**2.5. Wavelet analysis**

Wavelet analysis has become a widely used method to study variations of energy in environ‐ mental time series [25, 26]. The decomposition of a time series in the time‐frequency space allows the determination of dominant modes of variability and its variation modes in time [27]. Time series for TRMM and MOD16 were analyzed with continuous wavelet transform using the algorithm developed by [28]. Generally, continuous wavelet transform is used to visualize, in a three‐dimensional diagram, the relationship between components of different frequencies according to the time scale of the series studied [29]. Several functions are used to generate wavelets; in this study, the Morlet complex function was used, which is composed of a plane wave modulated by a Gaussian envelope, as shown in Eq. (2):

$$
\Psi(\eta) = \pi^{-\aleph\_4} e^{i\alpha\_0 \eta} e^{-\eta\_2^2} \tag{2}
$$

where *η* is the dimensionless time parameter, and *ω*0 represents the dimensionless frequency. Here, it is important to point out that Torrence and Compo algorithm was compiled in MATLAB version 7.9.0 and that the analysis was performed exclusively for forest remnants of Atlantic Rainforest. Therefore, values used to generate wavelets referred to the monthly average precipitation (TRMM) and evapotranspiration (MOD16) in the analyzed forest remnants.
