Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Response of bankfull flood magnitudes to Holocene climate change, Uinta Mountains, northeastern Utah Eric C. Carson† Department of Geography and Department of Geology and Geophysics, University of Wisconsin-Madison, Madison, Wisconsin 53706, USA James C. Knox Department of Geography, University of Wisconsin-Madison, Madison, Wisconsin 53706, USA David M. Mickelson Department of Geology and Geophysics, University of Wisconsin-Madison, Madison, Wisconsin 53706, USA ABSTRACT Long-term variations in Holocene flood magnitude were quantified from the bankfull dimensions of abandoned channels preserved on floodplain surfaces in the northern Uinta Mountains of northeastern Utah. Cross-sectional areas of abandoned channels were reconstructed, and relationships derived from the modern gage records were used to estimate bankfull discharges from bankfull cross-section areas. The results indicate systematic (nonrandom) variations of bankfull floods in the northern Uinta Mountains. Large floods, as much as 10%–15% greater than modern, dominated from 8500 to 5000 calendar yr B.P., and again from 2800 to 1000 cal yr B.P. Small floods, as much as 15%–20% less than modern, characterize the periods from 5000 to 2800 cal yr B.P., and from 1000 cal yr B.P. to near present. The middle and late Holocene record of bankfull flood magnitude compares well with independent evidence for climatic variation in the area. The early Holocene record indicates that larger than modern bankfull floods coincide with warmer than modern mean annual temperature. We hypothesize that an increased range of magnitude for seasonal solar radiation during the early Holocene favored the accumulation and rapid melting of deep snowpacks in the high Uinta Mountains, thus producing large floods despite warmer mean annual temperatures. The episode of smaller than modern bankfull floods between 5000 and 2800 cal yr B.P. † Current address: San Jacinto College Geology Department, 5800 Uvalde Road, Houston, Texas 77049, USA; [email protected]. coincides with records of increased forest fire frequency in the northern Uintas. Larger than modern floods from 2800 to 1000 cal yr B.P. coincide with a local decrease in forest fire frequency and evidence for minor local glacial readvances. The decrease in flood magnitudes following 1000 cal yr B.P. corresponds to numerous local and regional records of warming during the Medieval Climatic Anomaly. Keywords: Holocene, floods, climate, subalpine, Uinta Mountains, Medieval Climatic Anomaly. INTRODUCTION Durations of historic records of stream discharge are typically too short to provide a fully adequate assessment of the relationship between climate change and flood magnitudes. Moreover, in many regions, the impacts of land use (e.g., urbanization, cultivation) and fluvial system alterations (e.g., channelization, dam construction) hinder the ability to identify hydrogeomorphic responses to climate change. In the past two decades there has been a growing recognition of the limitations of flood-frequency analyses that are based on assumptions of stationarity of the mean and variance in time series, and a rise in emphasis on paleoflood studies that address this shortcoming of flood-frequency analyses (e.g., Knox, 1984, 1993, 1999, 2000; Baker et al., 2001; Redmond et al., 2002; Baker, 2003). Many of the investigations of Holocene paleoflood variability in the United States have focused on two regions, the Upper Mississippi Valley and the southwestern U.S. Upper Mississippi Valley studies have emphasized both high-frequency and low-frequency floods (e.g., Knox, 1985, 1993), whereas southwest studies have largely focused on the relative importance of North Pacific winter frontal systems and Pacific tropical cyclones, and the lesser importance of monsoonal weather systems moving north from Baja California, on flood genesis (e.g., review by Ely, 1997). The southwest research has been highlighted by work on the Pecos River in Texas (Kochel and Baker, 1982, 1988), the Escalante River in Utah (Webb et al., 1988), the Gila River in Arizona (Ely, 1992), the Verde River in Arizona (Ely and Baker, 1985; House et al., 2001), and regional syntheses of the relationship between paleoflood behavior and climate change (Ely et al., 1993; Ely, 1997). Owing to the concerted efforts to understand the significance of slackwater paleostage indicators from large-magnitude floods (e.g., Costa, 1978; Baker, 1987), minimal work in the southwest has been focused on Holocene variations in small-magnitude, high-frequency discharges. The research presented here attempts to partly fill this void. Flood regimes for streams on the north flank of the Uinta Mountains of northeastern Utah are strongly influenced by the amount of winter snowpack and the timing of spring snowmelt. The magnitude of the peak annual flood on streams in the Uintas is dependent on amount of precipitation during the preceding winter, amount of precipitation during the spring snowmelt season, and temperature during the spring snowmelt season. Winter precipitation is significant as the source of the annual snowpack; and spring precipitation and spring temperature are significant as an additional source of snowpack and the determinant of the rate of spring snowmelt. The historic stream gage records for streams in the Uintas show that larger than average peak annual floods tend to occur relatively late in the snowmelt season, resulting GSA Bulletin; September/October 2007; v. 119; no. 9/10; p. 1066–1078; doi: 10.1130/B25916.1; 8 figures; 3 tables. 1066 For permission to copy, contact [email protected] © 2007 Geological Society of America Uinta Holocene bankfull flood from unseasonably cool spring temperatures allowing the winter’s snowpack to persist relatively late into the melt season and then be released in a single pulse that is likely often augmented by rain on snow events (Carson, 2003). In the Uintas, the annual nival (spring snowmelt) flood dominates the peak-discharge series, making this area ideal for reconstructing responses of high-frequency floods to longterm Holocene climatic variations. The general objective of this research is to use floodplain alluvial stratigraphy and relict channel morphology for quantitative reconstruction of the magnitudes of former high-frequency floods in the context of their response to Holocene climatic changes. Knox (1985) developed a methodology for evaluating the magnitudes of former bankfull floods in the Upper Mississippi Valley based on the dimensions of relict channels (cutoff oxbows). In humid climate settings, the capacity of the bankfull channel and the elevation of the associated floodplain under construction are strongly related to the magnitude of the most frequently occurring (modal) flood discharge (Wolman and Leopold, 1957; Dury, 1974, 1976). Alluvial valley fills in low-gradient meadows on the north flank of the Uinta Mountains are favorable for the preservation of abandoned relict channels. Infilling of the abandoned channels by natural alluvial processes has preserved the cross-section areas that were likely attuned to the magnitudes of modal floods at the time that each respective channel was abandoned. Reconstruction of the discharge responsible for the respective cross-section capacities for these relict channels provides a record of the temporal variations in bankfull flood magnitudes. Here we compare the results of this reconstruction with independent records of Holocene paleoclimate in the Uinta Mountains and surrounding regions to evaluate the sensitivity of flood behavior to climate change in this area. SETTING The main area of study is the north flank of the Uinta Mountains in northeastern Utah. The Uinta Mountains extend ~200 km eastward from the Wasatch Mountains near Kamas, Utah, into northwestern Colorado (Fig. 1). The range is an east-west–trending, Laramide-age fore- 110o 30′ 110o 15′ 00 BUP 00 Uinta Mountains 3500 0 350 0 35 0 40o 45′ 30 study area Salt Lake City 3500 3000 3000 CHM BBM E. F k. B la 30 ck’s BSM nr y’s WFB 3000 Fo rk EFB HFN E. Fk. Sm it Bear Rive r a Bl Fork . k .F W rk Fo He 00 30 ’s ck 3000 00 W. Fk. Smith ’s 25 h’s Fork Fork o 41 00′’ 110o 45′ land uplift cored by Proterozoic rocks of the Uinta Mountain Group (Sears et al., 1982). The streams in the Uinta Mountains occupy valleys scoured by multiple late Wisconsin glaciations (Atwood, 1909; Bradley, 1936; Richmond, 1965; Hansen, 1969; Laabs and Carson, 2005; Munroe, 2005). Channel beds rest on bedrock in headwater reaches, on boulders in confined valley reaches, and on glacial outwash along mainstem reaches of the streams (Carson, 2005; Schmidt et al., 2005). The sites selected for this study are all located on stream reaches in broad subalpine meadows underlain by glacial outwash or bedrock. Auger drilling at several of the study sites confirmed that the glacial outwash underlying the floodplain sediments is composed of cobbles and boulders to ~0.5 m in diameter. The size of these sediments suggests that the streams have not incised into the outwash during the Holocene, and have instead predominantly meandered laterally across the outwash surface. Despite the high elevation of the study reaches (2650–2950 m above sea level), channel gradients are commonly <0.0005, much gentler than many streams in many mountainous settings, but within the 5 150 5 0 km Utah 10 C.I. = 500 m Figure 1. Location map showing study area on north flank of the Uinta Mountains in northeastern Utah. Shaded boxes indicate field sites for coring relict channels. Field sites: BBM—Broadbent Meadow; BUP—Buck Pasture; BSM—Boy Scout Meadow; CHM—Christmas Meadows; EFB—East Fork Black’s Fork; HFN—Henry’s Fork Narrows; WFB—West Fork Black’s Fork. Geological Society of America Bulletin, September/October 2007 1067 Carson et al. range of values associated with stepped glacial valleys, moraine-dammed valleys, and lowelevation alluvial streams. Bedload sediments that are moved through these reaches at or near bankfull stage consist of sand and gravel to small boulders derived primarily from resistant quartzites of the Uinta Mountain Group. The suspended-sediment loads largely consist of silt and clay derived from the Red Pine Shale formation of the Uinta Mountain Group. The study site is well located to investigate the response of flood hydrology to climate change. Uinta Mountain precipitation is derived in large part from two distinct meteorological systems that vary seasonally. Winter precipitation and resultant snow accumulation, the