EXTREME FLOODING in the DOLORES RIVER BASIN Intro Final

Home , Dolores River

Extreme flooding in the Dolores River Basin, Colorado and Utah: insights from paleofloods, geochronology and hydroclimatic analysis

Item Type text; Electronic Dissertation

Authors Cline, Michael Logan

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 10/10/2021 03:39:40

Link to Item http://hdl.handle.net/10150/195522

EXTREME FLOODING IN THE DOLORES RIVER BASIN, COLORADO AND UTAH: INSIGHTS FROM PALEOFLOODS, GEOCHRONOLOGY AND HYDROCLIMATIC ANALYSIS

Michael Logan Cline

______

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF GEOGRAPHY & REGIONAL DEVELOPMENT

In Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY

WITH A MAJOR IN GEOGRAPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2010

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Dissertation Committee, we certify that we have read the dissertation prepared by Michael Logan Cline entitled Extreme flooding in the Dolores River Basin, Colorado and Utah: insights from paleofloods, geochronology and hydroclimatic analysis and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy

______Date: 8/12/2010 Victor R. Baker

______Date: 8/12/2010 Connie Woodhouse

______Date: 8/12/2010 Stephen Yool

______Date: 8/12/2010 Katie Hirschboeck

Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

______Date: 8/12/2010 Dissertation Director: Vic Baker

______Date: 8/12/2010 Dissertation Director: Connie Woodhouse

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: Michael Logan Cline

ACKNOWLEDGMENTS

The single most important individual in my academic career has been Dr. Vic Baker, who inspired me to not only achieve the coveted ‘Ph.D.’ in my title, but to ponder truly scientific thoughts. Dr. Baker continually encouraged me to “listen to what nature is saying”, and when I finally figured out what he meant, I was able to ask better questions and provide better answers than I had previously considered. Dr. Baker provided a great deal of intellectual support, which translated into emotional support when I began to stumble through my research. I also wanted to acknowledge Pauline Baker for her kind words of support near the completion of this Ph.D.

Dr Connie Woodhouse served as my Co-advisor with Dr. Baker. I hadn’t initially intended to delve into paleoclimate science, although her academic support made this possible. Her very insightful and critical reading of these manuscripts vastly improved the quality of the science. Dr. Katie Hirschboeck provide very useful comments, particularly nearing my defense date, and these comments all led to significant improvements of the manuscripts. She also clarified numerous key concepts for me, which has, and will continue to improve my writing. Dr. Steve Yool has always made himself available throughout my dissertation research and during coursework as well. His jubilant demeanor was always a pleasant reminder that academia doesn’t need to feel like a dark, lonely place.

Finally, I want to thank my family and friends for their unwavering support. The love of my life, Amanda, was 110% supportive of my pursuing of a Ph.D., even at the worst of times, and never placed undue pressures on me. Without her support, I would have had no chance. The other love of my life—my son, Noah—arrived during the second year of my Ph.D. He might not understand this now, but he made my completion of this degree possible—talk about unwavering support! My Mother & Father, along with their spouses, provided a phenomenal amount of love and support. Their understanding of the stresses that accompany the Ph.D. process made my life much easier to cope with. Of course I can’t leave the rest of my family out. My sisters and brother in-laws made for many enjoyable Sunday evenings in Tucson. Both were always supportive and understanding of the stresses associated with this endeavor. Unfortunately for them, I am even more of a ‘know-it-all’ than I was before. My In-Laws were extremely supportive as well. Their presence in my life with Amanda and Noah has made so many things possible. Of course, I want to also thank my numerous field assistants: Steve Delong, James Tamerius, Derek Eysenbach and Jeff Garmany. Their unpaid help made much of the data collection possible, and the great conversations kept my morale up throughout this difficult process. Gosh, as I write this, I am realizing that the ‘perfect storm’ of people, and the support that came with them has made it possible for me to earn my Ph.D. Thanks everyone!

TABLE OF CONTENTS

LIST OF FIGURES...... 6

ABSTRACT...... 7

CHAPTER 1: INTRODUCTION ...... 9 1.1 Explanation of the problem ...... 9 1.2 Background ...... 10 1.2.1 Paleoclimate records ...... 12 1.2.2 Paleoflood hydrology ...... 13 1.2.3 Radiocarbon and Optically Stimulated Luminescence ...... 15 1.2.4 Physiographic setting of the Dolores River Basin ...... 16 1.3 Approach ...... 18 1.4 Future research directions...... 20 1.5 Organization of the dissertation ...... 21

CHAPTER 2: PRESENT STUDY...... 26 2.1 Description of Appendix A...... 26 2.2 Description of Appendix B...... 27 2.3 Description of Appendix C...... 28 2.4 Important results and conclusions from this dissertation...... 29

WORKS CITED...... 34

APPENDIX A: HYDROCLIMATOLOGY IN THE DOLORES RIVER BASIN IN THE SOUTHERN ROCKY MOUNTAINS, USA: HEMISPHERIC AND SYNOPTIC CONTROLS ON NONSTATIONARY PRECIPITATION AND STREAMFLOW REGIMES...... 39

APPENDIX B: HOLOCENE PALEOFLOOD CHRONOLOGY BASED ON OPTICALLY STIMULATED LUMINESCENCE AND AMS 14C FOR THE DOLORES WATERSHED, CO AND UT, USA...... 99

APPENDIX C: LUMINESCENCE DATING OF HOLOCENE PALEOFLOOD DEPOSITS: DOLORES WATERSHED, CO AND UT...... 175 6

LIST OF FIGURES

FIGURE 1.1: Location map of the Dolores River Basin relative to the major ENSO- impacted regions of the western U.S...... 24

FIGURE 1.2: Detailed digital elevation model of the Dolores River Basin with study sites and USGS stream gage locations...... 25

ABSTRACT

The complex hydroclimatic response of the Upper Colorado River Basin (UCRB) to climate circulation patterns and their descriptive indices creates significant challenges to water managers, especially given the uncertain future of the climate. This dissertation addresses fundamental questions that surround extreme flooding in the UCRB by combining paleoflood field techniques, two analytical geochronology techniques and several numerical climate data analysis techniques. The three manuscripts included in this dissertation focus on the Dolores River Basin (DRB), a sub-basin within the UCRB in order to answer theoretical questions about the timing and climate patterns associated with extreme floods.

It has become widely accepted that extreme flooding in the Lower Colorado River Basin

(LCRB) is linked to a period when the frequency and intensity of El Nino periods was higher. Within the UCRB, and more specifically, the DRB, the linkages are less clear.

The paleoflood chronology that we developed indicates that the peak episodes of flooding in the DRB occurred between roughly 300 A.D. and 1200 A.D. This period of flooding is out of phase with many floods in the LCRB, whose peak floods dominantly clustered in the last 700 years; a period of time coincident with the termination of large floods in the DRB. The chronology that I developed utilizes accelerator mass spectrometry radiocarbon (AMS 14C) and optically stimulated luminescence (OSL) to provide a detailed flood history, highlighting the importance of utilizing independent age control.

Alternative, or less accurate chronologies would have resulted had we used AMS 14C or

OSL alone, suggesting that some previously studied basins may have incorrect chronologies. The detailed flood chronology of this study has subsequently allowed us to contextualize extreme floods relative to middle—late Holocene climate variability.

In an effort to provide a hydroclimatic context for flooding in the DRB, numerical analyses were applied to contemporary climate and streamflow data in order to identify the possible mechanisms that modulate precipitation and streamflow in the Western U.S. and more specifically, the DRB. Results from these techniques indicate that the DRB maintains a complex response to a major North Pacific, low-frequency circulation pattern. The North Pacific circulation modulates the low-frequency component of the

DRB’s precipitation and flooding, although the high frequency modulation remains very poorly characterized.

CHAPTER 1: INTRODUCTION

1.1 Explanation of the problem

Flooding continues to be a dominant factor in destroying human lives and infrastructure.

The impacts imposed by floods are largely a function of improper planning and mitigation measures, suggesting that many of the damages could be avoided, given the proper identification of the real flood hazards. Even as we increase the financial burden for flood mitigation, the U.S. continues to experience an increase in flood damages

(Pielke and Downton, 2000; Pielke et al., 2002; Baker et al., 2002). Much of this problem is attributable to our lack of understanding of the flood magnitude and frequency relationship. Our present knowledge is based upon the assumption that the flood magnitude-frequency relationship is time-invariant. We now know that this assumption is flawed; extreme floods tend to cluster in time (and space), and the flood magnitude- frequency distributions of a given river basin is variable in time and space, dependent on the global, synoptic, regional and local-scale circulation patterns (e.g. Hirschboeck,

1988). Unfortunately the temporal limitations of our gaged and historical records do not permit a complete understanding of the hazards associated with extreme floods, because extreme events are rare in nature, and are therefore infrequently captured by gaged records (Klemes, 1994; Baker, 2002). This is especially true in the arid, western U.S. where systematic climate records rarely extend a single century. Within the Lower

Colorado River Basin (LCRB), paleoflood records have been used to extend the flood record 5000 years B.P., and in doing so, it has provided a hydroclimatic context that

allows us to understand the general flood frequency-magnitude and its variability, associated with shifts in climate (Ely et al., 1993; Ely, 1997). On the contrary, the Upper

Colorado River Basin (UCRB) responds very differently to the same shifts in climate because the large-scale mechanisms responsible for anomalously high streamflow and precipitation are different (e.g. Cayan et al., 1999). Unfortunately, no comprehensive paleoflood studies have been performed in the UCRB, leaving the Holocene flood history completely unknown.

1.2 Background

Because of its significance as a critical water resource, the Colorado River Basin (CRB) has been the focus of a great number of studies that have worked to characterize the

CRB’s hydroclimate. Due to the CRB’s size and spatial hydroclimatic variability, the basin has often been split into two regions for analysis: The Upper Colorado River Basin

(UCRB), and Lower Colorado River Basin (LCRB). The hydroclimatology and flood- hydroclimatology of the LCRB has been rather well characterized in terms of its associated teleconnections and their respective indices (Woodhouse, 1997; Cayan et al.,

1999; Ely et al., 1997; Ely, 1997; Hirschboek, 1988; House and Hirschboeck, 1997). One of the key characteristics of streamflow and precipitation in the Western U.S. is the

North-South dipole, where centers of action lay in the Pacific Northwest and the

Southeast. The dipole is largely a result of the variability in the El Nino Southern

Oscillation (ENSO) (e.g. Cayan et al., 1999, Redmond and Koch, 1991; Brown and

Comrie, 2004). The southern portion of the ENSO dipole encompasses the LCRB, and

because of this, the LCRB is a relatively well-characterized basin over interannual timescales. Other major teleconnections operate over longer timescales, including the Pacific

Decadal Oscillation (PDO) (Brown and Comrie, 2004; McCabe et al., 2004). Other associations have been described as well, such as the Pacific North American (PNA) index (e.g. Woodhouse et al., 1997). North of the southern portion of the ENSO dipole, the UCRB has led numerous researchers to describe the inherent complexity of the

UCRB’s hydroclimate—the hydroclimatology displays more of a mixed-response to the often-described, circulation indices such as the PDO, ENSO and the Atlantic

Multidecadal Oscillation (AMO) (e.g. Wise, 2010; Hidalgo and Dracup, 2003; McCabe et al., 2007; Canon et al, 2007). The Dolores River Basin (DRB), the focus of this study, lies just north of the ENSO impacted zone (Figure 1A and 1B). The region between the precipitation dipoles—known as the transition zone—has been shown to be spatially transient (e.g. Wise, 2010), suggesting that the DRB may exhibit a mixed-response to major circulation indices.

The complex response of the DRB’s hydroclimate to circulation indices makes analysis of its hydroclimate difficult. To understand the natural variability is a critical concern, particularly because of human-caused climate change. There is considerable agreement amongst climate models that hydrologic changes will be paramount in semi-arid regions of the world (Seager, 2007). Much less is known however, about the characteristics of extreme flooding and how they relate to climate change because of the limited temporal range of the gaged and historical record. There is a general hypothesis, echoed by the

IPCC (2007) that suggests flooding will increase in the western U.S. due to global

warming (e.g. Trenberth et al., 2003; Milly et al., 2002). This hypothesis generally relies on the results derived from global circulation models that are generally considered inadequate for predicting extreme hydrologic responses to climate change (Knox et al.,

2000; Trenberth et al., 2003).

1.2.1 Paleoclimate records

High-resolution Paleoclimate records, derived from tree rings have provided critical information to improve watershed management. The records have greatly improved our understanding of the hydroclimatic variability within the CRB at an annual scale.

Dendrochronology has famously exposed the significant degree of the Colorado River’s over-allocation, (Stockton and Jacoby, 1976). This finding has been further emphasized with an updated reconstruction of the CRB to A.D. 762, which indicates the presence of persistent, extreme droughts in the past (Meko et al., 2007), and other studies have clearly demonstrated the widespread nature of the drought footprint (Cook et al., 2004). Despite the powerful application of dendrochronology as an excellent proxy for average annual streamflow, the method is not suitable for identifying extreme floods in the paleorecord because above-average annual streamflow is not a prerequisite for extreme flooding, and in some regions, the probability for extreme flooding is similar during periods of drought

(Hirschboeck, 2000). For example, Webb and Betancourt (1992) noted that early-mid

20th century flooding in the southwestern U.S. maintained a similar mean, although there were significant changes in the variance of streamflow due to changes in hemispheric and synoptic circulation patterns. The weak correlation between wet conditions, and episodes

of increased flooding highlights the need to explicitly incorporate paleoflood hydrology as a tool for water-resource policy and risk mitigation. It should be noted, however, that dendrochronology has been used to reconstruct the timing and magnitude of large floods.

This is made possible when the floods in question occurred during the lifetime of the living tree and the tree was either partially inundated or directly impacted by flood-waters

(e.g. Yanosky and Jarret, 2002; Hupp, 1982; Sigafoos, 1961). To overcome the temporal limitations presented by the systematic streamflow record, and the inherent limitations on the use of tree rings for flood reconstructions, the field of paleoflood hydrology (Kochel and Baker, 1982) was developed.

1.2.2 Paleoflood hydrology

The science of paleoflood hydrology is employed to extend gaged and historical records to provide a context for extreme floods and their associations with climate variability, and to improves understandings of flood frequency and magnitude (Stedinger and Cohn,

1986; Hosking and Wallace, 1986). Paleoflood hydrology is the study of floods that have occurred without any human record (Kochel and Baker, 1982). It makes use of the geologic record where the deposition of flood debris has been preserved on the landscape.

These flood debris are actual indexical evidence of the stage the floods attained, and are not proxy records (Baker 1988a,b; Baker, 2006). The flood debris are referred to as paleostage indicators, because their landscape position informs us of the minimum paleoflood stage, which ultimately allows for potentially accurate paleoflood discharge estimations. One of the most commonly used paleostage indicators are slackwater

deposits (SWDs). SWDs are sedimentary deposits that form where flow velocities reduce below the critical level to maintain sediments in suspension (Kochel and Baker,

1982). SWDs often contain organic material or quartz sands, as they do in this study, which we can date analytically using either radiocarbon or luminescence dating.

Paleoflood hydrology has provided a great utility in understanding the timing of large floods, and their association with shifts in the climate. These studies have greatly improved our understanding of hydroclimatic variability (e.g. Webb and Betancourt,

1992, Hereford, 2002; Ely, 1997; Ely et al., 1993; Hirschboeck, 1988; Knox, 1993; Knox,

2000; House and Hirschboeck, 1997; Ely et al., 1994; Redmond et al., 2002). Paleoflood hydrology has also helped to constrain our understanding of the true flood hazards, by extending the record in time, which captures a broader window of flood magnitude realizations. Without the broader window of realizations, understandings of flood hazard are limited to theoretical flood magnitude-frequency curves derived from relatively short, gaged historical records (Baker, 2002). They often neglect the “upper-tails” of the distributions (Hirschboeck, 2003), which are critical, especially given the increasingly nonstationary world we are entering (Milly et al., 2008). Flood frequency analyses have improved since paleofloods were included in the analyses (e.g. Benito and Thorndycraft,

2005; Blainey and Webb, 2002) because the addition of paleodata downplays the nonstationary assumption that is fundamental to flood-risk analysis. This is a particular important in locations such as the western U.S., where gaged and historical records rarely exceed one hundred-years, and consequently, large floods are statistically under- represented (Benito and Thorndycraft, 2005). A very consistent finding in most

paleoflood studies is the presence of floods in the paleorecord, whose magnitudes greatly exceed those of the systematic records (e.g. Enzell et al., 1993), illustrating the need to incorporate paleoflood records into water infrastructure planning.

A large number of paleoflood and Holocene stratigraphic studies have been used to develop a hydroclimatic context for large and extreme floods (Ely et al., 1993). Some of these studies have provided evidence for the hypothesis that anthropogenic climate change may increase the magnitude and frequency of large floods (i.e. Ely, 1997;

Hereford, 2002). Unfortunately, these studies have focused on the LCRB, an area that correlates strongly with the modern ENSO signal (Kim et al., 2006; Cayan et al., 1999), leaving a critical spatial gap in our understanding of hydroclimatic regions that correlates weakly with ENSO.

1.2.3 Radiocarbon and Optically Stimulated Luminescence

Paleoflood Hydrology has historically been dependent upon analytical ages derived from organic materials that can be dated radiometrically. Radiocarbon is an excellent tool, although the ages are often derived from detrital material, which may have substantial landscape residence times associated with them (Gavin et al., 2003). Furthermore, radiocarbon relies on the calibration of radiocarbon to calendar years, which may result in significant errors (e.g. Guilderson et al., 2005). More recently, optically stimulated luminescence (OSL) has allowed for the direct dating of the sediments age of deposition, although its application has been poorly tested in paleofloods studies, and the results have often led to problematic dates (e.g. Porat et al., 2001). For this reason, a component of

this research is focused on assessing the application of OSL to paleoflood studies.

1.2.4 Physiographic setting of the Dolores River Basin

The DRB (Figure 2) drains 11,860 km2 in Southwestern Colorado and southeastern Utah.

The basin drains the southern San Juan Mountains and the Uncompahgre Plateau, of CO, and the La Sal Mountains, UT. The Dolores River reaches the Colorado River roughly

26 km north of Moab, UT, where it empties at a mean rate of ~20 m3s-1: roughly ten- percent of the Colorado River’s annual discharge at the rivers confluence. The flow is evenly split between the two rivers that comprise the watershed: the San Miguel and

Dolores rivers. Both of the rivers head near Lizard Head Pass in the San Juan Mountains near Telluride, CO in predominantly Cenozoic volcanic and sedimentary rocks in the upper reaches, and Mesozoic sedimentary rocks in the lower reaches. The San Miguel

River is one of the few significant rivers in the western, U.S. whose main-stem is not impounded, whereas McPhee Dam impounds the upper Dolores (construction from

1980—1985). In the lower reaches of the two rivers, and below their confluence, the rivers are superposed into Mesozoic sedimentary rocks where they meander though steep-sided, and very stable bedrock canyons.

The DRB is a high-relief river basin, ranging from 1200m at its lowest elevation, to over

4000 m in the high San Juan Mountains. The hydroclimatic variability therefore controlled by physiographic, oceanic and atmospheric factors, each operating at different spatial and temporal scales (Mock, 1996). The majority of the basin is semi-arid. Most

of the winter precipitation in the DRB is attributed to Pacific air masses (Bryson and

Hare, 1974) that enter the U.S. from its southern Pacific coast, although occasional northern storms may also reach the DRB (McGinnis, 2000). The synoptic mechanisms responsible for delivering the moisture are generally mid-latitude cyclones (McGinnis,

2000). These are relatively common features during the winter months, and depending on their source region, they bring the potential for significant precipitation. The synoptic lifting associated with the low-pressure systems are enhanced orographically, and can result in heavy precipitation for multiple days at higher elevations, especially if the air mass becomes cutoff or stalled (Mock, 1996). This makes topography the dominant factor responsible for the spatial variability in winter precipitation (Weare and Hoeschele,

1983; Cayan and Rhodes, 1984; Sheppard et al., 2002). For example: snowfall differences between low and high elevations may differ up to an order of magnitude, but ultimately, it is the storm tracking and storm source that determines the precipitation capacity of a storm. Storms that enter the southern San Juans from the south tend to produce greater amounts of precipitation than any other storms. These systems not only carry saturated air mass from the Pacific, but they also draw moisture from warm SSTs in the Gulf of California (Keen, 1996). These southern storm tracks from the warmer ocean

SSTs in the tropical and extratropical Pacific permit greater atmospheric moisture flux, bringing deeply saturated, warmer air to the region which results in even greater unsettled air masses. Summer convective storms draw their moisture from the Pacific and Gulf of

California, and to a lesser degree the Gulf of Mexico (Adams and Comrie, 1997). There is potential for dissipating tropical cyclones to extend far enough inward to deliver

significant precipitation to the DRB, although it is rare because of the DRB’s distance from major water bodies. It is worth noting, however, the largest flood on the Dolores

River was the result of a dissipating tropical cyclone in October of 1911 (Hansen and

Schwarz, 1981). During that event, the largest historical flood occurred within the San

Juan River Basin—the DRB’s neighboring basin. Average January temperatures in

Telluride, CO (1900-2008) range from a minimum of -15° C and a maximum of 2° C, and the average summertime high is 24° C. Average precipitation in Telluride is 590 mm yr-1 and snowfall averages 425 mm yr-1. In Gateway CO, one of the lowest regions of the

Basin, the maximum and minimum January temperatures are 16° C and -8° C respectively, with summertime highs of 34° C. The mean precipitation in Gateway, CO is 280 mm yr-1 and the mean snowfall is 410 mm yr-1.

1.3 Approach

Extreme floods have been shown to regularly cluster in time and space, as a result of the prevailing circulation patterns operating across multiple scales (e.g. Ely et al., 1993;

Hirschboeck, 1988). The clustering is related to the prevailing circulation patterns influence the dominant hydroclimatic regime, and these regimes have been shown to be highly sensitive to subtle shifts in the climate (e.g. Knox, 1993). A major limitation in developing an understanding the distribution of flood frequency and magnitude is the significant temporal limitation of our flood record. Paleoflood hydrology can be an excellent utility that extends the timescale of the flood record, well beyond the gaged and historical records, thus allowing researchers to identify the periods in which extreme

floods were more or less frequent. When these are compared with paleoclimate records from the hydroclimatic region, conclusions can be drawn about the prevailing climatological conditions, and in some cases, the circulation patterns that prevailed during the times of heightened flooding (e.g. Ely et al., 1993, Ely, 1997). To better understand the flood hydrology in the DRB, I have: 1) provided a hydroclimatic context by combining a suite of hydroclimatic analyses on the instrumental and paleohydrological record; 2) developed a paleoflood record that extends into the early- mid Holocene; and 3) tested OSL against AMS 14C, in an effort to test the OSL application to paleoflood hydrology and make the flood ages more robust.

The ultimate goal of this project was to begin developing a paleoflood chronology for the greater UCRB. The motivation for selecting the DRB is two-fold:

1) The UCRB’s climate is dramatically different than that of the LCRB, suggesting

that the timing of large floods is notably different. This would indicate that the

mechanisms responsible for large flooding are poorly understood. By

developing an understanding of the natural variability of large floods, water

resource mangers can use the information to develop a framework to

understanding flood hazards in a rapidly changing contemporary climate. The

DRB provides an excellent analog to large portions of the UCRB, since it is

physiographically similar, and like much of the UCRB, lies within the transition

zone—the regions defined as the hinge point between the Pacific Northwest

and Southwestern Dipole (Redmond and Koch, 1991; Wise, 2010).

2) Ongoing studies along the Upper Colorado River near Moab, UT have revealed

an extended chronology of over 40-floods that have occurred in the last 2500

years. The vast majority of these floods exceed the peak historical flood of

1884. The largest paleoflood in the record, which occurred at roughly 0 AD/BC

was more than three times the magnitude of the 1884 flood—the largest

historical flood. Sedimentological evidence from some of the flood deposits

suggests the Dolores River may be a source for some of the flood sediments,

which is an interesting point given that there has been a relative lack of

extreme floods in the instrumental record.

1.4 Future research directions

Within the DRB, extensive and continuous flood terraces line the canyon. The Holocene stratigraphy in these terraces may provide critical information to understanding the higher frequency, less extreme flooding in the DRB. Preliminary age control from the upper surface of the flood terraces suggest that flooding during the latest part of the Holocene

(last several hundred years) has been responsible for their deposition. This would suggest that they may fill a void in the paleoflood record, which only preserves floods within a specific rage of stages. Convolving the SWD record with the latest Holocene stratigraphic record may allow us to place depositional/erosional episodes recorded within the vertically accreting fill terraces, within their contemporary streamflow record, illuminating questions about higher frequency and lower magnitude flooding. In addition, the study would provide an opportunity to connect the Holocene stratigraphy

with the paleoflood stratigraphy that I developed for this dissertation. Tying together

Holocene stratigraphy with paleoflood stratigraphy is not a trivial matter (e.g. Hereford,

2002), but it may illuminate some of the geomorphic details about the incision and/or aggradation of the terraces during climate shifts of the late Holocene (e.g. Medieval

Climatic Optimum and the little ice age).

The UCRB as a whole remains largely ignored by paleoflood hydrology. The Green

River, a tributary that accounts for nearly 40% of the discharge on the Upper Colorado lies in a critical hydroclimatic region. In the summer of 2010, I was involved in preliminary paleoflood field studies of the Green River, UT with Dr. Vic Baker and Dr.

Noam Greenbaum (University of Haifa, Israel), and several others. We identified numerous high-standing (up to 10m) and spatially extensive paleostage indicators that indicate there have been numerous extreme floods that would eclipse those of the modern record. In particular, SWD and driftwood lines have indicated that a single flood produced a discharge roughly double the peak historical flood. We have not developed any age control at this point, however the implications are geomorphically significant, and indicate the presence of a significant threat for downstream resources. The Bureau of

Reclamation has funded this work, and has expressed interest in continuing with funding.

We will continue to seek resources for this project because of the critical implications for both flood geomorphology, as well as hydroclimatology.

1.5 Organization of dissertation

This dissertation is organized into three manuscripts: Appendices A, B and C. Each of

the manuscripts presented in this dissertation represent my independent work, with only minimal input from committee members and graduate student colleagues. It is my intent to submit the manuscripts to the following peer-reviewed journals:

Water Resources Research (Appendix A). Water Resources Research is multidisciplinary, and focused on publishing new data. Because of the significance of the

Dolores River Basin’s location in a region where the streamflow and precipitation’s response to climate variability is mixed, there are important implications to water policy and management. In addition, this manuscript presents multiple quantitative methods providing a relatively robust understanding.

Catena or Quaternary Science Review (Appendix B). Catena’s aim is to publish articles with new data that develop a better understanding of the physical environment, which would appeal to a broader audience. It is a multidisciplinary journal includes hydrology and geomorphology as key themes. Although the study is specific to paleoflood hydrology, its implications are far-reaching into geomorphology, hydrology, geochronology and paleoclimatology. As an alternative, Quaternary Science Reviews not only publishes review papers, but publishes papers with new data provided that also provide a substantive review component. This study is organized in such a way that it provides a review of regional, Southwestern Paleoflood hydrology and also presents new data and information; so the synthesis of “review” with “new data” is apparent.

Quaternary Geochronology (Appendix C). Quaternary Geochronology is focused on publishing articles that describe advances in developing accurate age control—not necessarily new techniques. It is not limited by case studies; it seeks to be disseminated

widely, so the actual results can be used to understand climatic shifts, and other physical processes that have shaped the landscape. Appendix C fits this mission well, given the new OSL age control on flood deposits—an application of OSL that has historically been fraught is issues.

Figure 1.1 (A) Shaded relief digital elevation model of the Dolores River Basin. Major tributaries are highlighted. (B) Location of the DRB (white shaded region). Areas in darker red (Blue) indicate regions whose precipitation and streamflow are anomalously high during El Nino (La Nina) years. Green shaded regions are weakly influenced by ENSO variability. Figure 1B modified from Cayan et al., 1999.

Figure 1.2 detailed digital elevation model of the Dolores River Basin. USGS stream gage sites, and SWD study site locations are shown.

CHAPTER 2. PRESENT STUDY

The research presented in the appendices of this dissertation addresses critical questions that pertain to paleoflood hydrology in the Dolores River Basin (DRB) and the greater

Upper Colorado River Basin (UCRB). The paleoflood chronology that I have developed for the DRB—the focus of this study—is the first paleoflood chronology developed for the UCRB. Numerous analyses of the hydroclimate in the DRB have been performed as well. Furthermore, optically stimulated luminescence (OSL) was tested against independent age control as a means to evaluate its application for paleoflood hydrology.

Combined, these components provide insights to hydroclimate variability since the mid-

Holocene within the greater UCRB and DRB.

2.1 Description of Appendix A

Appendix A focuses on the hydroclimatic impacts resulting from the variability of synoptic and global scale circulation patterns. The DRB is a critical region because its hydroclimate is very poorly characterized, as is much of the UCRB. The complex response of the UCRB to climate circulation patterns has made it a central focus of research (e.g. Hidalgo and Dracup, 2003; Wise, 2010; Canon et al., 2007; McCabe et al.,

2007; McCabe and Dettinger, 1999; Balling and Goodrich, 2007; Ellis et al., 2010;

Redmond and Koch, 1991). Each of these studies improved our understanding of hydroclimatic variability in the UCRB, although the mechanisms responsible for the variability remain unclear. The research I conducted utilizes a suite of numerical analyses on the PRISM (Parameter-elevation Regressions on Independent Slopes Model)

precipitation data (Daly, 1994), U.S. Geological Survey stream gage data from major tributaries within the UCRB, and the reconstructed annual paleostreamflow from the

Dolores River (Woodhouse et al., 2006). The focus of this research is both to highlight the complexity of the UCRB’s hydroclimate, and more importantly, to extract the principal features that are associated with hydroclimatic variability.

2.2. Description of Appendix B

Appendix B of this dissertation focuses on the geomorphological interpretations of the

Paleoflood chronology I developed. More than fifty chronometric ages were developed for the paleoflood chronology by combining OSL and AMS 14C ages. This study is the first comprehensive study of paleofloods in the DRB, an important sub-basin to the

UCRB. In fact, north of the San Juan River confluence, this is the first paleoflood study within any tributary of the UCRB. The UCRB as a whole is very poorly understood from a paleoflood perspective, and vast critical regions with unique hydroclimatic signatures remain entirely unstudied. The DRB represents of one of these critical regions—the western slope of Colorado. The DRB lies in a region whose hydroclimate is rather complicated (e.g. Wise, 2010), and the goal of this research component was to identify the periods of time throughout the middle-late Holocene when flooding was more or less prevalent. By comparing the periods to known climatic periods (e.g. Medieval Climatic

Optimum), hypotheses about the mechanisms and climate associations were made possible. The chronology presented here is the beginning of a broad network of studies needed in the UCRB. The National Science Foundation (NSF) provided the Doctoral

Dissertation Improvement Grant (DDIG), which I was awarded in the spring of 2009

funding for this component of my dissertation though. The U.S. Bureau of Reclamation

(BoR) took an interest in the study as well, and has identified the UCRB as a critical watershed in need of further paleoflood studies, so this study was in part, funded by a grant from the BoR to Dr. Vic Baker who supported my spring 2010 research assistantship, as well as several field trips and conference presentations.

2.3 Description of Appendix C

Appendix C focuses on testing OSL against independent age control. Most paleoflood studies have relied on locating organic materials for 14C dating. In some cases, particularly in arid regions, the organic materials cannot be interpreted in a stratigraphic context, rendering the confidence of the ages very low (Delong and Arnold, 2005).

Recent advances in OSL dating have allowed for the dating of SWDs—an application historically riddled with complications (e.g. Porat, 2001). The possibility of accurately dating flood deposits with OSL is a significant advance in the potential for accurately deriving age-control for water-lain sediments because OSL is used to directly date the sediments age of deposition (Wallinga, 2002) and does not require any problematic conversions to calendar years—a potentially large source of error in radiocarbon dating

(Guilderson, 2005). OSL remains in a relatively developmental stage in fluvial studies, however. OSL is a radiogenic technique, and works on the premise that visible light resets the luminescence signal in an individual grain of quartz or feldspar (Aitken, 1998).

Upon burial, quartz and feldspar grains absorb free electrons from the natural radioactive decay (the ‘charge’) from the surrounding material, and that charge increases linearly with time. It is that ‘charge’ which allows us to derive the age, because when it is

exposed to light, a small amount of light (luminescence) is released, and the amount of luminescence is directly proportional to the amount of ‘charge’ (Lian and Roberts, 2006).

Numerous issues inhibit the resetting process that is commonly referred to as bleaching or zeroing. The limiting factors to bleaching are light exposure, and in fluvial environments, especially in turbulent flooding, many factors limit the light source.

The manuscript presents OSL ages that I processed and analyzed at the Utah State

University Luminescence Lab. It compares them with accompanying AMS 14C ages that

I processed at the University of Arizona AMS lab. The results are used as a case study for applying OSL in paleoflood hydrology. The manuscript includes a discussion of the transport and depositional issues that impact the OSL ages, as well as the post- depositional issues. Standard statistical models (Galbraith et al., 1999), originally developed to address aliquot scatter are applied in the study to address issues associated with partial bleaching. As the OSL technique sees more widespread use, it is critical for paleoflood workers who utilize OSL to be aware of the limitations and sources of error, in order to minimize inaccuracy in flood chronologies. Funding for this aspect of the study was secured from the NSF, through the DDIG that I was awarded in Spring 2009.

2.4 Important results and conclusions from this dissertation

Using multiple geochronology techniques led to a more robust paleoflood chronology, and the numerical analyses of the systematic hydroclimatic data detailed the complexity of the hydroclimate in the DRB and UCRB. The most important findings of this study are described below.

