COPYRIGHT on the article "Use of bayesian networks in the assessment of water quality in an amazon hydrographic basin". DOI: http://doi.org/10.6008/CBPC2179-6858.2022.008.0008
ID Fraction: Collection Year 2022 #439 NFT Stock of CBPC (Brazil)

Abstract: In order to monitor water quality in large areas of the Amazon, modeling water resources is an important instrument in identifying the factors that change the quality in the hydrographic basin. In this work, the objective was to analyze the spatial and seasonal variations of chemical, physical-chemical parameters, and trace elements of the surface water of the Peixe-Boi River basin, using Bayesian Networks. Twenty-eight collection points were selected from the river that make up the basin and twenty-nine parameters of water quality including the chemical elements were analyzed by the technique of optical emission spectrometry with inductively coupled plasma. The cause-and-effect relationships of the selected variables were verified through Bayesian Networks. The results showed that the methodology used to evaluate the surface water of the Peixe-Boi river basin through Bayesian Networks was efficient and showed that only five parameters, conductivity, manganese, salinity, total dissolved solids, and vanadium exerted greater influence on the water quality of the basin and were used in the construction of the model. The graph produced showed three variables considered as parent nodes: seasonality, conductivity, and manganese. It can be concluded that seasonality had no influence on quality parameters, but it did influence the variation of vanadium concentration in the two seasonal seasons. Manganese was the most relevant parameter and the one chosen for the evaluation of the Peixe-Boi river water quality index, with an evaluation considered Excellent in both seasonal periods.