Metodologias de preenchimento de falhas de séries pluviométricas anuais: comparativo entre o método do vetor regional e a regressão linear simples

Abstract

Historical rainfall series are required for Hydrology; however this information often contains periods without data. There are several methods of gap filling, from the less complex (Regional Means) to the more sophisticated (Regional Vector). Previous studies, carried out in the Semi-arid region of Brazil, pointed to the Simple Linear Regression Method as the one that presented the best results. This study aims to verify if the greater complexity of the Regional Vector incurs better results than those of Simple Linear Regression. It was used 16 stations of four cities of the state of Ceará (Canindé, Quixadá, Quixeramobim and Sobral), located in the same climatic region. Of these stations, one was selected which had 20%, 30%, 40% and 50% of its data purposely excluded to simulate gaps. In the calculation of the Regional Vector, the software HYDRACCESS was used, whose results were compared with those obtained by Simple Regression. Despite of slightly better results, the Regional Vetor has a higher degree of complexity and demands higher efforts, with the average difference of deviations between the two methods being only 4%. Therefore, for the region studied, Simple Linear Regression is more advantageous for the simplicity employed.