source of spring runoff, are generally associated with North Pacific frontal storms associated with eastward-moving low-pressure systems (Pyke, 1972; Hansen and Schwartz, 1981). Summer precipitation is derived from eastern North Pacific extratropical cyclones that are steered into Utah by either a deep lowpressure trough or a cutoff low-pressure system in the upper-level westerly circulation along the west coast of the United States (Mitchell, 1976; Maddox et al., 1980). Summer convective storms can provide intense, localized rainfall. The stream basins in this study are sufficiently high in elevation that snowmelt runoff—and therefore winter snowpack accumulation and the timing and rate of spring snowmelt—is the dominant flood-producing mechanism (Jarrett, 1990). We compiled 25 local stream gage records, with an average length of record of 57.0 yr, for this study. The longest of these gage records are Duchesne River at Myton, Utah, and Weber River at Oakley, Utah, with uninterrupted peak annual discharge records dating from 1901 and 1905, respectively. From these data, the recorded peak floods have been associated with the spring snowmelt in >95% of gage years on Uinta mountain streams. Because the peak annual floods in the northern Uinta Mountains are almost exclusively derived from the spring snowmelt, these stream systems are sensitive to variations in the amount of snowpack accumulated during the winter months, and to the timing of snowmelt during the spring and early summer months. METHODS We sought to identify long-term variations in the magnitude of the bankfull-stage discharge for streams in the northern Uinta Mountains in response to past climate change. Therefore, accurate recognition of bankfull stage and discharge are critical to this study. Data from U.S. Geological Survey stream gages in the Uinta 1068 Mountains suggest that the statistically most probable (modal) flood has a recurrence interval of between 1.8 and 2.2 yr. For this study, the bankfull-stage discharge was defined as having a 2.0 yr recurrence interval; this was employed as the reference for comparing modern flood discharges against flood discharge estimates for ancient channels, and provides a link between discharges of a known recurrence interval with bankfull conditions that can be identified in the field. In this study we combined recognition of the upper limit of sand-size particles with the elevation of the floodplain under construction to define the bankfull stage. In the northern Uinta Mountains study area, the floodplain under construction is composed of a fining-upward sequence of lateral-accretion (point-bar) sediments. While there is significant local variation in the grain size of point-bar sediments, ranging from coarse pebbles to medium sand, these sediments contrast markedly with the silt- and clay-dominated vertical-accretion (overbank) sediments on the floodplain surface. This pronounced sedimentological break between the sand-size particles and finer materials corresponds with the tops of the modern point bars. Employing this sedimentological definition allows for consistent determination of bankfull stage in both modern channels that are surveyed at the surface and ancient channels that are preserved in floodplain and terrace alluvial deposits. This method therefore combines the close correspondence between the statistically defined 2.0 yr recurrence interval flood and bankfull conditions with the sedimentological definition of bankfull conditions that can be identified in both modern and ancient channels. Individual relict abandoned channels were used to estimate the magnitude of the channel-forming flood at the time they were cut off. Abandoned channels were cored at short intervals to measure the bankfull cross-section area, as identified by geomorphic evidence. For each cored channel, the bankfull cross-section area was employed to estimate bankfull discharge using a regression equation derived from modern hydrologic data from the northern Uintas. Where possible, the age of abandonment for each of the cored channels was established directly by radiocarbon dating basal of organic channel-fill sediment. For channels without sufficient material for radiocarbon dating, age was estimated using the thickness of verticalaccretion sediment on the associated point bar. To remove the effect of varying drainage areas for the paleochannels, the paleodischarge for each abandoned channel was then compared to the modern bankfull discharge of the stream immediately adjacent to it. Coring Abandoned Channels Abandoned alluvial channels are preserved within the floodplain sediments of seven meadows across the northern Uinta Mountains (Fig. 1). In order to accurately reconstruct bankfull paleochannel dimensions, coring transects must be oriented perpendicular to paleoflow direction. Therefore, abandoned channels were selected only if both the point bars and cutbanks of the abandoned channels were clearly evident from surface morphology. Most relict channels have been filled and have only 5–10 cm of topographic relief on the floodplain, but are often visible in aerial photography (Fig. 2) and during field inspection. From the 7 meadows used in this study, a total of 30 relict channels were identified for coring: 2 channels in Broadbent Meadow, 2 in Boy Scout Meadow, 1 in Buck Pasture, 4 in Christmas Meadows, 3 in East Fork Black’s Fork, 3 in Henry’s Fork Narrows, and 15 in West Fork Black’s Fork (Fig. 1). Transects across 26 channels were sampled with an Oakfield 2-cm-diameter soil probe; cores were spaced at 0.5–1.0 m intervals. Sediment was described as it was collected and discarded. Samples for radiocarbon dating were collected by taking multiple samples from organic-rich horizons at identified depths. Transects across 4 additional relict channels were sampled at a 1.0–1.5 m interval with a modified 7.5-cmdiameter Vibracoring system (Carson, 2003). Additional Oakfield cores were collected at increased spacing as far as 20 m beyond channel boundaries to verify that no significant channel incision or aggradation had occurred. (For full descriptions of all cores, see Carson [2003].) In the West Fork Black’s Fork meadow, two Vibracore transects were collected across the meadow and further confirmed that no significant channel incision or aggradation has occurred at this site (Carson, 2006). Sediment described and retrieved from the coring was used to identify relict channel geometry and bankfull stage. Point-bar sediment is typically a dark red-gray and is characterized by fining-upward tendencies, with grain sizes ranging from cobbles to sand (only particles smaller than pebbles were retrieved by soil probe, whereas the Vibracore could retrieve particles as large as small cobbles). Overbank sediment is distinctly finer grained, and is typically composed of laminated silt and clay. Channel-fill sediment is also typically finer grained than point-bar sediment, but may range from laminated silt and clay to horizontally stratified sand. Abundant organic material, ranging from fibrous peat to twigs, leaf litter, spruce cones, and isolated logs, is found in channel fills and in basal point-bar positions along channel margins. Geological Society of America Bulletin, September/October 2007 Uinta Holocene bankfull flood Figure 2. Aerial photograph of the southern portion of West Fork Black’s Fork meadow (WFB from Fig. 1), showing 7 of the 15 paleochannels cored in this meadow. Channels are identified by dashed lines marking the cutbank of the paleochannel. Solid line in southwest corner marks the coring transect shown in Figure 3. 0 50 Estimation of Paleobankfull Discharge 100 m SE NW Distance (m) 0 5 10 15 20 25 0 ground surface silt and clay Depth (m) 1 m. gray sand pebbles and cobbles The clear visual difference between the red-gray sandy point-bar sediment and the darker finergrained channel-fill and overbank sediments facilitated accurate identification of paleochannel geometries. Channel-bottom sediment is composed of cobbles and boulders where relict channels are exposed in modern cutbanks. This association is similar to modern channel beds in the area. Bankfull stage in the relict abandoned channels was identified as the sedimentological break between the top elevation for sandy lateral-accretion sediment and the overlying silty vertical-accretion sediment. The cross-section area of each channel was calculated on the basis of reconstructed channel geometry and identified bankfull stage (Fig. 3). An average of 18 soil probes per channel was used to determine the geometry of each paleochannel and adjacent point bar and cutbank. Bankfull width/depth ratios were similar for abandoned channels that were cored and modern meandering channels that were surveyed. silt and peat pebbles and cobbles 2 boulders boulders bankfull cross-section area = 6.48 m A relation between the 2.0 yr discharge and bankfull cross-section area for the northern Uinta Mountains was established using data from U.S. Geological Survey gage stations (Table 1). Discharge measurements taken roughly every month at active stations were acquired for 17 gage stations. These data show actual values for discharge and cross-section area for a wide range of discharge values, which were distilled to at-a-station equations relating discharge and cross-section area particular to each station (Table 2). For each site, the discharge for the 2.0 yr flood was estimated by the log Pearson type III method (Interagency Advisory Committee on Water Data, 1986), which was then applied to the at-a-station equations to define the cross-section area associated with that discharge at the gaging site. This establishes the downstream relationship between the 2.0 yr discharge (Q) and the cross-section area (CSA), as shown in Figure 4. This relationship is described by: Qbkf = 2.142 (CSA)0.854, 2 3 Figure 3. Cross-section view of a typical paleochannel used in this study. Horizontal and vertical distances are measured from an arbitrary datum used to survey ground surface. Soil probes, shown by vertical gray bars, were generally spaced 1 m apart, except in cases where the paleochannel profile was changing rapidly, such as at the paleocutbank at 7.5 m horizontal distance. Lateral-accretion point-bar sediment, channel-fill sediment, and vertical-accretion overbank sediment were identified in the soil probes. Bankfull stage was selected based on the maximum elevation of lateral-accretion sediments (horizontal line). The close spacing of cores relative to the width of the channel allows for