Análise de precipitação e temperatura do Rio Grande do Sul por meio de séries temporais

Abstract

Brazil is affected by differentiated precipitation regimes, mainly due to its large territorial extension, thus from north to south one can find many climates with different characteristics. Rio Grande do Sul’s economy is based in agriculture (soy, wheat, rice and corn), livestock and industry (of leather and shoes, food, textile, timber, metallurgical and chemical). The ability of predicting future characteristics of the meteorological variables has been the aim of many researches along time, and many methods have been used in the elaboration of climate predictions. In this context, the objective of this work was to identify a model of prediction for the precipitation and temperature of eight regions equally spaced in Rio Grande do Sul. For the identification, the Statistics Methodology for time series, specifically, the Box and Jenkins method was used. The data was provided by the “Britsh Atmospheric Data Centre – BADC” and it refers to monthly values of precipitation and temperature, during the period from 1970 to 2006. Many prediction models were evaluated, resulting in SARIMA prediction models. The methodology Box and Jenkins managed in some months to catch well the standard found in the series in study, however in other months the difference between the reality and the predicted one was quite expressive. It is possible to observe the example of the temperature of the regions RS01, RS06 and RS07, in which the foresight found for the future months, is almost the same as the real values. On the other side, the haste of the regions RS07 and RS08 they presented results not satisfactory for the foresight all that compared with the real values.