Smart Data Analysis for Water Quality in Catchment Area Monitoring

A water supply system that integrates catchment area monitoring, treatment and distribution are essential infrastructure both in urban and rural places all around the world. The water quality is becoming a key factor to evaluate life quality for most of the people. However, water quality is facing more and more challenges from industrial, agriculture and social pollution. Traditional water quality control research mostly focused on separate aspects, such as different types of physical, chemical or biological indicators. But their work lacks forward-looking for water quality estimation. This can bring huge influence for practical system on people's health. In this paper, we build a smart data analysis method to analyze and estimate the water quality, considering all the water quality standard indicators in a comprehensive environment. Considering as the water source, catchment area monitoring is the origins of water supply system, we select this area as our first stage for the data analysis process. We take an application from the water supply system in Oslo, Norway. Our method also provides Zero-Inflated Poisson Regression (ZIP) and Zero-inflated Negative Binomial Regression (ZINB) models to estimate biological water quality indicators based on the real historical data. For the results, we compare different influence factors, and show that our models can accurately estimate the biological indicators evolution process, especially for the trend. In addition, we also share interesting findings from the data analysis process, as well as providing future control decision support in water supply system.