Bidimensional geostatistics as an alternative for modeling time series under climate change scenarios

Abstract

The objective of this work was to verify the use of semivariogram models and ordinary kriging to characterize and to represent the temporal dependence of the pluvial precipitation and air temperature, by using years and months with coordinates. The observed the forecast data of future scenarios of climatic changes, from 2000 to 2099, of the municipal district of Juiz de Fora – MG, for the pluvial precipitation and air temperature were analyzed. The semivariograms were adjusted by considering isotropy and anisotropy conditions. The wave model presented better adjustment than the Gaussian and spherical models when the two-dimensional grid was used. The structure and the magnitude of temporal dependence of the two variables were characterized by considering the 0º (months) and 90º (years) directions. The pattern of the temporal variability of the precipitation and the air temperature was represented by using kriging. Possible changes in the patterns of the pluvial precipitation and the air temperature for the future scenarios were verified by using a two-dimensional grid and kriging. The greatest changes in the pattern of the two variables were verified in the most pessimist future scenario.