USGS - science for a changing world

Oregon Water Science Center

green dotHome green dotData green dotStudies green dotPublications/Media green dotInformation

Geomorphology and Flood-Plain Vegetation of the Sprague and Lower Sycan Rivers, Klamath Basin, Oregon

By Jim E. O’Connor, Patricia F. McDowell, Pollyanna Lind, Christine G. Rasmussen, and Mackenzie K. Keith

In cooperation with the University of Oregon and the U.S. Fish and Wildlife Service


Channel and Flood-Plain Conditions and Processes

Sprague River bank erosion

Bank erosion along the Sprague River. (Photograph by Jim O'Connor, USGS)

This section outlines key morphologic attributes of the channel and flood plain of the Sprague and Sycan Rivers, and how these attributes have changed at time scales encompassing the last century to the last several thousand years. These morphologic characteristics are in turn related to past and present processes of channel and flood-plain formation.

Historical data used in mapping and quantitative analysis of channel and flood-plain conditions were mainly from the cadastral surveys, aerial photographs, and LiDAR (table 4). Sediment transport measurements and a survey of effects of the 2006 high flow provide evidence of key processes presently affecting channels and flood plains. The augering transects and analysis of stratigraphic exposures (appendix C) provides information on flood-plain stratigraphy and channel position and processes over the last several thousand years, pre-dating substantial human flood-plain and channel alteration. Methods for analysis build upon previous studies of rivers in the Pacific Northwest, including the Queets and Quinault Rivers of western Washington (O’Connor and others, 2003c), the Deschutes River of central Oregon (O’Connor and others, 2003a), the Umatilla River of eastern Oregon (Hughes and others, 2006), and the Umpqua River of western Oregon (Wallick and others, 2011).

Historical Modification

Much of the analysis of status and trends for the channels and flood plains of the Sprague and lower Sycan Rivers stems from maps, photographs, and observations from the 1860s through 2010 (table 4). This period, however, coincides with the main period of human manipulation of the channel and flood plain for purposes such as agriculture, water supply and delivery, transportation infrastructure, flood protection, and channel and flood-plain habitat restoration. Consequently, analyses and observations are in part affected by these human-caused changes to the system, prompting this description of the major historical changes.

Early Observations of Flood-Plain and Channel Conditions

The earliest systematic descriptions of the study area were recorded in the maps and notes from General Land Office cadastral surveys (table 3). The maps and notes provide information on channel width and position as well as vegetation. Summary observations from the field notes accompanying GLO surveys have been transcribed onto modern USGS topographic quadrangles by the Oregon Institute of Technology and Shaw Historical Libraries (2011). These observations, spanning 1866–1892, typically describe the geomorphic flood plain of the wider valley segments as “land-level; soil-1st rate.” Many of these wider flood plain areas are also described as “prairie.” Unique among the wide flood-plain segments is the Upper Valley, where several section corners between FK 96.9 and FK 94.3, just upstream of the North Fork confluence, were noted as “swampy” during the course of the September 1866 survey of T36S-R14E, also indicated on the corresponding plat map for which parts of sections 16, 17, 20, 21, 22, 27, and 28 are annotated with swamp symbols. The GLO descriptions for the confined segments mainly note the rocky and forested terrains flanking the narrow flood plains.

These early GLO descriptions accord with subsequent maps and observations. A USGS land classification and timber survey encompassing the Klamath Quadrangle includes the study area downstream of FK 92 (Leiberg, 1900), showing most of the Sprague Valley as “non-forested as marshes, meadows and agricultural lands,” and, near the town of Sprague River, “bottom lands, mostly grass covered, bordering Sprague River...” (p. 416). Similarly, a preliminary report on water resources for the Klamath Indian Reservation (Henshaw, 1912) and field notes of botanist W.E. Lawrence summarizing observations in July 1922 and July 1934 (W.E. Lawrence Field Journals, Oregon State University Herbarium, Corvallis, Oregon), note extensive and untimbered bottom lands flanking the Sprague River.

Early USGS records of streamflow measurements and related activities also provide information on channel and flow conditions, as well as documenting the effects of diversions and log-driving activities on the Sprague River (USGS Oregon Water Science Center, Portland, Oregon, streamflow measurement station records). In particular, station notes and photographs for USGS streamflow measurement sites on the South Fork Sprague River near Bly, North Fork Sprague River near Bly, Fivemile Creek near Bly, the Sprague River near Beatty, Sycan River near Beatty, and Sprague River near Chiloquin provide evidence of channel conditions dating back to 1904, with extensive accounts for the period between 1912 and 1930. Observations for the 1920s along the South Fork Sprague River (Station ID 11495500) included accounts of “hardpacked gravel” substrate and that “willows may affect extremely high water.” Likewise for the North Fork Sprague River (Station ID 11496500), K.N. Phillips noted on 18 May 1925 the “Channel very crooked near gage; no well [de]fined control...Willows on banks, left bank low...” And for Fivemile Creek (Station ID 11497000): “A crooked channel lined with a thick growth of alders was not conducive to a stable stage-discharge relationship for the site used 11 September 1917, to 11 May 1919...The pumice sand composition of the soil also contributed to constantly shifting conditions.”

In the Beatty Gap area, 1912 reconnaissance notes by H. Kimble record “The Sprague River from its forks near the eastern Indian Reservation line [near FK 93.0] westward to mouth of Sycan R. flows in a very deep channel and at very sluggish velocity and has many stagnant bychannels” (USGS Oregon Water Science Center, Portland, Oregon, streamflow measurement station records; Station ID 11497500; Sprague River near Beatty). Several early 20th century accounts indicate mobile channel conditions and sand and gravel substrate for the Sycan River (Station IDs 11499000, 11499100), as well as measurement records affected by substantial late summer aquatic growth.

For the Sprague River within the Kamkaun Spring reach, station records for gage 11501000, located at FK 28.6 between July 1920 and Sept 30, 1931, note “Current sluggish everywhere; right bank well defined and high, left bank rather low, and some willows; bottom of hard pan overlain with pumice sand, probably shifting very slowly.” (USGS Oregon Water Science Center, Portland, Oregon, streamflow measurement station records). Additionally, records for this location summarize some of the perturbations to the Sprague River associated with early European settlement which ultimately forced relocation of the station. These alterations include construction of the OC&E Railroad, the 1926 construction of a diversion dam 2 km downstream at FK 27.2, upstream diversion dams affecting flow as early as 1920, occasional sudden fluctuations caused by operation of upstream regulating dams, and frequent log drives (figure 20) that apparently led to many log jams (USGS Oregon Water Science Center, Portland, Oregon, measurement station records for station 11501000).

Evidence for changing beaver, otter, and muskrat population was noted by some of the streamflow measurement station records as well as a 1943 wildlife resource report on the Klamath Reservation by A.W. Moore (unpublished 1943 Wildlife resource report by A.W. Moore, U.S. Fish and Wildlife Service, in Klamath Indian Agency records, National Archives, written commun. from Larry Dunsmoor, The Klamath Tribe, 2006). Moore reported “Otter at present are practically extinct on the Reservation although some are occasionally reported from the lower Williamson River. Reports are that at one time this animal was common to all of the area.” and “The muskrat is a relatively recent addition to the Reservations fauna. Introduction was through the establishment of a muskrat farm on Crooked [C]reek, a few miles north of the Agency.” And regarding beaver, “Evidences are that at one time the Reservation was abundantly supplied with this animal. A thorough reconnaissance along the Sprague river from the Modoc Point diversion dam [Chiloquin Dam; FK 1.3] to the island, in Section 26 R. 8 E. Twp. 34 S., a distance of about 9 miles [near FK 21.8], showed 32 old colony sites with their canals extending, in places, 100 yards from the river’s banks. At present but three beaver were along this area. One was a solitary migrant, while at Braymill, a pair had established themselves and seven slides existed.”

These early observations indicate pre-European-settlement flood plains that were typically grass-covered in the wider valley segments but forested in the narrow valley segments. The South Fork valley segment was swampy, even in September, an observation consistent with the organic-rich soils evident in the stratigraphy at the section described at FK 98.5 (figure 8). The Sprague River was locally multichanneled and composed of sand and gravel substrate thinly or patchily covering indurated silt and clay. Channels were at least locally flanked by willow and colonized by beaver and otter. The Sycan River within the Lower Sycan valley segment was sandy and mobile.

Ditches, Dams, Levees, Field Leveling, and Roads

According to Helfrich (1974), digging of diversion ditches and temporary dams along the Sprague River started as early as 1890, soon after first European settlement. By 1904, 37 km of ditches were irrigating parts of the South Fork, North Fork, and Upper Valley segments (Kent, 1905). And by 1920, 37.6 km2 were irrigated in the Sprague River basin, mainly by ditches within the wider flood-plain portions of the study area (La Rue, 1922). Additionally, Chiloquin Dam was constructed in 1916 at FK 1.3 to divert water from the Sprague River to the Modoc Irrigation District.

Extensive ditch and dike networks control flow within wider flood-plain segments, particularly in the South Fork valley segment (table 6; figure 36). By 1968, the South Fork valley segment had more than 120 km of ditches and nearly 60 km of levees partitioning, draining, and diverting flow along the 13.1 km length of flood plain. By 2000, total levee length exceeded 80 km. No other valley segment has the same density of flow manipulation structures, but all of the unconfined reaches do have substantial lengths of ditches and levees. The length of ditches and levees has increased in most valley segments for most time periods, but most construction was between 1940 and 1968, partly resulting from a U.S. Army Corps of Engineers channelization program in the 1950s (Rabe and Calonje, 2009, p. 2–7). Since 1968, the length of levees has generally increased, whereas total ditch length has remained the same.

Associated with the locally extensive ditch network are many small dams and diversions. The vast majority are in the South Fork valley segment, where 110 structures were evident on 1968 aerial photographs (table 6). The North Fork, Lower Sycan, Council Butte and Buttes of the Gods valley segments all had at least 10 diversion structures in 2000. Several low dams were built across the Sprague River between 1916 and 1926 to facilitate water diversions and mill and log transport operations (USGS Oregon Water Science Center, Portland, Oregon, measurement station records for station 11501000). Remnants of these structures, chiefly pilings and dumped rock, are still visible at many former diversion locations along the river. With the 2008 removal of Chiloquin Dam, no structures currently span the main-stem Sprague River.

The total road length and developed area within the flood plain has increased in all segments between 1940 and 2000, with the exception of the Chiloquin Canyon and Beatty–Sycan segments  (figure 36). The increase in road length noted from aerial photograph mapping is in addition to earlier wagon, auto, and rail roads constructed in conjunction with first settlement and expanded timber harvest in the early 1900s. In particular, the railroad alignment locally limits flood-plain inundation.

Another evident flood-plain modification is leveling or smoothing (figure 37), although it was not systematically measured or mapped during this study. Such smoothing and leveling, facilitated in recent years by global-position-system-controlled land leveling systems, aims to improve efficiency of flood irrigation—the most common irrigation practice in the Sprague River watershed (Natural Resources Conservation Service, 2009). Leveled fields within the Sprague River geomorphic flood plain appear as areas where channel and scroll-bar topography has been obscured, and are typically associated with dense and regular networks of irrigation ditches. Leveled areas are most common in the Upper Valley and South Fork valley segments, although it is likely that all unconfined segments have areas of some field leveling.

Channel Rehabilitation and Remeandering

More recently, several channel restoration and rehabilitation projects have involved channel and flood-plain manipulation. These projects aim to improve aquatic and flood-plain habitat conditions, and include activities such as riparian fencing, bank stabilization (with vegetation plantings and placement of rock and large wood), levee breaching, and channel manipulation. These activities generally date from the early 1990s, and most have been done in coordination with the U.S. Fish and Wildlife Service. The most physically manipulative of these activities have been the channel plugs and remeandering, whereby the channel course is altered or moved by plugging and excavation, generally to place the channel in a former course. In some instances, the new channel course is protected by rock and anchored logs placed along banks (figure 23). At least five such projects have been conducted within the study area since 2004, including sites in the Kamkaun Spring, Council Butte, Beatty-Sycan, North Fork, and Sycan valley segments (Sue Mattenberger, U.S. Fish and Wildlife Service, written commun., 2011).

Channel Planform: Spatial and Temporal Trends

Broad-scale attributes of flood-plain and channel planform that are geomorphically and ecologically important include channel pattern, channel sinuosity, channel width, and the number and type of secondary channels. Important finer-scale attributes include bank morphology, bar area, bar frequency, and habitat measures such as pool frequency. Our analyses provide quantitative assessments of temporal changes for several of these channel and flood-plain attributes, drawing mainly from the historical maps and photographs.

Channel Pattern

Channel pattern characteristics such as sinuosity and frequency of secondary channels and backwater areas are ecologically important because they increase in-stream habitat complexity at all flow stages and provide refugia for fish and other aquatic organisms during high flow stages (National Research Council 1992, 1996; Gregory and Bisson, 1997). Planform channel pattern varies from straight in confined reaches to highly sinuous in some of the unconfined valley segments (figure 17). For the Kamkaun Spring, Buttes of the Gods, Council Butte, Beatty-Sycan, Beatty Gap, North Fork, and Lower Sycan valley segments, sinuosity approaches or exceeds 1.4. For the Chiloquin Canyon, Braymill, S’Ocholis Canyon, and South Fork valley segments, sinuosity is less than 1.2 (figure 38). The South Fork valley is unconfined, but the low sinuosity here owes to channel straightening and confinement by levees (figure 31B), which in turn has resulted in channel slope more than twice that of the adjacent downstream Upper Valley segment. For all unconfined reaches, sinuosity has decreased between 1940 and 2000 (figure 38).

