Non point pollution predictions in river system using time series patterns in multi level wavelet-ANN model

Abstract

Herbicides, pesticides, and other chemicals are employed in crop lands to increase the agricultural food productivity. These chemicals increase the concentration of non point pollutant in river systems. Non point pollution affects the health of human and aquatic environment. The transport mechanism of chemical pollutants into river or streams is not straight forward but complex function of applied chemicals and land use patterns in a given river or stream basin which are difficult to quantify accurately. Development of models based on temporal observations may improve understanding the underlying the hydrological processes in such complex transports. Present work utilized temporal patterns extracted from temporal observations using wavelet theory at single as well as multi resolution levels. These patterns are then utilized by an artificial neural network (ANN) based on feed forward backpropogation algorithm. The integrated model, Wavelet-ANN conjunction model, is then utilized to predict the monthly concentration of non point pollution in a river system. The application of the proposed methodology is illustrated with real data to estimate the diffuse pollution concentration in a river system due to application of a typical herbicide, atrazine, in corn fields. The limited performance evaluation of the methodology was found to work better than simple time series.