2.2.4. Wavelet Analysis

The present study employed the Morlet wavelet which provides a good balance between time and frequency localization [41], especially for geophysical data. Wavelet analysis includes different wavelet functions such as the windowed Fourier transform, wavelet transform, normalization, wavelet power spectrum, etc. The main advantage of the wavelet analysis in comparison with other techniques is that it analyses localized variations of power within a time series. By decomposing a time-series into time-frequency space, one is able to determine the dominant modes of variability and their variation with time [42]. Wavelet transform coherence (WTC) is a good method for analyzing the coherence and phase lag between two time-series as a function of both time and frequency [43]. Therefore, we adopted the Monte Carlo wavelet and coherence analysis to quantify the relationships between climate forcing and the two data sets (rainfall and temperature) recorded at Conakry. More details about wavelets and wavelet coherence and phase are given by Torrence and Compo [42], Grinsted et al. [41] and others.

Basically, from a climatological point of view, the 1961–1990 normal (30 years) was used for this study. The models used thus show complementarity in the sense that the Mann-Kendall test gives information about the trend of the total time series and its significance. In addition, SQ-MK is important to determining both the trend variability in time and the trend change points in the time series. However, it is important to identify the relationship between the studied time-series and climate indices, for that purpose, the MLR and Wavelet are used. But the difference between these two models is that the multi-linear regression (MLR) helps to explain the relationship between one continuous dependent and two or more independent variables. The Wavelet analysis method helps to determine the dominant modes of variability and their variation with time, in addition it helps to quantify the relationships between climate forcings and the two data sets indicating the period when the correlation is significant as well. Furthermore, it also specifies whether the parameters are correlated or not and if so, whether they are in-phase or out-of-phase or if the causal relationship is identified or if there is simultaneity. The results from this methodology are then discussed in the following sections and some figures are plotted according to that done by Bilbao et al. [44].

#### **3. Results**
