 R
182 7 R
186 2 R
189 7 L
194 4 L
196 2 R
201 2 R
201 9 R
206 5 R
207 7-207 8 L
208 5-209 1 L
209 5 R
211 5 R

No. of
Catchments
2
3
1
3
1
5
12
1
2
2
10
3
4
2
5
6
8
1
3
4
4
1
1
2
1
1
1
1
1
5
2
2
6
9
3
1
119

Methods

13

The Null Hypothesis
H0: Terrace erosion rates and gully erosion rates on the terraces have remained unchanged from
pre-dam time to post-dam time.
Test 1: Determine when large gullies formed in Grand Canyon using air-photo interpretation
Hereford and colleagues (1993) initially assessed the timing of large gully (arroyo) development
on river terraces within a seven-mile reach of eastern Grand Canyon (RM 65-72). They concluded
that arroyo development was not present in that area in 1890, and that most of the larger arroyos
were present in 1965 air photos. Incipient gullies in the Upper Unkar area were noted in 1973 and
were well developed by 1984. Hereford and colleagues concluded that gully erosion began to
intensify sometime after 1973, and RCMP monitors indicate that it is continuing (Leap et al. 2000).
We chose a subset of 23 catchments in 10 of our study areas to assess when gullies formed or
when arroyos were rejuvenated in Grand Canyon. Using low-altitude air photos taken from 1965 to
1996, we developed a time sequence that depicts when a certain stage of gully incision occured at
the sites listed in Table 2.2. The basis for catchment selection was whether they presently incise the
full suite of pre-dam sand deposits (Figure 1.2). Present-day arroyos are typically box-like in crosschannel shape and are up to 1-2 m deep and 5 m wide. The degree of arroyo development in the predam period is represented only by the 1965 air photos (1:6,000), as this is the earliest low-altitude
photography available for the river corridor in Grand Canyon. A systematic assessment of all photos
was limited, because photos were missing, areas were shadowed, or the overflight had been too high
to see the area clearly. Furthermore, once an arroyo could be seen to have reached its present level
of incision, subsequent photos of that feature were not examined.
Another approach to examining this hypothesis would involve an extensive search for archived
oblique photos taken by river runners and researchers. Hereford and others (1993) found and
republished only two such photos, taken by R. C. Euler in 1965, which illustrate intact terrace
escarpments at Upper Unkar. They also published repeat photos taken in 1983 and 1991 that show
progressively more dissected and eroded slopes. Hubbard (2000) also shows a 1965 Euler photo and
a repeat photo taken in 1996 of the same slope but from a different angle. These photos show that a
pristine slope containing a pueblo wall in 1965 had collapsed into a gully by 1996.
A complete historical photo analysis exceeds the scope of this study as originally proposed and is
worthy of its own study and funding. Duane Hubbard (personal communication 1999) is presently
using historical repeat photography in a study centered around archaeologic areas. Results of this
project will no doubt contribute to our understanding of the timing of formation of arroyos and
gullies. We investigated several photo collections at various archives, including the Museum of
Northern Arizona, Northern Arizona University, and GRCA. Our investigations of over 1500 photos
in these collections taken by La Rue, Kolb, Lampland, McCullough, Stone, and Belnap proved to be
essentially fruitless for this particular test, as these photographers focused on subjects other than
pristine hillslopes that contain archaeology.

Table 2.2. Photographic Assessment of Arroyo Development in Grand Canyon
Area

Soap Creek

Palisades
Creek
Espejo Creek
Comanche
Creek
Tanner
Canyon
Lower Tanner

Upper Unkar
Lower Unkar

Arroyo Grande
RM 209.5

Catchment

A
B
C
H

5/14
1965
NG
NG
NG
PG*

6/16
1965
NA
NA
NA
NA

6/16
1973
PG
PG
PG
PG*

A
B
F
H
A
B
A
B
C
D
E
A
B
C
A
B
A
B
C

NG
NG
NG
NG
PG*
PG*
NA
NA
NA
NG
NG
PG
PG
PG
PG
NG
NA
NA
NA

NA
NA
NA
NA
NA
NA
NG
NG
NG
NA
NA
NA
NA
NA
PG
NG
PG
PG
NG

NG
NG
NG
NG
PG*
PG*
NG
NG
NG
NA
NA
PG
PG
PG
PG
PG
PG
PG
PG

Date Photo Taken
7/11
10/21
1980
1980
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
FG
PG
FG
FG
NG
NG
NG
NG
NG
PG
PG
PG
FG
PG
FG
FG
PG

FG
PG
NA
PG
NA
NA
NG
NG
NG
NG
NG
PG
PG
PG
NA
NA
NA
NA
FG

10/21
1984
FG
FG
FG
FG

10/11
1992
NA
NA
NA
NA

4/6
1996
NA
NA
NA
NA

FG
FG
FG
FG
FG
NA
FG
PG
FG
PG
PG
PG
FG
FG
FG
FG
NA
NA
NA

NA
NA
FG
FG
NA
NA
NA
NA
NA
FG
FG
FG
NA
NA
NA
NA
NA
NA
NA

NA
NA
NA
NA
NA
NA
FG
PG
FG
FG
FG
NA
NA
NA
FG
FG
FG
FG
FG

NG = no gully present in terraces or stream is terrace-based on ap/sa terrace
PG = partial gully development or stream is terrace-based on mt or pda terraces
FG = full gully development or stream is terrace-based on 1983 sand or below
NA = no assessment – photos are not available, gullies not discernible on photo, or assessment not
needed.

Test 2: Compare amount of gully erosion on terraces of a control section of river (Cataract Canyon)
to that in Grand Canyon
In this test, we ask if gully density and activity in the Cataract Canyon control area and a
similar geomorphic reach in Grand Canyon are different. The first step in making this comparison
was to demonstrate that Cataract Canyon, a 15-mile section of the Colorado River that drains into
Lake Powell (Figure 1.1), is a suitable analogue to the pre-dam setting of Grand Canyon. Cataract
Canyon was chosen as potentially suitable as a control because in this area the geomorphic setting
and climate are somewhat similar to those in Grand Canyon, but dams have relatively little
influence on the natural hydrograph and sediment flux. We determined that the climate and
physical setting of a study section in Cataract Canyon are comparable to conditions in the Furnace
Flats reach in Grand Canyon (Figure 1.1), where 39% of our study catchments are located. Not
only do the two areas lie in similar climatic zones, but the Colorado River carries through Cataract
Canyon a similar pre-dam sediment load with an annual hydrograph approximating that of Grand
Canyon (Van Steeter and Pitlick 1998). One difference is that many river recirculation zones in
Cataract Canyon are formed below constrictions caused by landslides and rockfalls rather than by

tributary side canyon debris fans. Nonetheless, river dynamics are similar to those found in Grand
Canyon. Table 2.3 shows the similarities in the comparative data for the two areas.
Table 2.3. Comparison of Physical Characteristics of Cataract Canyon and the Furnace Flats
Reach of Grand Canyon
Characteristic

Cataract Canyon

Grand Canyon
(Furnace Flats Reach)

2.42

1.95

11

13

Elevation (m)

1158

807

Mean annual precipitation (mm)

220

233

Mean monsoon season temperature (C)

22.2

23.8

Mean monsoon season precipitation (mm)

86

97

Number of precipitation events exceeding
25 mm/day (1964–1995)

6

8

9.1

9.0

101,000

120,000

River gradient (m/km)
Reach length (km)

Percent of months in monsoon season
receiving >50 mm precipitation
1957 flood stage* (cfs)

Note: Climate data used are from Moab, Utah, for Cataract Canyon and from Phantom Ranch for Grand
Canyon (Western Regional Climate Center 1999).
*1957 figures represents a typical flood stage before closure of Glen Canyon Dam; GC gage at Lees Ferry
and Cataract Canyon is the sum of flows recorded at Cisco and Green River.

The geologic strata of Cataract Canyon, like those in Grand Canyon, consist of
Pennsylvanian and Permian resistant limestones and sandstones interbedded with softer shales,
creating a cliff-bench topography. In both areas, the structural attitudes of strata are horizontal to
gently inclined with minor faulting and folding (Huntoon, Billingsley, and Breed 1982). The
geomorphic setting along the river in both canyons is similar in that periodic constrictions of the
channel cause a pool-and-drop morphology. Constrictions are caused by side canyons producing
tributary debris fans, talus slopes and rock falls reaching the edge of the riparian zone, or land
slumps (Toreva blocks) reaching the river; the latter are more common in Cataract Canyon (R. H.
Webb, personal communication 1998).
Since there is no published work on the alluvial history of Cataract Canyon, we conducted
field investigations to develop relative ages, absolute ages, and correlation of river terraces for
Cataract Canyon and Grand Canyon. Data collected on river terraces in Cataract Canyon included
tree-ring samples, radiocarbon samples, sediment samples, height and width of terraces, diagnostic
driftwood lines, diagnostic artifacts in driftwood piles, and vegetation community. Previous
dendrochronological data from netleaf hackberry collected in Cataract Canyon were supplied to us
by Alex McCord (personal field notes, unpublished), along with a publication of work in Grand
Canyon by McCord and others (Salzer et al. 1996). This work, in conjunction with our tree-ring
results, helped us correlate pre-dam flood deposits in Cataract and Grand canyons. We also
correlated historical streamflow data for Cataract Canyon with the primary river flood events that
formed terraces in both Cataract Canyon and Grand Canyon..

We then identified three study sites in Cataract Canyon that are geomorphically similar to
three sites in Grand Canyon. The Cataract Canyon sites are Rapid 4, Cross Canyon, and Rapid 12.
The Grand Canyon sites are Palisades Creek, Tanner Canyon, and Upper Unkar. Gully density and
individual gully depths were measured to compare the degree of erosion at each site.
The Climate-Variation Hypothesis
H1: Gully erosion on alluvial terraces associated with archaeology has increased from pre-dam
time to post-dam time due to variation in precipitation patterns.
Test 1: Evaluate previous research on decadal-scale precipitation variations that may be responsible
for increased erosion rates
Regional maps (Edwards and Batson 1990) indicate that the elevation of the Colorado
Plateau averages more than 1200 m (3940 feet) above sea level, whereas the average elevation of
the Basin and Range province in the Lake Mead area is less than 1200 m (3940 feet) above sea
level. The regional elevation gradients from the high plateaus to the low deserts strongly influence
precipitation patterns across the region.
Precipitation in this region is characterized by two wet seasons, one produced by frontal
storm systems in the winter and the other by monsoon systems in the late summer (Hereford and
Webb 1992; Hansen and Schwarz 1981). Frontal systems originate in the Pacific Ocean and track
eastward across the region. These systems tend to be more pervasive toward the north, dissipating
in energy toward the south. The summer systems originate from the Gulf of Mexico or the Sea of
Cortez and track northward into the southwestern United States before circulating eastward into the
Great Plains (Griffiths, Webb, and Melis 1996). Some researchers have proposed anomalous wet
periods of both warm season and winter season precipitation from the late 1970s through the 1990s
have accelerated gully erosion of alluvial terraces (Hereford et al. 1993; Webb, Griffiths, and Melis
2000).
Several researchers have suggested that warm-season rainfall is the primary cause of fluvial
erosion, sediment transport, and land surface change in the region (Graf, Webb, and Hereford
1991; Hereford 1989; Melis et al. 1996; Webb 1985). Warm-season rainfall generally occurs
between June 15 and October 15 (Hereford and Webb 1992) and is typically the result of isolated
thunderstorms of the southwestern monsoon or, less frequently, of dissipating tropical cyclones or
cutoff low pressure systems that produce widespread rainfall. In this paper, we use the term
"monsoon season" precipitation to refer to all warm-season precipitation from July 1 to October 31,
regardless of the type of storm system.
A relatively few, very large debris flows from tributary canyons have been attributed to
winter-season rainfall from large frontal storm systems (Cooley et al. 1977; Melis et al. 1994;
Webb et al. 2000). These storms may be intense and are commonly of longer duration than warm
season storms. Very intense localized convective storm systems and high monthly precipitation
have been related to the triggering of debris flows and tributary flooding in Grand Canyon. High
seasonal precipitation, on the other hand, does not correlate well with large tributary floods in
Grand Canyon (Griffiths et al. 1996; Melis et al. 1996, Webb et al. 2000).
For this test we examined previous research on variation in twentieth-century precipitation

patterns across the Grand Canyon and the southern Colorado Plateau to determine if there is a
cause-and-effect relationship between precipitation and arroyo cutting.
Test 2: Investigate variations in monsoon season rainfall at equivalent time periods before and
after closure of Glen Canyon Dam
To assess variation in monsoon season rainfall, we needed to pool as much data as possible
from weather stations throughout the southern Colorado Plateau. We then isolated local variation
in seasonal precipitation by subdividing the path of the Colorado River through the southwestern
Colorado Plateau into four distinctive topographic regions: Lower Colorado River, Western Grand
Canyon, Kaibab-Mogollon Plateaus, and Lake Powell–Canyonlands (Figure 2.1). We designated
each region by its predominant physiographic feature and used these names only for the purpose of
regional climate assessment in this report. The Lower Colorado River region west-southwest of
Grand Canyon lies mostly below 900 m (2950 feet) elevation. Winter-front moisture from the
Pacific Ocean generally passes over this low desert region south and west of the Colorado Plateau.
Weather stations in this region record precipitation ranging from 120 to 180 mm (5-7 inches) per
year (Western Regional Climate Center [WRCC] 1999). The Western Grand Canyon region lies
between Lake Mead and Kanab Creek in Grand Canyon. It is a series of high plateaus that
progressively ascend in elevation from 600 to 1800 m (1970-5900 feet) over a broad region that
receives 200-400 mm (8-16 inches) of mean annual precipitation (WRCC 1999).

Figure 2.1. General climatic zones of the southwestern Colorado Plateau based on physiography
and elevation.
From Kanab Creek to the Little Colorado River, the plateaus rise even more sharply to form
the Kaibab-Mogollon (K-M) Plateaus. This region is part of a broadly arcuate topographic high
that extends from the Mogollon Rim of central Arizona northward through the San Francisco
volcanic field into the Coconino and Kaibab plateaus of northern Arizona. The crest of the K-M
plateaus varies in elevation from 1800 to 2500 m (5900-8200 feet), with the Colorado River cutting
through its midsection at about 750 m (2460 feet) elevation. The K-M plateaus area receives 300600 mm (12-24 inches) of precipitation per year, while the mean annual precipitation at Phantom
Ranch averages 230 mm (9 inches) (WRCC 1999).
The Lake Powell–Canyonlands region lies in the rain shadow of the Kaibab uplift and
extends far to the northeast from the Little Colorado River to Moab, Utah. Toward the northeast,
the K-M highland drops over 1000 m (3280 feet) in elevation to form the semiarid Marble Platform
and Lake Powell region. This region lies between 950 and 1500 m (3115-4920 feet) in elevation
and receives between 125 and 250 mm (5-10 inches) of precipitation annually (WRCC 1999).
We assessed climate records to determine whether there are any obvious variations in
monsoon season precipitation in the post-dam period. Weather stations were identified with
records that include the period from 1932 to 1995, so that 32-year periods pre- and post-dam could
be compared. Weather stations included in this analysis lie within a broad region that parallels the
river corridor from western Colorado to western Arizona (Figure 2.2). For this analysis we used
two parameters of monsoon season rainfall: (1) large daily precipitation events and (2) large
monthly volumes of precipitation.

Figure 2.2. Location of weather stations from which 64-year daily precipitation data were
compiled.
Using available precipitation data, we counted the number of precipitation events greater
than 25 mm/day for each of the two 32-year periods. We did the same for the number of months in
the monsoon season with total precipitation above 50 mm. These data are summarized in Table
2.4. Monthly precipitation is given as percent values because of the difference in the number of
missing values for the two periods. One important weather station, Phantom Ranch, has records
only from 1948. In this case, figures for the shorter pre-dam period are multiplied by 2 to
extrapolate to the 32-year pre-dam period, and these data should be regarded as approximations.

Table 2.4. Summary of Data Used for Investigating Precipitation Changes Before and After Emplacement of Glen Canyon Dam
Weather Station

Period
Analyzed

Percent of
Records
Missing

Elevation (m)

Number of Days with
> 25 mm Precipitation

Percent of Months with
> 50 mm Precipitation

1932-1963

1964-1995

1932-1963

1964-1995

Lower Colorado River
Boulder City, NV

1932-1998

2.27%

767.8

12

23

3.23%

5.93%

Parker, AZ

1932-1998

0.00%

128

10

10

3.31%

4.03%

Zion National Park, UT

1932-1998

0.05%

1234

15

17

15.32%

20.97%

Seligman, AZ

1932-1998

0.06%

1600

16

19

21.70%

27.87%

Kingman, AZ

1932-1993

2.58%

1051

44

26

13.04%

14.04%

Escalante, UT

1932-1998

0.10%

1770

25

11

25.86%

14.52%

Phantom Ranch, AZ*

1948-1998

6.81%

783.1

14

8

12.50%

9.01%

Grand Canyon, AZ

1932-1998

3.90%

2096

23

40

32.77%

34.15%

Fort Valley, AZ

1932-1998

0.07%

2240

25

38

49.19%

52.42%

Grand Junction, CO

1932-1998

0.08%

1475

4

4

9.68%

6.45%

Moab, UT

1932-1998

3.08%

1225

14

12

12.82%

9.09%

Bluff, UT

1932-1998

0.08%

1316

11

5

8.87%

7.26%

Lees Ferry, AZ

1932-1995

9.41%

978.1

11

9

5.61%

5.26%

Western Grand Canyon

K-M Plateaus

Lake Powell-Canyonlands

*Phantom Ranch record is for the 16-year period before the dam and was standardized to the 32-year period by multiplying by 2.

The Base-Level Hypothesis
H2: Gully erosion on alluvial terraces associated with archaeology has increased from pre-dam
time to post-dam time due to base-level effects.
Test 1: Report on rebuilding of sand bars that buttress pre-dam terraces and infill gullies in
Grand Canyon and in Cataract Canyon
In Grand Canyon, previous researchers reported on sand-bar rebuilding and gully infilling
associated with archaeological sites (Burchett, Coder, and Hubbard1996; Yeatts 1996, 1998).
Most of this work, conducted by repeat topographic surveys and repeat photography, centers
around the experimental test flood of 1996 and the resilience of these sand bars following that
time.
Using the concepts employed by Yeatts (1996), we conducted topographic mapping in
Cataract Canyon to show volume change in three areas prior to and following the annual spring
flood of 1999. The three areas—Rapid 4 (RM 211.3L), Cross Canyon (RM 208.4L), and Rapid
12 (RM 206.7L)— were selected from an 8-mile survey reach of Cataract Canyon that is closely
analogous to the Furnace Flats reach in Grand Canyon. These particular areas were selected
because they have a complete terrace sequence and river- and/or terrace-based streams within a
small area. Stream profiles and mouths were surveyed before and after the annual flood to detect
any loss or gain in volume of sediment. This comparison provided quantitative evidence of sand
bar dynamics in an undammed system and in an equivalent dammed system.
We rephotographed these three sites and others in Cataract Canyon for degree of gullying
and flood-sand renewal for the two-year period of study. These photos can be compared to
photo documentation of sites by RCMP monitors in Grand Canyon following the 1996 BHBF,
providing a simple means of annotating changes in areas representing pre-dam and post-dam
conditions.
We also investigated historical photo archives to show the relative elevation and extent of
sand-bar rebuilding around archaeological areas in pre-dam time in comparison to conditions
today. The photo archives used for Test 1 of the null hypothesis were used for this test as well.
Test 2: Assess catchment and river processes at each study site in Grand Canyon
The base-level hypothesis was further tested by conducting field work at each of the sites
listed in Table 2.1. We identified all of the alluvial terraces present, using characteristics and
syntax described by Hereford and others (1993) (Figure 1.2).
Specifically, we noted at each site characteristics of the process of channel integration
with the river. Using a deductive approach, we could eliminate processes that were completely
unrelated to the dam and its operations and assign points accordingly (Table 2.5). The geometry
of the river was recorded, along with any evidence of sediment renewal in the time directly
preceding the dam and in post-dam time. Sediment renewal refers to freshly deposited river sand

or active eolian sand that replenishes the mouths of streams and interrupts channel downcutting
or helps maintain sand bar elevation. We recorded whether gullies drain toward a recirculation
zone (upper pool, separation, or reattachment bar), point bar, or cutbank (channel margin bar).
Thus, we recorded presence of the pre-dam alluvium (pda), 1983 flood sand, and any active
eolian sand derived from these deposits. Presence or absence of these deposits indicates how
they may have blocked or slowed downslope channel evolution in the time just before and after
emplacement of the dam. Collectively, these observations provide clues to the validity of the
base-level hypothesis. We then rated the strength (or weakness) of the base-level hypothesis at
each site, using the values in Table 2.5, according to the documented processes at work. If a
particular geomorphic process or sand deposit was present, we assigned that characteristic a
value of 1. Scores for each site were totaled and evaluated for base-level control properties.
Table 2.5. Scores Assigned to the Presence of Specific Geologic Units or Geomorphic Processes
Description

Score

Geomorphic Process
gully not caused by side-canyon erosion

1.00

bank retreat at river level not caused by natural geometry of river

1.00

no evidence of pre-dam arroyo cutting (determined from 1965 air
photos)

1.00

Sandy Deposit Present
pre-dam alluvium and 1983

1.00

eolian dune (reworking of flood sand)

1.00

Maximum Score Possible

5.00

Note: The Maximum Score is used to test the strength of the base-level hypothesis.
DEVELOPMENT OF A PREDICTIVE GEOMORPHIC MODEL
A geomorphic model was developed to explain processes that cause erosion of high sand
terraces. Specifically, the model was designed to predict the relative vulnerability of the terrace
sequence to gully erosion, because these terraces lie in the runoff path of local catchments. The
model can be used by resource managers and monitoring teams to effectively assess and mitigate
adverse impacts to cultural materials that would otherwise be lost to erosion. It also identifies
the highest-risk sites where data recovery of archaeological materials should be a priority.

