AN ANALYSIS OF THE BOX-JENKINS AND BOX-TIAO METHODOLOGY FOR SAZONAL PRECIPITATION FORECAST. CASE STUDY: NORTHEAST OF BRAZIL
Climate Classification of Köppen, ARIMA Method, ARIMAX Method.
The objective of this work is to perform a comparative study with adjustments of predictions models by the Box-Jenkins (ARIMA) and Box-Tiao (ARIMAX) method for monthly cumulative precipitation in six cities of the Northeast of Brazil, being chosen according to the classification Weather forecast for Köppen. Having as exogenous variables: sea surface temperatures of the Atlantic and Pacific Ocean. In all the cumulative rainfall series, the presence of the seasonal component was verified, and, due to the constant variance assumption and normality of the data not being met, it was applied in the original series to the Box Cox transformation. Of the final models by ARIMA and ARIMAX, we have that the ARIMAX model showed the best fit to the data under study, presenting lower values for AIC information criteria, mean error and mean square error.