Estimation of Open Channel Bathymetry and Discharge with Large-Scale Particle Image Velocimetry

The discharge of river is the most important information in the water resource planning. One of the non-contact measurement techniques named Large-Scale Particle Image Velocimetry (LSPIV), is widely used in measuring the surface flow velocity and estimating the discharge. The flow condition, however, is usually complex and not controllable in the field. These unfavorable factors directly raise the error in matching Interrogation Area (IA) when direct cross-correlation algorithm (DCC) used in LSPIV. Moreover, the cross-section area of the channel is usually unknown while measuring surface velocity. The unknown water depth would provide another uncertainty to estimate the discharge with LSPIV. In order to evaluate the feasibility of LSPIV for measuring flow and water depth variation, a series of experiments were conducted in a 30 m long and 1 m wide indoor flume with five different shapes of sub-bed structures with various flow rates under controlled illumination. To improve the aforementioned errors in algorithm, the Convolutional Neural Network (CNN) which is powerful on the object detection was conducted to take more characteristics on river surface into surface velocity measurements instead of DCC. The effect of IA size and CNN layers toward velocity estimation was also evaluated in this study. The bathymetry analysis uses the 2D surface coordinates and velocity data obtained by LSPIV, together with the Shallow water equations in a grid with Courant-Friedrichs-Lewy (CFL) condition using the leapfrog method. The LSPIV measurements will also be compared with the Acoustic Doppler Velocimetry (ADV) measurements to investigate the feasibility of the LSPIV technique for flow measurement. Keywords: LSPIV, CNN, Surface velocity, Open channel discharge, ADV