Forecast of Sea Surface Temperature (SST) in Tropical Atlantic with the use Artificial Neural Networks

Abstract Many studies show that the sea surface temperature in the tropical Pacific and Atlantic oceans is the main influencer physical variable climatic conditions in various areas of the globe. Thus, the observations and forecasts of ocean variables such as sea surface temperature, constitute a valuable tool for climate monitoring and better use of natural resources, particularly in regions that are vulnerable to the impacts of climate change, as is the case of Northeast Brazil. The present work aims to implement and evaluate a method that object to use Artificial Neural Networks to predict the sea surface temperature in the Atlantic Ocean Tropical and the meridional gradient of sea surface temperature anomalies in the Tropical Atlantic. The results showed that the correlations, Significant to 95% according to the Student t-test, were better for the FMA and MAM quarters, where the coefficients showed values greater than 0,75 for some regions of the ATN. To estimate the Meridional Gradient, the RNA showed similar performance to persistent. The best results occur when using the average of the TSM from the DJF quarter. For this configuration the correlation coefficient was 0,74.