The DRB paleoflood chronology extends roughly 9000 years B.P. In particular, the last

2,000-years of the record provides critical information about the timing, and allows for us to develop hypotheses about the potential mechanisms responsible for the floods. The modern hydroclimate (last 115 years) has not effectively produced any floods that approach the magnitude of those preserved by the slackwater paleoflood record. In fact, this apparent hiatus in extreme floods seems to extend from roughly 1200 A.D. to the present. From roughly 100 A.D. until 1100 A.D., extreme floods were frequent in the

DRB. In fact, no less than 13 extreme floods occurred during this time. In contrast, extreme flooding in the LCRB was most active from roughly 1000 A.D. to the present.

Prior to roughly 1000 A.D., the floods are noticeably scarce in the record (Ely et al.,

1993; Ely, 1997). It should be noted in this summary of results that natural, minimum and maximum flood stage filters have likely hindered the preservation of flood of floods, particularly at the most detailed SWD study site. A bedrock ceiling and lower base only permitted a roughly 2m vertical section of deposition. Furthermore, the alcove that stores the SWD only extends roughly 2m into the bedrock face, limiting the degree to which inset relationships may have formed, so it if feasible that many more flood units would have been preserved. This was a common feature among most of the SWDs in the study reaches. The limitation is a clear indicator that this record represents a minimum number of events.

The out of phase relationship between the UCRB and LCRB is probably indicative of the circulation patterns that were prevalent. For example, within the LCRB, it is thought that frequent catastrophic floods are a result of shortened return intervals and

strengthened intensity of El Nino events (Ely et al, 1997; Ely, 1997). Within the DRB and UCRB, the circulation patters responsible for streamflow, flooding and precipitation variability are less clear. In this study, the hydroclimate analyses of precipitation variability in the DRB indicate that ENSO variability has little effect on DRB streamflow and precipitation. In addition, it seems that the precipitation and streamflow may be associated with a low-amplitude, northern Pacific, PDO-like pattern, although not the specific phase of the PDO. Instead, there seems to be a flood response to heightened variance of streamflow-precipitation surrounding the years where the PDO is shifting from positive to negative, or vice-versa. This conclusion was made evident using multiple analyses of different data sets (precipitation and streamflow) within the

DRB. The use of multiple analyses made the result more apparent. Unfortunately extrapolating this result to the large floods that are preserved by the paleoflood record is ill advised, since the floods were significantly larger in the past. Nevertheless, The influence of a variable North Pacific influences historically large floods in the DRB, and may provide a utility to water mangers.

When the DRB’s paleoflood record is combined with concurrent paleoclimate records

(dendrochronology and lake cores), it is clear that periods of frequent, very large floods were coincident with anomalously dry conditions. The floods occurred well into the

Medieval Climate Anomaly (MCA) MCA (900-1300 A.D.; Hughes and Funkhouser,

1998; Benson et al., 2007; Cook et al., 2004). The MCA in the UCRB was a period of widespread (Cook et al., 2004) and severe droughts (Meko et al., 2007). Furthermore,

Fire frequency records derived from lake cores in the San Juan Mountains, indicate that the period was quite dry, and large, widespread fires were frequent (Anderson et al.,

2008). This is in opposition to the flood record of the Lower Colorado River Basin.

Within the LCRB, the flood frequency-magnitude decreased during the onset of the

MCA (Ely et al., 1997). The increased flooding in the LCRB was interpreted as the result of a cooler and wetter temperature/precipitation regime, brought on by frequent El

Nino events (Ely et al, 1993; Ely, 1997). These results were supported by multiple geochronology techniques: AMS 14C as well as OSL. These results are elaborated upon within the manuscripts, and several potential mechanisms for flooding are postulated; however a great deal of uncertainty still remains about the climate conditions that drive extreme floods. Unlike the LCRB, where truly extreme floods have occurred in the last several decades, (e.g. House and Hirschboeck, 1997), extreme flood-generating analogs are not available within adjoining basins, with the exception of the San Juan River, although the San Juan River’s flood hydroclimate is more similar to rivers in the LCRB making it a poor analog for the DRB and greater UCRB. Future research should identify basins with similar hydroclimates that may have experienced extreme floods during historical periods. Detailed studies of the flood-generating circulation, and the antecedent conditions would allow for extrapolation to the DRB, and other sub-basins within the UCRB.

The application of both OSL, as well as AMS 14C improved the quality of the age control for this study. By applying the independent methods, I was able to assess the quality of the OSL method, and results were satisfying. The OSL samples did require

careful analysis, however. Initial ages derived from small numbers of large aliquots

(sediment subsamples) were producing ages considerably older than the actual, depositional ages—in some cases, doubling the age estimate. Small aliquots led to better identification of partially bleached and bioturbated samples, ultimately leading to better age control. Partial bleaching is clearly a problem in paleoflood studies. To address this, I used statistical models based on the degree of aliquot scatter. There is a physical basis in fluvial studies to use some statistical models, as opposed to the mean age

(Galbraith et al., 1999; Arnold et al., 2007). There are few mechanisms that produce overly young estimates, but far more that produce estimates that are simply too old, and therefore the use of statistical models is justified and recommended. The models used in this study, the minimum age (MAM), and central age models (CAM) (Galbraith et al.,

1999), are examples of statistical models that can be used to derive the age of flood deposition. Which model is selected dependends on the degree of aliquot scatter. In this study, I found that in general, the younger samples required the MAM, whereas most of the older samples (>2000-years B.P.) required the CAM. Reasons for this are expanded upon in Appendix C. These results are significant to earth scientists who are interested in applying OSL to flood deposits. These results indicate that OSL may be a valid tool for paleoflood hydrology, particularly in the SW U.S.

WORKS CITED

Adams, D.K. and Comrie, A.C., 1997. The North American monsoon. Bulletin of the American Meteorological Society, 78(10): 2197-2213.

Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford Science Publications, 267 pp.

Baker, V.R., 1987. Paleoflood hydrology and extraordinary flood events. Journal of Hydrology, 96(1-4): 79-99.

Baker, V.R., Webb, R.H. and House, P.K. (Editors), 2002. The scientific and societal value of paleoflood hydrology. Ancient floods and modern hazards, principles and applications of paleoflood hydrology, 4. American Geophysical Union, Water Science and Application Series, Washington, D.C., 1-19 pp.

Balling, R.C. and Goodrich, G.B., 2007. Analysis of drought determinants for the Colorado River Basin. Climatic Change, 82(1-2): 179-194.

Benito, G. and Thorndycraft, V.R., 2005. Palaeoflood hydrology and its role in applied hydrological sciences. Journal of Hydrology, 313(1-2): 3-15.

Benson, L., Petersen, K. and Stein, J., 2007. Anasazi (pre-Columbian Native-American) migrations during the middle-12th and late-13th centuries - Were they drought induced? Climatic Change, 83(1-2): 187-213.

Blainey, J.B., Webb, R.H., Moss, M.E. and Baker, V.R. (Editors), 2002. Bias and information content of paleoflood data in flood-frequency analysis. Ancient floods, modern hazards, principles and applications of paleoflood hydrology, 5. American Geophysical Union, Water Science and Application Series, Washington, DC, 161-174 pp.

Brown, D.P. and Comrie, A.C., 2004. A winter precipitation 'dipole' in the western United States associated with multidecadal ENSO variability. Geophysical Research Letters, 31(9).

Bryson, R.A. and Hare, F.K., 1974. The climates of North America. Climates of North America, World Survey of Climatology II. Elsevier, New York.

Canon, J., Gonzalez, J. and Valdes, J., 2007. Precipitation in the Colorado River Basin and its low frequency associations with PDO and ENSO signals. Journal of Hydrology, 333(2-4): 252-264.

Cayan, D.R., Redmond, K.T. and Riddle, L.G., 1999. ENSO and hydrologic extremes in the western United States (vol 12, pg 2881, 1999). Journal of Climate, 12(12):

3516-3516.

Cook, E.R., Woodhouse, C.A., Eakin, C.M., Meko, D.M. and Stahle, D.W., 2004. Long- term aridity changes in the western United States. Science, 306(5698): 1015- 1018.

D.R., C. and J.O., R., 1984. Local relationships between United States west coast precipitation and monthly mean circulation patterns. Mon. Weather Rev, 112: 1276–1282. .

Daly, C., Neilson, R.P. and Phillips, D.L., 1994. A statistical topographic model for mapping climatological precipitation over mountainous terrain. Journal of Applied Meteorology, 33(2): 140-158.

DeLong, S.B. and Arnold, L.J., 2007. Dating alluvial deposits with optically stimulated luminescence, AMS C-14 and cosmogenic techniques, western Transverse Ranges, California, USA. Quaternary Geochronology, 2(1-4): 129-136.

Ellis, A.W., Goodrich, G.B. and Garfin, G.M., 2010. A hydroclimatic index for examining patterns of drought in the Colorado River Basin. International Journal of Climatology, 30(2): 236-255.

Ely, L.L., 1997. Response of extreme floods in the southwestern United States to climatic variations in the late Holocene. Geomorphology, 19(3-4): 175-201.

Enzel, Y., Ely, L.L., House, P.K., Baker, V.R. and Webb, R.H., 1993. Paleoflood evidence for a natural upper bound to flood magnitudes in the Colorado River Basin. Water Resources Research, 29(7): 2287-2297.

Gavin, D.G., Brubaker, L.B. and Lertzman, K.P., 2003. Holocene fire history of a coastal temperate rain forest based on soil charcoal radiocarbon dates. Ecology, 84(1): 186-201.

Guilderson, T.P., Reimer, P.J. and Brown, T.A., 2005. The Boon and Bane of radiocarbon Dating. Science 307(5708): 4.

Hansen, E.M.S., F.K., 1981. Meteorology of important rainstorms in the Colorado River and Great Basin drainages. In: N.O.a.A.A.H. Report (Editor), pp. 167.

Hereford, R., 2002. Valley-fill alluviation during the Little Ice Age (ca. AD 1400-1880), Paria River basin and southern Colorado Plateau, United States. Geological Society of America Bulletin, 114(12): 1550-1563.

Hidalgo, H.G. and Dracup, J.A., 2003. ENSO and PDO effects on hydroclimatic variations of the Upper Colorado River basin. Journal of Hydrometeorology, 4(1):

5-23.

Hirschboeck, K. (Editor), 1988. Flood hydroclimatology. Flood Geomorphology. John Wiley & Sons, 27‑49 pp.

Hirschboeck, K.K., 2003. Floods, Paleofloods, and Drought: Insights from the Upper Tails, CLIVAR/PAGES/IPCC Drought Workshop. University of Arizona.

Hirschboeck, K.K., Ely, L.a.M. and R.A. (Editors), 2000. Hydroclimatology of meteorologic floods Inland Flood Hazards: Human, Riparian and Aquatic Communities. Cambridge University Press, p. 39-72 pp.

Hosking, J.R.M. and Wallis, J.R., 1986. Paleoflood hydrology and flood frequency- analysis. Water Resources Research, 22(4): 543-550.

House, P.K. and Hirschboeck, K.K., 1997. Hydroclimatological and paleohydrological context of extreme winter flooding in Arizona, 1993. Storm‑Induced Geological Hazards: Case Histories from the 1992‑1993 Winter Storm in Southern California and Arizona, Geological Society of America Reviews in Engineering Geology, pp. 1‑24.

Hughes, M.K. and Funkhouser, G. (Editors), 1998. Extremes of Moisture Availability Reconstructed from Tree Rings for Recent Millennia in the Great Basin of Western North America. The Impacts of Climate Variability on Forests. Springer Verlag, Berlin 99–107 pp.

Hupp, C.R. (Editor), 1982. Plant Ecological Aspects of Flood Geomorphology and Paleoflood History. Flood Geomorphology. Wiley.

Intergovernmental Panel on Climate Change (IPCC), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Kim, T.W., Valdes, J.B., Nijssen, B. and Roncayolo, D., 2006. Quantification of linkages between large-scale climatic patterns and precipitation in the Colorado River Basin. Journal of Hydrology, 321(1-4): 173-186.

Klemes, V. (Editor), 1994. Statistics and probability: Wrong remedies for a confused hydrologic modeler. Statistics for the environment 2: water related issues. John Wiley & Sons Ltd.: Chisester, England., 345–370 pp.

Knox, J.C., 2000. Sensitivity of modern and Holocene floods to climate change.

Quaternary Science Reviews, 19(1-5): 439-457.

McCabe, G.J., Betancourt, J.L. and Hidalgo, H.G., 2007. Associations of decadal to multidecadal sea-surface temperature variability with Upper Colorado River flow. Journal of the American Water Resources Association, 43(1): 183-192.

McCabe, G.J. and Dettinger, M.D., 1999. Decadal variations in the strength of ENSO teleconnections with precipitation in the western United States. International Journal of Climatology, 19(13): 1399-1410.

McCabe, G.J., Palecki, M.A. and Betancourt, J.L., 2004. Pacific and Atlantic Ocean influences on multidecadal drought frequency in the United States. Proceedings of the National Academy of Sciences of the United States of America, 101(12): 4136-4141.

Meko, D.M. et al., 2007. Medieval drought in the upper Colorado River Basin. Geophysical Research Letters, 34(10).

Milly, P.C.D. et al., 2008. Climate change - Stationarity is dead: Whither water management? Science, 319(5863): 573-574.

Milly, P.C.D., Wetherald, R.T., Dunne, K.A. and Delworth, T.L., 2002. Increasing risk of great floods in a changing climate. Nature, 415(6871): 514-517.

Mock, C.J., 1996. Climatic controls and spatial variations of precipitation in the western United States. Journal of Climate, 9(5): 1111-1125.

Porat, N., Zilberman, E., Amit, R. and Enzel, Y., 2001. Residual ages of modern sediments in an hyperarid region, Israel. Quaternary Science Reviews, 20(5-9): 795-798.

R.F., G., Roberts, R.G., Laslett, G.M., Yoshida, H. and Olley, J.M., 1999. Optical dating of single and multiple grains of quartz from Jinmium rock shelter, northern Australia: Part I, Experimental design and statistical models. Archaeology 41: 339–364.

Redmond, K.T., Enzel, Y., House, P.K. and Biondi, F. (Editors), 2002. Climate variability and flood frequency at decadal to millennial time scales. in Ancient Floods, Modern Hazards: Principles and Applications of Paleoflood Hydrology. American Geophysical Union, 21-45 pp.

Redmond, K.T. and Koch, R.W., 1991. Surface climate and streamflow variability in the Western United-States and their relationship to large-scale circulation:. Water Resources Research, 27(9): 2381-2399.

Seager, R. et al., 2007. Model projections of an imminent transition to a more arid climate in southwestern North America. Science, 316(5828): 1181-1184.

Sheppard, P.R., Comrie, A.C., Packin, G.D., Angersbach, K. and Hughes, M.K., 2002. The climate of the US Southwest. Climate Research, 21(3): 219-238.

Sigafoos, R.S., 1961. Vegetation in relation to flood frequency near Washington, D. C. . In: U.S.G. Survey (Editor). Professional Paper pp. 248-249.

Stedinger, J.R. and Cohn, T.A., 1986. Flood frequency analysis with historical and paleoflood information. Water Resources Research, 22(5): 785-793.

Stockton, C.W. and Jacoby, G.C., 1976. Long-term surface-water supply and streamflow trends in the Upper Colorado River basin based on tree-ring analyses. Lake Powell Research Project Bulletin, 18: 1-70.

Trenberth, K.E., Dai, A., Rasmussen, R.M. and Parsons, D.B., 2003. The changing character of precipitation. Bulletin of the American Meteorological Society, 84(9): 1205-+.

Wallinga, J., 2002. Optically stimulated luminescence dating of fluvial deposits: a review. Boreas, 31(4): 303-322.

Weare, B.C. and Hoeschele, M.A., 1983. Specification of monthly precipitation in the western United States from monthly mean circulation. Journal of Climate and Applied Meteorology, 22(6): 1000-1007.

Wise, E.K., 2010. Spatiotemporal variability of the precipitation dipole transition zone in the western United States. Geophysical Research Letters, 37.

Woodhouse, C.A., 1997. Winter climate and atmospheric circulation patterns in the Sonoran Desert region, USA. International Journal of Climatology, 17(8): 859- 873.

Woodhouse, C.A., Gray, S.T. and Meko, D.M., 2006. Updated streamflow reconstructions for the Upper Colorado River Basin. Water Resources Research, 42(5).

Yanosky, T.M. and Jarrett, R.D. (Editors), 2002. Dendrochronologic evidence for the frequency and magnitude of paleofloods Ancient floods, modern hazards, principles and applications of paleoflood hydrology, 5. American Geophysical Union Water Science and Application Series, 77-89 pp.

APPENDIX A

HYDROCLIMATOLOGY IN THE DOLORES RIVER BASIN IN THE SOUTHERN ROCKY MOUNTAINS, USA: HEMISPHERIC AND SYNOPTIC CONTROLS ON NONSTATIONARY PRECIPITATION AND STREAMFLOW REGIMES

Michael L. Cline 40

Abstract

Hydroclimatic variability in the Upper Colorado River Basin (UCRB) is definitively complex in the way it responds to atmospheric and ocean circulation patterns. This study highlights the dominant modes of hydroclimatic variability that impact the Dolores River

Basin (DRB)—a sub basin of the UCRB that lies between the dominant storm tracks associated with variability of the El Niño Southern Oscillation (ENSO). Results suggest that the modes of variability may be nonstationary: at times the streamflow may modulated by an ENSO-like interannual variability, and at other times, precipitation and streamflow signals related to the ENSO-like pattern are minimal. A North Pacific, interdecadal circulation pattern, may at least in part, modulate the nonstationarity. 41

1. INTRODUCTION

Hydrology in the western U.S. is particularly sensitive to climate change (Seager et al.,

2007) and has been the focus of many studies investigating the mechanisms responsible for hydroclimatic variability (e.g. Cayan et al, 1999; Woodhouse, 1997; Hirschboeck,

1987; Kim et al., 2005; Hirschboeck, 1988; House and Hirschboeck, 1997; Ely et al.,

1993; Ely, 1997; Redmond and Koch, 1991; Zhang et al., 1997; Cayan, 1999; Timelsena et al., 2009; McCabe et al., 2007; Hereford and Webb, 1992; McCabe and Dettinger,

1999; Hidalgo and Dracup, 2003; Wise, 2010). Several of these studies have emphasized the differential responses between the Lower Colorado River Basin (LCRB) and the

Upper Colorado River Basin (UCRB), to synoptic and hemispheric circulation patterns

(e.g. Cayan et al., 1999). The LCRB lies in the center of action for the Northwest-

Southwest dipole (Redmond and Koch, 1991)—the spatial pattern of contrasting precipitation anomalies that result from variability of the El Niño Southern Oscillation

(ENSO). In contrast, much of the UCRB lies between these two areas of opposite association, which has caused it to remain a basin whose hydroclimatic variability is poorly constrained. Adding to the complexity, human-caused climate change is accelerating the changes between ocean/air circulation and precipitation delivery 42

mechanisms (Milly et al., 2008; Ellis et al., 2009; Vecchi et al., 2006; Milly et al., 2002).

This research investigates the synoptic and hemispheric-scale circulation patterns associated with hydroclimatic variability in the Dolores River Basin (DRB), a sub-basin of the UCRB (Figure 1). The DRB is emphasized in this study because it is positioned within the “transition zone”—the region that lies between the centers of action of the

Western U.S. precipitation dipole (Wise, 2010), making its hydroclimate very complex.

The motivation for this study is to characterize hydroclimatic variability for the DRB, in part to provide a context for the DRB’s paleoflood record that has been recently developed (Appendix B of this dissertation). The DRB paleoflood study is the first comprehensive paleoflood study in the UCRB, so a hydroclimatological study is critical to develop a context by which we can understand the occurrences of anomalously large floods. This manuscript is organized by first providing a general review of the hydroclimatic variability of the UCRB based on historical and paleodata. The DRB is then described based on the results from a suite of quantitative analyses of the DRB’s winter precipitation and snowmelt-season streamflow from 1895-2007. The principal modes of variability from these results are compared with circulation indices that 43

describe the spatial patterns of se and air circulation, which ultimately help to develop a general characterization of the precipitation and streamflow variability in the DRB.

2. BACKGROUND

The UCRB has been the center of much research in hydroclimatology and paleoclimatology because it is the single most critical surface-water resource to the southwestern U.S., and its annual water budget is highly vulnerable to climate warming

(e.g. Christensen et al., 2004, Ellis et al., 2009). Numerous researchers have worked to understand streamflow and/or precipitation using major circulation indices (e.g. Cayan et al, 1999; Woodhouse, 1997; Kim et al., 2005; Redmond and Koch, 1991; Zhang et al.,

1997; Cayan, 1999; Timelsena et al., 2009; McCabe et al., 2007; McCabe and Dettinger,

1999; Hidalgo and Dracup, 2003; Wise, 2010). Circulation indices help us to understand the variability of variable hydroclimatic time series (Hirschboeck, 2009), a critical element in developing a framework for understanding how hydrological regimes may change under a rapidly changing climate. The most attention has been paid to El Niño and La Niña—the interannual alternation between sea surface temperature (SST) anomalies along the equator, between Peru and the central Pacific Ocean. Accompanying 44

El Niño and La Niña is the Southern Oscillation (ENSO), which refers to the atmospheric pressure gradient between the central Pacific and Northern Australia. The alternating

SST and air pressure patterns are causally linked, and are therefore referred to as the El

Niño Southern Oscillation. Variability of ENSO causes much of the precipitation and streamflow variability in the western U.S. This is explained by the general location of the equatorial pacific anomaly SSTs, which control the location and trajectory of the subtropical jet (Seager et al., 2010)--a critical element to the western North American-

ENSO teleconnection. The location and trajectory of the subtropical jet stream appears to be the dominant cause of storm-frequency modulation in parts of the western U.S.

(Seager et al., 2010). It is responsible for steering transient eddies (i.e. developing cyclonic storm systems) on a more zonal route across the Pacific, directly into the southwest during El Niño years, or on a more meridional route into the Pacific Northwest during La Niña years (Seager et al., 2005; Seager et al., 2010).

Another circulation index that has been the focus of many studies in the UCRB is the

Pacific Decadal Oscillation (PDO), which describes the modulation of November—

March SSTs in the northern Pacific Ocean (20° N poleward, 1900—1993). The index is 45

based on the 1st Empirical Orthogonal Function (EOF) of residual SSTs (global monthly mean SST is removed, e.g. Mantua and Hare, 2002) over the North Pacific, from 20° N, poleward (developed by Hare, 1996; Zhang, 1996). The spatial patterns of the PDO are similar to ENSO, so it has been referred to as ENSO-like (Mantua et al., 1997), although the greatest variability (center of action) in the PDO is in the North Pacific, and it operates over much longer, decadal timescales. The PDO has been shown to modulate the degree to which ENSO is associated with wet conditions in the southwest (e.g.

Gershunov and Barnett, 1998), which explains the transience in the spatial hydroclimatic footprint of the ENSO impacted region (Wise, 2010). The oscillatory frequency of the

PDO is not well constrained, although two general periodicities exist: 15-25 years and

50-70 years (Mantua and Hare, 2002; Minobe, 1997).

ENSO and the PDO are the two most widely discussed teleconnections in the western,

U.S. hydroclimate because of their associations with extreme, climate-related events such as floods, droughts and wildfires (e.g. Redmond and Koch, 1991; Cayan, 1999;

Timelsena et al., 2009; McCabe et al., 2007; Zhang et al., 1997; Cayan, 1999; Hereford and Webb, 1992; Hidalgo and Dracup, 2003; Grissino-Mayer et al., 2004; Westerling et 46

al., 2003). PDO has impacts on precipitation patterns in the U.S. because it modulates the atmospheric flow patterns in the middle atmosphere between zonal flow patterns

(during dry periods in the Southwest), to meridional dominated patterns (during the wet decades) (Hirschboeck et al., 2000; Webb and Betancourt, 1992). In the western U.S., abrupt shifts in precipitation occurred in 1905/06, 1932/33, 1946/47, and 1976/77 (Zhang et al., 1997; Mantua et al., 1997; Herford et al., 2002). These shifts are thought to be associated with sign reversals of the PDO. The relationship between precipitation regime shifts and PDO phases has sparked a great deal of research on the PDO teleconnection

(Zhang et al., 1997; Hereford et al., 2002; Balling and Goodrich, 2007). In fact, Principal

Component Analysis (PCA) has demonstrated that the PDO has more of an impact on drought intensity in the western U.S., than ENSO (Balling and Goodrich, 2007), although, other dominant modes are debated to explain a great deal of the variability as well (e.g. Woodhouse et al., 2009). In an effort to explain additional variance in multidecadal variability, attention has been paid to the Atlantic Multidecadal Oscillation

(AMO) in recent years, because it has been shown to correlate with streamflow in some parts of the UCRB over multi-decadal timescales (e.g. McCabe et al., 2007; Wise, 2010), and it has also been suggested that the correct phase relationship between the PDO and 47

AMO is responsible for extreme drought episodes (Fye et al., 2003; Benson et al., 2007).

The extreme limitations on the degrees of freedom, however, limit our ability to incorporate it into this study.

The precise relationship between ENSO and the PDO is not fully understood. This has limited the utility in identifying individual circulation patterns and their associations with climate and weather in a specific region. Many circulation indices are intimately linked through complex ocean-air interactions (Quadrelli and Wallace, 2004), and some processes may represent alternative timescales of a similar process, e.g. ENSO and the

PDO (Lau and Nath, 1994; Zhang et al., 1996; Newman et al., 2003). For example, the

“atmospheric bridge” describes the (lagged 3-month) forcing of surface fluxes over the north Pacific, by anomalous SSTs in the tropical ENSO during the peak ENSO maxima

(i.e. Trenberth and Hurrel, 1994; Alexander et al., 2002). This forcing describes nearly

50% of the variance between Jan-Mar (Alexander et al., 2002). The high degree of temporal autocorrelation of the PDO (referred to as the linearly independent component of the ESNO-like feature [i.e. Zhang et al., 1997]) is likely a result of mixing deeper ocean waters, integrating them to the surface before the following warm season, which 48

reintroduces warm SSTs (Alexander et al., 1999). There is also evidence that the PDO may be entirely dependent on ENSO on all timescales, suggesting that the PDO may be thought of as a “reddened” ENSO (Newman et al., 2003).

The unique position of the DRB within the latitudinal position of the transition zone suggests that it may have a mixed response to major teleconnections such as ENSO and the PDO. Wise (2010) described the transition zone as being spatiotemporally dynamic—it has made notable latitudinal shifts in the past, particularly in the east of the

Great Basin. These shifts were shown to roughly coincide with phase shifts in the PDO.

Unfortunately, the analysis of the transition zone in the Wise (2010) study ended at the

Utah-Colorado border, omitting the DRB. The study highlighted the dynamic nature of the transition zone, which increased eastward, suggesting the DRB may respond in an even more variable manner to large-scale teleconnections. The development of a comprehensive paleoflood chronology has necessitated a closer investigation of the DRB, in particular, because it still remains a region whose hydroclimatic variability is poorly understood.

Long-term paleoclimate records are highly relevant to interpreting the DRB’s modern

hydroclimate because they place the DRB in a broader temporal context. Long-term

streamflow records serve as an excellent utility in developing this context because of

their high (annual) resolution. The extended records may also provide water managers

with valuable information because they help to inform about the range of possibilities

facing water resources, given an uncertain future. The allocation of the Colorado River’s

water famously exemplified the need to incorporate paleodata when paleohydrology

researchers realized that the 1922 Colorado River compact was based on the 2nd wettest

20-year period in the last 400-years (Stockton and Jacoby, 1976; Woodhouse et al.,

2006). Furthermore, the dendrochronology-based streamflow reconstructions of the

Upper Colorado River and the UCRB tributaries have shown the presence of extended droughts, ubiquitous throughout much of the western U.S. (Cook et al., 2003), and also the presence of extreme, and persistent droughts (Meko et al., 2007) that eclipse those identified by the systematic record. All of these reconstructions have brought to light the highly variable nature of the UCRB’s hydroclimate over the last several hundred to more than one thousand years (e.g. Stockton and Jacoby, 1976; Woodhouse, 2003; Cook et al.,

2004, Woodhouse, 2006; Meko, 2007). It should be noted that large floods—those 50

recorded in the paleoflood record—are not necessarily captured by dendrochronology because they need not be associated with wet periods. In fact, it has been shown elsewhere that floods generally associated with periods of increased variance in the hydroclimate and do not necessarily follow trends in the streamflow (e.g. Webb and

Betancourt, 1992; Hirschboeck et al., 2000, Hirschboeck, 2003).

3. STUDY SITE DESCRIPTION

The Dolores Basin is situated on the western slope of the San Juan Mountains, but it also includes the Southern Uncompahgre Plateau and the eastern flank of the La Sal

Mountains (Figure 2). Regional to local-scale hydroclimatic variability in the DRB is controlled at multiple scales by physiographic, oceanic and atmospheric factors, each operating at different spatial and temporal scales (Mock, 1996). The DRB lies at the northernmost extent of the typical air masses that arrive from the topographically low region in southern California and northern Mexico. Northern storms may also reach the

DRB, although they are generally weakened because the air masses are required to traverse numerous topographic features (McGinnis, 2000). Most of the winter precipitation in the DRB is attributed to Pacific air masses (Bryson and Hare, 1974); 51

however, relating specific synoptic upper-air features with precipitation variability has proven difficult because of the dominant control exerted by topography—the major factor controlling precipitation variance (Weare and Hoeschele, 1983; Cayan and Rhodes, 1984;

Sheppard et al., 2002). The synoptic mechanisms responsible for delivering the moisture are generally mid-latitude cyclones (McGinnis, 2000). These are relatively common features during the winter months, and, depending on their source region, they bring the potential for significant precipitation. The synoptic lifting associated with the lows are enhanced orographically, and can result in heavy precipitation for multiple days, especially if the air mass is cutoff or stalled (Mock, 1996), which further enhances the high degree of spatial variability caused by topography (e.g. Sheppard et al., 2002). For example: snowfall differences between low and high elevations may differ up to an order of magnitude. The amount of precipitation is dependent on a number of factors, but primarily the storm tracking and storm source. Northern storms are steered by the spatial and strength relationship between the subtropical anticyclone off the coast of California, and the Aleutian Low—a quasi-stationary statistical feature in the North Pacific, defined by the location of peak intensity of cyclonic intensity over a pre-defined period of time

(Rodionov et al., 2007). Storms that enter the San Juan Mountains from the south tend to 52

produce greater amounts of precipitation than any other storms. These systems not only carry saturated air mass from the Pacific, but they also draw moisture from warm SSTs in the Gulf of California (Keen, 1996). These southern storm tracks from the warmer ocean

California, and to a lesser degree the Gulf of Mexico (Adams and Comrie, 1997).

Dissipating tropical cyclones are rare because of the distance from water bodies; however, the largest flood on the Dolores River was the result of a dissipating tropical cyclone, suggesting that under the right conditions, these systems may be able to reach the intermountain region (Hansen and Schwarz, 1981).

4. DATA AND METHODS

Precipitation: Precipitation analyses are based on the PRISM (Parameter-elevation

Regressions on Independent Slopes Model) data from the University of Oregon. Data are resolved at the monthly scale from 1895—present, on a 2.5 arc-minute digital elevation model (DEM) (re-sampled from a 3 arc second DEM). This gridded dataset is ideal for 53

mountainous or topographically complex terrain because the PRISM model is applied to interpolate between station data using physiographic elements such as elevation and slope aspect (Daly et al., 1994, 2002, 2008; Daly, 2006). This study uses the winter monthly data (November-March) for the 111-year period from 1895—2005 for the western U.S., ranging from approximately 95—125 degrees West longitude. This region includes the entire coterminous Western U.S. from the Pacific to the western Great Plains.

Streamflow: Daily historical streamflow data were acquired from the U.S. Geological

Survey. These data were acquired for most of the major tributaries within the UCRB, specifically from upper sub-basin sites. Elevation has been shown to influence precipitation variability, particularly ENSO-precipitation variability (i.e. Hidalgo and

Dracup, 2003), so we aimed to collect data from sites at similar elevations. Efforts were made to collect historical data from gages that are minimally impacted by dams.

Unfortunately, the highly regulated nature of the western U.S. water resources led to a relatively sparse dataset. Estimated natural flows are available from the U.S. Bureau of

Reclamation, for most of the 20th century, although these flows are calculated at the 54

monthly level, and, while this study uses seasonally resolved streamflow totals, we are concerned with flooding that typically occurs over several days.

We focused our efforts on the spring runoff season (Jan-July). This period omits the majority of summer convective storm activity in the region, and captures the majority of the winter precipitation and snowpack. Supplementing these data, we used the dendrochronology-based reconstruction of the Dolores River’s annual streamflow to

1569 (Woodhouse et al., 2006).

Circulation indices: In order to provide a context for the precipitation variability in the

DRB, we have used circulation indices, specifically ENSO and PDO. These are statistically-derived numerical indices that relate the spatial structure of atmospheric pressure and/or sea surface temperatures, and represent two of the major patterns of circulation that are associated with variability in the UCRB’s hydroclimate. The sea-air patterns are responsible for either the genesis and/or modulation of the strength and tracking of storms centers (e.g. Seager et al., 2010). Numerous extreme floods and droughts in the western U.S. over historical and paleo timescales have been attributed to 55

ENSO and the ENSO-like circulation feature—the PDO (i.e. Zhang et al., 1997). These

associations motivated us characterize the spatial and temporal variability of the

hydroclimate in the DRB using these two indices. While numerous studies have linked

variability in the UCRB to the AMO (e.g. Fye et al., 2003; Benson et al., 2007; McCabe et al., 2007; Wise, 2010; Ellis et al., 2009; Matter et al., 2010), the length of its period relative to the instrumental record leaves it with too few degrees of freedom (n=1), thereby limiting it’s utility.

The ENSO index we selected is the BEST index (Smith and Sardeshmukh, 2000), which combines the atmospheric pressure-based index (i.e. SOI) and SST-based index (i.e.

