COMPARATIVE ANALYSIS OF MODELS FOR TIMES TO SERIES MODELING AND FORECASTING OF SCHEME OF AVERAGE MONTHLY STREAMFLOW OF THE DOCE RIVER, COLATINA ESPIRITO SANTO, BRAZIL

Wanderson de Paula Pinto, Gemael Barbosa Lima, Juliano Brás Zanetti

Abstract


http://dx.doi.org/10.5902/2179460X17143

This work shows a time series study of the average monthly flow using stochastic model of the SARIMA class (Seasonal Autoregressive Moving Average). The hydrology model is an important tool in the planning and management of water resources programs. This article aimed to analyse, comparatively, time series models, as well as to realize forecasting of average monthly flow taken Doce River as watershed, it’s situated in Colatina municipality/ES. The applied methodology in this work was based in the proposal by Box and Jenkins. The identification of the order of the models was made by graphic analysis of the autocorrelation function (ACF) e partial autocorrelation (PACF). Among possible models was selected those that obtained minor value of the following parameters: AIC, BIC, EQM and EAM. The chosen model was SARIMA(1,1,1)(1,1,2)12, that represented well the dynamic of the series studied and showed the best numerical results in both calibration and validation stages. The chosen model was used to make forecasting to months of 2014 and 2015. The average monthly flow forecasting showed, obtained with the considered models, can be used to analyse hydrology regime of the Doce river, mainly to the critical months of 2015, as well as it can propose preventive measures to reduce both environmental and economic impacts in this municipality.


Keywords


Recursos hídricos, ARIMA, sazonalidade.



DOI: https://doi.org/10.5902/2179460X17143

Copyright (c)



Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.