Data and code for diffuse anthropogenic water-pollution assessment under climate and land-use change

This dataset contains the processed data, core result tables, source-provenance metadata, and core analysis scripts supporting the study “Explainable machine learning for diffuse anthropogenic water-pollution assessment and adaptive watershed management under climate and land-use change”. The package supports reproducibility of the manuscript analyses, including harmonized nutrient station-year records, pesticide occurrence-screening data, environmental covariates, model evaluation outputs, interpretation results, health-screening outputs, and watershed-priority products. The workflow integrates water-quality monitoring records with climate, land-use, agrochemical, watershed, and monitoring-context information derived from open sources including Water Quality Portal, GEMStat, EEA Waterbase, GLORICH, CHIRPS, HYDE, and FAOSTAT. The uploaded archive includes cleaned/core analysis data, model-result tables, figure-supporting results, source-provenance documentation, and the core scripts required to reproduce the main analytical workflow. Large third-party raw source datasets are not redistributed in full; instead, source links, retrieval information, and processing metadata are provided so that users can trace the original data sources and understand the construction of the analysis-ready dataset. The dataset is intended for reproducing and checking the manuscript results, including temporal, spatial, and external validation of machine-learning models, feature-attribution analyses, uncertainty diagnostics, pesticide occurrence screening, and HUC8 watershed-priority outputs for adaptive monitoring and mitigation planning.