Abstract

Traditional techniques for river cross-section surveys are costly, time-consuming, and difficult to implement. They are also limited by logistical constraints. The practical limitation in the spacing and frequency of survey points restricts ground-based surveys to those of reach scale. This research has demonstrated the potential of generating river topography and sizing bed materials within complex shallow channels, using high-resolution multispectral and stereo images. The demonstration uses a 13-km long reach of meandering, alluvial river (the Goulais River in Ontario). Fluvial remote sensing provides a complimentary alternative to field surveys, in which a detailed synoptic view of the river may be acquired. The presented technique generates a river topography model (RTM) by combining the bathymetry map (the channel-bed to the free surface) and digital elevation map (the free surface and above). The former is generated from the depth-to-brightness ratio that is empirically estimated by correlating available field survey points to the digital numbers of the image. The latter is generated through a photogrammetric analysis of stereo images. A challenge arises in combining the maps when the images used to derive the bathymetry map and the digital elevation map are captured at different times (or at different stages). The difficulty is overcome by applying gradually-varied flow type technique (one-dimensional conservation of mass, energy and momentum) to resolve the discrepancy in the stages per cross section. The resulting RTM is a continuous digital terrain that encompasses the channel-bed, floodplains, and the dry terrain features. Qualitative observations of the RTM indicate a correct placement of geomorphic features, including pool-riffle zones, and point bars. The RTM is used as a model domain for simulations of depth-averaged two-dimensional hydrodynamics in order to estimate the boundary shear stress corresponding to the formative discharge. The median sediment diameter in the riffle zones derived from the simulated boundary shear stress compares well with field observations. The method presented offers a promising complimentary tool for river dynamics analysis and river management.