The accuracy of river velocity measurements plays an important role in the effective management of water resources. Various methods have been developed to measure river velocity. Currently, image-based techniques provide a promising approach to avoid physical contact with targeted water bodies by researchers. In this study, measured surface velocities collected under low flow and high flow conditions in the Houlong River, Taiwan, using large-scale particle image velocimetry (LSPIV) captured by an unmanned aerial vehicle (UAV) and a terrestrial fixed station were analyzed and compared. Under low flow conditions, the mean absolute errors of the measured surface velocities using LSPIV from a UAV with shooting heights of 9, 12, and 15 m fell within 0.055 ± 0.015 m/s, which was lower than that obtained using LSPIV on video recorded from a terrestrial fixed station (i.e., 0.34 m/s). The mean absolute errors obtained using LSPIV derived from UAV aerial photography at a flight height of 12 m without seeding particles and with different seeding particle densities were slightly different, and fell within the range of 0.095 ± 0.025 m/s. Under high flow conditions, the mean absolute errors associated with using LSPIV derived from terrestrial fixed photography and LSPIV derived from a UAV with flight heights of 32, 62, and 112 m were 0.46 m/s and 0.49 m/s, 0.27 m, and 0.97 m/s, respectively. A UAV flight height of 62 m yielded the best measured surface velocity result. Moreover, we also demonstrated that the optimal appropriate interrogation area and image acquisition time interval using LSPIV with a UAV were 16 × 16 pixels and 1/8 s, respectively. These two parameters should be carefully adopted to accurately measure the surface velocity of rivers.

An automated discharge imaging system (ADIS), which is a non-intrusive and safe approach, was developed for measuring river flows during flash flood events. ADIS consists of dual cameras to capture complete surface images in the near and far fields. Surface velocities are accurately measured using the Large Scale Particle Image Velocimetry (LSPIV) technique. The stream discharges are then obtained from the depth-averaged velocity (based upon an empirical velocity-index relationship) and cross-section area. The ADIS was deployed at the Yu-Feng gauging station in Shimen Reservoir upper catchment, northern Taiwan. For a rigorous validation, surface velocity measurements were conducted using ADIS/LSPIV and other instruments. In terms of the averaged surface velocity, all of the measured results were in good agreement with small differences, i.e., 0.004 to 0.39 m/s and 0.023 to 0.345 m/s when compared to those from acoustic Doppler current profiler (ADCP) and surface velocity radar (SVR), respectively. The ADIS/LSPIV was further applied to measure surface velocities and discharges during typhoon events (i.e., Chan-Hom, Soudelor, Goni, and Dujuan) in 2015. The measured water level and surface velocity both showed rapid increases due to flash floods. The estimated discharges from ADIS/LSPIV and ADCP were compared, presenting good consistency with correlation coefficient R = 0.996 and normalized root mean square error NRMSE = 7.96%. The results of sensitivity analysis indicate that the components till (τ) and roll (θ) of the camera are most sensitive parameters to affect the surface velocity using ADIS/LSPIV. Overall, the ADIS based upon LSPIV technique effectively measures surface velocities for reliable estimations of river discharges during typhoon events.

Image-based gauging stations offer the potential for substantial enhancement in the monitoring networks of river water levels. Nonetheless, the majority of camera gauges fall short in delivering reliable and precise measurements because of the fluctuating appearance of water in the rivers over the course of the year. In this study, we introduce a method for measuring water levels in rivers using both the traditional continuous image subtraction (CIS) approach and a SegNet neural network based on deep learning computer vision. The historical images collected from on-site investigations were employed to train three neural networks (SegNet, U-Net, and FCN) in order to evaluate their effectiveness, overall performance, and reliability. The research findings demonstrated that the SegNet neural network outperformed the CIS method in accurately measuring water levels. The root mean square error (RMSE) between the water level measurements obtained by the SegNet neural network and the gauge station's readings ranged from 0.013 m to 0.066 m, with a high correlation coefficient of 0.998. Furthermore, the study revealed that the performance of the SegNet neural network in analyzing water levels in rivers improved with the inclusion of a larger number of images, diverse image categories, and higher image resolutions in the training dataset. These promising results emphasize the potential of deep learning computer vision technology, particularly the SegNet neural network, to enhance water level measurement in rivers. Notably, the quality and diversity of the training dataset play a crucial role in optimizing the network's performance. Overall, the application of this advanced technology holds great promise for advancing water level monitoring and management in river systems.

An automated discharge imaging system (ADIS), a non-intrusive and safe approach, was developed for measuring river flows during flash flood events. ADIS consists of dual cameras to capture complete surface images in the near and far fields. Surface velocities are accurately measured using the Large Scale Particle Image Velocimetry (LSPIV) technique. The stream discharges are then obtained from the depth-averaged velocity (based upon an empirical velocity-index relationship) and cross-section area. The ADIS was deployed at the Yu-Feng gauging station in Shimen Reservoir upper catchment, northern Taiwan. For a rigorous validation, surface velocity measurements were conducted using ADIS/LSPIV and other instruments. In terms of the averaged surface velocity, all measured results were in good agreement with small differences, i.e., 0.004 to 0.39 m/s and 0.023 to 0.345 m/s when compared to those from acoustic Doppler current profiler (ADCP) and surface velocity radar (SVR), respectively. The ADIS/LSPIV was further applied to measure surface velocities and discharges during typhoon events (i.e., Chan-Hom, Soudelor, Goni, and Dujuan) in 2015. The measured water level and surface velocity both showed rapid increases due to flash floods. The estimated discharges from ADIS/LSPIV and ADCP were compared, presenting good consistency with correlation coefficient R = 0.996 and normalized root mean square error NRMSE = 7.96%. The results of sensitivity analysis indicate that components till (τ) and roll (θ) of the camera are most sensitive parameter to affect the surface velocity using ADIS/LSPIV. Overall, the ADIS based upon LSPIV technique effectively measures surface velocities for reliable estimations of river discharges during typhoon events.