The Small-Catchment Geomorphic System in the Grand Canyon
Geomorphic settings in which cultural sites exist can be characterized by underlying
rocky material, which reflects the nature of the local hillslope or stream environment. The
configuration of the river corridor near river level is controlled by its resistance to erosion.
Easily erodible shales (e.g., Moenkopi Formation, Hermit Formation, Dox Formation, Hakatai
Formation, and Bright Angel Shale) tend to form wider and more gently sloped sections in the
river corridor. In addition, erodibility is increased in fault zones, where the rock is often highly
fractured. A wider river corridor with a more gentle slope has greater accommodation space for
the deposition and preservation of river terraces through time. The opposite is true for the harder
rock formations that form the narrow and steep-walled reaches of the corridor (e.g., the Supai
Gorge, Granite Gorges, and Muav Gorge) (Schmidt and Graf 1990).
The shape and size of tributary debris fans largely control the distribution, thickness, and
extent of sandy river terraces in the corridor. Gravel debris fans are formed by sediment-laden
floods and debris flows that episodically flow from the tributary canyons. A debris fan causes a
narrowing of the river that is followed by a wider area with a recirculation zone (eddy). Within
these wider areas, sand is deposited in the slower currents (Schmidt 1990). The size, shape, and
composition of these debris fans are functions of the rock type, catchment size, and slope of the
tributary canyons (Potochnik and Reynolds 1990). These variations in debris-fan complexes
create different response mechanisms in small catchments trying to reach equilibrium with the
post-dam floodplain of the river. Therefore, we sub-categorized fan complex settings, from
largest to smallest, as tributary plain, alluvial fan, and debris lobe. A number of cultural sites lie
at the base of talus slopes, in alluvium at the fringe of or away from the tributary fan complex;
this context is termed the talus-slope setting. Finally, many cultural sites in our study are in a
dune field setting, in which channels have formed. Although these areas are often associated
with tributary fan complexes, eolian processes are the dominant mechanisms that control the
stream environment. The five unique geomorphic settings are illustrated in Figure 2.3. Many
catchments are influenced by more than one of these settings, but we feel that each can be
categorized on the basis of the primary setting present.
To fully understand processes at each of these geomorphic settings, we established type
geomorphic settings with associated catchments established for detailed study. Small-scale
topographic maps were created by Horizons, Inc., for areas representing each of the five
geomorphic settings. These maps were produced by computerized photogrammetric technology
from 1998 aerial photography at 0.25-m contour intervals. Individual boulders 0.5-1.0 m in size
are clearly visible features on these maps, as are gullies and large headcuts with measurable
widths and depths. We sought map bases with a large enough scale to plot details within the
channel, such as areas of channel roughness, nickpoint locations, and stream cross sections.
Other data gathered for these sites include soil infiltration rates, grain-size distributions (R.
Hereford, written communication1998), surficial geology (Hereford 1996; Hereford, Burke, and
Thompson 1998), and location of archaeological sites.

Figure 2.3. Schematic illustration of the five geomorphic settings along the Colorado River.
Terrace-based streams drain to or above the 1983 flood sand; river-based streams cut through the
1983 sand.
The areas selected for this detailed work, with their respective catchments (designated by
letters), are:
•
•
•
•
•

Alluvial fan setting – Nankoweap (catchments O, Q)
Tributary plain setting – Palisades (catchments H, H')
Talus slope setting – Upper Unkar (catchments A, E)
Debris lobe setting – 122 Mile (catchments B, C)
Dune field setting – Lower Tanner (catchments A, C)

Data collected at these sites were archived in ArcInfo; thematic data are shown as layers
on the baseline topographic maps (Plates 2-6, in folder at end of document). Characteristics of
each geomorphic setting and its catchments are described in Chapter 3.
U.S. Geological Survey (USGS) maps (Burke and Hereford 1999; Hereford 1996;
Hereford, Burke, and Thompson 1998, 2000a, 2000b) with their comprehensive and detailed
descriptions served as an invaluable resource for this study and substantially increased the
efficiency of office and field work. These maps were useful for quickly assessing the detailed
geomorphology of field sites, understanding geomorphic units and their interrelationships, and
conceptualizing a framework in which to do detailed geomorphic analyses.
Topography in four of the five type-setting maps was tested for accuracy by overlaying
available digital versions of the USGS map topography produced by the Hereford group. Further

accuracy checking included comparison of random measurements of stream cross sections taken
in the field with computer-generated measurements of the same cross sections from the maps.
However, we found the scale of USGS maps produced by Hereford and others too small for
documenting subtle variations in landforms and their change through time. Furthermore,
existing GIS and archaeological site maps produced by Grand Canyon Environmental Studies
(GCES) and GCMRC that include our sites did not have the detailed coverage or accuracy
needed to complete our study, since they were not intended to cover the upper talus/bedrock area
and lower terrace areas by the river.
Developing the Matrix of Geomorphic Conditions
During the inventory of study sites, we identified geomorphic factors that clearly
influence catchment and river processes. These factors are categorized as either intrinsic or
extrinsic to catchment processes (Schumm 1977). Intrinsic factors control rainfall-induced
runoff within the catchment area and the terrace sequence that lies below. Extrinsic factors
include processes related to restoration or erosion outside of internal catchment processes.
Several factors measured in the field were later found to be unnecessary for development of the
mathematical model but were useful for the field-rating model and site descriptions.
Processes operating intrinsically within the catchment can be further divided into driving
and resisting factors (Bull 1991). The upper catchment or talus/bedrock portion contributes
runoff and funnels it onto the terrace sequence, which absorbs this energy (Figure 2.4). This
relationship can be conceptualized simply as a funnel perched above a sponge. When water is
poured into the funnel, the energy becomes concentrated as it is delivered onto the sponge, which
then dissipates the energy. When the sponge has absorbed the maximum volume of water it can
contain, runoff occurs. We employ this concept in formulating the mathematical model, which is
discussed in the next section of this chapter.
Intrinsic Factors of Catchments
Factors driving erosion occur within the bedrock and gravel region of the catchment,
upslope of the terrace sequence. These factors are measured as follows:
Driving Factors (upper catchment)
•

Drainage Area

•

Stream Length

total drainage basin area in square
meters measured on air photos and in field by survey
equipment; subdivided into bedrock, gravel, and sand
portions
total length in meters of main
thalweg measured on air photos and in field by survey
equipment

Figure 2.4. Schematic illustration showing process of channel integration with the river. The
upper catchment acts as a funnel that delivers water onto the terrace "sponge." Alluvial fan
progradation across the terrace accommodates the channel in reaching the river.
•
•
•
•
•

Runoff Efficiency

coefficient based on weighted mean
of area containing three types of lithologies (bedrock,
gravel, sand)
Upper Catchment Length
distance in meters from highest point in
upper catchment to uppermost sand terrace
Catchment Relief
total relief in meters from highest
point in catchment to uppermost sand terrace
Slope of Bedrock/Talus Portion
degree of slope as measured with a clinometer;
measurement taken from base of talus to top of catchment
in bedrock
Shale Factor
percent area of shale covering upper
catchment; determined from air photo analysis

Resisting Factors (occur in terrace portion of catchments)
•

Ground Cover

percent ground cover measured by
the point-intercept method
(vegetative and non-vegetative)
along a transect and categorized according to
vegetation/soil type: (1) woody, (2) herbaceous, (3)
organic detritus (includes cryptobiotic soil, deadfall,

•

Grain Size Distribution

•

Permeability

•

Width

•

Height

•

Sand Depth

•

Number of Terraces

driftwood, leaf and grass litter), (4) rock, (5) bare ground
field assessment of grain sizes (U.S.
Department of Agriculture hand texture method); also
identified whether sediments of eolian or fluvial origin
infiltration rates in inches per second
measured with Gwelph permeameter at each of the "Type
Catchment" sites
width of terrace in meters
from bottom of riser to back of tread
height of terrace in meters
from bottom of riser to back of tread
depth of terrace in meters as
estimated from cross section in nearby side canyon or
arroyo
number of terraces a channel must cross to
reach the river; counted from base of upper catchment to
the 1996 stage height
Extrinsic Factors of Catchments

Extrinsic factors drive erosion and restorative factors outside of catchment internal
processes. They are measured or assessed as follows:
•

Human Mitigation

•

Eolian Infilling

presence of checkdams or
revegetation
percent of active windblown sand
over area

•
•

Flood Flow Infilling
Human Impact

•

Side Canyon Overflow

presence of 1983 and 1996 flood sands
qualitative assessment of degree of
trailing, categorized as: (1) slight (footprints only), (2)
moderate (trail intersects channel), (3) severe (stream
established downslope along a trail)
presence of side canyon overflow that partly
forms channel

Field data were collected during four river trips through Grand Canyon and four river
trips through Cataract Canyon between February 1998 and August 1999. Since field crews were
dispatched to collect data, the protocols defined above were established to assure consistency in
data collection. Customized field data sheets (see Appendix A) were issued for crew members to
ensure complete data gathering for each catchment. All data were archived into Access® to
streamline data management. Data were manipulated in both Access® and Excel® to
consolidate and graph results.
It should be noted that none of these measurements were repeated because of the short

duration of this project. However, all measurements were systematically collected and were
carefully recorded so that repeat data collection may be carried out in the future. A clear
understanding of geomorphic processes is an essential prerequisite for the establishment of an
effective long-term monitoring program that will provide significant results through time-series
analyses.
STEPS IN ASSEMBLING THE PREDICTIVE MODEL
The second main objective of this study was to develop a quantitative model that would
predict the vulnerability of high terraces to gully erosion from local catchments. We explored
four separate approaches and chose a process-based mathematical model as the best method.
Statistical Model Approach
The purpose of the model is to predict which areas are most vulnerable to erosion and
which will erode first. Our first approach was to apply Principle Component, Multiple
Regression, and Cluster analyses to draw out relationships among variables. Moreover, we tried
through these multivariate analyses to reduce the number of measured variables and identify
those that are most discriminating. We were hoping that patterns in the data could be identified
through statistical relations among intrinsic geomorphic parameters. Unfortunately, we could
not find consistent patterns relating these parameters to the actual degradation of study sites.
This failure did not reflect a lack of causal relationships, but resulted from the discrete
nature of major erosion events induced by rainfall. Catchment basin characteristics certainly
relate to vulnerability (Leopold and Maddock 1953; Schumm 1977); however all sites do not
equally show that vulnerability if they have not yet received the random high-rainfall event that
will cut a large channel. Some sites have received such events and thus exhibit high erosion
characteristics. Others with similar physical properties show little channel degradation, probably
because they have not received the same type of large rainfall events. Multivariate analyses will
thus fail to show relationships, because the entire range of erosion conditions is possible with the
same physical characteristics. This confounding problem resulted in the development of a more
mechanistically based model with response frequencies as an indicator of the model.
Sediment yield, which is the weight or volume of sediment eroded from a unit area, was
also investigated for its usefulness in modeling gully erosion in terraces. Several empirical
equations have been developed to estimate sediment yield, but these models, such as those
developed by Fournier (1960) and Flaxman (1972), generally relate sediment yield to climate
and watershed variables. Since it was impossible to measure rainfall at each Grand Canyon site,
we felt that a customized approach would be more useful. Moreover, these models are designed
for large catchments; sediment yield is extremely difficult to measure in the small-catchment
environment of our study sites, and figures obtained would falsely reflect the rate of gully
erosion. Critical variables such as those indicating rainfall/runoff relations, watershed
morphometry, sediment grain size, and vegetation are the focus of our conceptual process model.

We also explored use of a field-rating model for consistently recording observations on
each site and providing qualitative assessments of site vulnerability. The field-rating model uses
geomorphic parameters (identified in Table 2.6) that were considered to be the most critical in
assessing vulnerability in the field (Leopold and Maddock 1953; Schumm 1977). Other standard
methods for

Table 2.6. Point Scale for Parameters within the Catchment
Bedrock/Talus Portion
Catchment Area
(m2)

Slope of Talus/Bedrock (degrees)

Slope Aspect (degrees)

1

< 200

0-10

0-44

2

200-700

11-20

44-89, 315-360

3

700-5000

21-30

90-134, 270-314

4

5000–20,000

31-40

135-179, 225-269

5

20,000–100,000

41-50

180-224

6

> 100,000

51-60

Score
0

7

61-70

8

71-80

9

> 80

Sandy Portion
Score

Slope of Sandy

Width of Sandy

Bare Ground

Number of

Terraces (percent)

Terraces (m)

(percent)

Terrace Levels

> 250

<9

6

none

5

slight

4

moderate

3

severe

0
1

<6

201-250

2

6-10

151-200

3

11-15

101-150

4

16-20

50-100

5

21-25

<50

6

26-30

7

> 30

9-20

21-40

Human Trailing

1-2

41-60

Note: Quantitative breaks in data follow standard method of rating sediment yield (PSIAC rating [PSIAC 1968]) or are statistically
derived from frequency distributions.

estimating soil erosion (Universal Soil Loss Equation) or sediment yield (PSIAC Rating [Pacific Southwest Inter-agency Committee 1968]) use
these same parameters, with the exception of "number of terrace levels." Based on empirical evidence in Grand Canyon, we determined that the
stepped sequence helps to slow and dissipate stream power and that a higher number of terraces therefore increases resistance to erosion.
While this method helped us to identify high-risk sites in the field, it did not rank sites of low to medium risk very effectively.
Moreover, poor correlation was found when plotting the total vulnerability rating against actual degradation of the ap terrace (see Plate 1 for
terrace descriptions). Therefore, we investigated a mathematical model that focused primarily on process.

Mathematical Model Approach
We developed a mathematical model using both Grand Canyon and Cataract Canyon data. The model is based on the concept of the
upper catchment acting as a funnel for discharging water onto the terrace sequence below, with the terrace sequence acting as a sponge to absorb
water and dissipate stream power (Bull 1991) (Figure 2.4). We used simple hydrologic equations to derive the vulnerability of terraces for each
site in developing this process-based model.
For erosion-driving factors, we evaluated total runoff volume (Q) and peak discharge (Qp) from a hypothetical storm of 25 mm (1
inch)/hour to determine which of these factors is more accurate in representing the erosion-driving forces of a storm event. This event was
chosen based on previous research on debris flow initiation (Griffiths, Webb, and Melis 1996). We assume that when a terrace becomes saturated
with water and can no longer behave like a sponge, excess runoff induces gully cutting. Thus, the resistance of the terrace is fundamentally a
function of its volume and permeability (diffusion capacity).
The volume of a terrace is assumed to be laterally infinite. Therefore, the critical dimension of a terrace’s ability to absorb water is the
shortest distance water must travel to reach the terrace riser, which we call terrace width (Wt). The other critical dimension is the thickness of the
sand terrace above its rocky base, which we call terrace depth (Dt). In our model, cross-sectional area of a terrace (Axt) is the primary measure
of a terrace’s resistance to erosion, where:
Axt = Wt × Dt

(equation 1)

Other workers have suggested that a rainfall event of 25 mm/hour initiates debris and stream flows in tributary side canyons (Melis et
al. 1996). Storm systems of this type are assumed to be the same ones that cause severe erosion of terrace-based catchments (Hereford et al.
1993). For simplification of modeling, we assume unsaturated conditions and do not include possible antecedent conditions of partial soil
saturation. We therefore assume that lateral diffusion forces, which are inherent in saturated conditions (Fetter 1980), are negligible compared to
downward gravitational forces, which are dominant in unsaturated conditions.

In our model, the upper gravel-bedrock portion of the catchment delivers a flow event of 25 mm/hour to the terrace portion, which we
assume to be unincised for simplicity of modeling. Essentially, we are modeling the balance between the erosion-driving stream power of water
in the upper catchment and the diffusion capacity of terraces below, referred to as the geomorphic threshold (Schumm 1977). This basic part of
the model, which we address as Model 1, determines a site’s vulnerability to gully incision. Other variables unique to each site, incorporated into
the basic model as Model 2, help refine the vulnerability index of Model 1.

Model 1
First, Model 1 subjects the uppermost sandy terrace to total runoff volume (Q) from the hypothetical storm event. The vulnerability of
the terrace (Qvul) is expressed as a percentage of the Q ratio:
Q vul = 100 (Q ratio/maximum Q ratio),

(equation 2)

where Q ratio is a function of runoff from the upper catchment compared to resistance of the terrace below. The Q ratio in equation (2) is
expressed as:
Q ratio = ln Q / ln Axt

(equation 3)

Equation (3) simply normalizes the Q ratio for ease of graphing. The value Q is calculated using a modification of Fetter's (1980) Rational
Equation and is expressed as:
Q = (I) (At) (C),
where

Q = total volume of runoff (m3

(equation 4)

)

I = total rainfall (m)
At = total catchment area (m2)
C = runoff coefficient
The value for C in equation (4) varies with three typical geologic surfaces (bedrock = 0.85,
gravel = 0.35, and sand = 0.01 [American Society of Civil Engineers 1970]). It is calculated by
the following equation:
C = (Ab * 0.85) + (Ag * 0.35) + (As * 0.10)/At,
where Ab = bedrock catchment area (m2)
Ag = gravel/talus catchment area (m2)
As = sand catchment area (m2)

(equation 5)

Next, Model 1 subjects the uppermost terrace to peak flow discharge, Qp (m3sec-1), from
the hypothetical runoff event. We determine this terrace vulnerability (Qp vul) according to the
equation:
Qp vul = 100 (Qp ratio/maximum Qp ratio),

(equation 6)

where, similar to the terms of equation (2), Qp ratio is a function of peak flow discharge from the
upper catchment compared to resistance of the terrace below. The Qp ratio in equation (6) is
expressed as:
Qp ratio = ln Qp / ln Axt

(equation 7)

Qp is derived from the following equation (McCuen 1982, in Western Water Consultants, Inc.,
1992):
Qp = 484 * Q/(0.6665 * Tc),

(equation 8)

where Q is the total runoff volume in mi2-in and Tc is the time of concentration (the time it takes
a particle of water in the most distant part of the basin to reach the outlet) in hours. Tc is
expressed as:
Tc = (Ls/0.3047)1.15 / 7700 * (H/0.3047)0.385,

(equation 9)

where Ls is the length in meters of the main stream of the basin and H is the relief in meters
between the most distant drainage divide and the outlet (Fetter 1980).
Model 2
From our field work at type geomorphic settings (Plates 2-6), we can generally
distinguish pre-dam and post-dam sandy deposits. Based on grain-size research (R. Hereford,
written communication 1998), we can assume that all pre-dam river terraces are composed of
poorly sorted fluvial sand (silty fine sand) and have relatively uniform permeability (mean K =
0.005) (Table 2.7) compared to post-dam river terraces, which are composed of well-sorted fine
sand (mean K = 0.05). This comparison illustrates an order-of-magnitude difference between the
two main types of sediment. Furthermore, permeability of eolian deposits (mean K = 0.03)
roughly mimics that of post-dam sand. Therefore, we lump post-dam and eolian deposits
together when assessing their ability to absorb water.

Table 2.7. Infiltration Rates of Pre-dam, 1983, and Eolian Sand Deposits
Geomorphic
Feature
Talus
sa Terrace/Alluvial
ap Terrace
1983 terrace
Eolian sand
Gravel*
Kaolinite (clay)*

Hydraulic Conductivity K (× 10-3 cm/sec)
Mean

Range

Number

50.0
6.0
5.0
50.0
32.0
370.0
0.0007

2.0-191.0
5.0-7.0
1.0-10.0
1.0-97.0
11.0-68.0
N/A
N/A

5
3
5
5
3
N/A
N/A

*Not measured in the field; values from Davis and DeWiest (1996).
Four geomorphic variables within terraces are also incorporated into the model (Figure
2.5). Two of these variables reduce the resilience of the terrace "sponge," whereas the other two
increase it. The sum of these variables, which modifies the diffusion capacity of terraces, is
stated as follows:
TF = SF + WF + HF + EF, and -0.5 < TF < 0.5

(equation 10)

The shale factor (SF) is a function of shale comprising part of the upper catchment. This
factor decreases permeability of water into the terrace by depositing clay and providing material
to build alluvial fans across the terrace surface (Griffiths, Webb, and Melis 1996). SF is derived
from the equation:
if [(ln (As + 1)/ ln (Lt + 1)] / (max) * 100 > 10,
then SF = -0.5; otherwise SF = 0.0

(equation 11)

The numeric values representing how much SF and other factors contribute to TF were estimated
in the field.
In equation (11), As = the upper catchment area composed of shale in square meters, and max =
the maximum value of [ln(As + 1)/ln(Lt + 1)].
The woody vegetation factor (WF) decreases diffusion of water into the terrace by
increasing piping and seepage erosion along roots (Parker and Higgins 1990). It is calculated as
follows:
WF = (woody veg) * -0.5,

(equation 12)

where woody veg is the fraction of woody vegetation to terrace width (both in meters).

