Artificial neural networks have progressed in a rapid way in the field of soft computing, and it is widely used in forecasting. The work presented in this chapter is about the development of artificial neural network (ANN)-based models to forecast the water quality (WQ) in Nainital Lake, Uttarakhand, India. A dataset comprising pH, turbidity, and total dissolved solid (TDS) of time period 2018-2019 has been used and analyzed using MATLAB software. For experimentation purposes, four data partition strategies, 10 learning algorithms of back propagation neural network (BPNN), and different combinations of learning rates and training tolerance were evaluated. The performance of the model was evaluated using statistical methods such as MSE, RMSE, MAD, MAPE. The results of the experiment show the capability of the optimal ANN models to predict the WQ of Nainital Lake.
Top1. Introduction
Water is a crucial resource on the earth for the survival of living organisms. Besides it, water played a crucial role for several industries, irrigation, or many other fields. Worldwide approximately, 70% of water is utilized for farming and irrigation purposes, and 10% used for domestic purposes. The water level is decreasing very rapidly due to a speedy increase in population, which creates a worse situation in the future. Several scientific studies have reported that 80% of illnesses are based on poor water quality and sanitation conditions in developing countries (Kofi, 2003). Currently, water quality is a matter of serious concern because many water-borne diseases like cholera, diarrhea, dysentery, etc., affect a large population especially Rural area. India is one of the countries in the world most water-challenged developing. Uttarakhand is a Himalayan state and consists of Garhwal and Kumaon regions. Uttarakhand has special importance in Indian culture and traditions because it has the origin of several holy rivers, springs and lakes. Ganga and Yamuna both have their origin points reckoned to be sacred most in the country. Yamuna, Bhagirathi, and many other tributaries and distributaries (Semwal & Akolkar, 2006). Rivers, streams, springs, and lakes are the primary sources of water in the Kumaun division. More than 50% of the total population depends on these resources for their daily need of water (Singh & Rawat, 1985). Nainital district is called the Lake District of India, and the center of natural beauty in the Kumaun region. Therefore, it is a favorite tourist place of Uttarakhand and consists of seven lakes namely Nainital Lake, Sattal Lake, Sariyatal Lake, Khurpatal Lake, Naukuchiatal Lake, Bhimtal Lake, and Kamaltal Lake.
Nainital Lake is also known as Naini Lake and is the only source of drinking water supply for the population. But nowadays, the water of the Nainital lake is being polluted continuously due to many factors such as disposal of solid waste, anthropogenic activities and sewage effluents, etc. These factors affect the physicochemical and biological behavior of water, which affects the water quality of lakes (Alam et al., 2007; M. Alam et al., 2007; Chandra et al., 2006; Sharma & Kansal, 2011). Due to the high load of tourism activities and increasing population day by day, Lake is losing its rejuvenation and the self-purification capability and its water is becoming not suitable for domestic as well as agricultural purposes. Therefore, the need to monitor and maintain the quality of water becomes necessary for the safety of human beings.
The conventional method of water quality monitoring techniques consumes a lot of time and labor (Korostynska et al., 2013). So, there was a need for a technology-based water quality monitoring system for Nainital Lakes.
1.1 Objective
The research work focuses on the development of an optimal Artificial Neural Network model to predict the water quality of Nainital Lake which will work in an efficient and in accurate manner.
2. Literature Review
This research paper investigates the use of the Artificial Neural Network method to estimate water quality. ANN is quite popular due to its ability to model non-linear patterns, its self-adjusting nature and to provide an accurate result. Nowadays, ANN has been applied in a variety of prediction applications. Several parameters of water quality have been modelled using Artificial Neural Network, such as pH, temp, Total Dissolved Solids, turbidity, BOD by the researchers.
Seo and colleagues’ (2016) proposed a technique to minimize the influence of components such as trend, periodicity and stochastic by developing the Ensemble ANN model with stratified sampling technique. They used 8 parameters temp, DO, pH, Electric Conductivity, Tn, TP and Chlorophyll. The result stated that 7 parameters have higher value of R Squared than 0.85 and 5 parameters have higher value of RMSE than 1.0.
Sarkar and Pandey (2015) present the use of Artificial Neural Network technique to estimate the concentration of Dissolved Oxygen at the downstream of Mathura City, India. They applied feed forward error back propagation Neural Network technique to develop three models by taking different combinations of input variables and input stations. They used monthly datasets of 6 parameters such as flow discharge, temp, pH, biochemical oxygen demand (BOD) and dissolved oxygen. Statistical tool has been used to evaluate the performance of ANN technique. The result stated that the predicted value of DO show outstanding accuracy by giving high correlation (up to 0.9) between measured and predicted value.