Environmental Management DOI 10.1007/s00267-016-0695-1

Debris Flow Occurrence and Sediment Persistence, Upper River Valley, CO

1 1 2 3 K. J. Grimsley • S. L. Rathburn • J. M. Friedman • J. F. Mangano

Received: 16 September 2015 / Accepted: 23 March 2016 Ó Springer Science+Business Media New York 2016

Abstract Debris flow magnitudes and frequencies are debris flows, exceeding HRV, create persistent effects due compared across the Upper valley to assess to valley geometry and geomorphic setting conducive to influences on debris flow occurrence and to evaluate valley sediment storage that are easily delineated by valley con- geometry effects on sediment persistence. Den- finement ratios which are useful to land managers. drochronology, field mapping, and aerial photographic analysis are used to evaluate whether a 19th century Keywords Debris flow Á Sediment persistence Á earthen, water-conveyance ditch has altered the regime of Dendrochronology Á Valley confinement Á Colorado River debris flow occurrence in the Colorado River headwaters. Identifying any shifts in disturbance processes or changes in magnitudes and frequencies of occurrence is funda- Introduction mental to establishing the historical range of variability (HRV) at the site. We found no substantial difference in Debris flows are important processes of sediment transport frequency of debris flows cataloged at eleven sites of in montane ecosystems (Benda 1990; Korup et al. 2004; deposition between the east (8) and west (11) sides of the Stock and Dietrich 2006; Savi et al. 2013) that intermit- Colorado River valley over the last century, but four of the tently deliver material to the fluvial system (Dietrich and five largest debris flows originated on the west side of the Dunne 1978). Often initiated as slope failures in poorly valley in association with the earthen ditch, while the fifth sorted colluvium on sparsely or unvegetated hillslopes is on a steep hillslope of hydrothermally altered rock on the above tree line (Costa and Jarrett 1981) or as shallow east side. These results suggest that the ditch has altered the landslides in unchannelized hillslope hollows (Iverson regime of debris flow activity in the Colorado River et al. 1997), debris flows entrain additional sediment as headwaters as compared to HRV by increasing the fre- they travel down channels and scour to bedrock (Stock and quency of debris flows large enough to reach the Colorado Dietrich 2006). Debris flows are part of the natural dis- River valley. Valley confinement is a dominant control on turbance regime of fluvial systems and may constitute the response to debris flows, influencing volumes of aggrada- dominate mode of coarse sediment (Reneau and Deitrich tion and persistence of debris flow deposits. Large, frequent 1991) and wood delivery to channels, structuring bed morphology, creating important aquatic habitat (Mont- gomery et al. 2003), producing high erosion rates that are & S. L. Rathburn important for landscape evolution (Bennett et al. 2012) and [email protected] as a sediment transfer mechanism that couples hillslopes 1 Department of Geosciences, Colorado State University, and channels (Savi et al. 2013). Debris flows may be Fort Collins, CO 80523-1482, USA triggered naturally when steep hillslopes become unsta- 2 US Geological Survey, Fort Collins Science Center, 2150 ble due to disturbances like heavy rain, flooding, earth- Centre Ave, Bldg. C, Fort Collins, CO 80525, USA quakes, wildfires, or insect outbreaks (Wieczorek 1996)as 3 US Geological Survey, Oregon Water Science Center, 2130 well as failure of landslide dams (Costa and Schuster 1988; SW 5th Ave, Portland, OR 97201, USA Schuster and Highland 2007). In the semiarid Rocky 123 Environmental Management

Mountains, large naturally occurring debris flows are pri- counted to precisely date scarring events (McBride and marily associated with wildfire (Cannon et al. 2001; Laven 1976). The application of tree scars to the study of Wondzell and King 2003), and extreme rainfall (Shroba debris flows has become increasingly prevalent in previous et al. 1979; Menounos 2000; Godt and Coe 2007; Coe et al. research (Hupp et al. 1987; Benda 1990; Baumann and 2014) or rapid snowmelt (Brabb et al. 1989) that saturates Kaiser 1999; Fantucci and Sorriso-Valvo 1999; D’Agos- unstable hillslopes. Anthropogenic causes of debris flows tino and Marchi 2001; Grau et al. 2003; Rubino and include overtopping or piping through earthen dams and/or McCarthy 2004; Bollschweiler et al. 2008; Stoffel and ditches (Jarrett and Costa 1984; McDonald 1999; Clayton Bollschweiler 2008; Arbellay et al. 2010). and Westbrook 2008), failures associated with road cuts Although frequency-magnitude relations are crucial for (Swanson and Dyrness 1975; Wemple et al. 2001), or understanding recurrence intervals and volumes of sedi- clear-cutting and deforestation (Swanston and Swanson ment transfer in mountain watersheds, another useful 1976; Barnard et al. 2001). parameter is the persistence of debris flow deposits, espe- Associated hazards from all debris flows, both natural cially for management and restoration issues of natural and anthropogenic, have important management implica- environments. Sediment that is episodically delivered to tions because of the ever-mounting costs of repairing channels via debris flows is temporarily stored and grad- damaged transportation infrastructure and personal prop- ually released by fluvial erosion, and the storage sites of erty. Over US$670 million in federal funds were obtained debris flow sediment are strongly influenced by valley for recovery efforts after the 2013 devastating floods and geometry whereby, flat, wide valley bottoms effectively debris flows in the of Colorado (Coe et al. limit the export of coarse-grained material from alpine 2014). As such, research into the controls on debris flow environments to downstream main stem rivers (Barsch and occurrence, both natural and human–induced events, is Caine 1984; Caine 1986). A considerable body of research vital to an improved process-level understanding of debris is devoted to quantifying the lag time between delivery of flows to improve hazard prediction. Determining the rela- sediment and subsequent erosion to address transient ver- tive occurrence of natural and human-induced causes of sus persistent landforms (Brunsden and Thornes 1979), to debris flows in a particular watershed, and any shifts in the use as input into sediment routing models (Benda and controls on occurrence over time, requires an assessment of Dunne 1997; Lancaster et al. 2001), for use in habitat magnitude and frequency of both types. management (Phillips 1995), to evaluate channel response When minimal to no documentation exists on the timing to extreme sedimentation events in the context of restora- of historical debris flows, analysis of trees at the debris tion (Madej and Ozaki 2009; Rathburn et al. 2013), and to flow site using dendrochronological techniques is typically quantify long-term landscape evolution (Stock and Dietrich an effective approach (Bollschweiler and Stoffel 2010). 2006). Trees may preserve a remarkable amount of information on The abrupt and pronounced changes in valley and the timing and frequency of debris flows, as well as indi- channel geometry present along many Colorado rivers cators of relative flow magnitude (Bollschweiler and (Wohl 2001) control how geomorphic setting influences Stoffel 2010). Dendrogeomorphic evidence of debris flows response to debris flows and persistence of debris flow can be preserved either through the age of an entire stand or deposits. An ideal setting exists along the Upper Colorado through the characteristics of individual trees (Butler et al. River in Rocky Mountain National Park (RMNP) to assess 1987). When no trees survive a debris flow, the age of trees debris flow occurrence over the last 91 years, and to on the disturbed surface corresponds to a minimum age of evaluate valley geometry influences on sediment persis- the most recent disturbance (Bollschweiler et al. 2008). tence. In this case, we first ask: What is the influence of Assuming the oldest trees have been sampled, the age of anthropogenic activities on debris flow occurrence relative the surface can be accurately calculated by adding in the to natural processes inherent in a steep-gradient mountain colonization time gap (Pierson 2007). Tree rings also system? To address this question, we compared natural and provide more precise dates of debris flow occurrence over anthropogenic influences on debris flows to evaluate the other records such as aerial photographs which only role of a 19th century earthen ditch on debris flow activity. bracket events between years of repeat coverage. A second research question is: How large and frequent do When survivor trees are present in the disturbed area, debris flows have to be to create persistent effects given the they can preserve evidence of impacts or other stresses existing valley geometry along the Upper Colorado River? through a variety of mechanisms including (1) growth We contend that stream channel and valley geometry are decrease, (2) callus tissue, (3) reaction wood, and (4) the dominant controls on sediment persistence within the traumatic resin ducts (Bollschweiler and Stoffel 2010). By Upper Colorado River Valley, and possibly many other sawing out a thin wedge through the discontinuity in the fluvial systems in the semiarid . This cambium at the edge of a scar, warped tree rings can be paper explicitly examines how geomorphic setting 123 Environmental Management influences response to debris flows that appear to exceed because the abundance of till on the east side was below the the range of debris flow magnitudes and frequencies prior minimum mapping area although till was observed in the to human influences, referred to hereafter as the historical field. Late Oligocene rhyolite tuff is altered to bentonite in range of variability (HRV). the study area (Braddock and Cole 1990) although detailed mapping of the alteration has not been completed. Within Historical Range of Variability (HRV) the study area, the hydrothermally altered rhyolite tuff extends northwest–southeast at the north end of the valley HRV is defined as the range of environmental forms and (Fig. 1). processes that predate human use (Morgan et al. 1994) and Channel geometry of the Colorado River within the that have not been altered by human activity. For purposes study site is mainly pool-riffle planform (Montgomery and of this paper, the appropriate time period defining the HRV Buffington 1997) with stream gradients varying from up to in the Colorado River headwaters is roughly constrained by 4 % to less than 1 % through the Lulu City wetland on the the end of neoglaciation (3000 BP) and the start of logging south end of the study area (Fig. 1). Elevations within the and mining operations (150 BP). Debris flows are part of watershed range from 2670 m at the USGS Baker Gulch the natural disturbance regime of high elevation steep- gage (USGS Gage 09010500 Colorado River Below Baker gradient mountain channels in Colorado (Caine 1984; Gulch, Near Grand Lake, CO) to 3944 m at the summit of Menounos 2000), and the HRV framework allows for along the Continental Divide. Many of consideration of whether the magnitude and frequency of the tributaries to the Upper Colorado River originate in debris flows have been shifted by human influence relative glacial cirques and have a steeper gradient than the main to natural variability. The conceptual framework of HRV channel with largely step-pool planform (Woods 2000). allows for a better understanding of the significance of the The runoff regime is snowmelt dominated, with roughly changes that have occurred along a river corridor by 80 % of annual runoff occurring in May, June, and July determining baseline conditions. (Woods 2000) and an estimated 107 cm of average annual We rely on previous, local research to constrain the precipitation (Capesius et al. 2009). Hillslope vegetation is recurrence interval of debris flow activity within the HRV largely composed of Engelmann spruce (Picea engelman- period in RMNP. Caine (1984) ranked natural mass wast- nii), lodgepole pine (Pinus contorta), and subalpine fir ing activity in subalpine forests (between 2800 m and (Abies lasiocarpa) (Westbrook et al. 2006). A mountain treeline, similar to this study site) of the Colorado Front pine beetle outbreak within the last decade has severely Range as ‘slight’ relative to a ‘high’ frequency of mass affected this area, resulting in many dead or dying mature wasting in higher elevation alpine tundra. In addition, pines. Menounos (2000) found that debris flow frequency within two high elevation basins in the Front Range, with bedrock lithologies similar to the study basin, ranges from one to five debris flows per century in 1600 years of record. The Grand Ditch, a 19th century water-conveyance ditch, parallels the Upper Colorado River at approximately Study Area 3100 m above the valley bottom on its western side (Fig. 1). Construction of the ditch began in 1896 and This research was conducted in the headwaters of the continued for 40 years before it was completed in 1936 Colorado River on the western side of RMNP (Fig. 1). The (Woods 2000). The Grand Ditch diverts water across the Colorado River flows south from the Continental Divide Continental Divide from the in and is bordered by the Never Summer Mountains to the the Colorado River watershed into the Cache la Poudre west and Front Range to the east. The Never Summer basin to support irrigated agriculture on the Mountains are composed of late Oligocene and early of Colorado. The Ditch starts on the hillslope above the Miocene intrusive volcanic rocks, with till from the upper USGS Baker Gulch gage and extends north for over 25 km Pleistocene Pinedale glaciation covering the west valley at near 3100 m elevation. Approximately 50 % of annual hillslopes. The Front Range consists of mainly uplifted flow at Baker Gulch is captured depending on the amount Paleoproterozoic metamorphic rocks to the south and Oli- and timing of snowmelt (Woods 2000). In addition, Grand gocene-aged post-Laramide intrusive rocks towards the Ditch captures almost all of the receding limb flows north end of the Colorado River valley (Braddock and Cole resulting in a shorter, reduced recession (Woods 2000). The 1990). Till at the base of the east hillslopes is discontinuous flow reduction alters sediment transport capacity and relative to the west side, possibly because of valley lowers the water table, impacting riparian plant commu- asymmetry or structural control from north-east trending nities (Woods and Cooper 2005) and benthic macroinver- concealed faults mapped by Braddock and Cole (1990), or tebrates (Clayton and Westbrook 2008). 123 Environmental Management