Figure 2.5. Flow chart showing inter-relationships of geomorphic factors and model
development.
The herbaceous vegetation factor (HF) accounts for ground coverage of cryptobiotic soil,
grasses, forbs, and organic matter. It increases diffusion of the terrace “sponge” (Schumm 1977)
according to the equation:
HF = (herb veg) * 0.25,

(equation 13)

where herb veg is the fraction of herbaceous ground cover to terrace width (both in meters).
The eolian factor (EF) accounts for reworking and sorting of sand by eolian and river
processes. This factor increases permeability and diffusion of water into the terrace (Pye and
Tsoar 1990) by winnowing out the silt and clay fraction from the sand. If the geologic
description of a terrace is eolian or 1983 sand, then:
EF = 0.5; otherwise EF = 0.0
(equation14)
Taking these four factors (equations 11–14) into consideration, the raw vulnerability

quotient (Vr) of any individual terrace to erosion is independent of its position in a terrace
sequence. It is as if the individual terrace were placed at the base of the catchment area where it
would receive the full force of the runoff. This condition can be expressed as:
Vr = ln(Q or Qp)/ln[Axt * (1 + TF)]

(equation 15)

For ease of graphing, values for Vr are normalized and subsequently referred to as the final
vulnerability quotient for individual terraces (FVi), expressed as:
FVi = 100 * Vr / max(Vr),

(equation16)

where max(Vr) = maximum Vr for all upper terraces using Model 1.
In most cases, an individual terrace is located below another terrace, which modifies its
vulnerability quotient. A cumulative vulnerability quotient (FVC) was calculated for each
terrace in the sequence as the product of its individual vulnerability quotient (Fvi) and the
cumulative vulnerability from the previous terrace (FVCi-1):
FVCi = (FVi x FVCi-1)/100

(equation 17)

We obtained average vulnerability for each catchment by calculating the mean of all terraces
(FVCi) per catchment (listed with Site Descriptions in Appendix B).
The mathematical model includes factors that are intrinsic to the rainfall-runoff process
and its effects on a pristine terrace sequence. The model does not include extrinsic factors that,
on a case-by-case basis, may strongly dictate the vulnerability of the terrace sequence at a site.
Five factors, evaluated for each site, act to modify some of the predicted vulnerabilities from the
mathematical model: (a) human impact, (b) side canyon overflow, (c) climate gradient, (d) (river
sand renewal, and (e) eolian processes. These data are summarized in the Site Descriptions
(Appendix B).
DEVELOPMENT OF THE CLIMATE PARAMETER
FOR THE GEOMORPHIC MODEL
To detect a climate gradient as it pertains to our study area, we analyzed warm-season
(summer monsoon, July-October) precipitation data specifically at river-level weather stations
throughout the Colorado River corridor. The analysis extends to the fringe of the corridor that
receives monsoon-season precipitation, from Moab, Utah, to Parker, Arizona (Hereford and
Webb 1992; WRCC 1999). Analyzing only stations at river level eliminated local orographic
effects on precipitation trends, because higher-elevation weather stations are not reliable
indicators of precipitation patterns along the river (Melis et al. 1996). Data were analyzed for
post-dam time, so that a climate gradient is consistent with the post-dam era in its application to
the final vulnerability index. The data are summarized in Table 2.8 according to mean annual
precipitation, mean precipitation for the monsoon season (July-October), mean precipitation for

the peak month in the monsoon season, and number of days exceeding 25 mm (1 inch) of
precipitation for the monsoon season. The data are concerned primarily with warm-season
rainfall, because many arroyo-cutting episodes occur during that time (Hereford and Webb
1992). Table 2.8 summarizes the results of the climate parameter, which are discussed in
Chapter 3.
Table 2.8. Summary of Precipitation Data from River-Level Stations for Thirty-two Years
Following Emplacement of Glen Canyon Dam
Weather Station

Period
Analyzed

Mean
Elevatio Mean Annual
Mean
n
Precipitation Precipitation for Precipitation for
(m)
(mm)
Peak Month of
Monsoon
Monsoon
Season (mm)
Season
(mm)

Lower Colorado
River
Willow Beach, AZ

1967-1998

232

141

47

15

Davis Dam, AZ

1967-1998

177

148

39

13

Parker, AZ

1967-1998

128

122

44

16

1967-1998

783

233

97

33

Fruita, CO

1967-1998

1371

223

82

23

Dewey, UT

1967-1998

1255

236

81

32

Moab, UT

1967-1998

1225

220

86

27

Lees Ferry, AZ

1967-1995

978

153

74

26

K-M Plateaus
Phantom Ranch, AZ
Lake Powell–
Canyonlands

Note: No river-level stations available for Western Grand Canyon.

CHAPTER 3
RESULTS AND DISCUSSION
Kate S. Thompson, Andre R. Potochnik,
Ron Ryel, and Lynn A. Neal
HYPOTHESIS TESTING
Null Hypothesis
H0:

Terrace erosion rates and gully erosion rates on the terraces have remained
unchanged from pre-dam time to post-dam time.

Test 1: Determine when large gullies formed in Grand Canyon
Assessment of data, including air photos, of 23 representative catchments (see
Appendixes C and D) from three geomorphic reaches between Soap Creek and Granite Park
provides a synoptic view of large-arroyo development in pre-dam terraces in Grand Canyon.
The pre-dam dynamic equilibrium conditions between arroyo incision and river deposition in the
terraces are represented by photos taken in 1965.
Air photo analysis shows that erosion has profoundly increased since 1965 (Figure 3.1).
Less than 40% of today’s catchments were partially formed gullies in 1965, but by 1992 arroyos
had fully formed in about 95% of catchments. The time period of the most dramatic increase is
from 1973 to 1984, when new gullies were formed and most pre-existing gullies had matured
into arroyos (see Appendix C). These results coincide with those of Hereford et al.(1993) who
conclude that a cycle of arroyo-cutting began around 1973 and continued until 1984.
In extrapolating the rate of arroyo incision within this 32-year period to the 32-year
period before the emplacement of Glen Canyon Dam, we should expect that 100% of the
catchments examined would show some amount of terrace incision by1965. However, periods
of arroyo cutting in the Southwest are cyclical. An extended period of arroyo cutting from about
1890 to the early 1940s (Leopold 1976) was followed by several decades of arroyo filling and
floodplain formation (Hereford 1987). Despite extensive investigations, researchers have not
reached a clear consensus on the causes of these changes (Cooke and Reeves 1976; Graf 1983).
Any arroyos cut in Grand Canyon in pre-dam time probably coincided with turn-of-the-century
arroyo cutting throughout the Southwest, but this assumption can only be deduced at this point.
Nonetheless, evidence of a renewed period of arroyo cutting in Grand Canyon in post-dam time
suggests that the amount of gullying has increased since pre-dam time and therefore fails to
support the null hypothesis.

40

Chapter 3

Figure 3.1. Degree of arroyo development since 1965 based on air photo analysis at 23 catchments.
As discussed in Chapter 2, historical photos could help define the time of arroyo cutting in
Grand Canyon. There are numerous historical photos of the river corridor that could collectively
provide a time frame during which gullies have developed in the canyon. However, such photos
are extremely difficult to find, as proven by our extensive search in different archives; we found
none except those uncovered by previous researchers, which are used here for this test.
Test 2: Compare amount of gully erosion on terraces in Cataract Canyon to those in Grand Canyon
Understanding the relationship between correlatable terraces and the alluvial history of
Cataract Canyon provides a basis for comparison of gully incision on terraces in Cataract
Canyon and in Grand Canyon. The geomorphology in Cataract Canyon is very similar to that in
Grand Canyon, as discussed in Chapter 2. In our analysis of pre-dam annual hydrographs and
sediment supply for the two regions, differences in amplitude are evident, yet patterns in the data
roughly parallel each other.
The annual water yield through Cataract Canyon was, on average, 84% of the total water
yield at Lees Ferry for the period 1929-1986. Data for Cataract Canyon are approximated by
summing the water yields of the Colorado River at Cisco, Utah, and the Green River at Green
River, Utah. The difference in water yield between Cataract Canyon and Grand Canyon is due
primarily to input from the San Juan River (USGS 1999), which joins the Colorado River below
Cataract Canyon.

Results and Discussion

41

Annual flood flows in Cataract Canyon show a pattern similar to those at Lees Ferry (Table
3.1.). Of the highest flood peaks during the 1900s, those in Cataract Canyon average about 80%
of those at the Lees Ferry gauge (USGS 1999). The hydrograph for the last large annual flood
before the dam, in 1957, shows the same general pattern for Cataract Canyon and Lees Ferry.
Additional spikes in the hydrograph for Lees Ferry are a result of San Juan River monsoon flows
(USGS 1999).
Table 3.1. Historical Peak Flow Events in Cataract Canyon
Green River
(cfs)

Colorado River
Cisco/Fruita (cfs)

Date

Cataract Canyon1
(cfs)

95,0002

125,000

July 4, 1884

220,000

62,200

72,0003

June 10, 1909

134,000

68,100

76,800

June 27, 1917

130,000

65,500

89,100

June 15, 1921

44,700

65,000

42,900

Lees Ferry (cfs)
300,0002

(7/7/84)

154,000

220,000

(6/18/21)

June 3, 1928

109,700

127,000

(7/01/27)

64,200

June 10, 1957

101,000

126,000

(6/12/57)

44,800

61,900

June 27, 1983

106,000

97,300

(6/29/83)

48,300

70,300

May 27, 1984

115,000

58,200

(8/12/84)

29,800

51,900

June 18, 1995

82,000

45,000

(4/1/96)

1

Flow estimated by adding Green River and Colorado River flows; peaks are offset for each river gage
and do not necessarily add up to Cataract peak flow.
2
No gage recording; approximated from hydrologists' data (Dickenson 1944).
3
Compiled from Fruita gage plus estimate of Dolores River flow.

Historical records of annual suspended sediment supply vary considerably between the two
reaches. For the period 1930-1980 the San Juan River contributed 48% of suspended sediment
to Lees Ferry, compared to 52% from Cataract Canyon (USGS Water Resources Division data,
W. L. Graf, personal communication 1993). These analyses indicate that some terrace heights
and widths in Cataract Canyon may not necessarily match those in Grand Canyon, depending
upon the timing of San Juan River peak discharges.
We identified 13 sites in Cataract Canyon with clearly recognizable terrace sequences and
geomorphic settings similar to those in Grand Canyon. Table 3.2 provides a list of sites where
we have conducted field investigations, along with the type of data collected at each site. The
focus of our work was in the "survey reach" from Spanish Bottom (RM 213) to Range Canyon
(RM 205) (Belknap, Belknap, and Evans 1991).

Table 3.2. Study Sites in Cataract Canyon with Data Indicating River Terrace Sequence
Site Location
215 Mile

Geomorphic
Setting

Terrace
Level

Talus Slope

T3

(RM-215.0L)

Terrace
Height (m)

Talus Slope

T5

Talus Slope

(RM-213.4R)

Talus Slope

Hy (2)

86, 60

3

Grain Size
(% si+c)

mature tamarisk
7.8

hackberry
desert scrub

T5
T3

mature tamarisk

T2

hackberry

T1

desert scrub
1.6

T4

2.5

young tamarisk

T3

3.7

mature tamarisk

Tk

36

T2

5.5

mature tamarisk

GW

72

T1

6.7

desert scrub

T5

2.0

T4

3.3

young tamarisk

T3

4.0

mature tamarisk

T2

6.2

hackberry

Alluvial Fan

T5

2.5

(survey site)

T4

3.7

Talus Slope

(RM-212.1R)

(RM-211.3L)

44

T5

(RM-212.7L)

Rapid 4

GW

Artifacts Found

2.5

T1

Brown Betty Camp

Tree Ring
2
Count (yrs)

desert scrub

T3
T2

Red Lake Canyon

Tree Core
1
Sample

T2

(RM-213.7R)

213 Mile

Driftwood
Present

young tamarisk

T1
214 Mile

Vegetation
Community

young tamarisk

56

1999

10

1995

20

4

Table 3.2. Study Sites in Cataract Canyon with Data Indicating River Terrace Sequence, continued
Site Location

Geomorphic
Setting

Rapid 4, cont.

Cross Canyon
(RM-208.1L)

Terrace
Level

Terrace
Height (m)

Vegetation
Community

Driftwood
Present

Tree Core
1
Sample

Tree Ring
2
Count (yrs)

T3

5.1

mature tamarisk

1984

Tk

36

T2

6.1

Hackberry

1921

Hy

38

T1

10.3

dunes/desert scrub

Alluvial Fan

T5

(survey site)

T4

1.7

T3

3.0

T2

Rapid 12
(RM-206.7L)

Range Canyon

T1

4.2

Tributary
Plain

T5

3.1

(survey site)

T4

4.5

T3

Debris Lobe

(RM-204.8R)

Big Drop 1
(RM-202.8L)

Talus Slope

Artifacts Found

3

diagnostic aluminum
cans, bottles

Grain Size
(% si+c)
48
44

preceramic flaked
stone debitage, tools

24

1999

36

1995

36

mature tamarisk

1984

cottonwood/
hackberry

1921

desert scrub

1884?

Cd(2); Hy

77, 85; 159

diagnostic aluminum
cans, bottles

36

historic hearth (?)

46

prehistoric flaked stone

52

1999

34

young tamarisk

1995

30

7.4

mature tamarisk

1984,
1952?

Tk

43

T2

9.4

cottonwood/
hackberry

1917?,
1921

Cd, Hy

76, 65

T1

11.6

desert scrub

1884?

T5

0.9

T4

1.9

young tamarisk

1995

T3

3.5

mature tamarisk

1984

T2

4.5

hackberry
1999
scour line

1995

24

SCA5 bottles, wire
nails, saw-cut railroad
ties
46

1999

T5
T4

diagnostic cans and
bottles

diagnostic aluminum
cans, bottles

4

Table 3.2. Study Sites in Cataract Canyon with Data Indicating River Terrace Sequence, continued
Site Location

Teapot Canyon

Geomorphic
Setting

Debris Lobe

(RM-202.7R)

Ten Cent

Tributary
Plain

(RM-201.1R)

Imperial Canyon

Terrace
Level

Terrace
Height (m)

Vegetation
Community

Driftwood
Present

T3

mature tamarisk

1984

T2

Hackberry

1921?

T5

2.3

T4

4.2

young tamarisk

1984

T2

5.8

scour line

1921,
1957?

T1

8.9

desert scrub

1884

(RM-200.3R)

Tree Ring
2
Count (yrs)

Hy

young tamarisk

1995

wire nails
square nails in plank

1984

T3

4.4

mature tamarisk

T2

6.0

hackberry/Apache
plume

T1

8.8

desert scrub

T5

1.6

lake

T4

3.0

lake

T3

5.1

scour line

1984

1984 fire extinguisher

T2

7.5

scour line

1921

diagnostic cans and
bottles

T1

9.1

desert scrub

1884

square nails in beam

Cross-checked by Arizona State University Dendrochronology Laboratory.

3

Artifacts identified and recorded by Kate Thompson; identifications confirmed and expanded by Lynn Neal, SWCA Archaeologist.
Grain size reported as percent silt and clay as determined by hydrometer method.

diagnostic aluminum
cans, bottles

1884

GW = Goodding willow, BE = box elder, Hy = hackberry, Tk = tamarisk, Cd = cottonwood

SCA=sun-colored amethyst

35

1999

2

5

3

diagnostic cans and
bottles

1

4

Artifacts Found

1999

T5
T4

Tributary
Plain

Tree Core
1
Sample

Grain Size
(% si+c)

4

Results and Discussion

45

For each site, we evaluated catchment processes, river dynamics, terrace sequences, and
geomorphic setting. After completing a description of the terraces at 13 sites, we correlated terrace
sequences using standard field techniques (see Chapter 2). Data on terraces include vegetation
community, terrace height, dendrochronology, identification of modern and historical artifacts in
driftwood piles, and identification of prehistoric artifacts (Table 3.2).
Five distinct terraces can be correlated throughout Cataract Canyon, several of which directly
relate to the same flood events in Grand Canyon. Figure 3.2 shows a schematic illustration of the
complete terrace sequence. This reconstruction is based solely on data we have collected to date,
depicted in Table 3.2 and cross-referenced with the streamflow data in Table 3.1.
Stratigraphic levels are identified by age, from oldest to youngest. The youngest terrace, T5,
forms the modern floodplain, which was built by flows up to 50,000 cubic feet per second (cfs)
(Figure 3.2). This level provides low-water campsites for raft trips, is generally devoid of
vegetation, contains fresh, small driftwood piles, and is a well-sorted medium sand (10-36% silt
and clay) (Table 3.2).
The T4 level was formed by flows up to about 75,000 cfs. It was overtopped by the flows of
1995 and 1997 (Table 3.1.), which either buried immature tamarisks or ripped them away. The
result is a vegetation community that consists of immature tamarisk or tamarisk seedlings and
sparse annual grasses. Driftwood piles are somewhat larger than those on T5 and contain a high
proportion of modern-day trash. Sediment is typically a well-sorted sand with about 20-36% silt
and clay. The T4 and T5 terraces in Cataract Canyon cannot be correlated directly to the same
flood events in Grand Canyon because of the presence of Glen Canyon Dam and its regulated flow.
The T3 level was formed in the late 1950s by floods exceeding 100,000 cfs (R. H. Webb,
personal communication 1998) and was overtopped by the floods of 1983 and 1984 (Table 3.1).
Mature tamarisk up to 45 years in age commonly grow on this level and are often buried by the
1984 flood sand. Large driftwood piles are silver in color and contain the first all-aluminum
beverage cans (1957-1959) (Goodman 1998) and other diagnostic trash from the 1950s through the
1980s. This flood sand characteristically has 24-48% silt and clay. The T3 terrace is correlated
with the pda terrace in Grand Canyon, on which the 1957 flood driftwood line is a distinctive and
ubiquitous marker. The T3 is the key marker terrace by which critical channelization processes can
be directly compared to those in Grand Canyon.
The T2 level was probably formed by floods near the beginning of the twentieth century,
including a maximum flood of 155,000 cfs in 1921 (Table 3.1). This distinct terrace is often
associated with a mature hackberry or Apache plume community interspersed with rabbitbrush.
The oldest hackberry trees are 80-100 years in age, based on a previous study of hackberry
dendrochronology (Salzer et al. 1996). Our tree-ring results show the oldest hackberry to be 159
years in age, but this tree is likely associated with a slightly higher level. We found evidence of
1921 driftwood piles inset against the riser of the T1 terrace. The 1921 driftwood piles are
commonly checked and deeply weathered, with diagnostic artifacts such as wire and square-cut
nails,

46

Chapter 3

Figure 3.2. Schematic cross section of river terrace sequence, Cataract Canyon.
heavily weathered rough-hewn milled lumber, and UV-affected sun-colored amethyst glass. The
grain size is silty fine sand (36-46% silt and clay), and several flood-sand packages make up the
terrace. In the large side canyon at Rapid 12, a T2 cutbank exposure reveals three discrete river
flood deposits interbedded with side-canyon stream flow deposits. In addition, we observed cutand-fill structures within T2, suggesting that a balance between gully cutting and subsequent
infilling of those gullies by the river is a process that has occurred historically in Cataract Canyon.
The T2 terrace correlates with the lmt terrace of Grand Canyon in that both are said to have
been overtopped by the 1921 annual flood (Hereford, Burke and Thompson 1998), which was
220,000 cfs at the Lees Ferry gage and estimated to be 154,000 cfs in Cataract Canyon (USGS
1999). This information is roughly confirmed by our average tree-ring dates for the T2. We
assume that the older trees represented the initial germination conditions following a large flood
and that T2 has not been flooded since that time (R. H. Webb, personal communication 1998).
Driftwood and associated artifacts of the early 1900s are distributed across the entire T2 surface.
The highest terrace level, T1, is a prehistoric terrace that was built by floods exceeding about
225,000 cfs. These sands are commonly mounded into eolian dunes in which prehistoric artifacts
such as lithic flakes and tools can occasionally be found. On August 2, 1999, during the final
mapping trip through Cataract Canyon, SWCA archaeologist Lynn Neal accompanied the
geomorphology team to document cultural resources at two study areas. The first site is on the left

Results and Discussion

47

bank of the river above Rapid 4, creatively named the Rapid 4 site. The site’s artifacts are on T1
and are exposed primarily in deflated areas surrounded by low eolian dunes and debris-flow
mounds to the south. The site is in a low, west-northwesterly sloping swale bounded on the west
by a gully, on the east by a side canyon to the river, on the north by the riser of T1, and on the
south by the debris-flow mounds. Artifacts were flagged and described, and the site was then
mapped using a tape and compass. Flaked stone tools were point located on the map and were also
drawn (Figure 3.3). The size and quality of manufacture of these dart preforms and the associated
debitage suggest a preceramic or pre-Puebloan period (pre–A.D. 300-500) use or occupation.

