Medium-term projection for the National Hydro-Electrical System using wavelets

Abstract The Brazilian energetic matrix is predominantly based on hydroelectric plants and its planning is very sensitive to climate variability in different time scales. Natural Affluent Energy (NAE) is an established planning tool to project different scenarios of possible energy production, especially in an integrated system. This work aims to fill a gap between short-term (seasonal/ interannual) and long-term (climate change) planning scales by realizing NAE medium-term projections for the Brazilian National Interconnected System basins. The historical NAE series provided by the National System Operator was used for the years 1931 to 2014. The series was divided into two periods: from 1931 to 2003 for verification, and from 2004 to 2014 for calibration. The Wavelets Auto-Regressive (WAR) model was applied from low- and medium-frequency bands. The band signal was analyzed and the NAE was projected for the years 2014 to 2024. A relationship of the NAE variability with the Pacific Decadal Oscillation (PDO) climate index and the Atlantic Multidecadal Oscillation (AMO) was verified.