Contribution of cross time-frequency analysis in assessment of possible relationships between large-scale climatic fluctuations and rainfall of northern central Algeria

This paper is proposed for the investigation of possible relationships between the large-scale atmospheric circulation phenomena such as the North Atlantic Oscillation (NAO), Southern Oscillation (SOI), Mediterranean Oscillation (MO), Western Mediterranean Oscillation (WeMO) and rainfall of Sebaou river watershed (Northern central Algeria), covering a period of 39 years at monthly scale. Several time and scale-based methods were used: correlation and spectral analysis (CSA), continuous wavelet transform (CWT), multiresolution wavelet analysis (MRWA), cross wavelet analysis (XWT), wavelet coherence transform (WCT) and cross multiresolution wavelet analysis (CMRWA). The rainfall analysis by CSA and CWT has been clearly demonstrating the dominance of 1 year and 1–3-year modes, which they explain 30 to 51% and 25 to 28% of the variance respectively. However, the indices have shown that inter-annual fluctuations up to long-term explain between 60 and 90%. CWT and MRWA indicated significant fluctuations materialising a dry period more marked between the 1980s and 1990s with strong trend towards drier conditions starting from the 1980s, explained by the decadal components D7 and the approximation A7. In addition to the annual component, the XWT spectrums reveal strong coefficients for the SOI between 1992–2005 and 1986–2000 for the modes of 5–10 years and higher than 10 years respectively and less intense for NAO. The WCT between NAO and rainfall indicated the most significant relationship for 1 year, 1–3 years and 3–5 years approximately from the early 1980s corresponding to the dry period. However, the SOI affects rainfall only locally and with significant values more or less localised in the time-frequency space between MO, WeMO and rainfall, but this influence could be significant for low-frequency events. CWMRA shows that the components of 5–10 years and higher than 10 years are the most effective to represent climate index-rainfall significant relationships, where change in Daubechies wavelet properties can improve the correlation across the scales. Furthermore, has indicated that the short-term processes dominate the relationship index-rainfall, which masks the long-term phenomena whose influence can sometimes be very distant. As such, the rainfall variability of the study area has shown fairly significant links, at least locally with large-scale atmospheric circulation phenomena.