Figure 3.3. Diagnostic flaked stone tools from the Rapid 4 site: white silicified chert (PL1), white
chalcedony (PL2), red jasper (PL3), red jasper (PL4), and red jasper/limestone (PL5). Tools are
shown actual size, in plan and cross section of each broken edge.
The Rapid 4 site contains well over 1000 flaked stone artifacts densely concentrated in a
roughly 14 × 11-m area at the site’s center, surrounded by a moderately dense flaked stone
concentration. The entire site area measures 78 m east-west by 29 m north-south. The majority of

48

Chapter 3

artifacts are medium and small tertiary flakes generated from interior core reduction and initial
biface reduction; late-stage biface thinning pressure flakes are present as well. There are some
large tertiary flakes, a few secondary flakes, and no primary flakes. The cherts, jaspers, and
chalcedonies that are present are all of high quality and appear to be heat treated. Great care was
taken in securing premium raw materials to reduce (by percussion), heat, and then pressure flake,
making large, thin knives and, presumably, dart points. Colors and types most represented are red
jasper; white, mottled red and orange, brown, and gray (light and dark) cherts; white and pinkish
red translucent chalcedonies; and yellow jasper. Some of the chert is silicified, and some (browns
and oranges) has limestone attached; all of the material appears to have been acquired locally or
within a relatively short distance. In summary, this site is a preceramic period limited-activity
flaked stone scatter. The site’s artifacts indicate late-stage interior core reduction, meaning that the
raw materials were being reduced before being transported to the site, where they were further
reduced and shaped into tools.
The second site is also on the left side of the river, downstream from Cross Canyon and just
above Rapid 10. Neal documented what appeared to be the first evidence of a site being exposed
on August 4, 1999, consisting of eight light gray chert flakes (1 secondary, 4 tertiary, and 3 biface
thinning) in an area less than 5 m in diameter. These flakes represent a single-episode reduction of
an interior core and subsequent biface reduction. The artifacts are probably eroding from the
southern slope of a T1 dune, since the flakes were visible only in the vicinity of rodent holes.
These sites are not unlike those documented at roughly the same terrace level the Grand
Canyon. Preceramic sites in the Grand Canyon often date to the Late Archaic, as evidenced by
projectile point types (Fairley et al. 1994). These aceramic and Archaic sites are often found
deeply buried in alluvium of Puebloan age (ap), a possible T1 equivalent, or on or near the surface
of the striped alluvium (sa), which represents the highest T1 surface in Cataract Canyon.
The front edge of the T1 terrace was probably overtopped by the flood of 1884 (Table 3.1).
Extremely deteriorated driftwood lines set into the soil were observed in this position at several of
the sites. At Calf Canyon and Imperial Canyon enormous, extremely decayed driftwood piles,
containing less than 5% milled lumber with square-cut nails still in the beams, lie above both the
1984 and the 1921 piles (Table 3.2). The height of the piles, the deteriorated condition of the
wood, and the artifacts found seemingly correlate with the 1884 driftwood in Grand Canyon as
described by Hereford and others (1993). One barely discernible driftwood line was identified
beyond the 1884 line, but we found no associated artifacts or datable materials to indicate whether
this material is older than the flood of 1884 . The vegetation on this level consists of mature desert
scrub plants such as four-wing saltbush, Mormon tea, and prickly pear cactus, all of which are
growing out of very thick (3-5 cm), dark cryptobiotic soil. The sedimentology is also distinct,
consisting primarily of silt (46-56% silt and clay), except for dune portions that are primarily fine
sand (24-34% silt and clay).
The T1 terrace of Cataract Canyon is broadly equivalent to the ap and sa terraces of Grand
Canyon. Its height above the river, the xeric vegetation community, the gentle slope toward the
river, the presence of preceramic lithic materials on the surface, and its great thickness indicate that

Results and Discussion

49

the T1 is many hundreds of years in age. It is as laterally extensive in Cataract Canyon as the ap
and sa terraces in Grand Canyon. Driftwood was deposited on its front edge during the late 1800s
by either the 1884 flood (Table 3.1) or the more poorly known and probably much larger flood in
about 1862 (O’Connor et al. 1994). Evidence for the 1862 flood has not been identified in Grand
Canyon, although Hereford, Burke, and Thompson (1998) suggest that these two large floods
probably did not overtop the ap or sa terraces. Despite the lack of conclusive evidence, we propose
that the T1 terrace was deposited over many hundreds of years during sa and ap time and has only
recently been wind-scoured into a hummocky surface, leaving buried cultural materials exposed on
its modern surface.
With a correlation of terraces between Grand Canyon and Cataract Canyon demonstrated, we
can now compare the two settings for relative amounts of gully erosion. In comparing three similar
geomorphic areas in the two canyons, we found that gully density is higher for two of the areas in
Grand Canyon by 1.5 times (Table 3.3). We also investigated gully depth in the ap/T1 levels
(Table 3.4) and the number of catchments that are river based. Results show that gullies at the
three areas in Grand Canyon are over four times deeper than those in similar areas of Cataract
Canyon and that the number of river-based streams is two times greater in Grand Canyon than in
Cataract Canyon.
Table 3.3. Gully Density of Three Paired Areas, Grand Canyon and Cataract Canyon
Sites Compared

Length of Area
(m)

Number of
Catchments

Average Distance
between Gullies (m)

Gully Density
(m)

Palisades Creek

42

3

14

0.07

Cross Canyon

42

2

21

0.05

Upper Unkar

112

8

14

0.07

Rapid #12

108

4

27

0.04

112

4

28

0.04

83

3

28

0.04

Comparison 1

Comparison 2

Comparison 3
Tanner Canyon
Rapid #4

50

Chapter 3

Table 3.4. Comparison of Degree of Gully Incision, Grand Canyon and Cataract Canyon
Indicators of Gully Incision
Mean gully depth in ap/T1 terrace (m)
Range of gully depths (m)
River-based catchments (%)

Grand Canyon

Cataract Canyon

2.43 (n=15)

0.56 (n=9)

0.50 - 4.0

0.12 - 0.85

53

22

In general, gully degradation of the terrace sequence in Cataract Canyon is relatively minor
compared to that in Grand Canyon. Although the frequency of large monsoon season precipitation
events appears to be slightly less than in Grand Canyon, we postulate that the relative rarity of
gully-forming processes is largely the result of redeposition by large annual floods that continue to
occur in this reach.
Falsification of the Null Hypothesis
Results of the first test show that gully erosion has increased from pre-dam time to post-dam
time in Grand Canyon. Since we were limited in our analysis by the scarcity of photos from predam time (1965), in contrast to the many years of photography spanning post-dam time, in the
second test we used Cataract Canyon as a pre-dam analogue to sites in the Furnace Flats reach of
Grand Canyon. The amount of gullying in Cataract Canyon is far less than that in Grand Canyon.
Furthermore, the extent to which gullies have downcut in sandy terraces is minimal compared to
that in Grand Canyon. Results of these tests indicate failure to prove the null hypothesis. We
therefore pursued alternate hypotheses that could explain why gully erosion has increased from
pre-dam time to post-dam time.
Climate-Variation Hypothesis
H1:

Gully erosion on alluvial terraces associated with archaeology has increased from pre-dam
time to post-dam time due to variation in precipitation patterns.

Test 1: Evaluate previous research on decadal-scale precipitation variations that may be responsible
for increased erosion rates
Hereford and Webb (1992) conducted an examination of variation in warm season rainfall
across the southern Colorado Plateau for the period 1900-1985 in an attempt to relate precipitation
patterns to arroyo cutting and filling episodes. The results of their standardized anomaly index
show that the dam was constructed in the middle of the longest period of below-normal warm
season rainfall in this century, 1942-1980. Except for a dry period from 1933 to 1940, warmseason rainfall was generally higher than normal from 1906 to 1941 and became much higher than
normal again from about 1980 to the present (Hereford and Webb 1992; Hereford et al.1993;

Results and Discussion

51

Webb, Griffiths, and Melis 2000; Webb et al. 2000).
Hereford and Webb (1992) and Webb, Griffiths, and Melis (2000) determined that the driest
warm-season period in this century was from 1973 to 1980. The relatively stable condition of
arroyos from 1965 to 1973, as shown in air photo and repeat ground level photography of Hereford
and others (1993) and Hubbard (2000), correlates with this dry period . From those same analyses,
the wettest warm-season period in this century, from 1980 until the present, correlates with the
nearly complete down-cutting of arroyos in Grand Canyon.
Standardized seasonal precipitation data for the western Grand Canyon area (Webb et al. 2000)
and decadal standardized precipitation data for the larger region (Webb, Griffiths, and Melis 2000)
show very large multi-year winter rainfall periods in the first decade and last two decades of the
twentieth century. These three decades also experienced unusually high summer-season rainfall.
Winter-season rainfall was generally at or below normal during the intervening seven decades.
The evidence from these century-long analyses of winter and summer rainfall suggests that
arroyo cutting in Grand Canyon could have occurred in 1905-1933, 1940-41, and 1980-1995. The
intervening periods, particularly the period from 1942 to 1980, probably did not cause the amount
of gully erosion of terraces witnessed from 1980 to the present (Hereford et al. 1993; Leap et al.
2000). There appears to be a cause and effect relationship between periods of unusually high
precipitation and arroyo cutting during the past two decades. However, if arroyo cutting were to
consistently correlate with any period of high precipitation, then little of the alluvial terraces with
cultural remains would be left in Grand Canyon. That this is not the case points to some other
geomorphic threshold that has been exceeded, such as loss of sand in the river system.
Test 2: Investigate variations in monsoon season rainfall at equivalent time periods before and after
closure of Glen Canyon Dam
High-intensity daily and monthly precipitation data are plotted for weather stations between
Grand Junction, Colorado, and Parker, Arizona (Figure 3.4). The solid line with a perfect slope of
1 divides weather stations that recorded higher values for the pre-dam period from those recording
higher values in the post-dam period. Points that lie on the solid line show no change between the
pre-dam and post-dam periods.
Plot A in Figure 3.4 shows that stations recording more than 50 mm per month are evenly
distributed around the line of equal events and lie fairly close to the line (R2 = 0.90). Station data
grouped by physiographic region show that two regions, Lower Colorado River and Western Grand
Canyon, have recorded more monsoon precipitation in post-dam time, with increases in 50
mm/month events by 1% and 3%, respectively. The other two regions, the K-M Plateaus and Lake
Powell–Canyonlands, have recorded fewer of these events since closure of the dam, with decreases
of 3% and 2%, respectively.

52

Chapter 3

Figure 3.4. Relationships of monsoon precipitation events between equal periods in pre-dam time
and post-dam time: (A) percent of monthly precipitation events exceeding 50 mm, and (B) number
of days exceeding 25 mm.

Results and Discussion

53

Plot B in Figure 3.4 shows that, for the 25 mm/day data, stations are evenly distributed but are
widely scattered about the line of equal events (R2 = 0.30). The large scatter is probably due to the
sporadic and localized nature of these intense precipitation events. Station data grouped by
physiographic region show that two regions have increased in high-intensity events from pre-dam
to post-dam time: the Lower Colorado River region has increased by 46% and the K-M Plateaus
have increased by 7%. Fewer of these events have been recorded in the other two regions since
closure of the dam, by 3% for Western Grand Canyon and 22% for Lake Powell–Canyonlands.
Both plots show almost no difference between the best-fit line and the line of equal events. For
the period of record examined, general change in monsoon activity for the greater Colorado River
vicinity shows little difference between the pre-dam and post-dam time periods examined. Within
the Grand Canyon region, river level station data for Lees Ferry and Phantom Ranch show a
decrease in both indices of monsoon intensity from pre-dam to post-dam time. This is in contrast
to high-elevation station data for Grand Canyon and Fort Valley, which show an increase in both of
these parameters. Eleven of the 13 stations record the same direction of change for both
precipitation indices, suggesting that both indices are valid measures of monsoon season intensity.
Although these plots do not show a regional temporal change in monsoon season precipitation,
they do show an interesting spatial variation. All stations that lie east of a line along the crest of
the Kaibab-Mogollon plateaus show a decrease in post-dam monsoon precipitation, whereas nearly
all stations to the west of this geographic high show an increase in the same period. Further
research is needed to determine the validity of and possible explanations for this trend.
This temporal analysis of monsoon precipitation for the southern Colorado Plateau region is not
inconsistent with the standardized anomaly indices for monsoon precipitation developed by other
workers (Hereford and Webb 1992; Webb, Griffiths, and Melis 2000), which do not show an
obvious change in the fluctuation about the mean. Although both of these analyses show a very
large positive anomaly beginning about 1980, it is balanced out by an equally large negative
anomaly in the 1970s. Standardized anomalies for the pre-dam period fluctuate less around the
mean value than do those for the post-dam period. Nonetheless, it seems likely that the high
monsoon season anomalies in the 1980s and 1990s resulted in uncommonly high gully erosion
rates in Grand Canyon, especially considering the reduced sediment supply and absence of large
annual floods as a result of the emplacement of Glen Canyon Dam.
Base-Level Hypothesis
H2:

Gully erosion on alluvial terraces associated with archaeology has increased from pre-dam
time to post-dam time due to base-level effects.

Test 1: Report on rebuilding of sand bars that buttress pre-dam terraces and infill gullies in Grand
Canyon versus those in Cataract Canyon

54

Chapter 3

Pre-dam floods maintained sand bars at a higher base level than at present. During the 41-year
period prior to closing of the dam (1922 to 1962), annual peak floods of the river ranged from
105,000 cfs to 127,000 cfs (USGS 1999) and inundated the pda terrace at least 10 times (Hereford
et al. 1998). Likewise, the river flooded the level of the umt deposit at least nine times from
sometime between A.D. 1280 and 1470 up until 1883, as determined from the Lava Canyon profile
(Hereford 1996).
The 1884 annual river flood of about 300,000 cfs (USGS 1999) overtopped the lmt terrace
at most locations, with the exception of a few anomalous sites. This flood inundated the ap/sa
terrace at Axehandle Cove (RM 2) in Marble Canyon (O’Connor et al. 1994) and possibly
inundated the lowest ap levels at Palisades Creek (RM 65) in Furnace Flats as well. We observed
flood sand deposited inside of the McCormick cabin, which was built near the turn of the century
(Helen Fairley, personal communication 1999) on ap sand (Hereford 1996). There is also
evidence that the flood of 1884 or one in the mid 1800s may have overtopped the ap level at
Arroyo Grande in western Grand Canyon. A radiocarbon sample from a feature at the surface of
the ap level dates to 170±50 B.P. These pieces of evidence are profoundly significant, in that the
archaeological terraces could have been inundated with flood sand more often than has been
thought in the past.
Research conducted by Burchett, Coder, and Hubbard (1996) indicated that the beachhabitat-building flow (BHBF) of 1996 had an overall positive effect on 18 near-river cultural
resources studied. At three out of four locations studied by Yeatts and others (1996), new sand was
deposited in the mouths of arroyos. The authors of both studies express concern for the longevity
of these fresh deposits, as the volume of sand deposited was not large enough to offset the
pervasive effects of erosion occurring upstream in the catchment and at river level (Yeatts et al.
1998)
Duane Hubbard of the RCMP is currently rephotographing sites shown in historical photos.
A photo taken in the Palisades area in 1923 by Julius Stone (Figure 3.5A) shows the extent to
which sand was deposited near the river. While Hubbard has not yet matched this photo, the
setting is now gravel with two river-based streams and several terrace-based streams that drain past
this level (see Hereford 1998 map). Likewise, a photo taken at 222 Mile in 1923 by E. C. La Rue
(Figure 3.5B) shows a relatively high sand bar, which is equivalent to the pda level. Any streams
that crossed this area were periodically infilled, temporarily raising the local base level.
The photos in the Bill Belknap collection (NAU Special Collections) offer some general
insights about how pre-dam sand provided a higher base level for the mouths of channels in Grand
Canyon. Figure 3.6A is a view downstream taken at the "old Unkar camp" (RM 72.2R) by Bill
Belknap during a 1960 jet boat trip, prior to his landmark up-run of Grand Canyon with Buzz
Belknap (Buzz Belknap, personal communication 1994). This locality was favored as a campsite
by river-runners in the 1970s and early 1980s.
Figure 3.6B is the repeat photo taken in October 1998. Note that most of the white sand is
missing from the area downhill of the orthogonal boulder, leaving a strip of white sand behind and
to the right of the boulder (here mapped as 1983 flood sand). The upper slopes were severely

Results and Discussion

55

Figure 3.5. Historical photos showing elevation of sand bars below archaeological sites: (A)
Julius Stone photo of Palisades Creek area taken in 1909, and (B) E. C. LaRue photo of 222 Mile
taken in 1923.

Results and Discussion

57

Results and Discussion

57

Figure 3.6. Repeat photos showing change at "old Unkar camp": (A) Belknap photo of 1960, and
(B) repeat photo of October 1998.

Blank back of Figure 3.6

Results and Discussion

59

impacted by the summer monsoon of 1983, and the high flows of 1984-1986 caused the scarp
retreat seen today (Regan Dale, Kenton Grua, Andre Potochnik, personal communication 1999).
The removal of sand from beneath the large orthogonal boulder caused it to tilt downward toward
the river. Note also the high degree of tamarisk encroachment in Figure 3.5B. Furthermore, the
reddish sandy slope in the background was once covered with fresh eolian sand that was blown
upslope (Figure 3.6A). Today this slope is deeply incised, and archaeologic features are exposed in
gullies. We interpret the combination of rainfall-induced gullying and scarp retreat as accelerating
erosion of the upper terraces. We speculate that the cut-off of fresh sand supply from the river has
interrupted any renewal of wind-blown sand to upper slopes and terraces. This cut-off is due in
part to the encroachment of tamarisk near the river, blocking entrainment of sand particles to upper
terraces.
Figure 3.7A was taken by Belknap during his 1963 sportyak trip through the Grand Canyon on
flows of 1000 cfs following closure of Glen Canyon Dam. The location is a view downstream of
the large willow tree at Granite Park (RM 208.9L). Note the high embankment of sand on the river
side of the tree.
Figure 3.7B is a repeat photo taken in April 1996 just after the BHBF and during the low flow
of 8000 cfs. While the BHBF rebuilt the height of the sand bar, the amount of sediment deposited
was insignificant compared to what pre-dam floods were capable of depositing. The constant
renewal of fresh sand in pre-dam time maintained the level of terraces near the river, suggesting
that restorative processes of river deposition and eolian redistribution of sand were considerably
more active than today.
Repeat photography of several sand bars in Cataract Canyon provides evidence of annual flood
renewal to lower terraces. Figures 3.8, 3.9, 3.10, and 3.11 show the level of infilling after the
spring runoff of 1999. Figure 3.8 illustrates the recirculation zone at Cross Canyon, where a
cutbank recorded in March 1999 was buttressed with a large amount of sand in August 1999. The
associated gullies (all of which are terrace based) at this site remained stable during the two-year
period of study. Figure 3.9 shows the effects of infilling of a river-based gully at the channel
margin bar of Rapid 12. It appears that the face of the sand bar was partly eroded and redistributed
higher up, infilling and covering the pre-existing gully. It also appears that the bar has built
upward, creating a new, higher level that the channel must now cross. Terrace-based streams in
this area remained stable over the two-year study.
Figures 3.10A and B show the cycle of redeposition from the annual flood of two years at
Teapot Canyon. Photo 3.10C then shows subsequent gully erosion caused by a severe monsoon
season. Presumably these will be infilled and this level will be built upward again with the next
annual flood. Figures 3.11A and B show the cycle of monsoon-season gullying across the sand bar
at Brown Betty Camp and subsequent sand coverage. This series depicts severe gully erosion in
August 1997, with eolian infilling from floods of 1998 and 1999.

