Dam Monitoring Dataset and Code for Data Reconstruction and Forecasting of Incomplete Time Series

This repository contains selected, anonymized data and Python code supporting the manuscript “Data Reconstruction Strategies for Reliable Forecasting of Incomplete Dam Monitoring Time Series.”The complete raw dam-monitoring datasets are not included because of data-ownership and infrastructure-security restrictions. The archive includes a selected 646-hour GPS displacement dataset for Dam A represented through six missing-data and reconstruction scenarios (Types 1–6), reconstruction and validation logs, downstream XGBoost forecasting results, aggregated cross-dam performance and SHAP summaries for Dams A–C, selected figures, and executable Python code.Variables are hourly X-, Y-, Z-, and V-directional GPS displacement measurements in millimetres. See README.md and DATA_DICTIONARY.md for the reproducibility scope and limitations.