Long-Term Impacts of Bacteria-Sediment Interactions in Watershed-Scale Microbial Fate and Transport Modeling

Elevated levels of fecal indicator bacteria (FIB) remain the leading cause of surface water-quality impairments in the United States. Under the Clean Water Act, basin-specific total maximum daily load (TMDL) restoration plans are responsible for bringing identified water impairments in compliance with applicable standards. Watershed-scale model predictions of FIB concentrations that facilitate the development of TMDLs are associated with considerable uncertainty. An increasingly cited criticism of existing modeling practice is the common strategy that assumes bacteria behave similarly to "free-phase" contaminants, although many field evidence indicates a nontrivial number of cells preferentially associate with particulates. Few attempts have been made to evaluate the impacts of sediment on the predictions of in-stream FIB concentrations at the watershed scale, with limited observational data available for model development, calibration, and validation. This study evaluates the impacts of bacteria-sediment interactions in a continuous, watershed-scale model widely used in TMDL development. In addition to observed FIB concentrations in the water column, streambed sediment-associated FIB concentrations were available for model calibration. While improved model performance was achieved compared with previous studies, model performance under a "sediment-attached" scenario was essentially equivalent to the simpler "free-phase" scenario. Watershed-specific characteristics (e.g., steep slope, high imperviousness) likely contributed to the dominance of wet-weather pollutant loading in the water column, which may have obscured sediment impacts. As adding a module accounting for bacteria-sediment interactions would increase the model complexity considerably, site evaluation preceding modeling efforts is needed to determine whether the additional model complexity and effort associated with partitioning phases of FIB is sufficiently offset by gains in predictive capacity.