60

Chapter 3

Repeat topographic surveys of Rapid 4, Cross Canyon, and Rapid 12 depict the volume change
at each site that occurred after the 1999 spring flood. Survey results allow for quantification of
annual flood impacts on the terrace sequences and small catchments in Cataract Canyon that
otherwise cannot be determined from oblique photos.
To be able to make a confident comparison of pre– and post–annual flood conditions, we
conducted surveys under similar river flow conditions. The average estimated discharge through
Cataract Canyon for the pre-flood survey on March 20-22, 1999, was 9900 cfs; in comparison, the
post-flood survey on August 2-4, 1999, was 12,900 cfs. The stage height difference between preand post-flood survey was less than 0.5 m. This small variation in stage enabled reasonably
accurate comparison of the two topographic surveys. The hydrograph for the peak runoff season
(April-July) in 1999 was very similar to the long-term median for this period. The peak discharge
in 1999 of 47,000 cfs represents average annual runoff, as the long-term median peak discharge
through Cataract Canyon is 43,000 cfs.
Figures 3.12 through 3.20 show the results of the surveyed areas, depicting mapped areas with
volume change and cross-sectional views of change. Table 3.5 summarizes results of the two
survey periods with a standardized index of net change to facilitate comparative results between
areas.
Table 3.5. Volume of Sand Lost or Gained at Three Topographically Surveyed Terrace Catchment
Study Areas in Cataract Canyon
Study Site

Sand Volume
Cut (m3)

Sand Volume
Fill (m3)

Net Sand Volume
Change (m3)

Ratio of Net Change
by Volume/Area (m)

Rapid 4

-187.6

+145.5

-42.1

-0.02

Cross Canyon

-400.9

+2677.6

+2276.7

+0.75

Rapid 12

-811.2

+1.9

-809.3

-0.42

The Rapid 4 catchment area lies adjacent to a large pool above the rapid and is a channel
margin bar with no recirculation zone at moderate flows (Figures 3.12 and 3.13) (see Appendix E
for location of site). This site shows a minor amount of change (0.02 m/m2), which could be
attributed to error in the repeat topographic survey, but we will assume the change is real.
Generally the site remained unchanged except for about a 1-m cut from the sand embankment
(Figure 3.14) from a small area on the downstream end of the surveyed area. Just beyond the
upstream end of the survey area, we observed a new sand deposit 1-2 m deep about 25 m into the
mouth of the adjacent tributary drainage. This deposit serves as a temporary rise in base level for
the tributary drainage. Otherwise the annual flood did little to impact this site, and study
catchments remained stable.
The Cross Canyon study area is a large recirculation zone (Figures 3.15 and 3.16) (see
Appendix E for location of site) that had a 2.5 m vertical cut bank prior to the annual flood. The
eddy gained a very large amount of sand as a result of the annual flood (Table 3.5). The four
profiles show a dramatic change in the configuration of the bar complex (Figure 3.17). The annual
flood filled the entire eddy with a separation bar of sand more than 2 m deep and 2,677 m3 in
volume, eliminating

Results and Discussion

61

Figure 3.7. Repeat photos showing change at Granite Park recirculation zone: (A) Belknap photo
of 1963, and (B) repeat photo of October 1998.

Results

Blank back of Figure 3.7

Results and Discussion

Figure 3.8. Repeat photos showing effects of annual flood at Cross Canyon recirculation zone:
(A) March 1999, and (B) August 1999.

63

Blank back of Figure 3.8

Results and Discussion

Figure 3.9. Repeat photos showing effects of annual flood on Rapid 12 channel margin bar: (A)
March 1999, and (B) August 1999.

65

Blank back of Figure 3.9

Results and Discussion

67

Figure 3.10. Repeat photos (view upstream) of Teapot Canyon area showing cycle of periodic
deposition from annual flood and subsequent gully erosion from monsoon season: (A) September
1998, (B) March 1999, and © August 1999.

Blank back of Figure 3.10

Results and Discussion

Figure 3.11. Repeat photos showing gully infilling at Brown Betty Camp, Cataract Canyon: (A)
August 1997, (B) August 1998, and © August 1999.

69

Blank back of Figure 3.11

Figure 3.12. Cataract Canyon Rapid 4

Blank back of Figure 3.12

Figure 3.13. Cataract Canyon Rapid 4

Results

Blank back of Figure 3.13

Figure 3.14. Cataract Canyon Rapid 4

Blank back of Figure 3.14

Figure 3.15. Cross Canyon

Blank back of Figure 3.15

Figure 3.16. Cross Canyon

Blank back of Figure 3.16

Figure 3.17. Cross Canyon

Blank back of Figure 3.17

Figure 3.18. Cataract Canyon Rapid 12

Blank back of Figure 3.18

Figure 3.19. Cataract Canyon Rapid 12

Blank back of Figure 3.19

Figure 3.20. Cataract Canyon Rapid 12

Blank back of Figure 3.20

Results and Discussion

89

the recirculation zone at low flows. The separation bar is flat-topped with a return current channel
1-2 m deep scoured along its back side near the shore. Two large backwater ponds formed in the
return current channel as the river stage dropped during July. A part of the pre-flood cut bank
retreated farther during the annual flood, resulting in only a minor amount of sand erosion (400 m3)
compared to the large gain in the eddy. Catchments A and B remained stable, and the level to
which they will try to drain is now the newly formed separation bar. This recirculation zone is
somewhat similar to the one that Yeatts (1996, 1998) studied at Palisades Creek. He recorded a
ratio of net change at Palisades of +0.04 m/m2 after the 1996 BHBF of 45,000 cfs, a substantial
difference from the net change at Cross Canyon of +0.75 m/m2. We surmise that the difference is
the amount of sediment available for deposition in Cataract Canyon.
The Rapid 12 catchment area is a channel margin bar (Figures 3.18 and 3.19) (see Appendix E
for site location) below the rapid that lies across from a large mid-channel gravel bar. This site has
no recirculation zone at lower stages, and it could form a reattachment bar at flows above 100,000
cfs, which approximates the flow in 1984. This area lost 809 m3 from its embankment after the
1999 annual flood and did not gain as much new sand as appeared to be the case in the repeat
photos. The entire bar was cut back by 1-2 m (Figure 3.20). A deep gully in catchment D that had
incised through the entire terrace sequence during the 1998 monsoon season was infilled by the
flood, causing it to become terrace based on T4 rather than river based as it had been before the
annual flood. The mouth of the stream gained only 0.6 m3, suggesting that the terrace was beveled
off and the depth of gully cutting was therefore less apparent.
The results of this survey indicate that sand deposition does not necessarily occur at these sites
every year. Since sand deposition is dependent on stage level and flow duration as well as sand
supply, we predict that if the flood hydrograph had been of shorter duration and higher peak, less
scarp retreat would have occurred at the Rapid 12 area and that, in consequence, less of the
available sand supply would have been transported downstream into Lake Powell. Unfortunately,
we can only confidently compare results from Cross Canyon of Cataract Canyon to Palisades
Creek in the Furnace Flats reach of Grand Canyon. Although the value of percent gain (volume
per unit area) is higher for the Cataract Canyon site than for the Grand Canyon site, we need a
larger sample set to confidently cross-compare sites. The survey results do confirm that dynamic
equilibrium of sand bar restoration varies from site to site and from year to year. Moreover,
persistent recirculation zones such as the one at Cross Canyon consistently show deposition with
the annual flood (John Weisheit, Cataract Canyon guide, personal communication 1999).
A further piece of evidence on the process of gully infilling is the presence of cut-and-fill
structures, specifically, alluvial sands that were once dissected by small streams and since then
have been infilled by river sand or eolian sand. These structures are observed in the walls of large
gullies or tributary cuts at several sites in Grand Canyon, including Arroyo Grande, Lower Tanner,
Lava Canyon, and Palisades, for the most part in the umt and ap units. The same type of cut-fill
structures were observed at the tributary cut of Rapid 12 in the T3 level. This suggests quasiequilibrium conditions between catchment erosion and periodic sand deposition. The lack of
infilling of these large gullies in post-dam time confirms that the interruption of large annual

90

Chapter 3

sediment-laden floods by the dam has shifted the balance. Catchment erosion processes are now
dominant, accelerating erosion of cultural materials contained in those terraces. Only small sand
plugs in river-based gullies have been measured to date, but they cannot endure runoff events from
the catchment (Yeatts 1998).
A final piece of anecdotal evidence of gully infilling comes from investigations of the Paria
River floods of 1997-1999 (Richard Hereford, personal communication 1999). The Paria River
overtopped its bank and filled in many small catchments with sand that is 2-3 m deep. Today the
catchments have formed alluvial fans and are backing up with sediment due to their own base-level
change.
Test 2 - Assess catchment and river processes at each study site in Grand Canyon
Specific catchment and river processes were identified (see Table 2.2), and each was given a
score as to whether it was operable at a specific site. The total score from each site indicates
whether it supports or fails to support the base-level hypothesis. Characteristics of individual sites
are listed in Appendix B. Figure 3.21 shows the degree to which all sites support this hypothesis
(data provided in Appendix C).

Figure 3.21. Percent of surveyed sites (n = 119) that support or negate the current base-level
hypothesis. Each catchment evaluated using data in Table 2.2.
Only eight sites (7%) do not support the hypothesis (score=2), and 18 sites (15%) weakly
support the hypothesis (score=3). These sites exhibit pre-dam arroyo-cutting processes or are nondepositional areas, processes that are unrelated to the presence and/or operations of the dam. Most
base-level characteristics (listed in Table 2.2) are present at the remaining 93 sites (78%),
supporting the hypothesis (score > 3). Specifically, they are long-lived recirculation zones with no
pre-dam arroyo cutting and little evidence of pre-dam bank retreat. Scores for all sites are listed in
Appendix B.

Results and Discussion

91

In the post-dam era, streams must cross lower terrace levels that once were bars with higher
elevations. In considering possible sand renewal to sand bars in the post-dam setting, we assessed
the presence of post-dam flood sands at each site compared to the presence of pda sand (Figure
3.22). The 1996 and 1983 flood sands have generally benefited beach building in Grand Canyon
(Hereford et al. 1996; Kaplinski et al. 1999; Yeatts 1998). Figure 3.22 shows the number of sites
containing these deposits and the number with to the pre-dam annual flood deposits (pda). Over
half of all study catchments have either 1983 or 1996 deposits, yet only 44 catchments have both
deposits. In comparison, 80 catchments show evidence of pda deposits or remnants. Thus, the
number of sites with 1983 sand deposits closely approximates the number with pda deposits.
These results show the effectiveness of the 1983 event, which built nearly as many sand bars as
were built by pre-dam floods (pda unit).

Figure 3.22. Number of sites containing 1996, 1983, or pda flood deposits.
Finally, wind is a major mechanism in restoring sand to gullies in Grand Canyon and probably
maintains the level of higher terraces by blowing sand upslope. Figure 3.23 shows the results of

92

Chapter 3

active eolian sand that has directly infilled parts of study catchments. Half of the catchments have
some kind of eolian deposition, and 42% have active eolian deposition within the catchment area.

Figure 3.23. Percent of catchments that exhibit varying degrees of eolian activity. Units found
within catchments are: none = no eolian activity, eo = old eolian dune, and % ea = percent of
catchment with active eolian sand.
These results do not account for wind-deflated areas, nor do they include eolian deposits that lie
below or near the mouths of catchments. Results do suggest that wind is extremely influential in
resupplying sediment to higher terraces. However, periodic sediment renewal by the river is
imperative for this process to operate. A prime example is at the Palisades Creek type area (Plate
3), where deflation of dunes probably has promoted increased rates of gully incision because the
sediment supply by the river has been severely reduced. In 1973 this was a partial gully that was
blocked by the pda level and related dunes. By 1980 this channel had pushed through the dunes
and became river based. Another example of this is seen at the Lower Tanner area, where stream C
(type stream in Plate 6) has breached the sand dunes and is now river based..
Based on evidence presented in these two tests, we conclude that the base-level hypothesis
cannot be rejected. We needed to use as many lines of investigation as possible because the
concept of base-level change with regard to dams and their operations is poorly understood,
although it is a widely known and generally accepted concept with regard to stream development.
Several lines of investigation—using available historical photos and repeat photos, making
comparative assessments of sand bar renewal in Grand Canyon versus and in Cataract Canyon, and
examining processes operating at individual sites in Grand Canyon—aided us in understanding
base-level processes. Therefore, we feel that this hypothesis needs to be refined to include the
importance of sediment-laden flood flows and subsequent eolian action. We therefore propose a
refined hypothesis to be called the "restorative base-level hypothesis."

Results and Discussion

93

THE GEOMORPHIC MODEL
Results of hypothesis testing show that increases in precipitation in the 1980s correlate with a
large amount of gullying in Grand Canyon. These periods are not unusual, as above-normal
precipitation has occurred in several periods throughout pre-dam time. However, the amount of
arroyo cutting and gullying was not evident in 1965 aerial photos or from the few historical photos
uncovered by researchers, indicating that some geomorphic threshold besides climate has been
exceeded. This threshold component points to lack of sediment renewal to the terraces. The
potential for advanced gully erosion has increased because of decreased periodic deposition of fine
sediment on terraces. In this next section, we explore the kinds of geomorphic settings that control
the depositional environment.
Geomorphic Settings
The geomorphic setting exerts a profound influence on the type of sand terrace in each area and
how that area is affected by runoff events. Small streams that drain across a particular geomorphic
setting exhibit a high degree of similarity. Classifying streams into a geomorphic setting allows for
general characterization of a wide variety of small catchments along the river. Some catchments
divide or include two adjacent settings, in which case they are categorized according to the
dominant setting (see Figure 2.3). The type-area maps display many of the details of each setting,
which are discussed in sections that follow. Hydraulic conductivity values are taken from Table
2.7.
Alluvial Fan Setting
In wide sections of the canyon dominated by shale lithologies, alluvial fans are commonly
deposited by flashy runoff from small tributary canyons. A variation of this setting occurs where a
talus slope merges downslope with coalesced alluvial fans. These alluvial fans of shale commonly
coalesce into small alluvial fan complexes (see Figure 2.3), typified in the reach between
Nankoweap Creek and Unkar Creek. Gravel is transported by local flash floods toward the river,
where it is typically interbedded with high river sand deposits. Interbedding of shale with light
brown river sand is the feature for which the striped alluvium (sa) (Figure 1.2) was named
(Hereford 1996).
The type setting at Lower Nankoweap with paired catchments, N and Q, best illustrates the
alluvial fan setting (Plate 2). Catchment N is terrace based on the lmt terrace and has a
discontinuous channel across the ap terrace. Catchment Q is river based. Both catchment areas are
very large and similar in size. They drain from the east flank of Nankoweap Mesa down a very
steep series of cliffs and slopes of Cambrian strata onto a deeply dissected talus slope, where they
are reworked into short, broad alluvial fans of gravel. The alluvial fans are interbedded with the ap
and sa terraces, which are also very broad and well-vegetated. A robust xeric vegetation
community is a result of the eastern exposure and minimal evaporation. Because of the nature of
this setting, many channels cross the alluvial fan but then discontinue. Typically a thalweg will

94

Chapter 3

occupy a shallow swale for 15-30 m and terminate in a small alluvial fan. Farther downslope
another short thalweg segment will then appear. Large nickpoints have formed in the ap terrace
riser below, but no thalweg can be directly traced to it.
The discontinuous nature of gully development on the ap terrace despite large drainage areas is
likely due to alluvial fans creating distributary networks. Runoff in this area easily infiltrates (K =
0.007 cm/s) into the broad upper ap terrace through two mechanisms: interbedded gravel layers
from alluvial fans of Bright Angel Shale and the extensive roots of perennial acacia trees.
Numerous holes and large roots in the ap terrace riser have locally caused piping and subsequent
surface collapse and have initiated nickpoint migration.
Tributary Plain Setting
Large tributary canyons with abundant fine-grained source material in their headwaters tend to
form extensive, low-gradient fans at their mouths (Figure 2.3). Most of these deposits are of
debris-flow origin and are interbedded with hyper-concentrated flood flow and turbulent flow
deposits (Hereford, Burke, and Thompson1998). Typically, the side-canyon stream has incised a
channel into the fan complex, leaving a broad, perched plain rarely overtopped by the primary
stream channel. These low-relief gravel plains have been partly overtopped by large river flood
events, resulting in extensive terrace deposits (Hereford et al. 1996) that were periodically used by
prehistoric cultures living in Grand Canyon (Fairley et al. 1994; Schwartz, Kepp, and Chapman
1981). Local catchments in these settings can originate from the talus-bedrock slopes adjacent to
the canyon mouth or on any part of the broad fan surface. As a result, catchment area sizes vary,
ranging from extremely local "postage stamp" basins to broadly extensive ones. Eolian deposition
is common in large areas of the tributary plain, as at Palisades Creek.
The catchments H and H' at Palisades are designated as type catchments for the tributary plain
(or deltaic fan) setting (Plate 3). Their catchment areas are very large, extending to the top of a
high ridge of Tapeats Sandstone, Dox Formation, and Cardenas Lava. Catchment H is river based,
while H' is terrace based on the pda terrace. Both catchments drain from a playa lake–coppice
dune complex deposited on the ap terrace at the base of the talus slope (Hereford 1996).
Catchment H' is a shallow, incipient gully that has low vulnerability for ponding and overflow
because the playa is now being continuously drained by the deep gully system of Catchment H.
The coppice dunes had apparently once acted to impound surface water in the playa due to the
accumulation of clay and evaporites weathered from the overlying rock formations. Infiltration
tests on the playa show negligible hydraulic conductivity values (.0007 × 10-3 cm/sec). Numerous
tree roots project outward from the scoured flanks of coppice dunes closer to the river, suggesting
wind deflation. We interpret this to mean that wind deflation of the dune complex reduced its
ability to continually block water in the playa, resulting in overflow and gully cutting of Catchment
H. However, with a fresh supply of a large amount of flood sand, this area could be infilled again.
Evidence of this process was recorded by Yeatts (1998). One year after the 1996 BHBF, survey

Results and Discussion

95

results showed a gain in volume of sand in the area just above the high flow, which Yeatts
attributed to the eolian effects of sand redistribution to higher terrace levels.
Talus Slope Setting
Reaches of the river corridor that lie between extensive regions of tributary deltaic fans are
commonly dominated by talus slopes. In these reaches, lobate talus cones create an irregular
gravel embankment against which channel margin sand bars are deposited (Figure 2.3). The
bedrock headwater areas of these catchments are close to the river and have a very high runoff
efficiency because talus cones reach an angle of repose at 33o. The typically steep slope funnels
water onto the upper sand terrace. However, the lithology of the talus can influence runoff and
infiltration.
Gravel size and percent of interstitial space occupied by a fine-grained matrix are two very
important sedimentologic features that influence permeability and runoff from the talus slopes. For
example, the talus slopes in the Hermit Formation in upper Marble Canyon contain interstitial
material derived from the shale that causes relatively low permeability (K < 0.002 cm/sec) and high
runoff efficiency. Conversely, talus slopes composed of harder rocks, such as the Cambrian
limestones at 201 Mile, form sieve deposits with high permeability (K = 0.191 cm/sec) and,
consequently, low runoff efficiency.
The paired catchments A and E at Upper Unkar are designated as type catchments for the talus
slope setting (Plate 4). The catchment areas are small, narrow, steep, and largely devoid of
vegetation. The terrace sequence is narrow and drops steeply from the ap terrace to the river.
Eolian dunes are common and help dissipate runoff, but they show evidence of much wind scour.
The source of sand renewal has been shut off by post-dam operations, and the base of the terrace
sequence appears vulnerable to potential erosion by sustained high river flows.
Debris Lobe Setting
Steep-sloped, lobate boulder deposits are commonly present at the mouths of medium-sized
tributary canyons (Figure 2.3). Most of these deposits were likely deposited in middle to late
Holocene time (Hereford, Burke, and Thompson 1998) and are presently incised by the main
tributary canyon. Typically, small sand terrace remnants are perched on top of the debris lobe
immediately below the talus or bedrock slope. Catchment areas for these deposits tend to be very
small because they commonly originate a short distance upslope, either on the bedrock spur or on
an old debris fan segment. These small terraces lie in a protected position outside the path of both
side-canyon and river floods.
The debris lobe (or fan lobe) type setting is 122 Mile Canyon with its respective catchments B
and C (Plate 5). The catchment areas at 122 Mile are very small, with little to no talus portion, and
are dominated by Bright Angel Shale bedrock. Catchment B drops off the ap terrace and onto the

96

Chapter 3

bedrock floor of 122 Mile Canyon at the pda terrace level. Catchment C drops off the ap riser onto
the 1983 separation bar.
The ap terrace remnant is relatively wide for a debris lobe setting but drops steeply with only a
thin sliver of pda terrace before reaching the extensive 1983 and 1996 sand bars in the large eddy
complex. The wind has blown a large amount of 1983 sand upslope against the ap terrace riser,
helping to infill shallow gullies. Trailing severely exacerbates erosion of these terraces due to their
steep sand slope and lack of protective vegetation.
Dune Field Setting
Wind plays an influential role in reworking sandy alluvial terraces and depositing abundant
new sand in many sections of the river corridor. Locally, eolian sand accumulates as coppice
dunes and is often interbedded with alluvial sand in the higher and older terraces (Figure 2.3).
However, in some areas eolian deposition is the predominant surficial process and forms extensive
dunes. In this case, dunes control runoff because the sand is very well sorted and highly permeable
(K = 0.03 cm/s). Any gullies that do form tend to be quickly refilled with wind-blown sand.
However, these areas rely on a constant supply of fresh sand from the river in order to maintain the
balance of sand resupply to gullies.
The type setting at Lower Tanner with paired catchments, A and C, illustrates the dune field (or
eolian) setting (Plate 6). The catchment areas are large and drain the shale-rich Dox Formation.
Large alluvial fans coalesce at the base of the bedrock and talus slopes and in the mouths of local
drainages. The sa terrace has been extensively reworked into a very large coppice dune complex
that is well-vegetated and acts as a substantial, permeable barrier to runoff.
Alluvial fans of Dox shale have prograded into the interdunal areas of catchment A, thus
locally lowering hydraulic conductivity (K=0.006 cm/s). In reducing the permeability of the eolian
sand, runoff events become more concentrated through the dune complex. The gully of catchment
A has progressed half the distance of the coppice dune complex. The gully of catchment C is river
based and was able to break through the dune complex because of its steeper gradient to the river.
Cut-and-fill structures exposed in the walls of the deep gully clearly indicate a long history of gully
cutting and subsequent infilling by the river and dune sand. Eolian redistribution of river sand
across and up the slope may prove to be as important for reducing rates of erosion as deposition by
the river itself.
The Process-Based Conceptual Model
The process-based conceptual model describes the process-response mechanisms and
relationships that control the initiation and perpetuation of gully erosion. This model is a result of
intensive study at type areas and was used to evaluate the base-level hypothesis and to develop the