NINO3.4). The selected El Niño/La Niña winter years are determined as the upper 20th

percentile of both the SOI and the NINO3.4, based on a 5-month running mean (Smith

and Sardeshmukh, 2000): 1896/97, 1899/1900, 1902/03, 1918/19, 1925/26, 1930/31,

1941, 1957/58, 1965/66, 1972/73*, 1982/83*, 1987, 1991/92, 1997/98, and 2002/03; and

La Niña years as follows: 1889/90, 1917/18, 1950/51, 1955/56*, 1971, 1973/74*,

1975/76*, and 1988/89*, 2000/01*. The latest PDO index (Mantua et al., 1997; Zhang et

* indicates strong El Niño/La Niña y ears. 56

al., 1997) was downloaded from the University of Washington’s Joint Institute for the

Study of the Atmosphere and Ocean.

The seemingly chaotic nature of climate time series can often be decomposed into a handful of discrete variables, accounting for a large percentage of its variance (Vautard and Ghil, 1989). Multiple methods were applied in this study, following the precautions presented by Dommenget and Latif (2002) as it pertained to empirical orthogonal function (EOF) analysis; however, the general precautions are widely applicable to any mathematical signal-noise separation procedure. In this study, we used two techniques for a first order, exploratory analysis of the UCRB, to place the UCRB in a broader geographic context. Correlation fields between runoff from major upper basin tributaries, and their corresponding precipitation at specific USGS gage sites were constructed.

Correlations are based on the annual winter precipitation and its corresponding year’s winter and spring, snowmelt season runoff. The length of time represented by the correlation fields was limited by the length of time recorded by the USGS stream gage.

EOF analysis for the western U.S. winter precipitation from 1895-2006 based upon the

PRISM data. Computational limitations required us to reduce our data set. We selected 57

to degrade the spatial resolution as opposed to the spatial domain because of the high

level of domain dependence of EOF analysis. Preserving the spatial domain is likely

more critical to capturing the physical mechanisms responsible for precipitation

variability (Richman, 1986), so the data were sub-sampled for each 7th column and row pixel making the analyzed resolution of the grid roughly 28 km2 (17.5 minutes). To ensure that our selected sampling rate still captured the large-scale EOF patterns, we conducted several basic sensitivity analyses on a 111-year by 200x200 pixel matrix at full resolution, centered at various locations in the western U.S. We sub-sampled the data by iteratively degrading the spatial resolution by increasing the sampling rate by one. EOF structures were retained until the resolution was degraded to 20x20 pixels (a sampling rate of every 10th pixel). Calculations of the EOFs were based on singular value decomposition (SVD) of the correlation matrix, as opposed to a covariance matrix. The covariance matrix avoids the differential weighting based on elevation—a dominant influence on spatial precipitation variability, and therefore overshadows the larger-scale structures. A red noise (AR1) Monte-Carlo filter was applied to determine the significant

EOF, based on the principal component (PC) spectrum. To localize EOF patterns we used a Varimax rotation (Kaiser, 1958), a technique that relaxes the orthogonal spatial 58

constraints imposed by traditional EOF. A weakness of traditional EOF analysis is that the modes of variability are always orthogonal in space, an inconsistent assumption in nature (Dommenget and Latif, 2002). Rotating the EOFs allows them to be data-directed, providing a more realistic means of interpretation (Dommenget and Latif, 2001;

Richman, 1986). Varimax rotation also corrects for Buell pattern (domain dependent patterns) development of the un-rotated EOF patterns.

For a more detailed analysis of the DRB, We applied singular spectrum analysis (SSA) and wavelet analysis to extract the signals that represent the dominant modes of precipitation and streamflow variability. These modes were compared with 500mb pressure fields to understand their far-reaching relationships with atmospheric patterns using the NOAA Earth System Research Laboratory’s climate reanalysis toolbox. We used SSA, following the criteria described by Ghil et al, (2002). To solve the SSA algorithm, We used SVD on the standard covariance matrix. An advantage of SSA over more traditional methods, such as Fourier analysis, is that SSA is data adaptive, and it captures variability in both phase modulation and amplitude of a time series (Allen and

Smith, 1996) 59

Wavelet analysis was used to identify the dominant temporal modes of precipitation variability following the guidelines of Torrence and Compo (1998). Application of wavelets to noisy climate time series is ideal because the technique is less scale- dependent than Fourier analysis, or even SSA. Wavelet analysis provides information, not only about the dominant modes of variability, but also how those modes change in time (Torrence and Compo, 1998), making it more skilled in characterizing nonstationarity in a time series. From this, we can determine the wavelet power of the variables (e.g. the normalized variance of precipitation or streamflow) at any frequency within a preselected time domain (i.e. 2—64 years). Additionally, the time domain in which specific frequencies are dominant is reported because the wavelet transform places the times series into time-frequency domain. This study utilized the complex, non- orthogonal, Morlet mother wavelet function. A complex function is better for capturing oscillatory behavior in a time series because it provides information about the amplitude and phase (Torrence and Compo, 1998). The Wavelet transform assumes time series data is cyclic, so zeros were used to pad the ends of the time series to reduce edge effects. To 60

separate the signal from red noise, a Monte-Carlo filter was used to determine the 95% significance levels in the results.

We isolated spring runoff floods from the full period of record for the daily time series.

We calculated standardized scores (z-scores) based on daily stream gage records for the period of record. Using z-scores allows us to isolate large events relative to their seasonality, thereby removing the seasonal autocorrelation inherent in climate time series.

While the absolute flood magnitude is typically what informs us of the actual geomorphic work potential (e.g. Baker 1994; O’Connor et al., 1986), the standardized values make possible the determination anomalous flow conditions relative to the seasonal means--an important aspect to ecological health and climatic variability as well as being of critical concern to water managers. This is especially important in snowmelt-dominated streams, typical of the intermountain west because the z-scores account for and effectively filter the mean trends in snowmelt. By detrending the seasonal snowmelt, we are left only with the anomalous meteorological induced runoff.

5. RESULTS 61

The broader UCRB

Spatially contextualizing the DRB within the greater UCRB is an important aspect of this study because it highlights the distinctive regions whose principal modes of variability are poorly realized. This highlights the DRB’s position within the complicated transition zone described by Wise (2010). Figure 3 shows the correlation fields between precipitation and streamflow. These correlation patterns illustrate the patterns of precipitation that are associated with streamflow in the UCRB. The Dolores River is well correlated (R>0.6) with precipitation on much of the Colorado Plateau, extending as far south as the Mogollon Rim and to the west as well into Southern California. Other rivers within the UCRB are shown as well, to highlight their contributing precipitation domains.

When disaggregating the time series into El Niño years and La Niña years individually, the coherent spatial patterns of correlation break down and became muddled, with well- correlated regions showing up as localized regions to the NW as well as SW (Figure 4).

Further illustrating ENSO as a weak precipitation delivery modulator, the precipitation anomalies based upon ENSO years are shown in Figure 5. The figure shows the weak precipitation anomalies across the UCRB and more specifically, the DRB. The ENSO years are broken down by strength as well, showing the intensification of the winter 62

precipitation dipole on ENSO strength, which has very little effect on the UCRB. During both El Niño, as well as La Niña years, there is a very weak increase in the precipitation delivery to the DRB.

EOF analysis returned four significant modes of variability for the UCRB; the first two were rotated for analysis. The rotated EOF (REOF) patterns are shown in Figure 6. The coherent spatial patterns of the rotated EOFs show loadings that are consistent with past work—they display typical dipole precipitation patterns between the Pacific Northwest and Southwest, U.S., (e.g. Cayan et al., 1999; Cannon et al., 2007; Woodhouse et al.,

2009). Their corresponding principal components (PC) explain a combined, 42% of the precipitation variance (PC1=24% and PC2=18%), although the loading patterns within the UCRB are limited. One exception, however, is the spatial loading of REOF 1. The pattern indicates that it may impact precipitation in the part of the UCRB that includes the western periphery of DRB. The principal component associated with REOF1 (PC1) suggests the mode of variability is multidecadal, and is shown in Figure 7, superimposed over the PDO time series for comparison.

The Dolores River Basin

SSA successfully extracted a low frequency mode of precipitation variability. The low frequency mode was also the only mode of variability that held up as statistically significant (based on a Monte Carlo red noise filtering routine). This mode is shown as the reconstructed component (RC1), which results from the convolution of the first PC and it’s corresponding eigenvector. RC1 is shown in Figure 8, superimposed over the precipitation time series. It broadly captures the variably trending pattern associated with the Dolores Basin’s precipitation regime, explaining 15% of the precipitation’s variance.

Shifts in the mode are roughly coincident with shifts in PDO, but it poorly captures the major PDO shift in the late 1940s.

Wavelet analysis was performed on the same representative precipitation grid cell in the

Dolores Basin as SSA. Figure 9 shows the power spectra for the precipitation time series.

Relatively high wavelet power is shown in the 3-8 year scale during the late 1930s through the later 1940s, and again from the early 1970s to the middle 1980s. Not as apparent, is a region in the 16-year scale from the early 1930s through the late 1940s, and 64

a 32-year scale from roughly 1930-1980, although its interpretation is difficult due to its

proximity to the cone of influence (the region where edge-effects corrupt the results).

In addition to performing wavelet analysis on precipitation data, we used the technique

on the Dolores paleo-streamflow reconstruction provided by Woodhouse et al., (2006).

Results in Figure 10 show greater variability in the 3—8 year scale from the late 1600s to

early 1800s. Additionally, a 16—32 year scale is evident from the early—mid 1700s, through to the late 1800s. Wavelet analysis was also performed on the peak annual flood time series for the Dolores River at Dolores, CO (Figure 11). The period from 1920s through the late 1950s experienced a greater degree of variance in the 1-5 year scale domain. The particularly high spectral power in the early 1920s is in part, explained as an artifact of the missing data prior to that period, although observations from other gages in the region suggest it in not an artifact. During the early 1920’s a large region of the

UCRB experienced numerous large floods. During the late spring of 1921, widespread flooding occurred across much of the state of Colorado following rain-on-snow events.

At several sites, including the Colorado River at Lees Ferry, the peak instrumental flood was recorded in association with these rain-on-snow events. 65

6. DISCUSSION

The multiple analysis approach proved necessary to understand the mechanisms that are responsible for modulating precipitation and streamflow. Each technique has its own merits, although as stand-alone techniques, they leave an incomplete realization of the hydroclimatic variability of the Dolores River Basin. The variability of the Dolores

Basin’s mean winter precipitation (Figure 12), is highlighted by two major transitions: one in the early to mid 1930s, the other shift, a more abrupt shift, in the late 1970s which reflect major shifts in the dominant modes of variability. As previous studies have shown, much of the precipitation in the western U.S. is impacted by ENSO variability, although the UCRB is largely an exception, and the focus of this study, the DRB, lies in a regions defined as the “transition zone” (Wise, 2010), the hinge point between the two regions of opposing hydroclimatic association. The fact that the DRB lies within he transition zones makes it a fundamentally complex hydroclimatic basin.

The coherent spatial patterns of precipitation–streamflow correlations provide a generalized sense of the dominant storm-tracks for precipitation delivery to their respective basins. When the El Niño and La Niña years are separated, the patterns 66

quickly lose their coherency, indicating the low level of confidence associated with using

ENSO as a hydroclimatic predictor for the DRB. The very weak increase in precipitation delivery during the ENSO anomaly years also provides support that the DRB lies within the transition zone.

The EOF analysis of the western U.S. provides a context to investigate the UCRB, and

DRB proper. A critical aspect of placing the Dolores Basin in a regional hydroclimatic context is the basin’s rather weak relationship to significant REOF patterns associated with precipitation in the western U.S; however, because of the broad spatial scale of this the EOF analysis, the leading PCs should be interpreted very cautiously at the fine scale.

They are difficult to interpret because they represent a very broad region, and therefore, may represent multiple, uncorrelated modes of variability across space (e.g. Dommenget and Latif, 2001). The only relevant loading pattern is that of REOF1, which shows a peripheral loading pattern near the DRB (Figure 6a). This highlights the fact that the majority of the dominant modes of hydroclimatic variability in the western U.S. are not associated precipitation in much of the UCRB. This is an expected outcome of this analysis, given the results and conclusions of prior researchers (e.g. Cayan et al, 1999; 67

Kim et al., 2005; Redmond and Koch, 1991; Zhang et al., 1997; Cayan, 1999; Timelsena et al., 2009; Hidalgo and Dracup, 2003; Cannon et al., 2007; Wise, 2010). SSA allowed us to constrain our analysis to the DRB. RC1, the first mode of variability for SSA, captures the low frequency trending of the DRB rather well, explaining 15% of the annual winter precipitation variance (Figure 8). The smoothed RC1 timeseries loosely resembles the smoothed PDO time series (timing and magnitude). The first PC from the

EOF analysis also resembles the PDO shifts in a similar regard to RC1, although it more closely captures this (Figure 7). This suggests the DRB may in fact be dependent upon north Pacific variability. An additional line of evidence to support interpreting RC1 as a

North Pacific mode is its similarity to the first PC derived from tree rings in the Mojave

Desert, which has been interpreted as the PDO, and was therefore used to reconstruct the

PDO to AD 1661 (Biondi et al., 2001). To further test the presence of a teleconnection between the DRB’s hydroclimate and the PDO, we correlated RC1 with both SSTs, and also 500mb pressure patterns using NOAA’s Earth Systems Research Laboratory (ESRL)

Reanalysis Project tools. The results show similar trends between RC1 and PDO

(temporally and spatially), which provides support that RC1 may loosely represent variability associated with the PDO (Figure 12). 68

Wavelet spectral power from the annual winter precipitation on the Dolores River

indicates that periods of high spectral power in the 3-8 year scale are coincident with abrupt transitions in the PDO. In particular, the PDO switching during the late 1970s to early 1980s, and during the late 1940s to early 1950s are well highlighted in Figure 9.

This suggests that PDO switchings are associated with increased streamflow variance for the several years that surround the PDO transition. Prior to the early 1920s, the PDO exhibited more variable, or unsettled behavior between positive and negative PDO phases, which was coincident with a period of more variable winter runoff. The time frame was defined by numerous years of low amplitude and higher frequency variability, similar to the PDO.

The PDO and most other low-frequency modes of variability pose limitations for analysis because of their limited degrees of freedom. Nevertheless, similarities between the timeseries from various analyses in this study, as well as the similarity with other studies results (e.g. Biondi et al., 2001) suggest that a North Pacific multidecadal component may in fact be responsible for the low frequency modulation of the winter precipitation 69

on the Dolores River. Furthermore, wavelet analysis of the DRB’s paleostreamflow

reconstruction (i.e. Woodhouse et al., 2006) provides additional insights to this tentative

conclusion. High wavelet power in the in the 3-8 year scale is evident for extended periods from the late 1500s until the end of the 1800s. In particular, notable periods of enhanced 3-8 year scale occurred in the earlier half of the 1700s, and much of the 1800s.

A lower frequency, 16-32 year scale is notable from the early—mid 1700s, through to the late 1800s. When these are compared with the PDO reconstruction of Biondi et al.,

(2001), certain periods correspond in a rough sense. The heightened 5-8 year scale centered in 1725 is actually coincident with a relatively minor switch in the PDO reconstruction at that time. A major shift of the PDO near 1750, is however coincident a heightened spectral power in the 16-32 year scale. Two other notable times occurred in the middle 1800s, when a significant switch of the PDO occurred near 1850. The wavelet power at that time shows a notable increased for both the 16-year period, as well as the 3-8 year period.

Occurrence of large floods 70

The Dolores River is experiencing a hydroclimatic period in which extreme flooding,

relative to the floods whose evidence has been preserved in the sedimentological record,

is apparently suppressed, and this has eliminated the possibility of an explicit flood-

hydroclimatic study. Instead, we have characterized the basin’s flood hydroclimate for

moderate sized floods. Paleoflood evidence in the DRB suggests that numerous floods

have occurred since the mid Holocene that are at least four times the magnitude of the

largest historical flood (estimated 285 m3s-1 on Oct. 5th, 1911), which occurred during the fall season resulting from a dissipating tropical cyclone. This places that event outside of the context of this study. It should be noted, however, the same storm led to extreme flooding in the San Juan River and produced a discharge that was more than double that largest historical flood (Webb et al., 2001).

The correlation between peak annual floods and the annual winter runoff is strongly positive (R=0.83)—not surprising given the snowmelt dominated runoff regime. The results are shown in Figure 13 as standardized discharge values. With few exceptions

(e.g. 1949), the standardized scores for the spring runoff and peak floods share similar values. The general lack of very large floods does not allow for a flood-hydroclimate 71

study (Hirschboeck, 1988), so we are not able to isolate the synoptic mechanisms responsible for truly extreme flood events. Although, the timing of moderate floods can be placed in the context of global circulation indices and expand our knowledge about the range of possibilities, thus providing a means to evaluate the climate conditions responsible for large and extreme flood potential. A brief visual inspection, as well as a correlation analysis between peak annual flows and the ENSO BEST index was performed, and the results indicate a lack of relationship between the two, although in extreme cases, a possible linkage exists but is very inconsistent. For example, 1941 and

1983, both strong El Niño years saw significant flooding, although other notable floods occurred during La Niña or ENSO neutral years, suggesting that historical flooding is poorly connected to ENSO. The relationship between PDO and flooding, however, may be more significant. A visual comparison seems to identify a trend in large flood occurrences. During positive (negative) phase of the PDO, there seems to be a higher proportion of large (small) floods. From the PDO phase shift in the early 1920s, a notable increase in greater-than-average floods occurred until the next phase shift, in the mid-

1940s. The second largest winter flood (1941) occurred when the PDO index peaked, prior to the shift in the PDO toward the negative. As a counter point to the positive 72

relationship between PDO and large floods, the 1949 flood, which was the largest winter flood, occurred when the PDO was negative. In fact, the PDO index was nearly at its lowest point when this flood occurred. Possibly of greater significance than the flood-

PDO phase relationship is the association that floods have with the PDO when the PDO is less settled. For example, the large 1921 flood, which was concurrent with flooding across much of the UCRB, occurred during a time when the PDO was exhibiting low- amplitude, high-frequency variability between positive and negative phases. The large floods that occurred in 1941 and 1949 are associated with near-peaks in the PDO, although the peaks were strongly positive and strongly negative (respectively).

Furthermore, each of these floods occurred within only a few years of the mid-1940s shift in the PDO.

Wavelet analysis on the annual flood series at Dolores, CO shows that there is variable periodicity, and anomalous streamflow. Similar to other analyses in this study, these are loosely associated with shifts in the PDO. Wavelet power shows that high variance is present in the 3-7 year scale surrounding PDO switch in the mid 1920s. The period in the

1940s indicates a time of high flood variance, with a scale of roughly 2-5 years. The 73

higher flood variance extends into the 1950s, when the PDO reached its minimum index.

A third period of high flood variance is evident in the mid-late 1970s, when the PDO transitioned from a negative to positive phase. As a final note on the flood time series and its relationship to the PDO, the two largest winter floods (1941 and 1949) occurred during two extremes (+, -) of the PDO.

It is well understood that the relationship between the hydroclimate of the UCRB and major climate indices is very complex, especially in the peripheral regions of the UCRB, such as the Dolores Basin. Transitions in the PDO may at least in part, help to explain variance in the streamflow, precipitation and flooding. Furthermore, it is important to note that periods of heightened variance in the streamflow, as well as peak annual floods are associated with larger magnitude floods, a result consistent with other studies (e.g.

Webb and Betancourt, 1992). This suggest that the PDO is a modulating influence on precipitation and streamflow, and the transitional, or unstable periods are introducing a condition that may aiding the tracking storms and/or permitting increased rain-on-snow events during the spring snowmelt season. The PDO has already been demonstrated as an important modulating factor with ENSO and its spatial hydroclimatic extent (e.g. 74

Wise, 2010), although in the DRB, the lower frequency PDO seems to be the dominant mode of precipitation and streamflow variability.

7. CONCLUSIONS

Streamflow and precipitation in the Dolores River Basin show a weak but notable response to a low-frequency interdecadal mode of variability, which possibly represents the PDO. The relationship is not evident in the strict sense of the phase condition of the

PDO, but rather the variability of the PDO: winter streamflow, winter precipitation and peak-annual flood all indicate that their variability increases over the several years that surround PDO phase transitions. Furthermore, periods of increased flooding seem to occur during times of heightened variability in the peak annual floods. No significant high-frequency modes were identified in this study, which is consistent with past studies of the Colorado River Basin (e.g. Cayan et al., 1999). This study was not successful in identifying any specific mechanisms or high-frequency modes of variability that might directly influence the storm tracks responsible for delivering precipitation to the DRB; however, future modeling studies may help to assess the degree to which this occurs.

Factors associated with the warming of the climate, especially in vulnerable regions of the world (e.g. Seager et al, 2007) are continuing to complicate water resource planning and flood mitigation measures. Earlier season snowmelts as a result of general warming

(e.g. Cayan et al., 2001) or changes in the snow surface albedo due to increased dust deposition on snowpack (e.g. Painter et al., 2007) are all contributing to complex factors that are affecting water resource management. Nevertheless, this study highlights the

increasingly important, nonstationary patterns in precipitation, streamflow and flooding

associated with changes in the North Pacific Ocean. While these patterns are likely to

transform as the global climate change, this study may provide a framework to assess

how the UCRB’s hydroclimate will respond to global climate change. 76

REFERENCES CITED

Adams, D.K. and Comrie, A.C., 1997. The North American monsoon. Bulletin of the American Meteorological Society, 78(10): 2197-2213.

Alexander, M.A. et al., 2002. The atmospheric bridge: The influence of ENSO teleconnections on air-sea interaction over the global oceans. Journal of Climate, 15(16): 2205-2231.

Alexander, M.A., Deser, C. and Timlin, M.S., 1999. The reemergence of SST anomalies in the North Pacific Ocean. Journal of Climate, 12(8): 2419-2433.

Allen, M.R. and Smith, L.A., 1996. Monte Carlo SSA: Detecting irregular oscillations in the presence of colored noise. Journal of Climate, 9(12): 3373-3404.

Baker, V.R., 1994. Geomorphological understanding of floods: Geomorphology. Geomorphology, 10(1-4): 139-156.

Balling, R.C. and Goodrich, G.B., 2007. Analysis of drought determinants for the Colorado River basin. Climatic Change, 82(1-2): 179-194.

Biondi, F., Gershunov, A. and Cayan, D.R., 2001. North Pacific decadal climate variability since 1661. Journal of Climate, 14(1): 5-10.

Board, C.W.C., 2006. Dolores/San Juan/San Miguel Basin Fact Sheet, pp. 2.

Brown, D.P. and Comrie, A.C., 2004. A winter precipitation 'dipole' in the western United States associated with multidecadal ENSO variability. Geophysical Research Letters, 31(9).

Bryson, R.A. and Hare, F.K., 1974. The Climates of North America. Climates of North America, World Survey of Climatology II. Elsevier, New York.

Canon, J., Gonzalez, J. and Valdes, J., 2007. Precipitation in the Colorado River Basin 77

and its low frequency associations with PDO and ENSO signals. Journal of Hydrology, 333(2-4): 252-264.

Cayan, D.R., Redmond, K.T. and Riddle, L.G., 1999. ENSO and hydrologic extremes in the western United States (vol 12, pg 2881, 1999). Journal of Climate, 12(12): 3516-3516.

Christensen, N.S., Wood, A.W., Voisin, N., Lettenmaier, D.P. and Palmer, R.N., 2004. The effects of climate change on the hydrology and water resources of the Colorado River basin. Climatic Change, 62(1-3): 337-363.

Cook, E.R., Woodhouse, C.A., Eakin, C.M., Meko, D.M. and Stahle, D.W., 2004. Long- term aridity changes in the western United States. Science, 306(5698): 1015- 1018.

D.R., C. and J.O., R., 1984 Local relationships between United States west coast precipitation and monthly mean circulation patterns. Mon. Weather Rev, 112: 1276–1282. .

Daly, C., 2006. Guidelines for assessing the suitability of spatial climate data sets. International Journal of Climatology, 26(6): 707-721.

Daly, C., Gibson, W.P., Taylor, G.H., Johnson, G.L. and Pasteris, P., 2002. A knowledge-based approach to the statistical mapping of climate. Climate Research, 22(2): 99-113.

Daly, C. et al., 2008. Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. International Journal of Climatology, 28(15): 2031-2064.

Daly, C., Neilson, R.P. and Phillips, D.L., 1994. A statistical topographic model for mapping climatological precipitation over mountainous terrain. Journal of Applied Meteorology, 33(2): 140-158.

Dommenget, D. and Latif, M., 2002. A cautionary note on the interpretation of EOFs. Journal of Climate, 15(2): 216-225. 78

Ellis, A.W., Goodrich, G.B. and Garfin, G.M., 2010. A hydroclimatic index for examining patterns of drought in the Colorado River Basin. International Journal of Climatology, 30(2): 236-255.

Ely, L.L., 1997. Response of extreme floods in the southwestern United States to climatic variations in the late Holocene. Geomorphology, 19(3-4): 175-201.

Ely, L.L., Enzel, Y., Baker, V.R. and Cayan, D.R., 1993. A 5000-year record of extreme floods and climate change in the southwestern United-States. Science, 262(5132): 410-412.

Ely, L.L., Enzel, Y. and Cayan, D.R., 1994. Anomalous North Pacific atmospheric circulation and large winter floods in the southwestern United-States. Journal of Climate, 7(6): 977-987.

Engineer, C.O.o.t.S., 1912. Biennial Report of the Engineer State of Colorado for the years, 1911-192. Office of the State Engineer, Denver.

Friedman, J.M. et al., 2006. Transverse and longitudinal variation in woody riparian vegetation along a montane river. Western North American Naturalist, 66(1): 78- 91.

Fye, F.K., Stahle, D.W. and Cook, E.R., 2003. Paleoclimatic analogs to twentieth-century moisture regimes across the United States. Bulletin of the American Meteorological Society, 84(7): 901.

Gershunov, A. and Barnett, T.P., 1998. ENSO influence on intraseasonal extreme rainfall and temperature frequencies in the contiguous United States: Observations and model results. Journal of Climate, 11(7): 1575-1586.

Ghil, M. et al., 2002. Advanced spectral methods for climatic time series. Reviews of Geophysics, 40(1).

Goodrich, G.B., 2007. Influence of the Pacific decadal oscillation on winter precipitation and drought during years of neutral ENSO in the western United States. Weather and Forecasting, 22(1): 116-124. 79

Grissino-Mayer, H.D., Romme, W.H., Floyd, M.L. and Hanna, D.D., 2004. Climatic and human influences on fire regimes of the southern San Juan Mountains, Colorado, USA. Ecology, 85(6): 1708-1724.

Hansen, E.M, and Shwarz, F.K., 1981. Meteorology of important rainstorms in the Colorado River and Great Basin drainages: National Weather Service Hydrometeorological Report No. 50, 167 p.

Hare, S.R., 1996. Low Frequency Climate Variability and Salmon Production, Unpublished Ph.D. dissertation, University of Washington.

Hereford, R. and Webb, R.H., 1992. Historic Variation of warm-season rainfall, southern Colorado Plateau, southwestern, USA. Climatic Change, 22(3): 239-256.

Hereford, R., Webb, R.H. and Graham, S., 2002. Precipitation History of the Colorado Plateau Region, 1900–2000. In: U.S.G. Survey (Editor), Reston.

Hidalgo, H.G. and Dracup, J.A., 2003. ENSO and PDO effects on hydroclimatic variations of the Upper Colorado River basin. Journal of Hydrometeorology, 4(1): 5-23.

Hirschboeck, K. (Editor), 1988. Flood hydroclimatology. Flood Geomorphology. John Wiley & Sons, 27‑49 pp.

Hirschboeck, K.K. (Editor), 1987. Catastrophic flooding and atmospheric circulation anomalies. Catastrophic Flooding. Allen & Unwin, 23‑56 pp.

Hirschboeck, K.K., 2003. Floods, Paleofloods, and Drought: Insights from the Upper Tails, CLIVAR/PAGES/IPCC Drought Workshop. University of Arizona.

Hirschboeck, K.K., 2009. Future Hydroclimatology and the Research Challenges of a Post-Stationary World. 142: 4-9. 80

Hirschboeck, K.K., Ely, L.a.M. and R.A. (Editors), 2000. Hydroclimatology of meteorologic floods. Inland Flood Hazards: Human, Riparian and Aquatic Communities. Cambridge University Press, p. 39-72 pp.

Keen, R.A. (Editor), Weather and Climate. The Western San Juan Mountains: Their Geology, Ecology and Human History. University of Colorado Press., Boulder.

Knox, J.C., 1993. Large increases in flood magnitude in response to modest changes in climate. Nature, 361(6411): 430-432.

Knox, J.C., 2000. Sensitivity of modern and Holocene floods to climate change. Quaternary Science Reviews, 19(1-5): 439-457.

Knox, J.C. and Kundzewicz, Z.W., 1997. Extreme hydrological events, palaeo- information and climate change. Hydrological Sciences Journal-Journal Des Sciences Hydrologiques, 42(5): 765-779.

Lau, N.C. and Nath, M.J., 1994. A modeling study of the relative roles of tropical and extratropical SST anomalies in the variability of the global atmosphere ocean system. Journal of Climate, 7(8): 1184-1207.

Mantua, N.J. and Hare, S.R., 2002. The Pacific decadal oscillation. Journal of Oceanography, 58(1): 35-44.

Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M. and Francis, R.C., 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society, 78(6): 1069-1079. 81

Matter, M.A., Garcia, L.A., Fontane, D.G. and Bledsoe, B., 2010. Characterizing hydroclimatic variability in tributaries of the Upper Colorado River Basin- WY1911-2001. Journal of Hydrology, 380(3-4): 260-276.

McGinnis, D.L., 2000. Synoptic controls on upper Colorado River basin snowfall. International Journal of Climatology, 20(2): 131-149.

Meko, D.M. et al., 2007. Medieval drought in the upper Colorado River Basin. Geophysical Research Letters, 34(10).

Milly, P.C.D. et al., 2008. Climate change - Stationarity is dead: Whither water management? Science, 319(5863): 573-574.

Milly, P.C.D., Wetherald, R.T., Dunne, K.A. and Delworth, T.L., 2002. Increasing risk of great floods in a changing climate. Nature, 415(6871): 514-517.

Minobe, S., 1997. A 50-70 year climatic oscillation over the North Pacific and North America. Geophysical Research Letters, 24(6): 683-686.

Mock, C.J., 1996. Climatic controls and spatial variations of precipitation in the western United States. Journal of Climate, 9(5): 1111-1125.

Newman, M., Compo, G.P. and Alexander, M.A., 2003. ENSO-forced variability of the Pacific decadal oscillation. Journal of Climate, 16(23): 3853-3857. 82

O'Connor, J.E., 1985. Paleohydrology of pool and riffle pattern development: Boulder Creek, Utah, pp. 70 Geological Society of America Bulletin. (97)4: 410-420.

O'Connor, J.E. et al., 1994. A 4500-year record of large floods on the Colorado River in the Grand Canyon, Arizona. Journal of Geology, 102(1): 1-9.

Quadrelli, R. and Wallace, J.M., 2004. A simplified linear framework for interpreting patterns of Northern Hemisphere wintertime climate variability. Journal of Climate, 17(19): 3728-3744.

Richman, M.B., 1986. Rotation of Principal Components. Journal of Climatology, 6(3): 293-335.

Rodionov, S.N., Bond, N.A. and Overland, J.E., 2007. The Aleutian Low, storm tracks, and winter climate variability in the Bering Sea. Deep-Sea Research Part Ii- Topical Studies in Oceanography, 54(23-26): 2560-2577.

Seager, R., 2010. Adjustment of the atmospheric circulation to tropical Pacific SST anomalies: Variability of transient eddy propagation in the Pacific-North America sector. In: N.H. Naik, T.M. F, C.M. A, H. N and K. Y (Editors), Quarterly Journal of the Royal Meteorological Society.

Seager, R. et al., 2005. Mechanisms of ENSO-forcing of hemispherically symmetric precipitation variability. Quarterly Journal of the Royal Meteorological Society, 131(608): 1501-1527.

Seager, R. et al., 2007. Model projections of an imminent transition to a more arid climate in southwestern North America. Science, 316(5828): 1181-1184.

Sheppard, P.R., Comrie, A.C., Packin, G.D., Angersbach, K. and Hughes, M.K., 2002. The climate of the US Southwest. Climate Research, 21(3): 219-238.

Smith, C.A. and Sardeshmukh, P.D., 2000. The effect of ENSO on the intraseasonal 83

variance of surface temperatures in winter. International Journal of Climatology, 20(13): 1543-1557.

Stahle, D.W. et al., 1998. Experimental dendroclimatic reconstruction of the Southern Oscillation. Bulletin of the American Meteorological Society, 79(10): 2137-2152.

Timilsena, J., Piechota, T., Tootle, G. and Singh, A., 2009. Associations of interdecadal/interannual climate variability and long-term Colorado River Basin streamflow. Journal of Hydrology, 365(3-4): 289-301.

Torrence, C. and Compo, G.P., 1998. A practical guide to wavelet analysis. Bulletin of the American Meteorological Society, 79(1): 61-78.

Trenberth, K.E. and Hurrell, J.W., 1994. Decadal atmosphere-ocean variation in the Pacific:. Climate Dynamics, 9(6): 303-319.

Vautard, R. and Ghil, M., 1989. Singular Spectrum Analysis in nonlinear dynamics, with application to paleoclimatic time-series: Physica D, 35(3): 395-424.

Vecchi, G.A. et al., 2006. Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature, 441(7089): 73-76.

Weare, B.C. and Hoeschele, M.A., 1983. Specification of monthly precipitation in the western United-States from monthly mean circulation. Journal of Climate and Applied Meteorology, 22(6): 1000-1007.

