U.S. Drinking Water CCR Rich-Fields Dataset

Per-utility Consumer Confidence Report rich enrichment for 5,572 U.S. public water systems. PWSID-keyed. Derived from EPA SDWIS + state health department CCR archives parsed via three-tier OCR + LLM extraction. 27 fields per record (18 L1 + 9 L3 derived metrics including pfas_signal, lsl_progress_enum, softener_trigger, bilingual_required). Outlier-validated under ZipCheckup CCR display-bounds policy. Distinct from the companion ZIP-keyed weekly water-quality dataset (DOI: 10.5281/zenodo.20121062) — this version is utility-granularity for direct research use. License: CC-BY-4.0.