Estimating bankfull channel geometry, dimensions and associated hydraulic attributes using high resolution dem generated from lidar data

Abstract

Knowledge on river channel dimensions at bankfull flow is essential for flood forecasting, stream rehabilitation and bank stabilization works, environmental flow modelling, and streambank erosion modelling. The difficulty of collecting spatially distributed, high-resolution data on channel form and behaviour made it necessary for broad scale erosion models to adopt generalised approaches to bankfull flow estimation. These techniques are commonly based on simple hydraulic formulae applied to cross-sectional averages. It is recognized that these generalised estimations frequently fail to describe the non-uniform flow and transport conditions observed in natural rivers. Application of Dynamic SedNet in the catchment water quality modelling project of Paddock to Reef Integrated Monitoring, Modelling and Reporting Program (P2R) under Reef Plan uses relationships between channel width and height with contributing catchment area that are regionally generated to determine these river channel dimensions. There are obvious shortcomings of this approach including: (1) the fact that channel dimensions are greatly spatially variable even when values of the explanatory variable are the same (e.g., bankfull flow occurs more often on coastal plains), (2) limited data points used to generate the empirical equation. However, with recent advances in the area of generating high resolution digital elevation models (DEMs) from Light Detection And Ranging (LiDAR) data and availability of tools to process them, it may be possible to determine channel geometry and dimensions in a more robust and reliable way. Conventional topographic LiDAR, such as that used to generate the DEM used in this study, does not penetrate water bodies. In these situations, water penetrating LiDAR (Bathymetric LiDAR) will need to be employed. However, in low flow situations, as in the current study, and/or where flow depth can be determined from monitoring, it is possible to subtract flow depth from the water surface elevation to estimate channel-bed elevation and height. This study demonstrates how a high resolution DEM generated from LiDAR data in conjunction with estimates of flow depth, which was low at the time of LiDAR data acquisition, can be used to generate bankfull river channel width and height thereby allowing estimation of other river flow parameters. Figure 1 shows the workflow (steps followed) in order to achieve this. Bankfull flow parameter values estimated from the approach currently employed in P2R and those estimated in this study (e.g. flow cross-sectional area, wetted perimeter, and hydraulic radius) have been compared at two modelled stream reaches where LiDAR DEM is available. The comparison shows that the P2R approach overestimates all bankfull flow attributes. However, since Dynamic SedNet uses a user-specified recurrence interval to determine bankfull discharge and applies a calibration coefficient for adjusting bank erosion, the model may not necessarily overestimate streambank erosion. Nevertheless, the reliance on a calibration coefficient to account for input data limitations reduces confidence in model predictions, as it could lead to situations where the model gives the right answer for the wrong reason which will inhibit parameter transfer outside the calibration dataset. This paper has: (1) demonstrated that the approach adopted in this study using LiDAR DEM may be a more reliable alternative than determining river channel dimensions as a power function of catchment area, and the application of the concept of recurrence interval in estimating bankfull flow, and (2) shown that the assumption of rectangular river channel geometry in the application of Dynamic SedNet in P2R is overestimating bankfull flow and associated hydraulic attributes such as hydraulic radius in the case study reaches.