Results and Discussion

97

predictive mathematical model at all study sites. Intrinsic processes and antecedent conditions
ultimately drive a channel to cut across terraces to the river. Extrinsic processes also act upon the
catchments to either increase or decrease erosion rates of the terraces.
The Process of Channel Initiation
Channels form from rainfall and can progress across a surface. Two general pathways cause
channel formation across a flat surface of sand: overland flow and subsurface flow. Overland flow
causes channels to form by two mechanisms: alluvial fan progradation, and ponding and overflow.
Subsurface flow, especially if localized preferentially along a particular pathway, will cause
seepage erosion and surface collapse. These processes may cause a nickpoint to migrate upslope
across the terrace, eventually establishing a channel across the terrace (Higgins, Hill, and Lehre
1990). Through these mechanisms, a small ephemeral stream progressively deepens and extends
its channel in the downslope direction across a series of sand terraces (Figure 3.24).
Alluvial fan progradation is a process observed in areas where the source rock in the adjacent
catchment area is mostly shale (Figure 3.24A). Shale is typically very friable and produces an
abundance of small, angular, pebble-sized fragments that are easily transported by flow events
from the upstream talus, debris fan, or bedrock slopes. Successive storms deposit more shale
clasts, causing progradation of the alluvial fan across the terrace (Hereford et al. 1993). This
process establishes a slope across the flat terrace surface, which allows water to flow to the next
lower terrace. Again, once water flows down the steep gradient of the terrace riser, a nickpoint
forms and migrates upslope, incising the alluvial fan that produced it. Runoff then continues to
incise both the alluvial fan and the terrace until it becomes graded to the next lower terrace. An
example of this process can be seen at Lower Tanner Catchment A, where an alluvial fan of Dox
Formation shale has prograded partly across the interdunal area of an extensive eolian complex
(Plate 2).
Ponding and overflow occurs when the rate of water input onto a terrace exceeds the rate of
infiltration into the terrace (Figure 3.24B). This process causes a standing body of water to form,
which, when sufficiently deep, overtops the lowest point in the downstream edge of the terrace.
Water pours through this low point down the relatively steep terrace riser and creates a nickpoint.
As the nickpoint migrates upslope, it continues to funnel more of the ponded water. With
continued runoff, the channel will eventually incise the entire terrace width and become graded to
the next-lower terrace. An example is Palisades Creek, where runoff from the deeply weathered
Cardenas Lava and Dox Shale have caused the deposition of clays on the broad upper ap terrace,
forming an impervious playa lake bed (Hereford 1996). Other smaller scale examples are at Lower
Tanner and Comanche Creek, where small ponded areas have formed during past rainfall events
because a source of shale lies directly uphill.

98

Chapter 3

Figure 3.24. Processes causing stream integration with the river: (A) alluvial fan progradation, (B)
ponding and overflow, and (C) infiltration and piping. In all cases, a headcut forms at a terrace
riser, causing nickpoint migration upslope into the higher terrace.
Infiltration and piping (seepage erosion) is caused when subsurface flow becomes localized
along a particular pathway, such as large roots that extend through a terrace riser (Figure 3.24C).
The water gains sufficient power to cause seepage erosion, surface collapse, and nickpoint
migration across the terrace (Baker et al. 1990; Dunne 1990). Sand terraces behave like sponges to
various degrees depending upon their grain-size distribution and homogeneity. Well-sorted fine to
medium sand of eolian and post-dam fluvial origin has the highest permeability and homogeneity.
These sands quickly absorb large quantities of runoff, reducing the ponding potential on the terrace
surface.

Results and Discussion

99

Piping occurs when groundwater is localized along animal burrows, root cavities, layers of
interbedded gravel, or channels on the underlying bedrock surface. Small sinkholes commonly
form on the surface, with associated nickpoints. As a terrace becomes saturated, groundwater will
flow toward its closest outlet, the terrace riser. As seepage erosion removes sand from the
subsurface conduit, the terrace surface begins to sag. This sag may become an overflow point for a
pond, or it may simply result in collapse. Subsequent surface collapse initiates a nickpoint that will
migrate upslope from a combination of overland flow and subsurface flow. An example of
infiltration and piping occurs at the type area of Nankoweap, catchments N and P (Plate 2).
Other processes of channel initiation occur outside of the catchment. These extrinsic processes
include trailing and tributary overflow. Human and animal trails compact the surface and
channelize water, thereby decreasing diffusivity. Tributary canyon overflow can initiate a channel
where an adjacent side canyon flood event overtops its banks and flows across the terrace
sequence. These events occur infrequently but can cut a large and extensive gully in a short period
of time. Very clear examples are at Espejo A and B, Tanner A, and Lower Tanner D.
In summary, the establishment of an incised channel across any flat terrace is caused by
headward erosion. Headward erosion begins with a nickpoint that forms in the riser of the terrace
tread. The nickpoint may be generated by the leading toe of a prograding alluvial fan, by a low
swale in the terrace tread, or by collapse from seepage erosion. In addition, deflation of eolian
dunes and trailing may contribute to development of the subtle topographic changes that lead to
gully initiation. The absence of large historical river floods, especially on higher terraces, does not
allow these topographic variations to be periodically filled in or leveled out, as they were in predam time. We address these processes specifically for assessment of the base-level hypothesis and
for identifying anomolous sites in the predictive mathematical model.
Diffusion Capacity of the Terrace "Sponge"
A terrace sequence with greater cross sectional area will be able to absorb and diffuse more
runoff than one with lesser area. Terrace cross sectional areas are calculated using depth as
estimated from profile views. Sand thickness is of especially great importance in the more
permeable eolian and post-dam sands because of their ability to rapidly absorb incoming runoff.
The permeability of silty fine sand (pre-dam) reaches a point where the ponding rate supersedes the
infiltration rate, thereby increasing the chance of overflow onto the next lower terrace.
A greater number of terraces generally creates a wider sand area and interrupts the channelforming process due to the process described above; a terrace usually will not be incised until the
terrace above it has formed a channel. Thus, the time required for a terrace-based stream to
become a river-based stream is extended in direct proportion to the total width or number of all the
terraces.
Well-sorted eolian sand exhibits a very high infiltration capacity and can add to the overall
diffusion capacity of the terrace (Figure 3.25A). Likewise, terraces vegetated with annual grasses,

100 Chapter 3
forbs, and cryptobiotic crust are more resistant to channel development (Figure 3.25B). Their fine
root systems and extensive ground cover disperse ponded water, slow the rate of overland flow,
and increase diffusion capacity.
Eolian sand is commonly deposited on terraces and within gullies. Although the wind acts
independently of runoff events in catchments, it tends to influence catchment processes. Wind can
block channels from reaching the river (Plate 6) or deflate eolian dune complexes, leading to
ponding and overflow (Plate 3). The repeat photos in Cataract Canyon show the annual resupply
of sand to a beach with subsequent redistribution by wind into existing gullies (Figure 3.11).
Likewise, the 1996 and 1983 flood sands have been blown upslope in several areas in Grand
Canyon (Hereford et al. 1996; Yeatts 1998; Potochnik and Thompson, personal observations)
toward draining gullies.
River processes may either deposit or erode sand from the terraces near the mouth of a
catchment stream. If the river is eroding or "sapping" an embankment (Dunne 1990), the terrace
sequence is narrowed, allowing an easier avenue for a small stream to reach the river. If the river is
depositing sediment against an embankment, the terrace sequence is widened, reducing the ability
of a small stream to reach the river. Moreover, deposition by the river fills in existing river-based
gullies within its fluctuation zone (Yeatts 1996) and provides a sand source for redistribution by
the wind. The ubiquitous distribution of both 1983 and 1996 post-dam flood deposits at the type
catchments supports this concept (Plates 2-6).
Human mitigation includes efforts by the National Park Service and some tribes to stabilize
existing terraces by using various types of rock rip-rap, wood posts, and brush and may also
include re-vegetation efforts to increase the resisting framework of a terrace sequence. We do not
quantify the effectiveness of these efforts, but rather use them to modify the vulnerability of a
particular terrace sequence.
The Predictive Geomorphic Model
Our mathematical geomorphic model is designed to predict the erosional vulnerability of the
terrace sequence where it meets the rock slopes of the catchment area. The mathematical model
assumes as a starting condition that no channels cross the terraces, that is, that they are in a
"pristine" state. The mathematical model includes only geomorphic factors and processes that are
intrinsic to the catchment system. Extrinsic factors are described qualitatively in terms of their
current or potential impact on the catchment system. Measurements of the present-day degree of
degradation (as of April 1999) are used to test the mathematical model.
The three processes of gully integration—alluvial fan progradation, ponding and overflow, and
infiltration and piping—are applied to the predictive mathematical model as modifying factors to
the

Results and Discussion

101

Figure 3.25. Processes that resist erosion on terraces by increasing diffusivity: (A) increased
diffusion from build-up of eolian sand, and (B) increased diffusion from grasses and organic matter
covering surface.
basic model (Model 1). Model 1 relates potential runoff volume to dissipation capacity of each
terrace. We then applied values for the modifying factors, called Model 2, for each terrace:
(1) Shale factor:

relates the potential for alluvial fan progradation and ponding
on the terrace surface, thereby reducing diffusivity
(2) Woody vegetation factor:
increases the potential for seepage erosion within the terrace,
thereby decreasing diffusivity
(3) Herbaceous vegetation factor:relates the potential for increasing water-holding capacity,
thereby increasing diffusivity on the terrace sequence
(4) Eolian factor:
relates the potential for increased infiltration, thereby
increasing diffusivity
In its simplest form, the model assumes homogeneity within two types of substrate found in the
terrace sequences: well-sorted sand of very high permeability (eolian and post-dam river sand) and
silty fine sand of moderate permeability (pre-dam sand). For simplicity, we assume that the gravel
or bedrock beneath the terraces behaves as an impermeable boundary for the movement of ground
water.

102

Chapter 3
Model 1 Results

Total runoff volume (Q) and peak discharge (Qp) were analyzed separately against present-day
degradation (gully depth:total sand depth) to determine which of these factors more accurately
represents the erosion-driving forces of a storm event. The relation between the two parameters
shows a high correlation (R2 = .89), where Qp is a function of Q (Figure 3.26). We choose to use Q
in subsequent model development for three reasons: (1) Q is a simplifying assumption relating
terrace area to total runoff volume going into that area; (2) peak discharge (Qp) is a function of
total runoff volume (Q)and therefore Q should be used; and (3) runoff volume (Q) is the same unit
of measure as the storage volume of terraces. A plot of Q versus present-day degradation (Figure
3.27) shows the distribution of data. The threshold line is drawn according to the deepest gully
depth found at sites with the lowest vulnerability, and the line extends to the data point showing
maximum gully incision at 100% vulnerability.
In Model 1, the highest terrace of each sequence is assessed for its capacity to hold water (Axt)
as a function of total runoff volume (Q) from the hypothetical storm event. Therefore, terrace
vulnerability (Q vul) is expressed as a ratio of Q to Axt. Figure 3.26 shows the fundamental
structure of the model and the primary elements driving the system: runoff volume, Q, and terrace
cross-sectional area, Axt. It should be noted that several catchments have a negative Q vul value
due to the absence of upper catchment area leading into the terrace sequence. These catchments
are shown in Figure 3.25 as effectively having zero vulnerability to erosion.
Points are clustered within a triangular area in the lower part of the field (Fig. 3.27), with only
eight outliers above the threshold line. This line serves as a response function interpreted to
represent the quasi-equilibrium effect of today’s climate on the small-catchment geomorphic
system. It does not account for any extrinsic factors such as influence of the dam. Points lying
along the threshold line represent uppermost terraces that are close to reaching their maximum
gully incision, probably as a result of these catchments being subjected to a higher number of
episodic rainfall events than other catchments. Points lying along the bottom of the x-axis
represent terraces not yet incised. Points lying farthest to the right have the greatest potential for
erosion; points lying farther to the left have the least potential for erosion and are considered
mostly stable.
If reduction in restorative processes continues, we would expect to see points shift to the right.
Therefore, all points would have shifted to the right in 1963 when most of the river’s sediment
supply and large floods were eliminated by the closure of the dam. The loss of restorative sand
increases the vulnerability (Q vul) for each site by effectively reducing the volume of the terrace
"sponge."
Without restorative beach-habitat-building flows and with continued rates of erosion, we can
expect that all points will steadily creep upward and toward the right. Once they reach the
threshold line, maximum gully depth will be attained in a short period of time. At this point gullies
will progressively widen through lateral erosion of their banks, which will eventually remove the
whole terrace segment. However, if restorative processes are applied, we predict that points will
slow in

Figure 3.26. Relation of two runoff parameters, peak flow (Qp) and total runoff volume (Q), as
applied to the uppermost terrace at each site.

Figure 3.27. Vulnerability of top terrace versus gully depth ratio using Model 1.

104

Chapter 3

their ascent and be held in stasis. With restorative floods they may even shift slightly downward
and to the left, depending on the effectiveness of the restorative processes.
The position of the threshold line in Figure 3.27 is interpreted to be a function of the present
climate. If storm severity and frequency increase, we predict that the threshold line will pivot
upward and allow further deepening of the gullies. Conversely, if storm severity and frequency
decrease, we predict that the threshold line will pivot downward and allow restorative processes
to be more effective.
Model 2 Results
Model 2 is simply a refined version of Model 1 that accounts for intrinsic variations
found within terraces. The presence of any of the four geomorphic factors (shale, woody plants,
grasses and organic matter, and eolian sand), as discussed in Chapter 2, have been added to the
vulnerability of each terrace within each study catchment. Figure 3.28 shows the Model 2
vulnerability rating of the uppermost terrace plotted against gully depth ratio. Adding the
presence of these factors to each catchment results in shifting the points only slightly, thus
refining the fundamental structure of the predictive model. We present the vulnerability of the
uppermost terrace in this figure because that is where most of the cultural features are found in
Grand Canyon. This plot shows that Axehandle Cove (catchment A) represents maximum
vulnerability, with 100% of the terrace depth incised by its gully cutting through that terrace;
Comanche Creek (catchment E) represents the lowest vulnerability, with 15% of its terrace depth
incised by its gully.
Several extrinsic geomorphic characteristics, such as severe human impact/trailing, side
canyon overflow, and the fire-hose effect from waterfalls, interfere with the predicted
vulnerability of several catchments. The outlier points, which lie above the threshold line in
Figure 3.26, reflect such anomalies:
Paria (B) – severe human trailing
Axehandle Cove (A, B) – firehose effect from waterfall
Espejo (A, B) – overflow channels from adjacent side canyon
Granite Park (A, F) - unknown; catchment area could extend beyond measured
area
Arroyo Grande (A) – unknown; catchment area could extend beyond measured
area
We also plotted Cataract Canyon catchments as a comparison to vulnerability in Grand
Canyon (Figure 3.26). Most of the Cataract Canyon sites are predicted to have relatively high
vulnerabilities, yet gullies have incised a maximum of 15% of the terrace depth, the same ratio as
found with vulnerabilities equal to zero. We determined that processes in Cataract Canyon
mimic the effects of the monsoon season in Grand Canyon. Therefore, the low occurrence of
deep gullying is attributed to episodic redeposition of sand on the terrace sequence by annual
floods.

Results

Results

Figure 3.28. Vulnerability of uppermost terrace versus gully depth ratio using Model 2, which
includes geomorphic factors.
To test how effectively the model can distinguish between variations in geomorphic
characteristics from terrace to terrace, and how likely a terrace will be to degrade if it receives
the full impact of Q, we analyzed each individual terrace for its vulnerability to receiving the full
impact of Q from the upper catchment. A frequency distribution of each measured terrace
against a full range of vulnerability ratings (Figure 3.29A) shows a fairly wide and even
distribution of the data. This distribution confirms that the model can detect subtle differences in
geomorphic characteristics such as catchment area, type of substrate, and cross-sectional area of
the terrace.
To assess a site’s average vulnerability, we calculated cumulative vulnerability for each
terrace at each catchment. This calculation takes into account the vulnerability of the terrace
immediately upslope, assuming no degradation has occurred. In this way we can evaluate the
effectiveness of the terrace "sponge" across the whole suite of terraces. These weighted average
ratings are listed for each site in the Site Descriptions (Appendix B). Frequency of all terraces as
calculated for cumulative vulnerability is shown in Figure 3.29B. The strong skew toward the
left in the histogram shows the amount of resistance to erosion afforded by many terraces
because of their protected position below more resistant higher terraces. This skew suggests that

Results
most of the terraces in Grand

Results

Results
Figure 3.29. Frequency distribution showing vulnerability ranges of all measured terraces: (A)
terraces evaluated independently of their position within the suite of terraces, and (B) terraces
evaluated dependent on their position within the suite of terraces.

Results
Canyon are fairly protected from erosion, until the terrace above has been completely incised by
a channel.
When factoring in the amount of present-day degradation to the uppermost terrace at each
catchment (Figure 3.30), we find that gully depths are progressively greater with higher
vulnerability classes. This histogram is simply a test of the model’s predictive capability and its
accuracy in representing terrace vulnerability.

Figure 3.30. Relationship of terrace vulnerability (all terraces) to gully depth ratio.
To detect trends in vulnerability throughout the canyon, values for upper terrace
vulnerability are grouped according to geomorphic reach and geomorphic setting (Figure 3.31).
First we wanted to see if reaches closer to the dam would show a higher average vulnerability
(Figure 3.31A). This is not the case, as all reaches except the Aisles, which has substantially
lower average vulnerability, show similar average vulnerability. Evidence of eolian activity is
abundant at many of the sites in this particular reach, and terrace areas therefore generally tend to
be larger than those in other reaches.
We then grouped average vulnerability by geomorphic setting, which separates the data
more effectively (Figure 3.31B). The alluvial fan setting shows the highest average
vulnerability, because catchment areas are typically highest within this setting. Likewise the
tributary plain setting has large average catchment area dictating the high average vulnerability.
Interestingly, catchment areas are not as strongly related to vulnerability within the talus slope
and debris lobe settings. In these settings vulnerability is more a function of narrow terrace

Results
segments and therefore a relatively small

Results

Results
Figure 3.31. Mean vulnerability of uppermost terraces grouped by (A) geomorphic reach and
(B) geomorphic setting.

Results
terrace "sponge." The dune field setting shows the lowest average vulnerability, the result of
small catchment size combined with high cross-sectional area of the terrace segment.
Integrating a Climatic Parameter into the Model
A further refinement to the model is to integrate any consistent variation in precipitation
throughout the length of the Colorado River corridor. In plotting various analyses of
precipitation data against elevation of weather stations located along the length of the river
corridor, we detect consistently decreasing slopes in Figure 3.32 from A through C. This pattern
suggests that the relationship of mean annual precipitation to elevation is strong but that of
monsoon precipitation to elevation is weaker. Correlation coefficients range from 0.58 to 0.72,
which indicates variable distribution in rainfall patterns progressing from the southwest to the
northeast. Nonetheless, there is a general northeastward increase in annual and monsoon-season
moisture.
When focusing on the Grand Canyon stations containing our study sites, from Lees Ferry
to Diamond Creek, we would expect an incremental increase in moisture to the northeast, with
Lees Ferry receiving the most precipitation. However, Phantom Ranch values are notably higher
than those at other stations (Figure 3.32 A-C), because the orographic effect of the KaibabMogollon Plateaus induces greater annual and monsoonal precipitation at this location. Values
for Lees Ferry plot directly on or near the regression line and are controlled by the rain shadow
effect of the K-M Plateaus. Therefore, Lees Ferry receives less annual and monsoonal
precipitation than Phantom Ranch.
When frequency of rainfall events greater than 25mm/day is plotted, a strong correlation
(R = 0.88) exists between the number of large rainfall events and elevation; however, it is an
inverse relationship. The number of these particular high-intensity storms increases to the
southwest, probably because northerly tracking monsoon moisture in the Lower Colorado River
region is not influenced by the K-M plateaus. Lees Ferry and Phantom Ranch receive a similar
number of events despite their elevation difference, which attests to the influence of the K-M
plateaus and the ephemeral nature of these high-intensity storms.
2

The strong correlation between river elevation and frequency of big monsoon storms
suggests that this relation is the most useful plot of the four presented here for imposing a
climatic gradient for the geomorphic model. Table 3.6 shows how data are extrapolated for each
geomorphic reach using the regression equation in Figure 3.32D. This parameter is not included
in the final vulnerability index but is presented in this table simply to evaluate recurrence
frequency of intense storms. Future researchers can then account for the probability that
Western Grand Canyon will receive more of these types of events.

