Data-driven modeling of municipal water system responses to hydroclimate extremes

Abstract Sustainable western US municipal water system (MWS) management depends on quantifying the impacts of supply and demand dynamics infrastructure reliability vulnerability. Systems modeling can replicate interactions but extensive parameterization, high complexity, long development cycles present barriers to widespread adoption. To address these challenges, we develop Machine Learning Water Model (ML-WSM) – a novel application data-driven for MWS management. We apply ML-WSM framework Salt Lake City, Utah system, where benchmark prediction performance seasonal response reservoir levels, groundwater withdrawal, imported requests climate anomalies at daily resolution against an existing systems model. The accurately predicts all components; especially during supply-limiting conditions (KGE > 0.88, PBias < ±3%). Extreme wet challenged model skill communicated appropriate trends relationships component thresholds (e.g., dead pool). correctly classified nearly instances vulnerability (83%) peak severity (100%), encouraging its use as guidance tool that complements models evaluating influences performance.