accurate estimation of bankfull cross-section capacity. (1) where Qbkf is the 2.0 yr discharge (in cubic meters per second), and CSA is the bankfull cross-section area (in square meters). With this equation, 87% of the variance in the bankfull discharge is explained by the cross-section area of the associated bankfull channel. Applying this relation to the gage data shows that the average of the deviations between the 17 observed discharges and the 17 predicted discharges is 22% and the corresponding standard deviation is 15%. Complete Geological Society of America Bulletin, September/October 2007 1069 Carson et al. at-a-station and downstream hydrologic relations for the Uinta Mountains were provided in Carson (2003). While recognizing that some inherent variability through time is derived from physical conditions such as riparian vegetation, channel sinuosity, and sediment load, beyond the direct role of bankfull discharge in determining cross-section area, equation 1 can be used to provide a suitable estimate of the bankfull discharge for any modern or ancient channel based on its bankfull cross-section area. This provides bankfull discharge for each of the cored abandoned channels. One assumption of this process is that the stream channels in the study area have maintained a meandering planform throughout the Holocene, rather than converting to a braided pattern as a result of temporal variations in discharge and sediment load. Studies in the Uinta Basin to the south of the Uinta Mountains have found that the Duchesne River has significantly narrowed and altered its planform over the past century in response to municipal diversions (see summary in Schmidt et al., 2005). However, the width/depth ratios of the cored paleochannels do not vary significantly (p < 0.01) from those observed for the modern channels in the study area (Carson, 2003). The apparent lack of channel adjustment in the studied reaches as compared to the Duchesne River may reflect the extent of drainage diversion in the Uinta Basin. Schmidt et al. (2005) documented decreases in the 2 yr discharge on Duchesne River by as much as 60% over the past century, which are likely far greater than climate-derived streamflow variations. The meandering planform that has been maintained through the Holocene in these subalpine meadows probably results in similar hydrologic relations for both ancient and modern channels. Age of Channel Abandonment The ages of the abandoned relict channels were determined by two methods. Where possible, ages were determined directly by radiocarbon dating of wood or peat at the base of the filled channel (Table 3). For the remaining channels, the ages were determined indirectly by regressing the ages of the directly dated channels against the thickness of vertical-accretion sediment on the point bars of the relict channels (Fig. 5). Oakfield and Vibracore cores across the channel and floodplain study sites showed that channel bed elevations have remained nearly constant throughout the Holocene. Therefore, the accumulation of vertical-accretion sediment on relict point bars over time scales of thousands of years has a linear tendency expressed as: Age = 95.05 (VAS) – 1443, 1070 (2) TABLE 1. U.S. GEOLOGICAL SURVEY GAGE DATA FOR THE NORTHWEST UINTA MOUNTAINS Stream USGS* Drainage area Gage datum Annual duration series 2 gage no. (km ) (m asl) Period of record years (A.D.) Henry's Fork 09226000 145 2543 1943–1972 30 Henry's Fork 09228000 627 2171 1942–1954 13 Burnt Fork 09228500 137 2530 1944–1983 40 Burnt Fork 09229000 189 2164 1930–1942 13 Smith's Fork 09221500 497 2082 1942–1957 16 East Fork Smith's Fork 09220000 137 2582 1940–1999 60 West Fork Smith's Fork 09220500 96 2627 1940–1981 42 Black's Fork 09217900 337 2686 1938–1939, 1967–2000 36 Black's Fork 09218500 394 2595 1940–1998 59 Black's Fork 09222000 2126 1945 1938–1983 46 Black's Fork 09224700 8029 1868 1962–2000 39 West Fork Duchesne River 09276600 215 2098 1990–2000 11 Rock Creek 09279000 381 2210 1938–2000 63 Bear River 10011500 445 2428 1943–2000 58 Bear River 10020100 1948 1970 1962–2000 39 North Fork Provo River 10153800 63 2280 1964–1996 33 Weber River 10128500 420 2024 1905–2000 95 Note: asl—above sea level. *Data source: U.S. Geological Survey, http://water.usgs.gov. TABLE 2. AT-A-STATION EQUATIONS FOR ESTIMATING BANKFULL CROSS SECTION AREAS (CSA) CORRESPONDING TO THE 1.58 yr DISCHARGE (Q) 2 Stream name USGS gage no.* At-a-station equation R Sample size 1.729 Henry's Fork 09226000 CSA = 0.308 (Q) 0.96 56 1.549 Henry's Fork 09228000 CSA = 0.271 (Q) 0.69 52 1.603 Burnt Fork 09228500 CSA = 0.346 (Q) 0.90 89 1.339 Burnt Fork 09229000 CSA = 0.205 (Q) 0.83 60 1.451 Smith's Fork 09221500 CSA = 0.326 (Q) 0.83 75 1.489 West Fork Smith's Fork 09220500 CSA = 0.304 (Q) 0.84 12 2.904 East Fork Smith's Fork 09220000 CSA = 0.047 (Q) 0.89 13 1.501 Black's Fork 09217900 CSA = 0.345 (Q) 0.91 128 1.398 Black's Fork 09218500 CSA = 0.298 (Q) 0.93 106 1.347 Black's Fork 09222000 CSA = 0.250 (Q) 0.96 67 1.267 Black's Fork 09224700 CSA = 0.297 (Q) 0.90 218 1.549 Duchesne River 09276600 CSA = 0.276 (Q) 0.77 97 1.258 Rock Creek 09279000 CSA = 0.209 (Q) 0.77 179 1.588 Bear River 10016900 CSA = 0.258 (Q) 0.96 136 1.346 Bear River 10020100 CSA = 0.151 (Q) 0.85 127 2.019 North Fork Provo River 10153800 CSA = 0.256 (Q) 0.87 151 1.911 Weber River 10128500 CSA = 0.130 (Q) 0.91 222 *Data source: U.S. Geological Survey, http://water.usgs.gov. where Age is the basal age of the dated channels (in calendar yr B.P.) and VAS is the thickness of the vertical-accretion sediment on the point bars (in centimeters) (Fig. 5). One would expect that the accretion rate is likely greatest immediately following channel abandonment when the floodplain surface is at its lowest, and slows as the age of the point bar exceeds several thousand years. However, Knox and Daniels (2001) summarized radiocarbon dates from 53 relict channels and demonstrated a linear relationship between age of channel abandonment and thickness of overbank alluvium over the past ~10 k.y. in the Upper Mississippi Valley. The assumption of a nearly constant overbank sedimentation rate was extrapolated to this study, and a linear relation was used to approximate channel abandonment age based on the thickness of vertical-accretion sediment for the period of data in the Uintas. With this equation, ~96% of the variation in channel age is explained by the thickness of vertical-accretion sediments accumulated on the point bars. The standard error of estimate is ~690 yr. The relatively small standard error of estimate suggests that equation 2 provides a reasonable age assignment for relict channels for interpreting flood behavior at multimillennial time scales. Application of this equation to the Geological Society of America Bulletin, September/October 2007 Uinta Holocene bankfull flood 2.0-year discharge (m3/s) 100 y = 2.142 x R 2 = 0.87 Syx = 1.28 p < 0.001 collected from the WFB meadow identified the spatial distribution and thickness of this historic sediment (Carson, 2003), allowing that horizon to be removed from consideration when calculating thickness of Holocene overbank sediments at the affected paleochannel sites. 0.854 Estimation of Modern Bankfull Discharge at Each Coring Site 10 1 10 1 100 Channel cross-section area (m2) for 2.0 yr discharge Figure 4. Downstream relation between the 2.0 yr discharge and bankfull cross-section area. Data compiled from U.S. Geological Survey gage stations in the northwestern Uinta Mountains (see Tables 1 and 2). In the figure equations, the terms are as follows: y is the dependent variable for the regression equation, R2 is the correlation coefficient, Syx is the standard error of the estimate, and p is the p-value. TABLE 3. BASAL RADIOCARBON DATES FOR RELICT ABANDONED CHANNELS † 14 Sample number Channel C age Calendar years B.P. Material ‡ § identification (yr B.P.) (2σ) BETA-163182 BBM-1 2720 ± 70 2960–2740 Wood BETA-163184 BSM-1 760 ± 40 730–630 Peat BETA-163185 BUP-1 1320 ± 70 1330–1070 Peat BETA-163187 CHM-4 130 ± 60 300–0 Peat BETA-163188 EFB-3 350 ± 60 520–290 Peat BETA-153929 WFBF-1 7080 ± 40 7960–7820 Wood BETA-154583 WFBF-2 7220 ± 70 8170–7930 Peat BETA-153933 WFBF-3 630 ± 40 660–540 Wood BETA-154586 WFBF-4 6470 ± 80 7500–7250 Peat BETA-163190 WFBF-8 1610 ± 80 1700–1320 Wood BETA-163191 WFBF-10 830 ± 60 910–600 Peat BETA-164359 WFBF-11 8970 ± 70 10,230–9900 Wood † Beta Analytic, Inc. Three-letter prefix for channels same as meadows identified in Fig. 1. § Calibrated to calendar years from Stuiver et al. (1998, 2002). ‡ undated channels with known thicknesses of vertical-accretion sediment on the buried point bars allows estimation of ages for those channels. Accumulations of vertical-accretion sediment accelerated by human activity after ca. A.D. 1870 were not included in the formulation of equation 2. During the period ca. 1870–1915, the north slope of the Uinta Mountains was clearcut episodically for railroad ties (Baker and Hauge, 1913). Clearcut logging practices have been shown to favor increased rates of slope erosion (e.g., Swanston and Swanson, 1976; van Heeswijk et al., 1996; Shaw and Julin, 1997). The valley morphometry of the West Fork Black’s Fork meadow (site WFB in Fig. 1) makes this area particularly susceptible to increased rates of overbank alluviation in response to increased sediment supplies. The floodplain in the upper portion of this meadow is covered by as much as 40 cm of visually distinct reddish-brown silt and clay. Analyses of vertical-accretion sediment in this meadow for concentrations of 137Cs (e.g., Ely et al., 1992) identified a peak at ~16–20 cm depth, which corresponds to ca. A.D. 1950. This suggests that the entire package of sediment is related to the historic logging activity (Carson, 2006). More than 200 Vibracore and Oakfield samples The relict channels for which paleodischarge values have been reconstructed represent a range of drainage areas. To facilitate comparisons of these data, the paleodischarge values have been standardized by relating them to the modern bankfull discharge for the same drainage area at the same sites. To avoid any localized influence on channel size (e.g., bank failures, fallen trees, slumping of valley-wall sediment into the channel), a general relation of modern bankfull cross-section area to drainage area was established that could be applied to any location in the northern Uintas. We surveyed bankfull cross-section areas for 21 modern channel sites representing drainage areas between 12 and 280 km2 (Fig. 6). During