TerraceM 3.0: Advancing marine terrace mapping through integrated machine learning methods
Résumé
TerraceM is an open-source software written in MATLAB for mapping and analyzing marine terraces. In this latest release, TerraceM-3 has undergone significant evolution, which leverages the capabilities of machine learning to introduce an automated marine terrace mapping feature. This new version includes a neural network that has been meticulously trained with over 1000 mapped marine terraces. This allows TerraceM-3 users to effortlessly map marine terraces and precisely determine their elevation through the automated mapping of their shoreline angles. In addition, TerraceM-3 incorporates two new functionalities: 1) Photon profile mapping, which includes mapping of satellite LiDAR profiles from the IceSat-2 mission, which broadens the applicability of TerraceM-3 beyond the availability of topographic data. 2) Indicative meaning calculator that accounts for the factors that can alter the initial sea-level position using global datasets (wave conditions and tidal ranges). This method facilitates the direct assessment of uncertainties in the reconstructions of the paleo-sea-level based on marine terraces. TerraceM-3 is a complete toolkit for researchers and students engaged in marine terrace analysis by offering a unique blend of numerical methods, statistical analyses techniques and additional enhanced functionalities to precisely map marine terraces and using them as markers of tectonic deformation.