Detecting morphological change from DInSAR data in ephemeral, braided, gravel bed rivers

Estimating morphological change in braided gravel bed rivers is important to determine the geometry of the active part of the braidplain, i.e. the width of the area where bed material is transported and the number of active branches during a certain time interval. Repeated surveys and photogrammetry have been used to measure morphological change with different spatial and temporal resolution but characterizing the geometry of braided rivers remains a problem. Increased satellite coverage and frequency of revisiting times motivated satellite-based studies of river dynamics. Here we present a procedure to estimate morphological change in ephemeral, braided, gravel bed rivers from Differential Interferometry of Synthetic Aperture Radar (DInSAR) data, i.e. information on areas where ground displacement larger than 5-6 cm occurred. We hypothesize that such displacement is representative of bedload transport and can be used to describe the morphologically active braidplain. The procedure is applied to the Trionto River in Southern Italy. Results indicate that the proposed approach has the potential to capture different channel dynamics related to lateral confinement and flood hydrographs. It is found that, in agreement with field and laboratory observations reported in the literature, narrower morphological active width and fewer active branches are estimated where valley confinement is largest. In response to a frequent flood, active width remains independent of lateral confinement and one or two active branches are predicted. Conversely, a large, less frequent flood results in the reorganization of the channel network with larger active width and more active branches in the widest part of the braidplain. To improve scalability and reproducibility, the entire procedure has been automated through a connectivity-based algorithm. The identification of active channels and the measurement of morphological active width are both performed automatically, significantly reducing manual interpretation. To demonstrate the versatility of the code, the same workflow is also applied to field data acquired by drone surveys.