3D definition of the river bankfull stage from topographic LiDAR

Abstract

Determining bankfull width and depth parameters, as indicators of hydrosystem responses to pressures, is a critical step in hydromorphological analysis and in assessing the degree of river alteration. Over the last few decades, numerous studies have proposed methods for identifying bankfull stage from remote sensing data, generally at the cross‐section scale. The automatic Bf3D method developed in this study aims to extend the scale of analysis to homogeneous river reaches, by using a 3D definition of hydraulic depth. A high‐density light detection and ranging (LiDAR) point cloud (>10 pulses/m ² ) was used to test the method on 28 river reaches distributed throughout mainland France representing a wide range of hydromorphological characteristics and fluvial forms. The results were compared to three other bankfull width datasets: (i) field acquisition (), (ii) an expert delineation on a LiDAR DTM () and (iii) an automatic calculation based on 2D hydraulic depth ( Bf2D) . The Mean Absolute Percentage Deviation (MAPD) between the bankfull widths of the two manual references ( and ) is 19%. This deviation between and Bf2D reaches 37% , while it reaches 26% for Bf3D . Moreover, the Mean Percentage Deviation (MPD) in comparison to showed that the Bf3D methodology is significantly more robust across river hydromorphological characteristics than Bf2D . In comparison to bankfull widths from , the MPD of Bf3D is only 5%, close to the MPD of 4% and far from the Bf2D MPD of 33%. Analysis of the variability of bankfull width along reaches showed that the improved accuracy and robustness of Bf3D over Bf2D are mainly explained by avoiding reliance on individual cross‐sections, thus allowing a more homogeneous representation of the river channel section, particularly when applied in an automated system. However, automated methods, including Bf3D , still have certain limitations, particularly for the analysis of confined rivers and braided river systems. Bf3D provides a robust, automated framework for large‐scale hydromorphological assessments and improved monitoring of river‐system responses.