Space-time velocimetry to estimate discharge during the 2022 River Murray flood

: Streamflow discharge measurement underpins a range of assessments, policy, and management related to resource management. The standard methods to measure discharge can be costly due to the time consuming and labour-intensive manual measurements required by highly specialized staff, particularly in remote and difficult to access sites. Surface velocity measurements achieved through video image analysis are becoming increasingly popular methods to estimate velocity and discharge, driven by remote pilot aircraft (RPA, or drone) and camera technology. These methods have the advantage of being non-intrusive and hence improved safety during high flow measurements, are suited to low flows and depths and inexpensive measuring equipment can be deployed remotely not requiring staff to be present. This paper demonstrates the application of video-based surface velocity methods during the peak of a high flow event in the River Murray in late 2022, peaking at approximately 200 GL/d (an annual exceedance probability of approximately 1 in 50). Six videos were recorded with an RPA and a mobile phone camera at five locations between the townships of Renmark and Berri. The Space Time Image Velocimetry (STIV) method was used to compute surface velocities and available survey information was used to derive coordinates to orthorectify the video as well as river cross section bathymetry. The STIV method uses changes in brightness of the river surface in the direction of flow (a distance in space in the image) over time (between video frames) to produce diagonal lines on a combined image, with the slope of the line representing the surface velocity. Three methods to estimate surface velocity were tested in combination with two methods to convert the surface velocity to the mean channel velocity. The deep learning method with a log-law relationship to derive mean channel velocity was found to perform the best for the videos recorded when compared to more traditional Acoustic Doppler Current Profiler discharge measurements recorded at the same time. The results demonstrate that relatively accurate discharge estimates can be achieved with minimal equipment, just a phone camera on the riverbank. The other data requirements, survey of points to orthorectify the video into real-world distances and survey of the river cross section to compute discharge, and the water level relative to these points, become the more significant data requirements to estimate discharge