Webb, R.H. and Betancourt, J.L., 1992. Climatic variability and flood frequency of the Santa Cruz River, Pima County, Arizona. In: U.S.G.S. (Editor). U.S. Geological Survey Water-Supply Paper, Tucson, AZ.

Webb, R.H., Boyer, D.E., Orchard, K.L. and Baker, V.R., 2005. Changes in riparian vegetation in the southwestern United States: Floods and riparian vegetation on the San Juan River, southeastern Utah. U.S. Geological Survey Open-File 84

Report.

Wells, J.V.B., 1954. Compilation of records of surface waters of the United States through September 1950; Part 9, Colorado River Basin. In: U.S.G. Survey (Editor). U.S. Geological Survey Water-Supply Paper, Reston.

Westerling, A.L., Gershunov, A., Brown, T.J., Cayan, D.R. and Dettinger, M.D., 2003. Climate and wildfire in the western United States. Bulletin of the American Meteorological Society, 84(5): 595-+.

Wise, E.K., 2010. Spatiotemporal variability of the precipitation dipole transition zone in the western United States. Geophysical Research Letters, 37.

Woodhouse, C.A., 1997. Winter climate and atmospheric circulation patterns in the Sonoran Desert region, USA. International Journal of Climatology, 17(8): 859- 873.

Woodhouse, C.A., 2003. A 431-yr reconstruction of western Colorado snowpack from tree rings. Journal of Climate, 16(10): 1551-1561.

Woodhouse, C.A., Gray, S.T. and Meko, D.M., 2006. Updated streamflow reconstructions for the Upper Colorado River Basin. Water Resources Research, 42(5).

Woodhouse, C.A., Kunkel, K.E., Easterling, D.R. and Cook, E.R., 2005. The twentieth- century pluvial in the western United States. Geophysical Research Letters, 32(7).

Woodhouse, C.A., Russell, J.L. and Cook, E.R., 2009. Two Modes of North American Drought from Instrumental and Paleoclimatic Data. Journal of Climate, 22(16): 4336-4347.

Zhang, R. and Delworth, T.L., 2007. Impact of the Atlantic Multidecadal Oscillation on North Pacific climate variability. Geophysical Research Letters, 34(23).

Zhang, Y., Wallace, J.M. and Battisti, D.S., 1997. ENSO-like interdecadal variability: 1900-93. Journal of Climate, 10(5): 1004-1020.

Zhang, Y., Wallace, J.M. and Iwasaka, N., 1996. Is climate variability over the North Pacific a linear response to ENSO? Journal of Climate, 9(7): 1468-147.

UCRB

DRB

Figure 1. Location map of the Dolores River Basin (DRB) within the Upper Colorado River Basin (UCRB). o Elevation (m)

High : 4000

Low : 1200 0 gage sites o o flood deposits o

o o

Gage site used in this study o (Dolores River, Dolores, CO)

Figure 2. Location of the Dolores River Basin (DRB), USGS stream gages and SWD site locations 87 Dolores River Colorado River San Miguel River Dolores, CO Cisco, CO Placerville, CO

Yampa River Upper Colorado River Gunnison River Maybell, CO Dotsero, CO Gunnison, CO

Figure 3. Correlation fields between precipitation and streamflow for the full period of streamflow record. Locations of stream gages are shown by dark circles. Figure 4 88 Dolores River runnoff--PPT Dolores River runnoff--PPT El Nino years correlation La Nina years corrollation 1.0

0.6

-0.6

A -1.0 B Figure 4. Correlation fields between streamflow and precipitation at the Dolores River stream gage in Dolores, CO (shown by x) during El Niño (a) and La Niña years (b). 89 All La Nina years Strong La Nina years

2 x104

1.5

1.0

0.5

All El Nino years 0 Strong El Nino years -0.5

-1.0

-1.5

-2 x104 90 Figure 6 91

REOF1

-0.02low -0.01 0 0.01 0.02 high 0.03 Figure 6. (a) Rotated EOF 1 (REOF1) and (b) REOF2. DRB location shown by the dark ring. REOF1 shown moderate loading in the DRB region. PCI superimposed over PDO

1.5

0.5

-0.50

-1

-1.5 PDOPDO smoothed smoothed (8-yr) (8-year) PC1PC1 smoothed smoothed (8-yr)(8-year) andand scaledscaled -2 PDOPDO (annual (annual mean)

19001900 1910 19201920 1930 1940 1950 1960 1970 1980 1990 20002000 DateDate (yr)

FigureFigure 7. Smoothed8 PCI from the Western U.S. EOF analysis (red line) superimposed over the winter annual, and smoothed PDO (black bars and green line). 92 Dolores Basin winter precipitation (mean removed, 1895-2005) 4 st x 10 1 reconstructed component (15% variance explained) 3

ppt (0.1 mm) -1 ppt (0.1 mm)

-2

-3

RC1 -4 1880 1900 1920 1940 1960 1980 2000 2020 Figure 8. First reconstructed component (RC1) from SSA. RC1 (red line) is superimposed over the precipitationFigure 10 times series (blue line). This low frequency mode captures the variably trending mean of the DRB's winter precipitation. The mode expains 15% of the annual precipitaiton variance. 93 Dolores'RORUHV337 :LQWHUPRQWKV 1'-)0 Basin’s winter precipitation (NDJFM) 4 2 0 ï =éVFRUH Dolores PPT ï 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 'RORUHV337:DYHOHW3RZHU6SHFWUXP ppt wavelet power spectrum

3HULRG \HDUV 32

64 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

ï ï ï 0 8 x 10 2é\U6FDOHé2-16 yr scale-averageaverage time Time series Series

3 é

1 $YJYDULDQFH

(0.1 mm yr 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Time (year) Figure 9. Wavelet analysis of the DRB's winter precipitation. Upper box shows the standardized precipitation time series.Figure Middle 11 box shows the wavelet power spectra. Lower box shows the average variance on a 2-16 year scale- average. Regions outside of the black line indicate regions effected by edge effects, and the solid black line represents areas that are statistically significant. 94 Reconstructed Dolores River streamflow

0 Z−score −2

1600 1650 1700 1750 1800 1850 1900 1950 paleostreamflow wavelet power spectrum

32 Period (years) 64 1600 1650 1700 1750 1800 1850 1900 1950

−16 −14 −12 −10 −8 −6 −4 −2 0

5 x 10 3-32 yr scale-average time series 4 ) 1 − y 2 (MAF 0 Avg variance 1600 1650 1700 1750 1800 1850 1900 1950 Time (year) Figure 10. Wavelet analysis of the reconstructed DRB's streamflow. Upper box shows the streamflow time series. Middle box shows the wavelet power spectra. Lower box shows the average variance on a 3-32 year scale-average. Regions outside of the black line indicate regions effected by edge effects, and the solid black line represents areas that are statistically significant. 95 96

a) Annual Dolores floods (spring runnoff only) 10000 1) −

s 8000 3 6000

4000

discharge (ft 2000

0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 b) Peak Annual Dolores flood Wavelet Power Spectrum

16 Period (years) 32

64 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Low Hi 6 x 10 d) 1−5 yr Scale−average Time Series

1) 5 − yr

3 4 3 2 1 Avg variance (ft 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Time (year) 97

RC1--500mb correlation field

PDO--500mb correlation field

Figure 17 Winter WInterrunoff runoff - annual - annual peakpeak flood flood relationship relationship (R=0.83)(R=0.83) 4

1 Z-Score Z-score 0

-1

-2

1900 1920 1940 1960 1980 2000 date (yr) Figure 13. Standardized peak annual floods shown as stems with their associated winter runoff shown as bars. TheFigure two are 18 well correlated (R=0.83) correlated in most cases, with the major exception in 1911, which was not a winter flood. 98 99

APPENDIX B

HOLOCENE PALEOFLOOD CHRONOLOGY BASED ON OPTICALLY STIMULATED LUMINESCENCE AND AMS 14C FOR THE DOLORES WATERSHED, CO AND UT, USA

Michael L. Cline

100

Abstract

The paleoflood chronology of the Dolores River Basin is one of the first paleoflood studies in a concerted effort to characterize the Upper Colorado River Basin’s response to climate variability throughout the Holocene. The Dolores River Basin’s 8200-year paleoflood chronology is based on accelerator mass spectrometry radiocarbon and optically stimulated luminescence. A large number of floods occurred well into the onset of the Medieval Climatic Anomaly (c.a. 900-1300 A.D.), and stratigraphic evidence for large floods is entirely absent during a period commonly thought to be wetter (c.a. 1300- present) indicating the regional flooding is mostly out-of-phase with those in the Lower

Colorado Basin. The floods documenting this study suggest the hydroclimatic setting associated with very large floods in the Upper Colorado River Basin is very different today, than it was during episodes of high magnitude flooding. 101

1. Introduction

The Lower Colorado River Basin (LCRB) has been the focus of many flood studies, and the comprehensive data these studies have provided have led to the identification of the hydroclimatic anomalies that are associated with extreme floods (e.g. Ely et al., 1993;

Ely, 1997). Modern analogs for hydroclimatic circulation have supported the findings of these studies (e.g. House and Hirschboeck, 1997), making the LCRB a well-characterized basin in the context of extreme flood hazards. The Upper Colorado River Basin (UCRB), in contrast, has a remarkably poorly characterized flood history, which is surprising given its significance as the single most important water resource in the southwestern U.S. This may be in part due to its complex hydroclimatology (see appendix A of this dissertation), which has made the region the focus of many hydroclimatic studies that attempt to characterize its response to large-scale circulation patterns and/or indices (e.g. Wise,

2010; McCabe and Dettinger, 1999, Hidalgo and Dracup, 2003, Cayan et al., 1999). A renewed interest in extreme floods in the UCRB has been brought forth by recent studies on the Upper Colorado River near Moab, UT (i.e. Greenbaum et al., 2006), where paleoflood stratigraphy has shown a minimum of 40 extreme floods have occurred since the late Holocene, several of which lie near the flood envelope maximum for the

Southwestern U.S. (Enzel et al., 1993). The preliminary data from the Upper Colorado

River site suggest that extreme floods that are significantly larger than those of the historical record may pose a substantial risk to major infrastructure including the Atlas

Uranium Mine tailings in Moab, UT, and ultimately, human life. The paleoflood chronology of the Dolores River Basin (DRB), presented in this study is the first, 102

comprehensive study of a major tributary within the UCRB, and provides a context for understanding extreme floods in regions that are not typically associated with major teleconnections such as the El Niño-Southern Oscillation (e.g. Cayan et al., 1999). This study may provide a temporal analog to other regional sub-basins within the Western

Slope region of Colorado. This study serves as a first step in a larger effort to characterize the flood hydrology of the UCRB.

2. Background

Fluvial response to climate variability remains a critical area of research in the earth sciences, especially given the potential intensification of the hydrological cycle—a probable result of increasing greenhouse gasses (Milly et al., 2002; Trenberth et al.,

2003). Climate models agree that hydrological changes will be paramount, especially in arid regions of the world (Seager, 2007); however, with few exceptions (e.g. Seager et al.,

2005) the model projections are limited to the mean conditions and do not include transient eddy propagation or other anomalous circulation patterns, the leading causes of large, rare floods (personal communication, Hirschboeck, 2010). For these reasons, they are not well suited for the prediction of extreme floods (Knox and Kundzewicz, 1997;

Knox et al., 2000). Efforts directed at inferring future flood behavior have historically relied on systematic (gaged) and historical records, which are then extrapolated to predict the frequency-magnitude relationships of extreme events (Baker, 2002). This method is fundamentally flawed however, because middle-late Holocene flood behavior is poorly characterized using the assumption that floods are an independent and identically 103

distributed (i.i.d.) phenomenon (e.g. Redmond et al., 2002). In reality, the probability density functions that define flood hazards are dynamically evolving, dependent upon circulation patterns from global to local scales (Hirschboeck, 1988), as well as the changing land use practices and surface conditions of a given watershed. The problem with the i.i.d. assumption is further exemplified in parts of the world where the lengths of gaged records rarely exceeds a single century, making them inadequate because temporally limited gage records rarely capture the occurrence of truly extreme floods.

Adding to the difficulty of constraining probabilistic flood hazards in nonstationary hydroclimates, anthropogenic greenhouse climate forcing is enhancing the hydrological cycle, which is likely to result in an increase in the potential for large floods (e.g. Milly et al., 2002). This is especially the case in arid regions of the world where it is probable that climate change and its impacts will be intensified (e.g. Seager et al., 2007; Milly et al., 2008). The potential increase in large floods due to anthropogenic forcing is particularly difficult to discern from natural forcing, however, due to the high variability in their natural occurrence (Milly et al. 2002). All of these issues have called for alternative methods, statistical and numerical treatments of climate and hydrological data that incorporate principles of nonstationarity (Milly et al., 2008). The sentiment in moving beyond stationarity is echoed by Hirschboeck (2009), who advises future workers to address specific themes in research to better handle nonstationarity. One of these themes is to better understand hydrological extremes (i.e. floods and droughts).

Paleohydrological data can provide a more realistic set of scenarios used by water- resource managers because it effectively extends the gaged record, thereby capturing a 104

greater range of circumstances that can better inform managers of the real risks for flood and drought magnitudes.

Several landmark studies have provided a framework to understand flood nonstationarity.

These studies have used historical records (e.g., Cayan and Webb, 1992; Hereford and

Webb, 1992; Hidalgo and Dracup, 2003, Wise, 2010, House and Hirschboeck, 1997;

Hirschboeck, 1988; Hirschboeck et al., 2000), as well as paleoflood and/or stratigraphic records (e.g. Knox, 1993; Ely et al., 1993; Ely, 1997; Hereford, 2002; House and

Hirschboeck, 1997) to develop conclusions about flood sensitivity to shifts in climate.

Most of these studies have focused on the LCRB however, an area that correlates strongly with the modern El Niño-Southern Oscillation (ENSO) signal (Cayan et al.,

1999; Kim et al., 2006), and in at least one case, differences in methodological approaches have led to competing hypotheses about the climatic conditions which led to flood deposition and erosion (i.e. Hereford, 2002; Ely, 1997), exemplifying the complexity in these interpretations. Regardless of the differences in interpretations, it is clear that flood regimes have shifted dramatically in the past, and will continue to do so in the future. This requires that we continually update and refine data and techniques that are applied to understanding the potentially catastrophic effects of improper water management practices (Milly et al., 2008).

The UCRB is the most important surface water resource in the southwestern U.S., so there have been extensive efforts aimed at reconstructing average annual stream flows. 105

Using dendrochronology, the streamflow record for the Upper Colorado River has been

extended to A.D. 762 (e.g. Stockton and Jacoby, 1976; Meko et al., 2007). Streamflow

reconstructions have proven invaluable for paleoclimate and paleohydrology

reconstructions, highlighting, for example, the numerous paleo-droughts that have

eclipsed historical droughts (Cook et al., 2004, Woodhouse, 2005) and the severe over-

allocation of the modern Colorado River (Stockton and Jacoby 1976; Cook et al., 2004).

Despite the powerful application of dendrochronology as an excellent proxy for average

annual streamflow, extreme flood events cannot be resolved using streamflow

reconstructions derived from dendrochronology, unless the floods in question occurred

during the lifetime of the living tree and the tree was either partially inundated or directly

impacted by flood waters (e.g. Sigafoos, 1961; Hupp, 1982; Yanosky, 1982; Yanosky

and Jarret, 2002). In fact, higher-than-average annual streamflow is not a prerequisite for

extreme flooding; it has been shown that there may be an equal probability of extreme

floods during droughts (Hirschboeck, 2000). For example, Webb and Betancourt (1992)

noted that early-mid 20th century flooding in the southwestern U.S. maintained a similar mean, although there were significant changes in the variance of streamflow associated with large floods. The variance shifts were a result of changes to the hemispheric and synoptic circulation patterns impacting the region. The weak correlation between wet conditions, and episodes of increased flooding highlight the need to explicitly incorporate paleoflood hydrology as a tool for water-resource policy and risk mitigation. Paleoflood hydrology—the study of ancient or unobserved floods (Kochel and Baker, 1982), provides critical information that can be used to expand our knowledge of flood hazards 106

by extending gaged records with extreme flood data (Stedinger and Cohn, 1986; Hosking and Wallace, 1986). It utilizes indexical marks left by floods on the landscape (Baker

1988a,b; Baker, 2006). Paleoflood studies have provided corroborative evidence to the historical and systematic records for flood sensitivity to climate shifts, and have helped develop our understanding of hydroclimatic variability (e.g. Webb and Betancourt, 1992,

Hereford, 2002; Ely, 1997; Ely et al., 1993; Hirschboeck, 1988; Knox, 1993; Knox, 2000;

House and Hirschboeck, 1997; Ely et al., 1994; Redmond et al., 2002). Without paleoflood studies, perceptions of flood hazards are limited to theoretical flood magnitude-frequency curves derived from relatively short, gaged historical records

(Baker, 2002). They often neglect the “upper-tails” of the distributions (Hirschboeck,

2003), which are critical, especially given the increasingly nonstationary world we are entering (Milly et al., 2008). Flood frequency analyses clearly improve when paleoflood data are included (e.g. Benito and Thorndycraft, 2005; Blainey and Webb, 2002) because the addition of paleodata downplays the i.i.d. assumption that is fundamental to flood-risk analysis. This is particularly important in the western U.S., where gaged and historical records rarely exceed one hundred-years, and consequently, large floods are statistically under-represented (Benito and Thorndycraft, 2005). In order to understand the risk associated with extreme floods, the relationship between maximum flood-discharge and drainage-area relationships (i.e. maximum flood envelope) are used. Paleoflood information is used to supplement these data with paleoflood discharge values that frequently exceed those of the modern record (e.g. Enzell et al., 1993). This illustrates 107

the need to include paleoflood records in conjunctions with systematic gaged streamflow records as a tool for water-resource policy and risk mitigation.

2.1 Paleostage indicators and discharge estimates: There are numerous types of paleostage indicators (PSIs), but most paleoflood studies rely on some form of a slackwater deposit (SWD) to derive information about paleofloods. SWDs provide information to researchers about the minimum paleo water surface elevation and they often contain dateable organic material. They form in back-flooded tributaries and other areas of the channel margins where the energy required to keep sediment in suspension is reduced below the critical grain-size-dependent velocity threshold (Baker, 1987; Kochel and Baker, 1982). Paleoflood research is ideally conducted in deeply incised, stable bedrock channels, where the channel geometry is stable over several thousand years. In mixed bedrock-alluvial reaches, there are more limitations on paleoflood hydrology because of the dynamic nature of the channel morphology through time; nevertheless, studies in alluvial reaches have been successfully performed on ancient floods (e.g.

Knox, 1993; Knox, 2000) as well as modern, un-gaged basins (e.g. Pelletier et al., 2005).

Most discharge estimates rely on one-dimensional (1D) models that use step-backwater routines or slope-area methods. Multi-dimensional hydrological models have been used as well (e.g. Denlinger et al., 2002; Pelletier et al., 2005), and if the channel geometry is well constrained by stable margins and high-resolution measurements, the accuracy of the paleoflood discharge estimates can be within the same margin of error as gage records

(Webb and Jarret, 2002, O’Connor and Webb, 1988). The reliability is also highly 108

dependent on the type of PSI used (Jarret and England, 2003; Webb and Jarret, 2002;

Webb et al., 2002). During an extreme flood on the Verde River, the use of SWDs underestimated the flood stage (as determined by the emplacement of flotsam) by as much as 30%, for example (House, 1997). Ideally, flotsam and silt lines are used, but when they are not present, SWDs can be used but are always recognized as minimum stage estimates (Webb and Jarret, 2002; House and Hirschboeck, 1997, House et al.,

2002, Webb et al., 2002).

2.2 Age control: Comparisons of contemporary large flooding with meteorological and climate conditions provides context for understanding the mechanisms responsible for very large paleofloods (e.g. House and Hirschboeck, 1997); however, if a river basin has not experienced extreme floods during modern times, our understanding of the major flood generating processes is limited. In the case in this study, the gaged record does not contain any floods that approach the stages of the stratigraphically-derived, paleoflood discharges. We are left only to utilize the analytically-derived ages of (paleo)floods and concurrent (paleo)climate record. Unfortunately, there are significant scale issues between the resolution of age-control for most paleofloods and high-resolution proxy- based climate records such as tree rings. For example: analytically-derived ages using radiocarbon, under the most ideal conditions, are typically only accurate within a century, making the identification of specific anomalous circulation anomalies that drive flooding impossible. Nevertheless, individual paleofloods, or even periods where flooding was more or less frequent may be understood through bracketing the relevant climatic periods 109

suggested through paleoclimate records (e.g. Ely et al., 1993, Ely, 1997). In particular, annually resolved dendrochronology records provide high-resolution stream flow or other pertinent data that can be analyzed and compared with the bracketed periods of large flood occurrences. Where high-resolution proxy data do not exist, or if paleoflood records exceeds the timescales of the high-resolution data, paleoflood chronologies can simply be superimposed with paleoclimate proxy records such as those derived from tree rings (e.g. fire chronologies and streamflow reconstruction), and lake levels (e.g. Ely et al., 1993, Ely, 1997). Even under the most ideal circumstances (e.g. small analytical range of error in flood ages combined with high resolution paleoclimate records), there is still considerable uncertainty associated with the analytical techniques commonly used to derive the ages for flood deposits. A companion to this study discusses some of these issues in detail (appendix C of this dissertation). For 14C age control, the best-case resolution for most pre-bomb 14C is ± fifty-years but it is typically closer to ± one hundred-years, depending on what part of the dendrochronology-based calibration curve ! ! they reside (i.e. Reimer et al., 2009). The use of optically stimulated luminescence

(OSL) in paleoflood studies remains a relatively new technique in need of continual refining, and the results have been often problematic (e.g. Porat, 2001). The application of OSL to paleoflood hydrology is a complex process, and numerous steps are necessary in avoiding pitfalls inherent to it application.

2.3 Southwestern U.S. paleohydrology research: The southwestern U.S. is generally well-suited for conducting research in paleoflood hydrology because of its numerous 110

deeply incised stable channel margins, bedrock lithology, and its arid to semiarid climate

which increases the preservation potential for PSIs and organic material for 14C dating.

Reviews of the paleoflood literature in the southwestern U.S. have highlighted the temporal clustering of floods during the last 5,000 years in response to climate variability

(Ely et al., 1993; Ely, 1997). These studies have been limited to the LCRB, however, and the majority of the sites utilized in that study are hydroclimatically linked to ENSO variability (e.g. Ely et al., 1993; Ely et al., 1997). Most of the rivers that were included in the study showed the existence of numerous paleofloods that have significantly exceeded the largest historical floods, particularly over the last (e.g. O'Connor et al., 1994; Enzel et al., 1994; Webb et al., 1988), highlighting the misrepresentation of floods by historically based, flood frequency curves. Other studies have illustrated the importance of extreme flooding as the dominant channel-forming mechanism (e.g. O'Connor et al.,

1986; Baker, 1988), suggesting that paleoflood studies can provide important insights to fluvial geomorphic and fluvial landscape evolution studies.

2.4 Anomalous atmospheric conditions: When meteorological conditions provide precipitation or snowmelt that exceeds the reservoir storage capacity of a basin, floods result (Hirschboeck, 2000; O’Connor, 2002). By developing an understanding of the causal mechanisms that lead to extreme floods, we are able to develop frameworks to understand flood regime changes relative to climate variability (e.g. Hirschboeck, 1988;

House and Hirschboeck, 1997). This information is critical to water managers and other concerned stakeholders because if the information is applied correctly, it can be used to 111

prevent the loss of life and property. Modern flood records have shown that flood

probability distributions are variable in time and space, dependent on the circulation

patterns from global to local scales (Hirschboek, 1988). These circulation patterns are

directly responsible for the moisture delivery mechanisms that cause extreme flooding

(Hirschboeck 1987; Hirschboeck, 1988; House and Hirschboeck, 1997). Based on this,

paleoflood and stratigraphic records have been interpreted in a climatic context; many

paleoflood studies highlight the clustering of floods in time (and space). Work in the

Mississippi Basin (Knox, 1993; Knox, 1997; Knox, 2000), the lower Colorado Basin (Ely

et al., 1993; Ely 1997; Webb and Betancourt, 1992), and the Colorado Plateau (Herford,

2002; Ely et al., 1993, Ely et al., 1997) have suggested possible links to ENSO, nested

within regional and global-scale climate shifts, such as the transition from the Medieval

Climate Anomaly to the Little Ice Age, (c.a. 1300). Modern records have demonstrated

similar regime changes (e.g. Hereford and Webb, 1992; Webb and Betancourt, 1992).

Appendix A of this dissertation discusses the hydroclimatic patterns associated with

streamflow and flood variability in the DRB.

3. Study site

The DRB, whose location is shown in Figure 1, drains 11,860 km2 in Southwestern

Colorado and southeastern Utah. The basin drains the southern Uncompahgre Plateau,

CO, the southwestern San Juan Mountains, CO, and the northern and eastern flank of the

La Sal Mountains, UT. The Dolores River reaches the Colorado River roughly 26 km

north of Moab, UT, where it empties at a mean rate of ~20 m3s-1: roughly ten-percent of 112

the Colorado River’s annual discharge at the rivers confluence. The watershed is

comprised of two major rivers: the San Miguel, and the Dolores rivers. Both of the rivers

head near Lizard Head Pass in the San Juan Mountains near Telluride, CO in

predominantly Cenozoic volcanic and sedimentary rocks in the upper reaches, and

Mesozoic sedimentary rocks in the lower reaches. The San Miguel River is one of the

few rivers in the western, U.S. whose main-stem is not impounded, whereas McPhee

Dam impounds the upper Dolores (construction from 1980-1985). The average annual

discharge for the Dolores and San Miguel rivers are nearly equal, each providing roughly

9-10 m3s-1.

Further downstream, the two rivers meander though steep-sided, stable bedrock canyons, superimposed into Mesozoic sedimentary rocks. The steep-sided canyons of the rivers and their many tributaries are littered with bedrock alcoves that create opportunities for slackwater deposition and preservation. Numerous tributaries and tight radius meanders provide additional potential for the deposition and subsequent preservation of slackwater deposits.

Regional to local-scale hydroclimatic variability is controlled at multiple scales, by physiographic, oceanic and atmospheric factors, each operating at different spatial and temporal scales (Mock, 1996). The steep and rugged topography in the semi-arid

Dolores River Basin (DRB) make orography a critical element in the spatial delineation of precipitation (Sheppard et al., 2002). Most of the winter precipitation in the DRB is 113

attributed to Pacific air masses (Bryson and Hare, 1974). The DRB is situated at the northernmost extent of the typical air masses that arrive from the topographic low feature in southern California and northern Mexico. Northern storms may also reach the DRB.

The synoptic mechanisms responsible for delivering the moisture are generally mid- latitude cyclones. These are relatively common features during the winter months, and depending on their source region, they bring the potential for significant precipitation.

The synoptic lifting associated with the lows are enhanced orographically, and can result in heavy precipitation for multiple days, especially if the air mass is cutoff or stalled

(Mock, 1996). This makes topography the dominant factor responsible for the spatial variability in winter precipitation (Sheppard et al., 2002). For example: snowfall differences between low and high elevations may differ up to an order of magnitude. The amount of precipitation is dependent on a number of factors, but primarily the storm tracking and storm source. Northern storms are steered by the spatial and strength relationship between the subtropical anticyclone off the coast of California, and the

Aleutian Low—a quasi-stationary statistical feature in the North Pacific, defined by the location of peak intensity of cyclonic intensity over a pre-defined period of time

(Rodionov et al., 2007). Storms that enter the southern San Juans from the south tend to produce greater amounts of precipitation than any other storms. These synoptic storms are often referred to as Four-Corner lows; usually occurring in the form of cutoff lows.

The center of a Four-Corner low is usually positioned over Northern, AZ, and these systems not only carry saturated air mass from the Pacific, but they also draw moisture from warm SSTs in the Gulf of California (Keen, 1996). These southern storm tracks 114

from the warmer ocean SSTs in the tropical and extratropical Pacific permit greater atmospheric moisture flux, bringing deeply saturated, warmer air to the region which results in even greater unsettled air masses. Summer convective storms draw their moisture from the North American Monsoon (Adams and Comrie, 1997).

The steep and rugged topography make for a highly diversified climate in the Dolores

Basin. The majority of the basin is arid. Precipitation dramatically increases with elevation above 2200 m because the elevated regions directly influence precipitation in the Dolores Basin by providing lifting mechanisms for moist air masses. Dissipating tropical cyclones are rare because of the distance from water bodies; however, the largest flood on the Dolores River was the result of a dissipating tropical cyclone, suggesting that under the right conditions, moist tropical low cyclones are able to reach the intermountain region (Hansen and Schwarz, 1981). During the winter months, moist air masses that enter the region entrained in low pressure, mid latitude cyclones (MLCs) from the Pacific, and account for the majority of precipitation to the region. MLCs that that follow more southerly tracks tend to bring the heaviest precipitation because they carry moist subtropical Pacific air, and may draw additional moisture from the Gulf of

California as well (Keen, 1996). Stalled systems, or cutoff Lows that follow a southerly track, may bring prolonged precipitation. During the summer months, localized convective storms are frequent, particularly in mountainous regions where convection is initiated by orographically uplifted air. Most of the moisture for these storms, is derived from humid air in the sub tropical Pacific, Gulf of California and the Gulf of Mexico 115

(Adams and Comrie, 1997). Average January temperatures in Telluride, CO (1900-2008) range from a minimum of -15° C and a maximum of 2° C, and the average summertime high is 24° C. Average precipitation in Telluride is 590 mm yr-1 and snowfall averages

425 mm yr-1. In Gateway CO, one of the lowest regions of the Basin, the maximum and minimum January temperatures are 16° C and -8° C respectively, with summertime highs of 34° C. The mean precipitation in Gateway, CO is 280 mm yr-1 and the mean snowfall is 410 mm yr-1.

The rivers of the DRB, particularly the San Miguel River, has a long history of placer gold and silver mining that was most active during the early 1900’s, and its legacy remains all over the landscape in the form of hydraulically and conventionally mined gravels. Uranium mining came to the watershed shortly after, and has left a legacy of numerous mine tailings and shafts. The most notable legacy of the uranium-mining period is the Uravan Uranium Project Superfund Site that lines a portion of the banks of the lower San Miguel River (Environmental Protection Agency 2000).

The peak historical flood on the upper Dolores occurred on Oct. 5th, 1911. The flood resulted from a dissipating tropical cyclone that tracked northward from the Gulf of

California, then through eastern Arizona and western New Mexico, before interacting with a secondary low-pressure system from the extra-tropical Pacific (Hansen and

Schwarz, 1981). In Telluride, Colorado, nearly 7.5 cm of rain had fallen during the days leading up to flood event on Oct. 5th. That day, 5.0 additional cm’s fell (in town) on 116

already saturated grounds, causing rapid runoff and rapidly rising waters, ultimately

producing a flood with an estimated discharge of 283 cm3s-1 (Colorado, Office of the

State Engineer, 1912), and on the 4th and early morning of the 5th of October, the heaviest rainfall occurred in the higher elevations of the San Juan Mountains (Hansen and

Schwarz, 1981). Most of the precipitation occurred within the headwaters of the San

Juan River, which resulted in a catastrophic flood on the 6th of October with a peak

discharge of ~4200 m3s-1, more than double that of the record gaged flood of 1927

(LaRue, 1925; Webb et al., 2005). One rain gage, in Gladstone, CO, recorded 210 mm of precipitation in 24 hrs (Hansen and Schwarz, 1981). On the San Miguel, none of the historical, natural floods approach the magnitude of a dam breach flood that occurred in

September 1909, which had an estimated discharge of 283 m3s-1 as well. At Illium CO

(near the site of the dam breach), the return interval for a flood of this magnitude is nearly one order of magnitude greater than the calculated 100-year flood. Even at the bottom reach of the San Miguel River, at Uravan, CO the peak discharge is 1.1 times the 100- year flood (Friedman et al., 2006). The extreme magnitude of the event may have eroded preserved SWDs, particularly in the upper watershed prior to the flood-wave attenuation.

4. Methods

Field studies for this project were conducted between April 2007 and June 2010. During this time, two raft trips were used to access the more remote reaches of the river during

July 2008 and May 2009. It was important to perform this study across the entirety of the

DRB because no single SWD preserved a significant number of floods. In addition, the 117

inclusion of as many sites as possible helps to limit self-censoring and bias toward the youngest and largest flood deposits (e.g. House et al., 2002; House and Hirschboeck,

1997). This ensures the most continuous chronology possible because some SWDs will contain correlated deposits, whereas others may preserve additional flood deposits intercalated within the correlated units. Using multiple sights also may provide us with more robust discharge estimates because it may provide insights to the channel morphological change since the emplacement of the PSIs. PSIs at different locations that return notably different (similar) discharge estimates may suggest there has been a (an) significant (insignificant) amount of channel-geometric change. Changes in the channel geometry may lead to different patterns of erosion and deposition within the same, main river corridor leaving some SWDs to remain depositional zones, and others to either become eroded, or limit further deposition where older units have been preserved.

4.1 Site Selection: The DRB was selected because of its hydroclimatic significance—a significant basin in the UCRB whose streamflow and precipitation patterns are poorly correlated with the LCRB because of its position in the transition zone (Wise, 2010). In addition, paleoflood chronologies in the UCRB are nonexistent to date, making this site an ideal location to begin developing a regional chronology for the UCRB. The physiography of the DRB presents ideal conditions for a paleoflood slackwater study. It meanders through deeply-incised, stable bedrock canyons; contains an abundance of fine- sediments, ideal for demarking the upper water surface during high flood stages; and has 118

an abundance of quarts and feldspars in its sediment, allowing for OSL dating of the

SWDs.

