Monitorização da qualidade da água de barragens com redes neuronais artificiais

Abstract

The water quality control of dams is fundamental, since these constitute indispensable renewable water resources for a variety of purposes. A feasible low cost approach for monitoring water quality bio-optical quantities is to use satellite spectral measurements that allow to continuously and globally follow the evolution of the water surface spectral reflectance. The present approach takes advantage of the Medium Resolution Imaging Spectrometer (MERIS) measurements, onboard ENVISAT satellite, as well as of the in situ samples of chlorophyll-a, cyanobacteria and total suspended solids (TSS), regularly collected and analysed. The method is applied to Alqueva dam the largest artificial lake in Europe in terms of surface area (250 km2), located in the south of Portugal (Alentejo). It is based on the use of a neural network that, after trained with a proper set of data, is capable of predicting the concentrations of the bio-optical parameters from the atmospherically corrected satellite spectral radiance.