Finding Water Quality Trend Patterns Using Time Series Clustering: A Case Study
Leijun Huang, Hailin Feng, Ying Le · 2019
The increasing proliferation of online data provides great opportunities for mining useful knowledge that can be used to support decision-making. In this paper, we present our work flow where water quality measurements at numerous monitoring sites are continuously collected from online sources, statistical methods are applied to clean the data and extract the time-variant trends, and cluster analysis is applied to the trends to find common patterns among the sites. Our analysis results in several homogeneous groups of sites for each of the five water quality parameters (pH, CODMn, DO, NH3-N and TP), where sites of the same group share the same trend pattern but are geographically distant, implying a possible correlation between water quality and human activities. This discovery is helpful in identifying the root cause of water pollution when combining with analysis of human activities. Also, it provides support for the municipal governments of the same group to share their water quality control measures and experience.