4.2 Flood stratigraphy: All PSIs used in this study are fine-medium grained slackwater deposits (SWDs) that line the Dolores and San Miguel rivers. Some SWDs are preserved on the landscape due to their emplacement in protected alcoves or in back-flooded tributaries where subsequent flows and/or local erosion is unable to produce the necessary shear stress to erode the deposit (e.g. Kochel and Baker, 1982). At most locations, there are sequences of flood deposits, stacked vertically and/or laterally as inset deposits. Individual flood units within SWDs are often distinguishable by abrupt stratigraphic changes (Baker, 1987). Usually the abrupt changes reflect differences in pedogenic development, grain-size variation, changes in color, intercalated colluvial deposits, in-situ surface diagnostics (e.g. burned horizons, artifacts, etc.), and changes in the sediment color (e.g. Baker, 1987, House et al. 2002). Stratigraphic ambiguity is a significant problem at some sites due to bioturbation and the potential for variable hydraulics and flood source variability during a single flood event which may lead to abrupt stratigraphic changes same flood event (House et al., 2002), so wherever possible, emphasis was placed on evidence for sub-aerial exposure, such as burned surface horizons, intercalated colluvial deposits, and archeological evidence to differentiate individual flood units (Hattingh and Zawada, 1996). A serious dilemma when differentiating flood deposits is the potential for relatively short periods of time between floods of similar magnitude. During these clustered events, like those described in the 119

Verde River basin (House and Hirschboeck, 1997; House et al., 2002) floods occur with enough frequency that analytically separating floods using geochronology is likely to result in inconclusive results, and the length of time for sub-aerial exposure may be too limited to cause differences that are stratigraphically distinguishable. For this reason, as well as the concern for self-censorship (House and Hirschboeck, 1997; House et al.,

2002) described earlier, the number of floods differentiated in a stacked SWD sequence was considered a minimum number in this study.

4.3 Paleoflood discharge estimates: EDM (electronic distance meter) total-station surveys were conducted in November 2008 at the Big Alcove SWD, and during June and

July 2010 at the Tafoni SWD. These data were used to develop topographic cross sections for one-dimensional (1D) paleodischarge estimates. Cross-sections were measured for a length of the river, roughly 10-times the channel-width. Spacing between each cross section ranged from 80-220 m, dependent upon the channel geometry. 1D step-backwater modeling has become the standard for estimating the discharge from individual flood events due to the cost and time required for acquiring data for multidimensional modeling, although advances in two and three-dimensional techniques have supported, and continue to advances in the science (e.g. Denlinger et al., 2002;

Pelletier, 2004; Alho and Aaltonen, 2008). In constricted channels where abrupt transitions are uncommon (e.g. widening, constriction, complex distributary channel form) 1D models provide adequate discharge estimates. The HEC-RAS (Hydrological

Engineering Center, River analysis system) 1D hydrologic model, developed by the U.S. 120

Army’s Corps of Engineers (1998) was used to estimate the discharge of paleofloods using a step-backwater routine (Chow, 1959), by solving the conservation of mass, conservation of energy and Manning’s equations (Webb and Jarret, 2002). Water-surface profiles computed in HEC-RAS were iteratively run until the water-surface profiles matched the PSIs (Webb and Jarret, 2002, O’Connor and Webb, 1988). The water- surface-slope can be used to match paleostages of correlative SWDs, which may provide an additional source of confidence to the estimates (Greenbaum et al., 2001). A significant source of error is the presence of fill-terraces along most of the channel margins along the two rivers. They are generally composed of highly erodible fine to medium sands with very little induration. Given the excessive sheer-stress exerted by extreme floods, the terrace form is highly modifiable, making for discharge estimates of low confidence. For this reason, some of the channel geometry that we have derived is assumed to be somewhat arbitrary in the paleoflood discharge calculation. Nevertheless, even though our confidence in discharge estimates may be low (in most cases), they are assumed to represent a minimum flood discharge.

4.4 Geochronology: In an effort to provide a more robust paleoflood chronology this study has applied two geochronology techniques: AMS 14C and OSL. OSL has been applied in some paleoflood studies; however, the quality of the ages has not been well assessed. This paleoflood study is the first to focus efforts on assessing the applicability of OSL in flood studies. Appendix C of this dissertation provides a detailed discussion of 121

the geochronology in this study; the following sections provide generalized summary of

the geochronology in this study.

4.4.1 Radiocarbon: 14C is the most widely used geochronology method in fluvial studies

and the analytical techniques have been highly refined over the last half-century. The

key to interpreting the meaning of the 14C ages is that the material being dated lies in an interpretable stratigraphic context (Delong and Arnold, 2007). Between One and twelve

14C samples were collected at seven slackwater sites following a systematic sampling and interpretation protocol outlined by House et al. (2002) designed to constrain the ages based on rather they are maximum or minimum ages. Most of the charcoal that we collected was detrital, and for this reason we have assumed the ages to represent a maxima (old) for the flood. In the few cases where archeological artifact charcoal was used, the ages represent a minimum (young) for flood deposition. Whenever possible, we used growth ends (e.g. twigs). Landscape residence times for growth ends are much shorter than that of charcoal or larger wood, because they decompose on the landscape more readily, so they are likely more representative of the true age of the flood.

Clustered detrital charcoal may also indicate that landscape residence times were minimal. The clustering implies the least time-since-fire (Gavin et al. 2003), because it is more likely that large quantities of charcoal, following burns are mobilized en masse from basin’s hillslopes. Issues related to landscape residence time of allocthonous detrital organic material is a particularly large issue in paleoflood hydrology since extreme floods tend to erode otherwise stable, alluvial banks (Wohl, 2000). Along with 122

channel modification, stored organic material is inherited into the river and eventually

deposited in fluvial terraces of SWDs.

All radiocarbon samples were prepared and analyzed at the Arizona AMS Lab at the

University of Arizona. Sample targets were prepared by first pre-treating charcoal and/or

wood, using a standard ABA (acid-base-acid) protocol to remove secondary carbon

sources (e.g. humic acids and soil carbonates). Pretreated samples were combusted in a

high vacuum line to isolate the CO2. CO2 was then reduced to graphite targets, which were analyzed at the Arizona AMS laboratory. 14C ages were converted to calendar ages within 1σ and 2σ, using the Calib 6.0 software (Stuiver and Reimer, 1993). Although the calibration process increases the potential for error, it is a necessary step for applied paleoflood techniques in order to place it in a temporal context.

4.4.2 Optically Stimulated Luminescence: Theoretically, there are numerous reasons to use OSL for the dating of flood deposits: 1) 1t provides an independent source of corroboration for the radiocarbon ages; 2) it dates the sediments age of deposition directly, theoretically reducing issues with landscape residence time; 3) the analytical results from the single aliquot and single grain OSL methods allow the researcher to assess the quality of the age, based on the distribution of results—a major limitation of

14C dating; 4) the ages derived from OSL are direct, calendar ages; conversion steps from radiocarbon years to calendar years are a potentially large source of error (Guilderson et al., 2005), and are not required in OSL dating; 5) the use of OSL in paleoflood hydrology 123

has seen limited use, and the results have been of mixed quality, so this research additionally serves as an investigation into the application of OSL for paleoflood studies.

A detailed discussion of the geochronology, in particular, the OSL dating for this research can be found in Appendix C of this dissertation.

Luminescence dating relies on the principal that all things in nature produce small but measurable radioactive decay that produces free electrons (i.e. ionizing radiation) that can be stored by the crystal lattice in quartz and feldspar mineral grains. The amount of stored charge in a single grain is linearly proportional to the time since deposition. The charge in mineral grains will continue increase until it saturates or is exposed to light.

Visible light provides the energy required to evict stored electrons (i.e. the charge) from quartz and/or feldspar’s crystal lattice, effectively “zeroing” the charge (Aitkin, 1998).

The light sensitivity of samples requires careful collection and transport of the sample, so we pounded 8” x 2” opaque steel conduit tubes into the SWDs and sealed samples from light. We also representatively sampled sediment from the surrounding 10 cm of the sample tube to be analyzed for background ionizing radiation. Measurements were made to the top of the geomorphic surfaces to account for the influence of cosmogenic radiation, and subsamples from the original sample were taken in the lab for water content, to calculate the attenuation of ionizing radiation to the sample. We made efforts to avoid the sampling of bioturbated sections, aiming for original fluvial structures

(bedding).

124

All samples were opened and processed under dim amber safelight conditions within the lab. Sample processing follows standard procedures involving sieving, gravity separation and acid treatments with HCl and HF to isolate the quartz component of a narrow grain- size range, usually 90-150 µm. The purity of the samples is checked by measurement with infrared stimulation to detect the presence of feldspar. Sample processing procedures follow those outlined in Aitkin (1998) and described in Rittenour et al., (2003,

2005).

The USU Luminescence Lab follows the latest single-aliquot regenerative-dose (SAR) procedures for dating quartz sand (Murray and Wintle, 2000, 2003; Wintle and Murray,

2006). The SAR protocol includes tests for sensitivity correction and brackets the equivalent dose (De) the sample received during burial by irradiating the sample at five different doses (below, at, and above the De, plus a zero dose and a repeated dose to check for recuperation of the signal and sensitivity correction). The resultant data are fit with a saturating exponential curve from which the De is determined. The reported De is based on the mean and standard deviation from the measurement of at least 20 aliquots of sand mounted on a 1-2 mm diameter disks. Final De values used to derive the ages of burial have been determined using either the central age model (CAM) or minimum age model (MAM) (Galbraith et al., 1999). The age model we selected was based upon the degree of over-dispersion of the individual aliquots, and the age of the sample. In general, older samples (>2500-years B.P.) utilized the CAM, and young samples utilized the MAM. This is discussed in detail in appendix C of this dissertation. 125

5. Results

There are numerous SWDs preserved in the DRB, each containing incomplete flood records. Unlike prior paleoflood studies such as the Verde River in AZ, or the Paria

River in UT, where exceptional, semi-continuous deposits were emplaced due to downstream constrictions allowing the backfilling of tributaries or alcoves (House et al,

2002; Webb et al, 2002 respectively), the geometry of the Dolores River lacks the permanent constrictions capable of causing significantly back-flooded reaches, so SWDs are generally confined to 1) channel margin deposits in alcoves, 2) on the lee side of obstructions, or 3) on the inside-bend of meanders. Table 1 and Table 2 show the ages of individual samples and their corresponding stratigraphic units; the complete results are discussed in greater detail in the following sections. A minimum of sixteen individual flood units have been identified and deposited over seven SWDs, which are described below. These individual flood units have been named strictly by their chronological periods using the following chronology scheme: Hf1-Hf12, where Hf1 represents the oldest flood period and Hf12 represents the most recent flood period. Where chronological evidence bracketed deposits closely, but distinct flood units exist between them, a secondary lettering scheme was used to designate their depositional period relative to the age-constrained unit.

5.1 Slackwater flood deposit stratigraphy: From the summer of 2007 to the fall of 2009, we located over two-dozen potential SWD study sites and seven of these sites were 126

investigated further based upon the number if individual flood units preserved, and

degree to which the sites had undergone post-depositional disturbance. At each site, we

collected organic material for AMS 14C and at five of the sites sediment samples for OSL dating were collected. The Basin has been divided into three distinct regions, and the following conventions are used to refer to various river reaches: The San Miguel (refers to the entire reach), the Upper (Lower) Dolores refers to the section of the Dolores River located upstream (downstream) of its confluence with the San Miguel River. Each of the sites is described in detail below, and their locations are shown in Figure 2. All of the ages listed here have been converted to calendar years (B.P.) using the Calib 6.0 software

(Stuiver and Reimer, 1993) to 1σ, although Table 1 lists the calibration range to 2σ. In the case of OSL ages, are produced as calendar years and do not require calibration, and error ranges are analytically derived. OSL results are shown in Table 2.

5.1.1 Big Rock alcove SWD: This site is on the Dolores River, and is the furthest downstream SWD that we chose to sample. Figure 3 shows the stratigraphy and collection points for 14C, as well as OSL sample locations and their corresponding ages.

The SWD is linearly extensive, approximately 250m in length. The site we sampled lies on the downstream side of a large boulder protruding from the canyon wall and stands approximately 2.5 m tall. The top of the SWD stands 4m (preliminary) above the modern river. There are at least two individual flood units preserved at this site. Bedding is weakly preserved, and the entire SWD is generally massive, fine-medium grained sands, 127

although bedding has been lightly preserved in small pockets throughout the unit. The

oldest unit preserved at the site (Hf8) contains an abundance of charcoal where two

charcoal samples were collected. While no archeological evidence was identified, the

charcoal was burned in-situ and appears to have been disturbed following the burn. AMS

14C has returned corroborative ages for the following two samples in the bottom stratigraphic unit: 1250-1302 years B.P. (MC031809-05) and 1266-1311 years B.P.

(MC031809-06). Their in-situ context indicates the ages represent the youngest age estimate for the deposition of the lowest layer. This stratigraphic layer consists of fine-- medium sands, and is somewhat loose. Weakly developed pedogenic carbonate nodules

(stage I+) are present as well. The top several centimeters of the layer are intercalated with platy lithic fragments (sandstone) of probable colluvial origin, from overlying

Mesozoic sandstone cliffs. One charcoal sample was collected for 14C from the colluvial

layer. The age returned for this sample is 877-955 years B.P. (MC080507-03). The

charcoal was burned in-situ as well, so the age represents the oldest possible age for its

deposition, and a minimum age for the underlying deposit. The upper flood stratigraphic

unit of the Big Rock SWD is dominantly composed of fine sands. No 14C samples were

collected within this layer because of a complete lack of contextual organic material, so

we selected this site for OSL dating. OSL ages returned an age of 1.11 ± 0.17 ka B.P.

(USU-627). It can be assumed by the bracketing of ages, that Hf9 was emplaced roughly

0.94 to roughly 1.2 ka B.P. A subjective look at the aliquot distribution for this sample

would suggest the true age is on the younger tail of the distribution, and the actual age is

likely close to the analytically derived age and the spread of the aliquots is primarily a 128

function of partial bleaching. This layer contained the majority of the preserved bedding

suggesting minimal bioturbation of the sample, and the limited skew toward smaller De

Values supports this claim. When bracketed, the lower flood unit: Hf8 was likely deposited between 1.3-1.1 ka B.P., making Hf8 and Hf9 very similar in age.

5.1.2 Big Alcove SWD: This SWD site lies upstream approximately 3km from the Big

Rock SWD on the lower Dolores River. It has been deposited on the inside of a meander, under an overhanging alcove, stands approximately 1.2 m tall, extends outward 2m and it is approximately 5m in length. Figure 4 shows the stratigraphy of the unit. There is an abundance of archeological evidence at the site, suggesting it has been disturbed numerous times, initially by indigenous North Americans, and more recently by local placer mining operations near the turn of the 20th century. Most recently, mining

hobbyists occupy the site regularly. The upper flood layer (Hf9) has an upper surface

disconformity, which contacts reworked Plio/Pleistocene river gravels that have scoured

the upper surface due to localized slope wash from placer mining. An abrupt transition

between Hf9 and the lowest flood stratigraphic unit (Hf8) is evident by a transition in

color, dominant grain size, texture, and an apparent difference in pedogenic development.

Hf9 is 10-15 cm thick and is composed of primarily massive, medium to fine-grained

sands; however, cross- bedding is preserved in some places within the unit. There is no

apparent pedogenic development in this unit. Hf8 is >30 cm thick, has a granular texture,

has some preserved cross-bedding, is mostly composed of medium sands and displays

weak pedogenic carbonate development (Stage 1+ nodules). Nine charcoal samples and 129

one wood sample were collected at this SWD. Additionally, two OSL samples were

collected here. Hf8 has returned the following AMS 14C ages: 1110-1210 years B.P.

(MC052908-01), 1110-1220 years B.P. (MC080507-05), and 820-910 (MC081509-06).

The first two samples were burned in-situ, evidenced by the stone hearth in the unit and large angular charcoal fragments, so they represent a minimum time since the sediments deposition. The latter age is likely erroneous, because it was collected from a part of the deposit where there was ample evidence for bioturbation, specifically the translocation of charcoal from the upper unit. Additionally, it is more consistent with the age of Hf9.

The OSL age from Hf8 returned an age of 1.44 ± 0.18 Ka B.P. (USU-331), which is

relatively consistent with the 14C ages. An additional OSL sample was collected in the overtopping, Hf9 unit from the preserved bedding, and it returned an age of 0.99 ±0.08 yrs B.P. (USU-332). 14C samples collected in the top of the unit returned much younger

ages, ranging from modern to 330 years B.P., although these samples have likely been

reworked from above and may represent the age of reworked, bioturbated charcoal from

the placer reworking on the overlying gravel deposit. Bracketing ages at this site

suggests the Hf8 unit is likely 1.1-1.5 ka B.P., and the Hf9 unit is roughly 0.9-1.1 ka B.P.

5.1.3 Tafoni SWD: This paleoflood deposit is located on the San Miguel River,

approximately 2.3 km upstream of the confluence with the Lower Dolores River. It is

approximately 2 m tall, 12 m in width, and extends into the tafoni sandstone roughly 1.5

m. The site is protected from fluvial erosion, which has made possible the preservation

of the depositional surface. There are at least seven significant floods preserved at this 130

site. Figure 5a shows the stratigraphy at the site, and Figure 5b shows a detailed

stratigraphic section of the east end of the SWD. The oldest flood unit (Hf5) is

moderately bioturbated, so very little fluvial bedding is evident. There is also a weak

pedogenic carbonate development (stage 1). Hf5 is overlain by coarse lithic clasts, likely

derived from slope wash or of colluvial origin. No interpretable organic material was

present within the Hf5, so we were limited to collecting sediment samples for OSL

dating. The age returned 2.94 ± 0.32 Ka B.P. (USU-330). On the upstream (east) side of

the deposit, geochronology and stratigraphy provide the most continuous record

preserved at a single site. The upper unit (Hf11) has very few charcoal fragments. One

sample was collected here: MC031809-02; it returned an age of 750-970 years B.P. An

abrupt transition in grain size separates Hf11 from its underlying unit: Hf10. This unit is

roughly 7-10 cm thick, and contained no organic material for dating. Its limited thickness

made it impossible to sample for OSL as well, so its age can only be bracketed between

known age flood units. The next lower unit: Hf9 lays uncomformably beneath a slope

wash unit that scoured its surface following its deposition. No organic material was

collected in Hf9, although an OSL sample collected here returned an age of 1.10 ± 0.13 ka B.P. (USU-626). The base of the Hf9 unit is composed of several coarse, rip-up clasts, and underlying that is a fine-sand-silty cap. The abrupt transition in grain size delineated next older flood unit. No age control exists for the unit, so its age is uncertain; it is Hf7 or Hf8. The oldest unit in the section, likely associated with the Hf6 period, underlies the unit. The age is based upon a single detrital charcoal unit which returned an

AMS 14C age of 1849-2071 years B.P. (MC031809-03), so the confidence in this age 131

designation is low; however, the proximal upstream Mucky Alcove SWD provides some evidence this designation is allowable, based upon it’s stratigraphic position and geochronology. The youngest flood unit is not present within the vertically accreting flood stratigraphy. This is due to the limited geometry of the alcove site. It is present on the opposing side of the bedrock, however, but no age control exists, so we can only designate is as being younger than Hf11, making it Hf13.

5.1.4 Mucky Alcove SWD: This site is located approximately 0.7 km upstream from the

Tafoni SWD site, on the San Miguel River. Figure 6 shows the stratigraphy and ages of samples collected at the site. The flood deposit was emplaced within a bedrock alcove with an overhanging, unstable ceiling. The ceiling actively collapses large, rectangular blocks up to a meter in length. In some cases, these blocks help to differentiate different flood units. Unfortunately, the site has been disturbed by placer mining, and some of the tailing obscure the site. The SWD extends approximately 10m in width, it protrudes roughly 1-3 m from the bedrock face, and stands roughly 4 m above the modern river, with silt and sand units plastered to the bedrock alcove up to 5 m above the river. The site appears to have been site of human occupation, evidenced by the stone hearth and excessive amounts of large, in-situ, angular charcoal fragments. Over-topping the older archeological site are well-bedded flood deposits with abundant detrital charcoal fragments. The site’s stratigraphy is complex, and the upper part of the deposit is composed of coarse sands, indicating the top of the flood was likely at a much higher stage the SWD suggests, making this a poor site for discharge calculations. Within the 132

lower reaches of the unit, well-bedded, medium-coarse sands grade upwards into finer sands or charcoal layers suggest that up to seven distinct flood deposits have been preserved at this site; however, we densely dated the section using AMS 14C, and the results were inconclusive, although, for the most part, they were stratigraphically consistent. The charcoal fragments and growth ends were mostly of detrital origin and did not provide the necessary evidence with differentiate the same number of flood events at the stratigraphy suggested. In total, 12 samples were retained for AMS 14C analysis at this site. Correspondence between the ages suggest two distinct periods of flooding: one centered at approximately 1500 years B.P. and another slightly earlier, at

1800-2000 years B.P., however, as described above, the stratigraphy is suggestive that numerous floods were emplaced during the two periods of flooding. While the complex stratigraphy lacks the necessary age control to isolate the same number of independent floods with confidence, the range of ages in the stratigraphic unit make the bracketing of a range of frequent flooding allowable. It is also clear that the upper part of the flood unit has been removed by mining operations, so we have interpreted this site to be preserving between 4-7 distinct events. To maintain coherence in the naming scheme of the flooding periods, the results at this site range from Hf6-Hf7. The secondary, lettering designation is used to indicate the uncertainty in our age interpretation (clear, abrupt transitions are apparent between flood stratigraphic units, although the ages do not reflect this). The units are discussed in general terms below.

In the lower flood stratigraphic unit (Hf6), coherency between in-situ, and slightly older 133

detrital samples form the basis for bracketing the minimum and maximum age of the

flooding periods between 1680-1840 (MC081409-03) and 1960-2140 (MC111108-03)

years B.P. Within the Hf7 flooding period, the age ranges from a maximum of ~1700

years B.P. (as determined from the in situ burned material in the upper part of the Hf6

unit), and is as young as 1470 B.P. (as determined from a burned twig). There is strong

agreement between several of the samples, and therefore it provides a justifiable basis for

using its age as a close estimate of the minimum age (e.g. Gavin et al. 2003). The upper

portion of the SWD has been removed, however, so the young range of flooding is only

associated with the main stratigraphic section, and may not be associated with the

plastered medium-coarse sands that are indicative of the higher stage floods.

5.1.5 McPhee Alcove SWD: This is the most upstream site on the Upper Dolores River.

The actual deposit is 16 km downstream of McPhee Dam. The SWD is composed of two

flood stratigraphic units, Hf6 and Hf7. The SWD is adjacent to a well-maintained gravel

road, so it is possible that much of the SWD has been disturbed and removed by the road

construction. Figure 7 shows the stratigraphy and locations of the samples. The upper

charcoal-rich surface is allocthonous and detrital. The abundance of charcoal is

suggestive of a minimal time-since-fire (e.g. Gavin et al. 2003), so the sample from this

layer (MC081409-03 1680-1840 B.P.) may be the best indicator of the age of Hf6 at this

site. These detrital charcoals are rare in Hf6, and few samples were located. There is

well-preserved bedding within the lower portion of Hf6, and it grades into massive sands.

The entire deposit is composed of well indurated, medium to coarse sands. Highly 134

altered bedrock sheets are intercalated within the deposit, although, there is no evidence

to suggest the intercalation of the bedrock and flood sediments are indicative of multiple

floods. The intercalation, and high degree of bedrock alteration makes the distinguishing

between the bedrock and sediment interface vague. The top of Hf7 unit is difficult to

identify as well because the flood deposited materials that were higher in elevation than

the top of the alcove, similar to the Mucky SWD. One detrital charcoal was collected

here, and returned an AMS 14C age of 1360-1470 B.P. (MC111008-01). The fine-

medium grained sands were not well preserved over broad regions and only left in small

alcoves within the bedrock cliff. In addition, no organic material was recovered from the

smaller remnant deposits above the main SWD, so assuming they are associated with the

Hf7 unit is not supported.

5.1.6 Juniper SWD: This site was the first of the SWDs sampled on a float trip in May

2009. It is a large, overbank SWD on the inside meander (river left) within the Slickrock

Canyon reach of the Upper Dolores River. The dimensions of this site are difficult to

discern from the colluvial units along the canyon walls; however, the portion that we

sampled stands roughly 5.8 m above the modern river, and is approximately 20 meters in

width. We did not excavate the site in detail due to limitations on the time and equipment

available to us, although in total, four small trenches were dug, and two of these were

sampled for OSL and 14C. We collected three OSL samples and four 14C samples, although only two OSL and three 14C samples were analyzed. Figure 8 shows the stratigraphy and locations where samples were collected. Only the upper reach of the 135

deposit was sampled and this site remains a site for future paleoflood investigations. The

SWD is partially covered with large (~1m) angular colluvial boulders, and some boulders

are buried throughout the unit, which helps to delineate individual flood units. The flood

stratigraphy is generally massive, and composed of medium to fine-grained sands. At

least three floods are represented in this deposit. Detrital charcoal was collected from the

upper stratigraphic unit of the SWD, Hf3, and returned an infinite age. No other organic

material was located in Hf3, although an OSL sample collected here returned an age of

7.84 ± 0.66 yrs ka B.P. (USU-628). The next two flood-stratigraphic units are

differentiated by abrupt breaks in the color and texture of the sediment. The first break

occurs roughly 40 cm beneath the surface. Hf2 is composed of slightly finer sands than

Hf3; a 14C sample returned an age of 7940-8020 yrs B.P. (MC051909-04) and an OSL

sample was collected as well, which returned an age of 6.23 ± 0.47 yrs B.P. (USU-629).

Beneath Hf2, a thin colluvial or slope wash unit approximately 5 cm thick delineates a

third flood deposit (Hf1), mostly composed of medium to fine sands. A 14C age of 8820-

8880 B.P. (MC051909-02) was returned for charcoal that was collected in Hf1, making this flood stratigraphic unit the oldest in the chronology.

5.1.7 Tributary SWD: This particular site is located at the mouth of a minor tributary.

Similar to the Juniper SWD, It is also located along the Slickrock Canyon reach of the

Upper Dolores, making it a difficult site to reoccupy. This is the highest standing flood deposit that we identified; the top of the SWD stands 6.5 m above the modern river level, the SWD is approximately 30m in width, and extends outward, roughly 10 m from the 136

canyon’s colluvial slopes. Figure 9 shows the site’s stratigraphic context. The site is

ideal for SWD deposition: the deposit is emplaced at a notable expansion of the river,

which is further expanded by the tributary. Time spent at the site was minimal, so we

were limited to reconnaissance style excavations to describe the stratigraphy in the upper

reaches of the deposit and collect materials for radiocarbon and OSL dating. The upper

part of the SWD is composed of fine to medium-grained sands, and forms a massive,

loose deposit. No interpretable organic material was collected here, although its

stratigraphic position suggests it may be correlative with HF3. It overlays the burned soil

horizon and the top of the unit displays moderate rubification of the upper horizon,

suggesting it may be mid-early Holocene in age. The second flood-stratigraphic unit in

depth, Hf2, is only slightly younger than the Hf3 unit. Only one 14C sample and one OSL were analyzed within Hf3 at this site. The 14C samples returned an age of 7561-8269

(MC052009-02) yrs B.P. The detrital charcoal used for this sample was collected 0.8 m from the surface, and 40cm beneath the burned soil horizon that delineates the top of this deposit. The OSL sample was also collected in Hf1, 1.0 m beneath the surface and it returned an age of 8.21 ±0.79 ka yrs B.P. (USU-630). A subjective inspection of the aliquot distribution of the OSL sample suggests the leading tail of the distribution is more likely indicative of the actual depositional age of the deposit, placing it in excellent agreement with the 14C sample.

5.2 Correlative floods: A summary figure of all of the SWD stratigraphic sections is

shown in Figure 10. Correlations of individual flood periods are also shown in the figure. 137

Correlating the flood units at different SWDs allow for more robust estimates for the

timing and magnitude of the floods by providing more data points for developing the

chronology. Developing evidence to correlate deposits is not a trivial matter, however.

Correlating SWDs using 14C alone is problematic because the age of the flood is not

necessarily indicated by the 14C age at the site (Webb et al., 2002). The inclusion of OSL ages in the correlations does not allow for robust correlations between flood deposits solely on the merits of geochronology. Determining correlations on the basis of sediment color and texture between distal SWDs is difficult as well, because of the transitions in source regions (i.e. additional tributary influences) that bring varied sediment types. And finally, attributing SWD correlations using discharge estimates is not possible in most cases because it cannot simply be assumed that discharge estimates decrease or increase linearly downstream (e.g. flood-wave attenuation; tributary influences). Correlations are justified however, by developing multiple lines of evidence using any these variables.

The following section describes the relatively few correlative flood sites. The Hfn flood designations, unless otherwise noted, should be thought of as timeframes where flooding was more prevalent. That is to say, the similar designation based on the naming scheme need not be considered correlative to other similarly designated units.

The Hf9 flood stratigraphic unit at the Big Alcove SWD, Big Rock SWD and Tafoni

SWD are likely correlative. We correlated the Hf9 unit at these sites based on sediment color, grain size and geochronology. The SWDs are within several kilometers of one another, making the use of these data allowable. On the basis of OSL alone, a subjective 138

comparison between the aliquot distributions of each (USU-332, USU-627 and USU-

626) suggest similar bleaching characteristics, a characteristic that is not a requisite for, but is supportive of sediments deposited from the same flood event. Both sites have very limited amounts of charcoal in Hf9. At the Big Alcove site, radiocarbon ages in Hf9 ranged from modern to several hundred years in age. These erroneously young ages are likely the result of the translocation of the charcoal, into the Hf9 flood deposit from above. The issue of translocation of charcoal is not unique, and has been described in other flood studies as well (e.g. House et al., 2002). Hf9 at the Tafoni site lacked organic material, although the sediment texture, grain-size, and an OSL age of 1.10 ka B.P.

(USU-626) are all data that are highly supportive of this conclusion.

Due to the extremely rare nature of a flood with the magnitude to deposit the upper units at the Tributary and Juniper slackwater sites, it is likely that several of the units between the Juniper and Tributary SWDs are correlated. Hf2 and Hf3 represented at both sites share similar textural and color characteristics. The upper 5-10 cm of Hf2 show moderate rubification at the Tributary site. Rubification is present in the upper 10cm at the Juniper site, although it is not as evident. Texture and grain size characteristics are similar as well. AMS 14C ages are similar, although they are detrital and suggest a maximum age (7942-8017 at the Tributary and 7687-8044 at the Juniper SWDs). The

OSL ages disagree, however, as much as 2,000-years. At the tributary site, the OSL age corroborates the AMS 14C age, but at the Juniper site, it is much younger, and an age reversal is present, with the overlying Hf3 deposit having an older OSL age. One 139

possibility is that the sediment is poorly bleached at the Juniper Site, an issue that is

inherent to many flood-deposited OSL samples. However, the aliquot distribution of

USU-629 suggests the sediment is well-bleached. Another possibility at this site is the

improper calibration of the environmental dose rate. We did not account for bedrock or

colluvial boulder sources of radioactive decay, which may significantly offset the age.

The stratigraphic position however, is suggestive that the AMS 14C age may better

estimate the age of the unit.

5.3 Discharge estimates: During the summer of 2010, we developed cross-sectional

topographic data on the San Miguel River, at the Tafoni SWD. This was the most

suitable site for discharge modeling due to the channels’ consistency, and lack of

significant expansion/contraction geometric changes that may otherwise require 2D

modeling. In addition, it was one of the few sites that did not require extensive cutting of

invasive tamarisk vegetation. An EDM total-station was used to develop the topographic

data, so accuracy of the channel measures are within 1 cm. The upper unit of the Tafoni

site (Hf11) was emplaced at a minimum discharge of 1050 m3s-1. The lower range of the flood evidence at this site (Hf5) is roughly 750 m3s-1. Figure 11 shows the longitudinal profiles for paleoflood water surface-slopes at the Tafoni site for each flood episode described.

5.4 Paleoflood chronology: The paleoflood time series (Figure 12) shows the number of floods we confidently identified in the SWD record, so they represent a minimum number 140

of actual flood events. They are stacked within their respective periods. During the last

800 years, no evidence exists for large floods that exceeded the SWD censoring level of roughly 520 m3s-1—based on the minimum flood stage required to deposit SWD sediments in the SWD sites. This indicates that the contemporary hydroclimatic regime is not producing floods as large as those prior to 800 years B.P. An abrupt increase in large flooding episodes occurred at roughly 2000 years B.P., and there is evidence that the increase in the large flood frequency peaked in the middle of the last millennium.

Evidence for the enhanced flooding continued, until the SWD flood evidence indicated an abrupt end around 800-900 years B.P. The Hf1-Hf3 stratigraphic units show evidence of older flooding episodes between 7800 and 9000 years B.P.

6. Discussion

Unlike the LCRB, where a 5,000-year compilation of extreme paleofloods has been developed (Ely et al., 1993; Ely, 1993), there is a lack of paleoflood records for vast portions of the UCRB—a critical water resource region due to its extensive winter snowpack. This study represents early steps in developing a robust chronology for the

UCRB, but lacks evidence of similarly large floods in the gaged record, and therefore there are no flood analogs to allow for a flood-hydroclimatology study. Within the DRB, the largest historical and gaged floods occurred on Oct. 5th, 1911 (State Engineer of

Colorado, 1912) on the upper Dolores, and one on the San Miguel River, on Sept. 5th,

1909, which resulted from a dam breach (Friedman et al., 2006). These two historical floods do not come within 25% of the magnitude of the larger floods in the paleoflood 141

record. In fact, no historical floods are preserved within the SWD record, because none have achieved the ~520 m3s-1 censoring threshold that is required to deposit the SWDs in this study. This provides evidence to say that the flood-generating hydroclimatic setting was quite different during the period defined by the paleoflood record, than today, a time that seems to see very suppressed flood magnitudes. Such a pronounced difference between the paleo and gaged/historical record could indicate that the channel bed elevation may have adjusted though time. In response to that consideration, we compared the discharge with the flood and paleoflood envelope curve for the full

Colorado River Basin (Enzell et al., 1993), and the peak floods lie well within the expected range of discharge possibilities (Figure 13). We inspected the vertically accreting Holocene flood terraces that occupy extended reaches of the channel’s margins as well. These terraces show no coarse grained exposures—those that may indicate channel-bed materials. In fact, the modern channel margins appear to be dominated by fine-grained sediments—evidence for continuous filling by overbank flood deposition, so the paleochannel may have actually had a larger cross-sectional area than the modern channel.