channel surveys, individual measurements to construct the modern channel geometries were taken 0.5–1.0 m apart across the width of the channels, providing the same level of data for identifying modern channel geometry as collected with Oakfield cores for identifying paleochannel geometries. Bankfull stage was determined as the sedimentological break between sand and fine-grained sediment on the point bar. Where necessary, shallow trenches were dug on the point bars to accurately identify the elevation of the sedimentological break between lateral-accretion point-bar sediments and vertical-accretion overbank sediments. While this method does not necessarily conform to typical field methods of identifying channel bankfull geometries, it does standardize the methods used for identifying modern and paleochannel geometries in this study. The relationship between the cross-section area and the drainage area is expressed by: CSA = 0.868 (DA) 0.477, (3) where CSA is the bankfull cross-section area (in square meters), and DA is the associated drainage area (in square kilometers). Figure 6 shows that the size of the drainage area accounts for 93% of the variance in the bankfull cross-section area. Equation 3 was used to estimate the respective bankfull cross-section area for the modern channel with a drainage area equivalent to the drainage area for a corresponding abandoned relict channel that was cored. This indirect method provides a better estimation of bankfull Geological Society of America Bulletin, September/October 2007 1071 Carson et al. 9000 8000 Channel age (calendar yr B.P.) cross-section area than directly surveying the modern channel adjacent to each abandoned relict channel (Knox, 1985; Wharton et al., 1989; Wharton, 1992). The magnitude of the modern bankfull discharge at each site is then calculated by application of equation 1. This methodology provided paired values of bankfull discharge for each of the 30 abandoned relict channels and the modern channel with an equivalent drainage area at the same site. To remove the dependence of 2.0 yr discharge on drainage area and to provide for dimensionless comparisons of changes between sites, the magnitude of a bankfull discharge from each of the 30 coring sites is expressed as a percentage deviation from the magnitude of the modern bankfull discharge at that site. Calculated bankfull paleodischarges range between 15% larger than modern and 20% less than modern (Fig. 7). The propagated error associated with the application of the linear regression models averages ~±6%. Much of this error is associated with the application of equation 1 to the paired values of ancient and modern bankfull cross-section area for each site. A basic premise used in this study is that variations in bankfull discharge are the principal control on bankfull cross-section area. Therefore, the data were compiled as paired values of ancient and modern bankfull cross-section areas, and then converted to bankfull discharges via application of equation 1. This methodological step allowed the final data to be presented as percentage deviation from the magnitude of the modern bankfull flood, rather than as percentage deviation from the magnitude of the modern bankfull cross-section area. 7000 6000 5000 4000 2000 1000 0 0 50 100 150 Floodplain sediment thickness (cm) Figure 5. Relation between thickness of floodplain sediments and age of abandoned relict channels in the northern Uinta Mountains. Floodplain sedimentation is the measured thickness of vertical-accretion sediments overlying the lateral-accretion point-bar sediments. The regression represents a long-term average rate of floodplain vertical accretion that is ~0.013 cm/yr. Radiocarbon dates were associated with wood buried in lateral-accretion sediments underlying vertical-accretion sediments and in basal paleochannel fill sediments. In the figure equations, the terms are as follows: y is the dependent variable for the regression equation, R2 is the correlation coefficient, Syx is the standard error of the estimate, and p is the p-value. RESULTS AND DISCUSSION Cross-section area (m2) for 2.0 yr discharge 50 The modal flood is produced by spring snowmelt in ~95% of the gage years at gage stations within the research area (Carson, 2003). The very few years that summer precipitation produces the peak annual flood coincide with years of below average snowpack and snowmelt floods; these are cases of smaller than typical nival floods rather than extremely large latesummer floods. This pattern indicates that for the north-slope Uinta Mountains, the maximum flood discharges that are produced by summer weather patterns are significantly smaller than the average flood discharges produced by runoff from the winter snowpack. It is, therefore, the depth of the winter’s snowpack and the rate at which it melts in the spring and early summer that determines the magnitude of the modal flood in the Uinta Mountains. This relationship probably has persisted throughout the Holocene; thus, temporal changes in the amount of snow 1072 y = 95.05 x - 1443 2 R = 0.96 Syx = 686 p < 0.001 3000 y = 0.868 x 2 R = 0.93 Syx = 1.27 p < 0.01 0.477 10 1 10 100 500 2 Drainage Area (km ) Figure 6. Relation between bankfull cross-section area and drainage area for northwestern Uinta Mountains drainage basins. In the figure equations, the terms are as follows: y is the dependent variable for the regression equation, R2 is the correlation coefficient, Syx is the standard error of the estimate, and p is the p-value. Geological Society of America Bulletin, September/October 2007 Uinta Holocene bankfull flood (?) (?) (?) modern channel adjacent to these relict channels was surveyed in 2001. The measured modern bankfull cross-section area is 31% smaller than the expected cross-section area by application of equation 2, which supports our hypothesis. Two additional channels show bankfull-flood discharges between 35% and 44% greater than the modern bankfull flood. One channel is directly dated as 1320–1700 cal yr B.P., and the other is indirectly estimated to 2360 cal yr B.P. by application of equation 2. While both ages are within a period of persistently large floods (Fig. 7), the extreme values calculated are questionable. Both of these channel cross sections are located within tight meander bends visible on floodplain surfaces, and it is likely that the cross-section geometries have been influenced locally by anomalous cutbank erosion (Knox, 1985). Meyer et al. (1995) Munroe and Laidlaw (2002) Magnitude of bankfull flood as a percentage change from modern bankfull flood 20 15 10 5 0 -5 -10 -15 -20 Average 95% confidence interval for flood magnitude Average 95% confidence interval for indirectly dated channels -25 12,000 10,000 8000 6000 4000 2000 0 Comparison of Uintas Paleoflood Record to Regional Paleoclimate Data Calendar Years B.P. Channel dating method: = direct (radiocarbon); Relative forest fire frequency: = indirect (use of equation 3) = frequent; = infrequent Figure 7. Long-term variations in the magnitude of the bankfull discharge in the northern Uinta Mountains expressed as a percentage variation from the magnitude of the modern 2.0 yr discharge. Closed squares represent channels where the age was determined directly from radiocarbon dating of basal channel-fill sediments; open squares represent channels whose age was determined indirectly through application of equation 3 (see text). Relative frequency of forest fires was determined by analysis that quantified density of charcoal particles and sediment grain size in a sediment core from a lake in the Bear River drainage (Munroe and Laidlaw, 2002) and dating alluvial fan sediments in the Yellowstone, Wyoming, region (Meyer et al., 1995). The late Holocene peak in bankfull flood magnitudes at 2800–1000 cal yr B.P. roughly correlates to a period of almost no forest fires; these are both interpreted to reflect cool moist climatic conditions. In contrast, the preceding and following periods of smaller than modern bankfull floods occurred during more frequent forest fires. This is interpreted to reflect warmer, drier climatic conditions during these times. delivered to the Uinta Mountains and the rate at which it melts in the spring and early summer are the principal cause of temporal variations in the magnitude of the modal flood. The reconstructed magnitudes of Holocene bankfull floods for the northern Uinta Mountains (Fig. 7) show that bankfull flood magnitudes have varied systematically (nonrandomly) through time. Former bankfull-discharge magnitudes in north-slope watersheds of the Uinta Mountains were apparently as much as 15%– 20% smaller than modern ca. 500–1000 cal yr B.P. and again ca. 2800–5000 cal yr B.P. In contrast, bankfull-discharge magnitudes were as much as 10% larger than modern at 1000– 2800 cal yr B.P. and were as much as 10%–15% larger than modern at 5000–8500 cal yr B.P. A few problematic data points have been excluded from the reconstruction of bankfull discharge. Three channels, located near each other along the East Fork Black’s Fork, all indicated bankfull discharges more than 40% smaller than modern. Two of these channels dated to ca. 400–450 cal yr B.P., and the third dated to 3100 cal yr B.P. While these coincide with the periods of smaller than modern floods, the extremity of the values is questionable. These three channels are all located immediately north (downstream) of where the Mississippian Madison Limestone intersects the valley floor; we hypothesize that recharge into the aquifer has depleted the surface-water flow in the stream and pervasively reduced the bankfull discharges at these locations. A single cross section of the The temporal pattern of variations in bankfull-flood magnitudes (Fig. 7) corresponds with changes in climate during the middle and late Holocene determined independently by a range of paleoclimate proxies. The two most recently abandoned channels in this study indicate that the floods were smaller than modern during the past ~400 yr. Dendrohydrologic reconstructions of streamflow in the southern Uintas (Carson and Munroe, 2005) and for the entire Upper Colorado River Basin (Woodhouse et al., 2006) indicate that persistent drought conditions prevailed from A.D. 1750 to 1900, and possibly earlier. Meyer et al. (1995) inferred from limited evidence for lateral channel migration and floodplain widening that the Little Ice Age in the Yellowstone region was relatively dry; a paucity of fire-related sedimentation in the Yellowstone region also suggested this time