Image Analysis and Optical Algorithms for Streamflow Sensing

Since the 1990s, streamflow observations have been conducted through image-based velocimetry techniques, such as Large Scale Particle Image Velocimetry. Since then, surface streamflow sensing has entailed the installation of permanent and cost-effective gauge-cams in the proximity of water bodies to continuously monitor flow dynamics. These automated systems comprise digital cameras, controlling units, and, in some cases, laser systems for fully remote photometric calibration. Gauge-cams collect a large volume of images of the water surface that can be off-line analyzed to inspect the stream flow regime at high temporal resolution. However, image-based techniques and gauge-cams are rarely systematically implemented in practical engineering operations probably due to the lack of consistent image processing protocols.

Towards the establishment of shared good practices, considerable experimental work has been conducted based on image data gathered through gauge-cams. These studies have highlighted the promise of particle tracking velocimetry algorithms for reconstructing surface flow kinematics. In particular, the combination of feature extraction methods and sparse optical flow is very suitable for real-time surface flow velocity estimation in noisy images taken in outdoor conditions. Further, the integration of this approach with trajectory-based filtering constraints allows for identifying reliable trajectories that can provide accurate flow velocity estimation in diverse flow regimes.

In this poster, a novel optical flow procedure for surface flow velocity estimation in stream systems is briefly outlined. Also, two gauge-cams recently installed in Italy are illustrated and several case studies are presented in two Italian riverine systems in different flow regimes.