The paleoflood record described here presents a classic example of a paleoflood chronology whose flood magnitudes dramatically exceed the largest historical floods.

The paleoflood chronology in this study provides evidence for much larger floods that have occurred in the basin, and therefore expands our recognition of the range of possibilities for flood magnitudes. These results provide a truer understanding of the 142

actual flood hazard, especially important given our rapidly changing climate. This study has provided an empirical record of floods within the basin that may be used to begin to develop a context of flooding, relative to climate variability in the UCRB.

The most prominent features of the paleoflood chronology is the abrupt termination of large floods preserved in the SWD record over the last ~800-years. Additionally, there is a lack of large flood evidence in the middle-late Holocene, prior to 0 A.D. (with only a single exception: Hf5). The lack of very large flood evidence in the gaged record and

SWD record over the last ~800-years, suggests the hydroclimatic setting has remained relatively stable in terms of it’s flood-generating potential over the last 800 years. This is rather contradictory to the paleoflood record in the LCRB, which indicates that there have been frequent extreme floods since roughly 1300 A.D. (Ely et al., 1997). The timing of the paleofloods in the DRB, relative to the paleoflood chronology in the LCRB (Ely et al., 1993; Ely, 1997) indicates the periods of frequent, extreme flooding are generally out-of-phase with the one another with the exception of a short but weak period of overlap between the chronologies from ~800-1100 A.D. (Figure 12). While we remain cautious about this finding due to the relatively small data set this study provides, preliminary paleoflood results on the Upper Colorado River near Moab, UT, indicate similar trends in their timing. (Greenbaum, et al., 2006). These similarities suggest the

DRB may provide a reasonable analog for the eastern half of the UCRB.

143

It is not entirely surprising that the LCRB and DRB paleofloods do not share similar

timing. Studies within the LCRB have highlighted the association of anomalously high

precipitation and streamflow with periods of higher frequency (and intensity) warm-

phase ENSO, or ENSO-like events, and the ENSO pattern is most dominant within the

southern part of the Colorado River Basin. The data backing this includes modern,

systematic data (e.g. Cayan et al., 1999), geomorphologic data from arroyo cut-fill (e.g.

Herford, 2002), dendrochronological data (e.g. Woodhouse et al., 1997) and paleoflood

records (Ely et al., 1993; Ely, 1997). On the other hand, studies within the UCRB remain

far less conclusive, as to what the primary mechanisms are that are responsible for

anomalously high streamflow and precipitation (e.g. Wise, 2010; Hidalgo and Dracup,

2003; McCabe et al., 2007; Canon et al, 2007; Appendix A this dissertation).

The abrupt termination of large floods in the DRB roughly 800-years ago is broadly

coincident with the end of the MCA (900-1300 A.D.; Hughes and Funkhouser, 1998;

Benson et al., 2007; Cook et al., 2004)—a period noted by widespread (Cook et al.,

2004) and severe droughts (Meko et al., 2007) . On the contrary, the flood frequency/magnitude regime in the LCRB was generally increasing following the transition out of the MCA. The increased flooding in the LCRB was interpreted as the result of a cooler and wetter temperature/precipitation regime, brought on by frequent El

Nino events (Ely, 1993; Ely, 1997). During the early part of the MCA, prior to the peak drought in the mid 1100’s (i.e. Meko et al., 2007), evidence for large floods is evident in the SWD chronology, and both AMS 14C and OSL support this. The Palmer Drought 144

Severity Index (PDSI) indicates there was a period of frequent, moderate droughts in the

LCRB (the period from roughly 500-1000 A.D.) (Cook, 2004), prior to more intense drought that occurred in the mid 1100’s A.D. (Meko et al., 2007). Archeological evidence supports this interpretation. The Anasazi evacuated from the Dolores River

Basin region during the MCA in two phases: during the mid 1100’s and again in the late

1200’s (Benson et al., 2007). The intensive droughts were interpreted as a precipitation response to phase coherency between the Pacific Decadal Oscillation (PDO), and the

Atlantic Multidecadal Oscillation (AMO), more specifically, during PDO(-) and

AMO(+).

If flooding within the DRB, and likely the UCRB are not associated with the same hydroclimatic factors as the LCRB (based on the similarity between the preliminary study results of the UCRB), then the question remains: what are the responsible mechanisms for driving large floods in this region? Also, could a relatively modest change in the average drought condition change the magnitude and frequency of flooding in such a dramatic way? Knox (1993) demonstrated that very modest shifts in the climate are capable of this. It is feasible that it may not be the direct precipitation that is responsible for most of the flood volume, especially in snowmelt-dominated rivers. Specific antecedent conditions are often required for extreme flooding (e.g. Hirschboeck, 2003).

For example, during dry periods, increased episodes of wildfire alter the surface hydrology and can work to increase the rate of runoff. Using the wildfire chronology from lake cores in the San Juan Mountains provides some evidence for this. Strong 145

agreement between numerous records in the San Juan Mountains indicates that periods of very high fire episode frequency from roughly 2000-1000 years B.P. (Anderson et al.,

2008), is coincident with increased flooding in the DRB. This period of enhanced fire return intervals are attributed episodic droughts from O to 1000 A.D. (Anderson et al.,

2008). An additional explanation may have to do with the dry period in the southwestern deserts. Modern anthropogenic expansion of the dust source region for the San Juan

Mountains (Neff et al, 2008) has been shown to intensify snowmelt and subsequent runoff by up to 35 days (Painter et al., 2007). Dust deposition on snowfields is well known to facilitate rapid snowmelt and intensify runoff by decreasing the snow’s surface albedo. Intensified glacial runoff and ice ablation have been linked to dust blanketing as well, even when meteorological conditions remain unchanged (Fujita, 2006). Based on this, it is feasible that if periods of droughts (i.e. high dust deposition), coupled with a single wet episode during a critical time of year may facilitate the necessary conditions to trigger extreme floods.

With respect to global and regional climate change, these types of flood mechanisms are a cause for concern. Arid regions of the world are expected to undergo more frequent and persistent droughts as a direct result of climate change (e.g. Seager et al., 2007).

Wildfires are increasing in the western U.S. (Westerling et al., 2006), and in the southwestern U.S. dust source regions are expanding (Neff et al., 2008) and are therefore becoming greater sources of atmospheric dust (Tanaka and Chiba, 2006). In the San Juan

Mountains, the rate of dust deposition has dramatically increased due to anthropogenic 146

land use changes (Neff, 2008) and has subsequently reduced the spring snowmelt season

(Painter et al., 2007). If dust deposition is at least in part, responsible for high runoff in

the region-a reasonable assumption given the snowmelt dominated flow regimes, then

there are important implications for future flooding in the region and it may inform us of

the antecedent conditions that help to set up the conditions that will allow for extreme

flooding.

Storm tracking in the past was probably a significant contributing factor as well. Within

the LCRB, it has been determined that the large, frequent floods have been dominantly

driven by winter storms (House and Hirschboeck, 1997), and to a lesser degree,

dissipating tropical cyclones. The large paleofloods in the LCRB have likely resulted

from zonal storm tracks associated with increases in transient eddy propagation from

frequent and intense El Nino events. Within the DRB, the fact the floods appear to be

controlled by other synoptic patterns suggest that it is probable that the storm genesis is

farther to the north, resulting from enhanced meridional jet patterns associated with a

North Pacific influence. Anomalously high flows (albeit, they are not large floods) appear to be associated with a North Pacific influence in the gaged record, supporting this

hypothesis (Appendix A, this dissertation). Conclusive evidence for this will not be

available however, until further paleoflood studies develop chronologies for tributaries in

the UCRB.

7. Conclusion 147

The paleoflood record we present here highlights the need to incorporate as many SWDs as possible to capture the maximum number of floods in the paleorecord. Even in cases where SWDs are only within 1-2 km of each other, differing chronologies may result due to local hydraulic conditions (i.e. the opportunity for deposition), or alcove characteristics

(i.e. upper and lower stage thresholds for deposition). No single SWD site with an extensive, semi-continuous record is present in the basin, unlike most studies that have successfully developed entire chronologies based on a single site. To address the shortcoming, numerous sites were used to both extend, and to fill the gaps in the paleoflood record.

Future paleoflood studies within the region will add to our understanding of hydrologic responses to climate change, and modern analogs in other sub-basins will provide a better context to understand the actual mechanisms responsible for extreme flooding. This study provides a starting point for future flood studies in the UCRB. 148

WORKS CITED

Adams, D.K. and Comrie, A.C., 1997. The North American monsoon. Bulletin of the American Meteorological Society, 78(10): 2197-2213.

Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford Science Publications, 267 pp.

Alho, P. and Aaltonen, J., 2008. Comparing a 1D hydraulic model with a 2D hydraulic model for the simulation of extreme glacial outburst floods. Hydrological Processes, 22(10): 1537-1547.

Anderson, R.S., Allen, C.D., Toney, J.L., Jass, R.B. and Bair, A.N., 2008. Holocene vegetation and fire regimes in subalpine and mixed conifer forests, southern Rocky Mountains, USA. International Journal of Wildland Fire, 17(1): 96-114.

Baker, V.R., 1987. Paleoflood hydrology and extraordinary flood events. Journal of Hydrology, 96(1-4): 79-99.

Baker, V.R., 1994. Geomorphological understanding of floods: Geomorphology. Geomorphology, 10(1-4): 139-156.

Baker, V.R., 2004. Palaeofloods and global change. Journal of the Geological Society of India, 64(4): 395-401.

Baker, V.R., 2006. Palaeoflood hydrology in a global context. Catena, 66(1-2): 161-168.

Baker, V.R., 2008. Paleoflood hydrology: Origin, progress, prospects. Geomorphology, 101(1-2): 1-13.

Baker, V.R., Kochel, R.C. and Patton, P.C., 1988. Flood geomorphology. Wiley, New York, 503 p. pp.

Benito, G. and Thorndycraft, V.R., 2005. Palaeoflood hydrology and its role in applied hydrological sciences. Journal of Hydrology, 313(1-2): 3-15.

Blainey, J.B., Webb, R.H., Moss, M.E. and Baker, V.R. (Editors), 2002. Bias and 149

information content of paleoflood data in flood-frequency analysis. Ancient floods, modern hazards, principles and applications of paleoflood hydrology, 5. American Geophysical Union, Water Science and Application Series, Washington, DC, 161-174 pp.

Board, C.W.C., 2006. Dolores/San Juan/San Miguel Basin Fact Sheet, pp. 2.

Bryson, R.A. and Hare, F.K., 1974. The climates of North America. Climates of North America, World Survey of Climatology II. Elsevier, New York.

Canon, J., Gonzalez, J. and Valdes, J., 2007. Precipitation in the Colorado River Basin and its low frequency associations with PDO and ENSO signals. Journal of Hydrology, 333(2-4): 252-264.

Cayan, D.R., Redmond, K.T. and Riddle, L.G., 1999. ENSO and hydrologic extremes in the western United States (vol 12, pg 2881, 1999). Journal of Climate, 12(12): 3516-3516.

Cook, E.R., Woodhouse, C.A., Eakin, C.M., Meko, D.M. and Stahle, D.W., 2004. Long- term aridity changes in the western United States. Science, 306(5698): 1015- 1018.

Denlinger, R.P., O’Connell, D.R.H. and House, P.K. (Editors), 2002. Robust determination of stage and discharge: an example from an extreme flood on the Verde River Arizona. Ancient Floods, Modern Hazards: Principles and Applications of Paleoflood Hydrology in Water Science and Application, 5. American Geophysical Union, Washington, D.C., 127-146 pp.

Ely, L.L., 1997. Response of extreme floods in the southwestern United States to climatic variations in the late Holocene. Geomorphology, 19(3-4): 175-201.

Ely, L.L. and Baker, V.R., 1985. Reconstructing paleoflood hydrolgoy with slackwater deposits – Verde River, Arizona. Physical Geography, 6(2): 103-126.

Ely, L.L., Enzel, Y., Baker, V.R. and Cayan, D.R., 1993. A 5000-year record of extreme floods and climate change in the southwestern United-States. Science, 262(5132): 410-412.

Ely, L.L., Enzel, Y., Baker, V.R., Kale, V.S. and Mishra, S., 1996. Changes in the magnitude and frequency of late Holocene monsoon floods on the Narmada River, central India. Geological Society of America Bulletin, 108(9): 1134-1148. 150

Ely, L.L., Enzel, Y. and Cayan, D.R., 1994. Anomalous North Pacific atmospheric circulation and large winter floods in the southwestern United-States. Journal of Climate, 7(6): 977-987.

Engineer, C.O.o.t.S., 1912. Biennial Report of the Engineer State of Colorado for the years, 1911-192. Office of the State Engineer, Denver.

Enzel, Y., Ely, L.L., Martinez-goytre, J. and Vivian, R.G., 1994. Paleofloods and a dam failure on the Virgin River, Utah and Arizona. Journal of Hydrology, 153(1-4): 291-315.

Freeman, W.B., 1909. Flood on the San Juan River September, 1909. Monthly Weather Review, 37: 648-649.

Friedman, J.M. et al., 2006. Transverse and longitudinal variation in woody riparian vegetation along a montane river. Western North American Naturalist, 66(1): 78- 91.

Fujita, K., 2007. Effect of dust event timing on glacier runoff: sensitivity analysis for a Tibetan glacier. Hydrological Processes, 21(21): 2892-2896.

Fuller, J.E., 1987. Paleoflood hydrology of the alluvial Salt River, Tempe, Arizona, pp. iv, 92 leaves (some unnum.).

Gavin, D.G., Brubaker, L.B. and Lertzman, K.P., 2003. Holocene fire history of a coastal temperate rain forest based on soil charcoal radiocarbon dates. Ecology, 84(1): 186-201.

Greenbaum, N., 2007. Assessment of dam failure flood and a natural, high-magnitude flood in a hyperarid region using paleoflood hydrology, Nahal Ashalim catchment, Dead Sea, Israel. Water Resources Research, 43(2).

Greenbaum, N., Enzel, Y. and Schick, A.P., 2001. Magnitude and frequency of paleofloods and historical floods in the Arava basin, Negev Desert, Israel. Israel Journal of Earth-Sciences, 50: 159–186.

Greenbaum, N., Margalit, A., Schick, A.P., Sharon, D. and Baker, V.R., 1998. A high magnitude storm and flood in a hyperarid catchment, Nahal Zin, Negev Desert, Israel. Hydrological Processes, 12(1): 1-+.

Greenbaum, N., Porat, N., Rhodes, E. and Enzel, Y., 2006a. Large floods during late 151

Oxygen Isotope Stage 3, southern Negev desert, Israel. Quaternary Science Reviews, 25(7-8): 704-719.

Greenbaum, N., Schick, A.P. and Baker, V.R., 2000. The palaeoflood record of a hyperarid catchment, Nahal Zin, Negev Desert, Israel. Earth Surface Processes and Landforms, 25(9): 951-971.

Greenbaum, N., Weisheit, J.S., Harden, T. and Dohrenwend, J.D. (Editors), 2006b. Paleofloods in the Upper Colorado River near Moab, Utah, May 2006. editors, 2006, The Moab Mill Project: A technical report towards reclaiming uranium mill tailings along the Colorado River in Grand County, Utah. Living Rivers, Moab, Utah. Living Rivers, Moab, UT.

Guilderson, T.P., Reimer, P.J. and Brown, T.A., 2005. The Boon and Bane of radiocarbon Dating. Science 307(5708): 4.

Hansen, E.M.S., F.K., 1981. Meteorology of important rainstorms in the Colorado River and Great Basin drainages. In: N.O.a.A.A.H. Report (Editor), pp. 167.

Hattingh, J. and Zawada, P.K., 1996. Relief peels in the study of palaeoflood slack-water sediments. Geomorphology, 16(2): 121-126.

Hereford, R. and Webb, R.H., 1992. Historic Variation of warm-season rainfall, southern Colorado Plateau, southwestern, USA. Climatic Change, 22(3): 239-256.

Hereford, R., Webb, R.H. and Graham, S., 2002. Precipitation History of the Colorado Plateau Region, 1900–2000. In: U.S.G. Survey (Editor), Reston.

Hidalgo, H.G. and Dracup, J.A., 2003. ENSO and PDO effects on hydroclimatic variations of the Upper Colorado River basin. Journal of Hydrometeorology, 4(1): 5-23.

Hirschboeck, K. (Editor), 1988. Flood hydroclimatology. Flood Geomorphology. John Wiley & Sons, 27‑49 pp.

Hirschboeck, K.K. (Editor), 1987. Catastrophic flooding and atmospheric circulation anomalies. Catastrophic Flooding. Allen & Unwin, 23‑56 pp.

Hirschboeck, K.K., 2003a. Floods, Paleofloods, and Drought: Insights from the Upper Tails, CLIVAR/PAGES/IPCC Drought Workshop. University of Arizona. 152

Hirschboeck, K.K., 2003b. Respecting the drainage divide: a perspective on hydrological change and scale: (126).

Hirschboeck, K.K., 2009. Future Hydroclimatology and the Research Challenges of a Post-Stationary World. 142: 4-9.

Hosking, J.R.M. and Wallis, J.R., 1986. Paleoflood hydrology and flood frequency- analysis. Water Resources Research, 22(4): 543-550.

House, P.K., 1997. An unconventional, multidisciplinary approach to evaluating the magnitude and frequency of flash floods in small desert watersheds. In: A.G. Survey (Editor). Open-File Report, Tucson, AZ, pp. 61.

House, P.K., 2002. Ancient floods, modern hazards : principles and applications of paleoflood hydrology. Water science and application. American Geophysical Union, Washington, DC, vi, 385 p. pp.

House, P.K. and Baker, V.R., 2001. Paleohydrology of flash floods in small desert watersheds in western Arizona. Water Resources Research, 37(6): 1825-1839.

House, P.K., Pearthree, P.A. and Klawon, J.E. (Editors), 2002. Historical flood and paleoflood chronology of the lower Verde River, Arizona: Stratigraphic complexity and related uncertainties. Ancient Floods, Modern Hazards: Principles and Applications of Paleoflood Hydrology, 5. American Geophysical Union, Washington, D.C., 267-294 pp.

Hupp, C.R. (Editor), 1982. Plant Ecological Aspects of Flood Geomorphology and Paleoflood History. Flood Geomorphology. Wiley.

Jarrett, R.D. and Tomlinson, E.M., 2000. Regional interdisciplinary paleoflood approach 153

to assess extreme flood potential. Water Resources Research, 36(10): 2957-2984.

Keen, R.A. (Editor), Weather and Climate. The Western San Juan Mountains: Their Geology, Ecology and Human History. University of Colorado Press., Boulder.

Klawon, J.E., Paleoflood hydrology and historic flood analysis in the Upper Verde River Basin, Central Arizona (MS), Tucson, Arizona University of Arizona, 1997.

Klawon, J.E. and Arizona Geological Survey., 1998. Paleoflood hydrology and historic flood analysis in the upper Verde River basin, central Arizona. Open-file report / Arizona Geological Survey. Arizona Geological Survey, Tucson, Ariz., 93 p. pp.

Knox, J.C., 1985. Responses of floods to climatic change in the Upper Mississippi Valley. Quaternary Research, 23(3): 287-300.

Knox, J.C., 1993. Large increases in flood magnitude in response to modest changes in climate. Nature, 361(6411): 430-432.

Knox, J.C., 2000. Sensitivity of modern and Holocene floods to climate change. Quaternary Science Reviews, 19(1-5): 439-457.

Knox, J.C. and Kundzewicz, Z.W., 1997. Extreme hydrological events, palaeo- information and climate change. Hydrological Sciences Journal-Journal Des Sciences Hydrologiques, 42(5): 765-779.

Kochel, R.C. and Baker, V.R., 1982. Paleoflood Hydrology. Science, 215(4531): 353- 361.

LaRue, E.C., 1925. Water power and flood control on the Colorado River below Green River, UT. U.S. Geological Survey water supply paper (556): 176 pp.

Magilligan, F.J., Goldstein, P.S., Fisher, G.B., Bostick, B.C. and Manners, R.B., 2008. Late Quaternary hydroclimatology of a hyper-arid Andean watershed: Climate change, floods, and hydrologic responses to the El Nino-Southern Oscillation in the Atacama Desert. Geomorphology, 101(1-2): 14-32.

Martinezgoytre, J., House, P.K. and Baker, V.R., 1994. Spatial variability of small-basin paleoflood magnitudes for a southeastern Arizona mountain-rang. Water Resources Research, 30(5): 1491-1501.

McCabe, G.J., Betancourt, J.L. and Hidalgo, H.G., 2007. Associations of decadal to multidecadal sea-surface temperature variability with Upper Colorado River flow. 154

Journal of the American Water Resources Association, 43(1): 183-192.

Meko, D.M. et al., 2007. Medieval drought in the upper Colorado River Basin. Geophysical Research Letters, 34(10).

Milly, P.C.D. et al., 2008. Climate change - Stationarity is dead: Whither water management? Science, 319(5863): 573-574.

Milly, P.C.D., Wetherald, R.T., Dunne, K.A. and Delworth, T.L., 2002. Increasing risk of great floods in a changing climate. Nature, 415(6871): 514-517.

Mock, C.J., 1996. Climatic controls and spatial variations of precipitation in the western United States. Journal of Climate, 9(5): 1111-1125.

Murray, A.S. and Wintle, A.G., 2000. Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol. Radiation Measurements, 32(1): 57-73.

Murray, A.S. and Wintle, A.G., 2003. The single aliquot regenerative dose protocol: potential for improvements in reliability. Radiation Measurements, 37(4-5): 377- 381.

Neff, J.C. et al., 2008. Increasing eolian dust deposition in the western United States linked to human activity. Nature Geoscience, 1(3): 189-195.

O'Connor, J.E., 1985. Paleohydrology of pool and riffle pattern development: Boulder Creek, Utah, pp. 70 leaves (some unnum.).

O'Connor, J.E. et al., 1994. A 4500-year record of large floods on the Colorado River in the Grand Canyon, Arizona. Journal of Geology, 102(1): 1-9.

O'Connor, J.E. and Webb, R.H. (Editors), 1988. Hydraulic modeling for paleoflood analysis. F. John Wiley and Sons, New York, 393-402 pp.

Orchard, K.L., 2001. Paleoflood hydrology of the San Juan River, southeastern Utah, USA, pp. 138 leaves.

Painter, T.H. et al., 2007. Impact of disturbed desert soils on duration of mountain snow cover. Geophysical Research Letters, 34(12).

Partridge, J. and Baker, V.R., 1987. Paleoflood Hydrology of the Salt River, Arizona. Earth Surface Processes and Landforms, 12(2): 109-125. 155

Pelletier, J.D. et al., 2005. An integrated approach to flood hazard assessment on alluvial fans using numerical modeling, field mapping, and remote sensing. Geological Society of America Bulletin, 117(9-10): 1167-1180.

Petersen, K.L., 1988. Climate and the Dolores River Anasazi, 113. University of Utah Press, Salt Lake City.

Petersen, K.L., 1994. Climatic Change. Climatic Change, 26(2-3): 243-269.

Porat, N., Zilberman, E., Amit, R. and Enzel, Y., 2001. Residual ages of modern sediments in a hyperarid region, Israel. Quaternary Science Reviews, 20(5-9): 795-798.

R.D., J. and J.F.J., E. (Editors), 2002. Reliability of Paleostage Indicators for Paleoflood Studies. Ancient floods, modern hazards: principles and applications of palaeoflood hydrology. American Geophysical Union, Washington, D.C., 111- 125 pp.

Redmond, K.T., Enzel, Y., House, P.K. and Biondi, F. (Editors), 2002. Climate variability and flood frequency at decadal to millenial time scales. in Ancient Floods, Modern Hazards: Principles and Applications of Paleoflood Hydrology. American Geophysical Union, 21-45 pp.

Reimer, P.J. et al., 2004. IntCal04 terrestrial radiocarbon age calibration, 0-26 cal kyr BP. Radiocarbon, 46(3): 1029-1058.

Reimer, P.J. et al., 2009. INTCAL09 and MARINE09 radiocarbon age calibration curves, 0-50,000 years Cal B.P. Radiocarbon, 51(4): 1111-1150.

Rittenour, T.M., Goble, R.J. and Blum, M.D., 2003. An optical age chronology of Late Pleistocene fluvial deposits in the northern lower Mississippi valley. Quaternary Science Reviews, 22(10-13): 1105-1110.

Rittenour, T.M., Goble, R.J. and Blum, M.D., 2005. Development of an OSL chronology for Late Pleistocene channel belts in the lower Mississippi valley, USA. Quaternary Science Reviews, 24(23-24): 2539-2554. 156

Saint-Laurent, D., 2004. Palaeoflood hydrology: an emerging science. Progress in Physical Geography, 28(4): 531-543.

Seager, R., Kushnir, Y., Herweijer, C., Naik, N. and Velez, J., 2005. Modeling of tropical forcing of persistent droughts and pluvials over western North America: 1856- 2000. Journal of Climate, 18(19): 4065-4088.

Seager, R. et al., 2007. Model projections of an imminent transition to a more arid climate in southwestern North America. Science, 316(5828): 1181-1184.

Sheppard, P.R., Comrie, A.C., Packin, G.D., Angersbach, K. and Hughes, M.K., 2002. The climate of the US Southwest. Climate Research, 21(3): 219-238.

Sigafoos, R.S., 1961. Vegetation in relation to flood frequency near Washington, D. C. . In: U.S.G. Survey (Editor). Professional Paper pp. 248-249.

Stedinger, J.R. and Cohn, T.A., 1986. Flood frequency analysis with historical and paleoflood information. Water Resources Research, 22(5): 785-793.

Stuiver, M. and Reimer, P.J., 1993. Extended C-14 data-base and revised Calib 3.0 C-14 age calibration program. Radiocarbon, 35(1): 215-230.

Tanaka, T.Y. and Chiba, M., 2006. A numerical study of the contributions of dust source regions to the global dust budget. Global and Planetary Change, 52(1-4): 88-104.

Thorndycraft, V.R. and Benito, G., 2006a. Late Holocene fluvial chronology of Spain: The role of climatic variability and human impact. Catena, 66(1-2): 34-41.

Thorndycraft, V.R. and Benito, G., 2006b. The Holocene fluvial chronology of Spain: evidence from a newly compiled radiocarbon database. Quaternary Science Reviews, 25(3-4): 223-234.

Thorndycraft, V.R. et al., 2005. A long-term flood discharge record derived from slackwater flood deposits of the Llobregat River, NE Spain. Journal of Hydrology, 313(1-2): 16-31.

Trenberth, K.E., Dai, A., Rasmussen, R.M. and Parsons, D.B., 2003. The changing character of precipitation. Bulletin of the American Meteorological Society, 157

84(9): 1205.

Webb, R.H., Blainey, J.B. and Hyndman, D.W. (Editors), 2002. Paleoflood hydrology of the Paria River, southern Utah and northern Arizona, USA. Ancient floods, modern hazards, principles and applications of paleoflood hydrology, 5. American Geophysical Union, Washington, DC, 295-310 pp.

Webb, R.H. and Jarrett, R.D. (Editors), 2002. One-dimensional estimation techniques for discharges of paleofloods and historical floods. Ancient floods, modern hazards, principles and applications of paleoflood hydrology, 5. American Geophysical Union, Water Science and Application Series, Washington, DC, 111-125 pp.

Webb, R.H., O'Connor, J.E. and Baker, V.R. (Editors), 1988. Paleohydrologic reconstruction of flood frequency on the Escalante River, south-central Utah. Flood geomorphology. John Wiley and Sons, New York, 403-418 pp.

Westerling, A.L., Hidalgo, H.G., Cayan, D.R. and Swetnam, T.W., 2006. Warming and earlier spring increase western US forest wildfire activity. Science, 313(5789): 940-943.

Wintle, A.G. and Murray, A.S., 2006. A review of quartz optically stimulated luminescence characteristics and their relevance in single-aliquot regeneration dating protocols. Radiation Measurements, 41(4): 369-391.

Wise, E.K., 2010. Spatiotemporal variability of the precipitation dipole transition zone in the western United States. Geophysical Research Letters, 37.

Wohl, E.E., 1998. Uncertainty in flood estimates associated with roughness coefficient. Journal of Hydraulic Engineering-Asce, 124(2): 219-223.

Wohl, E.E., 2000a. Inland flood hazards : human, riparian, and aquatic communities. Cambridge University Press, Cambridge, U.K.; New York, xiv, 498 p., 4 p. of plates pp.

Wohl, E.E., 2000b. Inland flood hazards: human, riparian, and aquatic communities. Cambridge University Press, Cambridge, U.K.; New York, xiv, 498 p., 4 p. of plates pp. 158 Woodhouse, C.A., 1997. Winter climate and atmospheric circulation patterns in the Sonoran Desert region, USA. International Journal of Climatology, 17(8): 859- 873.

Woodhouse, C.A., Gray, S.T. and Meko, D.M., 2006. Updated streamflow reconstructions for the Upper Colorado River Basin. Water Resources Research, 42(5).

Woodhouse, C.A., Kunkel, K.E., Easterling, D.R. and Cook, E.R., 2005. The twentieth- century pluvial in the western United States. Geophysical Research Letters, 32(7).

Yanosky, T.M., 1982. Hydrologic inferences from ring widths of flood-damaged trees, Potomac River, Maryland. Environmental Geology, 4(1): 43-52.

UCRB

DRB

Figure 1. Location map of the Dolores River Basin (DRB) within the Upper Colorado River Basin (UCRB). 109°30'0"W 109°0'0"W 108°30'0"W 108°0'0"W 107°30'0"W

Elevation (m) Big Rock High : 4000 Big Alcove

Tufoni Low : 1200 38°30'0"N 38°30'0"N Mucky

Tributary

38°0'0"N 38°0'0"N

Juniper

37°30'0"N McPhee 37°30'0"N

109°30'0"W 109°0'0"W 108°30'0"W 108°0'0"W 107°30'0"W

FigureFigure 2. 2. Paleoflood Location deposits of sites the within Dolores the Dolores River River Basin Basin SWD study sites 160 161

Hf9 USU-627 1.11 ± 0.17 ka

x x

MC080507-03 ** Colluvial 877-955 B.P. MC031809-05 ** 14 Hf7 1250--1302 C Sample x detrital * MC031809-06 ** 1266--1311 OSL Sample in-situ **

Figure 3. Detail of stratigraphy at the Big Rock SWD.

Placer 14 and C Sample x organic OSL Sample materials

x USU-332 Hf9 0.99 ± 0.08 Ka MC080507-07 x x 0--30

USU-331 1.44 ±0.18 ka MC080507-05 1110--1210 Hf8 MC052908-01 1110--1220 (both in-situ)

Figure 4. Big Alcove SWD site with detailed stratigraphy shown, as well as sample locations. 162

detrital * in-situ **

14 C Sample x OSL Sample

MC031809-02 * Hf11 752--973 Flood (?) Colluvial Hf10 x Flood (?)

Colluvial Hf9 USU-626 1.10 ±0.13 Ka Hf5

Hf(7 or 8) USU-330 x 2.94 ±0.32 ka MC031809-03 * Hf6(?) 1849--2071

Figure 5a. Tafoni SWD site with detialed startigraphy shown, including sample locations. An additional unit (Hf12) is not shown here, but is location is shown in Figure 5b. 163 164

0 Hf11

15 Hf10 23

30 Silty sands Hf9 40 Medium sands Hf12 Hf7 or 8 Fine Sands 55

Rip up clasts

OSL sample Hf6 14C sample

Bedrock contact 90

Figure 5b. Statigraphic detail of the Tafoni SWD. A bedrock contact exists between the main statigraphic section, and a younger onlapped unit to the right (i.e. Hf12) 14 C Sample x detrital * in-situ ** x

MC081409-02a* xx 1593--1695 x Hf7 Hf7a--e MC081409-02b * (?) 1471--1568 x MC111108-04 * MC081509-01 * 1428--1533 1575--1772

MC111108-03 * 1963--2142 x x MC111108-02 * x MC081409-03** 8414--8825 Hf6 Hf6a,b 1678--1841 MC110808-02b * (?) 2122--2338

Figure 6. Mucky Alcove SWD. Detailed stratigraphy and sample locations are shown. 165 Bedrock x MC111008-01 * Hf7 x 1356--1470 B.P.

MC081409-03 * 1678--1841 B.P.

Hf6

x MC111008-02 * 1411--1516 B.P.

14 C Sample x OSL Sample

detrital *

in-situ **

Figure 7. McPhee Alcove SWD. Detailed statigraphy and sample locations are shown. 166

detrital * in-situ **

14 C Sample x OSL Sample

Hf3 USU-628 7.84 ka ± 0.66

Hf2

x MC051909-04 * 7942--8017 x Hf2 USU-629 6.23 ka ± 0.47 MC051909-02 * Hf1 8818--8880

Figure 8. Juniper SWD site. Detailed stratigraphy and sample locations are shown. 167 Hf3

x USU-630 8.21 ± 0.79 ka Hf2 mc052009-02 7690--8040 (detrital)

14 C Sample x OSL Sample

Figure 9. Tributary SWD site shown, as well as the sample location. Only the upper 4m are shown here. 168 169

San Miguel R. Tafoni

Lower Hf11 Mucky Dolores R. Hf10 Big Rock

Hf7,8 ? Hf7 Big Alcove Hf9

Hf7,8 ? Hf9 Hf6 Hf6 Hf8

Juniper Tributary McPhee ? ? ? ?

