Locally relevant ambient water quality criteria to protect human health

The U.S. Environmental Protection Agency (USEPA) uses a deterministic risk-based framework to derive national recommendations for ambient water quality criteria (AWQC) protective of human health through the ingestion of water and aquatic organisms. States are required to either adopt these recommendations or propose scientifically defensible alternatives. The deterministic approach has faced criticism for relying on multiple high-percentile input parameters, leading to criteria disconnected from actual risk. Consequently, although some states adopt USEPA's criteria recommendations in their entirety, while others utilize different input parameters or alternative approaches to derive criteria that better represent local exposure conditions. Probabilistic risk assessment (PRA) represents a scientifically robust alternative that offers transparency and flexibility by using full data distributions rather than point estimates to define exposures. This enables a clear linkage between the acceptable risk targets and affected population subgroups. Although USEPA has provided guidance supporting the use of PRA in other regulatory programs, direct guidance on implementing a PRA approach for deriving state-specific AWQC is lacking. This work explores USEPA's risk-based framework and applies both deterministic and probabilistic approaches to quantify patterns in AWQC under different criteria derivation scenarios that alter assumptions of exposure and risk. We implement an open-source R Shiny tool designed to reduce technical barriers and facilitate practical adoption by state agencies, including those without specialized modeling expertise. Outcomes highlight how exposure assumptions, risk thresholds, and derivation approaches affect AWQC, offer a practical guidance for environmental agencies to derive locally relevant and scientifically defensible criteria, and may serve as a basis for a future USEPA technical support document on the use of PRA for AWQC derivation.