Analysis of Real-Time Streambed Scour Data from Bridges in Alaska

Four years of stage and bed-elevation data have been collected at 17 bridge sites throughout Alaska to assess streambed-scour in near real-time and to identify shortcomings of existing methods used to estimate scour at bridges. These sites range from small and steep creeks to large glacial rivers. Bridges have been instrumented with sonars at the pier nose for measuring distance to the streambed and transducers on the pier or bridge deck to measure stage. Data were transmitted via satellite every 6 hours and uploaded to a website with graphical presentation of bed and water-surface elevations at the pier. Data collected at these sites illustrates short-term scour and fill as well as longer duration channel degradation and aggradation cycles. Several sites showed little change in bed elevation, despite being subjected to 100-year recurrence interval flooding. Lack of streambed scour at these, sites is valuable information because they were classified as scour-critical bridges using data generated by hydraulic models and predictive scour equations. Observed changes in bed elevation were attributed to (1) short-term scour and fill during high flows, (2) seasonal aggradation and degradation, or (3) channel migration. Data from the glacial-fed Knik River near Palmer, Alaska, show an annual cycle of channel aggradation and degradation punctuated by shorter periods of scour and fill. The annual vertical bed-elevation change exceeds 6 meters and is an interplay of sediment supply, discharge, and instream hydraulic structures. The short scour and fill cycles exposed the pier footing and subfooting three times from 2002–2006. Infilling after scour was rapid in all instances and would have been missed by traditional bridge surveys unless the surveys were made during the high flow. Data from a bridge on the Copper River showed 6 meters of long-term scour and fill that resulted from a change in local flow hydraulics caused by a lateral migration of the channel. These data sets include several dynamic rivers and indicate that scour can result from a reactions to changes in hydraulic variables, timing and duration of streamflow, and the source of the high flow. These factors are not typically included in the engineering assessment of streambed scour. Measured scour was less than the estimated scour at four of five sites that experienced large floods. The disparity between estimated and observed scour was attributed to changes in flow distribution within the channel since initial scour analysis and bed armoring, a factor not considered in the initial scour analyses.