Hybrid Deterministic-Stochastic Rainfall-Flow Model for Projecting Seasonal and Annual Flows in Sub-Basins of the State of Paraná

Abstract

Estimating and simulating hydroclimatological variables is important for regional water management, since it has the purpose of mitigating socioeconomic and environmental effects. In this context, rainfall-flow models based on seasonal and annual periods can be an important tool for management of regional water resources. The objective of this work was to develop a hybrid deterministic-stochastic rainfall-flow model based on historical hydroclimatological records of the Lower Ivai River and Upper Tibagi River sub-basins, in the state of Paraná, Brazil. Monthly fluviometric and pluviometric data of 10 locations were obtained from the Paraná Waters Institute, and converted into seasonal and annual periods. The concept of equivalent flow was applied to develop a regressive model between the observed and estimated accumulated flows. The results were evaluated by the Shapiro-Wilk (S-W), Komogorov-Smirnov (K-S), Wilcoxon (WCX), and Brown-Forsythe (B-F) tests at 5% level. Significant differences in homogeneity were found for the estimated data series; thus, statistical procedures were carried out for correcting bias through stochastic simulation. The S-W, K-S, WCX, and B-F tests were carried out again, indicating very good results, as the observed and projected flow distributions, means, and variances were very similar. In addition, a graphical analysis showed that the data of observed and projected flows presented the same trend. Thus, the rainfall-flow model proposed showed a simple applicability, presenting no deformities due to local factors, as the sub-basins are in different climatic regions.