Geosmin off-flavor compromises product quality in recirculating aquaculture systems (RAS), yet predictive tools for managing this challenge remain inadequate. Existing bioconcentration models treat water concentration with simplifying assumptions and omit removal pathways operating in commercial systems, producing parameter inconsistencies that undermine confidence in their predictions. This dissertation develops and validates a mechanistic modeling framework for geosmin biokinetics in Atlantic salmon (Salmo salar) RAS, integrating process engineering approaches with experimental aquaculture science. The research addressed three objectives through complementary experimental and modeling efforts. Controlled trials evaluated TiO₂/UV + H₂O₂ advanced oxidation for accelerating

geosmin depuration, demonstrating 4.3-fold faster tissue clearance relative to controls while revealing that dissolved organic carbon substantially reduces treatment efficacy under production

conditions. Tissue distribution studies across multiple organs established strong blood-tissue correlations (R² > 0.89) supporting blood-based monitoring as a non-lethal alternative to fillet

sampling, and identified ovarian tissue as a preferential geosmin reservoir with potential implications for depuration in maturing fish.

These experimental findings informed development of an enhanced two-compartment model incorporating bacterial biodegradation, volatilization, and dissolved organic carbon partitioning

corrections. Application to published rainbow trout data suggests that unaccounted bacterial degradation may explain the 6-fold discrepancy between fish-fitted and water-fitted rate constants reported in prior work. A three-compartment extension enabled quantitative evaluation

of removal technologies. Model application yielded the first experimentally-derived kinetic coefficients for market-size Atlantic salmon under the tested conditions: k₁ = 30–40 L kg⁻¹ d⁻¹, k₂ = 0.20–0.60 d⁻¹, kₘ = 0.03–0.06 d⁻¹, corresponding to an elimination half-life of

approximately 3.3 days. 

The framework provides a foundation for predicting geosmin dynamics and evaluating treatment technologies, though transferability of parameters to other systems and conditions requires further investigation.