period was relatively cold there. The correspondence between these data from the Uintas and Yellowstone with the current paleoflood record over the past several hundred years suggests that the Holocene paleoflood data will accurately describe flood regimes from earlier time periods. Further comparisons between the Holocene paleoflood record and paleoclimate proxies from the Uintas reinforce this idea. Growth curves of Rhizocarpon geographicum lichen from the headwaters of Black’s Fork tributaries (Fig. 1) indicate no Little Ice Age glacial advances in the area (Munroe, 2002). The stability of lichen-supporting rock surfaces during this period likely indicates insufficient moisture supply to the Uinta Mountains for a renewed glacial advance, a conclusion also supported by evidence of drought inferred from a tree-ring record on the south flank of the range (Carson Geological Society of America Bulletin, September/October 2007 1073 Carson et al. and Munroe, 2005). Numerous studies in the region corroborate the occurrence of small floods through 1000 cal yr B.P., coincident with the occurrence of the Medieval Climatic Anomaly. Submerged tree trunks in growth position from Jenny Lake in Grand Teton National Park date to 600 cal yr B.P. and suggest that water levels at that site were low at that time in response to warm, dry conditions (Stine, 1998). Similar stumps in Mono Lake (Stine, 1990, 1994), Owens Lake (Stine, 1998), and Walker Lake (Stine, 1998) on the east flank of the Sierra Nevada reflect contemporaneous lake lowstands. In western Nebrasaka and eastern Colorado, the most recent of several Holocene periods of eolian dune mobilization began between ca. 1000 and 700 cal yr B.P. (Madole, 1994). Gridded drought reconstructions for most of the western United States confirm regional drought conditions during the Medieval Climatic Anomaly (Cook et al., 2004). Lichenometric evidence for late Neoglacial readvances in the headwaters of Black’s Fork tributaries agrees with the paleoflood record of large floods between 2800 and 1000 cal yr B.P. (Munroe, 2002). A continuous core from Henry’s Fork (Fig. 1) produced Picea/Pinus ratios that indicate that the middle Holocene in the Uinta Mountains was characterized by warm and dry conditions with mean annual temperature ~1 ºC above modern, with maximum temperatures between 6500 and 5400 cal yr B.P. (Munroe, 2003). Climate cooling began ca. 5400 cal yr B.P., and between ca. 5000 and 2800 cal yr B.P., bankfull floods were generally smaller than modern floods of the same recurrence frequency (Fig. 7). Fossil climate proxy data indicate a return to more cool and moist conditions that favored larger than modern nival floods by ca. 2800 cal yr B.P. (Fig. 7). In general, during the middle and late Holocene, warm and dry conditions have favored smaller than modern floods, whereas cool and wet conditions have favored larger than modern floods. Proxy data for forest fire frequency in the northern Uintas further support the association of smaller than modern bankfull floods during periods of warm dry climate and larger than modern bankfull floods during periods of cool moist climate. Munroe and Laidlaw (2002) identified periods of relatively frequent and infrequent forest fires over the past 5500 yr in the northern Uintas on the basis of density of charcoal in a lake sediment core. The study site is at Lily Lake, at the confluence of the East and Hayden Forks of Bear River, within the present study area. Lily Lake is located within dense lodgepole pine cover, where infrequent standreplacing fires tend to occur during drought conditions. These data suggest that forest fires 1074 occurred relatively frequently during 1500–850 cal yr B.P. and 5500–2900 cal yr B.P., but infrequently during 2900–1500 cal yr B.P. (Fig. 7). Although the changes in flood regime and fire frequency do not correspond exactly, there is nevertheless a strong tendency for fires to be infrequent during the period of larger than modern bankfull floods and more frequent during periods of smaller than modern bankfull floods. Evidence for infrequent fires between 2900 and 1500 cal yr B.P. agrees closely with a period of larger than modern floods during 2800–1000 cal yr B.P., and reinforces an interpretation that this was a period of cooler, moister conditions than at present. Similarly, periods of frequent forest fires from 5500 to 2900 cal yr B.P. overlap periods of smaller than modern floods between 5000 and 2800 cal yr B.P. This overlap supports an interpretation that the Uinta Mountains had a warmer, drier climate than at present during those periods. The Uinta paleoflood record corresponds similarly to the Holocene record of wildfire and alluvial sedimentation in the Yellowstone, Wyoming, region (Meyer et al., 1992, 1995). The temporal frequency of forest fires was determined using numerous radiocarbon dates associated with valley-margin alluvial fan deposits. While this approach has produced a higher-resolution record showing more subtle variations in landscape response to climate variability, there is general agreement between the Uinta paleoflood record and the Yellowstone wildfire record (Fig. 7). The data of Meyer et al. (1995) show three clusters of infrequent fire-related events between ca. 3000 and 1300 cal yr B.P., roughly coincident with the most recent period of larger than modern floods in the Uintas. Fire-related events were more frequent between ca. 1300 and 700 cal yr B.P., during the recent period of smaller than modern floods in the Uintas. The earlier portions of the Holocene show less correlation between the two records, although this may be partially due to the decrease in data points in both records during this time. Evidence for persistently larger than modern floods during the apparently warm and dry early to middle Holocene presents a paradox when compared to the middle and late Holocene climate and flood relations described here. Munroe’s (2003) pollen study contributes unique paleoclimate data specific to the Uinta Mountains. Fossil pollen ratios indicate that the Uintas were as much as 1 ºC warmer than modern through much of the early Holocene, and that warm, dry conditions persisted throughout the early to middle Holocene, with maximum warmth occurring at 5400 cal yr B.P. For regional comparison, Fall et al. (1995) identified a warm mid-Holocene interval in parts of the Wind River Range, Wyoming. Their data suggested that temperatures remained as much as 1.0 ºC above modern from 12,600 to 3200 cal yr B.P. (10,600–3000 14C yr B.P.), and that maximum warmth and aridity occurred ca. 6200 cal yr B.P. (5400 14C yr B.P.). Whitlock and Bartlein (1993) differentiated between summerdry and summer-wet regions in the Yellowstone area of northwestern Wyoming. The summerdry areas, which are most similar to the study area, had warm and dry conditions from 9000 to 5000 cal yr B.P. Feiler et al. (1997) estimated that warm, dry conditions prevailed in northwestern Colorado from 9100 to 5400 cal yr B.P. (8100–4600 14C yr B.P.). The general agreement among these studies indicates that large portions of the central Rocky Mountains had warm, dry conditions during the early to middle Holocene, during which time the Uinta Mountains apparently had larger than modern modal floods. Influence of Large-Scale Atmospheric Circulation Patterns on Flood Magnitudes While precipitation over the Uintas is derived from both North Pacific winter frontal storms and late summer eastern North Pacific tropical cyclones, it is the former that controls the flood regime of the Uinta Mountains. The accumulation of winter snowpack derived from winter frontal systems plays a critical role in determining the magnitude of the nival flood. The lowpressure systems that produce winter storms travel in the middle-latitude upper troposphere circulation, and are associated with relatively strong and persistent hemispheric-scale atmospheric patterns over the North Pacific (Pyke, 1972; Hirschboeck, 1985). Dry winters occur predominantly as a result of a persistent highpressure ridge over the Pacific Ocean along the west coast of the United States. Frontal systems then flow to the north of the ridge and enter the continent over the Pacific Northwest, depriving the central Rocky Mountains of the snowpack-producing winter storms. Conversely, wet winters occur when the high-pressure ridge is displaced to the west farther off the west coast. Under these conditions, a low-pressure trough in the upper atmospheric circulation develops over the western United States, the associated circulation shifts the preferential storm track farther to the south, and winter storms in the Uintas occur more frequently. Climatic trends in North America throughout the twentieth century coincided with persistent shifts in the prevailing upper atmospheric circulation patterns (Diaz and Quayle, 1980; McQuirk, 1982; Bradley et al., 1987). Zonal flow was relatively common from the 1930s through 1950s, followed by more persistent meridional Geological Society of America Bulletin, September/October 2007 Uinta Holocene bankfull flood flow conditions that favor an anchored trough over the central Rockies (Barry et al., 1981; Balling and Lawson, 1982). These shifts in prevailing hemispheric upper atmosphere circulation patterns favored increased winter precipitation in the Uinta Mountains and coincided with a significant shift in the Uinta Mountains flood regime toward an increased mean annual flood magnitude and increased frequency of extremely large floods (Carson, 2003). Long-term variations in flood magnitudes through the Holocene likely represent such shifts in large-scale atmospheric circulation patterns. In particular, we attribute the unexpected early Holocene association of large floods with apparently warm and dry climatic conditions to be a consequence of the effects of the retreating Laurentide Ice Sheet on large-scale upper tropospheric circulation patterns combined with amplified seasonality of solar radiation during the early Holocene. Between ca. 12,000 and 6000 cal yr B.P., perihelion occurred during the Northern Hemisphere summer, resulting in increased solar radiation relative to today during the summer and decreased solar radiation during the winter. In addition, orbital obliquity was greater during the early Holocene than today (24.2° at 9000 cal yr B.P. as compared to 23.5° today). Kutzbach (1981) estimated that at 9000 cal yr B.P., solar radiation was as much as 7% greater than modern during July and 7% less than modern during January. The maximum