Dynamical spatial model of heavy metals in Kendari bay using Bayesian geographical weighted regression

Abstract

Abstract Kendari Bay has been designated as an ecotourism area and as the main route for local trading in Kendari City, Province of Southeast Sulawesi, Indonesia. An earlier study conducted by Armid et al [2] found that water quality in Kendari Bay has been polluted by heavy metals from household and factory wastes. Such metalsare spatially distributed throughout Kendari Bay area, but the main source of contributors to pollutants has not yet been identified clearly. A study on the distribution of heavy metals in the aquatic system of Kendari Bay is imperative to determine the source and status of pollution. This study aims for analyzing the main source of the largest pollutant contributors in Kendari Bay in order to maintain the water quality in this bay. The model for analyzing spatial effect is geographical weighted regression (GWR), and Bayesian Markov Chain Monte Carlo (MCMC) is used to estimate GWR parameters. The data of this study originated from 32 sampling sites spread across the Kendari Bay area, referring to a previous study by Armid et al [2]. Based on these data, numerical simulation results were obtained with prior r = 35 and δ = 10 which produced the best BGWR model with the highest R 2 value of 86.75% and the lowest MSE value of 0.02290; suggesting that 86.75% of the pollutants are caused by heavy metals Pb, Cd, and Cr, while 13.25% is caused by other factors. There are two sampling sites that have significant effects on the number of pollutants in Kendari Bay, both site 3 (downstream of the Wanggu River) and site 29 (Port area).