Wavelet Transform and Deep Learning approach to predict physico chemical parameters of water

Abstract

Alton water reservoir located within the stunning countryside of Suffolk in UK has a history of cyanobacterial bloom, that are single-celled organisms that live in fresh, brackish, and marine water. The traditional approaches to monitoring water reservoirs are often limited by the need for data collection which often is time-consuming and expensive. In addition, Chlorophyll-a, algae and turbidity are important variable for the analysis of water quality, that are significant not only for human populations but also essential for plant and animal diversity. The main objective of this study is to predict these chemico physical parameters from 2014 to 2019 using time series analysis, satellite imagery, wavelet transform and deep learning. The satellites images were used to predict these parameters in Alton reservoir, manually selected samples were employed to validate estimated parameters using Wavelet Neural Networks. The results predicted by the neural network show good results, and good approximation to laboratory results, suggesting that the proposed model is suitable for environmental monitoring, contributing to monitor water quality parameters.