Some conceptions about sediment rating equations.

Abstract

This paper examines the anomalies in the sedimentation rating equations in terms of existing notions of statistical analysis and concludes that c=aqb is the preferred relationship. The method of estimating parameters a and b through ordinary least squares and the method of prediction using the log normal probability distribution of the error component z (c=aqbz) is presented using data for the Mathare river at Kabete, Kenya. The need for nonlinear least squares for estimation of parameters a and b is discussed in relation to the additive nature of the error component (c=aqb + z) n the nonlinear form of rating equation. The equation c=Faqb (where F is a correction factor arising due to log normal distribution of the error term z) predicted the sediment yield of the Mathare catchment quite well for 61 days during the long rainy season of 1991.