All of the bacteria load in the Mill Creek watershed originates from nonpoint sources. The nonpoint sources of bacteria are mainly agricultural and include land-applied animal waste and manure deposited on pastures by livestock. A significant bacteria load comes from cattle and wildlife directly depositing feces in streams. The Hydrological Simulation Program – FORTRAN (HSPF) was used to simulate the fate and transport of fecal coliform bacteria in the Mill Creek watershed. To identify localized sources of fecal coliform within the Mill Creek watershed, Mill Creek was divided into 7 sub-watersheds, based primarily on confluences with tributaries. As no observed flow data was available for the watershed and the neighboring Hawksbill Creek watershed had recently been modeled with HSPF as part of a TMDL study, the calibrated hydrology parameter values from Hawksbill Creek were applied to Mill Creek.

Contributions from various sources within the Mill Creek watershed were represented in HSPF to simulate bacteria loadings for existing conditions. Average annual loads were calculated using meteorological inputs for the same 1990-2002 period used in neighboring Hawksbill Creek. Forty-one percent of the fecal coliform in the mean daily fecal coliform concentration comes from cattle directly depositing in the stream, 52% from upland areas due to runoff, 5% from wildlife directly depositing in the stream, and approximately 2% from straight pipes, sanitary sewer overflows, runoff from impervious areas, and contributions from interflow and groundwater.

For existing conditions in Mill Creek, simulated bacteria concentrations exceeded the calendar-month geometric mean water quality standard 77% of the time, and by larger amounts during low flow periods and the summer. During the summer when stream flow is lower, cattle tend to spend more time in streams, increasing direct fecal coliform deposition to streams when water for dilution is least available. Based on amounts of bacteria produced in different locations, monthly bacteria loadings to different land use categories were calculated for each subwatershed for input into the respective models. For the TMDL allocation scenarios, a target of zero violations of both the instantaneous and geometric mean water quality standards was used. Various allocation scenarios were developed to meet the standard as shown in the Table below.

Scenarios 4 and 5 represent alternative Stage 1 reduction scenarios with single standard criterion violation rates less than 10%. Scenario 6 shows that violations of the single sample criterion cannot be eliminated even with 100% reductions from everything except background wildlife sources. The last scenario – Scenario 7 – was developed to eliminate all violations of both the calendar-month geometric mean and the single sample criteria and requires, in addition to all the previous reductions, a 40% reduction in the direct deposits in land deposited waste from wildlife, whose reductions are exempted under current state guidelines.