Water Quality in the Upper Anacostia River, Maryland: Continuous and Discrete Monitoring with Simulations to Estimate Concentrations and Yields, 2003–05

By Cherie V. Miller, Angélica L. Gutiérrez-Magness, Brenda L. Feit Majedi, and Gregory D. Foster

Prepared in cooperation with
Prince George’s County Department of Environmental Resources,
the Maryland Department of the Environment,
the U.S. Environmental Protection Agency, and George Mason University

Abstract

From 2003 through 2005, continuous and discrete waterquality data were collected at two stations on the Anacostia River in Maryland: Northeast Branch at Riverdale, Maryland (U.S. Geological Survey Station 01649500) and Northwest Branch near Hyattsville, Maryland (Station 01651000). Both stations are above the heads of tide for the river, and measurements approximately represent contributions of chemicals from the nontidal watersheds in the Anacostia River. This study was a cooperative effort between the U.S. Geological Survey, the Prince George’s County Department of Environmental Resources, the Maryland Department of the Environment, the U.S. Environmental Protection Agency, and George Mason University. Samples were collected for suspended sediment, nutrients, and trace metals; data were used to calculate loads of selected chemical parameters, and to evaluate the sources and transport processes of contaminants. Enrichment factors were calculated for some trace metals and used to interpret patterns of occurrence over different flow regimes. Some metals, such as cadmium, lead, and zinc, were slightly enriched as compared to global averages for shales; overall, median values of enrichment factors for all metals were approximately 15 to 35.

Stepwise linear regression models were developed on log-transformed concentrations to estimate the concentrations of suspended sediment, total nitrogen, and total phosphorus from continuous data of discharge and turbidity. The use of multiple explanatory variables improved the predictions over traditional rating curves that use only streamflow as the explanatory variable, because other variables such as turbidity measure the hysteretic effects of fine-grained suspended sediment over storm hydrographs. Estimates of the concentrations of suspended sediment from continuous discharge and turbidity showed coefficients of determination for the predictions (multiple R2) of 0.95 and biases of less than 4 percent. Models to estimate the concentrations of total phosphorus and total nitrogen had lower values of multiple R2 than suspended sediment, but the estimated bias for all the models was similar. The models for total nitrogen and total phosphorus tended to under-predict high concentrations and to over-predict low concentrations as compared to measured values.

Annual yields (loads per square area in kilograms per year per square kilometer) were estimated for suspended sediment, total nitrogen, and total phosphorus using the U.S. Geological Survey models ESTIMATOR and LOADEST. The model LOADEST used hourly time steps and allowed the use of turbidity, which is strongly correlated to concentrations of suspended sediment, as a predictor variable. Annual yields for total nitrogen and total phosphorus were slightly higher but similar to previous estimates for other watersheds of the Chesapeake Bay, but annual yields for suspended sediment were higher by an order of magnitude for the two Anacostia River stations. Annual yields of suspended sediment at the two Anacostia River stations ranged from 131,000 to 248,000 kilograms per year per square kilometer for 2004 and 2005. LOADEST estimates were similar to those determined with ESTIMATOR, but had reduced errors associated with the estimates.