Center for the Analysis and Prediction of River Basin Environmental System (CARES)

Adaptive Implementation of TMDLs:
Interpretation and Application

A Proposal for a Workshop and Monograph

Kenneth H. Reckhow
Center for the Analysis and Prediction of River Basin Environmental Systems
Duke University
Leonard Shabman
Resources for the Future

Objectives
To conduct a workshop and develop a monograph on the interpretation, analytics, and execution of adaptive implementation of a total maximum daily load (TMDL). The purpose of this monograph will be to serve as the scientific basis for the US EPA to develop guidance on analytic strategies and recommendations for research needs in support of adaptive implementation of TMDLs.

Background
A cornerstone of the 2001 NRC review of the USEPA TMDL program is the recommendation for “adaptive implementation” of TMDLs in recognition of the often substantial inaccuracies in the predictions of water quality models used for TMDL development. Further, the “learning while doing” that is an essential feature of adaptive implementation is expected to result in improved management actions when there are multiple stressors on a waterbody, including pollution and pollutants.
In essence, adaptive implementation means that the implementation strategy is designed not only to meet the water quality standard, but also to accommodate post-implementation monitoring that supports continuing model development, refinement of the water quality standard, and improved targeting of management actions over time. One compelling argument in favor of adaptive implementation is the fact that the uncertainty in TMDL model forecasts is not likely to change soon (if at all). There are several reasons for this, but foremost among them are the complexity of ecological systems and the nonlinearities in system response that are unlikely to be well understood (or mathematically represented).

Format
Approximately fifteen distinguished scientists will be invited to participate in a set of coordinated workshops at the Center for the Analysis and Prediction of River Basin Environmental Systems at Duke University. The format will be similar to that used by the NAS National Research Council for its committees (such as the Committee to Assess the Scientific Basis of the Total Maximum Daily Load Approach to Water Pollution Reduction). The initial workshop, of 3 days duration, will begin with a set of presentations and background papers to facilitate discussion; the outcome of this workshop will be a draft outline for the monograph and writing assignments for the workshop participants. Between workshops, drafts of papers will be shared, critiqued, and revised. Approximately 4-5 months later, a second workshop will be held to discuss the draft papers, the overall monograph, and preparation of the final draft. The draft will undergo external peer review, leading to a manuscript suitable for publication. While the time frame for completion of the monograph is ambitious, it reflects the urgent need for guidance on adaptive implementation.

Working in conjunction with the AI Science Panel will be an Adaptive Implementation Review Committee composed of professionals engaged in TMDL assessment and implementation. This committee of approximately fifteen members will include federal, state, and local officials, plus professionals in the private sector, each of whom is actively engaged in the science, policy, engineering, and/or regulatory aspects of TMDLs. The purpose of this committee is to provide critical comment and insight on the practical aspects of the guidance and recommendations resulting from the Science Panel. To facilitate the work of both the AI Science Panel and the AI Review Committee, some members of each group will be asked to participate in the meetings of the other group.

Key issues
Chapter 5 in the NRC report (“Assessing the TMDL Approach to Water Quality Management”) outlines the perspective and recommendations of the NRC committee. This document serves as an excellent starting point for the workshop. Some of the adaptive implementation issues to be addressed by the workshop participants are:

What should be the goal of the adaptive implementation strategy? Specifically, should the implementation strategy be adjusted only to reach a pre-determined water quality criterion or should the water quality criterion itself be adjusted over time? If the latter, how is progress and success in making water quality improvements to be measured?
How should the adaptive implementation strategy be executed? Specifically, how would the optimum combinations of spending on actions for implementation be selected from among the following:
those actions that are most cost effective (based on calculations of pollutant mass delivery reduction to the waterbody or other stressor reduction per unit of measured cost),
- those actions that promise the greatest opportunity for “learning while doing”
- those actions that are organized around an experimental design in the watershed?
What analytic strategies (e.g., Bayesian inference, Kalman filtering) should be considered for integrating TMDL forecasts with post-implementation monitoring to refine the water quality management strategy over time?
- The revised TMDL forecast based on adaptive implementation could reflect monitoring waterbody response (the water quality criterion), actual reduced pollutant loads (thereby substituting measured loads for predicted loads in the model), or experimental work to reduce model parameter error. What should the post-implementation monitoring strategy be?
- Given expected lags in system response to reduced loads, how should post-implementation monitoring data be interpreted and modeled?
- Can software be developed to facilitate the modeling/monitoring integration?
What statistical inference and sampling design procedures should be used to maximize the amount of information that can be extracted from monitoring programs that will likely have limited funding? How might monitoring resources for adaptive implementation be expanded?
What decision making process should be responsible for the execution of an adaptive implementation approach, given the inevitability of decision making under uncertainty? What is the decision authority of the EPA region, the state, and local stakeholders in making the implementation choices over time? What modeling procedures are most likely to produce information that will assist these decision making bodies.