Fig. 1 Location of the study site in Rocky Mountain National Park. study site. Solid shading indicates hydrothermal alteration in the a View of Rocky Mountain National Park between Long Draw vicinity of the Grand Ditch interpreted in the field and from initial Reservoir and Grand Lake, with the location of the USGS Gage mapping by Sanford (2010). Alteration at the top of Specimen 09010500 Colorado River below Baker Gulch. b Close up of the Mountain (hatched shading) is interpreted from remote observations

The Grand Ditch is earthen, unlined, and traverses a on hillslope processes within the study site given its limited hillslope that contains exposures of hydrothermally altered contributing area and extent. rock (Fig. 1) and abundant, easily mobilized glacial till and colluvium. The Ditch bisects several smaller channels Hydrothermal Alteration of Bedrock tributary to the Upper Colorado including Lulu Creek (Fig. 1), and has the potential to send unnaturally large Multiple authors have cited a relationship between water flows down these channels when the Ditch overtops hydrothermal alteration and slope failure (Crowley and or breaches. These large water flows can entrain sufficient Zimbelman 1997; Reid et al. 2001; Lopez and Williams sediment along the tributaries to create debris flows cap- 1993). Hydrothermal alteration tends to lower shear able of reaching the valley bottom along the Colorado strength of rocks (Watters and Delahaut 1995) and to River. Specimen Ditch is a short earthen ditch (Fig. 1) that facilitate debris flow occurrence in combination with steep exists on the east side of valley and exerts minor influence slopes (Reid et al. 2001). In the Upper Colorado River

123 Environmental Management valley, field observations indicate that zones of altered wetland have occurred sporadically over the last rhyolitic tuff are weaker than the surrounding rock. 4000 years, but with increased frequency in recently Changes in mineralogy, especially production of clay aggraded sediments as a result of anthropogenic activities minerals, alter both the appearance and strength properties associated with logging, mining and water diversions in the of these rocks (Sanford 2010). Clasts of altered rhyolitic basin (Rubin et al. 2012). tuff are observed to crumble in situ or in transport, and represent a large fraction of debris flow deposits in the northern areas of the study site. Research on the extent and Methods potential impacts of hydrothermal alteration suggests a heightened risk of slope failure (Sanford 2010). Debris Flow Mapping

