Image-based river discharge estimation by merging heterogeneous data with information entropy theory

An information entropy based approach for the discharge measurements is evaluated for the gaging of the Isere river at the Grenoble university campus. Over a four month period, six discharge measurements were made using a vessel-mounted aDcp. Simultaneously, particle tracking velocimetry (PTV) from video images was used to estimate surface velocities. The surface velocities are projected along the regularly surveyed river section of the Isere-Campus gaging station. The vertical velocity profile at each stream-wise location is approximated by a 1D entropy profile. Information entropy 1D velocity vertical profile depends on two parameters which are fitted using aDcp and surface velocity measurements. The inclusion of the surface velocities reduces the dispersion of the estimated entropy parameters. The measurements show that the two parameters are linearly related with a slope that is stage dependent and thus, surface velocity dependent. From there, the information entropy theory for 1D velocity distribution offers a protocol by which surface velocities only are used to compute the discharges. The protocol is calibrated with both aDcp and surface velocity measurements. It is finally validated with several events during which only surface velocities are measured. For the high water flood event the estimated discharge falls within 2% of the one estimated with the rating curve of the gaging station.

Automated summary

An information entropy based approach is used to measure the discharge of the Isere river, with parameters that are linearly related to surface velocities, reducing dispersion. The protocol is calibrated with both aDcp and surface velocity measurements, and validated during events where only surface velocities are measured, showing accurate results.