Smart survey on recent trends in water level, drought and water quality analysis system

Over 200 million yearly reports of diseases identified with scarce water and sanitation conditions, 5-10 million deaths occurred worldwide. Water quality checking has subsequently gotten important to supply clean and safe water. This survey work depicts the fundamental explanation behind the requirement for robust and productive Water level, Drought, and water quality control in the level framework, which will keep human assets healthy, sustainable and diminish water use for household purposes. Climate change and variability have so many significant impacts caused by the natural environment’s water system. Incredible methods, collection of water samples are tested alone and analyzed in water laboratories. However, it is not always easy to capture, analyze, and rapidly disseminate information to relevant users to make timely and well-informed decisions. The review work encompasses traditional methods based on Machine Learning (ML), and Deep Learning (DL) approaches.


Introduction
Due to population growth and economic growth, India faces a significant natural resource problem in water [Ranjan et al. 2020]. Most of the water levels have been contaminated due to excess pollutants, which are mostly human-made. Thus certifying water purity is a significant challenge. With rapid industrialization and the latest technology, greater emphasis on agricultural development, and fertilizers and pesticides, large-scale pollution occurs in the aquatic environment that drives the decline in marine life water quality [4]. Water bodies discharge from sewage and discharge industries due to run-off agricultural sectors, run-off urban areas, floods, droughts including pollution point and non-source, pollution, and lack of educational awareness facility among users.

Figure 1. Sample of Polluted Water
User involvement in observing aspects such as sanitation, environmental hygiene, storage, and disposal are distinctive elements for maintaining water bodies' quality. The strength of lakes, rivers, and other water bodies and their biological diversification are directly associated with the health of almost every element of the ecosystem. Waterborne diseases such as water contaminated by environmental factors cause death in the environment, and socio-economic progress is slowly spreading. About 5 million people are said to have died because of waterborne diseases worldwide (Indian Water Resources Information System, 2017). The sample of contaminated water is affected by water quality, as shown in 'figure 1', for six reasons: agriculture, wastewater, and sewage, oil pollution, radioactive waste, urban development, and plastics. User involvement in observing aspects such as sanitation, environmental hygiene, storage, and disposal are distinctive elements for maintaining water quality. The strength of lakes, rivers, and other water bodies and their biological diversification are directly associated with the health of almost every element of the ecosystem. Waterborne diseases such as water contaminated by environmental factors cause death in the environment, and socio-economic progress is slowly spreading. About 5 million people are said to have died because of waterborne diseases worldwide (Indian Water Resources Information System, 2017). The sample of contaminated water is affected by water quality, as shown in 'figure 1', for six reasons: agriculture, wastewater, and sewage, oil pollution, radioactive waste, urban development, and plastics. The primary point of the structure taking care of consistent water quality is to evaluate physical, material properties, and earth quality standards, with a definite objective to identify the water parameters' types and give an early warning of the dangers. The structure additionally provides a consistent examination of the accumulated information and recommends reasonable medical measures when water pollution slows down. Ideal water ranges following WHO standards are listed in Table 1. This work's objectives were to survey the activities and quality monitoring systems at the smart water level regarding the application using communication technology and sensors based on a different machine and deep learning methods. The task considers a facility to inform users and officials concerned over the water level's variability.

Literature survey
Various technical works on the Assessment of water quality worldwide have presented at the research level from which this survey work referred many papers for study.

Manual water quality monitoring
Water plays an essential role in all aspects of our lives, and its quality is deteriorating with everincreasing pollution due to urbanization, industrialization, and population growth. To sustain the quality of life, it is imperative to detect water pollutants, causing contamination. Typically, water quality detection is time-consuming and cumbersome, requiring manual lab analysis and statistical inferences (Gazzaz et al., .2012). Daud et al. (2017), in a research study, collected various water samples and tested for different parameters that were compared against the World Health Organization (WHO) standards and National Environmental Quality Standards (NEQS).

Machine learning classification methods based on water quality monitoring
Classification is a major problem in machine learning and data mining. It is widely used in many realworld applications. To build an assortment, a user first collects a set of tutorials on training examples/instances of pre-classes. A classification algorithm then uses specialized classification phase training data to assign limited classes to test cases (for evaluation) or future events (application). This section discusses some machine learning classification methods.  2018) led an examination to decide the water quality in the repository. Four notable artificial insight strategies, Artificial Neural Networks (ANN), support vector machines, classification and regression trees, and linear regression were utilized to dissect reference situations and sets. A simple to-utilize interface was built up that incorporates a metaheuristic regression model to assess prescient execution and contrast it with those of the two constituent situations.
The ANN model was more precise than the other extraordinary models, sets, and meta-heuristic regression hybrids. (Chou et al., 2018), proposed another anomaly recognition algorithm for water quality information utilizing double development windows after continuously recognizing chronicled design oddity information. The algorithm depends on measurable models, autoregressive straight mix model. The algorithm has tried utilizing water quality PH information three months from a genuine water quality checking station on a waterway framework.

Deep leaning methods based water level, drought and water quality monitoring
Deep Learning is a subfield of Machine Learning where a since quite a long-known algorithm a Neural Network (NN) or an Artificial Neural Network (ANN) or a DeepNeural Network (DNN) or a Feed-Forward Neural Network (FFNN), alluding to a similar algorithm, is utilized to plan highlights into input, hidden and output layer. Framed by an input layer, hidden layers, and an output layer, ANNs present an effective method to learn straight and non-direct connections among input and output sets appeared in 'figure 2'. This section discusses the operation of different deep learning methods based water quality analysis system.    [28]. Nasser, Ali, Karim, and Belhaouari (2013) planned the actual use of energy efficiency, self-configuration, and reusable based on the WSN-water quality observation framework. The ZigBee sensor hub's proposed system that opens the device includes (Squidbee), a server farm associated with the Internet, and includes a data entry channel and a fixed client station for network workers. Lambrou (2014) planned and established a minimal workload in-pipe sensor network architecture to observe drinking water quality at the user's destination. The majority of pre-existing techniques have certain conditions and issues because it points out some of them are ignored:

Gaps in literature
1. Water level, Drought, and water quality recognition using feature selection techniques integration can further increase accuracy. 2. Most of the existing techniques should be as far as the essential aspects of Water level, Drought, and water quality are concerned. 3. The integration of feature selection techniques and classification mechanisms has been neglected to improve the accuracy ratio to recognize Water level, Drought, and water quality.