[RETRACTED ARTICLE] ARTIFICIAL NEURAL NETWORKS APPLIED TO FLOOD FORECASTING FOR THE PANTANAL REGION IN MATO GROSSO DO SUL
[RETRACTED ARTICLE] Artificial neural networks applied to flood forecastins for the Pantanal region in Mato Grosso do Sul
DOI:
https://doi.org/10.5016/geociencias.v39i1.13644Abstract
Since 1941, there have been reports of flooding in the city of Aquidauana and Anastácio. In this context, this study aimed at forecast flood the urban area of the city of Aquidauana, using the technique of Artificial Neural Networks (ANNs). The RNA model adopted was the Perceptron Multi Layers (MLP), using data from the quotas of the Aquidauana river and the accumulated precipitation from 1995 to 2014 in points of the river basin in the city of Aquidauana. To reduce the number of meteorological data, was applied to the technical data of the Principal Component Analysis (ACP), with the calculation of the correlation coefficient r. The best results of the correlation coefficients were for one and three days before the forecast, with values equal to 97% and 86%, respectively. The results indicate that the RNAs provide means of detection or risk of flooding in the city of Aquidauana, with a forecast of up to five days before.