Within most of the study area, low flow is confined to a single channel. In total, secondary channels in which flow is conveyed during base flow conditions occur along about 25 percent of the flood plain of the main-stem Sprague River and South Fork Sprague River (figure 39). Few or no secondary channels convey flow in the Lower Sycan and North Fork valley segments. The vast majority of secondary channels are in the unconfined reaches, particularly the Kamkaun Spring and Council Butte valley segments, which also have the lowest overall channel slopes.

The types of secondary channels present in the Sprague system include (1) short channels at the scale of a single meander, where a chute has developed across the meander core or neck but both chute and meander are still open and flowing and (2) anabranches that are longer than a single meander (figure 40). Such anabranches are distinct from channel braids in that they are separated from the main channel by vegetated and stable flood-plain surfaces rather than active bars. The degree of anabranching varies among segments in the Sprague River basin, but anabranches are particularly distinctive in the upper part of Council Butte Segment and just downstream of the Sycan River confluence where they attain lengths of up 2.8 km and have been present and stable since 1940 (figure 40c). The stability of these anabranches is consistent with the overall absence of temporal trends in the area of secondary channels between 1940 and 2000 (figure 38).

Backwater areas, defined as side channels connected to the primary channel but without through-flow at low flow (sometimes termed blind channels), occupy about half of the total area of secondary channels (figure 39). Their spatial distribution follows that of secondary channels, with 90 percent of the total backwater area being in the unconfined Kamkaun Spring, Buttes of the Gods, and Council Butte valley segments. For these three valley segments, the total area of backwater has decreased by 32 percent between 1940 and 2000, although backwater area has increased since 1968 in the Council Butte valley segment (figure 38).

Channel Width

Channel width is an ecologically important attribute affecting in-channel hydraulic conditions, flow conveyance and stream temperature (by affecting exposure to solar radiation). Channel width may change over time as a result of many causes, commonly including human disturbance, land-use, and changes in hydrologic regime (Wolman, 1967; Hammer, 1972; Kauffman and Krueger, 1984; Simon, 1989; Trimble and Mendel, 1995; Simon and Rinaldi, 2006). Consequently, stream width is a system attribute commonly scrutinized in fluvial assessments. Watershed assessments of the upper and lower Sprague River basin conducted by the Klamath Basin Environmental Foundation proposed that there has been historical stream incision and widening in some streams within the Sprague watershed, but that the extent of this disturbance is unknown (Connelly and Lyons, 2007; Rabe and Calonje, 2009). Similar observations and inferences throughout the Upper Klamath Basin have led to the regional management goal to decrease channel widths (Oregon Department of Environmental Quality, 2002).

Within the Sprague River study area, historical changes in channel width were assessed in two manners. The first relies on measurements associated with the GLO cadastral surveys of the late 1800s in comparison with channel width as shown on the aerial photographs from 2000. The second approach documents changes since 1940 on the basis of our channel mapping from the sequential aerial photograph analysis.

Analysis of Channel Width from General Land Office Surveys

The original GLO surveys in the Sprague River basin date from 1866 to 1892 (table 3), corresponding to first European settlement but predating significant disturbance of the valley bottom and channel system. These surveys aimed to facilitate land parceling and distribution; consequently, information regarding the channel was incidental to these goals. Nonetheless, the notes and maps recorded by the surveyors as they measured the township grids and surveyed bank lines provide information on channel conditions and geometry. In general, channel width was noted in two ways: (1) For all surveys of the township boundaries and its subdivisions (1-mile square sections), surveyors indicated locations along section lines where they crossed water and the width (in the direction of the surveyed line) of the water body. (2) For the meander surveys, which included the Sprague River only within parts of the Chiloquin Canyon, Council Buttes, and Beatty–Sycan valley segments, bank lines were surveyed to better document river location and available acreage for homesteading and cultivation. The bank lines of these meander surveys were marked by posts or stones at intersections with section lines.

In analyzing the river-width observations from the GLO surveys, an important uncertainty is the nature of the measured width—was it actual water width at the time of survey, or bank-to-bank width (similar to bankfull or active channel width)? In many situations, meander surveys can provide continuous channel-width information (for example, O’Connor and others, 2003c), but for the Sprague River, meander surveys were done for only a limited section of the river, and only one bank was surveyed, so no actual measurements of river width can be obtained for these meander surveys except for where they were monumented at section-line intersections. The observations along section lines are more difficult to extract, requiring access to the surveyor notes, but may provide better information on actual channel width (Knox, 1977). The 1855 manual to GLO surveyors instructed surveyors to measure the widths of water features on the section lines, including “all creeks, rivers and smaller streams of water which the line crosses,” with no mention of banks (White, 1983, p. 17), although practices varied from surveyor to surveyor and with time (Collins and others, 2003).

To clarify the distinction between the widths indicated by the meander surveys and those from the surveys along the section lines, we assessed the GLO surveys for T36S R10, 11, and 12E in the Council Butte valley segment. The section lines for these townships were surveyed in 1866, during which river width was measured along each section line as the surveyor crossed it. In July and August 1872, another surveyor returned to these townships to survey subdivisions of some sections (into 20 acre lots) and to survey the meander course of the river. While this later survey only surveyed the south bank line, surveyed monuments of both banks were placed at the intersections of the section lines, allowing direct comparison with the channel width as noted in the 1866 survey. This comparison shows that for 13 of the 17 section lines intersected by the meander survey, channel width noted in the 1866 section-line survey was less than the 1872 width from the meander survey, ranging from 22 to 94 percent (median value 67 percent) of the 1872 width (table 7). Of the four 1872 measurements showing a narrower channel width, two appear erroneous since the noted measurements do not close properly along the length of the section line (resulting in a negative calculated 1872 channel width). The other two measurements indicating a channel narrower by 10–15 m in 1872 may also reflect survey error or possibly channel change in the 6 years between surveys. In sum, this comparison indicates narrower measurements associated with the section line crossings of 1866 relative with the meander surveys of 1872, consistent with the inference that the meander surveys are of bank edges whereas the section line surveys denote actual water edges, as has been assumed in previous studies (Williams, 1971; Wallick and others, 2006).

Assuming that the section line surveys were measurements of actual wetted width, we assessed 143 section line crossings in the study area from the surveyors’ notes, available from the Bureau of Land Management (2011). For section lines where the river was reported as crossing the section line perpendicularly, the GLO river width was not adjusted. But for locations where the intersection between the channel and section line were not perpendicular, measurements of channel width along the section line were converted to width perpendicular to flow at the time of survey by multiplying the observed width by the sine of the angle between the direction of river flow and the surveyed section line. For most section lines, surveyors noted river flow direction only to the nearest cardinal or intercardinal direction, but for some sections it was recorded to the nearest degree. In addition, for 16 section lines in the Council Butte and Beatty Sycan valley segments, river flow directions were obtained from the plat maps showing 1872 meander surveys, assuming little change since the 1866 section-line survey.

Sixty of the 143 crossings were not fully analyzed because (1) they were highly oblique (> than 60 degrees) to the channel direction, (2) they were at a location where comparison to modern channel widths would be challenging, such as at apex of a meander bend or at a tributary confluence, or (3) no channel direction information was provided. Another 21 crossings were excluded because the GLO survey records of channel width and orientation were inconsistent with topography, indicating errors in the GLO survey notes.

Excluding the less reliable measurements left 62 late 19th-century channel-width observations at known locations (table 7). The spatial distribution of these measurements is uneven, ranging from only one observation along the combined length of the Chiloquin Canyon and Braymill valley segments, to 14 observations in the Council Butte valley segment. The resulting channel widths (perpendicular to flow) ranged from 6 to 58 m, with many of the widest sections being in the Kamkaun Spring and Council Butte valley segments. The narrowest channel widths were in the South Fork, North Fork and Coyote Bucket valley segments. The GLO width measurements were compared to channel width as measured from the 2000 aerial photographs at the same locations. At the six sites where the 2000 aerial photographs showed bars flanking the channel at the section line crossing, the width of the bar was included in the channel width. All measurements from the 2000 aerial photographs were perpendicular to flow direction. Similar to the GLO measurements, measured channel width in 2000 ranged from 7 to 51 m.

Although the discharge at the time of the GLO surveys is unknown, the water widths on the GLO surveys and on the 2000 photographs are likely comparable because all were measured at relatively low flow. The 2000 aerial imagery was flown between late July and mid-August (table 4). The GLO surveys were between July and November, the five lowest flow months of the year based on long term average discharge at Chiloquin (figure 3).

Whereas different discharges associated with the surveys and measurements may introduce a small but unknown bias, more significant are transcription and survey errors that affect some of the observations, as apparent from the clear miscues noted in comparing the 1872 meander survey and 1866 GLO section-line surveys. An even more likely pervasive error results from adjusting the GLO measured channel widths to those perpendicular to flow direction. For locations where the angle of intersection was within 25 degrees of perpendicular, likely errors in the adjusted width are small—less than 10 percent. But, a ± 25 degree uncertainty about a 45 degree intersection angle results in a ± 30 percent uncertainty in the resulting adjusted width. For lower incidence angles, the resulting uncertainty is even greater.

The resulting 62 comparisons between channel width surveyed by the GLO between 1865 and 1892 (most were surveyed between 1865 and 1869) and those measured at the same location from 2000 aerial photographs show a wide range of change (appendix F; figure 41). While there are substantial discrepancies at some sites—in places the channel in 2000 is as much as four times as wide as it was at the GLO survey and in others it is less than half of its previous width—some overall trends are evident. In general, channels that were wide during the GLO surveys became narrower and those that were narrow became wider. The trend of the measurements crosses the 1:1 line between channel widths of 20 and 30 m (figure 41a), indicating that this width range generally separates the channels that widened from those that narrowed. Overall, the median value of the ratio between the 2000 width and the GLO width is 0.96, indicating slight overall narrowing of the measurement sites, but the difference is small relative to the variance and the measurement uncertainty.

Stratifying the observations by valley segments reveals trends within specific reaches (figure 41b). Both the Kamkaun Spring and North Fork valley segments have apparently narrowed between the GLO surveys and year 2000, as possibly has the Beatty Gap valley segment with three of its four measurements showing narrowing. By contrast, four of the five comparisons in the South Fork segment indicate widening. Overall widening trends are also possible in the Council Butte, Beatty-Sycan and Coyote Bucket segments, although data are either few or the differences are small.

Analysis of Channel Width from Aerial Photographs

More systematic but covering a shorter period are observations of channel width derived from historical aerial photographs. For this analysis, channel width for each of the 1940, 1968, and 2000 photograph sets was measured on transects spaced 200 m apart along the channel centerline. For these measurements, the wetted width orthogonal to the channel centerline was considered the channel width. Because the channel centerline shifted between years, transect locations are not identical for each time period. But the large number of observations—for example, 932 width measurements within the study area from the 1968 aerial photographs—permits observation of general trends in channel width both spatially and temporally (figures 42 and 43).

The overall median primary channel width for 682 measurements along the main-stem Sprague River in 2000 was 30.3 m. For the North Fork Sprague River, median width (for 79 measurements) was 10.0 m, for the South Fork Sprague River (74 measurements) 9.3 m, and for the Sycan River along both the Lower Sycan and Coyote Bucket segments (152 measurements) 14.1 m. For all time periods, the channel is generally widest in the unconfined segments, particularly the Buttes of the Gods, Council Butte (where width locally exceeds 100 m) and Beatty–Sycan segments. The channel is also wide in parts of the Kamkaun Spring segment as well as in the lower part of the Chiloquin Canyon segment near the Sprague River confluence with the Williamson River and where impounded by Chiloquin Dam at FK 1.6.

The channel was generally widest in the 1968 aerial photographs, when the median width of the main-stem Sprague River was 32.4 m compared to 29.9 m for 1940 and 30.3 in 2000. This was the case for most individual valley segments as well (figure 43); of the valley segments with complete aerial photograph coverage for all three years, only the Braymill and Council Butte valley segments had wider channels in either 1940 or 2000. For six segments (Kamkaun Spring, S’Ocholis Canyon, Buttes of the Gods, Beatty–Sycan, Beatty Gap, and Upper Valley), widening was substantial, ranging up to 60 percent of the 1940 width in the case of Upper Valley. Between 1968 and 2000, most segments narrowed to widths similar to those measured from 1940 photographs. For some valley segments, such as Lower Sycan and Beatty-Sycan, narrowing decreased median channel width by as much as 20 percent. The Council Butte valley segment is unique in having an apparent systematic decreasing trend in median width between 1940 and 2000 (figure 43). The South Fork valley segment shows a similar trend, but the 1940 aerial photograph coverage is incomplete for this segment. The variance in channel width has decreased since 1940 for several valley segments, particularly upstream of S’Ocholis Canyon (figure 43).