Past Debris Flow Activity Debris flow location, timing, and magnitude within the study area were constrained using multiple approaches In May 2003, a large debris flow entered Lulu Creek when including field mapping, aerial photograph analysis, and the Grand Ditch overtopped from rapid melting of the dendrochronology. All identifiable major debris flow abundant snowpack (Clayton and Westbrook 2008; Rubin deposits that reached the valley bottom of the Colorado et al. 2012; Rathburn et al. 2013). Impacts of the 2003 River were cataloged and mapped. These flows tend to be event are still visible along the Colorado River, and include large enough to scour channels and deposit large amounts up to 2-m high unstable berms, a large sediment fan, lateral of sediment along tributaries to the mainstem Colorado and mid-channel bars, and scarred trees. The Lulu City River. General field mapping of debris flow locations was wetland (Fig. 1), a sensitive and ecologically valuable area conducted by traversing the valley bottom to locate berms of Rocky Mountain National Park impacted by the 2003 along tributaries or valley wall concavities where debris debris flow, was buried in up to 1 m of sand and gravel. flows are most likely to occur. These potential debris flow The 2003 event mobilized substantial quantities of the deposits were traced up- and down-slope to verify the light-colored, hydrothermally altered rhyolite tuff, making source location and make measurements and observations. it easy to identify debris flow material in the field and on Detailed field mapping was then conducted and relied on aerial photographs. Analysis of aerial photographs from exposed debris flow berms and fans, and extended from the nearly every decade spanning 1937–2001 indicates that at upstream end of the study site to the downstream end of least two other large debris flows have resulted from the Lulu City wetland (Fig. 1). Prior observations and aerial overtopping of the Grand Ditch (Rubin 2010) in areas photographs indicated that this was a key area of debris outside the bedrock hydrothermal alteration. Observed flow activity, and the wetland is of particular ecological establishment of conifer trees dating to the 1920s and and geomorphologic interest. Other sites of debris flow 1950s on easily identifiable debris flow deposits below deposition were mapped using historical aerial photographs Grand Ditch and along the Colorado River indicate debris from 1937, 1954, 1971, and 1999 with field verification as flow disturbance prior to conifer establishment. There is needed. also evidence of debris flows from the Grand Ditch near An estimate of debris flow volume within the Colorado Baker Gulch, a tributary south of the study site on the west River Valley was used as a proxy for debris flow magni- side of the Colorado River. Occurrence of the Holzwarth tude. Beginning at a break in slope where large-scale debris flow near Baker Gulch was documented on June 16, deposition was evident, width and depth measurements of 1978, and another, earlier debris flow with a similar path deposition were taken on a transect perpendicular to the originating at Grand Ditch is reported to have occurred flow direction. Debris flow deposit width was measured in between 1969 and 1974 (Braddock and Cole 1990). the field along a transect, with transect endpoints recorded Rubin et al. (2012) used ground penetrating radar sur- using a handheld Garmin GPS eTrex (\15 m position veys and radiocarbon dating of valley-bottom sediments to accuracy due to steep, remote terrain). Average deposit evaluate depositional rates and processes in the Lulu City depth along each transect was extrapolated from points wetland. They estimate that aggradation rates over the past where stream incision had exposed the stratigraphy of two centuries, the period of intensive human influence, are debris flow deposition below the ground surface. Depth approximately six times higher than the long-term average measurements were continued down gradient towards the of the last several millennia. Aggradation on the east side Colorado River to an end of discernible debris flow of the wetland has been dominated by peat and overbank deposition. Field coordinates of the extent of each deposit deposition, while the two main sources of sediment on the were imported onto a digital elevation model (DEM) and west side have been debris flows and overbank flood used to create a polygon shapefile in ArcGIS representing deposition. Debris flow and flood processes within the the upper surface of the debris flow deposit (current ground 123 Environmental Management surface). A second polygon representing the lower surface m2. Valley confinement associated with areas of mapped of the debris flow (previous ground surface before debris debris flow deposits was calculated as the ratio of valley-

flow occurred) was constructed using the depth measure- bottom width (Wv) to morphologic bankfull channel width ments taken at various points within the exposed debris (Wc). A confinement ratio of Wv \ 2XWc was considered flow deposit. The upper surface was determined by using confined; a ratio of Wv 2–10XWc was partly confined, and the GPS coordinate to mark off the perimeter on the DEM. Wv [ 10XWc was unconfined (modified from Wohl et al. To find the lower surface, we subtracted the depth mea- 2012). surements taken at various points within the debris flow deposit and then interpolated into a continuous surface of Tree Core and Scar Collection depths and cropped to the extent of the debris flow boundary coordinates. Interpolating between field-derived In this study, debris flow ages were evaluated by sampling points to create the lower surface/polygon assumes that the wedges or slabs from 26 scarred trees and collecting 14 tree deposit depth changes smoothly between each measured cores (Fig. 2). Only trees that were located within or on the point, a reasonable simplification based on the topography. fringe of debris flow berms with an appropriate scar shape Finally, the ArcGIS cut/fill tool was used to calculate and orientation were sampled for wedges/slabs. Death deposit volume between the upper and lower surfaces of dates of beetle-killed trees were determined using cross- each debris flow deposit. In cases where field relations dating (Stokes and Smiley 1996) and therefore, did not indicated the deposit consisted of multiple, overlapping create a source of uncertainty in debris flow ages. Where events, we estimated a minimum volume based on the no pre-existing trees survived a debris flow, we used an exposed surface area, such that volumes are minimum increment borer (Grissino-Mayer 2003) to determine the estimates. age of the oldest trees growing on the deposit. Cored Calculated volumes are minimum estimates of total species included lodgepole pine, Douglas fir, and Engel- debris flow deposition because of the potential for erosion mann spruce. Cores were taken as close to the ground of material following the debris flow, burial of deposited surface as possible to minimize the underestimation of the material below levels evident from stream cutbanks, germination age (Grissino-Mayer 2003). In most cases, the deposition upslope of the mapped area, or GPS errors. largest four trees at the site were selected for sampling. Additionally, in situations of multiple debris flows, it is Tree rings were counted and measured using a Velmex difficult to distinguish the material deposited by an indi- microscope setup. The computer program ARSTAN (Cook vidual debris flow, and therefore, to ascertain contributions and Holmes 1986) was used to normalize and detrend to the total volume of deposition. This issue was partially individual samples. The resulting ring width series were overcome using historical aerial photography to identify cross-dated based on years of very low growth, according debris flow scars and the extent of associated deposits. to Stokes and Smiley (1996), and shifted where appropri- Aerial photographs provide a useful remote sensing ate. The final standardized ring width series were then supplement for determining the spatial and temporal averaged to create a representative master chronology for occurrence of debris flows. Large debris flows commonly remove patches of trees from the landscape, and therefore leave scars on hillslopes that are visible in aerial pho- tographs. By delineating scarred areas on several images and comparing the results, time periods containing the most significant debris flow activity were determined. Aerial photographs for the Upper Colorado River valley are spaced on approximately a decadal interval since 1937, with sufficient resolution to confidently identify debris 1923 flows in aerial photographs (5 m2 open areas identifiable in all but one aerial photo). The results for this technique were used both to confirm source locations and assess recent debris flow activity. Finally, valley confinement ratios from 2012 airborne lidar-derived maps were extracted at locations of mapped debris flow deposits along the main stem Colorado River. The accuracy of the privately contracted lidar was less than Fig. 2 A scar sample collected with a chainsaw from a debris flow berm along Sawmill Creek. The staining on the outer portion of the 0.5 m horizontally, less than 0.15 m vertically, and the wood is a result of pine beetle infestation that killed the tree. The scar average shot density for bare earth points was 0.7 points/ occurred in 1923, the oldest dated debris flow in the study area 123 Environmental Management the study area (Stokes and Smiley 1996). Ring width series from scar samples were checked against the representative chronology using visual comparison and the COFECHA quality control software (Holmes 1983) to improve dating accuracy.

Hydrologic Influences on Debris Flow Occurrence

Historical snowpack data of snow depth and snow water equivalent were acquired from the USDA snow course sites at Grand Lake (elevation 2550 m) and Long Draw Reser- voir (elevation 3080 m; Fig. 1), approximately 15 km downstream and 5 km upstream from the study site, respectively. Continuous records from 1949 to 1971 exist at both snow courses and data were averaged for compar- ison. Annual peak instantaneous discharges at the Baker Gulch gage (USGS Gage 09010500 Colorado River below Baker Gulch) from 1953 to the present were also used. High levels of runoff and hillslope saturation resulting from either snowmelt or heavy rainfall are the most com- mon triggering mechanisms for debris flows (Rebetz et al. 1997). Considering both the snowpack and peak flow records for the study area gives a sense of the total water availability during the snowmelt season as well as the largest flow achieved through snowmelt or intense summer thunderstorms. While thunderstorms are typically of sec- ondary importance to snowmelt for causing high discharges at subalpine elevations of the Colorado Rocky Mountains, they may occur with local intensities that are sufficient to trigger debris flows (Jarrett and Costa 1988).

