Ensemble-based data assimilation for estimation of river depths

Abstract

A method is presented for estimating bathymetry in a river, based on observations of depth-averaged velocity during steady flow. The estimator minimizes a cost function that combines known information in the form of a prior estimate and measured data (including measurement noise). State augmentation is used to relate the measured variable (velocity) to the unknown parameter (bathymetry). Specifically, the unknown consists of deviations in depth about a known along-channel mean. Verification of the method is performed using a simple 1D channel geometry as well as for two real-world reaches. In all cases, the verification is based on nominal river depths of 3-10 m, channel widths of 50-100 m, and Froude numbers much less than one. Further tests are performed to assess the usefulness of various observation types and sampling schemes for this type of estimation. © 2012 American Meteorological Society.