Applying image recording and identification for measuring water stages to prevent flood hazards

Information on river stages can be transmitted to relevant management offices over a network by using an automatic stage gauge so that management offices can effectively evaluate whether a river stage is exceeding the warning line and take action if necessary. However, current river stage measurement methods can present this information only in data form because the methods cannot simultaneously obtain images such as the rising or overflow of the river stage. In addition, the stage gauge can fail or be washed away when the river stage is high. To solve these problems, this research evaluates a novel measurement method involving a video surveillance system that exhibits features such as easy installation, low maintenance cost, and low failure possibility. Through on-site image recording, this measurement method involves using image identification technology to read water level figures automatically. This method offers instant river stage figures and on-site video so that disaster prevention measures can be implemented accordingly. The results of a dynamic water flow test conducted in an indoor experimental channel indicated that all of the average absolute error levels of river stage identification were less than ±1.2 %, meaning that the image identification technology could achieve identification results at any flow level. By contrast, the findings of a rainfall simulation experiment suggest that the average absolute error of river stage identification was less than ±2.5 %, meaning that the measurement technology and method used in this research are useful and feasible at various rainfall intensities.     

Potential hazard analysis from the viewpoint of flow measurement in large open-channel junctions

A robust and reliable risk assessment procedure for hydrological hazards deserves attention. The transport of mixed material in the types of hazard, such as flash floods and debris flows, plays an important role and should be taken into consideration. At present, however, the mixed material transport phenomenon is not systematically and simply included in the procedures for the elaboration of hazard analysis. The consequence is the risk of losing prediction accuracy and of underestimating hazard impacts. The phenomenon of open-channel junction flow such as in irrigation ditches and at wastewater treatment facilities is an important aspect of hydrological engineering. Extensive experimental studies of the flow characteristics have been conducted. In this study, an Acoustic Doppler Velocimeter is used to monitor the flow in large open-channel junctions at the Chihtan purification plant in Taipei. The condition of the experiment is made up of different discharge. The result of the measurement data shows that the mean velocity profile, aspect ratio, measured vertical line location and the ratio of the distance of the vertical line maximum velocity from the bed to water depth. An artificial neural network (ANN) is used to simulate the mean velocity profiles in this study. The results show that the ANN can accurately and reliably simulate the mean velocity profiles. The result is believed to provide hints for the consideration of the effects, induced by mixed material transport, in the elaboration of hazard analysis.