Regresar

Finding teleconnections from decomposed rainfall signals using dynamic harmonic regressions: a Tropical Andean case study

Abstract:

Global climate is a multi-scale system whose subsystems interact complexly. Notably, the Tropical-Andean region has a strong rainfall variability because of the confluence of many global climate processes altered by morphological features. An approach for a synthetical climate description is the use of global indicators and their regional teleconnections. However, typically this is carried out using filters and correlations, which results in seasonal and inter-annual teleconnections information, which are difficult to integrate into a modeling framework. A new methodology, based on rainfall signal extraction using dynamic-harmonic-regressions (DHR) and stochastic-multiple-linear-regressions (SMLR) between rainfall components and global signals for searching intra-annual and inter-annual teleconnections, is proposed. DHR gives non-stationary inter-annual trends and intra-annual quasi-periodic oscillations for monthly rainfall measurements. Time-variable amplitudes of quasi-periodical oscillations are crucial for finding intra-annual teleconnections using SMLR, while trends are better suited for the case of inter-annual ones. The methodology is tested over a Tropical-Andean region in southern Ecuador. The following results were obtained: (1) trans-Niño-Index (TNI) and Tropical-South-Atlantic signals are strongly connected to inter-annual and intra-annual time-scales. (2) However, TNI progressively weakens its relation with intra-annual components; meanwhile, El-Niño-Southern-Oscillation 3 gains ground for such time-scales. (3) Finally, an inter-annual connection with the North-Atlantic-Oscillation (NAO) is revealed. These results are consistent with previous literature, although the TNI and NAO connections are interesting findings, taking into account the differences in the connected scales. These results show the methodology’s capability of unraveling global teleconnections in different space and time scales using attributes embedded in an integral mathematical framework, which could be interesting for other purposes—such as the analysis of climate mechanisms or climate modeling.