Results

Debris Flow Deposits and Persistence

Nineteen mappable, age-constrained debris flows at 11 sites were located within the survey area (Fig. 3). Seven of these sites, with 11 debris flows due to multiple events, are on the west side of the Colorado River valley, while the remaining four sites, with eight debris flows are on the east side of the valley (Table 1). Substantial deposition along the active channel of the Colorado River is observed at four sites on the west side (Specimen Creek West, Lady Creek, Lulu Creek, and Little Dutch Creek; denoted with a superscripta in Table 2) and one site on the east side (Little Yellow East). One site from the west (Lulu Creek) and one from the east side (Little Yellow East) of the Colorado River valley clearly originate within the zone of bedrock hydrothermal alteration (Fig. 1). The large, unstable debris flow berms and banks paralleling Lulu Creek, formed during the 2003 Ditch breach, are composed of unconsol- idated cobbles and boulders of welded rhyolite tuff, altered 123 Environmental Management b Fig. 3 Mapped debris flow deposits extending to the valley floor measured to the nearest 0.25 m. The smallest and more along the Upper Colorado River. Deposits from the west side are spatially fragmented of the deposits are labeled as ‘Mis- represented with light shading, those from the east with dark shading, and an area of mixed deposition with hatched shading. Two cellaneous’ because they are associated with small overlapping deposits are mapped on Big Dutch Creek. Approximate unnamed tributaries on poorly channelized hillslopes and extent of deposition from the 2003 Grand Ditch debris flow along are otherwise indistinct. No deposit associated with scarred Lulu Creek is indicated with a dashed black line, which also shows and aged trees on either side of Lulu Creek at the conflu- the debris flow path down the hillslope from Grand Ditch ence with the Colorado River was identified in the field. The scar ages are used to infer an older debris flow or flood, but because of a lack of depositional evidence, this in places to bentonite within a sand matrix. Median grain event was not included in the inventory of debris flows. size of the hydrothermally altered material from the upper The deposited sediment volume from the 2003 debris flow 6 cm on the Lulu Creek fan surface is medium gravel along Lulu Creek (dashed lines in Fig. 3) is the only well- (8–16 mm) (Rathburn et al. 2013). The other flows origi- constrained deposit at this site (36,000 m3 in Table 3; nating on the west side of the valley (Lady Creek, Little Rubin et al. 2012; Rathburn et al. 2013). Dutch Creek, Misc West) are sourced in Quaternary glacial Uncertainty in volume measurements increases with age till of Pinedale age, forming distinctive linear berms that of the deposit, especially where field evidence of multiple parallel the tributary of origin. The coarsest clasts in debris deposits exists (e.g., Big Dutch Creek) and erosion and flow deposits are primarily composed of crystalline igneous subsequent burial of earlier deposits occurred. In these or metamorphic subangular to subrounded boulders and cases, minimum estimates were developed such that the cobbles whereas clasts of the hydrothermally altered tuff total volume was reported without discerning individual are reduced in size to gravel and finer. deposit contributions (Table 2). Debris flow volumes were Estimated debris flow volumes ranged from a few determined using methods that were internally consistent 3 thousand m (Miscellaneous East and Miscellaneous West) such that estimates retain the relative differences in field- 3 to the largest at 138,000 m (Crater Creek) (Table 2). measured dimensions and are assumed to be more accurate Deposit surface areas are shown as polygons on Fig. 3, than order-of-magnitude, sufficient for comparison of rel- with deposit thicknesses ranging from 0.25 to 3.5 m, ative debris flow magnitudes.

Table 1 Basin characteristics for study sites with debris flow deposits that reached the Colorado River valley Site Basin area Mean Notes (km2) slope (%)

West Specimen Cr. W 0.91 (0.30) 39.7 (29.1) Channel scour and deposition visible below Ditch. Highly confined tributary (SCW) Lady Cr. (LC) 0.47 (0.17) 39.5 (23.7) Highly confined source valley. Distinctive material observed much further downstream Lulu Cr. (LU) 6.17 (0.71) 41.4 (23.6) Berms up to 2 m high and downed trees. Abundant coarse sediment in fan at confluence, deposition extends to wetland. Confinement calculated from field surveys Sawmill Cr. 5.22 (0.48) 48.3 (22.9) CO River Valley ‘‘confined’’ by older debris flows. Berms up to 1 m high. Deposition along CO (SC) River buried by 2003 LU debris flow Little Dutch Partially confined valley. Extensive fan deposit partially buried by older debris flows. No Cr(LD) scarred trees. Age bracketed by aerial photos. Misc. West 0.26 (0.13) 33.7 (29.9) Lack of a well-defined drainage. Minor evidence of berms and scarred trees (MW) Big Dutch Cr. 5.22 (0.35) 47.6 (33.9) Multithread channel pattern in the vicinity of highest deposition. Field determination of (BD) confinement because beyond lidar coverage East Little Yellow 0.04 66.4 Confined tributary. Basin too steep for tree growth. Deposition along CO River overlies LC East (LYE) deposition. Ellen’s Trib 0.63 50.2 Two distinct areas of deposition, larger volume associated with unconfined CO River (ET) Misc. East (ME) 0.24 50.1 Lack of well-defined drainage. Includes hillslope between ET and CC. Confined by CC Crater Cr. (CC) 3.67 48.4 Expansive area of deposition, but little recent debris flow activity. Field determination of confinement because beyond lidar coverage Area and slope below the Grand Ditch are shown in parentheses for sites on the West side of the valley

123 Environmental Management

Table 2 Approximate debris Site Approximate volume (m3) Confinement Ratio flow deposit volumes for Colorado River tributaries and West confinement ratios of the a Colorado River valley where the Specimen Creek West 10,000 \2 deposit occurs Lady Creeka 11,000 \2 Lulu Creeka 36,000b 11c Sawmill Creek \5000 4–6 Little Dutch Creeka 10-15,000 10–12 Misc. West \5000 10–12 Big Dutch Creek 53,000d [10 East Little Yellow Easta 16,000 2–4 Ellen’s Tributary 33,000 10–12 Misc. East \5000 – Crater Creek 138,000 [10 Volumes are minimum estimates of total preserved deposition and may consist of multiple, overlapping events. Deposits at some sites may predate the tree-ring record, such that volumes may or may not reflect dated debris flows. Confinement ratio is the ratio of valley-bottom width to bankfull channel width (Wv/Wc). Information on the confinement ratio is described in Table 1 ‘‘Notes’’ column and listed here for com- parison with debris flow volume and potential storage along the Colorado River valley a Indicates widespread debris flow deposition within the Colorado River valley proximal to the active channel (labeled circles in Fig. 5) b Total volume of the 2003 debris flow is 36,000 m3, of which 7200 m3 was deposited in the Lulu Creek debris flow fan. After the 2011 snowmelt runoff, current fan volume is 5550 m3 (Rathburn et al. 2013) c Confinement of Lulu Creek relative to its valley at the site of the 2003 debris flow fan d Total volume from Big Dutch Creek includes evidence for two separate events; no data are available on the relative contributions of each debris flow

The primary photographic evidence of debris flows is valley width (confinement ratio [10) at Lulu Creek, Little the destruction or re-growth of trees on disturbed surfaces. Dutch, Big Dutch, Ellen’s Tributary and Crater Creek. Photographs of Lulu Creek (Fig. 4) just below the Grand Interestingly, older debris flow deposits at the mouth of Ditch from 1999 to 2007 show the change from nearly Sawmill Gulch with mixed east and west source areas solid forest cover to a path of exposed ground caused by (hatched polygon, Fig. 3) effectively confine the Colorado the 2003 debris flow. Lulu Creek provides a useful example River valley at that location, limiting additional debris flow from a recent event where the characteristics of the asso- deposition to events that would exceed existing terrace ciated debris flow are known, and therefore allows for heights which are several meters above the Colorado River comparison with debris flow scars evident on older pho- thalweg at this location. tographs. Removal or recovery of trees along Crater Creek, Sawmill Creek, Lady Creek, Specimen Creek West, and Debris Flow Chronology and Frequency Little and Big Dutch Creeks confirm the occurrence of debris flows within the time periods constrained between or Using tree rings and scar data, the record of mapped debris before images. In one instance, aerial photographic evi- flows extends back over 91 years from present, with the dence of a debris flow is not supported by tree scar ages in oldest debris flow along Sawmill Creek in 1923 (Fig. 5). the field. On the 1937 aerial photograph, a debris flow path The other 18 documented debris flows all occurred within down Lady Creek clearly originates at Grand Ditch, how- the past 65 years, and four of the largest five debris flows ever scarred trees from the deposit along the Upper Col- have occurred since 1958 (Fig. 5). Multiple debris flows orado River immediately downstream from Lady Creek occur within 1 year or consecutive years three times: indicate a mid-1960s age. 1949–1954, 1970–1971, and 1975–1977. These account for Confinement ratios for mapped debris flow deposits 9 of the 19 documented debris flows, or 10 of the 19 with (Table 2) indicate the unconfined valley reaches of the Little Dutch Creek occurring prior to 1953. Upper Colorado River have adequate room for deposition The ‘‘large’’ relative scale of debris flows (Fig. 5) and channel adjustment. The largest volumes of sediment denote larger deposits that are proximal to the active storage and persistence are located in areas of unconfined Colorado River channel (Fig. 3) and may present ongoing

