Abstract
This study aims to explore the remote climate influences, through patterns of climate variability, and regional, through precipitation and streamflow in contribution of basins, given the meaning of river flow variations knowledge to the planning of water multiple uses. Therefore, empirical models of monthly streamflow in the lower of basin the Parana River at the site Itaipu Dam were developed. These models were tested with different groups of predictors: only the lagged climate indices; only precipitation in homogeneous rainfall regions; only streamflow at upstream points and even the streamflow of Itaipu; and the set of predictors. Through stepwise method the most significant predictors were chosen, being highlighted the best predictors: the indices of the El Niño Southern Oscillation and temperature anomalies of the sea surface in the tropical South Atlantic; rainfall in southern basin sites; and proper lagged streamflow in Itaipu. The models were validated, indicating generally higher performance in shorter lags, when considered predictors of streamflow and precipitation, in the longer lags, it was found that the best performance is for the model considering only the climate indices. Therefore, the results of this study demonstrate the importance of considering the remote climate influences on streamflow estimates, especially for longer-term forecasts.
Keywords:
multiple linear regression; streamflow forecasting; climate patterns