Bars are significant because they are indicative of bedload transport and contribute to physical habit complexity. Active sand and gravel bars are relatively sparse for much of the Sprague River and its major tributaries (figure 39). In total, bar area in the study area at the time of the 2000 aerial photographs was about 330,000 m² (33.0 hectares [ha]), less than 10 percent of the corresponding primary channel area. Bars are more broadly distributed than secondary and backwater channels, with a few bars flanking the channel in all valley segments (figure 18). Bars are generally more common and larger in the unconfined reaches, although the confined Beatty Gap and S’Ocholis Canyon valley segments both have abundant bars. Bars are sparse in the narrowly confined Chiloquin Canyon and Braymill valley segments.

Bars are locally abundant where the North Fork Sprague River and Sycan River exit steeper and more confined segments. For the North Fork Sprague River, this corresponds to the upper end of the study reach where the North Fork emerges from the Fremont National Forest. For the Sycan River, the greatest bar frequency is at the exit from the steep and confined Coyote Bucket valley segment ( figures 18 and 39). Similarly, many of the few small bars in the South Fork valley segment are at the upstream end of the study area, where the river emerges from the foothills east of Bly. Bars are also concentrated near the Sycan River confluence within the Council Butte and Beatty–Sycan valley segments.

In nearly all valley segments, bar area was greater in 2000 than it was in 1940, with the total area (excluding the North Fork and South Fork valley segments for which the 1940 photographs are incomplete) increasing from 18 ha to 29 ha (figure 38). It is not clear, however, that this is a long-term trend because the area of bars was far larger in 1968, when bar area totaled 67 ha. This high total in 1968 almost certainly owes to enhanced sediment transport and consequent bar formation and expansion during the 422 m3/s flow of 26 December 1964, the largest flood since at least 1920 (figure 2).

Channel Migration

Closely related to processes affecting channel planform are those that control the rates and styles of lateral channel migration. For the Sprague River study area, we drew inferences from the stratigraphic analyses as well as from the sequential photographs, which together document the relative importance of processes such as lateral migration, meander cutoffs, and avulsions in controlling the overall channel pattern and rates of evolution.

Observations from Stratigraphy and Morphology

The stratigraphic exposures, augering transects, and surficial morphology indicate lateral bar growth and channel migration, followed in many instances by channel abandonment, presumably by meander cutoff or avulsion. Lateral flood-plain building is evident from the dipping beds visible in stratigraphic exposures along the North Fork Sprague River (figures 9 and 11) and Sycan River (figure 14). The scroll-bar morphology locally evident on low and young flood-plain surfaces (figure 13) indicates lateral channel migration in conjunction with sandy point bar development, an observation consistent with point bar deposition during the 2006 high flows (figure 15).

Many of the meander bends, however, appear to have been abandoned rapidly, leaving distinct and isolated oxbows or swales. The stratigraphy of deposits filling these former channels show abrupt transitions, with basal bedrock strath surfaces, gravel, and sand associated with active channel environments overlain by peat, silt and clay (figures 12 and 13). At some sites, such as 9/22/05-4 at FK 48.7 (figure 12), where the paleochannel is substantially filled with sand and gravel, meander cutoff or channel avulsion may have been promoted by point-bar growth blocking sharp bends. Prehistorical avulsions, particularly in the more densely timbered confined reaches, may have been triggered by log jams.

Also evident from the stratigraphic analyses is that the channel migration only slowly works its way across the flood plain. Some of the preserved paleochannels are nearly 4,000 years old (figure 12), and many flood-plain sections have been accumulating overbank deposits for several hundreds and, in places, thousands of years subsequent to the last channel occupancy (figures 5, 8, 9).

Centerline Analysis

More quantitative description of channel migration can be derived from the historical maps and photographs, although rates and mechanisms observed since the first maps of the 1860s and the photographs beginning with 1940 have likely been affected by land use and channel manipulation. Channel migration was measured in the manner of O’Connor and others (2003c), in which channel centerline position for each time period was evaluated in reference to flood-plain transects spaced 200 m along the flood-plain centerline. The change in position of the intercept between successive channel centerlines and the flood-plain transects indicates the net lateral movement with respect to the flood-plain reference system. Migration rates were calculated on the basis of this net movement divided by the time interval between mapped positions. Centerlines were mapped for the meandered sections of the GLO maps in the Council Butte and Beatty-Sycan valley segments, as from the 1940, 1968, 1975, 2000, and 2005 aerial photograph and map sources (table 4). Portions of the South Fork, North Fork, Lower Sycan, and Coyote Bucket valley segments are incompletely covered by the 1940 or 1975 photographs and maps; consequently, migration rates encompassing those time periods and segments are incomplete.

Calculated migration rates for various periods between about 1870 and 2005 are mostly less than 0.5 m/yr but range up to 30 m/yr (figures 44 and 45). Migration rates vary spatially and temporally, with the overall greatest migration rates in the unconfined valley segments, particularly the Kamkaun Spring, Buttes of the Gods, and Council Butte valley segments. The unconfined South Fork valley segment is an exception, with low migration rates (particularly since 1975) similar to those measured for the Chiloquin, Braymill, and S’Ocholis Canyon valley segments.

Migration rates are spatially and temporally correlated; reaches of high and low mobility tend to extend for several kilometers and to persist for multiple measurement periods (figure 44). Additionally, some reaches for some periods have exceptional migration rates exceeding 5 m/yr. These very high rates owe to the channel moving far distances laterally by translocation across the flood plain rather than by continuous migration from lateral bank erosion and point bar development. In most cases where migration rates exceeded 1 m/yr, avulsion or meander cutoff was responsible. The median migration rates, ranging from about 0.06 to 0.3 m/yr for most segments (figure 45), probably reflect rates of continuous lateral shifting by bank erosion.

For most valley segments, the 1968–1975 period had the highest migration rates (figure 45). But this period appears singular within an overall declining trend of historical migration rates, with the 2000–2005 period having the lowest median migration rate for 10 of the 13 analyzed valley segments. This declining trend is particularly evident for the Council Butte and Beatty–Sycan valley segments for which migration rates were calculated dating back to the 1870’s GLO meander maps.

Channel migration rates on the Sprague River are slow in relation to many other alluvial rivers of western North America. Mean migration rates range between 0.16 and 0.59 m/yr for all analyzed time periods on the main-stem Sprague River. By contrast, the gravel-bed rivers of the Olympic Peninsula have migration rates ranging from 2.1±1.7 m/yr for the lower Elwha River (Draut and others, 2011) to 8.8±4.1 m/yr for the Quinault River (O’Connor and others, 2003c). Nanson and Hickin (1986) reported average migration rates of 0.57 to 8.3 m/yr for 18 river reaches in western Canada, and Giardino and Lee (2011) document an average migration rate of 3.28 m/yr for the Brazos River, Texas, for an analysis period extending back to 1929.

Expressed in terms of channel width, most segments had median migration rates of less than 0.01 channel widths per year (figure 46). Such rates place the Sprague River well into Hooke's (2003) class of stable meanders and in the low end of the range of other low-gradient rivers like the Sprague River for which meander migration rates range from 0.0003 to 0.23 channel widths per year (Dietrich and others, 1999; Gilvear and others, 2000; Brooks, 2003; Micheli and others, 2004; Rodnight and others, 2005; Gautier and others, 2007).

Channel cutoffs and avulsions

The locally high channel migration rates evident for short reaches within the unconfined valley segments (figure 44) resulted from meander cutoffs and channel avulsions. Such events result in rapid translocation of the primary channel, leaving secondary or backwater channels in their wake (figure 40). Mapping changes in channel position from the historical aerial photographs and GLO maps (where the channel was meandered in 1872 within the Council Butte and Beatty–Sycan valley segments) shows cutoffs and avulsions widely distributed throughout the Sprague and Sycan River system but more common in the wider and unconfined valley segments (figure 47). When normalized by length of channel available for comparison and time period, it is evident the periods between 1940 and 1975 had the greatest frequency of avulsions, with rates more than twice that of any other time period (table 8). This finding is consistent with the locally very high lateral migration rates determined for those time periods from the flood-plain transect analysis of lateral migration (figure 44) and probably reflect effects of the 1964 flood or possibly human modifications of the channel and flood plain.

Channel Incision

Although long-term channel incision or downcutting is the inevitable consequence of drainage systems eroding landscapes, short-term episodes of channel incision into flood plains can lead to a host of morphologic, ecologic, and water-quality consequences to the fluvial system (Williams and Wolman, 1984; Schumm, 1999; Simon and Rinaldi, 2006; Shields and others, 2010; Cluer and Thorne, 2013). Such short term incision can be triggered by land-use changes, changes in hydrologic regime, and channelization, among other causes. Once incision begins in alluvial channel systems, it typically propagates upstream and leads to an enduring sequence of channel evolution involving downcutting and widening before establishment of a new flood-plain inset within the original flood-plain surface (Simon and Hupp, 1986; Cluer and Thorne, 2013). In certain environments, recovery to pre-incision states requires decades or centuries (Beechie and others, 2008), while in others a compound cross-section consisting of a narrow flood plain laterally constrained by paired terraces may be more-or-less permanent over the same timescale. Consequently, preventing or reversing channel incision is a common goal of watershed and channel restoration (Palmer and others, 2005; MacWilliams and others, 2010), and has been an issue of concern in the Sprague River basin (Connelly and Lyons, 2007; Rabe and Calonje, 2009).

Stratigraphic Analysis

The stratigraphic records and augering transects provide a century-to-millenial scale perspective on historical channel incision. Along the main-stem Sprague River and lower Sycan River, the stratigraphic sections recording older stratigraphy locally show pre-Mazama fluvial gravel as much as 2 m above the present channel bed (figure 4), indicating channel and flood-plain formation above present levels at times prior to 7,700 years ago. Subsequent incision may reflect overall valley excavation, but more likely the higher channels and flood plains (now terraces) of the late Pleistocene and early Holocene reflect elevated sediment loads and flow regimes, similar to conditions in the Willamette Valley (O’Connor and others, 2001) and the Oregon Coast Range (Personius and others, 1993) where greater sediment (owing to more intense physical weathering) and water fluxes during colder Pleistocene climates resulted in regional aggradation.

But subsequent to the Mazama eruption of about 7,700 years ago, the stratigraphic exposures and augering indicate little or no overall channel incision for most of the Sprague River study area. The base of paleochannels (abandoned as much as 3,800 cal yr BP) probed during the augering transects near FK 31 and 48 are as much as 1 m lower than the present channel to 0.2 m higher, indicating little net incision and perhaps more general aggradation of the valley bottom in the Kamkaun Spring and Buttes of the Gods valley segments during the late Holocene (figures 12 and 13). Similarly, the bank exposures of post-Mazama flood-plain stratigraphy along the main-stem Sprague River, North Fork Sprague River, and Sycan River show channel gravels nowhere more than 40 cm above the present base-flow water surface (figures 5, 8, and 9). Modern (2006) coarse sand and gravel deposition to similar or higher levels on point bars during high flows also indicate that the present channel is not incised (figure 14). Similar to the auger probes, several sections show channel deposits of the last 1,000–2,000 yr as much as 50 cm below the present channel bottom, also indicating late Holocene channel aggradation.

Stratigraphic exposures along both the North Fork Sprague River and Sycan River indicate episodes of flood-plain erosion and construction in addition to possible episodes of different hydrologic regimes. Along the North Fork Sprague River, an actively aggrading flood plain less than 300 yr old is inset into an older flood plain as much as 3,500–4,700 yr old (figures 9 and 11). The older flood-plain deposits along the North Fork Sprague River are distinctive because of a 0.5 m thick peat and diatomaceous horizon, indicating a significant period of wet flood-plain conditions sometime after 4,330 cal yr BP (figure 9).

Similarly, the Sycan River is flanked by low flood-plain surfaces that began accumulating overbank deposits as recently as 600 cal yr BP (figure 5). In some locations, this younger flood plain is inset against taller and older flood-plain deposits that began accumulating soon after the approximately 7,700 cal yr BP Sycan flood (Lind, 2009). For both the Sycan River and North Fork Sprague River, however, there is no evidence of substantial and persistent changes in channel elevation associated with these multiple episodes of flood-plain formation and erosion.

Channel incision is evident, however, in the stratigraphy exposed at the single investigated site along the South Fork Sprague River (figure 8). Here, the base of post-Mazama channel and bar deposits is about 50 cm above the present thalweg; and the top of a 60-cm thick diatomaceous peat layer, probably actively accumulating deposits at the time of the first GLO surveys (judging from surveyor descriptions of a wet flood plain during their September 1866 traverse), was more than 1.6 m above the water surface at the time of description.

Channel Centerline Elevation Changes

Trends in channel incision are more challenging to document quantitatively from historical sources such as aerial photographs than are planform changes. In some instances, repeat longitudinal topographic and cross section surveys can indicate spatial and temporal trends in channel incision (for example, Wallick and others, 2010; 2011). But, for the Sprague River study area, we know of no such surveys. So to evaluate possible channel elevation changes with time, we have employed the channel centerline mapping in conjunction with the LiDAR topography to estimate changes in channel elevation with time within the study area (figure 48).