123 Environmental Management

Table 3 Tree scar and core data Minimum Number of for the Upper Colorado River Year Locaon Scar 1 Age Scar 2 Age Scar 3 Age study site Core Age Cores 1923 SM 1923 <1937 LC a N/A N/A 1949 SCW 1949 1950 LUb 1950 1951 1957 1951 CC 1951 1952 1963 1 1953 LDc N/A N/A 1954 ET 1955 3 1954 MW 1954 1956 1957 1956 BD 1956 1958 SCW 1958 1962 1964 4 1961 LYE 1961 1968 2 1965 LC 1965 1967 6 1970 LYE 1970 1975 1976 1977 5 1971 BD 1971 1974 1975 1975 CC 1975 1983 1 1976 ME 1976 1977 LC 1977 1986 2 1985 ET 1985 1986 1987 1988 CC 1988 1992 MW 1992 1992 2003 LU 2003 2003

Year (column 1) was determined from highest quality scar/core where time since debris impact was clearly demarcated in rings. Two additional debris flow ages are included in italics for Lady Creek and Little Dutch Creek (LC and LD) based on aerial photographic evidence only SM Sawmill Creek, SCW Specimen Creek West, LU Lulu Creek, CC Crater Creek, LD Little Dutch Creek, ET Ellen’s Trib, MW Misc West, BD Big Dutch Creek, LC Lady Creek, LYE Little Yellowstone East, ME Misc East a 1937 is maximum age for Lady Creek debris flow based on aerial photographic coverage in 1937 b No mappable deposit evident in the field; scarred trees only c 1953 is maximum age for Little Dutch Creek based on aerial photographic coverage after 1937 but before 1953

Fig. 4 Aerial photographs of Lulu Creek below the Grand Ditch from 1999 to 2007. The white arrow on the 2007 image indicates the 2003 debris flow path

sediment management issues due to reworking of sediment. (Specimen West, Lady, Lulu, Little Dutch). Three of the Four of these five largest debris flow deposits near the west-side derived debris flows are clearly related to activ- active channel originated on the west side of the valley ities along Grand Ditch; Lady Creek was created as a

123 Environmental Management

flows on the west side (four listed above plus Sawmill) may be associated with Grand Ditch. Tree ages in this analysis are assumed to be correct to within plus or minus 2 years (Table 3), due to slightly increased uncertainty associated with the inconsistency of ring growth after trees are impacted by debris flows. The length of this record is limited by a combination of deposit mobilization or burial by the Colorado River and tree recovery. In addition, cross dating is difficult due to the complacency of tree ring series at the study site. Most of the trees sampled in this study were growing along channel margins where the proximity of the river results in less year-to-year variation in growth as precipitation becomes less of a limiting factor (Dudek et al. 1998). Furthermore, tree ring chronologies from nearby sites (, Onahu Creek, and Cameron Pass) do not correspond well with each other or the representative site chronology developed in this study (Grimsley 2012), which may reflect the sensitivity of ring growth to elevation, aspect, and other site conditions of trees in the region (Oberhuber 2004).

Debris Flows and Hydrologic Indictors

Debris flow ages compared to snow depth and discharge records indicate that four of the five largest debris flows with significant deposition along the Colorado River occur in years with at least 140 % of mean discharge (Fig. 6). Although debris flows from Little Dutch Creek also con- tributed substantial sediment deposition within the Color- ado River valley, debris flow ages are poorly constrained and hence it was excluded from Fig. 6. No such correlation is observed between the four largest debris flows and snowpack, however. Annual snowpack may be high but melt over a long duration and not cause hillslope instabil- ities. Peak discharge may be a better proxy than overall snow pack for the amount of moisture available to the hillslope during the wettest conditions and could therefore Fig. 5 Relative volumes and age of debris flow occurrence on the better reflect the likelihood of slope failure and debris flow east and west sides of the Colorado River valley. ‘‘Large’’ debris flows resulted in widespread, observable deposition along the occurrence. It is also possible that the highest mean peak Colorado River proximal to the active channel. ‘‘Small’’ debris flows discharges may correlate with changes in ditch manage- are either of small mapped volume or at a great distance from the ment and flow releases, about which additional information active channel. The dashed bar indicates debris flow(s) along Little is unavailable. Dutch Creek that occurred after 1937 but before 1953 based on aerial photographs. Deposits at Little Dutch Creek were not mapped in the field because other upstream sources partially buried the deposits. SCW Specimen Creek West, LU Lulu Creek, LC Lady Creek, LYE Discussion Little Yellowstone East Anthropogenic Activities, Historical Range wasteway to release water at a northern point along the of Variability and Debris Flow Occurrence ditch and multiple sets of aerial photos show debris flow paths the length of Lady Creek; Lulu Creek resulted from The east side of the Colorado River valley is used as a the 2003 breach in the ditch; and incision of Specimen proxy for pre-Ditch conditions because although similar Creek West originates at Grand Ditch. It is likely, but not human activities of logging and mining, beginning definitively determined, that all five of the large debris approximately 150 years ago, are likely to have occurred 123 Environmental Management

Fig. 6 Four of the five largest 1.80 debris flows along Specimen Snow Creek West (1958), Lady Creek Discharge (1965), Little Yellow East 1.60 Debris Flows (1970), and Lulu Creek (2003) are associated with discharges more than 140 % of normal. 1.40 Gage data derived from Baker

Gulch (USGS Gage 09010500 1.20 Colorado River below Baker Gulch) approximately 13 km downstream from the study site 1.00

0.80 Variation relative to mean Variation

0.60

0.40

0.20 1949 1959 1969 1979 1989 1999 2009 Year throughout the valley, the east side lacks the spatially other four largest debris flows originate on the west side of extensive influence of the earthen Grand Ditch. This design the valley outside of the mapped bedrock alteration. The requires the assumption that lithology and hillslope aspect lack of debris flow material from the southern and eastern is not a key factor affecting debris flow processes. Igneous catchments therefore indicates that the northern portion of and metamorphic bedrock is similar between the west and the study area—where the Grand Ditch coincides with east sides of the valley, with the exception that extensive hydrothermally altered rock and where Ditch management glacial till covering bedrock on the west side is thin and allows for rapid releases of water to sustain levels in Long discontinuous on the east side. The hydrothermal alteration Draw Reservoir—has recently been more prone to slope of rhyolite in the northern portions of the study area spans failure. The 2003 debris flow from the Grand Ditch was both sides of the valley. Solar insolation is comparable unusually large (Table 2) relative to other debris flows in because the hillslopes face predominantly east and west. In the time period of human observation. the Colorado Front Range, aspect-related differences in Forest growth provides additional evidence for inferring channel head location are more prominent at lower eleva- timing of debris flows. The largest total volume of depo- tions, below that of this study site, and aspect does not sition was observed along Crater Creek, along with a few appear to impact channel initiation processes at subalpine tree scars located on berms. However, the advanced soil elevations (Henkle et al. 2011). Ultimately, the spatial development and mature forest age ([200 years) at Crater variability in controlling factors on debris flows con- Creek are indicative of a relatively undisturbed environ- strained us to work within the valley, rather than use other ment. Well-developed forest structure, with a wide range of areas as reference sites. tree ages including some trees that have reached or are Initial comparison of frequencies and volumes of approaching old growth age, suggests that at least a few deposition shows only small, insignificant differences centuries and probably a much longer time has passed since between the east and west hillslopes (8 vs. 11 events). major stand removal along Crater Creek. However, multiple pieces of evidence suggest that sites on The high volume of deposition along Crater Creek the west side of the valley have had more recent high (138,000 m3), as well as Big Dutch Creek (53,000 m3) and magnitude debris flows. First, the only deposits currently Ellen’s Tributary (33,000 m3), must be attributed to larger, visible along the main channel of the Colorado River are older debris flows. The cause for an earlier period of more those from Specimen Creek West, Lady Creek, Little intense debris flow activity is uncertain, but may be related Yellow East, the 2003 Lulu Creek debris flow, and Little to a shift in climate or to one or more catastrophic events. Dutch Creek. Of these, Little Yellow East originated on the The possible existence of an earlier period of more intense east side of the valley, Lulu Creek on the west, and both debris flow activity is supported by a study of sediments within the zone of mapped hydrothermal alteration. The deposited in Sky Pond, an alpine lake in the Colorado Front

