Variabilidade decenal de chuva e vazão e seu impacto no cálculo de vazões de referência para séries históricas de usinas hidrelétricas do centro, sul e sudeste do Brasil

Abstract

The non-stationarity of historical flow series is a reality in most reservoirs in the Midwest, Southeast, and South of the National Interconnected System (SIN). In these Brazilian regions, the break in stationarity is remarkable in the mid-1970s, so that most negative anomalies occur before that decade and most positive anomalies occur after. In 1976, there was a phase shift in the Pacific Decadal Oscillation (PDO) with significant impacts on climatology around the globe. The objective of this work is to analyze the influence of decadal variabilities that are present in the hydrographic basins. Firstly, 6 incremental basins were selected in the Paraná hydrographic region from 1950 to 2020, and the dates of minimum and maximum accumulated anomalies were extracted, which normally occur in the 1970s and 2010s, respectively. Firstly, dividing the flow series between the year 1950 and the date of minimum accumulation and between the dates of minimum and maximum accumulation, it was verified that precipitation between these two periods increases from 13% to 44%, and that for natural flow, this increase is from 30% to 102%. The rainfall and flow curves resemble the variability pattern of the MEIv2 index, which represents El Niño Southern Oscillation (ENSO) events. For the other basins, these minima occur before the aforementioned "Pacific shift". Subsequently, for the same 6 incremental basins, the contribution of precipitation to the increase in flow was isolated, disregarding land use. For this, a methodology that does not involve detailed hydrological modeling is used, but the results indicate that the role of rainfall is predominant with contributions from 68% to 84% for the increase in flows. Next, 50 historical series of daily natural flow from 1931 to 2019 were analyzed, verifying by parametric and non-parametric tests that 98% of the series are non-stationary. Finally, the reference flows Q_(95 ) and Q_(7,10 ), mean, and standard deviation were calculated with the same temporal partition methodology. The average variations of these four parameters, from the first period of predominantly negative anomalies to the second period of predominantly positive anomalies, for the 50 plants were 35%, 17%, 33%, and 36%, respectively. An open-source code for processing and visualizing time series was developed and made available in a repository, representing the past availability of rainfall with a single input of a contour file located anywhere in the Brazilian territory. With a clear decadal variability pattern in such a strategic region for the Brazilian hydroelectric system, the extension of historical series and scientific advances in understanding ocean-atmosphere interactions, new methodologies can be created as a tool also in the scope of water resources planning. Observations of fluctuating trends in flow series and a climate prognosis on the horizon of decades can be evaluated for a better definition of reference flows for water resources management and hydraulic projects.