anomalies in seasonal insolation relative to modern conditions occurred between ca. 12,000 and 10,000 cal yr B.P., but the Laurentide Ice Sheet was still sufficiently large enough to significantly influence large-scale upper tropospheric circulation and delay the maximum effect of seasonal insolation anomalies on climate in North America at that time (Kutzbach and Ruddiman, 1993; Webb et al., 1993). Enhanced summer solar radiation throughout the central Rocky Mountains has been cited as the cause for higher temperature and lower effective moisture during the summer months of the early Holocene (Thompson et al., 1993; Whitlock and Bartlein, 1993). However, the flood hydroclimatology of the Uinta Mountains is strongly dependent on winter conditions, and resultant snowpack, rather than summer conditions. The presence of a significant ice mass on the northern portion of the continent during the early Holocene displaced storm tracks southward, thus steering moist air masses toward the central Rockies (Webb et al., 1993). Despite the warmth and aridity prevalent through the summer months of the early Holocene, suppressed solar radiation during the winter months and an abundant moisture supply would have allowed snowpack accumulation to meet or exceed current levels. We therefore hypothesize that a southward displacement of storm tracks coupled with cooler winters followed by relatively rapid and more intense summer warming accounts for the relatively large floods that characterized the north-slope Uinta Mountain streams during the early Holocene. Our hypothesis regarding the flood-climate relation in the early Holocene is supported by observed relations between modern nival-flood magnitudes and melt season conditions. Comparisons of historical streamflow and climate data for the Uintas indicate that nival floods are largest in years in which abnormally cool early spring temperatures are followed by a rapid temperature rise in the late spring (Carson, 2005). Cooler than average early spring temperatures allow thick winter snowpack to persist later into the melt season than usual, and an abrupt temperature rise in late spring produces unusually large nival floods. Enhanced seasonality of solar radiation during the early Holocene would have increased the temperature gradient between winter and summer, and fostered conditions similar to those that currently produce large nival floods. Southwest versus Uinta Paleoflood Records Numerous analyses of late Holocene paleoflood chronologies have been conducted in the southwestern United States (Kochel et al., 1982; Ely and Baker, 1985; Webb et al., 1988; Enzel, 1992; Ely et al., 1993; O’Connor et al., 1994). Most of the chronologies in this region have been constructed from flood slackwater deposits. Ely’s (1997) synthesis of Holocene paleofloods in Arizona and southern Utah indicates that large overbank floods clustered into distinct time periods (Fig. 8B). Ely found that large overbank floods were relatively frequent between 5800 and 4200 cal yr B.P., and again after 2400 cal yr B.P. Large floods were relatively rare between 4200 and 2400 cal yr B.P., and were underrepresented between 800 and 600 cal yr B.P., during the Medieval Climatic Anomaly. Ely attributed the nonrandom clustering of floods to changes in regional and global climate that affected hemispheric circulation patterns and the delivery of regional precipitation. Periods of frequent large floods were associated with cool, wet climate in the southwestern United States, whereas periods of less-frequent large floods were associated with a warmer and drier climate. Comparison of the Uintas paleoflood chronology with the synthesis of paleofloods in the desert southwest (Figs. 8A, 8B) indicates that the two regions underwent broadly synchronous changes in flood regimes. Frequent large floods in the southwest between 5800 and 4200 cal yr B.P. roughly correspond to larger than modern bankfull discharges in the Uintas prior to ca. 5000 cal yr B.P. Similarly, frequent large floods in the southwest between 2400 and 800 cal yr B.P. coincide with bankfull discharges in the Uintas that also were larger than modern. From 4200 to 2400 cal yr B.P., infrequent large floods in the southwest correspond with a period of smaller than modern bankfull discharges in the Uintas. During the later part of the Medieval Climatic Anomaly, between 800 and 600 cal yr B.P., the frequency of large floods in the southwest declined slightly, and bankfull discharges in the Uintas were smaller than modern. The broad synchroneity between the nature of flood magnitudes in the southwestern U.S. and Uinta Mountains underscores the role of climate variation as a principal forcing factor accounting for nonstationarity in the two flood series. Modern gage and climate data indicate that the flood hydrologies of both regions are sensitive to the occurrences of winter frontal storms. Therefore, we infer that cool, wet periods involving winter storms are very important contributors to anomalous high frequencies of large floods in these two regions. Ely et al. (1993) and Ely (1997) demonstrated that large floods preserved in sedimentary records in the southwestern U.S. were most likely caused by intense winter storms or storms related to dissipating tropical cyclones. Delivery of winter storms to the central Rocky Mountains and southwestern U.S., as well as delivery of tropical cyclones to the southwest, is associated with a strongly developed meridional flow pattern in the large-scale upper troposphere circulation. Development of a strong low-pressure system or trough in the upper atmosphere along or off the coast of the western United States deflects winter storm tracks southward. The similarity of the paleoflood records from the Uintas and the southwestern U.S. reflects this common source for flood-generating weather patterns. Upper Mississippi Valley versus Uinta Paleoflood Records Magnitudes of Upper Mississippi Valley bankfull paleofloods were estimated using the same procedure employed in the current study, whereas large overbank paleofloods in the Upper Mississippi Valley were estimated from the dimensions of transported cobbles and boulders found in overbank alluvium (Knox, 1985, 1993). Holocene floods for small tributaries from a few to a few hundred square kilometers in southwestern Wisconsin indicate that both bankfull floods and large overbank floods underwent episodic nonrandom variation over the Holocene. A period of small floods occurred from ca. 5500 to 3300 cal yr B.P. (Figs. 8C, 8D). Proxy climate data for the midwestern U.S. indicate that this Geological Society of America Bulletin, September/October 2007 1075 Carson et al. 20 Magnitude of bankfull flood as a percentage change from modern bankfull flood Figure 8. (A) Uinta Mountains bankfull paleoflood chronology. (B) Record of large overbank floods in Arizona and southern Utah based on slackwater deposits (modified from Ely, 1997). (C) Bankfull paleoflood chronology from the Upper Mississippi Valley (UMV) identified using the same methodology as the Uintas chronology (modified from Knox, 1985, 2000). (D) Chronology of large overbank floods in the Upper Mississippi Valley based on competent depth calculations derived from channel-bed cobbles deposited on the floodplain surface (modified from Knox, 1993, 2000). The two shaded areas represent periods of smaller than modern bankfull floods recognized in the Uintas. A Uinta Mountains 10 0 -10 -20 10,000 8000 6000 115 4000 2000 0 4000 2000 0 4000 2000 0 4000 2000 0 B Southwest U.S. 1076 Number of floods 30 20 10 0 6000 Magnitude of bankfull flood as a percentage change from modern bankfull flood 40 C UMV bankfull floods 30 20 10 0 -10 -20 -30 -40 10,000 5 Ratio of Flood Depth (d) to Bankfull Depth (dbkf) period had mean annual temperatures ~1.5 ºC warmer than the early Holocene and mean annual precipitation ~15% less than modern (Bartlein, 1982; Winkler et al., 1986; Baker et al., 2002). After ca. 3300 cal yr B.P. magnitudes of both bankfull floods and large overbank floods in the Upper Mississippi Valley tributaries increased abruptly. This abrupt increase corresponds with fossil pollen evidence that is indicative of a change to cooler and moister climate (Bartlein, 1982; Winkler et al., 1986; Baker et al., 2002). The early Holocene record of floods in the Upper Mississippi Valley is not well understood, in part because of a lack of data for large overbank floods (Knox, 2000). However, existing data show that prior to ca. 7000 cal yr B.P. bankfull floods were significantly smaller than modern bankfull floods (Fig. 8C). Comparison of the paleoflood record for the Uinta Mountains with those for the Upper Mississippi Valley (Fig. 8) suggests a relationship between flood regimes in the two regions. The low-resolution age control does not allow detailed correlations, but certain broad patterns are evident. Particularly noteworthy is the early Holocene, when bankfull floods in the Uinta Mountains and Upper Mississippi Valley tend to be out of phase with one another. Prior to ca. 7000 cal yr B.P., bankfull floods in the Uinta Mountains were 10%–15% larger than modern, whereas bankfull floods in the Upper Mississippi Valley were 15%–30% smaller than modern. Proxy climate data indicates that the Uinta Mountains were relatively warm (Munroe, 2003), whereas Knox (2000) suggested that the Upper Mississippi Valley was responding to a persistent influx of dry air masses at this time due to the wasting of the Laurentide Ice Sheet. Beginning ca. 7000 cal yr B.P. and lasting until 5500 cal yr B.P., the two regions shifted to an in-phase episode of frequent large floods in both 8000 6000 D UMV large-magnitude floods 4 3 2 1 Ave. 95 percentile C.I. for age Ave. 95 percentile C.I. for flood ratio 0 10,000 8000 6000 Calendar Years B.P. Geological Society of America Bulletin, September/October 2007 Uinta Holocene bankfull flood locations. This period coincides with independent records of generally warm and dry climatic conditions in western North America (although capable of producing large bankfull floods in the Uintas, as previously discussed), although it is possible that the shift to large floods in the Upper Mississippi Valley reflected the final wastage of the Laurentide Ice Sheet in northern Canada. In both regions, relative flood magnitudes declined beginning ca. 5500 cal yr B.P., followed by a period of sustained smaller than modern floods between ca. 5000 and 3000 cal yr B.P. This decrease in flood magnitudes in both regions coincides with evidence