123 Environmental Management

Range (Menounos 2000), in which heightened frequency Sediment Persistence and magnitude of debris flows were interpreted during the period of regional climatic amelioration from 8000 to The persistence of debris-flow-related sediment within the 4500 years BP. Although the clustering of the more recent Colorado River valley is ultimately controlled by geo- debris flow occurrences in the study area between 1949 and morphic setting, especially valley confinement. Natural 1976 relative to hydrologic patterns is beyond the scope of resource managers would take less notice, and persistent this analysis, previous research in different basins suggests sedimentation effects would likely be reduced, if, when that coincidental occurrence of multiple debris flows natural or anthropogenically induced debris flows occurred, reflects a site-wide influence on triggering mechanisms subsequent snowmelt runoff reworked and transported a such as weather (Hupp et al. 1987). majority of the sediment downstream. Given the valley The occurrence of the 2003 Lulu Creek debris flow does geometry of the upper Colorado River, debris flow deposits appear to fall within the historical range of variability of mapped in this analysis are retained because of valley geomorphic processes in the Colorado River headwaters, geometry controls, with deposition occurring and persisting based on the older deposition observed along Crater Creek, within unconfined reaches, such as at sites from Little Big Dutch Creek, and Ellen’s Tributary, and previous Dutch Creek, Crater Creek and Big Dutch Creek, where research documenting flood or debris flow deposition over confinement ratios are [ 10. Two end-member scenarios the last 4000 years (Rubin et al. 2012). However, to revisit apply on the Upper Colorado River that influences the our first research question (What is the influence of balance between sediment transport energy and sediment anthropogenic activities on debris flow occurrence relative supply (Fig. 7). Debris flow sediment reaching the main to natural processes inherent in a steep-gradient mountain stem Colorado River is transported through more confined system?), the multiple large debris flows triggered from the valley reaches and retained where valley geometry widens west side of the valley, where anthropogenic influences abruptly. Key controls of transport (valley/channel geom- exist, over approximately the last century constitutes an etry, position within the river segment, discharge, and increase in frequency that is outside the HRV. These volume and grain size distribution of the debris flow findings are in agreement with Menounous (2000) who deposit), are similar to controls on sediment storage with found that on average one to two debris flows occurred per one important distinction; valley/channel geometry century over the last 1600 years. Furthermore, Rubin et al. becomes the dominant factor controlling deposition if (2012) in an analysis of aggradation rates in the down- discharge is relatively constant downstream, as is the case stream Lulu City wetland, found no apparent increase in on the Upper Colorado River where tributary inflow is the magnitude of debris flows evident from ground-pene- minimal, especially during peak water diversion season trating radar (GRP) surveys within the wetland, but did when all headgates along Grand Ditch are closed. As such, document an increased rate of aggradation attributed to areas of persistent debris flow sediment are easily identified increased debris flow frequency over the last two centuries, by calculating valley confinement and targeting areas the period of human intervention. The increased frequency where confinement ratios are [10. of debris flows within the last 200 years translates into a It is important to note that valley confinement may short relaxation time between anthropogenic events with change as older debris flow sediment from a once-uncon- sediment inputs from multiple events limiting full recovery fined valley reach now ‘‘confines’’ the valley creating a time (Rathburn et al. 2013). more constricted condition for future sediment transport. The dendrochronologic record of debris flows along this This is the case on the Upper Colorado River illustrated by portion of the Upper Colorado River is limited to the last the hatched, mixed deposit of Fig. 3, which are terraces of century. The short length of this record can be attributed to old deposits elevated above the current thalweg and no three factors, including: (1) riparian areas in the Colorado longer accessible to flow except for extreme discharges River headwaters are dynamic and therefore trees are many times greater than bankfull. Sediment transport unlikely to persist to old-growth status ([200 years), (2) capacity is now higher in this area and it acts as a transport logging for both the Lulu City mining camp (mid-1800s) reach even during the 2003 debris flow. Cross section A–A0 and construction of the Grand Ditch removed much of the (Fig. 7) illustrates the transition from partially confined to older forest, and (3) debris flow scars eventually heal over confined valley geometry with high sediment persistence. and are no longer readily visible on the outer surface of the Although a remote analysis of confinement may identify tree, so old trees that survived both of the previous fates this reach as partially confined, field verification estab- may have been missed in sampling. This short record lished that the older, high elevation deposits impose a length adds to the difficulty of isolating the impact of the confined condition. Grand Ditch on debris flow occurrence, as diversions began Differences in confinement of the Colorado River valley in the late 1890’s. lead to a disparity in the preservation of debris flow 123 Environmental Management

Transport energy > sediment supply Confined valley; coarse sediment near source

Debris Flow W v,t1 W Sediment Transport and sand A v,t0 A’ A’ cobble A W Persistence c,t1 W c,t0 t : W /W = 8 parally confined 0 v0 c0 t : W /W < 2 confined Valley/channel Posion in 1 v1 c1 geometry river segment High sediment persistence B B’

Wv B’ B Discharge and Wc Grain size Wv/Wc = 4 parally confined Low sediment persistence C’ sand C gravel W C v C’

Wc

Wv/Wc > 10 unconfined sand Moderate sediment persistence

Transport energy < sediment supply Unconfined valley; fine sediment far from source

Fig. 7 Conceptual model showing different scenarios of sediment decrease in valley width. Further downstream, valley width increases, persistence based on valley confinement along the Upper Colorado channel gradient shallows, multiple threads of flow exist, and River in Rocky Mountain National Park. Controls on sediment sediment grain size decreases. At B–B0 the valley is partial confined, persistence are shown in the rounded rectangles. Once debris flow with debris flow deposits as lateral bars of low persistence due to sediment is introduced into the main stem, a first order control is the proximity to active channel processes. At C–C0, valley width is transport capacity relative to the sediment supply, with valley/channel similar to B–B0 but channel width decreases so the valley is geometry and position in the river segment providing a second order unconfined and sediment persistence becomes moderate. At the most control. Finally, discharge and grain size exert a final control as downstream reaches, transport energy exceeds sediment supply due to sediment gets episodically stored, re-entrained and sorted by grain a shallow gradient and wide valley, and depending on the grain size 0 size downstream. At A–A during t0, the channel is partially confined, and discharge, sediment will be variably persistent. Flow direction is has a steeper gradient with moderate transport energy. Debris flow from top of the figure to the bottom, as indicated by the arrow within deposition at t1 decreases valley confinement and increases sediment the channel boundary persistence, a change in valley confinement due to an effective deposits, with limited preservation along northern tribu- volumes with sufficient transport capacity to reach the taries such as Specimen Creek West, Lulu Creek, and Colorado River valley (volumes 30,000 m3 in this case) in Sawmill Creek, but relatively good preservation of debris areas with confinement ratios [ 10 show the greatest sed- flow deposit volumes along southern tributaries such as Big iment persistence in the study area. Dutch and Crater Creeks, where the Colorado River valley Although sediment persistence is high for unconfined is up to 1000 % wider than confined reaches. The role of reaches like Big Dutch and Crater Creeks, we find that valley width in controlling fan formation from debris flow sediment management issues in these areas may be of deposits at tributary mouths has been previously recog- lesser concern because of the greater distance to the active nized in the Oregon Coast Range (May and Gresswell channel. Debris flow deposits at Big Dutch, Crater Creek, 2004). and Ellen’s Creek have high persistence because the To address our second research question: (How large deposits are either elevated above the active channel or and frequent do debris flows have to be to create persistent setback sufficiently to be mobilized only during excep- effects given the existing valley geometry along the Upper tionally large magnitude, low frequency events. In the three Colorado River?), we argue that debris flows on the Upper conceptual cross sections of Fig. 7, sediment persistence is Colorado River over the past century are of sufficient not only governed by channel geometry but also likelihood magnitude and frequency to create lasting sedimentation of entrainment by the active channel. Using the 2003 Lulu effects in unconfined valley reaches. Large debris flow Creek debris flow sediment as an example, high flows have

