MORPHOLOGICAL CHANGE DETECTION AT PIPELINE CROSSINGS USING REMOTE SENSING: A PROOF-OF-CONCEPT STUDY

A proof-of-concept research and development study was undertaken to determine to what degree changes in river morphology could be detected from satellite imagery in support of pipeline geohazard monitoring programs. Algorithms for the detection of the wetted extent and river centerline were developed in the Python programming language and applied to synthetic aperture radar and multispectral imaging data. Additional algorithms were developed to highlight meaningful changes in wetted extent and river centerline occurring between subsequent data acquisitions. The results showed that it was possible to detect morphological change using satellite imagery, but that the process was effective for rivers with a higher signal to noise ratio represented by the ratio of river width to image pixel size. For smaller rivers (20-50 m), the change in the centerline location was a more useful metric of morphological change. The change detection algorithms were successful at highlighting possibly important changes between subsequent data acquisitions but were limited by the relatively short catalogue of available images to compare changes against. Future work involves the improvement and deployment of the algorithms in operational software systems to support pipeline geohazard management programs. Keywords: Earth Observation, Machine Learning, Computer Vision, Morphological Change Detection, Synthetic Aperture Radar, Multispectral Imaging, Geohazard Management