Water managers and stakeholders usually face uncertainty in water availability due to the challenge of incorporating the dynamic nature of precipitation into the water management system. Surface water rights are commonly related to the baseflow component, which is part of the precipitation incident on a watershed. This study proposes an empirical linear model to predict baseflow in perennial streams based on a moving average of antecedent rainfall data and the basin time response. The short-term responses of three nested basins were estimated using cross Fourier spectral analysis, and the proposed model was applied to two nested basin scales (1,867 and 3,519 km 2 ), located in southeastern Brazil. Results indicate that the aquifer stores the rainfall water with regulation times of approximately 60 days for the fast-subsurface flow and approximately 2-3 years for the slow groundwater flow in both basins. Differences between our model results and monthly 95% exceedance discharge (Q 95 ) were as high as 10 m 3 s −1 between September and November in the largest basin, revealing how conservative Q 95 can be as a criterion for water allocation purposes.Despite the simplicity, our empirical rainfall-based model is structurally consistent and robust in representing the hydrological processes involving precipitation, groundwater storage and baseflow interactions at multiple scales by using few inputs and calibration parameters. Because it considers a range of rainfall periods, from past to present, our model contributes to a dynamic, predictive, and integrated water rights management.
K E Y W O R D Sbaseflow coefficient, spectral analysis, response time, time series analysis