A mixture model approach to analyzing major element chemistry data of the Changjiang (Yangtze River)

In this article we study the statistical distributions of major chemical compositions (HCO3, Ca; charges are neglected for simplicity) and the total dissolved solid (TDS) concentration in the river water of the Changjiang (Yangtze River) of China. We propose a Bayesian finite mixture model with an unknown number of components for the multi-year averages of continuously monitored data over the period 1958-1990 at 191 stations in the drainage basin. A discretization-based Monte Carlo sampling approach is used to estimate the posterior distributions of the parameters in the model. Two sub-populations are identified for the levels of TDS, HCO3 and Ca, and observations from the 191 stations are classified into two groups using the posterior classification probabilities. Copyright (c) 2005 John Wiley & Sons, Ltd.