Hf7 Hf3 Hf8

Hf6

? ? ? ? Hf2

? ? ? ? Hearth stones colluvial or slope-wash Biturbated sediment massive sands Silt lines fine sands with silt Hf1 Bedrock bedded sands ? ? ? ? soil or peleosol Upper Dolores R. Figure 10. Fence diagram illustrating correlative food units. Each river reach is within its own box. Note that the sections are not drawn to scale. Clear divisions between flood units are shown as solid lines and unclear divisions are idicated by dashed lines. 110

108 1053: peak paleoflood 1020 988 106 962 934 254: peak historical flood 104 96: mean annual flood bed surface meters above abitrary datum meters above 102

100 0 38 70 103 125 168 218 meters upstream of abitrary datum Figure 11. Discharge estimates (m3s-1) from the Tafoni SWD site. Estimates are based on one-dimensional HEC-RAS results. Significant floods are described. Peak paleoflood stage is more than double the peak historical flood, and more than four-times the discharge. 170 2000-years, B.P. from Elyetal.,(1993)and Ely, (1997).Notethebreakintimescalepriorto the DRB(shownasagradient).Datausedtoproduce thegradientwastaken rectangles), comparedwiththeperiodsofincreased large magnitudefloodingin Figure 12.PeriodsoffloodingidentifiedintheDRB watershed(shownasred 9000 * Hf1

indicatesalinearbreakinthetimescale 8000 High Low Hf2 Hf3

Individual floodperiodsintheDRB 7000 Ely, 1997) Periods ofheightened,large magnitude flooding (basedonElyetal.,1993;

6000

5000

4000 calibrated years B.P. 3000 Hf5 * 2000 Hf6b Hf6a

Hf7a Hf7b Hf7c Hf7d Hf7e

Hf10 Hf8 Hf9 Hf11 1000 Hf12 Hf13

0 171 Flood envelope curve, Colorado River Basin 105 San Miguel paleoflood (peak) San Miguel gaged flood (peak) 4 10 Dolores gaged flood (peak) ) -1 s 3 103

102 peak discharge (m peak discharge

10 Flodds described in this study Colorado River flood envelope LCRB Paleofloods

1 1 10 102 103 104 105 106 area (km2)

Figure 13. Flood envelope curve for the Colorado River Basin (reproduced from Enzell et al., 1993). The locations of the Dolores and San Miguel River peak historical floods are shown by the grey shaded circles. The peak paleoflood on the San Miguel River shown by the large black shaded circle. Other paeoflood discharge estimate are shown by the black shaded squares. 172 Table 2. Calbrated calendar year ranges (BP) corrected delta relative relative median Sample code SWD site Lab code Description 14C age +/- 13C lower upper probability lower upper probability probability 080507-03 Big Rock AA82500 charcoal 991 37 -24.7 803 971 1 913 080507-05 Big Rock AA82502 charcoal 1226 38 -22 1060 1255 1 1160 031809-05 Big Rock AA85261 charcoal 1337 36 -20.7 1202 1323 1 1273 031809-06 Big Rock AA85262 charcoal 1359 36 -24.3 1229 1335 1 1288 080507-06 Big Alcove AA82503 charcoal 267 36 -21.9 280 456 0.992 328 080507-07 Big Alcove AA82504 charcoal 162 68 -25.4 -3 302 1 163 080507-08 Big Alcove AA82505 charcoal 139 37 -22.6 9 272 0.995 148 052908-01 Big Alcove AA82506 charcoal 1231 38 -21 1066 1259 1 1166 081509-05 Big Alcove AA86680 charcoal 1271 34 -22.4 1122 1279 1 1213 081509-06 Big Alcove AA86681 charcoal 947 33 -21.1 781 933 1 864 081509-07 Big Alcove AA86682 wood 130 33 -22 54 150 0.432 174 276 0.402 130 111008-01 McPhee AA84122 charcoal 1526 53 -22.2 1325 1528 1 1418 111008-02 McPhee AA84123 charcoal 1578 45 -21.5 1370 1560 1 1465 081409-03 McPhee AA86674 charcoal 1822 67 -17.2 1585 1897 1 1753 111108-02 Mucky AA84126 charcoal 7800 200 -30.2 8199 9153 1 8659 111108-03 Mucky AA84127 charcoal 2089 70 -21.2 1904 2204 0.914 2213 2293 0.086 2063 111108-04 Mucky AA84128 charcoal 1596 43 -26.9 1383 1573 1 1483 110808-01-a Mucky AA85263 charcoal 3322 62 -29.6 3412 3690 1 3555 110808-02-b Mucky AA85264 charcoal twig 2212 97 -29.8 1947 2368 0.978 2388 2402 0.005 2210 031809-X5 Mucky AA85267 charcoal 1342 50 -20.8 1164 1342 1 1272 081509-01 Mucky AA86675 charcoal 1750 38 -27.4 1566 1763 1 1665 081509-02-a Mucky AA86676 charcoal 1735 38 -21.8 1555 1743 1 1648 081509-02-b Mucky AA86677 wood 1626 38 -12.1 1408 1603 1 1520 081509-03 Mucky AA86678 charcoal 1821 39 -27.3 1653 1854 1 1758 111108-05 Mucky AA84129 charcoal 121 82 -25.6 -4 292 1 147 111108-06 Mucky AA84130 charcoal 2680 64 -26.7 2710 2935 1 2798 031809-01 Alluvial reach AA85257 wood 169 34 -26.6 129 230 0.52 -3 37 0.189 174 031809-02 Tafoni AA85258 charcoal 893 51 -24.2 717 919 1 813 031809-03 Tafoni AA85259 charcoal 1995 94 -26.5 1713 2182 0.996 1955 051909-01 Juniper AA86018 charcoal 34900 0 -27.4 inf inf na 0 051909-02 Juniper AA86019 charcoal 8033 55 -24.2 8695 9059 1 8913 051909-04 Juniper AA86021 charcoal 7153 48 -25.7 7881 8065 1 7979 052009-02 Tributary AA86024 charcoal 7060 190 -25.5 7561 8270 1 7892 all calibrations listed are 2 sigma 173 relative probabilities <0.16 were ommited Table 1. # Equivalent dose Dose Rate USU # Sample # location aliquots* (De), Gy (Gy/ka) OSL Age (ka)

USU-330 MC-052908-OSL-01 lower San Miguel: Tafoni 23 (35) 7.29 ± 1.61 2.48 ± 0.10 2.94 ± 0.321 USU-331 MC-052908-OSL-02 lower Dolores: Big alcove 26 (38) 3.47 ± 0.79 2.41 ± 0.09 1.44 ± 0.182 USU-332 MC-052908-OSL-03 lower Dolores: Big alcove 25 (38) 2.35 ± 1.09 2.37 ± 0.09 0.99 ± 0.082 USU-626 MC-031809-OSL-01 lower Dolores: Big rock 22 (28) 2.58 ± 0.54 2.35 ± 0.10 1.10 ± 0.132 USU-627 MC-031809-OSL-02 lower San Miguel: Tafoni 21 (30) 2.89 ± 0.87 2.60 ± 0.11 1.11 ± 0.172 USU-628 MC-051909-OSL-01 upper Dolores: Juniper 27 (38) 21.83 ± 4.23 2.79 ± 0.12 7.84 ± 0.661 USU-629 MC-051909-OSL-02 upper Dolores: Juniper 33 (40) 14.52 ± 2.24 2.33 ± 0.10 6.23 ± 0.471 USU-630 MC-052009-OSL-04 upper Dolores: Tributary 24 (41) 20.79 ± 5.53 2.53 ± 0.12 8.21 ± 0.791

*Number*Number ofof aliquotsaliquots used used for for age age calculation, calculation, number number of aliquots of aliquots measured measured in parentheses. in parentheses. AgeAge analysisanalysis using using the the SAR SAR procedure procedure of Murrayof Murray and and Wintle Wintle (2000) (2000) on quartz on quartz sand usingsand using1-mm 1-mmaliquots, aliquots, 220o C 220preheats.o C preheats. O.D.Error Overdispersion on De is 1 σ ,represents error on errorage includeson De beyond random instrumental and systematic error. errors caluclated in quadrature. 1 OSL age calculated using the Central Age Model (CAM) of Galbraith et al. (1999). Error on De is 1 σ, error on age includes random and systematic errors caluclated in quadrature. 2 OSL age calculated using the mimimum age model (MAM) of Galbraith et al. (1999). 1 OSL age calculated using the Central Age Model (CAM) of Galbraith et al. (1999). 2 OSL age calculated using the mimimum age model (MAM) of Galbraith et al. (1999). 174 175

APPENDIX C

LUMINESCENCE DATING OF HOLOCENE PALEOFLOOD DEPOSITS, DOLORES WATERSHED, CO AND UT

Michael Cline

176

Abstract

Optical ages for middle-late Holocene floods were compared with accompanying

accelerator mass spectrometry radiocarbon (AMS 14C) ages to assess the application of optically stimulated luminescence (OSL) to Paleofloods. Results from 7 of 8 of the optical ages provided a coherent and stratigraphically consistent chronology that was corroborated with AMS 14C independent age control. Optical ages were derived using

the single aliquot regenerative dose (SAR) protocol. This study was performed to

provide a case study in the application of SAR optically stimulated luminescence (OSL)

for paleoflood hydrology—an application that has seen mixed results in the past. Results

from this study demonstrate that applying appropriate age models should generally

provide acceptable results using standard SAR protocols on small aliquots of quartz sand.

177

1. Introduction

Fluvial geomorphic response to climate change remains a critical question, particularly as it pertains to a warming planet. The impressions that rivers leave on the earth’s surface provide us with critical information that relates climate and the fluvial system (e.g. Bull,

1991). In the context of extreme hydrological events, paleoflood hydrology has proven invaluable for illustrating the relationships between climatic variability and flood frequency (e.g. Ely et al., 1993; Ely, 1997; House and Hirschboeck, 1997; Knox, 1993;

Knox, 2000). The most critical element allowing us to relate the occurrence of large floods with climate variability over geological timescales is the quality of absolute age control; it allows for the bracketing of time periods when extreme flooding has been more or less prevalent. The chronologies can then be superimposed onto existing paleoclimate records which can form the basis that underlies our current understanding the relationship between extreme flooding and climatic shifts (e.g. Ely et al., 1993; Ely, 1997; Webb and

Betancourt, 1992; Knox 1993; Knox et al., 2000). Advances in traditional chronometric techniques such as accelerator mass spectrometry radiocarbon (AMS 14C), and developments and advances of newer techniques such as cosmogenic radionuclides and optically stimulated luminescence (OSL) have advanced our ability to develop more robust age constraints for fluvial deposits (Baker, 2008). Unlike 14C, which relies on

contextual relationships, OSL explicitly dates the age of deposition for fluvial sediments,

a particularly important aspect at locations where dateable organic material is either not

present, or uninterpretable (e.g. Greenbaum et al., 2001; Thomas et al., 2007).

Significant advances in luminescence techniques have been made in recent years that are

178

increasingly allowing for accurate dating of fluvial sediments (e.g. Murray and Roberts,

1998; Murray and Wintle, 2000; Olley et al. 1998; Duller et al. 1999). There are still

questions, however, regarding the application of OSL to paleoflood hydrology, especially

for relatively young sediments (middle-late Holocene) (Olley et al., 1998). This study

presents results from eight OSL and accompanying AMS 14C ages and from middle to early Holocene flood deposits in the Dolores Watershed in southern Colorado. With few exceptions (i.e. Porat et al., 2001; Greenbaum et al., 2006; Kale et al. 2000; Thomas et al., 2007; Greenbaum et al., 2001; Jaiswal et al., 2009; Zha et al., 2009), OSL has seen limited use in paleoflood studies, and results have been mixed. This study provides a case study, testing the application of OSL dating in an arid to semi-arid river systems against independent age control (AMS 14C). Included is a detailed discussion of the

potential pitfalls associated with OSL dating, specifically as it pertains to its application

to paleoflood hydrology. For thorough technical discussions of the OSL process and its

many applications, see review papers by Rittenour (2008), Wallinga (2002), and Lian and

Roberts (2006). For a detailed flood chronology of the Dolores Watershed, stemming

from this study, see appendix B of this dissertation.

2. Background

The field of paleoflood hydrology has been largely dependent on radiocarbon dating for

developing flood chronologies. Radiocarbon dating made the development of absolute

age control of fluvial deposits possible. For more than half a century, the technique has

seen continual refinement, and modern AMS techniques allow for the dating of

179

minuscule amounts of carbon rich materials (less than 1.0 mg) (Baker, 2006). The ages

derived from 14C require the presence of organic material in an interpretable context

(Delong and Bailey, 2007); however, even when it is interpretable, significant landscape residence times may be undetectable, which may lead to inaccurate chronologies. Other post-depositional, biochemical factors may corrupt the ages as well (e.g. Gillespie et al.,

1992). While analytical errors associated with AMS 14C are minimal (Burr, 2007), issues

associated with converting radiocarbon ages to calendar years remains a critical source of

uncertainty because the atmospheric 14C production rate is not constant, nor spatially

homogenous (Stuiver and Quay, 1981). The conversion has been shown to produce

centennial-scale errors in some cases (Guilderson et al., 2005). These concerns all

suggest a need to derive robust age control using multiple techniques, and that stand-

alone radiocarbon derived ages may not be adequate. A clear alternative, one that may

address the issues associated with 14C dating is OSL, because OSL requires no contextual organic material (as is often the case in arid environments); it relies on the use of quartz and feldspar grains—two of the most common sediments in the world; it requires no conversions to calendar years because the OSL age is calculated as calendar years (B.P.); and the quality of the OSL age can usually be directly assessed. The application of OSL to fluvial studies has seen continual refinements, and recent studies have reported excellent results (e.g. Rittenour, 2005; Jain et al., 2004); however, early studies lacked confidence in the derived ages (e.g. Olley et al. 1998; Olley et al., 1999). Numerous issues have impeded using OSL for fluvial processes. Some are inherent to the atomic properties of the mineralogy, other are inherent to the sediment transport and depositional

180

processes. This study focuses on the latter.

The luminescence process

To understand the factors that influence OSL ages, the luminescence process must be understood. Lian and Roberts (2006) provide a detailed explanation that is summarized here. Optically stimulated luminescence refers to the small but measurable amount of light (i.e. luminescence) that is released from mineral grains after sufficient exposure to visible light. The luminescence that is released after light exposure is limited by the amount of ionization energy trapped in the crystal lattice of quarts and feldspar grains, so additional light exposure will not result in additional luminescence. The natural defects in the mineral grains accommodate the storage free electrons (i.e. the luminescence charge), which are produced when alpha, beta and gamma particles from the natural radioactive decay of the mineral grains, and cosmic rays are emitted (i.e. ionizing radiation). The defects, or ‘traps’ store the electrons, and require differing amounts of visible light exposure to evict them. When the sediment is hidden from light, the luminescence charge increases by continually intercepting more electrons until either the traps become saturated, or the sediment is exposed to light, subsequently evicting the charge. The luminescence process in the fluvial system is fundamentally based upon the assumption that grains are adequately exposed to light during transport, and subsequently removed from light upon burial, which resets the luminescence ‘clock’ (Aitken, 1998).

The analytical OSL technique measures the sediment’s luminescence signal response to a

181

beam of light (either a light emitting diode for single aliquot (SA) techniques, or a laser for single grain (SG) measurements). The luminescence signal results from the eviction of the luminescence charge that is stored within the traps, and the intensity of the luminescence is a function of the amount of charge that is stored. Following the eviction of free electrons from the original, natural dose (that which is collected in the field), the individual aliquots (small disks with tens-hundreds of individual grains) are irradiated with a known radioactive source, and then re-exposed to the light source. This process is repeated to ultimately match the natural luminescence signal, thereby informing us of the actual stored radiation in the grains. This study utilizes the single aliquot regenerative dose (SAR) protocol that is described in detail by Murray and Wintle, (2000; 2003) and most recently by Wintle and Murray (2006). It is generally described in our methodology.

Pitfalls in OSL dating

The greatest issue facing luminescence dating of fluvial sediments is the potential for insufficient resetting of the luminescence ‘clock’ by light (e.g. Godfrey-Smith et al.,

1988; Wallinga, 2002). Zeroing, or fully bleaching sediments in fluvial systems is a complicated issue. In fluvial environments, the bleaching of quartz and feldspar occurs during transport in daylight conditions. Complete zeroing of sediments can range from seconds to hours (Figure 1, recreated from Wallinga, 2002; Berger and Luternauer, 1987;

Spooner 1994a,b). The time-span of zeroing is a function of local conditions, and the inherent properties of the minerals. These conditions are described in more detail below.

182

In the event that sediment is deposited prior to complete zeroing, erroneously old ages will result because of the residual dose that is trapped in the minerals of interest. In fluvial environments, the degree of sediment exposure to light is a function of a number of conditions. First, transport distance plays a role; long transport tends to mix grains more homogeneously, for longer periods of time, which allows for greater potential for light exposure. The depth of sediment in the water column plays a role as well, because of the light attenuating properties of water, and the preferential wavelength attenuation of sediment loaded water (Berger and Luternauer, 1987; Godfrey-Smith et al., 1988;

Spooner, 1994a,b Wallinga, 2002). Fluvial processes in high latitude regions of the world, where solar radiation is limited intensity and daylight hours are highly seasonally dependent present unfavorable conditions for zeroing of sediments. It is also feasible that the azimuthal trend of deeply entrenched canyon reaches may limit light exposure, requiring longer transport times or high sun angles to bleach sediments.

Inherent properties of the grains affect the bleaching potential as well. Bleaching appears to be grain size dependent, and counter-intuitively, the medium—coarse sand fraction seems to bleach more readily (e.g. Olley et al., 1998; Truelsen and Wallinga, 2003). The responsible mechanisms for this are poorly understood, although Rittenour (2008) postulated that coarse grains have longer total transport times allowing more significant light exposure, whereas silts and very fine sands are easily transported long distances, or to their final base levels under very average flow conditions. It is also likely that they have undergone more erosional cycles. Medium—coarse sands are transient on the

183

channel bars and channel margins, so continual erosional episodes, punctuated by short- term storage will eventually zero sediments. As a final explanation, fine sands and silts may be transported as aggregates, limiting their total light exposure (Fuchs, 2005;

Kadereit, 2000; Lang, 1996).

The zeroing of grains is mineralogy dependent. Several important studies have worked to identify the bleaching characteristics of quartz and feldspar. In highly sediment charged rivers (minimal light attenuation), the feldspars have shown to bleach more rapidly

(Spooner 1994b). The use of quartz grains is more appropriate however, due to the inherent instability of trapped charge in the feldspar grains (Huntley and Lamothe, 2001;

Wallinga et al., 2000). The use of quartz grains was further supported by a study of modern flood sediments (Fuchs et al., 2005). Although no grains had been thoroughly zeroed, the bleaching characteristics of the quartz grains was notably greater, limiting the age errors to 0.1 ka—1.0 ka (as opposed to 1.0 ka—10 ka for feldspar derived ages), suggesting the use of Quartz OSL is acceptable for catastrophic flood studies.

In addition to the standard concerns of zeroing fluvial sediments, other inherent issues exist for paleoflood studies. It has even been suggested that the application of OSL on flood deposits is of limited accuracy (e.g. Porat et al., 2001). Floods are generally short lived, sediment travels at high velocities and the water is highly turbulent. The high shear stress that is exerted by floods creates highly sediment charged rivers that limit the transmission of light (Berger and Luternauer, 1987; Wallinga, 2002b; Kale et al., 2000).

184

High shear stress exerted by floods lead to channel bank and channel bed erosion, adding mixed-age sediment populations (Rittenour, 2009). Older sediments that are stored in fluvial terraces, only affected by extreme flood stages, require longer periods to zero, adding to the mixed-age grain populations. The high shear stress may also erode primary bedrock, whose luminescence properties are often very weak (e.g. Porat et al. 2001). In areas where summer advective storm dominate flooding in the watershed, it is not uncommon for peak flooding to occur during the night, in which case no solar resetting may occur.

Young samples have posed considerable issues to fluvial OSL studies (Jain et al. 2004).

In paleoflood studies, common temporal scales range from middle—latest Holocene, requiring careful analysis. Young samples are particularly problematic because minor variations in dose residuals propagate into large errors (Rittenour, 2009). A possible advantage to using OSL on slackwater deposits associated with large floods is that young samples probably require relatively medium sands to effectively derive depositional ages.

As described above, the coarser fraction of fluvial sediment (medium to coarse sand) is more transient in channel bars and channel banks, which increases the probability of undergoing numerous erosional cycles. This would limit the residual dose stored, so even incomplete zeroing will lead to adequate age control in most settings.

Floods are episodic in nature. Long hiatus’s in the flood record are usually punctuated by periods of frequent flooding, depending on the climatic conditions that persist (e.g. Ely et

185

al., 1993). The episodic nature of paleofloods presents challenge because of potentially rapid changes in the environmental dose rate and the cosmogenic dose rate caused by rapid burial of SWD units may effectively shut down, or at a minimum, reduce the cosmogenic input. Subsequent erosion causes changes to the rates as well. The changes in both the terrestrially derived dose rates (e.g. Li et al., 2008) and cosmogenic dose

(Munyikwa, 2000) have been shown to erroneously shift the OSL ages of sediments.

Figure 2 illustrates this conceptually.

There are two possible types of partial bleaching according to Duller, (1994) (referenced in Wallinga, 2002): 1) when all the grains are homogeneously partially bleached and, 2) when the grains are heterogeneously bleached. The first type is highly unlikely. To satisfy this case, a unique set of events is required for the same relic charge to be present at the time of burial (Wallinga, 2002). All of the sediment that is deposited must be sourced from sediments of equal age, e.g. from a single stratigraphic unit, a highly unlikely scenario during extreme flooding. To occur, during transport, all grains would receive equal light exposure, and all grains would have to be equally sensitive to light.

As suggested by Wallinga (2002) it is widely accepted that this first scenario is not feasible, and only the second scenario is highly probable. The second case

(heterogeneity) is the focus on most studies working to identify partial bleaching in grains.

Inter-grain variation of the OSL signal is the greatest issue relevant to identifying partial

186

bleaching. High variance in the inter-grain variation easily identifies partial bleaching of grains (Wallinga 2002). Another means of identifying partially bleached grains is by the shape decay curve (‘shine-down’ curve) of the luminescence signal. There are deep medium and shallow traps, and deep traps are thought to be the most difficult to bleach, whereas the shallow traps tend to me less stable. The decay curve can be modeled with three components that identify the fast, medium and slow components of the OSL decay curve (Bailey et al. 1997). If the deepest traps return the anticipated shine-down profile, they were likely well-bleached prior to burial. For a complete review of the means of detecting partially bleached grains, see Wallinga (2002).

Bioturbation can pose significant problems to OSL dating as well (Bateman et al. 2007).

While this is not truly a form a partial bleaching, it produces scatter in the results similar to partial bleaching; although, the scatter tends to skew in the positive direction because of the addition of younger grains in the sediment. Translocation of individual grains by flora and fauna from overtopping flood units, into underlying units may introduce younger grains to the sample. For this reason, sampling sediments that may have been bioturbated should be done with great care.

Modern techniques such as single grain (Olley et al. 1998; Duller et al. 1999), and the

SAR protocol (Murray and Roberts, 1998; Murray and Wintle, 2000) have addressed many of these issues however, allowing for assessments partial bleaching and the degree of bioturbation. For the SAR protocol, single aliquots are used, systematically reducing

187

their size with each sample run. The use of small aliquots (<100 grains) can alleviate age discrepancies between SG and SA techniques because small aliquots allow for the identification of partially bleached samples. Large aliquots obscure partially bleached aliquots by averaging many grains within a single aliquot. Numerous statistical treatments have been developed for specific types of deposits, each with their own advantages (e.g. Rhodes et al., 2003; Olley et al. 1998; Lepper et al., 2000; Galbraith et al., 2005; Galbraith et al., 1999; Bailey and Arnold, 2006). The statistical treatment that is applied is dependent on the mechanism that causes the scatter (Bailey and Arnold,

2006; Arnold et al., 2007; Rittenour, 2005).

OSL in Paleoflood studies

Numerous paleoflood studies that have used OSL for developing chronologies have highlighted some of the methods pitfalls and uncertainty. A paleoflood study using infrared stimulated luminescence (IRSL) in the southern Arava Valley, Israel, discovered issues of sediment immaturity (i.e. eroded from bedrock for the first time), where the only well bleached sediments were located in areas where numerous cycles of erosion had taken place (Porat et al. 2001). Another study used IRSL and SAR OSL to develop a flood chronology in the southern Negev, Israel. SAR OSL techniques showed too much scatter to be of any use, suggesting that significant partial bleaching was hindering the results, either due to sediment immaturity or limited exposure to light. The ages derived from IRSL in the study, showed age reversals and large errors, although minimal scatter was observed suggesting that bleaching was readily occurring. The age reversals and

188

analytical errors made individual flood units indistinguishable, and the derived chronology was relegated to being of very low-resolution. Thomas et al., (2007) demonstrated a higher success in southern India using 19 samples, and analyzed them with both SAR and SG protocols. The samples were determined to be well-bleached, however, the study demonstrated that the SAR technique produced ages that were systematically younger than those derived from SG. Delong and Arnold (2007) found opposing results in a fluvial stratigraphy study in southern California, especially for samples younger than 1 ka. In the Thomas et al., (2007) study, the resultant younger

SAR ages were explained by the grains sensitivity to the wavelength of light stimulation.

In the Delong and Arnold (2007) study, the older SAR derived ages were explained by a polluting of the multi-grain aliquots with partially bleached grains.

3. Methods

Field Site Description

The location of the Dolores Watershed is shown in Figure 3. The watershed drains the southern Uncompahgre Plateau and the western San Juan Mountains providing the

Colorado River with a mean discharge of ~20 m3s-1: roughly ten-percent of its annual discharge at their confluence. The drainage basin encompasses 11,860 km2 and is divided into two major tributaries: the San Miguel and upper Dolores Rivers. Both rivers meander through deeply incised, (San Juan geology, upper Dolores and San Miguel) and Mesozoic sedimentary rocks (lower Dolores); Pliocene-early Pleistocene gravels (Hunt, 1969); and

189

Holocene, vertically accreting, flood terraces. At the confluence of the two rivers, the

Dolores River continues though incised, canyon topography until its confluence with the

Colorado River, ~26 km north of Moab, UT. The rugged topography causes steep precipitation and temperature gradients within the Dolores Watershed. The San Miguel

River has a long history of placer gold and silver mining, as well as Vanadium mining, and subsequently two Environmental Protection Agency (EPA) Superfund Sites line the river: the Uravan (EPA, 2000) and the Idarado Mine Natural Resource Damage Site

(EPA, 1995).

Paleoflood field techniques

Paleoflood hydrology utilizes indexical markers on the landscape to identify the stage of given flood (Baker, 2006). The majority of paleostage indicators (PSIs) that line the

Dolores and San Miguel rivers are fine-grained slackwater deposits (SWDs). SWDs form when flow velocities during a flood are reduced below the critical shear stress to maintain sediments in suspension (Kochel and Baker, 1982; Baker, 1987). This usually occurs in eddies caused by obstructions in the channel form, rapid channel expansions, or in back-flooded tributaries. Some SWDs are preserved on the landscape due to their emplacement in area protected from subsequent erosion. This work has historically relied on the use of radiocarbon derived age control to contextualize the channel-bed dynamics. Slackwater deposits are the most common indexes used, and indicate the minimum flood stage. They form in areas where flow velocities are reduced beneath the critical threshold that allows sediments to fall out of suspension (Kochel and Baker,

190

1982; Baker, 1987). Identifying SWDs requires time in the field, and sometimes the use of aerial imagery for reconnaissance work. There long-term preservation is relatively rare due to subsequent erosion following floods. We located numerous slackwater deposits (SWDs) along the Dolores and San Miguel Rivers. Figure 4 shows the locations of the deposits. After identifying the SWDs in the field, their stratigraphy was described in detail. Within each named SWD, we applied a flood unit-naming scheme, Hf1-Hf12, where the Hf1 unit represents the oldest flooding period, and the younger periods are represented by larger numbers. Each Hfn is representative of a specific flood event. In this study, only those with OSL samples collected are described. The naming scheme is based upon the scheme used for the larger study of the region.

OSL field Data Collection

All sediment samples that we analyzed for OSL were collected in either 1.5 inch or 2- inch metal conduit tubing, cut to 8 inches in length to ensure that samples would not be exposed to any light during collection or transport. The collection tubes were hammered horizontally into freshly exposed sediment, aiming for original bedding structures.

Emphasis was placed on avoiding heavily bioturbated areas, clay lenses, bedrock and we aimed for a minimum distance of 10 cm away from depositional or erosional contacts.

Avoiding these increases the chance for a uniform dose rate at the sample location since time of burial. We also collected a representative sample within a 30 cm radius around the sample for analysis of the environmental dose rate. Representative samples are critical, because they are used to calculate the OSL age directly. The OSL age is

191

calculated by the following equation: D age = e D r

Where the age (ka) is equal to the quotient of the equivalent dose (De) and the ! -1 environmental dose rate (Dr) measured in grays ka . Laboratory measurements of the luminescence determine the equivalent dose, whereas in-situ field measurements using a gamma spectrometer, or laboratory geochemical analysis are required to determine the environmental dose rate. The environmental dose rate measurement were performed at

CHEMEX labs in Sparks, NV using inductively coupled plasma mass spectroscopy (ICP-

MS) and inductively coupled plasma atomic emission spectroscopy (ICP-AES) for concentrations of U, K, Th, and Rb: the primary sources for ionization radiation in sediments. 3% and 5% errors were assumed for chemical data dose rate conversions

(Olley and Murray, 2002) and subsample dose rate heterogeneity, respectively.

Cosmogenic dose rates were estimated based on burial depth below the original geomorphic surfaces, latitudinal position and elevation (Prescott and Hutton, 1994).

Estimations of water content in the sample since the time of burial were made due to the attenuating properties of water on radionuclides.

Luminescence measures

The SAR OSL dating technique has significant advantages over older thermoluminescence (TL) and multiple aliquot OSL techniques in that individual age estimates are obtained from each aliquot of sample allowing for the identification of

192

incomplete bleaching. Additionally, the complete bleaching of grains (i.e. zeroing) can occur relatively quickly with OSL, in comparison to TL, which takes hours to reset. This is particularly important for fluvially transported sediment.

The Utah State University Luminescence Lab is equipped with the latest generation Risø

OSL Dating System (Model Risø TL/OSL-DA-20) with a single-grain attachment that allows for the dating of individual sand grains. The Risø instrument is fully automated with heating and irradiation steps conducted internally, has a 48-sample holding chamber, and comes with its own data analysis software developed by Risø National Labs in

Denmark. For Luminescence measurements, we followed the latest SAR procedures for dating quartz sand (Murray and Wintle, 2000, 2003; Wintle and Murray, 2006). Sample processing followed the procedures outlined in Aitken (1998) and described in detail in

Rittenour et al. (2003, 2005). Sample tubes were opened under dim amber safelight conditions within the lab. The outer two inches of sediment in the sample tubes were removed to limit the possibility of modern light contamination. Samples were wet-sieved to a narrow grain-size range, either 75-125 µm or 90-150 µm, treated with HCl to remove carbonate materials, and then bleach to remove organic matter. Samples were then separated by gravity, using a heavy liquid and a centrifuge. The remaining grains were then etched 3 µm—6 µm with HF acid in order to remove effects of alpha irradiation.

Initial measurements to bracket the equivalent dose were done with 5 mm aliquots of the coarsest grain size fractions of 75-125 and 90-150 µm quartz-isolate sands. Final measurements were made on 1 mm aliquots because they are approaching SG accuracy

193

and allow more precise determinations of bleaching. The purity of the samples is

checked by measurement with infrared stimulation to detect the presence of feldspar.

The SAR protocol includes tests for sensitivity correction and brackets the equivalent

dose (De) that the sample received during burial by irradiating the sample at five different

doses (i.e. below, at, and above the De, plus a zero dose and a repeated dose to check for

recuperation of the signal and sensitivity correction) using a Strontium-90 beta source.

The resultant data are fit with a saturating exponential curve from which the equivalent

dose is determined. A minimum of 23 accepted aliquots were used for final age

estimations, although many more aliquots were used (up to 40), and many were rejected

based upon acceptance criteria (Rittenour 2005): recuperation ratios >10% were

excluded due to changes in sensitivity from progressive irradiation of aliquots; feldspar

contaminated aliquots were rejected as well (based on IRSL); samples with weak signal

to noise ratios (natural response is too close to the background response) were rejected;

finally, aliquots that had higher De than the highest regenerative dose (i.e. fitting

parameters) were rejected. De values were determined using a saturating exponential growth curve, and De errors were calculated as standard errors.

AMS 14C field data collection

We collected mostly charcoal from SWDs where it was interpretable. We preferentially collected in-situ charcoal, and clustered charcoal fragments because the clusters indicate minimal fluvial reworking (Gavin et al., 2003). Charcoal was generally sparse however, in the form of small flecks (see table 2).

194

14C sample processing

All pre-treatment and AMS 14C analysis was conducted at the AMS Laboratory at the

University of Arizona. We mechanically removed visible contaminants form the samples, then used a standard acid-base-acid pre-treatment protocol to remove the carbonates and humates from samples to remove contaminants, thus minimizing critical issues associated with radiocarbon dating (e.g. Gillespie et al., 1992). Combustion steps for the extraction of CO2 and graphite target preparation were performed in a low- pressure vacuum, and then final AMS analysis of graphite targets was performed.

Radiocarbon ages were converted to calendar ages using the Calib 6.0 software (Stuiver and Reimer, 1993), which utilizes the latest Intcal98 calibration standards by Reimer et al., (2009).

