Estimativa da precipitação média anual considerando a variabilidade espaço-temporal através do uso da regressão linear múltipla no Estado da Paraíba.

Abstract

The variability of precipitations in space and time is directly related to the hydrological cycle, land use and occupation, and local economy. In this sense, regional models for rainfall may provide relevant information, supporting the decision-making process associated with water resources planning and management. The objective of this work was to calibrate a multiple linear regression equation to relate average annual precipitations to spatial localization of 145 stations placed on three mesoregions of Paraíba State, Brazil, Borborema, Agreste Paraibano, and Mata Paraibana. For the study case, stations with rainfall series with more than 10 years of daily data were considered. The posts that their few flaws in their series had the data filled in by the IDW deterministic method. The results were moderate, with a determination coefficient of 0.62 for the studied mesoregions. The average percentage errors of the estimate showed that most of the analyzed surface is classified with errors between 8.3% and 22,8%. Using the multiple regression statistical tool, it was possible to obtain a linear equation to estimate the average annual precipitation in three mesoregions of the Paraíba State.