for increased forest fire frequency in the northern Uintas (Munroe and Laidlaw, 2002) and warm dry conditions in the Midwestern U.S. (e.g., Bartlein et al., 1984; Winkler et al., 1986). The late Holocene records show a more complicated relationship, reflecting increased variability in flood behavior. The Uinta Mountains show a sustained period of larger than modern floods from ca. 2800 to 1000 cal yr B.P. During this time, the Upper Mississippi Valley displays shorter time-scale fluctuations, with floods approaching modern values from 2800 to 2000 cal yr B.P., then decreasing to 10%–15% smaller than modern from 2000 to ca. 1300 cal yr B.P., and finally increasing rapidly to as much as 20% larger than modern at 1000 cal yr B.P. Lichenometric data from the Uintas indicate that the large floods at this time correspond to locally renewed glacial activity. The final decrease in flood magnitudes in the Uintas after 1000 cal yr B.P. corresponds to evidence for warming during the Medieval Climatic Anomaly that is recorded both locally and regionally. Based on the modern hydroclimatology of the western United States, we tentatively hypothesize that the patterns of flood behavior identified in the Uintas and Upper Mississippi Valley through the Holocene may reflect shifts in the location and pattern of hemispheric circulation and storm tracks. A persistent long-wave trough located over the eastern United States anchored by the melting Laurentide Ice Sheet would have steered winter storms into the Uintas while simultaneously directing dry arctic air into the Upper Mississippi Valley, producing the observed pattern of larger than modern floods in the Uintas and smaller than modern floods in the Upper Mississippi Valley. A more welldeveloped meridional circulation pattern during the early to middle Holocene would have provided flood-producing moisture to both regions. A persistent zonal circulation pattern would have steered storm tracks to the south of both regions, producing the smaller than modern floods observed during the middle Holocene. Weak meridional circulation may have favored the varying flood behavior identified in the two regions during the late Holocene. This hypothesis requires further testing. CONCLUSIONS Reconstructions of bankfull channel cross sections and discharges across the northern Uinta Mountains indicate that the average magnitude of the modal flood has varied systematically (nonrandomly) throughout the Holocene. Bankfull-discharge magnitudes in the Uinta Mountains were apparently as much as 15%– 20% smaller than modern from ca. 5000 to 2800 cal yr B.P. and again from 1000 to 500 cal yr B.P. In contrast, bankfull-discharge magnitudes were as much as 10% larger than modern from 2800 to 1000 cal yr B.P. and were as much as 10%–15% larger than modern from 8500 to 5000 cal yr B.P. The record of bankfull discharge for the middle to late Holocene (ca. 5000 B.P. to present) corresponds well with independent paleoclimate data for the Uinta Mountains. During this period, the magnitude of the modal flood is smaller than modern during warm dry intervals and greater than modern during cool wet intervals. The early to middle Holocene record (8500–5000 cal yr B.P.) indicates that bankfull-flood magnitudes were greater than modern bankfull floods despite regionally prevailing warm and dry conditions, especially during the summer months. We hypothesize that increased summer insolation and decreased winter insolation during the early Holocene favored deep snowpacks during the winter and late and rapid spring snow melt, which fostered larger than modern nival floods despite warmer mean annual temperatures. We also hypothesize that the temporal variations in flood magnitudes identified in the Uintas and Upper Mississippi Valley through the Holocene may reflect variations in hemispheric upper atmosphere circulation patterns. ACKNOWLEDGMENTS This research represents a portion of Carson’s doctoral dissertation conducted under the guidance of Knox and Mickelson. We thank S. Green (U.S. Geological Survey [USGS], Cheyenne, Wyoming), L. Herbert (USGS, Salt Lake City, Utah), and J. Schmidt (Utah State University) for assistance compiling discharge measurement notes necessary for this research. We also thank M. Devito, B. Hess, J. Munroe, N. Oprandy, and L. Wasniewski for assistance in collecting field data for this research. Input from J.M. Daniels, D. Douglass, D. Koerner, M. Rhodes, and M. Wartes greatly improved the final product. Review comments by G. Meyer and J. Schmidt improved the manuscript, as did reviews of an earlier draft by L. Ely, L. Mark, and J. Schmidt. Partial funding for this research was provided by National Science Foundation grant BCS-0081896; Geological Society of America grants 6672-00 and 6858-01; Sigma Xi grant-in-aid; the Ashley and Wasatch-Cache National Forests; the Evjue-Bascom Foundation; and the Morgridge Distinguished Graduate Fellowship. REFERENCES CITED Atwood, W.W., 1909, Glaciation of the Uinta and Wasatch Mountains: U.S. Geological Survey Professional Paper 61, 96 p. Baker, F.S., and Hauge, A.G., 1913, Report on the operation of the Standard Timber Company on the Uinta National Forest, 1912–1913: Uinta National Forest. Salt Lake City, Utah: U.S. Forest Service, 32 p. Baker, R.G., Bettis, E.A., III, Denniston, R.F., Gonzalez, L.A., Strickland, L.E., and Krieg, J.R., 2002, Holocene paleoenvironments in southeastern Minnesota—Chasing the prairie-forest ecotone: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 177, p. 103–122, doi: 10.1016/S0031-0182(01)00354-6. Baker, V.R., 1987, Paleoflood hydrology and extraordinary flood events, in Kirby, W.H., et al., eds., Analysis of extraordinary flood events: Journal of Hydrology, v. 96, p. 79–99. Baker, V.R., 2003, Paleofloods and extended discharge records, in Gregory, K.J., and Benito, G., eds., Palaeohydrology: Understanding global change: London, John Wiley & Sons, Ltd., p. 307–323. Baker, V.R., Webb, R.H., and House, P.K., 2001, The scientific and societal value of paleoflood hydrology, in House, P.K., et al., eds., Ancient floods, modern hazards: Principles and applications of paleoflood hydrology: American Geophysical Union Water Science and Application Series Volume 5, p. 1–20. Balling, R., and Lawson, M., 1982, Twentieth-century changes in winter climatic regions: Climatic Change, v. 4, p. 57–69. Barry, R.G., Kiladis, G., and Bradley, R.S., 1981, Synoptic climatology of the western United States in relation to climatic fluctuations during the twentieth century: Journal of Climate, v. 1, p. 97–113, doi: 10.1002/joc.3370010202. Bartlein, P.J., 1982, Streamflow anomaly patterns in the U.S.A. and southern Canada, 1951–1970: Journal of Hydrology, v. 57, p. 49–63, doi: 10.1016/0022-1694(82)90102-0. Bartlein, P.J., Webb, T., III, and Fleri, E., 1984, Holocene climatic change in the northern Midwest: Pollen-derived estimates: Quaternary Research, v. 22, p. 361–374, doi: 10.1016/0033-5894(84)90029-2. Bradley, R., Diaz, H., Eischeid, J., Jones, P., Kelly, P., and Goodess, C., 1987, Precipitation fluctuations over Northern Hemisphere land areas since the mid-19th century: Science, v. 237, p. 171–175, doi: 10.1126/science. 237.4811.171. Bradley, W.H., 1936, Geomorphology of the north flank of the Uinta Mountains: U.S. Geological Survey Professional Paper 185-I, p. 163–199. Carson, E.C., 2003, Fluvial responses to environmental change, Uinta Mountains, northeastern Utah [Ph.D. thesis]: Madison, University of Wisconsin, 482 p. Carson, E.C., 2005, Fluvial geomorphology and hydrology of the sub-alpine streams of the Uinta Mountains, in Dehler, C., et al., eds., Uinta Mountain geology: Utah Geological Association Publication 33, p. 171–187. Carson, E.C., 2006, Hydrologic modeling of flood conveyance and impacts of historic overbank sedimentation on West Fork Black’s Fork, Uinta Mountains, northeastern Utah, USA: Geomorphology, v. 75, p. 368– 383, doi: 10.1016/j.geomorph.2005.07.022. Carson, E.C., and Munroe, J.S., 2005, Tree-ring-based streamflow reconstruction for Ashley Creek, northeastern Utah, U.S.A.: Implications for paleohydrology of the southern Uinta Mountains: The Holocene, v. 15, p. 602–611, doi: 10.1191/0959683605hl835rp. Cook, E.R., Woodhouse, C.A., Eakin, M.C., Meko, D.M., and Stahle, D.W., 2004, Long-term aridity in the western United States: Science, v. 306, p. 1015–1018, doi: 10.1126/science.1102586. Costa, J.E., 1978, Holocene stratigraphy of flood-frequency analysis: Water Resources Research, v. 4, p. 626–632. Diaz, H., and Quayle, R., 1980, The climate of the United States since 1895, spatial and temporal changes: Monthly Weather Review, v. 108, p. 249–266, doi: 10.1175/15200493(1980)108<0249:TCOTUS>2.0.CO;2. Dury, G.H., 1974, Magnitude-frequency analysis and channel morphology, in Morisawa, M., ed., Fluvial geomorphology: State University of New York Publications in Geomorphology, p. 91–121. Geological Society of America Bulletin, September/October 2007 1077 Carson et al. Dury, G.H., 1976, Discharge prediction, present and former, from channel dimensions: Journal of Hydrology, v. 30, p. 219–245, doi: 10.1016/0022-1694(76)90102-5. Ely, L.L., 1992, Large floods in the Southwestern United States in relation to late-Holocene climatic variations [Ph.D. thesis]: Tucson, University of Arizona, 326 p. Ely, L.L., 1997, Response of extreme floods in the Southwestern United States to climatic variations in the late Holocene: Geomorphology, v. 19, p. 175–201, doi: 10.1016/S0169-555X(97)00014-7. Ely, L.L., and Baker, V.R., 1985, Reconstructing paleoflood hydrology with slackwater deposits; Verde River, Arizona: Physical Geography, v. 6, p. 103–126. Ely, L.L., Webb, R.H., and Enzel, Y., 1992, Accuracy of postbomb (super 137) Cs and (super 14) C in dating fluvial deposits: Quaternary Research, v. 38, p. 196–204. 