Author(s): Soo Yeon Choi, Il Won Seo

Linked Author(s): Soo Yeon Choi

Keywords: Bayesian inference; Formal and informal likelihood; Pollutants mixing in rivers; Transient storage model; Uncertainty estimation;

Abstract: The parameter uncertainty of the transient storage model (TSM) should be accompanied when TSM is used for prediction and analysis of the mixing characteristics. In this study, Bayesian inference was employed for the parameter and the resulting prediction uncertainty. Especially, in order to apply proper likelihood for TSM, a new type of likelihood based on curve segmentation and finite mixture distribution was suggested. The properness of the suggested likelihood was estimated by comparing the results of the informal likelihood based on rootmean-
squared-error (RMSE) which has been applied to TSM in the previous studies. As a result, it was found
that the suggested likelihood yielded reasonable uncertainty estimation by showing the better coverage rates of the observation than the RMSE-based likelihood.

DOI: https://doi.org/10.3850/38WC092019-1598

Year: 2019