4. Results

Eight SAR OSL ages were analyzed for their stratigraphic accuracy, and 7 returned promising results that allowed us to constrain the ages of flood deposition throughout the middle to late Holocene, at a reasonably high resolution. Table 1 shows the OSL ages, and table 2 shows the AMS 14C samples used in this study. Subsequent figures (figures

5-9) show the stratigraphic context of the OSL samples, as well as their accompanying

AMS 14C ages. Corroborating AMS 14C ages allowed for further refinement of the ages of the OSL samples, and provided a basis for determining the best age models for the young (<2.0 ka B.P.) sediments, and older sediments as well. While any OSL age within

195

the Holocene is generally considered relatively young, it is the latter Holocene sediments

that pose considerable issues to the OSL technique because of the proportionally larger

impact that residual doses have on aliquots with smaller De’s (i.e. younger samples).

We applied one of two age models to determine the final age of the flood deposits: the minimum age model (MAM) and the central age model (CAM) (Galbraith et al., 1999).

These models, particularly the MAM are relatively sophisticated and these have been well-tested (e.g. Bailey and Arnold, 2006; Arnold et al., 2009). They provide a

reasonably robust approach to dealing with aliquot scatter, and this study provides an

additional case study to validate their application in the use of OSL dating of paleoflood

SWDs. The CAM essentially calculates the mean De of the aliquots, weighted by the error of the individual aliquots. The CAM is appropriate for samples with relatively little scatter in the De distributions. The MAM is more sophisticated: De is estimated by a

truncated log normal distribution where the truncation point determines the De that is

most representative of only the well-bleached grains, leading to a more accurate age in

poorly bleached samples (Galbraith et al., 1999; Bailey and Arnold, 2006). Selection of

the model in this study was done analytically, using the ‘over-dispersion‘ parameter

(Olley, 2004; Bailey and Arnold, 2006). Over-dispersion is calculated as the fraction of

the De spread that is not accounted for by the individual De scatter. If all of the De scatter

is accounted for by the analytical error, then the over-dispersion parameter in zero. In the

event that over-dispersion approaches 40%, it can be assumed that the sample is partially

bleached, and the MAM model should be used; if significantly <40%, the CAM model

196

should be applied (for SA, not SG, Olley, 2004). Figure 10 shows the probability density

functions (PDFs) of the De distributions, with error bars, for each aliquot used in each sample. The relative spread of the aliquot’s De, relative to their error determined which model was selected.

Stratigraphic results

When applying the correct age model, OSL ages were mostly stratigraphically consistent in this study and the independent age control provided by the AMS 14C corroborated the

OSL ages. Figures 5-9 show the stratigraphic context of the OSL and AMS 14C collection sites. At each site, stratigraphic consistency is evident, regardless of the bleaching characteristics of the samples. Figure 11 shows a fence diagram for all of the sites, relative to their longitudinal channel position.

Big Rock and the Big Alcove SWDs: these two sites lay within 2.8 km of each other, and they contain two correlative flood units. One OSL and one charcoal sample were collected at the downstream site—the Big Rock SWD (Figure 5). A flood unit associated with the Hf9 flood period represents the upper part of the SWD. Here, OSL sample

USU-626 was collected, and an in-situ charcoal sample was collected in the colluvial unit that underlies it. The charcoal sample (MC080507-03) returned an age that was slightly younger than the OSL sample, which we calculated using the MAM. Numerous known factors have been shown to impact the AMS 14C ages (e.g. Gillespie, 1992), and the ages in the figures and text have been reported to 1σ. Regardless of the known errors, the

197

corroboration between the two ages provides a robust age determination, and combined,

they provide a maximum age constraint on the flood. The AMS age suggests the ‘older‘

range of the OSL sample is likely irrelevant, since the charcoal was burned in-situ,

making it an age maxima on the underlying colluvial unit. Based on these data, the flood

likely occurred on the younger tail of the OSL age distribution, <1.0 ka B.P. At the Big

Alcove SWD, two OSL samples were collected and analyzed, and compared with three

charcoal samples (Figure 6). In the upper, younger flood unit, which is likely correlative

with the Hf9 unit at the Big Rock site, an OSL sample (USU-332) returned a near

identical age, using the MAM, as the Big Rock SWD downstream, providing an

additional corroborating age for Hf9. The OSL samples associated with the Hf9 period at

the Big Rock and Big Alcove sight have differing degrees of partial bleaching. The

USU-627 sample has an over-dispersion of 47%, and the PDF (Figure 10) clearly shows

a mixed population of aliquot Des. USU-332, however, is a well-bleached sample, with only 13% over-dispersion. In the flood unit, which underlies the Hf9 unit at the Big

Alcove SWD, an OSL sample (USU-331) was corroborated by two charcoal samples that were burned in-situ. This particular flood deposit (Hf8) is highly altered by human activity, evidenced by fire-shattered river stones, and presence of a hearth. The OSL sample we collected (USU-331) was within a part of the Hf8 unit with preserved bedding, suggesting it was undisturbed, whereas the charcoal was collected from a part of the unit that was clearly disturbed, so in-situ charcoal fragments were expected to return younger ages than the OSL sample from the same unit. As anticipated, the older Hf8 unit is several hundred years younger, with an age estimate close to ~1.4 ka B.P.

198

Tafoni SWD: Up to seven flood units are preserved at this site, two of which were sampled for OSL. This is the site of the most numerous individual floods preserved, and the stratigraphic detail is shown in Figure 7. The oldest at this location, the Hf5 unit was sampled for OSL (USU-627) was calculated using the CAM, due to its modest over- dispersion. No organic material was identifiable within the unit in any interpretable context, so no independent age control is available to corroborate the OSL age (2.9 ka

B.P.). Nevertheless, it is stratigraphically consistent. A younger flood unit is likely correlative with Hf9 was collected here, and compared with two detrital charcoal samples

(MC031809-02, MC031809-03): one located in a younger, overtopping unit, and the other, within the same flood unit as the OSL sample (USU-626). The OSL sample age is nearly identical to the Hf9 samples at the Big Rock and Big Alcove sites, which allows us to identify a fairly distal correlative SWD in this case. The detrital charcoal samples provide corroborating ages as well, but do little to further constrain the floods age. The

USU-626 sample was over dispersed (57%), suggesting it was partially bleached so we applied the MAM.

Juniper and Tributary SWDs: The Juniper SWD (Figure 8) showed age reversals in the

OSL and AMS 14C chronology in the unit associated with the Hf3 and Hf2 periods of flooding. In the stratigraphically older, Hf2 unit: a sample age of 7.84 ± 0.66 (USU-628 in Hf3) is stratigraphically positioned directly above a sample age of 6.23 ± 0.47 ka

(USU-629 in Hf2). An AMS 14C age in the Hf2 returned an age very similar to the USU-

199

628 OSL sample in Hf3, which may lead one to argue that the two units may have been deposited during the same flood. The abrupt stratigraphic change in color and texture, accompanied by a thin slope wash unit provided ample evidence to suggest the stratigraphic break was produced by two, distinct floods, as opposed to a single, multi- phase flood. In addition, the AMS 14C age from the upper unit is detrital in origin, so it represents a maxima age for the deposition of the sediment, and its similarity to the lower

OSL sample is probably coincidental. In both cases, the CAM model was applied. The over-dispersion was relatively low, at 26.8% (USU-628) and 17.1 (USU-629) (table 1).

The application of the MAM resulted in similar reversals and artificially young ages.

Several other sources of error are possible at this site, including the presence of ‘dim’ aliquots that may offset the age of the sample toward a younger range. Another possible source of error is the presence of substrates or materials that may force variability of the environmental dose rate. For example, variability in ground water (groundwater seeping, etc.) could produce temporal inhomogeneity in the dose rate (Li et al., 2008). Spatial inhomogeneity is possible as well: if large boulders that are buried within the deposit provide differential dose rates, omitting the source would pose issues to the age control.

A laboratory analysis of the dose rate may not account for the inhomogeneous source if its location were not field-sampled. A field-based, beta dosimeter may provide an additional source of information that would rule out this possibility, or provide a more accurate assessment of the dose rate.

At the Tributary SWD, the Hf2 and (likely) the Hf3 flood units are preserved (Figure 9),

200

so it provides a correlative flood stratigraphy to the Juniper site. One OSL and one AMS

14C sample were analyzed here. The AMS 14C is based on a very small fleck of detrital charcoal. Very little charcoal was available at this site, and all of it is detrital. The two ages agree well, where the OSL sample (USU-630) and the AMS 14C sample returned

ages roughly within 200-300 years of each other. The detrital charcoal actually returned

an age slightly younger, although the CAM (over-dispersion 33%) was applied to the

OSL sample, so it may slightly overestimate the age of the flood. Still, overall agreement

was excellent, and may provide a better constraint on the problematic results from the

Juniper site.

5. Discussion

The flood chronology derived in this study, provided a surprisingly coherent chronology,

suggesting its application to paleoflood studies will see increasingly promising results, at

least in study regions with similar physiographic characteristics as the Dolores River

Basin. The age quality seems to be highly dependent on the statistical model selection.

As both Galbraith (2005) and Arnold et al. (2009) have cautioned, however, the selection

of an appropriate statistical model should maintain a physical basis—an understanding of

the sediment transport and depositional regime that would impact the bleaching

characteristics, and ultimately De of individual grains.

The primary considerations for paleoflood workers in developing chronologies are as

follows:

201

1) Partially bleached or mixed De distributions should be anticipated, do in part to

poor light-penetration of the water column (e.g. Berger and Luternauer, 1987).

Statistical models that preferentially weight the aliquots or grains with lower Des

should be applied in many occasions. For example, the MAM (Galbraith et al., 1999)

was used in this study, and it returned a strongly coherent chronology. There is a

physical basis this presumption. Variability in the sediment source may impact the

inter-aliquot and inter-grain scatter because extreme floods exert enough energy on

the channel bed and channel margins to erode sediments in regions of long-term

storage, so residual doses may be significant.

2) Very fine sands and silt should be avoided, where possible. In light-limited

environments (e.g. deeply entrenched canyons or night time flooding), a reasonable

age might still be attained using the MAM. Coarser grains, those that are closer to the

average stream competence should be used because they are likely to move slowly

through the system. This relative grain size is more likely to temporarily store in

channel bars under mean flow conditions, allowing for numerous transport and

bleaching cycles ranging from days, to a number of years. The residual age

associated with the sample would be so miniscule in this event, that it would do little

to offset the burial age since the residuals would be nearly undetectable.

This study has demonstrated that partial bleaching limits the application of OSL only minimally; however, we suspect this is true on a case-by-case basis. While OSL is does

202

not provide the same level of precision as AMS 14C, the user can determine the quality of the dates with careful analysis. One significant advantage of OSL, at this point is that it remains in a relatively experimental stage in fluvial studies compared with radiocarbon dating. Because of this, nearly every published study utilizing OSL for fluvial studies investigates the degree of bleaching. These discussions provide the readers a greater sense of transparency for evaluating each age on a case-by-case basis. The direct dating of depositional age can provide a higher level of confidence for the selection process of

AMS 14C ages, and limit the false age estimates that are occasionally provided by reworked organic samples. Furthermore, the technique allows for dating much older deposits, beyond the typical 40 ka age of 14C, and can date deposits where there is no organic material preserved (Greenbaum et al., 2001). Ideally, future paleoflood studies with ample funding will apply multiple, independent age control, which can be used to both, corroborate, and further constrain the age of burial.

6. Conclusions

This research was undertaken to provide a case study for the application of OSL for use in paleoflood studies. Results were promising and should provide information to future workers. SAR protocols and use of the MAM model was able to isolate the well- bleached grains for analysis. The MAM places more emphasis on grains that are determined to contain a minimal residual dose, although this requires an assumption that partial bleaching is an issue that pertains to SWD sediments. In particular, the MAM was applied to young deposits because greater residual doses in the aliquots introduce a

203

proportionally greater error than it does for older samples. In older samples, the CAM model was applied, simply because residual doses are less disruptive to the calculated age. The SAR protocol provided ample evidence for identifying partially bleached grains, particularly when small aliquots (1.0 mm) were used. It is feasible that SG techniques may be necessary for very young (<500 years) SWDs, although this study did not identify a need to use the method. Due to the emplacement of SWDs near bedrock, and other potentially inhomogeneous sources of dose rates, a well-calibrated field dosimeter may be necessary to better capture spatial dose rate heterogeneity. These results should provide an informative framework for future SWD paleoflood studies, particularly in basins with mature, quartz-rich sediments, such as the Colorado Plateau.

204

REFERENCES

Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford Science Publications, 267 pp.

Arnold, L.J., Bailey, R.M. and Tucker, G.E., 2007. Statistical treatment of fluvial dose distributions from southern Colorado arroyo deposits. Quaternary Geochronology, 2(1-4): 162-167.

Arnold, L.J., Roberts, R.G., Galbraith, R.F. and DeLong, S.B., 2009. A revised burial dose estimation procedure for optical dating of young and modern-age sediments. Quaternary Geochronology, 4(4): 306-325.

Bailey, R.M. and Arnold, L.J., 2006. Statistical modelling of single grain quartz D-e distributions and an assessment of procedures for estimating burial dose. Quaternary Science Reviews, 25(19-20): 2475-2502.

Baker, V.R., 1987. Paleoflood hydrology and extraordinary flood events. Journal of Hydrology, 96(1-4): 79-99.

Baker, V.R., 2006. Paleoflood hydrology in a global context. Catena, 66(1-2): 161-168.

Baker, V.R., 2008. Paleoflood hydrology: Origin, progress, prospects. Geomorphology, 101(1-2): 1-13.

Bateman, M.D. et al., 2007. Detecting post-depositional sediment disturbance in sandy deposits using optical luminescence. Quaternary Geochronology, 2(1-4): 57-64.

Berger, G.W. and Luternauer, J.J., 1987. Preliminary field work for thermoluminescence dating studies at the Fraser River delta, British Columbia. . Geological Survey of Canada Paper 87/IA, : 901–904.

Bull, W.B., 1991a. Geomorphic responses to climatic change. Oxford University Press, New York, xviii, 326 p. pp.

Bull, W.B., 1991b. Geomorphic responses to climatic change. Oxford University Press, New York, xviii, 326 p. pp.

DeLong, S.B. and Arnold, L.J., 2007. Dating alluvial deposits with optically stimulated luminescence, AMS 14C and cosmogenic techniques, western Transverse Ranges, California, USA. Quaternary Geochronology, 2(1-4): 129-136.

Duller, G.A.T., 1994. Luminescence dating of poorly bleached sediments from Scotland. Quaternary Science Reviews, 13(5-7): 521-524.

Ely, L.L., 1997. Response of extreme floods in the southwestern United States to climatic

205

variations in the late Holocene. Geomorphology, 19(3-4): 175-201.

Ely, L.L., Enzel, Y., Baker, V.R. and Cayan, D.R., 1993. A 5000-year record of extreme floods and climate change in the southwestern United-States. Science, 262(5132): 410-412.

Galbraith, R.F., Roberts, R.G. and Yoshida, H., 2005. Error variation in OSL paleodose estimates from single aliquots of quartz: a factorial experiment. Radiation Measurements, 39(3): 289-307.

Gavin, D.G., Brubaker, L.B. and Lertzman, K.P., 2003. Holocene fire history of a coastal temperate rain forest based on soil charcoal radiocarbon dates. Ecology, 84(1): 186-201.

Gillespie, R. et al., 1992. AMS dating of alluvial sediments on the Southern Tablelands of New-South-Whales, Australia. Radiocarbon, 34(1): 29-36.

Godfrey-Smith, D.I., Huntley, D.J. and Chen, W.H., 1988. Optical dating studies of quartz and feldspar sediment extracts. Quaternary Science Reviews, 7(3-4): 373- 380.

Guilderson, T.P., Reimer, P.J. and Brown, T.A., 2005. The Boon and Bane of radiocarbon Dating. Science 307(5708): 4.

Hunt, C.B., 1969. Geologic history of the Colorado River: US Geol. In: U.S.G. Survey (Editor). Survey Prof. Paper.

Huntley, D.J. and Lamothe, M., 2001. Ubiquity of anomalous fading in K-feldspars and the measurement and correction for it in optical dating. Canadian Journal of Earth Sciences, 38(7): 1093-1106.

Jain, M., Tandon, S.K. and Bhatt, S.C., 2004. Late Quaternary stratigraphic development in the lower Luni, Mahi and Sabarmati river basins, western India. Proceedings of the Indian Academy of Sciences-Earth and Planetary Sciences, 113(3): 453-471.

206

Jaiswal, M.K., Chen, Y.G., Kale, V.S. and Achyuthan, H., 2009. Residual luminescence in quartz from slack water deposits in Kaveri Basin, South India: a single aliquot approach. Geochronometria, 33: 1-8.

Kadereit, A., 2000. IR-OSL-datierte Kolluvien als Archive zur Rekonstruktion anthropogen bedingter Landschaftsveränderungen. Ibidem-Verlag, Stuttgart

Kale, V.S., Singhvi, A.K., Mishra, P.K. and Banerjee, D., 2000. Sedimentary records and luminescence chronology of Late Holocene paleofloods in the Luni River, Thar Desert, northwest India. Catena, 40(4): 337-358.

Knox, J.C., 1993. Large increases in flood magnitude in response to modest changes in climate. Nature, 361(6411): 430-432.

Knox, J.C., 2000. Sensitivity of modern and Holocene floods to climate change. Quaternary Science Reviews, 19(1-5): 439-457.

Kochel, R.C. and Baker, V.R., 1982. Paleoflood Hydrology. Science, 215(4531): 353- 361.

Lang, A. 1996. Die Infrarot-Stimulierte Lumineszenz als Datierungsmethode für holozäne Lössderivate. PhD thesis, University of Heidelberg, 137pp.

Lepper, K., Larsen, N.A. and McKeever, S.W.S., 2000. Equivalent dose distribution analysis of Holocene eolian and fluvial quartz sands from Central Oklahoma. Radiation Measurements, 32(5-6): 603-608.

Li, B., Li, S.H. and Wintle, A.G., 2008. Overcoming environmental dose rate changes in luminescence dating of waterlain deposits. Geochronometria, 30: 33-40.

Lian, O.B. and Roberts, R.G., 2006. Dating the Quaternary: progress in luminescence dating of sediments. Quaternary Science Reviews, 25(19-20): 2449-2468.

Munyikwa, K., 2000. Cosmic ray contribution to environmental dose rates with varying overburden thickness. Ancient TL, 18: 27-34.

Murray, A.S. and Roberts, R.G., 1998. Measurement of the equivalent dose in quartz using a regenerative-dose single-aliquot protocol. Radiation Measurements, 29(5): 503-515.

Murray, A.S. and Wintle, A.G., 2000. Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol. Radiation Measurements, 32(1): 57-73.

Murray, A.S. and Wintle, A.G., 2003. The single aliquot regenerative dose protocol: potential for improvements in reliability. Radiation Measurements, 37(4-5): 377- 381.

207

Olley, J.M., Caitcheon, G.G. and Roberts, R.G., 1999. The origin of dose distributions in fluvial sediments, and the prospect of dating single grains from fluvial deposits using optically stimulated luminescence. Radiation Measurements, 30(2): 207- 217.

Olley, J.M. et al., 2004. Optical dating of deep-sea sediments using single grains of quartz: a comparison with radiocarbon. Sedimentary Geology, 169(3-4): 175-189.

Prescott, J.R. and Hutton, J.T., 1994. Cosmic-ray contributions to dose-rates for luminescence and ESR dating - large depths and long-term time variations. Radiation Measurements, 23(2-3): 497-500.

Reimer, P.J. et al., 2009. INTCAL09 and MARINE09 radiocarbon age calibration curves, 0-50,000 years Cal B.P. Radiocarbon, 51(4): 1111-1150.

Rittenour, T.M., 2008. Luminescence dating of fluvial deposits: applications to geomorphic, palaeoseismic and archaeological research. Boreas, 37(4): 613-635.

Spooner, N.A., 1993 The Validity of Optical Dating Based on Feldspar. PhD dissertation, Oxford University. 207 pp.

Spooner, N.A., 1994. On the optical dating signal from quartz. Radiation Measurements, 23(2-3): 593-600.

Stuiver, M. and Quay, P.D., 1981. A 1600-year-long record of solar change derived from atmospheric 14C levels. Solar Physics, 74(2): 479-481.

Stuiver, M. and Reimer, P.J., 1993. Extended 14C data-base and revised Calib 3.0 14C age calibration program. Radiocarbon, 35(1): 215-230.

Thomas, P.J., Juyal, N., Kale, V.S. and Singhvi, A.K., 2007. Luminescence chronology of late Holocene extreme hydrological events in the upper Penner River basin,

208

South India. Journal of Quaternary Science, 22(8): 747-753.

Truelsen, J.L. and Wallinga, J., 2003. Zeroing of the OSL signal as a function of grain size: investigating bleaching and thermal transfer for a young fluvial sample. Geochronometrica, 22: 1-8.

Wallinga, J., 2002. Optically stimulated luminescence dating of fluvial deposits: a review. Boreas, 31(4): 303-322.

Wallinga, J., Murray, A. and Wintle, A., 2000. The single-aliquot regenerative-dose (SAR) protocol applied to coarse-grain feldspar. Radiation Measurements, 32(5- 6): 529-533.

Zha, X.C., Huang, C.C. and Pang, J.L., 2009. Paleofloods recorded by slackwater deposits on the Qishuihe River in the Middle Yellow River. Journal of Geographical Sciences, 19(6): 681-690.

209

10 2.5 Light intensity (W m 2.0

1 1.5

1.0 -2 0.1 nm Bleaching efficiency Quartz -1 ) Feldspar 0.5 Light spectrum at the surface Light spectrum at 4 m depth 0.01 400 500 600 700 Wavelength (nm) Figure 1. (Modified from Wllainga, 2002; Berger and Luternauer, 1987; Spooner, 1994a,b). Light spectrum as a function of depth and its relative impact on 4 bleaching efficiency. Light spectrum at 4m depth has been multiplied by 4x10 210

Flood event Measured D (Gy) (light exposure) e Burial Sample collection (Gy)

e B D

Episodic flood events Measured D (Gy) e Incomplete bleaching

0 1 2 Time Figure 2. Conceptual model of (a) idealized bleaching and dose rate; (b) conceptual model of realistic bleaching and episodic changes in the dose rate through time, as it pertains to extreme flood deposition. Differences result in chanages in the measured De values. Vertical lines represent (time=1) initial bleaching/burial resulting from a large flood; (time=2) sample collection point in time. 211

UCRB

DRB

Figure 3. Location map of the Dolores River Basin (DRB) within the Upper Colorado River Basin (UCRB). 212

109°30'0"W 109°0'0"W 108°30'0"W 108°0'0"W 107°30'0"W

Elevation (m) BiBig g Rock Roc k

High : 4000 BiBig g Alcove Al c ove

Tuf oni Tafoni Low : 1200 38°30'0"N 38°30'0"N

TrTibutary i but a r y

38°0'0"N 38°0'0"N

JJuniper uni pe r

37°30'0"N 37°30'0"N

109°30'0"W 109°0'0"W 108°30'0"W 108°0'0"W 107°30'0"W

Figure 4. Location of the OSL SWD study sites in the DRB. Solid green dots indicate the locations of other SWDs. 213

USU-627 Hf9 1.11 ± 0.17 ka

x MC080507-03 Colluvial 880-960 (in situ)

older flood unit 14 C Sample x OSL Sample

Figure 5. Detail of stratigraphy and sample collection points for the Big Rock SWD.

Placer 14 and C Sample x organic OSL Sample materials

x USU-332 0.99 ± 0.08 Ka Hf8 x MC080507-07 x 0--30

USU-331 1.44 ±0.18 ka MC080507-05 Hf8 1110--1210 MC052908-01 1110--1220 (both in-situ)

Figure 6. Detail of stratigraphy and sample collection points for the Big Alcove SWD. 214 14 C Sample x OSL Sample Younger MC031809-02 flood units 750--970 x (detrital)

USU-626 Hf9 1.10 ±0.13 Ka

Hf5

Older USU-330 flood units 2.94 ±0.32 ka MC031809-03 1850--2070 x (detrital)

Figure 7. Detail of stratigraphy and sample collection points for the Tafoni SWD. 215 216

Hf3 USU-628 7.84 ± 0.66 ka

MC051909-04 x 7940--8020 (detrital) Hf2 USU-629 6.23 ± 0.47 ka

14 C Sample x OSL Sample Figure 8. Detail of stratigraphy and sample collection points for the Juniper SWD. 14 C Sample x Hf3 OSL Sample

x USU-630 8.21 ± 0.79 ka Hf2 mc052009-02 7690--8040 (detrital)

Figure 9. Detail of stratigraphy and sample collection points for the Tributary SWD. 217 USU-330 MC-052908-OSL-01 USU-331 MC-052908-OSL-02 USU-332 MC-052908-OSL-03 USU-626 MC031809-OSL-01 14.0

12.0

10.0

8.0

6.0

-5 0 5 10 15 20 -5 -3 -1 1 3 5 7 9 11 13 15 -2 0 2 4 6 8 -4 -2 0 2 4 6 8 10 12 14 De (Gy) De (Gy) DeDe (Gy) (Gy) DeDe (Gy) (Gy)

USU-627 MC031809-OSL-02 USU-628 MC051909-OSL-01Cumulative Probability Curve USU-629 MC051909-OSL-02 USU-630 MC-052009-OSL-04

-4 -2 0 2 4 6 8 10 12 14 16 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 De (Gy) De (Gy) De (Gy) De De(Gy) (Gy) Figure 10. OSL Sample probability density functions of the aliquot De distributions. dashed curve=Average of weighted De’s solid curve=Sum of weighted De’s 218 219

San Miguel R. Lower Tafoni Dolores R. Big Rock

Big Alcove 1.10 Ka ± 0.13

? 1.11 ka Hf9 ±0.17 0.99 Ka ±0.08 Hf9 Hf8 Juniper ? ? ? ? Tributary 1.44 ka ±0.18

7.84 ka Hf8 ±0.66

Hf3 ?

6.23 ka Hf2 ±0.47 8.21 ka ± 0.79

? ? ? ?

? ? ? ? Upper Dolores R.

Hearth stones colluvial or slope-wash Biturbated sediment massive sands Silt lines fine sands with silt Bedrock bedded sands soil or peleosol Figure 11. Fence diagram illustrating correlative food units. Each river reach is within its own box. Note that the sections are not drawn to scale. Clear divisions between flood units are shown as solid lines and unclear divisions are idicated by dashed lines. Table 1. # Equivalent dose Dose Rate USU # Sample # location aliquots* (De), Gy (Gy/ka) OD (%) OSL Age (ka) USU-330 MC-052908-OSL-01 lower San Miguel: Tafoni 23 (35) 7.29 ± 1.61 2.48 ± 0.10 33.2 2.94 ± 0.321 USU-331 MC-052908-OSL-02 lower Dolores: Big alcove 26 (38) 3.47 ± 0.79 2.41 ± 0.09 46.8 1.44 ± 0.182 USU-332 MC-052908-OSL-03 lower Dolores: Big alcove 25 (38) 2.35 ± 1.09 2.37 ± 0.09 12.7 0.99 ± 0.082 USU-626 MC-031809-OSL-01 lower Dolores: Big rock 22 (28) 2.58 ± 0.54 2.35 ± 0.10 56.8 1.10 ± 0.132 USU-627 MC-031809-OSL-02 lower San Miguel: Tafoni 21 (30) 2.89 ± 0.87 2.60 ± 0.11 46.9 1.11 ± 0.172 USU-628 MC-051909-OSL-01 upper Dolores: Juniper 27 (38) 21.83 ± 4.23 2.79 ± 0.12 26.8 7.84 ± 0.661 USU-629 MC-051909-OSL-02 upper Dolores: Juniper 33 (40) 14.52 ± 2.24 2.33 ± 0.10 17.1 6.23 ± 0.471 USU-630 MC-052009-OSL-04 upper Dolores: Tributary 24 (41) 20.79 ± 5.53 2.53 ± 0.12 33.1 8.21 ± 0.791

depth grain size cosmic USU # Sample # (m) (µm) H20 % K2O % Rb2 0 % Th (ppm) U (ppm) (Gy/ka) USU-330 MC-052908-OSL-01 1.30 75-125 2.0 2.16 ± 0.05 60.6 ± 2.4 2.8 ± 0.3 1.3 ± 0.1 0.23 ± 0.02 USU-331 MC-052908-OSL-02 0.40 75-125 0.4 1.82 ± 0.05 56.0 ± 2.2 4.3 ± 0.4 1.4 ± 0.1 0.26 ± 0.03 USU-332 MC-052908-OSL-03 0.15 75-150 0.8 2.06 ± 0.05 59.2 ± 2.4 3.1 ± 0.3 0.8 ± 0.1 0.27 ± 0.03 USU-626 MC-031809-OSL-01 0.40 90-150 3.0 2.10 ± 0.05 53.1 ± 2.1 2.5 ± 0.2 1.1 ± 0.1 0.27 ± 0.03 USU-627 MC-031809-OSL-02 0.70 90-150 3.0 1.95 ± 0.05 61.9 ± 2.5 5.6 ± 0.5 1.8 ± 0.1 0.25 ± 0.03 USU-628 MC-051909-OSL-01 1.20 90-150 3.0 1.84 ± 0.05 65.4 ± 2.6 7.3 ± 0.7 2.5 ± 0.2 0.24 ± 0.02 USU-629 MC-051909-OSL-02 1.70 90-150 3.0 1.75 ± 0.04 55.7 ± 2.2 4.4 ± 0.4 1.8 ± 0.1 0.22 ± 0.02 USU-630 MC-052009-OSL-04 1.00 90-150 3.6 1.72 ± 0.04 56.3 ± 2.3 6.2 ± 0.6 2.2 ± 0.2 0.24 ± 0.02 *Number of aliquots used for age calculation, number of aliquots measured in parentheses. Age analysis using the SAR procedure of Murray and Wintle (2000) on quartz sand using 1-mm aliquots, 220o C preheats. O.D. Overdispersion represents error on De beyond instrumental error. Error on De is 1 σ, error on age includes random and systematic errors caluclated in quadrature. 1 OSL age calculated using the Central Age Model (CAM) of Galbraith et al. (1999). 2 OSL age calculated using the mimimum age model (MAM) of Galbraith et al. (1999). 220 Table 2. Calbrated calendar year ranges (BP) corrected delta relative relative median Sample code SWD site Lab code Description 14C age +/- 13C lower upper probability lower upper probability probability 080507-03 Big Rock AA82500 charcoal 991 37 -24.7 803 971 1 913 080507-05 Big Rock AA82502 charcoal 1226 38 -22 1060 1255 1 1160 031809-05 Big Rock AA85261 charcoal 1337 36 -20.7 1202 1323 1 1273 031809-06 Big Rock AA85262 charcoal 1359 36 -24.3 1229 1335 1 1288 080507-06 Big Alcove AA82503 charcoal 267 36 -21.9 280 456 0.992 328 080507-07 Big Alcove AA82504 charcoal 162 68 -25.4 -3 302 1 163 080507-08 Big Alcove AA82505 charcoal 139 37 -22.6 9 272 0.995 148 052908-01 Big Alcove AA82506 charcoal 1231 38 -21 1066 1259 1 1166 081509-05 Big Alcove AA86680 charcoal 1271 34 -22.4 1122 1279 1 1213 081509-06 Big Alcove AA86681 charcoal 947 33 -21.1 781 933 1 864 081509-07 Big Alcove AA86682 wood 130 33 -22 54 150 0.432 174 276 0.402 130 111008-01 McPhee AA84122 charcoal 1526 53 -22.2 1325 1528 1 1418 111008-02 McPhee AA84123 charcoal 1578 45 -21.5 1370 1560 1 1465 081409-03 McPhee AA86674 charcoal 1822 67 -17.2 1585 1897 1 1753 111108-02 Mucky AA84126 charcoal 7800 200 -30.2 8199 9153 1 8659 111108-03 Mucky AA84127 charcoal 2089 70 -21.2 1904 2204 0.914 2213 2293 0.086 2063 111108-04 Mucky AA84128 charcoal 1596 43 -26.9 1383 1573 1 1483 110808-01-a Mucky AA85263 charcoal 3322 62 -29.6 3412 3690 1 3555 110808-02-b Mucky AA85264 charcoal twig 2212 97 -29.8 1947 2368 0.978 2388 2402 0.005 2210 031809-X5 Mucky AA85267 charcoal 1342 50 -20.8 1164 1342 1 1272 081509-01 Mucky AA86675 charcoal 1750 38 -27.4 1566 1763 1 1665 081509-02-a Mucky AA86676 charcoal 1735 38 -21.8 1555 1743 1 1648 081509-02-b Mucky AA86677 wood 1626 38 -12.1 1408 1603 1 1520 081509-03 Mucky AA86678 charcoal 1821 39 -27.3 1653 1854 1 1758 111108-05 Mucky AA84129 charcoal 121 82 -25.6 -4 292 1 147 111108-06 Mucky AA84130 charcoal 2680 64 -26.7 2710 2935 1 2798 031809-01 Alluvial reach AA85257 wood 169 34 -26.6 129 230 0.52 -3 37 0.189 174 031809-02 Tafoni AA85258 charcoal 893 51 -24.2 717 919 1 813 031809-03 Tafoni AA85259 charcoal 1995 94 -26.5 1713 2182 0.996 1955 051909-01 Juniper AA86018 charcoal 34900 0 -27.4 inf inf na 0 051909-02 Juniper AA86019 charcoal 8033 55 -24.2 8695 9059 1 8913 051909-04 Juniper AA86021 charcoal 7153 48 -25.7 7881 8065 1 7979 052009-02 Tributary AA86024 charcoal 7060 190 -25.5 7561 8270 1 7892 all calibrations listed are 2 sigma 221 relative probabilities <0.16 were ommited