Sharma and Dickinson [1980], Fitzgerald and Karlinger [1983], Gurnell and Fenn [1984], Olive and Rieger [1988] and Lemke [1991] found that residuals from rating curves tend to be serially correlated, violating the assumption of independent errors. They present transfer function models that address this concern. Gurnell and Fenn [1984] analyzed hourly suspended sediment concentrations from a proglacial stream (glacier do Tsidjiore Nouve) using several techniques, and found that a transfer function model provided the best fit. Thompson et al. [1987] endorses use of rating curves with autoregressive error components. Lemke [1991] applied a transfer function model in two drainage basins in Iowa.