John Beck
Streambank erosion hazard mapping has received much less attention than flood inundation mapping in the past due to the complexity of the task as well as bank protection works that have reduced bank erosion and unfortunately, the ecological functions of our watercourses at the same time. Damages due to streambank erosion in some flooding contexts are greater than the flood water damages (Loat and Petrasheck, 1997). For these reasons, streambank erosion hazard mapping should be an integral part of flood hazard mapping and methods must be developed to accomplish it. This research proposes a methodology for mapping streambank erosion hazards based on the directives of the Swiss Federal Office for Water and Geology (now within the Swiss Federal Office for the Environment). It permits the calculation of bank failure widths and their probability as opposed to future channel migration paths. This research also investigates the input data necessary for streambank erosion hazard modeling. Geomorphological mapping must be the first step to streambank erosion hazard mapping as it permits the identification of the sediment movements in the catchment. After this step, modeling of streambank erosion can be undertaken. Geofluvial models that combine hydraulic sediment transport and geotechnical modeling are well suited for streambank erosion modeling. The model CCHE1D is such a model and was adapted for the calculation of the streambank erosion hazard on an 8 kilometer reach of the Lower Venoge River, Switzerland. CCHE1D performs one dimensional hydraulic calculations. A shear stress correction function based on channel curvature distributes mean boundary shear stress appropriately to outer and inner bank toes and the phase lag of maximum toe shear stress compared to the apex of the bend curvature is ensured by a convolution of upstream shear stresses. Tension cracking was added to the slab failure algorithm due to its significant effect on bank failure widths. After a bank failure, the cross section shape does not change which allows the flow conditions to remain the same and in turn allows the probability of failure for the modeled bank profile to be evaluated. To gain a better understanding of streambank erosion on the Lower Venoge River, detailed erosion and flow depth monitoring were done on two 1 kilometer river reaches from November 2003 through September 2005. These measurements showed the mass failures to be mainly soil falls and cantilever failures. Measured bank erosion was linearly related to the product of maximum discharge and flood volume. Bank and bed sediment data were also collected for the study reach. Eighty-two cross sections were surveyed in 2004 in the 8 kilometer study reach. A new cross section tool was developed to properly reproduce scour holes in bends. It calculates transverse position and distance, graphs the cross section to allow identification of bank definition points, linearly interpolates, calibrates bed topography parameters based on surveyed cross sections, and interpolates with respect to channel curvature. Hydrological modeling allowed for the generation of input hydrographs for the period January 1979 - February 2005. This information was combined with historically based low probability floods to construct three 300 year discharge series. Flow and erosion measurements allowed for the calibration of roughness and critical shear stress parameters, respectively, in the CCHE1D model. Where detailed erosion measurements were not available, past channel migration served as a guide for estimating the critical shear stress. Calibrated critical shear stress was poorly correlated with measured bank properties indicating the necessity of measuring critical shear stress. A regression equation involving the percentage of fine sand – large silt and the fraction of non-vegetated bank explained 75% of the variation. The three 300 year discharge series were simulated with CCHE1D. Bank failures series were output for each of the banks of the 1149 computational nodes in the study reach. Empirical frequency was used to determine the bank failure width of a given probability. These bank failure widths and their probabilities were used to calculate a streambank erosion danger. To further qualify this danger, it was mapped with a bar proportional to the mean annual simulated erosion rate. The extreme failure width for the entire reach was determined by multiplying the maximum simulated failure width by a safety factor. The erosion hazard in straight reaches is too high showing that the shear stress reduction due to the highly vegetated banks needs to be taken into account better in the shear stress correction function. This research has demonstrated the feasibility of streambank erosion hazard mapping, although the quantity of input data necessary is prohibitive. Data acquisition methods must be researched and improved to reduce costs, and research must be continued to improve understanding of bank failure processes to be included in geofluvial models.
EPFL2006
