Análise espectral do sinal geofísico - exemplos de aplicação da transformada de ondeleta

Abstract

The study of natural signals demands robust techniquest to allow comprehension of active physical mechanisms. However, many of these signals show non-stationary characteristics, whose average rate varies over time, preventing the applications of classic spectral methods, such as the periodogram. The wavelet transform is a robust tool for non-stationary time series analysis and its use has spread in this field due to its ease of application through subroutines in several softwares. In this work, we present wavelet transform applied to two series of natural signals: number of sunspots and precipitation in Brasilia, Central Brazil. The goal was to identify cycles in both series and relate them to phenomena described in the literature, through the wavelet transform. Results showed a spectral peak at 11 years in Sunspot Series, related to Schwabe's cycle. For the Precipitation Series, a primary spectral peak of 21 years was identified, probably related to the temperature variance in the Pacific Ocean, and secondary cyclicity in 10-16 months. These results confirm the wavelet transform as a robust and satisfactory tool in the spectral study of non-stationary series.