123 Environmental Management routinely entrained, transported, and redeposited the accuracy. Dendrogeomorphic techniques are most effective material downstream (Rathburn et al. 2013), especially in when paired with evaluation of debris flow deposit char- lateral channel bars that are inundated annually as in B–B0 acteristics, historical aerial photographs, and any other (Fig. 7) resulting in ongoing management and maintenance observations which may provide additional context. of park trails and footbridges. Persistence of debris-flow-related sediment within the Natural resource managers may thus focus more atten- Colorado River valley is controlled by geomorphic setting, tion on unconfined valley bottoms to address future sedi- especially valley confinement. Given the valley geometry ment persistence and associated hazards within fluvial of the upper Colorado River, debris flow deposits mapped systems, especially if important structures may be threat- in this analysis are retained in unconfined reaches where ened. As additional measures of confinement emerge confinement ratios are [10 and sediment volumes are (Fryirs and Brierley 2013; Fryirs et al. 2015), managers sufficient to reach the main stem Colorado River. Persis- overseeing the continued migration of people into the tence of deposition occurs in wide valley reaches near a urban-wildland interface would benefit from additional sediment source. Older debris flow sediment may be the awareness of river diversity, valley geometry and the most persistent because these deposits altered valley associated sedimentation likelihood resulting from debris geometry by creating a more constricted reach for future flows and floods. In this way, potential hazard areas may be sediment transport. Reach-scale valley confinement is identified through measures of valley confinement and easily delineated such that land managers may use it as an proximity to the active channel, both of which could be index of potential sediment storage in areas where addressed in zoning and land use codes. In protected areas important buildings, roads, and associated infrastructure like natural parks, park staff may use valley confinement to exist. evaluate sediment persistence after disturbances to assess hazards to park resources while letting natural processes Acknowledgments The authors thank Ellen Wohl, Brian Bledsoe prevail, and simultaneously protect park infrastructure to Greg Auble, Julian Scott, Peter Brown, David Cooper, Paul McLaughlin, Judy Visty, Harold Pranger, Gary Smillie, and Ben maintain visitor safety. Bobowski for input, and logistical and financial support. Research funding to support KG was provided by Rocky Mountain National Park, the National Park Service (Geologic Resources and Water Conclusions Resources Divisions), and the Warner College of Natural Resources and Department of Geosciences at CSU. Greg Grosicki, Benton Line, Ryan Burbey, Matt Grey, Jonathan Garber, Kevin Pilgrim, and At least 20 debris flows with have occurred along the Amanda Koons helped with field work, and Jim Finley provided a Upper Colorado River over the last century, with 19 thorough review and improved figures. KG thanks everyone in the mappable and aged deposits identified in the field. This is Rathburn and Wohl research group for their encouragement and friendship. Any use of trade, firm, or product names is for descriptive considerably higher than debris flow frequencies docu- purposes only and does not imply endorsement by the US Govern- mented in previous research. We found no substantial ment. Two anonymous reviewers provided insightful comments that difference in the overall frequency of debris flows cata- improved the manuscript. loged at eleven sites of deposition between the east (8) and west (11) sides of the Colorado River valley over the last century, but four of the five largest debris flows originated References on the west side of the valley in association with an earthen ditch, while the fifth is on a steep hillslope of hydrother- Arbellay E, Stoffel M, Bollschweiler M (2010) Dendrogeomorphic reconstruction of past debris-flow activity using injured broad- mally altered rock on the east side of the valley. The Grand leaved trees. Earth Surf Process Landf 35:399–406 Ditch has altered the natural debris flow regime toward Barnard P, Owen L, Sharma M, Finkel R (2001) Natural and human- more frequent occurrence of debris flows large enough to induced landsliding in the Garhwal Himalaya of Northern India. reach the Colorado River that exceed HRV. Likelihood of Geomorphology 40:21–35 Barsch D, Caine N (1984) The nature of mountain geomorphology. debris flow occurrence is augmented by steep, glacial till Mt Res Dev 4:287–298 covered slopes and hydrothermally altered rock in the Baumann F, Kaiser K (1999) The Multetta debris fan, eastern Swiss vicinity of the ditch. At least two of the five largest debris Alps: a 500-year debris flow chronology. Arct Alp Res flows occurred within localized zones of rhyolitic tuff that 31:128–134 Benda L (1990) The influence of debris flows on channels and valley is weakened by hydrothermal alteration. floors in the Oregon Coast Range, US. Earth Surf Process Landf Dendrogeomorphic approaches can be effective tools for 15:457–466 dating debris flows in the Rocky Mountains, but may be Benda L, Dunne T (1997) Stochastic forcing of sediment routing and limited to recent time periods in highly disturbed areas. storage in channel networks. Water Resour Res 33:2865–2880 Bennett GL, Molnar P, Eisenbeiss H, McArdell BW (2012) Erosional Dendrogeomorphology is unique in its ability to precisely power in the Swiss Alps: characterization of slope failure in the determine the age of debris flows to within seasonal Illgraben. Earth Surf Process Landf 37:1627–1640 123 Environmental Management

Bollschweiler M, Stoffel M (2010) Tree rings and debris flows: recent setting on river forms and processes. Earth Surf Process Landf. developments, future directions. Prog Phys Geogr 34:625–645 doi:10.1002/esp.3893 Bollschweiler M, Stoffel M, Schneuwly DM (2008) Dynamics in Godt JW, Coe JA (2007) Alpine debris flows triggered by a 28 July debris-flow activity on a forested cone—a case study using 1999 thunderstorm in the central Front Range, Colorado. different dendroecological approaches. Catena 72:67–78 Geomorphology 84:80–97 Brabb EE, Wieczorek GF, Harp EL (1989) Map showing 1983 Grau HR, Easdale TA, Paolini L (2003) Subtropical dendroecology- landslides in Utah: U.S. Geologial Survey Miscellaneous Field dating disturbances and forest dynamics in northwestern Studies Map MF-2085, scale 1:500,000 Argentina montane ecosystems. For Ecol Manag 177:31–143 Braddock WA, Cole JC (1990) Geologic map of Rocky Mountain Grimsley K (2012) A debris flow chronology and analysis of controls National Park and vicinity, Colorado. US Geological Survey. on debris flow occurrence in the Upper Colorado River valley, Miscellaneous Investigations Series, Map I-1973, 1:50,000, Rocky Mountain National Park, CO. MS Thesis, Colorado State University Brunsden D, Thornes JB (1979) Landscape sensitivity and change. Grissino-Mayer HD (2003) A manual and tutorial for the proper use Trans Inst Br Geogr 4:463–484 of an increment borer. Tree Ring Res 59:63–79 Butler DR, Malanson GP, Oelfke JG (1987) Tree-ring analysis and Henkle JE, Wohl E, Beckman N (2011) Locations of channel heads in natural hazard chronologies: minimum sample sizes and index the semiarid Colorado Front Range, USA. Geomorphology values. Prof Geogr 39:41–47 129:309–319 Caine N (1984) Elevational contrasts in contemporary geomorphic Holmes R (1983) Computer assisted quality control in tree-ring dating activity in the Colorado Front Range. Studia Geomorphol and measurement. Tree Ring Bull 44:69–75 Carpatho-Balc 18:1–30 Hupp CR, Osterkamp W, Thornton JL (1987) Dendrogeomorphic Caine N (1986) Sediment movement and storage on alpine slopes in evidence and dating of recent debris flows on Mount Shasta, the Colorado Rocky Mountains. In: Abrahams AD (ed) Hillslope northern California. US Geol Surv Prof Pap 1396B:1–39 processes. The Binghamton symposium in geomorphology. Iverson RM, Reid MN, LaHusen RG (1997) Debris-flow mobilization International Series No. 16. Allen and Unwin, Winchester, from landslides. Annu Rev Earth Planet Sci 25:85–183 pp 115–137 Jarrett RD, Costa JE (1984) Hydrology, geomorphology, and dam- Cannon SH, Kirkham RM, Parise M (2001) Wildfire-related debris- break modeling of the July 15, 1982, and flow initiation processes, Storm King Mountain, Colorado. Cascade Lake Dam failures, Larimer Country, Colorado, Open Geomorphology 39:171–188 File Report 84-612, US Geological Survey, 107 Capesius JP, Stephens VC, Board CW (2009) Regional regression Jarrett RD, Costa JE (1988) Evaluation of the flood hydrology in the equations for estimation of natural streamflow statistics in Colorado Front Range using precipitation, streamflow, and Colorado. US Department of the Interior, US Geological Survey paleoflood data, U.S. Geological Suvey. Water Resources Clayton J, Westbrook C (2008) The effect of the Grand Ditch on the Investigations Report 87-4117, abundance of benthic invertebrates in the Colorado River, Rocky Korup O, McSaveney M, Davies T (2004) Sediment generation and Mountain National Park. River Res Appl 24:975–987 delivery from large historic landslides in the Southern Alps, new Coe JA, Kean JW, Godt JW, Baum RL, Jones ES, Anderson GS Zealand. Geomorphology 16:189–207 (2014) New insights into debris-flow hazards from an extraor- Lancaster ST, Hayes SK, Grant GE (2001) Modeling sediment and dinary event in the Colorado Front Range. GSA Today 24:10. wood storage and dynamics in small mountainous watershed. In: doi:10.1130/GSATG214A.1 Dorava JM, Montogomery DR, Palcsa BB, Fitzpatrick FA (eds) Cook ER, Holmes RL (1986) Users manual for program ARSTAN. Geomorphic processes and riverine habitat, vol 4. American Laboratory of Tree-Ring Research, University of Arizona, Geophysical Union, Washington, D.C., pp 85–102 Tucson Lopez DL, Williams SN (1993) Catastrophic volcanic collapse: Costa JE, Jarrett RD (1981) Debris flows in small mountain stream relation to hydrothermal processes. Science 260:1794–1796 channels of Colorado and their hydrologic implications. Bull Madej M, Ozaki V (2009) Persistence of effects of high sediment Assoc Eng Geol 18:309–322 loading in a salmon-bearing river, nothern California. In: James Costa JE, Schuster RL (1988) The formation and failure of natural LA, Rathburn SL, and Whittecar GR (eds) Management and dams. Geol Soc Am Bull 100(7):1054–1068 restoation of fluvial systems with broad historical changes and Crowley JK, Zimbelman DR (1997) Mapping hydrothermally altered human impacts, vol 451. GSA Special Paper, pp 43–56 rocks on Mount Rainier, Washington, with airborne visible/ May CL, Gresswell RE (2004) Spatial and temporal patterns of infrared imaging spectrometer (AVIRIS) data. Geology debris-flow deposition in the Oregon Coast Range, USA. 25:559–562 Geomorphology 57:135–149 D’Agostino V, Marchi L (2001) Debris flow magnitude in the Eastern McBride JR, Laven RD (1976) Scars as an indicator of fire frequency Italian Alps: data collection and analysis. Phys Chem Earth Part in the San Bernardino Mountains, California. J For 74:439–442 C 26:657–663 McDonald GN (1999) Technical reports for 1998 applied geology Dietrich WE, Dunne T (1978) Sediment budget for a small catchment program, Report of Investigation 242, Utah Geological Survey, in mountainous terrain. Z Geomorphol 29:191–206 Utah Department of Natural Resources, pp 36–41 Dudek DM, McClenahen JR, Mitsch WJ (1998) Tree growth Menounos B (2000) A holocene debris-flow chronology for an Alpine responses of Populus deltoides and Juglans nigra to streamflow Catchment, Colorado Front Range. In: Slaymaker O (ed) and climate in a bottomland hardwood forest in central Ohio. Am Geomorphology, human activity, and global environmental Midland Nat 140:233–244 change. Wiley, Chichester, pp 117–149 Fantucci R, Sorriso-Valvo M (1999) Dendrogeomorphological anal- Montgomery DR, Buffington JM (1997) Channel-reach morphol- ysis of a slope near Lago, Calabria (Italy). Geomorphology ogy in mountain drainage basins. Geol Soc Am Bull 30:165–174 109:596–611 Fryirs KA, Brierley GJ (2013) Geomorphic analysis of river systems: Montgomery DR, Massong TM, Hawley SCS (2003) Influence of an approach to reading the landscape. Wiley, Chichester debris flows and log jams on the location of pools and alluvial Fryirs KA, Wheaton JM, Brierley GJ (2015) An approach for channel reaches, Oregon Coast Range. Geol Soc Am Bull measuring confinement and assessing the influence of valley 115:78–88

