The massive wastage of water occurs due to irregular heavy rainfall and water released from dams. Many statistical methods are of the previous techniques used to predict water level, which give approximate results. To overcome this disadvantage, gradient descent algorithm has been used. This gives more accurate results and provides higher performance. K-means algorithm is used for clustering, which iteratively assigns each data point to one of the k groups according to the given attribute. The clustered output will be refined for further processing in such a way that the data will be extracted as ordered datasets of year-wise and month-wise data. Clustering accuracy has been improved to 90.22%. Gradient descent algorithm is applied for reducing the error. It also helps in predicting the amount of water to be stored in watershed for future usage. Watershed development appears to be helpful in terms of groundwater recharge, which benefits the farmers. It can also be used for domestic purposes.
Top1. Introduction
Water scarcity is the major issue in today’s world. Massive wastage (J.Nittin Johnson et al, 2013; Yun Hwan Kima et al, 2013; Parneet Kaur et al, 2015) of water occurs due to uneven rainfall which results in water being released from dams. Because of this, districts, towns and villages suffer from water scarcity. Over 95 thousand million cubic feet (TMC) to 110 TMC water is released from dams but not even 4 TMC water is reaching surrounding towns as per research carried out We are pushed to a situation where we have to save water in every possible way. To save water and to avoid wastage of water we propose to develop a watershed in Kumbakonam town situated in Thanjavur district. The town has three water resources such as Vennar, Vettar, and Kaveri. We would like to predict the water level measurement in watershed. A watershed is an area of land that stores rain water into one location such as a stream, lake or wetland. These watersheds supply the water, which can be used for drinking purposes, irrigation besides providing habitation to various plants and animals. To make weather pattern recognition efficiently many artificial neural network and Machine learning model had been used in extreme rainfall conditions (N. Q. Hung et al, 2009;V. K. Somvanshi et al,2006; Teegavarapu, R.S.V,2014). The development of physically based models of frequently needs in-depth knowledge and capability regarding hydrological constraints, reported to be highly inspiring as stated in reference (c. Liu et al, 2006), heavy association of sea surface temperature (SST) anomalies with regional and global climate has also been well documented in several studies (Maria C et al, 2005; Sahai AK et al 2000; French et al,1992).
Spatial interpolation methods as observations from different sites are used in local or global variants of these methods for assessment of missing data and it suggested bias‐correction methods related to those used in climate change studies for correcting missing precipitation estimates provided by the best spatial interpolation method (Hall et al. 1993, N.Q. Hung et al 2009) and (Hsu et al. 1995; Zhan Y, Shen D, 2005; Box, G.E.P. and Jenkins, 1970) have recognized artificial neural network for rainfall- runoff modeling. (Goswami et al 2009) has used ANNs with three layers, namely, input layer, hidden layer and output layer for experimental forecasts of all India Summer Monsoon Rainfall. (French et al, 1992) discussed on rainfall forecasting using neural networks. Here an attempt to represent the rainfall process in terms of a single–hidden layer feed forward Neural Network is made (Kulshrestha et al, 2006)