Surface Water Observations through Remote Optics-Based Techniques

nclimate1908-f1Autore del lavoro candidato: Flavia Tauro

SINTESI CONTENENTE UNA BREVE DESCRIZIONE DEL LAVORO SVOLTO E DEI RISULTATI OTTENUTI: This research focus is to advance the current understanding of the water cycle by establishing novel methodologies for the observation of surface waters. Specifically, these novel methods aim at reconstructing the surface velocity field of natural water bodies based on non-invasive acquisition of digital images. The surface water velocity field is computed by advanced image treatment and through application of high-performance velocimetry algorithms. Compared to existing traditional measurement methods, the approach enables continuous and remote kinematic characterization of small to large scale ungauged water environments that are not accessible via standard instrumentation. This is expected to directly impact flash flood forecasting and environmental risk management. Major results of the presented research activity can be summarized as follows. – Kinematic characterization of difficult-to-access water environments has been enabled through integration of high-visibility surface tracers and image-based analysis. – A novel class of high-visibility and environmentally-friendly hydrological tracers has been synthesized for studying surface flows physics. – For the first time in Earth sciences, drones have enabled the characterization of large scale water systems at high space and time resolutions. The approach is based on the use of low-cost drones and image analysis for the generation of accurate surface flow maps of water bodies. This result is remarkable and expected to open novel frontiers in flood risk management and engineering practice. – A novel set of computationally inexpensive tools for image enhancement and velocity extraction has been established. They include a template-based high-speed correlation algorithm, morphological segmentation procedures, and image classification methods based on manifold learning approaches. In particular, unsupervised machine learning has proved instrumental to rapidly unravel flow features and to automatically detect the regime of unknown flows. – Integrated, resilient, and high-performance sensing platforms for noninvasive and continuous surface flow monitoring have been designed and developed