Abstract
In situ fluorescent dissolved organic matter (fDOM) sensors provide high-frequency proxies for dissolved organic carbon (DOC), but turbidity interference reduces accuracy in sediment-rich environments. A physically grounded, three-parameter (3P) exponential turbidity correction was developed for YSI EXO fDOM sensors, anchored to laboratory-based fluorescence measurements of discrete samples. Paired in-situ and laboratory datasets from 2021–2024 across 16 United States Geological Survey (USGS) continuous monitoring stations in the Sacramento–San Joaquin Delta were used to fit and compare regional (Delta-wide), sub-regional, and site-specific correction models. Each model related turbidity to the ratio between temperature-corrected field fDOM and laboratory measurements collected on a benchtop instrument standardized to quinine sulfate units. Three-parameter models substantially reduced turbidity effects relative to the single parameter correction used widely across USGS monitoring stations. Median normalized root mean squared error decreased from 0.082 to 0.038 (~54% improvement), while median concordance correlation coefficient increased from 0.934 to 0.984. Three-parameter models were most frequently selected as the top model using Akaike Information Criterion and other metrics. The sub-regional model remained stable at higher turbidity values, whereas single-station corrections were under-constrained at a few stations, resulting in erroneous spikes during high turbidity events when applied to the full timeseries. The resulting framework is physically grounded using laboratory data and transferable to any monitoring network where co-located field and discrete measurements exist. By improving fDOM accuracy under variable turbidity, this approach enhances DOC tracking in dynamic, high turbidity systems. Use of these new corrections will harmonize long-term fDOM records and improve DOC tracking, supporting ecosystem science and source-water management in aquatic ecosystems.