This analysis entailed determining elevations from the 2004 LiDAR at intercepts between the flood-plain transects, spaced 200 m apart for the length of the study area, and channel centerlines, as mapped from each of the historical map and aerial photograph sources (figures 48 and 49). No analysis is possible at the many such intercepts where LiDAR elevations are missing because of specular reflection during the 2004 acquisition, mainly in the Kamkaun Spring valley segment. Also, for locations where the channel had not moved over the analysis period, the elevation of the intercept remained constant and, consequently, trends cannot be determined. But, for locations where the channel had shifted and LiDAR elevation is available, the difference in elevation can be determined between the channel intercept of the first location and that of subsequent channel positions. This difference owes to the combined consequences of (1) actual channel elevation changes, (2) differential rates of overbank aggradation and paleochannel filling within abandoned channel traces, and (3) errors in the LiDAR elevation data. Resolution of these errors is possible only where thalweg gravel elevations can be precisely surveyed for multiple channel centerlines at a single transect, as has been done with the augering transects for the older paleochannels (figures 12 and 13).

Because such surveys have not been conducted for the 1940–2005 centerlines, evaluation is approximate and emphasizes overall spatial and temporal trends rather than changes at specific transects. Local differences of less than about 1 m in channel elevations between time periods could relate to LiDAR elevation errors as well as paleochannel filling or lateral flood-plain accretion rather than changes in actual channel elevation, especially for the more recent and shorter time periods. Larger changes owing to deposition are plausible for the 1870–1940 period. Coherent patterns over several kilometers, and particularly over multiple time periods, however, likely indicate systematic changes in channel elevation with time.

From this perspective, the strongest evidence for channel lowering is within the South Fork and Upper Valley segments for most time periods (figure 49). Observations for the South Fork are incomplete because of limited aerial photo coverage for some periods, but channel lowering is apparent for much of the Sprague River above FK 90.0, perhaps extending downstream to FK 86.0 in the Beatty Gap valley segment. For the 1940–1968 period, apparent lowering exceeded 1 m in many locations with these three valley segments, with lowering also possibly evident in the Council Butte valley segment during this time period. But after 1968, changes in channel centerline elevation show no systematic changes along the main-stem Sprague River.

Incision in the North Fork Sprague River valley segment is difficult to assess because of limited aerial photo coverage, although local incision, or more likely, lateral migration with rapid bar and flood-plain building, may be reflected by up to 1.3-m difference in present centerline elevations between NFFK 2.0 and 4.0 for 2000 and 2005 channel locations (figure 49). The Sycan River shows little spatial or temporal systematic patterns in channel elevation changes, except for the downstream part of the Coyote Bucket valley segment, which likely reflects bar and flood-plain building.

This centerline analysis gives results consistent with the stratigraphic records: no evidence of substantial incision throughout most of the Sprague River corridor. The main areas of likely historical incision are in the South Fork valley segment, and possibly in parts of the Upper Valley and Beatty Gap valley segments, although this incision mostly pre-dates 1975.

Effects of the 2006 High Flows

High flows in 2006 (figure 50) provided an opportunity to investigate channel and flood-plain processes associated with the greatest flooding since WY 1999 (since WY 1996 on the Sycan River and on the Sprague River near Beatty Gap; figure 2). In WY 2006, mean daily discharge exceeded 80 m³/s on the Sprague River near Chiloquin (11501000) for multiple days in December and January as a consequence of a winter rain-on-snow event, with the largest instantaneous discharge of 101.7 m³/s on 5 January 2006. Earlier in this event, peak flow exceeded 86 m³/s on 1 January 2006 at the Sprague River gage (11497500) near Beatty Gap. Later, regional snowmelt raised flows in April and May, 2006, resulting in two peaks of about 90 m³/s near Chiloquin. The Sycan River also had its greatest flow of the year during this late April and early May snowmelt flow, peaking at 67 m³/s on 1 May 2006. The combined effects of the December-January and April-May runoff episodes were two peaks on the main-stem Sprague River, both exceeding 90 m³/s, but the first of shorter duration compared to the longer spring runoff to which the Sycan River also contributed significantly (figure 50).

Distribution and Character of Flood Deposits

Between 23 June and 15 July 2006, we surveyed the effects of these large flows by mapping and measuring deposit characteristics along most of the main-stem Sprague River. At each of 71 measurement sites (figure 51), spaced on average 1.7 km apart, we measured deposit thickness, texture, and elevation relative to the low-flow water surface. Where possible, we also surveyed the elevation of the highest evidence of flow from WY 2006 (figure 52; appendix G).

The goal of these surveys was to sample sites of evident deposition at about 1-km intervals along the river corridor. Sparse deposits in some reaches necessitated longer intervals, as did time constraints imposed by far-spaced river access locations. Consequently, intervals between sample locations are not even, with larger gaps between sites mainly in the Council Butte, Buttes of the Gods, and S’Ocholis Canyon valley segments. Each depositional site was classified as either a bar or flood-plain deposit from more detailed descriptions of each sampling site (appendix G).

At each sampling site, we excavated through the 2006 flood deposits to the preexisting surface, apparent at all sites by buried in-situ vegetation (figure 53). At some locations, we could infer separate deposits resulting from the multiple 2006 flood peaks on the basis of stratigraphic breaks or more definitively by young vegetation which had germinated in the intervening period between the early January peak and the April and May high flows. In general, however, distinguishing the deposits from the specific high flows was not possible, so the total thickness and texture of the deposit was measured and summarized for analysis. Texture was field-categorized into broad grain-size classes ranging from gravel to silt/clay.

The top of the excavated sequence at each site was surveyed relative to the Sprague River water surface by hand level and stadia. Additionally, we surveyed in the highest preserved evidence of 2006 flow stage, commonly consisting of flotsam or detritus entangled in vegetation and fencing. In locations of low flood plains, this evidence provides a minimum local estimate of the maximum 2006 flood stage. During the course of mapping, the Sprague River flow varied from 8.7 m³/s to 12.7 m³/s. Consequently, all surveyed elevations were adjusted to a reference discharge of 11.5 m³/s (which slightly exceeds the median daily flow of 9.7 m³/s) on the basis of the stage-discharge rating for the USGS gage 11501000 Sprague River near Chiloquin, Oregon, at FK 8.3. This resulted in elevation adjustments ranging between -0.22 and 0.08 m.

Maximum flood stages along the main-stem Sprague River were highest in the confined valley segments, such as Beatty Gap and S’Ocholis Canyon, and downstream in the Kamkaun Spring valley segment (figures 52 and 54). In several reaches, maximum flood stage was more than 2 m above the elevation of the 11.5 m³/s reference discharge. The lowest flood stages were in the wide and low-gradient Council Butte and Buttes of the Gods valley segments, where maximum 2006 flood stages were 1.2–1.5 m above the reference elevation. Even in these reaches, however, maximum flood stages were sufficient to inundate all or most of the flood plain.

Deposit distribution and characteristics varied among the valley segments (figure 52). The thickest, coarsest and some of the highest deposits are in the Beatty Gap valley segment, where coarse sand was deposited as high as 1.85 m above the reference discharge elevation. By contrast, we found no flood deposits more than 1 m above the reference discharge elevation in the Beatty–Sycan, Council Butte, or Buttes of the Gods valley segments. Coarse sand and gravel was typically deposited within 1 m of the reference discharge stage. Finer sand, silt, and clay were the dominant constituents of higher deposits.

The thickest and coarsest deposits were on bars, whereas finer and thinner deposits typified flood-plain depositional sites (figures 52 and 53). For example, the median 2006 accumulation on bars was 21 cm, whereas the median accumulation on flood plains was about 4 cm. All bar deposits were sand and gravel. By contrast, more than 60 percent of the flood-plain sites accumulated fine sand, silt or clay, typically at elevations higher than bar sites. Where stratigraphy was evident, inclined bedding indicated that most of the bars were formed by lateral accretion, contrasting with the vertical accretion indicated by the horizontally bedded flood-plain sediments. This pattern is consistent with the longer term stratigraphic records, especially those on the North Fork Sprague River and Sycan River where inset flood-plain relations are exposed; showing laterally accreted coarse sand transitioning up into vertically accreting finer sand and silt (figures 11 and 14).

Nearly all the sand and gravel particles, particularly in the coarse sand to granule range of 1–4 mm, were light-colored pumice grains (figure 53B). This observation is also consistent with the stratigraphic records from the length of the study reach, which show nearly all coarse sand deposits that post-date the Mazama eruption dominantly composed of pumiceous materials.

Measured Sediment Transport

In a sediment study commissioned by The Klamath Tribes, Graham Matthews & Associates (2007) measured streamflow and sediment transport at 14 sites in the Sprague River basin between February 2004 and January 2006. Included in this measurement period is the high flow resulting from the winter storm of late December 2005 and early January 2006 (figure 50). During these elevated flows, Graham Matthews & Associates (2007) report four to eight suspended sediment measurements from each of seven sites along the main-stem Sprague River, Sycan River, South Fork Sprague River, and North Fork Sprague River (figure 55). Five of these sites are within the study area; the other two are on the North Fork and South Fork Sprague Rivers less than 5.5 km upstream of the alluvial portion of the river corridors within the study area.

Flow at all sites peaked on 31 December 2005 except for the main-stem Sprague River site at Godawa Springs Road, which peaked the next day on 1 January 2006, and the Sycan River site at SYFK 2.3, which peaked a week earlier on 23 December 2005 (Graham Matthews & Associates, 2007). Sediment concentration measurements were associated with discharges ranging from 2.61 m³/s (with a suspended sediment flux of 0.0055 kilograms per second (kg/s)) on the North Fork Sprague River at Road 3411 (5.5 km upstream of the study area) to a discharge of 74.2 m³/s, with suspended sediment load of 38.8 kg/s, on the Sprague River at Godawa Springs Road (FK 78.2) (Graham Matthews & Associates, 2007). This latter measurement at the most downstream site was very close to the peak flow of 75.4 m³/s on 1 January 2006. In addition, this measurement had the highest measured sediment concentration of 523 milligrams per liter (mg/l). Measurements also at discharges within 95 percent of the peak discharge were obtained on 31 December 2005 for the North Fork Sprague River at NFFK 0.7, South Fork Sprague River at Picnic Area (1.5 km upstream of where the South Fork Sprague River enters the study area; equivalent to FK 107.6), South Fork Sprague River above Fishhole Creek (FK 100.1), and South Fork Sprague River at Ivory Pine Road (FK 94.3).

Aside from substantially elevated sediment concentrations near the time of peak discharge, all sites showed substantial hysteresis (changing sediment concentration relative to streamflow during a period of high flow) resulting from general declines in sediment concentrations and flux rates during the course of the flood (figure 55). This is most evident for the upstream and tributary sites, where sediment fluxes diminish by a factor of 10 or more relative to discharge on the North Fork Sprague River at Road 3411, South Fork Sprague River at Picnic Area, and Sycan River at Drews Road. Hysteresis is present but to a lesser degree on the other downstream sites, particularly the main-stem Sprague River at Godawa Springs Road, where sediment concentration declined by a factor of five during the flood.

Such hysteresis is commonly attributed to sediment depletion during the event. Sediment depletion may be more rapid in smaller streams and during initial wet-season flows (Williams, 1986; Smith and Dragovich, 2009). The pronounced peak in sediment fluxes and concentrations associated with peak flows in the Sprague River indicates entrainment from additional sediment sources at high flows, likely bank and channel materials, as well as from flood-plain diversion channels (Michael Hughes, Oregon Institute of Technology, written commun., 2013),  

The spatial evolution of sediment concentrations also clarifies sediment sources, especially along the South Fork and main-stem Sprague Rivers, for which closely timed measurements were made at four sites, including measurements at very close to peak discharge (figure 56; Graham Matthews & Associates, 2007). Particularly evident is the increase in suspended sediment concentration between the measurement sites at South Fork Sprague River at Picnic Area (FK ~107.6) and the next downstream site above Fishhole Creek (FK 100.1). At peak discharge and for the three measurements leading up to the peak flow, sediment concentration increased in this 6.5-km-long reach by factors ranging between 1.4 and 4.8, despite drainage area only increasing by a factor of 1.2. For the December 21, 23, 30, and 31 measurements, sediment concentrations diminished downstream from the measurement site above Fishhole Creek. These observations indicate significant sediment entrainment in the South Fork valley segment above FK 100.1. Because tributaries are few in this reach, some of this suspended sediment likely resulted from channel and bank erosion. Additional sources of sediment in this reach likely include eroding bank crevasses or cuts, which facilitate conveyance of sediment entrained from flood-plain surfaces and irrigation channels into the South Fork Sprague River (Michael Hughes, Oregon Institute of Technology, written commun., 2013).

Entrainment patterns apparently changed at or soon after peak flow. On 1 January 2006, sediment concentrations and total flux increased downstream, culminating with a sediment concentration of 523 mg/l (and a suspended load flux of 38.8 kg/s) on the Sprague River at Godawa Springs Road (FK 78.2). This pattern, although at much reduced levels, was also indicated by the 6 January 2006 measurements. The downstream increase in sediment concentration and load must owe to sediment input from sources downstream of the North Fork Sprague River confluence, because the 1 January North Fork suspended load at NFFK 0.07 accounts for less than 1 percent of the load for a nearly coincident measurement on the Sprague River at Godawa Springs Road. This sediment input was most likely derived from channel and bank erosion or tributary inputs in the Upper Valley or Beatty Gap valley segments.