Figure 3.32. Mean annual precipitations for monsoon season, in millimeters, 1967-1998: (A)
mean annual precipitation; (B) mean annual precipitation for monsoon season.

Figure 3.32. Mean annual precipitations for monsoon season, in millimeters, 1967-1998:
(C) mean precipitation for peak month of monsoon season; (D) number of events >25mm/day.

Table 3.6. Climate Parameter for the Predictive Model

Length of
Reach (River
Miles)

Representative
Catchment
Location in
Reach

Location of
Catchment
Miles from
Lees Ferry

Elevation of
Location (m)

Cataract Canyon

12

Rapid 12

-212

1158

Upper Marble Canyon

31

Soap Creek

12

929

Lower Marble Canyon

31

Willy's Grave

46

859

Furnace Flats

11

Basalt Canyon

69

807

Aisles

18

Owl Eyes

134

594

Western Grand
Canyon

36

194 Mile Canyon

194

472

Reach

Note: Data interpolated from the regional trend shown in Figure 3.33D for reaches
examined in this study.

>25
Even
Reac
of

CHAPTER 4
CONCLUSIONS AND RECOMMENDATIONS
Kate S. Thompson and Andre R. Potochnik
RESULTS AND CONCLUSIONS OF THE CURRENT STUDY
Results of this study indicate that the balance between catchment erosion and flood-sand
deposition in Grand Canyon has been disrupted since emplacement of Glen Canyon Dam. The
geomorphic response to this condition is gully rejuvenation, where down-cutting is initiated and
accelerated. Also, terraces lying below cultural sites are no longer being sustained at their higher
pre-dam level. The reduction of fine sediment supply (sand, silt, and clay) has propagated
headward erosion at terrace risers, providing a link for gullies draining onto higher terraces to be
integrated with the river. The result is increased rates of erosion on the cultural terrace through
channel deepening and widening. In arriving at these conclusions, we investigated two
hypotheses that might account for these changes and developed a mathematical model for
predicting erosion of pre-dam terraces.
Hypotheses
The climate-variation hypothesis states that increasing intensity of rainfall has accelerated
erosion processes. From an examination of previous work it appears that warm-season
precipitation was more intense than average during the decades before1932, was less intense
from 1932 to 1980, and has been more intense again since 1980 (Hereford et al. 1993; Hereford
and Webb 1992). Previous work on winter-season rainfall shows a similar pattern of high
precipitation in the first decade and last two decades of the twentieth century (Webb, Griffiths,
and Melis 2000). Our conclusion from examining weather stations throughout the Colorado
River system is that overall warm-season rainfall intensity did not significantly change in the 32
years before and after emplacement of the dam. Note that this comparison is based simply on the
summation of the two 32-year time spans and does not take into account pulses of wet and dry
periods. We cannot reject the hypothesis that increases in precipitation have caused increased
gullying in the 1980s and 1990s. However, these pulses have also occurred in pre-dam time and
likely were associated with arroyo cutting in that period as well, yet we do not see the evidence
of such large-scale arroyo cutting in the 1965 air photos.
We also considered local variation in monsoon intensity as an explanation for the
different degrees of erosion found in Cataract Canyon and Grand Canyon. The number of days
that monsoon-season precipitation exceeded 25 mm from 1967 to 1998 is slightly greater in
Grand Canyon than in Cataract Canyon, which could account for this difference. However, the
restorative effects of annual floods probably outweigh subtle differences in precipitation patterns.

We tested the current base-level hypothesis as a possible explanation for apparent
accelerated rates of erosion in archaeological river terraces of Grand Canyon. Through several
lines of investigation, we concluded that the base-level hypothesis needed to include the critical
role of eolian process in restoring sand to gullies and terraces. We propose the term "base-level
restoration hypothesis" to incorporate the importance of periodic sand renewal to the bases of
gullies and terraces. Furthermore, erosion of some sites in Grand Canyon has little relation to the
presence and/or operation of Glen Canyon Dam. These sites generally show pre-dam arroyo
cutting and bank retreat with little to no pre-dam annual flood deposits (pda) or 1983 sand
deposits. The five sites of this type in our study are listed in Table 4.1; in the Site Description
section (Appendix B) they have a "base-level hypothesis rating" of less than 3.
Table 4.1. Catchments in Grand Canyon that Do Not Support the Restorative Base-Level
Hypothesis
Site

Catchmen
t

Reasons

Axehandle Cove A, B, C

Little evidence of pda sand; pre-dam arroyo cutting;
pre-dam bank retreat

Little
Nankoweap

A

Scarp retreat caused by Little Nankoweap tributary;
no pda or 1983 flood sands; no eolian sand at mouth

Basalt Canyon

B

Pre-dam arroyo cutting; pre-dam scarp retreat; no
pda sand

Arroyo Grande

A

Pre-dam arroyo cutting; pre-dam scarp retreat; no
pda sand

Granite Park

E, F

Pre-dam scarp retreat from Granite Park Canyon; no
pda or 1983 sand; no eolian sand at mouth

Studies of repeat aerial and oblique photography in Grand Canyon show that before
closure of Glen Canyon Dam few gullies were incised to the degree found in 1999. Only the
largest arroyos, such as those at Axehandle Cove, Tanner Canyon, Lower Unkar, and Arroyo
Grande, were present in pre-dam time. Historical photos taken by Stone and La Rue in the early
1900s and repeat photography using Belknap photos of the early 1960s show that river-side sand
bars were much higher and more extensive than they are today. Also, the slope at the "Old
Unkar Camp," which is currently incised and exposing archaeological materials, appears in a
Belknap photo to be a smooth, unincised slope with a layer of fresh eolian sand, and we found
well-sorted eolian sand interbedded with silty fine sand at this locale. Presumably this slope,
among many others, was periodically blanketed with fresh wind-blown sand derived from annual
flood deposits in pre-dam time. We also believe that in many areas the new tamarisk riparian
zone blocks eolian redistribution of flood sands to higher terrace levels. Furthermore, the dying
off of mesquite in the old high-water zone has likely rejuvenated eolian deflation of

archaeological sites. These hypotheses need to be further tested at each site.
Cataract Canyon served as an analogue for studying the pre-dam processes that
presumably took place in Grand Canyon. The general fluvial setting and flow history of the two
reaches are remarkably similar. Although terraces in Cataract Canyon are not extensive, they
display a sequence that generally correlates to that in Grand Canyon. However, thorough
analysis of the correlation requires some additional work.
In our studies in Cataract Canyon we found relatively little gully incision of the terrace
sequence compared to that in Grand Canyon. Catchment erosion processes and geomorphic
settings in Cataract Canyon are very similar to those in Grand Canyon, yet terraces appear to be
more stable than those in Grand Canyon. We believe the exception is that the ratio of upper
catchment runoff to the capacity of the terrace "sponge" to absorb water is greater in Cataract
Canyon. One would therefore expect to see more arroyos and dissected terrace sequences in
Cataract Canyon, but this is not the case, as shown by our vulnerability assessments. The
highest-risk catchments in Cataract Canyon have vulnerability ratings between 50 and 65, yet
none of the gullies have incised more than 20% of the terrace depth. The same range of
vulnerabilities in Grand Canyon shows gullies incising up to 53% of the terrace depth (after
omitting outlier catchments) with an average incision of 35% of the terrace depth.
Based on our limited two years of repeat photography at Cataract Canyon study sites, we
conclude that: (1) gullies draining broad, sandy upper terraces have been mostly stable and have
not been recently rejuvenated; (2) active gullies as documented in certain study areas are
periodically infilled; (3) sand deposition by the annual flood tends to fill existing gullies; and (4)
lower terraces in large recirculation zones are likely rebuilt every year, but channel margin bars
are rebuilt less often. Therefore, the elevation and/or extent that gullies must cross to reach the
river is increased periodically, perhaps at least every two years. However, this estimate needs
repeat observations over a longer period of time. Finally, several instances of cut-and-fill
structures were observed in arroyos in both Cataract Canyon and Grand Canyon, indicating that a
dynamic equilibrium between erosion and deposition probably existed in pre-dam Grand
Canyon, much like that in Cataract Canyon today.
According to our results, the climate-variation and restorative base-level hypotheses
cannot be rejected. We conclude that erosion of river terraces containing archaeological
resources is due to large precipitation events of the last two decades that coincide with the loss of
the restorative processes of sand-bar rebuilding caused by operations of Glen Canyon Dam.
Predictive Mathematical Model and Mitigation Measures
We determined that a process-based mathematical model is the most effective way to
understand the current geomorphic system and to predict which sites are most vulnerable to
erosion. This type of model avoided the uncertainties that might have stemmed from short-term
monitoring within the period of this two-year study. The model is connected to actual process-

response relations of the geomorphic setting, and it accounts for vulnerability based on crosssectional area of the terrace sequence and the predicted runoff volume from the upper catchment.
In testing the effectiveness of this model, we concluded that catchment area is the main factor
that drives gully-erosion processes and that terrace volume is the main factor that resists this
erosion, through its capacity to diffuse runoff.
Vulnerability ratings were assessed with the assumption that terraces are not incised and
that all are subjected to the same precipitation events. However, we compare the degree of
present-day degradation of terraces with the vulnerability rating, which can be used in several
ways: (1) we test the effectiveness of the model, (2) we use both vulnerability and gully-depth
ratio to prioritize sites for mitigation, and (3) we establish a baseline for future monitoring of
gullies. The plot of this relationship (see Figure 3.28) shows the spread of terrace vulnerability
against gully-depth ratio for each study catchment and should be used as a baseline for studying
rates of gully incision. Future monitors can easily measure changes in gully depth at our preestablished points of measurement, track how points shift on this scatter plot, and evaluate the
rates at which a terrace is degrading or aggrading.
The highest-priority sites are summarized in Table 4.2, along with gully-depth ratios.
High vulnerability ratings and gully-depth ratios greater than 0.5 indicate that a catchment has
already incised most of its terrace depth and that mitigation efforts should therefore focus on data
recovery. Sites with high vulnerability ratings and gully-depth ratios less than 0.5 are good
targets for checkdams and/or diversion bars. Generally, any mitigative action such as placement
of checkdams should be focused at the talus slope/sand terrace interface; such efforts are already
ongoing at 122 Mile, catchment A. Since catchment area is such a dominant component driving
gully erosion, efforts should be concentrated on increasing diffusion capacity of the terrace by
slowing water from the upper catchment, or even diverting water away from the cultural site of
concern. We observed mitigative structures at many sites within Grand Canyon, but in only a
few places are efforts concentrated at the talus/terrace interface. Of further note is the need to
concentrate water in one channel. If diversion structures are used, efforts to channelize water are
imperative to prevent multiple gullies from forming. Individual site vulnerability ratings are
listed with the Site Descriptions in Appendix B; air photos of catchment locations are provided
in Appendix D. We recommend that these appendixes be used in the field to guide monitors and
managers in prioritizing sites for remedial action.
Gully erosion processes and river processes can be generalized for each type of
geomorphic setting. Generally, alluvial fan, tributary plain, and talus slope settings show the
highest vulnerability ratings. The catchment areas in alluvial fan and tributary plain settings are
typically very large, whereas terrace volumes are very small in the talus slope settings despite
typically small catchment size. Dune field settings are the least vulnerable to erosion because
they typically have smaller catchment areas and large areas of sand with high diffusion capacity.
Debris lobe settings and talus slope settings typically show moderate vulnerability to erosion
because upper catchment areas are of relatively small size, and terrace segments tend to be
narrow. We suggest that all archaeological monitoring sites be grouped according to geomorphic
setting to improve predictions about site vulnerability. The predominant geomorphic setting is

listed for each site in Appendix B.
Table 4.2. Ranking of Highest-Risk Sites Based on Vulnerability Rating of Top Terrace
Site

Catchment

Vulnerability of Top
Terrace

Gully-Depth Ratio1

Axehandle Cove2

A

100

1.00

Nankoweap

N

88

0.11

Palisades

B

88

0.35

Nankoweap

P

75

0.32

Axe Handle Cove

C

74

0.03

Palisades2

B'

72

0.57

Comanche
Creek2

H

71

0.55

Comanche Creek

F

70

0.25

Lava/Chuar

A

69

0.07

209 Mile

A

69

0.19

196 Mile

A

67

0.02

209 Mile

B

67

0.42

Lower Tanner

F

66

0.19

Kwagunt

A

66

0.40

Palisades

H'

65

0.08

Nankoweap

K

65

0.23

Nankoweap

V

64

0.06

Soap Creek

C

62

0.03

Soap Creek

A

62

0.37

201 Mile

A

59

0.30

Arroyo Grande2

A

59

1.00

Nankoweap

O

58

0.09

1

Included to identify sites where data recovery is recommended.
Gully at this site has incised over half of the terrace depth; site should be mitigated
through a data recovery program.
2

Restorative efforts should be directed toward replenishing sand to lower terraces. Our
results show that wind is an important mechanism, transporting sand to upper terrace risers and
periodically infilling gullies. Currently dunes are being depleted because of a lack of new source
material. Periodic replenishment of sand by beach-habitat-building flows (BHBFs) is the most
viable and efficient means of sustaining the terrace "sponge," a concept discussed in Chapter 2
and Chapter 3. Presence or absence of the post-dam1983 and 1996 flood sands as well as the
most recent pre-dam terrace (pda) at individual sites is included in the site descriptions
(Appendix B). Presence of these flood deposits offers clues for implementing future BHBFs as
part of Adaptive Management in Grand Canyon. These flows should be greater than 45,000 cfs,
up to a maximum of 120,000 cfs, given adequate sediment storage in the river. We selected this
maximum based on a recurring flood level since the 1921 flood. The flood of 1957 was
estimated at 126,000 cfs, which was the last big flood before closure of Glen Canyon Dam.
Floods of this nature should be planned as close to main-tributary flooding as possible to prevent
mining of the Colorado River channel (Kaplinski et al. 1999). Furthermore, based on our
infiltration results, well-sorted sand of the post-dam era is very effective at diffusing runoff and
behaves much like eolian sand. In spite of the lack of silt and clay, we believe that the 1983 sand
is one of the most effective terraces in resisting erosion because of its high diffusion potential,
more so than pre-dam sand. It is fairly homogeneous and permeable and behaves as a sponge in
the mouths of existing arroyos and at the base of higher pre-dam terraces.
Cataract Canyon should be considered as a potential study area for conceptualizing
BHBF scenarios in Grand Canyon because of the similarities between the two in geomorphic
setting, flow history, and hydrologic features. Through periodic topographic surveys of beaches
in Cataract Canyon, researchers can predict, to some degree, areas of deposition and erosion
from a particular flood, given a known sediment load. This model could perhaps be projected for
Grand Canyon, given its restricted sediment load. Case scenarios of annual floods in Cataract
Canyon provide opportunities for experimentation that may be too risky in Grand Canyon.
Implementation of this recommendation would greatly reduce costs associated with "test flood
flows" in Grand Canyon. BHBFs should be regularly used as a management tool to curtail
accelerated erosion of cultural sites.
Photogrammetrically produced topographic maps for each type of geomorphic setting
provide a possible means of monitoring gullies and obtaining gully erosion rates without
conducting in-field surveys. In checking resolution levels of these maps, we found that most of
the topography is accurately represented by 25-cm contour intervals (field check observations by
Andre Potochnik). However, a few areas show that the digital model probably extrapolates 25cm contours between 1-m contours, thus concealing any subtle topographical changes. The

resolution of these maps needs further improvement before they can be used for monitoring.
Testing for accuracy of this method might require digitally reproducing an area twice from the
same air photo and checking contours to make sure they match exactly. It is imperative to find
out whether changes measured from these maps are real or just an artifact of map inaccuracy.
Archaeological monitoring of sites in sandy deposits should be based on a geomorphic
monitoring program. Such a program would include quantitative monitoring of river terraces
and dunes that could contain cultural sites and features. We propose a modification to the
current River Corridor Monitoring Program (RCMP) to include simple repeat measurements at
each catchment and/or photogrammetric-style repeat mapping to avoid extensive field work and
the resultant impacts to sites. The work conducted thus far on the large sample set of sites
should be expanded to include the remaining sites located in sandy deposits.
RECOMMENDATIONS FOR FUTURE WORK
The most immediate concerns for future work regard the predictive model. To refine its
predictive capability, we suggest the following:
(1)
Improve accuracy of gully depth ratio measurements. Sand depth down to gravel
or bedrock was measured if cross-sectional view was available at a site. Many sand
depths had to be estimated at sites if no cross section could be measured. Technology
such as ground-penetrating radar could be used to improve this parameter of the model.
(2)
Improve accuracy of morphometric measurements. Once GCMRC has obtained
orthophotos for all archaeological sites in the RCMP, then catchment areas and stream
lengths can be accurately calculated for all sites..
(3)
Incorporate upper catchment slope and terrace slope into the predictive model.
Qp would then be recalculated and applied to the upper terrace, and vulnerabilities would
be recalculated accordingly.
(4)
Incorporate work at type catchment areas with S. M. Wiele's work in progress on
modeling of potential flood-deposited sand at areas with archaeological materials. To
fully understand restorative base-level processes, a one-dimensional flow model should
be projected in catchments to predict amount of channel aggradation in gullies.
Other work would involve a thorough test of the hypotheses presented in this paper,
including the following:
(1)
Determine extent of arroyo cutting in pre-dam time by conducting a repeat
photography study using historic photos. This study should also include photos that
further test the restorative base-level hypothesis.

(2)
Climate re-analysis. Precipitation records at related stations throughout the
southern Colorado Plateau need to be analyzed as a time series using daily precipitation
records. Analyses could then account for antecedent soil moisture conditions from
previous daily precipitation. A climatologist should undertake this type of study.
(3)
Determine where historical floods could have overtopped prehistoric terraces.
Preliminary investigation at Palisades Creek and Arroyo Grande show that the ap unit
was overtopped by floods in the 1800s or perhaps by the flood of 1921. It is possible that
many archaeological sites were periodically covered by high-magnitude historical floods
that kept these resources buried and protected from accelerated erosion.
(4)
Study process of eolian deposition or erosion around sites. Develop a thorough
understanding of how eolian processes relate to the reduction of annual flooding and
sediment supply caused by Glen Canyon Dam.

DEVELOPMENT OF A GEOMORPHIC MODEL TO PREDICT
EROSION OF PRE-DAM COLORADO RIVER TERRACES
CONTAINING ARCHAEOLOGICAL RESOURCES

Prepared for

GRAND CANYON MONITORING
AND RESEARCH CENTER

Submitted by

SWCA, INC.
Environmental Consultants

February 18, 2000

DEVELOPMENT OF A GEOMORPHIC MODEL
TO PREDICT EROSION OF PRE-DAM COLORADO RIVER
TERRACES CONTAINING ARCHAEOLOGICAL RESOURCES
Agreement No. 1425-98-FC-40-22610

Kate S. Thompson and
SWCA Cultural Resourc
BOR Cooperative

REFERENCES
American Society of Civil Engineers
1970 Manuals and Reports on Engineering Practice No. 37. ASCE, New York.
Andrews, E. D.
1991 Sediment Transport in the Colorado River Basin. In Colorado River Ecology and
Dam Management, Proceedings of a Symposium, May 24-25, 1990, Santa Fe, New
Mexico, edited by G. R. Marzoff and Committee on Glen Canyon Environmental
Studies, Water Science and Technology Board. National Academy Press,
Washington, D.C.
Baker, V. R., R. C. Kochel, J. E. Laity, and A. D. Howard
1990 Spring Sapping and Valley Network Development. In Groundwater Geomorphology,
edited by C. G. Higgins and D. R. Coates, pp. 235-266. Geological Society of
America Special Paper No. 252. Boulder, Colorado.
Belknap, B., B. Belknap, and L. Belknap Evans
1991 Canyonlands River Guide. Westwater Books, Evergreen, Colorado.
Belknap, B., and L. Belknap Evans
1997 Grand Canyon River Guide. Westwater Books, Evergreen, Colorado.
Belknap Photo Collections
1999 Northern Arizona University, Cline Library Special Collections, Flagstaff.
Bull, W. B.
1991 Conceptual Models for Changing Landscapes: in Geomorphic Responses to Climatic
Change. Oxford University Press, New York.
Burchett, T. W., C. M. Coder, and D. M. Hubbard
1996 Monitoring of Archaeological Sites between Glen Canyon Dam along the Colorado
River in Response to the Experimental Habitat Building Flow of 1996. In Mitigation
and Monitoring of Cultural Resources in Response to the Experimental Habitat
Building Flow in Glen and Grand Canyons, edited by J. R. Balsom and Signa
Larralde, pp. 26-92. Grand Canyon National Park, Arizona, and Bureau of
Reclamation, Upper Colorado Region, Salt Lake City.
Burke, K. J., and R. Hereford
1999 Map Showing Surficial Geology and Geomorphology of the Lower Tanner Area,
Grand Canyon, Arizona. U.S. Geological Survey Open-File Report. USGS,
Flagstaff. In press.
Carothers, S. W., and B. T. Brown
1991 The Colorado River through Grand Canyon: Natural History and Human Change.
University of Arizona Press, Tucson.

