Research Accomplishment Reports 2010

Modeling for TMDL Development, and Watershed Based Planning, Management and Assessment

D.R. Edwards
Department of Biosystems and Agricultural Engineering

 

Non-Technical Summary

The Clean Water Act (CWA) employs regulatory and nonregulatory tools to reduce direct pollutant discharges into waterways, finance wastewater treatment facilities, and manage polluted runoff. These tools are used to restore and maintain the chemical, physical, and biological integrity of the nation's waters.

Starting in the late 1980s, efforts to address polluted runoff have increased significantly. For "nonpoint" runoff, voluntary programs, including cost-sharing with landowners, have been used as the key tools. Evolution of CWA programs over the last decade has included a shift to more holistic watershed-based strategies, with equal emphasis placed on protecting healthy waters and restoring impaired ones. Involvement of stakeholder groups is another hallmark of this approach.

The CWA Section 303(d) fact sheet indicates a total of 38,698 impaired waters. Due to the immensity of the stream miles, lakes and estuaries involved and the jurisdictional differences within impaired watersheds, tools are needed to better understand the causes and processes that can be used to restore and protect these water bodies. Total Maximum Daily Loads (TMDLs) are quantitative objectives and strategies to achieve water quality standards. The water quality standards constitute the goals required to fully support designated uses of streams, lakes, and wetlands. In general terms, the TMDL development process involves assessing the causes and amounts of pollution, identifying the best corrective actions and a monitoring strategy to ensure effectiveness.

There is a need to evaluate existing tools and to develop new ones based on the best science available. This project will develop tools to guide the use of these policies so stakeholders can understand what practices are available and why they should implement them. An important outcome of the project will be increased knowledge of the appropriateness of various TMDL development tools for application in agricultural watersheds. In addition, existing TMDL development tools will be enhanced and some new tools may be developed. This outcome will improve models used for TMDL development.

Another important outcome of the projects is improved software interfaces. This outcome will employ advances in information technology to allow data to be entered more easily in models and to aid in the interpretation of results. Another outcome of the project is the collection of data for TMDL model evaluation and for BMP effectiveness assessment. Available data will be utilized where possible, but some additional data collection will be required.

The overall outcome of the project will be the evaluation and development of watershed models, economic, and social analysis tools that can be used for TMDL development and implementation in agricultural watersheds. Project accomplishment will ensure that techniques used for TMDL development and implementation in agricultural watersheds are based on the best science available and that proposed TMDLs are feasible. The ultimate beneficiaries will be the agricultural community, land users, home owners and other stakeholders who will be impacted by the TMDL program.

2010 Project Description

Previously reported work has been extended to quantify and demonstrate the amounts of and interactions between hydrologic/water quality model uncertainty and model parameter uncertainty, both of which are related to Tasks 9 and 10 of Objective 1. These results have been presented and discussed, thus far, at professional seminars and invited presentations. Presentation is scheduled for the next professional meeting with preparation of a peer-reviewed publication to follow.

2010 Impact

In contrast to earlier work, which examined model and parameter uncertainty conditioned on particular existing data sets, recent work has investigated whether it is possible to make inferences under conditions of no data or very limited data. These findings are much more site-specific and context-specific than previous findings, since at least some data (even if from a different site and/or context) are required to extend results to a situation of no data.

The basic conclusion is that uncertainty inferences, similar to accuracy of the models themselves, depend on the similarity between the situation to be modeled and situations that have already been modeled. Furthermore, more analyses of a comparable nature will have to be completed before uncertainty statements prior to data collection can be made with appreciable precision.

The effect of small data sets on uncertainty appears to be highly complex; in the case of very simple models, small data sets enable parameter uncertainty to be determined nearly completely, with little marginal reduction in uncertainty associated with additional data. In the case of relatively complex models, the effect of additional data appears to depend on the model's intrinsic capability to model the situation, with the ultimately more accurate models demonstrating parameter uncertainty reductions but with little (if any) reduction in the cases of less accurate models.

With models being used increasingly in litigation situations, the findings can be of assistance in both selecting models of choice and in interpreting the results, most notably in the context of what can be inferred from the results.