Estimating river bathymetry from multisource remote sensing data

Abstract

River bathymetry is a key variable in estimating river discharge by remote sensing. However, traditional methods of mapping river bathymetry are expensive and time-consuming, which hinders the estimation of river discharge. In this study, we propose and test a method for estimating river bathymetry by combining remotely sensed observations of water level (z) and river width (w) with a nonlinear z-w relationship derived from a generalized channel cross-sectional shape. The results from five reaches of the upper Yangtze River show that the absolute relative error for the estimate is between 1.25% and 6.18%, indicating the ability of this method to provide accurate estimates. The channel exposure (e) is the main limitation of the proposed method, and the river bathymetry estimates improve significantly with increasing e. For the upper Yangtze River, averaging reaches over an appropriate length can reduce variability in river bathymetry estimate, especially when data are reach-averaged over a length of 10 km. Considering that a priori bathymetric information is not required and its computational process is relatively simple, the proposed method could open up the possibility of bathymetry modeling of global rivers.