HydraSave : A Unique River Water Quality Forecasting and Pollution Control System

Technology has contributed a lot to the betterment of the environment, yet there is still a lot to be done in terms of sustaining nature and its resources. Water pollution is one of the most challenging tasks to be harnessed even with present technology. Thus, continuous monitoring and management of water resources is of prime importance. The river and its tributaries are a vital water source for millions of people. Rivers are not only an important source of water but also considered as habitat for animals and plants. Unfortunately, many rivers and streams are significantly polluted all around the world. Therefore, modelling and predicting water quality is of great importance in controlling water pollution. Water is of different quality and hence it can be used for various purposes. The water quality is measured in terms of WQI (Water Quality Index). With the calculated WQI water can be classified into five different grades. This grading allows us to know for what all purposes water of the predicted quality can be used. Even though there are many existing models for predicting water quality, there is still space for improvements as the existing models focuses only on the learning process and also do not perform missing value imputation. To overcome these problems we are proposing a model which is a combination of CNN (Convolutional Neural Network) and Bi-LSTM (Bidirectional Long Short-Term Memory) models for predicting water quality. The preprocessed data is fed into a WQI Generator whose output WQI value is used for predicting WQI for the next year. This predicted WQI is then fed into the LightGBM (Light Gradient-Boosting Machine) machine for grading the water. The dataset considered for HydraSave is from the water quality data collected from the Kerala State Pollution Control Board (KSPCB). HydraSave can be considered as a supporting hand for the pollution control campaigns conducted by the Government of India. Early prediction allows us to take the necessary steps to prevent water from getting highly polluted. Hence, our proposed model acts as a unique river water quality forecasting and pollution control system.