Sampling TMDL Implementation Rates and Patterns in the North Central US

Abstract

The U.S. Environmental Protection Agency (EPA) Total Maximum Daily Load (TMDL) program has accounted for the listing of over 40,000 impaired waters nationwide and the development of over 37,000 TMDLs since the program’s creation. Case-specific accounts of implementation are widespread, but the actual rate of implementing TMDLs nationally or regionally has remained virtually unknown because full census and tracking of every implemented practice would be an overwhelming if not impossible task. To gain insights on implementation, EPA’s TMDL Program Results Analysis Project conducted a sample-based analysis of TMDL implementation rates and characteristics in the six EPA Region 5 states (IL, IN, MI, MN, OH and WI). A probabilistic sample was drawn from all TMDLs established through FY2007. Sampled TMDLs were allocated proportionally to states based on each state’s total TMDL production. Regional but not specific state-level statistics were the goal of the study. Subpopulations of interest contrasted older (through FY2003) versus newer (FY2004 – FY2007) TMDLs, and nonpoint-source (NPS)-only TMDLs versus point source (PS)-only and mixed (PS/NPS) TMDLs. The project team extracted information on each of the 138 sample TMDLs and their proposed NPS and PS controls from EPA data systems in advance of working with each state to verify implementation rates and patterns across the Region. Findings demonstrated that, within a +/- 10% margin of error at 90% C.I., an estimated 80.3% of Region 5 TMDLs were at least partially implemented. Full implementation was uncommon. No implementation was observed in approximately 20% of the sample, but the diffuse nature of control practices typical of many TMDLs made complete verification of every practice difficult. Among subpopulations, implementation rates did not differ significantly between older or newer TMDLs, but the mixed TMDLs implementation rate exceeded the NPS-only rate by 16.1%. Implementation plans existed for 79.6% of TMDLs, and NPS-only TMDLs showed more plans than the mixed TMDL subpopulation. TMDLs generated as part of large watershed, multiTMDL efforts comprised 13.2% more of the newer TMDLs subpopulation than the older TMDLs subpopulation. Post-analysis steps included exploring GIS data on the watershed traits of each sampled TMDL for other possible associations with patterns of implementation. Analysis of predominant land cover in the samples’ watersheds did not reveal significant differences among implementation rates associated with urbanized, heavily agricultural, lightly agricultural, and rural non-agricultural watersheds.