The overarching objective of this project was to further demonstrate the Better Assessment Science Integrating point and Nonpoint Sources modeling system (BASINS.MIL), which was developed under the SERDP-funded Project RC-1547 as a tool for watershed-based decision making on military installations. The specific objectives of the BASINS.MIL demonstration/validation are related to modeling capabilities, model performance, and cost assessment/comparisons of the technology. The combined host sites offer an opportunity to meet the objectives in each of the three categories. Table 1-1 lists the components of the BASINS.MIL demonstration/validation associated with each host site. Components of the demonstration/validation have been performed on Fort Benning (FB) and Fort AP (Amborse Powell) Hill (FAPH). The FB portion of the demonstration/validation is focused on demonstration of the BASINS.MIL capabilities and performance, whereas the objective for the FAPH components is focused on the transferability of the BASINS.MIL modeling framework to a new installation, and to demonstrate the ability of BASINS.MIL to address total maximum daily load issues and small-scale assessments.
Components of BASINS.MIL Demonstration/Validation Associated with Each Host Site
Ft. Benning - BASINS.MIL was used to build a continuous computer simulation model of hydrology and water quality for the watersheds on and surrounding Fort Benning, GA. This model is referred to as the FB Model (or FB Enhanced Baseline Model). Preliminary model applications of the FB Model were performed to provide proof-of-principle demonstration of the modeling system and the model enhancements to support watershed management decisions on the installation. The demonstration/validation of BASINS.MIL would be the next step to fully demonstrate the validity of the technology to meet the Department of Defense's (DoD) need for tools to evaluate watershed hydrology and water quality for system-level assessments. The technology transfer of BASINS.MIL demonstration/validation leverages the watershed model developed on FB by conducting further modeling applications on FB, and by developing a watershed model for another installation that was used to further demonstrate the technology. The demonstrations and validations that were performed at FB include the following tasks: data richness versus model performance, unpaved road sediment erosion modeling application, model demonstration for climate non-stationarity and land use change impact analysis, and demonstration of sensitivity and uncertainty analysis.
AP Hill - FAPH, VA was selected as the second site since it provides a unique opportunity to (a) demonstrate the transferability of the BASINS.MIL modeling framework to a new installation and (b) demonstrate the ability of BASINS.MIL to address Total Maximum Daily Load (TMDL) and Best Management Practice (BMP) issues at a smaller subwatershed scale. The demonstrations and validations that were performed at FAPH include the following tasks; model set up and model calibration/Validation, assessment and comparison of model loading rates and small-scale storm water runoff analysis.
Ft. Benning - Recall that Imhoff and others (2010) derived a possible range of 2 to 20 tons/acre/year for unit area edge-of-road sediment washoff from unpaved roads. Elliot’s estimation of 10 tons/acre/year for edge-of-road erosion from Fort Benning’s GAB roads fits comfortably into the middle of this range. However, this report reflects that the 70 tons/acre/year estimate that Elliot derived using the monitoring data (and generated rainfall) from the unimproved road site more closely approximates a worst case scenario: the road section that Fort Benning and USFS staff selected as the monitoring site was steeper than most roads in Fort Benning’s unpaved road network, and it experienced heavier traffic than most roads, including tank traffic. As a result, rutting was prevalent, and erosion was extreme.
AP Hill - The three storm events were run to evaluate the effectiveness of the ASP BMP. The main focus for this application was determining peak flow rate out of the ASP detention pond and the maximum storage required for each storm. The peak flow rates were 0.605, 1.58, and 2.52 cfs for the 2-year, 10-year, and 100-year storms, respectively. The required storage volumes were 868, 1126, and 1374 cu-ft for the 2-year, 10-year, and 100-year storms, respectively. The flow and storage volume results were also shown graphically. Using only the developed rating and simulated storage volume data, it was determined that the existing detention pond can easily handle a 100-year, 24-hour storm event because the storage required (1374 cu-ft) never exceeded the storage provided based on the stated design volume (3840 cu-ft).
This project presents the procedures used in the sensitivity and uncertainty analyses performed for the Upatoi Creek Watershed (UCW) model in Fort Benning, Georgia, along with the analyses results. For the sensitivity analyses, a sensitivity factor was calculated as the ratio of the percent change in model output to the percent change in input/parameter value (expressed as a percentage). These sensitivity factors allowed the input and parameters to be ranked in terms of the highest to lowest impacts on model outputs. This ranking method provided the means for selecting the most sensitive inputs and parameters for the subsequent uncertainty analyses. The uncertainty analyses were conducted with a Monte Carlo procedure whereby the most sensitive parameters were assigned probability distributions, random values were drawn from these distributions, the model was run for each parameter selection combination, 1,000 model runs were performed, and the model results were analyzed to produce the outputs with 90 percent confidence bounds to reflect and quantify the model uncertainty for each output variable of interest. Results reflect the following:
1. Model uncertainty increases from median flow rate to and high flow rates and low flow rates (extremes).
2. Model uncertainty increases from median flow volume to high and low flow volumes.
3. Model uncertainty in mean TSS concentration are similar to the uncertainty in mean flow (around 30 percent); however, the uncertainty in daily TSS load is greater (about 40 percent).
4. Uncertainty estimates at Reach 46 (McBride Bridge) and Reach 74 (Upatoi Creek Outlet) are essentially the same; the differences in the Percent Uncertainty between these two sites is considered insignificant.
However, based on the past experiences in other watersheds, the results presented here are reasonable and realistic.