Riparian forest vegetation is widely believed to protect riverbanks from erosion, but few studies have quantified the effect of riparian vegetation removal on rates of river channel migration. Measured historical changes in a river channel centreline, combined with mapped changes in floodplain vegetation, provide an opportunity to test how riparian vegetation cover affects the erodibility of riverbanks. We analysed meander migration patterns from 1896 to 1997 for the central reach of the Sacramento River between Red Bluff and Colusa, using channel planform and vegetation cover data compiled from maps and aerial photography. We used a numerical model of meander migration to back-calculate local values for bank erodibility (i.e. the susceptibility of bank materials to erosion via lateral channel migration, normalized for variations in near-bank flow velocities due to channel curvature). A comparison of migration rates for approximately 50 years before and after the construction of Shasta dam suggests that bank migration rates and erodibility increased roughly 50%, despite significant flow regulation, as riparian floodplains were progressively converted to agriculture. A comparison of migration rates and bank erodibilities between 1949 and 1997, for reaches bordered by riparian forest versus agriculture, shows that agricultural floodplains are 80 to 150% more erodible than riparian forest floodplains. An improved understanding of the effect of floodplain vegetation on river channel migration will aid efforts to predict future patterns of meander migration for different river management and restoration scenarios.

No abstract

The objectives of this study were to develop and use a linked groundwater and vegetation model to simulate groundwater and vegetation distributions in a riverine and reservoir‐fringe system under different reservoir operations scenarios. This study was conducted where Little Stony Creek flows into East Park Reservoir on the east front of the Coast Range, northern California. A numerical groundwater model was used to model mean depth to groundwater during the growing season for water years 1980–1999 for each of five community types identified on the study site. Multiple vegetation models were developed, each of which described the probability that a given community type would occur primarily as a function of modeled mean depth to groundwater during the growing season and secondarily as a function of flooding. Four scenarios representing four different reservoir operations were simulated: existing condition, existing condition with late drawdown, full drawdown, and full pool. A groundwater backwater effect caused by the imposed reservoir stage extends to portions of the terrace, but the most pronounced effects occur on the delta. Consequently, the most pronounced changes in vegetation distributions also occur on the delta. Compared to the existing‐condition scenario, modeled vegetation distributions do not change under the existing condition with late‐drawdown scenario, a xeric herbaceous community type is greatly expanded under the full‐drawdown scenario, and mesic herbaceous, scrub‐shrub, and forested community types are greatly expanded under the full‐pool scenario. The results of this study are twofold. First, the linked groundwater and vegetation model is relatively simple to construct and can be used to efficiently simulate multiple surface‐water and groundwater management scenarios. Second, changes in reservoir operations can have pronounced effects on shallow groundwater and associated vegetation distributions in riverine and reservoir‐fringe systems. Thus, the effects of changing reservoir operations must be considered if the management of shallow groundwater and associated plant and wildlife habitat resources is to be successful.

Bank erosion along a river channel determines the pattern of channel migration. Lateral channel migration in large alluvial rivers creates new floodplain land that is essential for riparian vegetation to get established. Migration also erodes existing riparian, agricultural, and urban lands, sometimes damaging human infrastructure (e.g., scouring bridge foundations and endangering pumping facilities) in the process. Understanding what controls the rate of bank erosion and associated point bar deposition is necessary to manage large alluvial rivers effectively. In this study, bank erosion was proportionally related to the magnitude of stream power. Linear regressions were used to correlate the cumulative stream power, above a lower flow threshold, with rates of bank erosion at 13 sites on the middle Sacramento River in California. Two forms of data were used: aerial photography and field data. Each analysis showed that bank erosion and cumulative effective stream power were significantly correlated and that a lower flow threshold improves the statistical relationship in this system. These correlations demonstrate that land managers and others can relate rates of bank erosion to the daily flow rates of a river. Such relationships can provide information concerning ecological restoration of floodplains related to channel migration rates as well as planning that requires knowledge of the relationship between flow rates and bank erosion rates. (KEY TERMS: bank erosion rates; fluvial processes; meander migration; rivers/streams; stream power; surface water.)

We measured patterns of river channel migration and cutoff between 1904 and 1997 on a 160 km meandering alluvial reach of the Sacramento River by intersecting a sequential set of river channel centrelines mapped from a field survey and aerial photography. We identified approximate dates and locations of cutoffs and quantified cutoff dimensions. Twenty-seven chute and 11 partial cutoffs occurred over this 93-year time interval, with an average of one cutoff approximately every 2.5 years or 0.0029 cutoffs per kilometre per year. The average rate of lateral channel change was over the study period was 5.5 AE 0.6 m year À1 (approximately 0.02 channel widths per year) due to progressive migration and cutoff combined. An average of 5% of the total channel length moved laterally via chute cutoff at a rate of 22.1 AE 3.3 m year À1 versus 94% of channel length that moved via progressive migration at a rate of 4.7 AE 0.5 m year À1. The remaining 1% of channel length migrated via partial cutoff at a rate of 13.0 AE 2.8 m year À1. Although channel cutoff was less predominant mode of channel change than progressive migration in terms of channel length, an average of 20% of the total floodplain area change between successive centrelines was attributable to cutoffs. Peak cutoff frequency was concentrated temporally between 1964 and 1987 and was also spatially clustered in specific active sub-reaches along the valley axis over the entire study period.We hypothesize that the probability of channel cutoff is a function of both channel geometry and discharge. Bends that experienced chute cutoff displayed an average sinuosity of 1.97 AE 0.1, an average radius of curvature of 2.1 AE 0.2 channel widths and an average entrance angle of 111 AE 78, as opposed to average values for bends migrating progressively of 1.31 AE 0.01, 2.8 AE 0.1 and 66 AE 18, respectively. The sinuosity of Sacramento River bends experiencing chute cutoff appears to have been consistently declining from 2.25 AE 0.35 channel widths in 1904 to 1.54 AE 0.23 channel widths in 1987. We hypothesize that this trend may be due in part to the influence of land-use changes, such as the conversion of riparian forest to agriculture, on the 'erodibility' of bank and floodplain materials. For the post-dam flow regime (1937 on), cutoff frequency was significantly correlated with an estimate of cumulative overbank flow.