Use of descriptive statistics and SARIMA model in the study of precipitation in the Southeast Region of Mato Grosso.

Abstract

The models of the ARIMA class show great potential in the study of time series of rainfall data. Present in the Southeastern region of the State of Mato Grosso, the municipality of Rondonópolis stands out in the agricultural scenario and is one of the main economic poles in Mato Grosso. Given the importance of the application of good performance models, and the scarcity of works using this method in this region, this work had the objective of obtaining a SARIMA model for the rainfall phenomena between the years 2004 and 2015 in the municipality of Rondonópolis-MT. Descriptive statistical parameters, referring to the data population, were analyzed for the identification of climatic seasons in this period. For that, were used data referring to the period from 2004 to 2015, belonging to INMET (National Meteorological Institute). For the application of the SARIMA model, the methodology and performance indicators proposed in the literature were adopted. The average annual rainfall accumulated in this period (2004 to 2015) was 1330 mm with a standard deviation of 145.6 mm, with rain and dry seasons defined by the October-March and April-September months, respectively. The SARIMA (2,0,0) x (0,1,2) model was obtained as the most adequate, presenting better performance in relation to the data observed.