An assessment of the reference watershed approach for TMDLs with biological impairments

Abstract

The Reference Watershed Approach (RWA) is used in TMDL development for biological and other impairments with narrative water quality criteria. The questions addressed in this research include: (1) With the RWA, how do land use classifications from different data sources (Digital Ortho-Quarter Quads (DOQQ) and National Land Cover Dataset (NLCD)) impact simulated pollutant loadings? (2) With the RWA, what is the impact of alternative water quality models (GWLF and SWAT) on pollutant loadings? (3) With the RWA, how sensitive are simulated pollutant loadings to the use of different reference watersheds? Various combinations of land use, water quality model, and reference watershed resulted in pollutant reductions that ranged from −13 to 88%. In a comparison of the use of NLCD versus DOQQ-based land use data with the GWLF model, differences in required reductions ranged from 13 to 74% for three different reference watersheds. Use of alternative water quality models with DOQQ land use data resulted in differences in required sediment reductions from −43 to 8% for the three reference watersheds. Differences in load reduction requirements were also seen when different reference watersheds were used, regardless of the water quality model or the land use source used. With DOQQ land use, required sediment reductions ranged from 18 to 88% and 37 to 82%, for the SWAT and GWLF models, respectively. These results suggest that explicit margins of safety used to account for modeling uncertainties during TMDL development may need to be substantially greater than 20% when using the reference watershed approach.