Estimating river bathymetry from multisource remote sensing data

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 ba-thymetry 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.