An AI-ready shoreline forecasting dataset under wind–wave conditions and relative sea-level rise scenarios

Description AI4Shore is an open, AI-ready coastal dataset developed to support machine learning and deep learning applications for shoreline change analysis and forecasting along the Polish Baltic coast. The dataset integrates multi-source environmental observations into a harmonized transect-based database spanning 1987–2025 and provides future sea-level forcing scenarios extending to 2150 under SSP3-7.0 scenario. The dataset combines annual satellite-derived shoreline positions, shoreline change metrics, wind–wave characteristics, historical sea-level anomalies, and regional sea-level rise projections into a standardized format suitable for spatiotemporal modelling of shoreline evolution. In addition, all processing scripts, technical validation results, and a baseline Long Short-Term Memory (LSTM) benchmark model are provided to facilitate reproducibility and future methodological development. Repository contents This repository contains: Annual shoreline vectors (1987–2025) in GeoJSON format DSAS-generated transects and shoreline change metrics AI-ready training dataset (1987–2025) Future sea-level forcing dataset (2026–2150; IPCC AR6 SSP3-7.0) Wind–wave feature tables derived from Copernicus Marine reanalysis Historical sea-level anomaly dataset Baseline LSTM benchmark model Google Earth Engine scripts for shoreline extraction Python scripts and Jupyter notebooks for data processing, feature engineering, validation, benchmarking, and visualization Applications AI4Shore is intended for: shoreline change modelling and forecasting; deep learning and machine learning research; coastal hazard and erosion assessment; climate change impact studies; coastal management and adaptation planning; benchmarking of spatiotemporal prediction models.