In 1991 the US EPA designated the Carson River as part of a Superfund site
due to its high level of mercury contamination. Water column total mercury
concentrations have been observed as high as 61 mu g l(-1) by the USGS, whi
le some fish have methyl-Hg concentrations four times greater than the huma
n health limit for consumption 1 ppm (1 mu g g(-1)). Hg river bank concentr
ations are extremely high (avg. = 64,242 g Kg(-1)) and more than 95% of the
total Hg in the water column is associated with particles. The importance
of bank erosion on the transport and fate of mercury in this system is ther
efore apparent. Using the US EPA RIVMOD, WASPS and MERC4 numeric codes, thi
s study focuses on program modifications that predict bank erosion rates an
d Hg bank concentrations related to longitudinal slope. Calibration is perf
ormed for both Hg and methyl-Hg water column concentrations using data coll
ected during medium flows. Contribution of MeHg from bank erosion is calibr
ated with data collected during a high flow event. Subsequent verification
is accomplished using data from low and high flow regimes. The model simula
tes inorganic mercury (n = 15) with a root mean squared error (RMSE) of 490
ng l(-1). The average error is 21% with a bias of under prediction by 88 n
g l(-1). Methylmercury concentrations (n = 13) are simulated with a RMSE of
0.90 ng l(-1). This corresponds to an average error of 32% and tendency to
over predict MeHg by 0.14 ng l(-1). Post 1997-flood data, collected during
a low flow regime, indicate that the system may have significantly changed
in terms of mercury concentrations in the channel bottom sediments. (C) 20
00 Elsevier Science B.V. All rights reserved.