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, v. 262, p. 410–412, doi: 10.1126/science.262.5132.410. Enzel, Y., 1992, Flood frequency of the Mojave River and the formation of the late Holocene playa lakes, southern California, U.S.A.: The Holocene, v. 2, p. 11–18. Fall, P.L., Davis, P.T., and Zielinski, G.A., 1995, Late Quaternary vegetation and climate of the Wind River Range, Wyoming: Quaternary Research, v. 43, p. 393–404. Feiler, E.J., Anderson, R.S., and Koehler, P.A., 1997, Late Quaternary paleoenvironments of the White River Plateau, Colorado, U.S.A: Arctic and Alpine Research, v. 29, p. 53–62, doi: 10.2307/1551836. Hansen, E., and Schwartz, F., 1981, Meteorology of important rain-storms in the Colorado River and Great Basin drainages: National Oceanic and Atmospheric Administration Hydrometeorological Report 50, 167 p. Hansen, W.R., 1969, The geologic story of the Uinta Mountains: U.S. Geological Survey Bulletin 1291, 144 p. Hirschboeck, K.K., 1985, Hydroclimatology of flow events in the Gila River Basin, central and southern Arizona [Ph.D. thesis]: Tucson, University of Arizona, 335 p. House, P.K., Pearthree, P.A., and Klawon, J.E., 2001, Historical flood and paleoflood chronology of the Lower Verde River, Arizona: Stratigraphic evidence and related uncertainties, in House, P.K., et al., eds., Ancient floods, modern hazards: Principles and applications of paleoflood hydrology: American Geophysical Union Water Science and Application Series Volume 5, p. 295–310. Interagency Advisory Committee on Water Data, 1986, Guidelines for determining flood flow frequency: Interagency Advisory Committee on Water Data, Hydrology Subcommittee, Bulletin 17B, 28 p. Jarrett, R.D., 1990, Paleohydrologic techniques used to define the spatial occurrence of floods: Geomorphology, v. 3, p. 181–195, doi: 10.1016/0169-555X(90)90044-Q. Knox, J.C., 1984, Fluvial responses to small scale climate changes, in Costa, J.E., and Fleisher, P.J., eds., Developments and applications of geomorphology: Berlin, Springer-Verlag, p. 318–342. Knox, J.C., 1985, Responses of floods to Holocene climatic change in the Upper Mississippi Valley: Quaternary Research, v. 23, p. 287–300. Knox, J.C., 1993, Large increases in flood magnitude in response to modest changes in climate: Nature, v. 361, p. 430–432, doi: 10.1038/361430a0. Knox, J.C., 1999, Long-term episodic changes in magnitudes and frequencies of floods in the Upper Mississippi River Valley, in Brown, A.G., and Quine, T.A., eds., Fluvial processes and environmental change: New York, John Wiley & Sons, p. 255–282. Knox, J.C., 2000, Sensitivity of modern and Holocene floods to climate change: Quaternary Science Reviews, v. 19, p. 439–457, doi: 10.1016/S0277-3791(99)00074-8. Knox, J.C., and Daniels, J.M., 2001, Watershed scale and the stratigraphic record of large floods, in House, P.K., et al., eds., Ancient floods, modern hazards: Principles and applications of paleoflood hydrology: American Geophysical Union Water Science and Application Series Volume 5, p. 237–255. Kochel, R.C., and Baker, V.R., 1982, Paleoflood hydrology: Science, v. 215, p. 353–361, doi: 10.1126/science.215. 4531.353. Kochel, R.C., and Baker, V.R., 1988, Paleoflood analysis using slackwater deposits, in Baker, V. R., et al., eds., Flood geomorphology: New York, John Wiley & Sons, p. 357–376. 1078 Kochel, R.C., Baker, V.R., and Patton, P.C., 1982, Paleohydrology of southwestern Texas: Water Resources Research, v. 18, p. 1165–1183. Kutzbach, J.E., 1981, Monsoon climate of the early Holocene; climate experiment with the Earth’s orbital parameters for 9000 years ago: Science, v. 214, p. 59– 61, doi: 10.1126/science.214.4516.59. Kutzbach, J.E., and Ruddiman, W.F., 1993, Model description, external forcing, and surface boundary conditions, in Wright, J., et al., eds., Global climates since the Last Glacial Maximum: Minneapolis, University of Minnesota Press, p. 12–23. Laabs, B., and Carson, E., 2005, Glacial geology of the southern Uinta Mountains, in Dehler, C., et al., eds., Uinta Mountain geology: Utah Geological Association Publication 33, p. 235–253. Maddox, R., Canova, F., and Hoxit, L., 1980, Meteorological characteristics of flash flood events over the western United States: Monthly Weather Review, v. 108, p. 1866–1877, doi: 10.1175/1520-0493(1980)108<1866: MCOFFE>2.0.CO;2. Madole, R.F., 1994, Stratigraphic evidence of desertification in the west-central Great Plains within the past 1000 yr.: Geology, v. 22, p. 483-486. McQuirk, J., 1982, A century of precipitation variability along the Pacific coast of the North America and its impact: Climatic Change, v. 4, p. 41–56. Meyer, G., Wells, S., Balling, R., Jr., and Jull, A., 1992, Response of alluvial systems to fire and climate change in Yellowstone National Park: Nature, v. 357, p. 147–150. Meyer, G., Wells, S., and Jull, A., 1995, Fire and alluvial chronology in Yellowstone National Park; climatic and intrinsic controls on Holocene geomorphic processes: Geological Society of America Bulletin, v. 107, p. 1211–1230, doi: 10.1130/00167606(1995)107<1211:FAACIY>2.3.CO;2. Mitchell, V.R., 1976, The regionalization of climate in the Western United States: Journal of Applied Meteorology, v. 15, p. 920–927, doi: 10.1175/15200450(1976)015<0920:TROCIT>2.0.CO;2. Munroe, J.S., 2002, Timing of post-glacial cirque reoccupation in the northern Uinta Mountains, northeastern Utah, U.S.A: Arctic, Antarctic, and Alpine Research, v. 34, p. 38–48, doi: 10.2307/1552507. Munroe, J.S., 2003, Holocene timberline and palaeoclimate of the northern Uinta Mountains, northeastern Utah, USA: The Holocene, v. 13, p. 175–185, doi: 10.1191 /0959683603hl600rp. Munroe, J., 2005, Glacial geology of the northern Uinta Mountains, in Dehler, C., et al., eds., Uinta Mountain geology: Utah Geological Association Publication 33, p. 215–234. Munroe, J.S., and Laidlaw, J., 2002, A 5000-year record of fire and sedimentary response for the Uinta Mountains, northeastern Utah: Geological Society of America Abstracts with Programs, v. 34, no. 6, p. 469. O’Connor, J.E., Ely, L.L., Wohl, E.E., Stevens, L.E., Melis, T.S., Kale, V.S., and Baker, V.R., 1994, A 4500-year record of large floods on the Colorado River in the Grand Canyon, Arizona: Journal of Geology, v. 102, p. 1–9. Pyke, C., 1972, Some meteorological aspects of the seasonal distribution of precipitation in the western United States and Baja California: University of California Water Resources Center Contribution 139, 205 p. Redmond, K., Enzel, Y., House, P., and Biondi, F., 2002, Climate variability and flood frequency at decadal to millennial scales, in House, P., et al., eds., Ancient floods, modern hazards: Principles and applications of paleoflood hydrology: American Geophysical Union Water Science and Application Series Volume 5, p. 21–45. Richmond, G.M., 1965, Glaciation of the Rocky Mountains, in Wright, H.E., Jr., and Frey, D.G., eds., The Quaternary of the United States: Princeton, New Jersey, Princeton University Press, p. 217–230. Schmidt, J., Smelser, M., Stamp, M., and Gaeuman, D., 2005, Channel adjustment in response to surface-water development in the Duchesne River watershed, in Dehler, C., et al., eds., Uinta Mountain geology: Utah Geological Association Publication 33, p. 189–204. Sears, J.W., Graff, P.J., and Holden, G.S., 1982, Tectonic evolution of lower Proterozoic rocks, Uinta Mountains, Utah and Colorado: Geological Society of America Bulletin, v. 93, p. 990–997, doi: 10.1130/00167606(1982)93<990:TEOLPR>2.0.CO;2. Shaw, C.G., and Julin, K.R., 1997, Assessments of wildlife viability, old-growth timber volume estimates, forested wetlands, and slope stability: Pacific Northwest Forest and Range Experiment Station, USDA Forest Service General Technical Report PNW 65 p. Stine, S., 1990, Late Holocene fluctuations of Mono Lake, eastern California: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 78, p. 333–381, doi: 10.1016/0031-0182(90)90221-R. Stine, S., 1994, Extreme and persistent drought in California and Patagonia during Mediaeval time: Nature, v. 369, p. 546–549, doi: 10.1038/369546a0. Stine, S., 1998, Medieval climatic anomaly in the Americas, in Issar, A.S., and Brown, N., eds., Water, environment and society in times of climatic change: Dordrecht, Netherlands, Kluwer Academic, p. 43–67. Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S., Hughen, K.A., Kromer, B., and McCormac, G., van der Plicht, J., and Spurk, M., 1998, INTCAL98 radiocarbon age calibration, 24,000–0 cal BP: Radiocarbon, v. 40, p. 1041–1083. Stuiver, M., Reimer, P.J., and Reimer, R.W., 2002, CALIB 4.4: http://www.calib.org. Swanston, D.N., and Swanson, F.J., 1976, Timber harvesting, mass erosion, and steepland forest geomorphology in the Pacific Northwest, in Coates, D.R., ed., Geomorphology and engineering: Stroudsburg, Pennsylvania, Dowden, Hutchinson & Ross, Inc., p. 199–221. Thompson, R.S., Whitlock, C., Bartlein, P.J., Harrison, S.P., and Spaulding, W.G., 1993, Climatic changes in the Western United States since 18,000 yr B.P, in Wright, H.E., Jr., et al., eds., Global climates since the last glacial maximum.: Minneapolis, University of Minnesota Press, p. 468–513. van Heeswijk, M., Kimball, J.S., and Marks, D., 1996, Simulation of water available for runoff in clearcut forest openings during rain-on-snow events in the eastern Cascade Range of Oregon and Washington: U.S. Geological Survey Water-Resources Investigations Report 95-4219, 67 p. Webb, R.H., O’Connor, J.E., and Baker, V.R., 1988, Paleohydrologic reconstruction of flood frequency on the Escalante River, south-central Utah, in Baker, V.R., et al., eds., Flood geomorphology: New York, John Wiley & Sons, p. 403. Webb, T.W., III, Ruddiman, W.F., Street-Perrott, F.A., Markgraf, V., Kutzbach, J.E., Bartlein, P.J., Wright, J.H.E., and Prell, W.L., 1993, Climatic change during the past 18,000 years: Regional syntheses, mechanisms, and causes, in Wright, J., et al., eds., Global climate since the Last Glacial Maximum: Minneapolis, University of Minnesota Press, p. 514–535. Wharton, G., 1992, Flood estimation from channel size; guidelines for using the channel-geometry method: Applied Geography, v. 12, p. 339–359, doi: 10.1016/0143-6228 (92)90013-D. Wharton, G., Arnell, N.W., Gregory, K.J., and Gurnell, A.M., 1989, River discharge estimated from channel dimensions: Journal of Hydrology, v. 106, p. 365–376, doi: 10.1016/0022-1694(89)90080-2. Whitlock, C., and Bartlein, P.J., 1993, Spatial variations of Holocene climatic change in the Yellowstone region: Quaternary Research, v. 39, p. 231–238. Winkler, M., Swain, A., and Kutzbach, J., 1986, Middle Holocene dry period in the northern Midwestern United States; lake levels and pollen stratigraphy: Quaternary Research, v. 25, p. 235–250. Wolman, M.G., and Leopold, L.B., 1957, River flood plains; some observations on their formation: U.S. Geological Survey Professional Paper 282-C, p. 87–109. Woodhouse, C.A., Gray, S.T., and Meko, D.M., 2006, Updated streamflow reconstructions for the Upper Colorado River Basin: Water Resources Research, v. 42, W05415, doi: 10.1029/2005WR004455. MANUSCRIPT RECEIVED 16 OCTOBER 2005 REVISED MANUSCRIPT RECEIVED 17 FEBRUARY 2006 MANUSCRIPT ACCEPTED 26 FEBRUARY 2007 Printed in the USA Geological Society of America Bulletin, September/October 2007