Coastal Wetlands Dynamics (Brittany, France) from 1990 to 2020 Dataset

Coastal Wetlands Ecosystems Data Repository (Brittany, France) This dataset supports the forthcoming publication in Ecological Informatics: "Monitoring the dynamics of Coastal Wetlands Ecosystems in Brittany (France) using LANDSAT time series and Machine Learning" It provides the full set of data and processing tools used in the study, including training and validation datasets, classification results, topographic indices, and a reproducible machine learning workflow. 📂 Dataset Structure The repository is organized as follows: Fig/ Preview figures used in the manuscript: 1990_raster.png, 1990_vec.png: example visualizations of classification results (raster and vector formats). Results/ Contains final outputs of the coastal wetland classification: 🔹 Classification_Raster/ Classification_1990.tif, Classification_2020.tif: LANDSAT-based wetland classification maps for 1990 and 2020 (raster format). 🔹 Classification_Vec/ Classification_1990.gpkg, Classification_2020.gpkg: corresponding vector versions of the classifications (GeoPackage format). 🔹 Pre-identification_of_coastal_wetlands/ Potential_wetland_map.tif, Pre_localization_wetland.tif: intermediate outputs identifying likely wetland areas before classification. 🔹 Shoreline_dynamics/ Audierne_change_rate_line_1990_2020.gpkg, audierne_change_rate_polygons_1990_2020.gpkg: spatial data showing shoreline change rates from 1990 to 2020. 🔹 Symbology/ QGIS symbology files: Symbology_raster.qml, Symbology_shoreline_lines.qml, Symbology_shoreline_polygons.qml, Symbology_vector.qml Toolbox/ Index_calculator.ipynb: Python notebook used for spectral index computation. ML_Classifier_RF_.model3: graphical model for Random Forest classification (QGIS/OTB format). Topographic_Index/ Topographic layers derived using SAGA GIS: Topographic_Wetness_Index.tif Topographic_Position_Index.tif Valley_Depth.tif Module_Multiresolution_Index_of_Valley_Bottom_Flatness.tif Terrain_Ruggedness_Index.tif Training_Dataset/ Training.gpkg, Validation.gpkg: vector data used for training and validating the classification model. AOI.gpkg A file defining the Area of Interest (AOI) for the study. Satellite_image_list.txt:A plain text file listing: All LANDSAT scenes used Acquisition dates Scene IDs Sensors and processing levelsThis enables users to retrieve the raw images from USGS EarthExplorer or other public platforms. 📌 Key Components Summary Topographic IndicesDerived with SAGA GIS and provided as GeoTIFFs. Training and Validation DataVector layers in GeoPackage format (.gpkg). Classification ResultsRaster and vector maps of wetland dynamics for 1990 and 2020. Shoreline Dynamics and Wetland Pre-IdentificationOutputs related to shoreline evolution and pre-classification wetland localization. QGIS-Compatible Graphical Model (Random Forest)Provided in .model3 format, requiring Orfeo Toolbox (OTB) integration. Python Workflow (Jupyter Notebook)For spectral index processing and workflow replication. Symbology FilesReady-to-use .qml styles for visualizing all data layers in QGIS. 🧾 Citation If you use this dataset, please cite the forthcoming article in Ecological Informatics (citation format to be updated upon publication). You can also cite this Zenodo dataset directly using the provided DOI (10.5281/zenodo.15721404). ⚠️ Usage Notes The graphical model is designed for use in QGIS with Orfeo Toolbox (OTB) installed. Raw satellite images are not included due to storage constraints. A metadata file with acquisition dates and identifiers is provided to retrieve them from open sources (USGS here). Contact: Adrien Le Guillou – [e-mail] you for your interest in this research. Feedback and collaboration are welcome!