Modelos estadísticos para la homogenización de variables meteorológicas monitoreadas por el GEAA

Abstract

Meteorological variables are elements that characterize the state of the weather, which allows us to know the current or future behavior of the atmosphere. The objective of this research work was to homogenize the meteorological variables, for the study the meteorological records of the Alternative Energies and Environment Group (AEEG) were used. These data were analyzed and validated, where the absence of inconsistent data and additional missing data were found. We worked with the 14 meteorological mutable in the 11 stations belonging to the cantons of Chimborazo province, for that the orthogonal regression model of type II was used, employing the climatol library which is integrated into the statistical software R, it was applied the multivariate dendrogram technique, obtaining 3 of the 14 variables, 3 similar groups and in the 11 eleven variables, 2 groups which share similar characteristics, great variability in the information was observed, achieving the homogenization and correction of the time series, presenting histograms of normalized anomalies, in this way the data that exceeded 5 standard deviations were withdrawn, histograms of maximum values of the standard normal homogeneity test (SNHT) were generated, with which the high repetition of low values was found in the temporal series. After the homogenization of the variables was carried out, the quality that it presents was analyzed, thus obtaining stations that present different microclimates. Finally, databases with variables correctly homogenized by intervals of hours were obtained, the comparison of the original series with the series was also carried out. imputed and homogenized by Climatol, with the effect that they conform to its trend line. It is recommended to emphasize research work of this type, given that the country lacks information about homogenization methods and models.