COMPORTAMENTO DA PRECIPITAÇÃO PLUVIOMÉTRICA NO ESTADO DE SÃO PAULO NOS ANOS DE 2014 A 2018: ANÁLISES GEOESTATÍSTICAS ATRAVÉS DO MÉTODO DE KRIGING

Abstract

The present work refers to the study of the annual rainfall behavior in an area of the São Paulo coast delimited by the Baixada Santista and Litoral Norte Water Resources Management Units. Geostatistical analyses were performed using the simple kriging interpolation method for data from 2014 to 2018, using the ArcGIS 10.6.1 software. Through the chosen software, it was possible to generate histograms, which provide several statistical information regarding the mean, maximum and minimum values, standard deviation, asymmetry, among other parameters that help identify the sample quality used. Through the Geostatistical Analyst tool provided by ArcGIS 10.6.1, semivariograms were generated, which help in understanding the spatial autocorrelation between the distributed rainfall stations, demanding the choice of the model that generates the best results for each of the years studied. The quality verification of the model to be adopted was elaborated through the study of the comparative statistical analysis tables between the measured and predicted values, as well as through the scatter and error graphs, constructed based on the predicted values by the generated semivariogram. By defining the semivariograms and their respective adopted models, it was possible to perform the interpolation of the data through simple kriging, generating sample surfaces that spatialize annual rainfall indices for the entire studied area, in each of the 5 years studied. Finally, the analysis and comparison of the resulting spatializations allowed the understanding of local rainfall estimates and the identification of an existing tendency for a standard rainfall behavior in the study area.