Channel and Flood-Plain Processes

The geomorphic mapping and historical observations, in combination with the observations and trends evident from the map and photograph analyses and synoptic survey, provide a basis for interpretation of channel and flood-plain processes. In addition, these observations can support inferences as to how processes have been affected by system changes, either owing to long-term geologic conditions or human influences.

Channel Processes

Channel processes are broadly influenced by watershed-scale conditions of flow and sediment supply (in particular bed material) and, at the segment scale, channel slope. Consequently, attributes such as channel width, flow, and slope are strongly correlated. But, specific local morphologic characteristics and associated physical habitat conditions relate to local interactions among sediment transport, bed and bank materials, depositional processes, and lateral and vertical erosion.

Sediment Transport

Bed-material and suspended sediment transport are important channel-forming processes in the Sprague River basin. Judging from the stratigraphy, suspended load of very fine sand, silt, and clay is the dominant component of flood-plain deposits, therefore influencing bank erosion processes and riparian vegetation. Transport and deposition of bed-material composed of sand and gravel builds bars and consequently is a factor in building new flood-plain surfaces as well as promoting lateral channel migration, including cutoffs and channel avulsions.

Measurements of suspended load during 2004–2006 by Graham Mathews & Associates (2007) show that the South Fork Sprague River contributes about 60 percent of the total suspended load to the Sprague River, with the North Fork Sprague River (27 percent) and Sycan River (13 percent) contributing markedly less. Their overall measurement program as well as the measurements during the 31 December 2005 and 1 January 2006 flood show that a substantial portion of the suspended sediment derives from the North Fork and South Fork valley segments (figures 55 and 56). This sediment is possibly the result of lateral channel erosion, bed incision or tributary input. The historical evidence of possible incision in both the North Fork and South Fork valley segments suggests that bed incision, either within the two forks themselves, or possibly by incising tributaries and irrigation ditches graded to the two forks, is probably an important factor in the elevated suspended sediment loads. This sediment may be derived in part directly from bed erosion, but probably more significantly by enhanced channel widening within incising reaches.

Some of the suspended load is deposited on flood-plain surfaces, resulting in the vertical accretion evident in all of the stratigraphic sections of post-Mazama flood plains in the study area (figures 5, 8, and 9). Downstream of the confluence of the North Fork and South Fork Sprague Rivers, the sequence of high flows in WY 2006 deposited up to 20 cm of sediment (median thickness 4 cm) presumably composed of suspended load onto flood-plain surfaces as high as 1.5 m above the low-flow water surface. This deposition might be an important factor in the overall sediment budget. Suspended sediment load estimates by Graham Mathews & Associates (2007) indicate that the mass of suspended sediment passing the measurement site near the basin outlet near Chiloquin was only 63 percent of the suspended sediment delivered by the North Fork, South Fork, and Sycan Rivers between 2 November 2005 and 6 January 2006 (dominated by the late December and early January high flows), indicating substantial net deposition along the main-stem Sprague River. The presence of Cs137 in several flood-plain samples from 10 to 20 cm depth (appendix D) also indicates significant post-1950 deposition on many flood-plain surfaces.

Flood-plain deposition requires flood-plain inundation. During the 2006 flood, much of the flood plain was inundated along the main-stem Sprague River as well as along at least the lower North Fork Sprague River (figures 9, 52, and 54). If, however, the frequency of flood-plain inundation is reduced, by either channel incision, levee construction, or reduced formation of low flood-plain surfaces (mainly from channel migration), downstream sediment loads may increase because of the combined effects of less overbank deposition and greater in-channel flow velocities accelerating bank erosion and increasing transport capacity.

Bed material transport is evident by the bars along the length of the study reach (figures 15, 18, and 39), but little is known about actual transport rates. Graham Mathews & Associates (2007) measured bedload on 14 occasions among 5 locations during WY 2004 and 2005. From six measurements of the Sycan River at Drews Road (SYFK 2.2), they estimated that bedload composed about 30 percent of the total sediment load, a similar value to the 35 percent estimated from five measurements on the South Fork Sprague River picnic area just upstream of the study area at FK 107.6. Both of these sites are steeper than most of the study area, and it is likely that bed-material transport rates are lower and a smaller fraction of the total sediment load in the lower-gradient main-stem valley segments.

The reaches of greatest bed-material transport are those where both transport capacity (chiefly a function of channel slope and flow depth) and bed-material supply are high. These conditions can be inferred in part from the spatial and temporal distribution of bars. Bars are more abundant where the South Fork and North Fork emerge onto the alluvial valleys from the uplands surrounding the study area into the South Fork and North Fork valley segments. But the greatest frequency and area of bars is near the confluence of the Sycan River in the Beatty–Sycan and Council Butte valley segments (figures 18 and. 39), pointing to the Sycan River as a key source of bed material (Hughes and others, 2009). Bars are also abundant in the Buttes of the Gods segment. The dearth of  bars in the Chiloquin Canyon and Braymill segments is the consequence of the greater slope and confinement of these reaches, which results in most bed-material sediment delivered from the low-gradient and unconfined reaches upstream being transported efficiently through these segments.

The relatively high volume of bed material associated with the Lower Sycan compared to the North Fork and, even more so, the South Fork valley segments (figures 18 and 39) owes to the much greater sand supply in the Sycan River. The much thicker accumulation of Mazama pumice in the Sycan River watershed (Sherrod and Pickthorn, 1992) in conjunction with locally eroding banks of Sycan flood deposits (figure 7) provide abundant sand-size pumiceous material to the Sycan River. The North Fork Sprague River and South Fork Sprague River drainages were blanketed with respectively thinner accumulations of Mazama tephra—particularly sand-sized accumulations. Consequently, these drainages have had less bed-material transport during the late Holocene. As a result, bar frequency is less upstream of the Sycan River confluence, and bars are very sparse upstream on the South Fork Sprague River. During late Holocene and historical periods of wetland soil formation, evident in the stratigraphy of both the North Fork (figure 9) and South Fork (figure 8) Sprague River segments, there was likely little bed-material transport in these valley segments.

The much greater bar frequency mapped from the 1968 aerial photographs in comparison with the 1940 and 2000 aerial photographs indicates the importance of large peak flows in transporting bed material. In this case, the 26 December 1964 flood is almost certainly responsible for significant bed-material transport and resulting bar formation and expansion, similar to but to a greater degree than the depositional patterns documented by the survey of effects of the 2006 high flows. The relatively greater importance of high flows in transporting bedload is consistent with the few measurements by Graham Mathews & Associates (2007) and the observations from more systematic bed-material transport measurements and modeling studies for other western Oregon rivers (Wallick and others, 2010; 2011).

Overall bed-material transport rates are likely low in the Sprague River. The absence of continuous sand and gravel bed-material cover of the channel bottom (all valley segments have significant reaches of channel bottom composed of consolidated silt and clay) indicates that the bed-material supply is exceeded by transport capacity at decadal to century time scales, despite low transport capacities in most reaches because of low channel gradients. Supply is limited because the basin geology of mainly volcanic rocks does not produce substantial bed material, similar to the Deschutes River basin to the north (O’Connor, and others, 2003b). Judging from the composition of modern and late Holocene bed-material deposits, pumiceous sand from the 7,700 cal yr BP Mazama eruption, derived either from upstream portions of the watershed or by reworking of flood-plain deposits, continues to be a major source of bed material. The low supply and transport rates in part explain the overall scarcity of bars as well as the low channel migration rates of the main-stem Sprague River.

Bar Building

In addition to providing an index to bed-material transport rates, bar deposition affects channel and flood-plain morphology by promoting channel migration and as a foundation for young and low flood-plain surfaces. Bare bars also provide germination sites for riparian vegetation, particularly rushes, sedges, and riparian grasses. As noted above, most of the bed material forming bars is pumiceous sand, mainly derived from the Sycan River basin but also to a lesser degree from the North Fork and South Fork Sprague Rivers. An unknown but probably sizeable fraction of modern bar material is derived from reworking of older (but post-Mazama) flood-plain deposits, including the Sycan flood terrace, as a consequence of lateral erosion and channel migration.

Where visible in stratigraphic section and during the 2006 survey, dipping beds of sandy material indicate that bars grow laterally. In most situations, bar growth is in the form of point bars, but also locally as mid-channel islands and accumulations in flow separation zones formed by woody debris accumulations, natural and artificial channel obstructions, and channel bifurcations. Stratigraphic sections along the Sycan River and North Fork Sprague River (figures 11 and 14) show that these point bars gradually evolve into flood-plain surfaces, with bedding indicating a transition from lateral to vertical accretion.

Bars are a factor promoting bank erosion and lateral migration. Bars deposited on the inside of channel bends and in mid-channel will force additional flow to the outsides of bends and channel margins, thereby enhancing bank erosion and lateral migration. Bars are sparse in the Sprague River study area compared to many alluvial rivers; nevertheless, their frequency is positively correlated with migration rates (figure 46A). This finding is also consistent with the evident transformation recorded by the stratigraphy of the fluvial system at the time of the Mazama eruption from one of low-energy and possibly locally unchanneled flood-plains and wetlands to a connected fluvial system consisting of channels migrating across the flood plain.

Bank Erosion

In conjunction with bar building and channel migration, bank erosion is a ubiquitous process in the Sprague River study area. Judging from the sediment transport measurements, bank and channel erosion, especially in the Upper Valley and South Fork valley segments, is a significant contributor to suspended sediment loads. Bank erosion is a complex process (Simon and others, 1999) involving fluvial processes of direct entrainment and undercutting, gravity driven mass movements and raveling, and disturbances such as trampling by animals (including livestock) and impacts from debris, waves, and vehicular traffic. The relative effectiveness of different processes is affected by the nature of the bank materials, bank and channel geometry, flow conditions and vegetation (reviewed by Florsheim and others, 2008).

Bank erosion is most evident in the study area in the form of vertical banks, commonly flanked at their base with down-dropped blocks of flood-plain sediment (figure 57). While approaches are now available for quantitatively assessing bank erosion, especially for low-energy and fine-grained fluvial systems like the Sprague River (Simon and others, 1999; Pollen-Bankhead and Simon, 2008),  our observations are chiefly qualitative, drawn from inspections of banks during the course of stratigraphic analyses and the survey of the effects of the WY 2006 high flows.

Banks in the Sprague River study area are mainly formed in young, fine-grained flood-plain deposits, particularly in the unconfined reaches. For example, the Kamkaun Spring, Buttes of the Gods, Council Butte, Beatty-Sycan, Upper Valley, South Fork, and North Fork valley segments all have more than 80 percent of their 2000 channels wholly contained within the geomorphic flood plain (figure 16), which is mostly younger than the Mazama eruption of 7,700 years ago. The case for the Lower Sycan valley segment would be similar except that the channel is within 10 m of the Sycan flood terrace for 23 percent of its length.

The different bank materials apparently correlate with different mechanisms and consequences of bank erosion. The young flood-plain deposits, with their typical stratigraphy of unconsolidated pumiceous sand overlain by more cohesive very-fine sand, silt, and clay locally strengthened by dense root networks, are subject to undercutting, especially on the outside of bends where boundary shear stress is greatest (Dietrich and others, 1979). In many cases, such undercutting during WY 2006 promoted gravitational failure of cohesive blocks of overlying flood plain, resulting in stepped bank forms. These down-dropped blocks, which appear to have moved by toppling and slip failure, provided substrate for colonization by more saturation tolerant species (figure 57A). The Sycan River flood plain, with its more prevalent pumiceous sands in conjunction with overall greater channel slopes, appears to be especially susceptible to bank erosion. Young flood-plain deposits in the South Fork valley segment do not have the thick pumiceous sands at the base of their sections, perhaps inhibiting bank erosion by undercutting and mass wasting, although the sediment transport analyses indicate that this segment is a major contributor of suspended sediment.

Bank erosion where the channel impinges against older surfaces is less common because contact between the active channel and older surfaces is rare in most reaches (figure 16). In the few locations where the present channel is eroding edges of pre-Mazama terraces and fans, erosion rates appear to be slow with less undercutting because of the more consolidated character of sand and gravel at the base of these sections compared to the pumiceous sand at the base of the late Holocene flood-plain deposits. Tall raw banks formed in older materials locally extend along the channel for distances exceeding 10s of meters. These banks appear to erode slowly, with freeze-thaw, desiccation, and biogenic disturbance possibly being the main processes loosening bank materials, thereby facilitating fluvial entrainment and helping to maintain unvegetated faces.

A distinctive situation is where the Sycan River abuts the Sycan flood terrace, which is the case for about 23 percent of its length within the Lower Sycan valley segment (figure 16). Here the bank stratigraphy is composed of coarse and granular pumiceous sand overlying compact silt and clay, with the contact as much as 2.5 m above the low-flow water surface (figures 5 and 7). Whereas the pre-Mazama fine-grained flood-plain deposits resist erosion, the loose pumiceous sands above are exceptionally erodible, especially where vegetative cover is disturbed. Erosion is enhanced by water seepage concentrated at the basal contact of the granular sand and the underlying and less permeable silt and clay. This seepage locally conveys the overlying sand to the channel edge. In places, this seepage appears to be enhanced by irrigation of the Sycan flood terrace.

