Precipitation pattern in Viçosa-MG: a case study via time series

Abstract

This paper studies monthly precipitation time series in Viçosa MG, Brazil. We aimed to detect serial patterns in precipitation like trend and seasonality and make predictions for the 2019 year. The idea was to understand water shortage events that occur in Viçosa as well as the challenges regarding water supply for human consumption and agricultural production faced by urban and rural citizens. We used time series and SARIMA (1,0,0) x (0,1,1) model approaches selected based on Bayesian and Akaike Information criteria values (BIC and AIC, respectively). In addition, the ARCH (2) model, selected through AIC, was used to fit SARIMA (1,0,0) x (0,1,1) residues with heteroscedasticity. Our results reveal no changes in precipitation for Viçosa, MG, Brazil, with large variations observed only for specific periods.