Harmonic analysis of monthly rainfall for the cities of Pelotas, Rio Grande and Santa Vitória do Palmar, the extreme south of Rio Grande do Sul, Brazil

Abstract

In Rio Grande do Sul (RS), the agricultural sector accounts for a third of gross domestic product (GDP) and its performance shows fluctuations caused by the variability of meteorological elements, being water the key element. In order to contribute to studies and planning policies for the sector and for predicting shortages in water use, a study of the variability of monthly totals of rainfall was performed, from a time series of 60 years of data (1948-2007), for the cities of Pelotas, Rio Grande and Santa Vitória do Palmar, which constitute a homogeneous group regarding the variability of the element studied. Aiming to obtaina model that can represent the behavior of the average monthly rainfall, a harmonic analysis of data previously adjusted to 30-day months was performed, after checking the properties of normality, homogeneity of variance and stationarity. The obtained model is comprised of three sine waves and represents 87% of variability in the data. Stood out the harmonics: semiannual with 64% of representativeness, quarterly, and annual with 12% and 11%, respectively. The models were validated for predicting future values, by means of residual analysis, which showed them to be appropriate at the 5% probability level.