Modelagem estatística da precipitação mensal e anual e no período seco para o estado de Minas Gerais

Abstract

This study aimed at adjusting statistical linear models for prediction of total mean precipitation associated to monthly (in the wet season), annual and dry season periods, based on geographical coordinates (latitude and longitude) and altitude for the State of Minas Gerais, Brazil. Daily precipitation data from the "Agência Nacional de Águas" (ANA) for 209 pluviometric stations were applied, 197 for modeling adjustment and 12 for final validation. Coefficient of determination adjusted (r2), mean absolute error (%), prediction bias (%) and estimated parameters significance were considered for evaluation of models. The monthly and annual precipitation models presented good statistical validation coefficients, with r2 greater than 0.70, mean error smaller than 10% and bias not significant (< 2% in relation to mean value). However, the dry season model presented an overestimation of precipitation, showing that more variables associated to topographic characteristics would be necessary to produce a more accurate model. Nevertheless, the adjusted models present good conditions for practical applications, forming an important tool for environmental management in the State of Minas Gerais.