A fuzzy neural network approach to quality assessment of water reservoirs

Alton Water, Suffolk, U.K. is a reservoir that has a history of Cyanobacterial blooms, which are single-celled organisms that live in fresh, brackish, and marine water. The blooms of Cyanobacteria are not always helpful and traditional approaches to monitoring are often limited by the need of data collection, which is time consuming, expensive and non-continuous. The main goal of this study is to analyze some environmental parameters in the reservoir in a period of five years, from 2014 to 2018. We aim at forecasting the concentration of some physico-chemical parameters, which are indicators of Cyanobacteria presence, by using satellite imagery reflectance in connection with neural and fuzzy neural regression models. Both Landsat 8 and Sentinel 2 satellite images are used to predict concentrations of Chlorophyll-a, Algae and Turbidity level. The experimental results prove that the predicted values have good accuracy, therefore suggesting that satellite sensors and fuzzy neural networks are suitable tools for realtime monitoring of water reservoirs.