Advance of Big Data in Water Quality Monitoring

  • Yiheng Chen

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)

Abstract

The automated measurement of water quality is usually more difficult than the measurement of water quantity. Due to this nature of water quality measurement, the amount of water quality data is much lower than that of water quantity data. While we are on the era of the technology boom, this thesis focuses on the water quality data collection and analysis using the latest technology and tools such as in-situ sensors, Internet of Things, artificial intelligence, hyperspectral camera, aiming to bring up the water quality monitoring and modelling to the big data era.
The beginning of the thesis provides a review on big data with its relevance to hydroinformatics which is the first in the field. An exploration of precipitation big data is discussed as well. At the same time, the water quality data is found lagging behind the water quantity data in terms of data availability. It helps decide the main theme of the thesis to focus on novel methods to collect water quality data.
Bristol Floating Harbour keeps the water level constant along the world's second-largest tide in the Bristol Channel. The first smart water quality monitoring system is developed with the advance in in-site water quality sensors and the Internet of Things based on Bristol Is Open infrastructure. The high-frequency water quality data shows much more details of the water quality variation than the existing weekly sampling scheme. The diurnal variation of water quality and its response to the operating of the floating harbour are observed only by the new system. A video camera is also used in the system to capture the image of the water surface which is then used to estimate water quality through the ANN model.
The application of the visible and near-infrared hyperspectral camera on water quality is explored. A simple but robust model is found between the reflectance spectrum and turbidity. An unmanned surface vehicle is designed and crafted to mount the in-situ water quality sonde and hyperspectral camera simultaneously as the first one of the kind to carry out water quality survey in the floating harbour.
The last research question is the feasibility to simulate and forecast the water quality in Bristol Floating Harbour. A water quality model using meteorology inputs with artificial neural networks for simulation and forecasting is developed. A long short-term memory (LSTM) model is used to estimate the water quality. The hourly weather data is usually available while the collection of high-frequency water quality data is still challenging. The model helps determine the impact of weather on water quality and demonstrate the possibility to predict the water quality with the help of weather data.
Date of Award25 Jun 2019
Original languageEnglish
Awarding Institution
  • University of Bristol
SupervisorDawei Han (Supervisor)

Cite this

'