Although bank erosion can increase sediment loads, cause property and infrastructure loss, and perhaps locally adversely affect aquatic and riparian habitat conditions, bank erosion is also a key process for maintaining channel migration and other key ecologic processes (Florsheim and others, 2008). Bank erosion is probably a significant source of bed material for the Sprague River and North Fork Sprague River, in particular by providing a supply of the pumiceous sand, which is an important bar-building material. Locally, bank erosion also creates diverse bank and channel-margin conditions, especially where blocks of flood-plain sediment have down-dropped to low-water levels. Together these features create variable hydraulic environments and provide many of the low, near-channel surfaces for colonization by aquatic and riparian flora (Hughes and Leeseberg, 2009).


The common cut banks in the Sprague River study area are sometimes cited as an indicator of recent channel incision (Connelly and Lyons, 2007; Rabe and Colonje, 2009). But, as described above, evidence of historical channel lowering is primarily restricted to the upstream parts of the study area, including the South Fork, and possibly parts of the North Fork, Upper Valley, and Beatty Gap segments. No local observations document the processes and specific timing, but in these locations incision has probably proceeded by upstream migration of knickpoints as is commonly the case for channels with fine-grained or cohesive beds (Simon and Rinaldi, 2006). As for many incised streams (Cooke and Reeves, 1976; Schumm, 1999), some of these knickpoints may have been initiated by diversion or diking leading to flow concentration, or by clearing or removal of wood accumulations, including beaver dams. For the South Fork Sprague River, particularly for the straightened and leveed section between FK 97.0 and 101.1, channel incision was probably at least partly caused by direct excavation of materials from the channel for levee construction. In-channel excavation, documented by historical observations in the Kamkaun Spring valley segment (Greg Harris, oral commun., 13 September 2007), may be a factor in other local areas of possible historical channel incision in the Upper Valley and Council Butte valley segments.

The irregular channel profile along the South Fork Sprague River upstream of the diversion structure at the Campbell Road Bridge at FK 101.4 (figure 32) may reflect active channel erosion. Migrating knickpoints in this upper part of the South Fork valley segment would be consistent with the elevated suspended loads derived from this reach during the 2006 high flow (figure 56). Enhanced sediment loads could result either from direct sediment introduction to the flow as the knickpoint erodes or from incision of diversion channels and bank crevasses connected to the lowering channel.

Channel incision can be mitigated locally by grade-control structures as well as natural processes such as beaver activity and in-channel wood accumulation. The existing diversions on the South Fork Sprague River at FK 100.0 and FK 101.5, as well as the shorter structure on the North Fork Sprague River at NFFK 1.1 and the in-channel rock placement at NFFK 5.2, are all currently protecting upstream reaches from incision (figure 32). Judging from overall profile trends, removal of these structures, either planned or by flood, without mitigating grade protection measures, would likely trigger incision extending perhaps several kilometers upstream.

Channel Substrate

Channel substrate varies across the study area, ranging from coarse boulder material in the steep and confined Chiloquin Canyon and Coyote Bucket valley segments to indurated silt and clay where the channel is flowing directly on older fine-grained channel and flood-plain materials and Tertiary lacustrine sediment. Additionally, most valley segments have patchy to nearly continuous sand and gravel on the channel bottom and margins. Placed rock marks some fords and former bridge crossings, as well as channel restoration and bank protection sites.

The coarse cobble and boulder substrate in the Chiloquin Canyon and Coyote Bucket valley segments as well as other locations where the channel impinges on or near to bedrock is probably rarely mobilized. Stable rock accumulations may locally promote erosion and deposition of other substrate materials by changing local hydraulic conditions. Placed rock, however, may be susceptible to displacement in locations of bank erosion or channel incision.

Indurated silt and clay substrate is common in several valley segments. During our 2006 synoptic survey, we noted extensive reaches of this substrate in the Buttes of the Gods and the Kamkaun Spring valley segments. It is also common in the lower part of the Lower Sycan valley segment (figure 34) and was the main substrate visible at road crossings of the South Fork Sprague River in the South Fork segment. Impenetrable silt and clay marked the bottom of most augering attempts in the Buttes of the Gods and Kamkaun Spring valley segments (figures 12 and 13). In the Lower Sycan, Buttes of the Gods, and Kamkaun Spring valley segments, this indurated substrate is inferred to be chiefly Tertiary lacustrine sediment on the basis of the pervasive jointing, similar to that evident on roadcut exposures in the area. In some places, however, particularly in the Upper Valley and South Fork valley segments, the indurated silt and clay substrates may be late Quaternary low-energy flood-plain and wetland deposits.

The silt and clay substrate erodes into forms commonly associated with soft bedrock channels (Richardson and Carling, 2005), including flutes, furrows, and potholes. Some furrows, most evident in recent meander cutoffs, exceed depths of 2 m. These erosional features seem to be stratigraphically and structurally controlled, with plucking and erosion taking advantage of horizontal weaknesses associated with softer strata and vertical zones of weakness caused by joints. At low flow in the Buttes of the Gods valley segment, several small and short water-surface drops result from spill over ledges formed of resistant beds. This hard substrate probably inhibits channel incision in several of the valley segments where it forms much of the channel bottom. Where incision has or is occurring, as is probably the case for the South Fork valley segment, the resistant silt and clay probably forms persistent and slow-migrating knickpoints or short reaches of high channel slope.

In most valley segments, patchy to continuous sand and gravel (and in places, soft mud) locally cover the indurated silt and clay substrates. This sediment reflects current bed-material supply and transport conditions. These materials are generally loose, and the sand sized component dominated by pumice grains. Gravel, typically with a diameter less than 2 cm, is locally common along the channel thalweg in several valley segments, and is chiefly composed of volcanic clasts. In the steeper upstream parts of the Lower Sycan, North Fork, and South Fork valley segments, gravel and cobble channel bottoms and marginal bars are more common, in places forming continuous alluvial channels extending for several kilometers. In some reaches, particularly in the Lower Sycan and North Fork valley segments, sand and gravel bed material has been organized into pool and riffle sequences (figure 33). But as described above, the patchiness of alluvial cover in the Sprague River channel network indicates that overall bed-material supply does not match the transport capacity.

The augering transects, stratigraphic studies and observations from the 2006 survey indicate that bed material locally accumulates in areas of reduced flow velocity during high flows, particularly downstream of obstructions, on the inside of channel bends and in entrances of side channels. In addition to mid-channel and point bars, such deposition locally creates sandy bars at the entrance of side channels and can promote flow diversion into nascent meander cutoffs.

Spring channels, connecting springs to the main channel through the geomorphic flood plain, invariably have well sorted pumiceous sand bottoms. For these channels, it appears that spring flow is sufficient to mobilize fine sediment, leaving well-sorted sand. In situations where river channel migration or avulsions result in formation of new spring channels, it is likely that in time these new channels will also evolve to having well-sorted sandy beds under conditions of relatively constant spring discharge.

As described above, maintaining an alluvial cover on the channel bottom in part depends on maintaining a supply of bed material. Over millennial time scales, as the supply of pumice sand diminishes from the upper parts of the basin, the importance of bank erosion and reworking of flood-plain accumulations of sand and gravel will grow as a source of sand and gravel channel substrate.

Flood-Plain Processes

The overall observations of historical and current flood-plain conditions and trends, as well as inferences regarding important channel processes, provide a framework for understanding the major flood-plain forming processes. These include the mechanisms and consequences of lateral channel movement across the flood plain—in the Sprague River study area, meander growth counteracted by meander cutoff and channel avulsion—as well as the overbank depositional and erosional processes by which flood-plain surfaces are built, eroded and modified.

Meander Growth

Meander growth is a main driver of flood-plain morphology in the Sprague River study area. Meander enlargement provides opportunities for channel cutoffs and meander abandonment that result in the overall native flood-plain morphology of abandoned channels (and associated ponded water and side-channel environments) and multi-level flood-plain surfaces.

On the Sprague River, meander growth is by the tightly coupled processes of bar formation and lateral erosion. Judging from the presence of measureable migration rates and bar presence in all valley segments, these processes are evidently important to some degree for the entire study area. It is primarily the unconfined (either by valley margins or levees) segments, however, where meander loops can enlarge to promote high sinuosity (figure 38).

Bar morphology can also indicate patterns of meander growth. In particular, the orientation of scroll bars—concentric ridges developed on a point bar surface over time as it evolves (Lobeck, 1939; Hickin 1974)—can be used to interpret the direction and type of channel movement. Hooke (1997) identified five major types of bar movement: downstream translation, upstream translation, expansion (growing toward the flood-plain margin), rotation, and development of a compound or double-headed meander. Scroll bars are more evident in the wide valley segments. The dominant types of meander movement indicated by scroll bars in Kamkaun Spring, Buttes of the Gods, Council Butte, and Sycan-Beatty segments are downstream translation (figure 12) and expansion (figure 54), although some bars display evidence of upstream translation and development of compound meanders.

The slow rates of meander growth for the Sprague River and its major tributaries are a consequence of both the low overall bed-material transport rates and the low stream power. Channel confinement and slope are local factors also apparently exerting control on migration rates at the segment scale. The confined canyon segments and the South Fork segment confined by levees all have low migration rates compared to the unconfined segments (figure 45). Channel slope is positively correlated to migration rates (figure 46B), except for the steeper but closely confined Chiloquin Canyon segment, consistent with observations that stream power (which is directly proportional to slope) is correlated to bank erosion and channel migration (Nanson and Hickin, 1986; Richard and others, 2005).

Cutoff and Avulsion Processes

Cutoffs and avulsions counteract meander growth by isolating meander loops from the main channel. Such cutoffs and channel avulsions typically reduce local sinuosity and increase local channel slope as well as leave a suite of backwater and side-channel environments. A cutoff is the breaching of a meander bend at or near its base, typically by one of two processes: (1) Neck cutoffs result from bank erosion, narrowing the neck until a breakthrough occurs between the upstream and downstream meander limbs. (2) Chute cutoffs result from new channels shortcutting between upstream and downstream meander bends, commonly without narrowing of the meander base (Thomas and Goudie, 2000) (figure 40A, B). Prolonged meander growth inevitably leads to meander cutoff, as sinuosity increases to a critical or stable level. Sinuosity may be dynamically maintained by the concomitant processes of meander growth and cutoffs (Hooke, 2003). An avulsion is when the channel suddenly moves to a new location on the flood plain without substantially eroding or reforming the intervening flood-plain surface. Avulsions may result in anabranches that bypass several meander loops, forming new primary or secondary channels (figure 40C). The new location may be a new channel incised into the flood plain or an older channel that is re-occupied. In some locations, chutes or neck cutoffs may involve secondary channels formed by avulsions (figure 40A).

Avulsions and chute cutoffs require overbank flow. Lateral migration and resulting neck cutoffs may result from flows confined to the channel (Hooke, 1997). Once a secondary channel forms, channel infilling and bar growth in the older channel may accelerate diversion of primary flow into the newly formed channel, transforming the older channel into low flood plain and backwater areas (figure 12). Although, in some cases, it appears that multiple channels may convey flow for several decades. These dynamics are indicative of the manner by which cutoffs and avulsion connect channel and floodplain morphology.

The main mechanism for meander abandonment in the Sprague River study area is chute cutoff, accounting for 42 of the 63 instances of observed meander abandonment for which process could be determined (table 8). Only 12 avulsions were evident, but their effects are more significant than individual chute cutoffs because they typically result in abandonment of multiple meander loops. Only nine examples of neck cutoff were evident from the historical aerial photograph and map analysis. The relative rate of cutoff and avulsion processes is consistent with the overall geomorphic regime of frequent overbank flooding (which facilitates chute cutoffs and avulsions) and slow migration rates (which reduces the occurrence of neck cutoffs).

A process possibly promoting chute cutoffs in the Sprague River basin is the formation of ice jams. Residents report that the Sprague River is subject to ice-flow jams that lead to backwater and overbank flow. Small incised channels on the flood-plain surface that drain from meander interiors to the main channel (figure 58) may have been formed by overbank flow returning to the main channel during ice-jam floods, and they may serve as pilot channels that initiate chute cutoffs, as has been observed for other rivers (Smith and Pearce, 2002; Prowse and Culp, 2003).

Chute cutoffs in the Sprague River basin may also be promoted by channel and flood-plain modifications imposed by levees and railroad grades. This mechanism is evident in the Council Butte valley segment, where incipient chutes appear to be forming along levees (figure 40B), where overbank flow depth (and erosive power) has been enhanced by confinement and where new channel formation may be taking advantage of areas of flood-plain excavation for levee construction. In general, flow confinement will increase the likelihood of cutoffs, as will ground disturbing activities such as vehicle tracks or animal paths that disrupt flood-plain vegetative cover and thereby facilitate new channel formation.

Some avulsions may result from similar processes as chute cutoffs but with the new channels bypassing multiple meander loops. In particular, the 12 avulsions since 1940 appear to be of this type (figure 40A). These avulsions are generally short, cutting off only two or three meander loops. Contrasting with these are the longer anabranches near the Sycan River confluence (figure 40C). Long anabranches are typically associated with aggrading rivers, particularly those with high sediment loads (Slingerland and Smith, 2004). The long anabranches in the Council Butte near the Sycan River confluence have been stable since at least 1940 and may in part be relic features related to aggradation of the lower Sycan River and parts of the Sprague River after the Mazama eruption and the Sycan flood.

