Semi-arid Ceará (NE Brazil) is highly sensitive to climate variability, yet the non-stationary, multi-scale nature of its hydroclimate hampers traditional analyses and forecasting. To address this complexity, this study aims to characterize the multi-scale variability of rainfall and streamflow and quantify their lagged teleconnections with global climate drivers. Monthly rainfall and streamflow for 1987–2022 were decomposed with Complete Ensemble Empirical. Mode Decomposition with Adaptive Noise (CEEMDAN) to isolate intrinsic mode functions (IMFs) and a residual trend. Teleconnections with 16 climate indices were quantified via cross-correlation (lags 0–12 months) and bidirectional stepwise regression using the top-correlated candidates per IMF; index-selection frequencies were summarized irrespective of lag. Interannual variability (IMFs 3–6) was dominated by ENSO-family indices, especially NINO4/NINO12/ONI and BEST, while low-frequency variability and residual trends were primarily linked to Atlantic modes, with AMO unsmoothed most prominent. Spatially, Pacific influence was stronger at coastal gauges, whereas interior basins were more strongly conditioned by low-frequency Atlantic variability. These scale-aware relationships have operational value: TNA–TSA gradient supports within-season decisions, ENSO indices inform annual planning, and AMO unsmoothed provides multi-year context for carryover storage and drought readiness. The findings provide an organizing principle for forecast-informed water management in Ceará.
Rolim et al. (2026) studied this question.