122

References

Cooke, R. V., and R. W. Reeves
1976 Arroyos and Environmental Change in the American Southwest. Clarendon Press,
Oxford.
Cooley, M. E., B. N. Aldridge, and R. C. Euler
1977 Effects of the Catastrophic Flood of December, 1966, North Rim Area, Eastern
Grand Canyon, Arizona. U.S. Geological Survey Professional Paper No. 980.
Washington, D.C.
Davis, S. N., and R. J. M. DeWiest
1966 Hydrogeology. John Wiley and Sons, New York.
Dickenson, W. E.
1944 Summary of Records of Surface Waters at Base Stations in Colorado River Basin
1891-1938. USGS Water Supply Paper No. 918. Washington, D.C.
Dunne, T.
1990 Hydrology, Mechanics, and Geomorphic Implications of Erosion by Subsurface
Flow. In Groundwater Geomorphology, edited by C. G. Higgins and D. R. Coates,
pp. 1-28. Geological Society of America Special Paper No. 252. Boulder, Coloardo.
Edwards, K., and R. M. Batson
1990 Experimental Digital Shaded Relief Maps of Southwestern United States.
Geological Survey Map I-1850, 1:2,000,000. USGS, Flagstaff.

U.S.

Euler, R. C., and W. W. Taylor
1966 Additional Archeological Data from Upper Grand Canyon: Nankoweap to Unkar
Revisited. Plateau 39:26-45.
Fairley, H. C., P. W. Bungart, C. M. Coder, J. Huffman, T. L. Samples, and J. R. Balsom
1994 The Grand Canyon River Corridor Survey Project: Archaeological Survey along the
Colorado River between Glen Canyon Dam and Separation Canyon. Cooperative
Agreement No. 9AA-40-07920, Grand Canyon National Park. Prepared in
cooperation with the Bureau of Reclamation, Glen Canyon Environmental Studies,
Flagstaff.
Fetter, C. W., Jr.
1980 Applied Hydrogeology. Bell and Howell, Columbus, Ohio.
Flaxman, E. M
1972 Predicting Sediment Yield in the Western United States. Journal of Hydraulics
Division (American Society of Civil Engineers) HY12:2073-2085.
Fournier, F.
1960 Climat et Erosion. Presses Universitaires de France, Paris.
123

References

Gilbert, G. K.
1879 Geology of the Henry Mountains (Utah). U.S. Geographical and Geological Survey
of the Rocky Mountain Region. U.S. Government Printing Office, Washington, D.C.
Goodman, John D.
1998 Late Historic Artifacts Field Recording and Dating Guide. Prepared for the Spring
Meeting Workshop of the Arizona Archaeological Council. Ms. on file, SWCA, Inc.,
Environmental Consultants, Flagstaff.
Graf, W. L.
1983 The Arroyo Problem: Paleohydrology and Paleohydraulics in the Short Term. In
Background to Paleohydrology, edited by K. S. Gregory, pp. 279-302. John Wiley &
Sons, London.
Graf, W. L., and H. J. Verville
1993 Partial data bank for the Upper Colorado River Basin, a reduced version of BIGRED
file, version 931203. Ms. on file, Dept. of Geography, Arizona State University,
Tempe.
Graf, J. B., R. H. Webb, and R. Hereford
1991 Relation of Sediment Load and Flood-plain Formation to Climatic Variability, Paria
River Drainage Basin, Utah and Arizona. Geological Society of America Bulletin103:
1405-1415.
Griffiths, P. G., R. H. Webb, and T. S. Melis
1996 Initiation of Debris Flows in Tributaries of the Colorado River in Grand Canyon,
Arizona. U.S. Geological Survey Open-File Report 96-491. On file, USGS, Flagstaff.
Guy, H. P.
1970 Fluvial Sediment Concepts: Techniques of Water-Resources Investigations of the
United States Geological Survey. U.S. Geological Survey, Flagstaff.
Hansen, E. M., and F. K. Schwarz
1981 Meteorology of Important Rainstorms in the Colorado River and Great Basin
Drainages. U.S. Department of Commerce, National Oceanic and Atmospheric
Administration, Hydrometeorological Report No. 50. Washington, D.C.
Hereford, R.
1987 The Short Term: Fluvial Processes since 1940 on the Colorado Plateau. In
Geormophic Systems of North America, edited by W. L. Graf, pp. 276-288.
Geological Society of America Centennial Special Volume 2. Boulder, Colorado.

124

References
1989

Variation of Warm-season Rainfall Frequency in The Southern Colorado Plateau and
Its Effect on Runoff and Alluvial-channel Morphology: A Preliminary Analysis. U.S.
Geological Survey Open File Report No. 89-330. USGS, Flagstaff.

1996 Map showing surficial geology and geomorphology of the Palisades Creek
archeologic area, Grand Canyon National Park, Arizona. U.S. Geological Survey
Miscellaneous Investigations Map I-2449. On file, USGS, Flagstaff.
Hereford, R., K. J. Burke, and K. S. Thompson
1998 Quaternary Geology and Geomorphology of the Nankoweap Rapids Area, Marble
Canyon, Arizona. U.S. Geological Survey Geologic Investigations Series Map I2608. USGS, Flagstaff.
2000a Map Showing Quaternary Geology and Geomorphology of the Lees Ferry Area,
Marble Canyon, Arizona. U.S. Geological Survey Miscellaneous Investigation Series
Map I-2663, scale 1:2000, with discussion. USGS, Flagstaff.
2000b Map Showing Quaternary Geology and Geomorphology of the Granite Park Area,
Grand Canyon, Arizona. U.S. Geological Survey Miscellaneous Investigation Series
Map I-2662, scale 1:2000. USGS, Flagstaff.
Hereford, R., H. C. Fairley, K. S. Thompson, and J. R. Balsom
1993 Surficial Geology, Geomorphology, and Erosion ofArcheologic Sites along the
Colorado River, Eastern Grand Canyon, Grand Canyon National Park, Arizona.
U.S. Geological Survey Open-File Report 93-517. USGS, Flagstaff.
Hereford, R., K. S. Thompson, and K. J. Burke
1997 Dating Prehistoric Tributary Debris Fans, Colorado River, Grand Canyon National
Park, Arizona, with Implications for Channel Evolution and River Navigability. U.S.
Geological Survey Open-File Report 97-167. USGS, Flagstaff.
1998

Numerical Ages of Holocene Tributary Debis Fans Inferred from Dissolution Pitting
on Carbonate Boulders in the Grand Canyon, Arizona. Quaternary Research 50:139147.

Hereford, R., K. S. Thompson, K. J. Burke, and H. C. Fairley
1996 Tributary Debris Fans and the Late Holocene Alluvial Chronology of the Colorado
River, Eastern Grand Canyon, Arizona. Geological Society of America Bulletin
108(1):3-19.
Hereford, R., and R. H. Webb
1992 Historic Variation in Warm Season Rainfall on the Colorado Plateau, U.S.A.
Climatic Change, v. 22, pp. 239-256. Kluwer Academic Publishers, The Netherlands.

125

References

Higgins, C. G., B. R. Hill, and A. K. Lehre
1990 Gully Development. In Groundwater Geomorphology, edited by C. G. Higgins and
D. R. Coates, pp. 139-156. Geological Society of America Special Paper No. 252.
Boulder, Colorado.
Howard, A. D., and R. Dolan
1981 Geomorphology of the Colorador River in the Grand Canyon.
89: 259-298.

Journal of Geology

Hubbard, Duane M.
2000 Long-term Photo Replications. The River Corridor Monitoring Project Report No. 72.
Northern Arizona University, Flagstaff, and Grand Canyon National Park.
Huntoon, P. W., G. H. Billingsley, Jr., and W. J. Breed
1982 Geologic Map of Canyonlands National Park and Vicinity, Utah. 1:62,500.
Canyonlands Natural History Association, Moab, Utah.
Jett, S. C.
1968 Grand Canyon Dams, Split-Twig Figurines, and "Hit and Run" Archaeology.
American Antiquity 33:341-351.
Jones, A. T.
1985 A Cross-Section of Grand Canyon Archeology: Excavation at Five Sites along the
Colorado River. Western Archeological and Conservation Center Publications in
Anthropology No. 28. USDI National Park Service, Tucson.
Kaplinski, M. A., J. E. Hazel, Jr., S. S. Beus, L. E. Stevens, and H. B. Mayes
1993 The Effects of Interim Flow Operations from Glen Canyon Dam on Colorado River
Sand Bars in the Grand Canyon, Arizona. Abstract in GSA Cordilleran and Rocky
Mountain Section Annual Meeting Abstracts with Programs. Geological Society of
America 25(5):60.
Kaplinski, M. A., J. E. Hazel, Jr., M. F. Manone, R. A. Parnell, and A. Dale
1999 Final Report. Colorado River Sediment Storage in Grand Canyon. Northern Arizona
University Department of Geology for Grand Canyon Monitoring and Research
Center, USDI Bureau of Reclamation, Flagstaff.
Leap, L. M., N. B. Andrews, J. L. Kunde, C. M. Coder, and D. C. Hubbard
1996 1996 Summary Report: Monitoring of Archaeological Sites along the Colorado River
Corridor in Grand Canyon National Park. Cooperative Work Order 8005-8-002,
Grand Canyon National Park and Northern Arizona University. Prepared for Grand
Canyon National Park, Arizona.

126

References

Leap, L. M., T. W. Burchett, J. L. Kunde, N. B. Andrews, and D. C. Hubbard
1998 1998 Summary Report: Archaeological Site Monitoring and Management along the
Colorado River Corridor below Glen Canyon Dam. Cooperative Agreement 8210-97002, Grand Canyon National Park and Glen Canyon National Recreation Area,
Arizona, and Northern Arizona University, Flagstaff.
Leap, L., T. W. Burchett, J. L. Kunde, N. B. Andrews, and D. C. Hubbard
2000 Grand Canyon Monitoring Project 1992-1999: Synthesis and Annual Report, FY
1999. Grand Canyon National Park and Glen Canyon National Recreation Area,
Arizona, and Northern Arizona University, Flagstaff. In preparation.
Leopold, L. B.
1976 Reversal of Erosion Cycle and Climatic Change. Quaternary Research 6:557-562.
Leopold, L. B., and W. B. Bull
1979 Base Level, Aggradation, and Grade. Proceedings, American Philosophical Society
123: 168-202.
Leopold, L. B., and T. Maddock, Jr.
1953 The Hydraulic Geometry of Stream Channels and Some Physiographic Implications.
U.S. Geological Society Professional Paper No. 252. Washington, D.C.
Lucchitta, I.
1991 Quaternary Geology, Geomorphology, and Erosional Processes, Eastern Grand
Canyon, Arizona. Administrative Report. U.S. Geological Survey, Flagstaff.
Lucchitta, I., G. Curtis, M. E. Davis, and S. W. Davis
1995a Quaternary geological map of the Palisades Creek-Comanche Creek area, eastern
Grand Canyon, Arizona. U.S. Geological Survey Open-File Report 95-832. USGS,
Flagstaff.
Lucchitta, I., G. Curtis, M. E. Davis, S. W. Davis , and T. Turrin
1995b Quaternary Geology of the Granite Park Area, Grand Canyon: DowncuttingAggradation Cycles, Calibration of Soil Stages, and Response of Fluvial System to
Volcanic Activity. U.S. Geological Survey Open-File Report 95, USGS, Flagstaff.
Preliminary.
Lucchitta, I., and L. B. Leopold
1999 Floods and Sandbars in the Grand Canyon. GSA Today 9(4):1-7.
McCuen, R. H.
1982 A Guide to Hydrologic Analysis Using SCS Methods. Prentice-Hall, New York.

127

References

Melis, T. S., W. M. Phillips, R. H. Webb, and D. J. Bills
1996 When the Blue-green Waters Turn Red: Historic Flooding in Havasu Creek, Arizona.
U.S. Geological Survey Water Resources Investigations Report 96-4059. USGS,
Flagstaff.
Melis, T. S., R. H. Webb, P. G. Griffiths, and T. J. Wise
1994 Magnitude and Frequency Data for Historic Debris Flows in Grand Canyon National
Park and Vicinity, Arizona. U.S. Geological Survey Water Resources Investigations
Report No. 94-4214. USGS, Flagstaff.
Neal, Lynn A. (editor)
2000 Cultural Resources Data Synthesis within the Colorado River Corridor, Grand
Canyon National Park and Glen Canyon National Recreation Area, Arizona. SWCA
Cultural Resources Report No. 98-85. Flagstaff.
O'Connor, J. E., L. L. Ely, E. E. Wohl, L. E. Stevens, T. S. Melis, V. S. Kale, and V. R. Baker
1994 A 4500-Year Record of Large Floods on the Colorado River in the Grand Canyon,
Arizona. Journal of Geology 109:1-9.
Parker, G. G., and C. G. Higgins
1990 Piping and Pseudokarst in Drylands. In Groundwater Geomorphology, edited by C.
G. Higgins and D. R. Coates, pp. 1-28. Geological Society of America Special Paper
No. 252. Boulder, Colorado.
Potochnik, A. R., and S. J. Reynolds
1990 Side Canyons of the Colorado River, Grand Canyon. In Grand Canyon Geology,
edited by S. S. Beus and M. Morales, pp. 461-481. Oxford University Press, Oxford.
Pacific Southwest Inter-agency Committee
1968 Factors Affecting Sediment Yield in the Pacific Southwest Area and Selection and
Evaluation of Measures for Reduction of Erosion and Sediment Yield. Report of the
Water Management Subcommittee. PSIAC, Seattle.
Pye, K. and H. Tsoar
1990 Aeolian Sand and Sand Dunes. Unwin Hyman, London.
Ritter, D. F.
1978 Process Geomorphology. Wm. C. Brown, Dubuque.
Rubin, D. M., J. C. Schmidt, and J. N. Moore
1990 Origin, Structure, and Evolution of a Reattachment Bar, Colorado River, Grand
Canyon, Arizona. Journal of Sedimentary Petrology 60(6):982-991.

128

References

Salzer, M. W., V. A. S. McCord, L. E. Stevens, and R. H. Webb
1996 The Dendrochronology of Celtis reticulata in the Grand Canyon: Assessing the
Impact of Regulated River Flow on Tree Growth. In Tree Rings, Environment, and
Humanity, by J. S. Dean, D. M. Meko, and T. W. Swetnam. Radiocarbon 1996:273281.
Schmidt, J. C.
1990 Recirculating Flow and Sedimentation in the Colorado River in Grand Canyon,
Arizona. Journal of Geology 98:709-724.
Schmidt, J. C., and J. B. Graf
1990 Aggradation and Degradation of Alluvial Sand Deposits, 1965 to 1986, Colorado
River, Grand Canyon National Park, Arizona. U.S. Geological Survey Professional
Paper No. 1493 and Open-File Report 87-555. USGS, Flagstaff.
Schumm, S. A.
1977 The Fluvial System. John Wiley and Sons, New York.
1985

Explanation and Extrapolation in Geomorphology: Seven Reasons for Geologic
Uncertainty. Transactions of Japanese Geomorphological Union 6:1-18.

Schwartz, D. W., R. C. Chapman, and J. Kepp
1980 Unkar Delta. Grand Canyon Archeology Series No. 3. School of American Research
Press, Santa Fe.
Schwartz, D. W., J. Kepp, and R. C. Chapman
1981 Archaeology of the Grand Canyon: The Walhalla Plateau. Grand Canyon
Archeology Series No. 2. School of American Research Press, Santa Fe.
Taylor, W. W.
1958 Two Archaeological Studies in Northern Arizona. The Pueblo Ecology Study: Hail
and Farewell and A Brief Survey through the Grand Canyon of the Colorado River.
Museum of Northern Arizona Bulletin No. 30. Flagstaff.
Thompson, K. S., K. J. Burke, and R. Hereford
1996 Topographic Map Showing Drainage Basins Associated with Pre-dam Terraces in the
Granite Park Area, Grand Canyon, Arizona. U.S. Geological Survey Open-File
Report 96-298. USGS, Flagstaff.
Thompson, K. S., K. J. Burke, and A. R. Potochnik
1997 Effects of the Beach-Habitat Building Flow and Subsequent Interim Flows from Glen
Canyon Dam on Grand Canyon Camping Beaches, 1996: A Repeat Photography
Study by Grand Canyon River Guides (Adopt-a-Beach Program). Glen Canyon
Environmental Studies, USDI Bureau of Reclamation, Flagstaff.

129

References

U.S. Bureau of Reclamation, National Park Service, Hopi Tribe, Hualapai Nation, Navajo
Nation, Southern Paiute Consortium, and Zuni Pueblo
1997 Final Draft Historic Preservation Plan for Cultural Resources Affected by Glen
Canyon Dam Operations. June. USDI BOR, Upper Colorado Region, Salt Lake
City, and USDI National Park Service, Glen Canyon National Recreation Area and
Grand Canyon National Park, Arizona.
U.S. Geological Survey
1999 National Water Information Service, West, Website.http://waterdata.usgs.gov/nwis-w
Van Steeter, M. M., and J. Pitlick
1998 Geomorphology and Endangered Fish Habitats of the Upper Colorado River: 1.
Historic Changes in Streamflow, Sediment Load, and Channel Morphology. AGU
Water Resources Research 34(2):287-302.
Webb, R. H.
1985 Late Holocene Flooding on the Escalante River, South-central Utah. Unpublished
Ph.D. dissertation, University of Arizona, Tucson.
1996

Grand Canyon, A Century of Change: Rephotography of the 1889-1890 Stanton
Expedition. University of Arizona Press, Tucson.

Webb, R. H., P. G. Griffiths, and T. M. Melis
2000 Sediment Delivery from Ungaged Tributaries to the Colorado River in Grand
Canyon. Administrative Report for Grand Canyon Monitoring and Research Center,
Flagstaff. In preparation.
Webb, R. H., T. M. Melis, T. S. Wise, and J. G. Elliot
1996 The Great Cataract: Effects of Late Holocene Debris Flows on Lava Falls Rapid,
Grand Canyon National Park and Hualapai Indian Reservation, Arizona. U.S.
Geological Survey Open-File Report No. 96-460. USGS, Flagstaff.
Webb, R. H., P. T. Pringle, and G. R Rink
1989 Debris Flows from Tributaries of the Colorado River, Grand Canyon National Park,
Arizona. U.S. Geological Survey Professional Paper No. 1492 and Open-File Report
87-118. USGS, Flagstaff.
Webb, R.H., T.S. Melis, P.G. Griffiths, J.G. Elliott, T.E. Cerling, R.J. Poreda, T.W. Wise and
J.E. Pizzuto
2000 Lava Falls Rapid in Grand Canyon: effects of Late Holocene debris flows on the
Colorado River. U.S. Geological Society Professional Paper 1591, 90 p.

130

References

Western Regional Climate Center (WRCC)
1999 Website: http://www.wrcc.sage.dri.edu
Western Water Consultants, Inc.
1992 User’s Manual for TriHydro. vol. 2.2. WWC, Sheridan, Wyoming.
Wiele, S. M.
1997 Modeling of Flood-deposited Sand Distributions in a Reach of the Colorado River
below the Little Colorado River, Grand Canyon, Arizona. U.S. Geological Survey
Water Resources Investigations Report No. 97-4168. USGS, Flagstaff.
Wiele, S. M., and E. R. Griffin
1997 Modifications to a One-dimensional Model of Unsteady Flow in the Colorado River
Through the Grand Canyon, Arizona. U.S. Geological Survey Water Resources
Investigations Report No. 97-4046. USGS, Flagstaff.
Yeatts, Michael
1996 High Elevation Sand Deposition and Retention from the 1996 Spike Flow: An
Assessment for Cultural Resources Stabilization. In Mitigation and Monitoring of
Cultural Resources in Response to the Experimental Habitat Building Flow in Glen
and Grand Canyons, edited by J. R. Balsom and Signa Larralde, pp. 123-158. Grand
Canyon National Park, Arizona, and Bureau of Reclamation, Upper Colorado Region,
Salt Lake City.
1998

High Elevation Sand Retention from the 1996 Spike Flow. Grand Canyon Monitoring
and Research Center, USDI Bureau of Reclamation, Flagstaff.

APPENDIX A
DATA SHEETS

APPENDIX B
SITE DESCRIPTIONS AND VULNERABILITY RATINGS

APPENDIX C
SUMMARY OF CATCHMENT DATA USED IN MODEL

APPENDIX D
AIR PHOTOS OF GRAND CANYON CATCHMENT AREAS

APPENDIX E
AIR PHOTOS OF CATARACT CANYON CATCHMENT AREAS