The 35 cutoffs and avulsions in the 1940–1968 period exceeds the 28 instances in the ensuing 37 years up to 2005. The significant overbank flow during the 1964 flood was probably a contributing factor to this period of significant channel change. It is also possible that the extensive levee building and diking in the 1940–1968 period may have accelerated or promoted cutoff formation.

Where channels avulse or meanders are truncated, the resulting abandoned channels typically become backwater or secondary channels, oxbow lakes, and wetland areas. These channels eventually fill with sediment and organic materials, depending on the nature of the connection with the primary channel. For isolated abandoned channels, the topographic channel form and resulting oxbow lakes or seasonally wet areas may persist for millennia before filling with overbank deposits. For connected abandoned channels, infilling may be much more rapid, as is the case for the 1940 channel at FK 48.2 (figure 12).

Flood-Plain Building

Together, all of these channel movement and sediment erosion, deposition, and transport processes create the geomorphic flood plains of the study area with their diverse forms, including oxbows, scroll bars, overflow channels, and flood-plain surfaces of varied elevations. In addition to providing sources of bedload and suspended load for building downstream bars and flood-plain surfaces, channel migration provides new surfaces for flood-plain formation. In many areas of lateral migration, such surfaces typically begin as low elevation bars formed of laterally and obliquely accreted bed material (figure 15). In areas of more rapid channel migration and high bed-material supply, scroll bars may form in conjunction with lateral accretion. As the surfaces grow and become vegetated, vertical accretion becomes dominant. In most situations, rates of vertical growth gradually diminish as surfaces accumulate material and grow taller and as the active channel moves farther away. The highest flood-plain surfaces may accumulate sediment only during exceptional floods.

The pattern of flood-plain deposition can change if controlling factors vary. Most broadly, flood-plain deposition is affected by watershed and climate factors that affect the frequency, magnitude, and duration of flooding. Channel migration, incision, and aggradation can locally change the frequency of overbank flooding and deposition. Local land use can also affect flood-plain deposition by changing the depth, velocity, and sediment concentration of overbank flows. A hint of such changes is provided by the stratigraphy of the geomorphic flood-plain, which in many cases shows coarser deposits near the top (figures 8 and 9), indicating higher velocity flows associated with overbank deposition. Plausible explanations for this in light of current understanding of changing basin and flood-plain conditions include (1) confinement (and resulting acceleration) of flow by levees and embankments and (2) decreased surface roughness (by smoothing and vegetation removal).

Geologic Controls and Legacy

As with many fluvial systems, channel and flood-plain processes in the Sprague River system are strongly influenced by geologic controls and past events. At the broad scale, the valley segments and their morphologies reflect tectonics and volcanism of the last several million years. The widespread volcanic rocks and Tertiary lacustrine sediment in intervening basins result in relatively low quantities of sand-to-gravel size bed material available for fluvial transport, leading to a general condition of supply-limited fluvial systems flowing on or close to bedrock, even in the alluvial valleys of the study area. In places where channels flow directly on Tertiary lacustrine sediment, the potential for channel incision is limited. The North Fork, Upper Valley, and South Fork valley segments appear to be all within a basin of thicker alluvial fill, perhaps making these segments more susceptible to channel incision.

The valley bottoms in most segments are in part bounded by alluvial fans and terraces deposited during the Quaternary. But while these features locally confine the flood plain and form tall banks, they probably provide little sediment to the modern rivers.

The Mazama eruption of 7,700 cal yr BP has probably been the most important geologic event affecting modern flood-plain and channel conditions. Flood-plain stratigraphy and morphology, bar distribution, and the deposits of the 2006 high flows all indicate the major role of pumiceous sand derived from that eruption in controlling channel migration and flood-plain processes for the Sprague River and its major tributaries by enhancing bar deposition and channel movement. The historical condition of the South Fork Sprague River—extensive wetlands flanking a poorly defined channel—is probably similar to what much of the Sprague River was like prior to 7,700 cal yr BP. Because the Sycan River was the most affected by the Mazama eruption, including the ensuing Sycan flood, this tributary has been the most dynamic in response, with multiple episodes of flood-plain formation and erosion over the last several thousand years (Lind, 2009). Judging from the overall distribution of bars, the Sycan River, and to a lesser extent, the North Fork Sprague River, continue to be a major source of bed material to the Sprague River fluvial system. The multiple anabranches in the Council Buttes segment downstream of the Sycan River confluence (figure 40C) may also reflect bed-material deposition and resulting channel avulsions.

In the absence of additional disturbance or substantial change in basin conditions, the long term trajectory of the Sprague River and its major tributaries would likely be reversion back to pre-Mazama characteristics as the supply of sandy bed material diminishes. Thus, more active channel and bar environments may shift to low-energy wetland and low-flood-plain fluvial systems. This long-term transition may not be gradual either in time or space, but depending on climate, hydrology, and sediment supply, occur as back and forth transformations. The North Fork Sprague River has apparently already alternated between such conditions since the Mazama eruption, perhaps indicating future sequences on other parts of the fluvial system.

Human Influences

Overlaying geologic controls of flood-plain and channel processes and conditions is the locally strong influence of human manipulation. Alterations range from subtle, such as historical vegetation changes as discussed in more detail in the “Flood-Plain and Riparian Vegetation” section, to substantial, such as channelization and confinement of the South Fork Sprague River. The types of human influences have changed over time, from flow diversion, beaver eradication and log transport practices in the early 1900s, to levee construction and channelization in the mid-1900s, followed by more recent channel and flood-plain manipulation undertaken during the course of restoration projects (NewFields River Basin Services and Kondolf, 2012). The various scales, intensities, and confounding effects of all these activities challenge direct attribution of specific human activities to specific consequences for the Sprague River study area, but we draw some general conclusions from overall observed changes to flood-plain and channel conditions and known linkages among disturbance factors and channel and flood-plain processes.

Channel and Flood-Plain Manipulation

The most evident human manipulation with direct and continuing effects on channels and flood plains in the study area is channelization and flood-plain confinement. The South Fork valley segment in particular has been straightened and confined, directly causing the low sinuosity and high slope of this valley segment compared to other unconfined valley segments. Channelization occurred largely prior to 1940 and perhaps mostly subsequent to 1925 judging from the 18 May 1925 notes by USGS streamgager K.N. Phillips of a “very crooked” channel (USGS Oregon Water Science Center, Portland, Oregon, measurement station records for station 11495500). Confinement, channelization, and the consequent loss of sinuosity and increase in slope has almost certainly led to the transformation from a wet meadow fluvial system at the time of the GLO surveys to the locally straight and incised present channel. Incision likely commenced at the time of channelization but has evidently continued in the lower part of the valley segment through at least 1968. This incision likely contributed to local channel widening in this segment since the GLO surveys (figure 41). The locus of present incision is probably upstream of Campbell Road Bridge (FK 101.5), judging from the elevated suspended sediment concentrations derived from that reach (figure 55) as well as the irregular channel profile (figure 32). Incision for parts of the South Fork valley segment has apparently been locally mitigated by diversion structures at FK 100.0 and 101.5 (figure 32).

Although the South Fork Sprague River is the valley segment with most substantial channel realignment and confinement and most clear consequences, other unconfined valley segments have also likely been affected in less evident and directly attributable manners by confinement. In particular, the overall loss of sinuosity since 1940 for most unconfined valley segments, as well as the decrease in channel migration rates, are both consistent with the effects of confinement by roads, railroad grades and levees. These effects are most evident for the Kamkaun Spring, Buttes of the Gods, and Council Buttes valley segments. Additionally, the flood-plain stratigraphy along the main-stem, South Fork, and North Fork Sprague River shows coarsening toward the top, including sequences of post-1950 deposition as indicated by the presence of Cesium 137 (figures 8 and 9). This coarsening possibly reflects higher velocity flood flows resulting from flood confinement. Similarly, the locally high incidence of avulsions and chute cutoffs (figure 59) may reflect confinement resulting in deeper, faster, and more erosive overbank flow.

Roads, ditches, and railroad embankments likely have substantially reduced lateral connectivity (Poff and Ward, 1990) within the Sprague River flood plain, although this is not quantifiable on the basis of our analyses. Two factors contribute to this assertion: (1) In the incised reaches, including the South Fork and possibly parts of the Upper Valley, Beatty Gap, and Council Butte segments, flood-plain inundation has probably decreased, although without detailed understanding of historical topographic conditions, documenting the magnitude of change is not possible. (2) The areal extent of seasonal flood inundation is reduced by levees and dikes, preventing sediment, water, and organic material transfers between flood plain and channel during floods. The thin but extensive overbank deposition during the 2006 high flows (figure 52) as well as the young overbank deposits in the stratigraphic exposures of flood-plain deposits (figure 8) confirms the importance of overbank flooding in flood-plain formation.

Longitudinal connectivity has been affected by dams and diversions since the first diversion dams in the South Fork valley segment in the 1890s. The most significant such blockage was Chiloquin Dam at FK 1.3. Other small dams were constructed throughout the study area during the early and mid-1900s for log storage and transport and water diversions. With the advent of pumping and the demise of the local timber industry, nearly all of these structures have been removed. At present, the only diversions affecting the channel profile are the 1-to-2-m-high structures diverting flow along the North Fork and South Fork Sprague Rivers (figure 32). Although these structures certainly affect low-flow conditions by diverting water from the channel into irrigation networks, the ecologic and geomorphic effects of these structures are uncertain except for their likely control on upstream migration of incision-related knickpoints.

More recent channel manipulation has been motivated by efforts to restore channel and flood-plain conditions (NewFields River Basin Services and Kondolf, 2012). A primary objective has been to restore sinuosity by plugging meander cutoffs and avulsion channels and relocating the primary channel into historically occupied locations (figure 23). Such efforts have focused on the Kamkaun Spring, Council Butte, and Beatty-Sycan valley segments, where meander cutoffs have been frequent over the last several decades (figures 47 and 59). Where implemented and stable, these channel manipulations do increase sinuosity and decrease local channel slope, but they typically include some form of bank hardening (commonly rock or anchored wood) which suppresses channel migration processes. Additionally, channel remeandering in itself typically does not address some of the local factors that may have contributed to increased rates of meander cutoffs, such as flood-plain confinement by levees and dikes.

Grazing Management

If too intense, livestock or wild ungulate grazing in riparian areas can lead to diminished streamside vegetation and bank trampling, both of which increase susceptibility to fluvial erosion. Overgrazing has been associated with channel widening, incision or aggradation depending on the particular stream characteristics, and loss or diminution of other processes related to aquatic habitat (Armour and others, 1991; Trimble and Mendel 1995; National Research Council 2002). Grazing effects were noted at some locations along the Sprague and Sycan Rivers in the form of trampled banks and heavily browsed shrubs or grasses, although affected areas are limited in extent at present. Fencing and other methods to reduce livestock impact on the riparian zone have been adopted for some properties along the Sprague River. It is likely that grazing impacts on the channel banks and riparian zone were more extensive in the past, although no direct evidence is available for documenting changes in grazing intensity. Vegetation changes documented in the following section on “Flood-Plain and Riparian Vegetation,” particularly a decrease in short woody shrubs, are consistent with more intense grazing in the past.

Other Historical Activities

Besides the ongoing influences of channelization, diversions, levees, flood-plain confinement, and grazing practices, other historical activities may also contribute to present channel and flood-plain conditions in the study area. In particular, past timber practices and beaver eradication may have continuing effects on channel conditions, although the several intervening decades since these activities obscures evidence of their effects.

Timber harvest, particularly of large ponderosa pine in headwaters as well as from surfaces flanking the flood plain, especially in the confined valley segments where ponderosa pine colonizes flood-plain and flanking surfaces, has likely reduced the volume of transported large wood in the study area. While past and present wood flux is uncertain in the Sprague River system, it was (and is) likely small compared to forested flood plains. Nevertheless, in-channel wood provides a variety of channel structures and habitats (Gregory and others, 2003; Montgomery and others, 2003), and in supply-limited systems such as the Sprague River can promote deposition of bed material (Massong and Montgomery, 2000). Decreased transport of large wood has also possibly affected flood-plain environments. Wood deposition on flood-plain surfaces can provide ecological functions—well documented for forested alluvial valleys (O’Connor and others, 2003c; Collins and others, 2012), but less understood in sparsely timbered environments such as the wide valley segments of the study area. Contrasting with a likely overall long-term diminishment of large wood transport in the Sprague River fluvial system is the period of substantial fluvial wood transport during the first few decades of the 1900s, when the Sprague River was used for the storage and transport of saw logs (figure 20). These operations resulted in substantial flow fluctuations (USGS Oregon Water Science Center, Portland, Oregon, measurement station records for station 11501000). Log transport during this time likely scoured banks of sediment and vegetation, possibly widening channels, but no quantitative measurements are available to document specific and persistent effects. Log drives and in-river storage probably diminished substantially after the 1928 completion of the OC&E railroad to Bly.

