The periodic time series analysis of different aspects of urban areas is essential owing to rise in population, stressed resources, and lack of technology-based solutions. In this regard, temporal analysis of water quality holds paramount importance, and for this purpose, the data from satellite remote sensing, geographic information system (GIS), and internet of things (IoT) are collected to perform water quality trend analysis. The study area is Rawal Dam, where data are processed to derive water quality parameters (WQPs) and thereafter water quality index (WQI). The monthly, yearly, seasonal, pre- and post-COVID-19 temporal analyses are performed to analyze the trends of different WQPs and overall WQI, using suitable machine learning (ML) models over the last eight years (2013-20). The water quality classification is performed using neural networks (NN) with an accuracy of 80%, and predictions are made using vector auto-regression (VAR) and long short-term memory (LSTM) networks with an average root mean squared error (RMSE) of 25.63 and 2.664, respectively.
TopIntroduction
Water is vital to the survival of life on Earth. Pollution and population growth are causing a shortage of water supply and an alarming decline in water quality. The increasing environmental changes and problems have affected the quality of water available for drinking and other purposes. The poor water quality can directly result in the death of aquatic life and can cause long-lasting and deadly effects on wildlife, birds, plants, and soil. It can be a root cause for many fatal diseases like diarrhea, cancer, etc. in human beings. In short, it has impacted the whole ecosystem.
Pakistan, being an agricultural country, suffers from a reduction in both water supply and water quality. Many reasons are contributing to deteriorating water quality in Pakistan, including overpopulation, deforestation, and industrial waste discharges into water bodies. According to a UN study, only 20% of the population in Pakistan has access to clean drinking water (WHO/UNICEF Joint Monitoring Program (JMP) for Water Supply, 2015). A report by Pakistan Medical Association (PMA) shows that in Pakistan, 30% of all diseases and 40% of all deaths are due to poor water quality, with diarrhea being the leading cause of death in infants and children. (M. K. Daud, et al., 2017).
This raises the question of how water quality can be monitored so that the harmful effects of poor water quality can be avoided. This study performs a periodic analysis and prediction of water quality using multisource data to help take preventative measures for deteriorating water quality.
The water quality depends on more than 35 parameters including temperature, precipitation, pH, total dissolved solids (TDS), Total Suspended Solids (TSS), minerals, etc. The measurement of these parameters for a given water body can be used to calculate and analyze its water quality. The World Health Organization (WHO) provides guidelines on the presence of safe amounts of these parameters in water. If the quantity of these parameters differs from the provided guidelines, the water quality is affected.
A Water Quality Index (WQI) is a number or a class that can be derived from the Water Quality Parameters (WQPs). WQI can serve as a quick reference for determining the level of water quality for a given sample. Several WQIs have been developed over the years to measure water quality under various conditions and for different purposes. (R.M., et al., 2007)
Water quality data can be collected from different sources such as i) On-site data collection using IoT sensors, ii) Geographical Information Systems (GIS), and Remote Sensing (RS) satellites. The data collected on-site using IoT sensors is the most accurate and reliable, however, the data collected is point data and does not provide full coverage of the water body. To increase the data points, a greater number of IoT nodes need to be added, which will make the entire study costlier, and sometimes it is impossible to cover the whole area due to the limitation posed by the water body. The manual laboratory analysis of water quality is similarly hectic and slow, requiring skilled personnel and a variety of chemicals. Other sources of data are also needed to scale the process of collecting WQPs data.
GIS and RS provide an easy way to collect data at various scales and resolutions. Images of Earth are readily available from European Space Agency (ESA) and National Aeronautics and Space Administration (NASA)’s Sentinel and Landsat satellites, respectively. Therefore, large areas of data can be processed at once. Although, this does come at the cost of less accurate data, and time-consuming preprocessing of data to derive the WQPs using indirect methods.
In order to provide a scalable, robust, and fairly accurate system, in the proposed study, the data from all of these sources are analyzed. A time series analysis is performed to analyze the trend of water quality. The first part of the study involves collecting and acquiring data from GIS, RS, and IoT. The Rawal Lake data is collected from Landsat 8 (OLI/TIRS) and Sentinel 2 Level 2A satellites spanning the years 2015 to 2020. Rawal Lake Filtration Plant provided GIS data that included WQPs data from 2013 to 2020. Additionally, IoT data is provided, but only for the year 2019.