Filling and validating rainfall data based on statistical techniques and artificial intelligence

The study of the hydric regime rainfall helps in management analysis and decision-making hydrographic basins, but a fundamental condition is need for continuous time series data. Therefore, this compared gap filling methods precipitation data validated them using robust statistical techniques. Precipitation from municipality Itirapina, which has four monitoring stations, were used. Four techniques used, namely: normal ratio method, inverse distance weighting, multiple regression artificial neural networks, period 1979 to 1989. For validation performance evaluation, coefficient determination (R²), mean absolute error (MAE), squared (RMSE), Nash-Sutcliffe (Nash), agreement index (D), confidence used (C) through non-parametric with Mann-Witney Kruskal-Wallis test. Excellent performances real verified comparison estimated data, values above 0.8 (R²) Nash. Mann-Whitney tests not significant Stations C1 C2, demonstrating that there difference between proposed methods. It was concluded network showed best performance. From study, efficient tools found fill gap, thus promoting better operation water resources.
 Keywords: regression, method.