Quantile Regression-Based Probabilistic Estimation Scheme for Daily and Annual Suspended Sediment Loads

Abstract

Assessing suspended sediment loads in rivers is important since it affects water quality, hydraulic-facility design, and many other sediment-induced problems. Sediment-load estimation heavily depends upon empirical approaches such as a sediment rating curve, which is the empirical relationship between sediment load and river discharge. However, the sediment rating curve is insufficient to describe the inevitable scatter between sediment and discharge. This study aims to develop a probabilistic estimation scheme for daily and annual suspended sediment loads using quantile regression. All recorded daily suspended sediment load and discharge data are employed to construct quantile-dependent sediment rating curves. The empirical probability distribution of daily suspended sediment load is then built by integrating the conditional estimations associated with the corresponding quantiles for a given discharge. The probability distribution of a cumulative sediment load over a longer period can also be derived by the obtained daily sediment-load probability distributions and convolution theorem. The proposed approach is applied to the Laonung station located in southern Taiwan. The results indicate that the proposed approach provides not only the probabilistic description for daily and annual suspended sediment loads, but also the single estimations including the mean, median, and mode of the derived probability distribution. For the 1,110 recorded data of Laonung station during the 1959–2008 period, the proposed mean and median estimation schemes outperform the traditional sediment-rating-curve approach for less mean absolute errors.