Muttil, Nitin and Chau, Kwok-wing (2007) Revealing patterns in coastal water quality data using statistical analysis. In: MODSIM 2007 International Congress on Modelling and Simulation, 10-13 December 2007, Christchurch, New Zealand.

Abstract

A major impact of eutrophication is the stimulation of algal growth and the production of harmful algal blooms (HABs). HABs can have profound negative effects on the environment, which include severe dissolved oxygen depletion, fish kills, discoloration of marine water, beach closures, etc. Owing to the extremely complicated ecological processes, previous research efforts indicate that it is far from accurately unravelling the causality and dynamics of HABs and predicting their occurrence with acceptable accuracy and lead-time. Thus, techniques to better understand the complex interrelated processes in the ecosystem are needed. In this study, visual data mining using box plots and multivariate statistical analysis using factor analysis are employed for a spatio-temporal analysis of coastal water quality data from Tolo Harbour, Hong Kong (Figure 1). To study spatial patterns, data from the seven monitoring stations (shown in Figure 1) are used. To study temporal patterns, two datasets from different periods were used: one when the water quality in Tolo Harbour was at its worst (in the late eighties) and the second dataset being more recent, when the water quality had improved significantly. This study demonstrates the use of data mining techniques to exhibit prominent spatial and temporal patterns and to reveal dominant variables governing key ecological processes and thus provide further insight into the HAB dynamics.

Search Google Scholar

Repository staff login