123 Environmental Management

Morgan P, Aplet GH, Haufler JB, Humphries HC, Moore MM, Shroba RR, Schmidt PW, Crosby EJ, Hansen W, Soule JM (1979) Wilson WD (1994) Historical Range of Variability: a useful tool Geologic and geomorphic effects in the Big Thompson Canyon for evaluating ecosystem change. J Sustain For 2:87–111 area, Larimer Country, Park B of Storm and Flood of July Oberhuber W (2004) Influence of climate on radial growth of Pinus 31-August 1, 1976, in the Big Thompson and Cache la Poudre cembra within the alpine timerline ecotone. Tree Physiol River Basins, Larimer and Weld Counties, Colorado. US Geol 24:291–301 Surv Prof Pap 1115:87–148 Phillips JD (1995) Biogeomorphology and landscape evolution: the Stock JD, Dietrich WE (2006) Erosion of steepland valleys by debris problem of scale. Geomorphology 13:337–347 flows. Geol Soc Am Bull 118:1125–1148 Pierson TC (2007) Dating young geomorphic surfaces using age of Stoffel M, Bollschweiler M (2008) Tree-ring analysis in natural colonizing Douglas fir in southwestern Washington and north- hazards research—an overview. Nat Hazards Earth Syst Sci western Oregon, USA. Earth Surf Process Landf 32:811–831 8:187–202 Rathburn SL, Rubin ZK, Wohl EE (2013) Evaluating channel Stokes MA, Smiley TL (1996) An introduction to tree-ring dating. response to an extreme sedimentation event in the context of University of Arizona Press, Tucson historical range of variability: Upper Colorado River, USA. Swanson FJ, Dyrness CT (1975) Impact of clear-cutting and road Earth Surf Process Landf 36:391–406 construction on soil erosion by landslides in the western Cascade Rebetz M, Lugon R, Baeriswul P (1997) Climatic change and debris Range, Oregon. Geology 3:393–396 flows in high mountain regions: the case study of the Ritigraben Swanston DN, Swanson FJ (1976) Timber harvesting, mass erosion torrent (Swiss Alps). Clim Change 36:371–389 and steepland forest geomorphology in the Pacific Northwest. In: Reid ME, Sisson TW, Brien DL (2001) Volcano collapse promoted Coats DR (ed) Geomorphology and Engineering, pp 199–221 by hydrothermal alteration and edifice shape, Mount Rainier, Watters R, Delahaut W (1995) Effect of argillic alteration on rock Washington. Geology 29:779–782 mass stability. Clay and shale slope instability. Geol Soc Am Reneau SL, Deitrich WE (1991) Erosion rates in the southern Oregon Rev Eng Geol 10:139–150 Coast Range: evidence for an equilibrium between hillslope Wemple BC, Swanson FJ, Jones JA (2001) Forest roads and erosion and sediment yield. Earth Surf Process Landf geomorphic process interactions, Cascade Range, Oregon. Earth 16:307–322 Surf Process Landf 26:191–204 Rubin ZK (2010) Post-glacial valley evolution and post-disturbance Westbrook CJ, Cooper DJ, Baker BW (2006) Beaver dams and channel response as a context for restoration, Upper Colorado overbank floods influence groundwater-surface water interac- River, Rocky Mountain National Park. MS Thesis, Colorado tions of a Rocky Mountain riparian area. Water Resour Res State University 42:W06404 Rubin ZK, Rathburn SL, Wohl EE, Harry DL (2012) Historic range of Wieczorek GF (1996) Landslide triggering mechanisms. Landslides variability in geomorphic processes as a context for restoration: 247:76–90 Rocky Mountain National Park, Colorado, USA. Earth Surf Wohl EE (2001) Vitural Rivers. Yale University Press, New Haven Process Landf 37:209–222 Wohl E, Dwire K, Sutfin N, Polvi L, Bazan R (2012) Mechanisms of Rubino DL, McCarthy B (2004) Comparative analysis of dendroe- carbon storage in mountainous headwater rivers. Nat Commun cological methods used to assess disturbance events. Den- 3:1263. doi:10.1038/ncomms2274 drochronologia 21:97–115 Wondzell SM, King JG (2003) Postfire erosional processes in the Sanford B (2010) Implications of hydrothermal alternation for Pacific Northwest and Rocky Mountain regions. For Ecol Manag landscape stability and environmental geochemistry In: Rocky 178:75–87 Mountain National Park Research Conference Proceedings, Woods SW (2000) Hydrologic effects of the Grand Ditch on streams Estes Park and wetlands in Rocky Mountain National Park, Colorado. MS Savi S, Schneuwly-Bollschweiler M, Bommer-Denns B, Stoffel M, Thesis, Colorado State University Schlunegger F (2013) Geomorphic coupling between hillslopes Woods SW, Cooper DJ (2005) Hydrologic factors affecting initial and channels in the Swiss Alps. Earth Surf Process Landf willow seedling establishment along a subalpine stream, Color- 38:959–969 ado, USA. Artic Antarct Alp Res 37:636–643 Schuster RL, Highland LM (2007) Overview of the Effects of mass Wasting on the Natural Environment. Environ Eng Geosci 13(1):25–44

123