An even earlier activity, beaver eradication, may also have affected channel and flood-plain conditions in the study area. Several beaver lodges currently flank the channel in the study area, locally impounding flow in side channels and irrigation canals. But beaver were evidently more common in the early 1900s in the Sprague River basin, judging from A.W. Moore’s 1943 survey of 32 abandoned colony sites in the Chiloquin Canyon, Braymill, and Kamkaun Spring valley segments (1943 Wildlife resource report by A.W. Moore, U.S. Fish and Wildlife Service, in Klamath Indian Agency records, National Archives, provided by Larry Dunsmoor, The Klamath Tribes, 2006). Moore further concluded that “while the beaver population has been severely depleted, it is believed that where trapping of the animal stopped, a satisfactory population would build up in from seven to ten years.” Bailey (1936, p. 223) noted that “in 1914 L.J. Goldman reported them [beaver] on the west slope of the Yamsay Mountains, on the Sprague and Yamsay Rivers [upper Williamson], and in Klamath Marsh,” and that “in July 1927 there were still a few beavers in Sprague River and its branches north and east of Bly...” As was the case throughout western North America (Jenkins and Busher, 1979), beaver populations were largely eradicated by fur traders throughout Oregon by the mid-1850s (Bailey, 1936), although the timing of likely population reductions in the Sprague River area is not known.

Also uncertain is the exact role of beaver on channel and flood-plain morphology in the Sprague River basin. In many low-gradient montane basins of western North America, beaver can substantially affect channel morphology, in places creating extensive areas of wet meadows and promoting valley-bottom aggradation (Ives, 1942; Naiman and others, 1988; Kramer and others, 2012). In the Sprague River basin, it is plausible that beaver dams played a role in the formation and maintenance of the historical wet meadow complexes historically occupying the South Fork valley segment as well as earlier episodes of wet meadows evident in the bank stratigraphy flanking the North Fork Sprague and Sycan Rivers (figures 5 and 9). Beaver-chewed stick fragments were commonly found within these exposures. Also, notes by USGS hydrologist K.N. Phillips, describing conditions in 1926 at the North Fork Sprague River measurement site then located at Bailey Flat, a 2-km-long alluvial reach 3 km upstream of the upstream extent of the North Fork valley segment, described the “channel in general very crooked, winding through beaver dam land...” (USGS Oregon Water Science Center, Portland, Oregon, measurement station records for station 11496500). Given these observations, it is possible that local reductions in beaver population contributed to loss of wet-meadow environments in the South Fork Sprague and North Fork Sprague valley segments and possibly to channel incision in the Upper Valley and Beatty Gap valley segments. Available evidence, however, does not show whether loss of wet-meadow area and incision was due to removal of beaver or by human activities such as channel straightening and confinement.

Timber harvest and land-use can indirectly affect flood-plain morphology by altering hydrologic regimes, which in turn can affect channel and flood-plain processes. Timber harvest in the basin, starting in a significant manner in the late 1920s and peaked in the 1950s, has been shown for some locations in the Oregon Cascade Range to increase peak flows (Jones and Grant, 1996, Beschta and others, 2000; Grant and others, 2008). Such changes to peak flows could have affected Sprague River stream channels, but the magnitudes are not known and the effects are likely to have been small compared to direct flood-plain and channel modifications. The streamflow measurement record does not show clear evidence of greater peaks during the time period of intense timber harvest (figure 2). Low flows and total runoff have more likely been substantially affected for the Sprague River study area, particularly by implementation of irrigation systems, as indicated by hydrologic modeling (Natural Resources Conservation Service, 2009). Some of the historical precipitation-runoff changes noted by Risley and Laenan (1999) may be a consequence of timber harvest and changing irrigation practices, but while these changes may affect riparian vegetation, they have probably had little effect on physical channel and flood-plain processes.

Summary of Physical Flood-Plain and Channel Conditions, Processes and Trends

Strong geologic controls have resulted in distinct valley segments within the alluvial portion of the Sprague River watershed, ranging from low-gradient reaches with expansive flood plains to steep and narrow canyon segments. The wide flood-plain valley segments are broadly similar; most contain a sinuous, low-gradient channel that migrates slowly across the valley bottom. The narrow valley segments include the steep, boulder-and-cobble-bed Chiloquin Canyon, at the downstream end of the study area, and the Coyote Bucket valley segment as the Sycan River exits the Sycan Canyon. The other confined valley segments—Braymill, S’Ocholis Canyon, and Beatty Gap—have similar gradients and substrates as adjacent unconfined valley segments, but much lower sinuosities. Although the geologic setting of the expansive South Fork valley segment resulted in historical conditions of sinuous and poorly defined channels and wet meadows, flanking levees now narrowly confine the channelized South Fork Sprague River for much of its length.

The fine-grained extrusive volcanic rocks and lacustrine sediment that compose most of the Sprague River basin produce little sand and gravel bed material. In the absence of substantial bed material prior to the Mazama eruption of 7,700 cal yr BP, the main rivers of the study area were apparently flanked by wetlands and low flood plains, and migration rates were probably very low. The eruption, however, covered much of the northern watershed with sand- and granule-size pumice clasts, transforming the fluvial system by increasing bed-material transport and promoting bar formation and channel migration, particularly for the Sycan River, North Fork Sprague River, and the Sprague River downstream of the Sycan River confluence. The South Fork Sprague River, which had much less Mazama pumice deposited in its catchment, remained a low flood plain and wet meadow fluvial system until historical channelization and diking.

Despite the pumiceous sand input and the general low gradients of the main channels within the study area, the present overall bed-material sediment regime is supply limited, meaning the transport capacity exceeds the supply of bed material. This is locally manifest by the discontinuous alluvial cover of the channel bottom, the sparse bars in and flanking the channel and the overall slow channel migration rates compared to many western North American rivers.

Lateral channel movement is an important process for maintaining channel and flood-plain materials, form, and function, even at slow rates. Movement is by continuous lateral migration associated with bar formation, lateral accretion, and bank erosion, as well as by translocations by avulsion and meander cutoffs. Lateral migration forms and expands meanders, increasing sinuosity. In unconfined valley segments, sinuosity values typically exceed 1.6 (figure 60). The positive correlation between migration rates and bar area attests to the strong coupling between bed-material deposition and lateral channel movement.

Meander enlargement is countered by avulsions and meander cutoffs, including chute and neck cutoffs, which reduce sinuosity. Avulsions and chute cutoffs are much more common in the study area than neck cutoffs. Avulsions and chute cutoffs require overbank flow, whereas neck cutoffs do not. In some instances overbank flow and resulting cutoffs may be promoted by ice jams. The decrease in sinuosity since 1940 for nearly all the unconfined reaches is in part due to decreased migration rates but owes chiefly to cutoffs and avulsions, mainly between 1940 and 1975 (figure 60). The frequency of avulsions and meander cutoffs may have been enhanced historically by (1) flood-plain confinement by levees, dikes, roads, and railroads leading to deeper and faster overbank flow thereby promoting erosion of new flood-plain channels and (2) flood-plain disturbances such as trails, ditches, and vegetation manipulation or eradication, thereby locally concentrating overbank flow and more generally decreasing surface resistance to channel erosion.

The Sprague River generally is not incised to the extent that floods fail to inundate the flood plain. The 2006 flood left overbank deposits over flood-plain surfaces along the entirety of the main-stem Sprague, lower Sycan, and lower North Fork Sprague Rivers, consistent with the stratigraphic evidence of substantial flood-plain deposition and construction over the last several hundred years. The stratigraphy of exposed bank sections and along augering transects also indicates little or no incision over the last 4,000 years within most of the study area, including the North Fork and Lower Sycan valley segments. Up to 1.6 m of historical incision is evident in the upstream part of the study area, chiefly in the South Fork valley segment but possibly also including parts of the Upper Valley and Beatty Gap valley segments. For the South Fork valley segment, incision (and ditching and levee construction) has transformed areas of wet meadow at the time of the GLO surveys to flood plain that is only rarely inundated. The locus of ongoing incision in the South Fork valley segment appears to be primarily upstream of FK 101.4.

The general absence of channel incision for much of the study area is consistent with little evidence of channel widening (figure 60). Neither analyses using the GLO surveys (figure 41) nor those of historical aerial photographs (figure 43) substantiate overall widening or narrowing trends in channel width. Since the GLO surveys, wide channels have narrowed and narrow channels have widened. For measurements going back to the earliest aerial photographs of 1940, the channel was generally widest in the 1968 aerial photographs, probably reflecting channel widening during the high flow of December 1964. Between 1968 and 2000, most segments narrowed to widths very similar to those measured from 1940 photographs. The Council Butte valley segment is unique in having a systematic narrowing trend for the entire time period. The variance in channel width has decreased since 1940 for several valley segments, particularly upstream of S’Ocholis Canyon (figure 43).

All measurements, stratigraphic analyses, and other observations point to the important role of flooding in creating and maintaining channel and flood-plain conditions. Floods promote channel movement, including lateral migration and meander cutoffs, which together create a dynamic mosaic of channel and flood-plain landforms. Floods deposit water, sediment, and organic materials in the channel and overbank areas. Large floods, as in 1964, and moderate floods, as in 2006, can have persistent effects that include bank erosion, channel widening, bar building, and overbank flood-plain deposition as well as creating new channels from avulsions and cutoffs.

The Sycan River is a unique system because of the profound influence of the 7,700 cal yr BP Mazama eruption on its catchment. The Sycan flood, shortly after the eruption, formed a surface underlain by loose and coarse pumiceous sand flanking the channel and younger flood plains. Where disturbed, this surface contributes substantial sand to the Sycan River. Subsequent to the flood, episodes of watershed-scale enhanced sediment supply resulted in construction of sandy flood plains flanking the channel, which continue to be susceptible to lateral erosion. This situation results in the Lower Sycan valley segment having abundant bars, wide channels, and high migration rates relative to other valley segments. Additionally, these local sediment sources, as well as pumiceous sand still being transported from upstream parts of the watershed, result in the Sycan River being a major source of bed material for the Sprague River fluvial system, reflected in the high bar frequency and bar-area measurements in the Council Butte valley segment near the Sycan River confluence (figure 60).

The South Fork Sprague River was little affected by the Mazama eruption; consequently, the historical condition of its fluvial systems was probably markedly different from that of the Sycan River and, to a lesser extent, the North Fork Sprague River. At the time of the GLO surveys and for thousands of years prior, the South Fork Sprague River was probably a poorly defined and stable (and locally multi-thread) channel and wet-meadow complex fringed by willows. The expansive valley bottom was likely seasonally flooded and slowly aggrading with peats and organic-rich marsh and low-energy fluvial deposits. Beavers likely had a strong role in controlling local flood-plain and channel conditions.

The South Fork valley segment has been the most clearly transformed since first historical observations. The present channel is incised, straightened and separated from the flood plain by levees for much of the valley segment. Significant portions of the flood plain have been drained and leveled. Multiple in-channel structures currently divert flow into irrigation ditch networks. This valley segment also appears to be a major source of suspended sediment to the Sprague River fluvial system.

Since first European settlement in the 1860s, human actions have locally affected Sprague River channel and flood-plain conditions. Early activities such as beaver trapping and timber harvest (and associated activities) almost certainly had consequences for channels and flood plains, although their persistent effects are unknown. Similarly, the current consequences of late 19th and early 20th century diversions, dams, timber harvest and saw-log transport are unclear. More evident are the contemporaneous and persistent effects of channelization and flood-plain confinement by levees, roads and the former OC&E railroad alignment, mainly dating from the 1920s through the 1970s. Channelization before 1940 in particular has directly transformed the South Fork Sprague River between FK 97.0 and 101.0. Less evident are the effects of levees, but flood-plain confinement has probably increased the frequency of avulsions and cutoffs, resulting in reduced sinuosity and overall greater channel slopes since 1940. Riparian grazing and channel and bank trampling by cattle have locally affected bank and channel conditions, as has local accumulations of dumped rock at fords and former bridge crossings. Channel remeandering, chiefly since 2000, has locally resulted in filling of chute cutoffs and avulsions and placing the primary channel back into historical positions. These restoration efforts, however, are commonly engineered with anchored materials to reduce future channel migration.

Despite these broad-scale and local effects of human disturbances, many of the fundamental processes that have shaped the Sprague River fluvial systems over the last several thousand years are still operating to various degrees. Most importantly, the overall bed-material transport regime, including sediment supply and water flow, which drives processes of bar building and channel migration, has not been substantially affected by upstream impoundments interrupting either flow or sediment. Also important is the absence of systemwide channel incision, which fundamentally alters channel and flood-plain processes, particularly lateral exchanges of water, sediment and organic materials, and can significantly limit restoration opportunities (Simon 1989; Simon and Rinaldi 2006; Cluer and Thorne, 2013). Areas of the fluvial system that have been incised, particularly the South Fork valley segment, are in the upstream part of the study area, limiting the likelihood that geomorphic effects (aside from changes in flow and sediment transport) will migrate to other parts of the study area, since channel incision typically migrates upstream. These factors suggest that restoration of many of the historical physical conditions and processes is possible for much of the Sprague River study area without substantial physical manipulation of current conditions.


For more information, contact:

Director, Oregon Water Science Center
U.S Geological Survey
2130 SW 5th Ave
Portland, OR 97201
Phone: (503) 251-3200

Part or all of this report is presented in Portable Document Format (PDF); the latest version of Adobe Reader or similar software is required to view it. Download the latest version of Adobe Reader, free of charge.

Accessibility FOIA Privacy Policies and Notices logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Oregon Webteam
Page Last Modified: Tuesday, 11-Jun-2013 19:49:51 EDT