ENSO, Tropical Atlantic variability, and scale-dependent wind complexity in Northeast Brazil: a calendar-preserving multifractal and circulation assessment
Year: 2026
Type: article
Abstract: Hourly wind series combine diurnal periodicity, synoptic variability, and ocean-atmosphere variability, complicating the interpretation of finite-scale multifractal estimates. We analyze twelve automatic stations in Pernambuco, Northeast Brazil, with a regular hourly calendar, non-overlapping monthly sequences, multifractal detrended fluctuation analysis (MF-DFA), and circulation diagnostics. The primary sample contains 1,487 station-months spanning 216 calendar months. An empirical transition near 24 h separates fitted 8–24 h and 24–168 h bands, but does not establish two scale-invariant regimes. Removing a leave-one-year-out diurnal climatology changes median $$\:\text{h}\left(2\right)$$ from 1.083 to 0.943 below 24 h and from 0.635 to 0.817 above it. Periodicity-preserving controls likewise generate apparent short-band scaling. Rolling 720-h spectra describe temporal evolution without entering climate inference. The observed distributions retain 57,575 zeros, and seasonal-hourly summaries reveal substantial spatial heterogeneity. In 150 state-months, direct climate coefficients for surface wind and upper-band scaling remain unresolved under false-discovery-rate control. An 80-month common-support analysis detects tropical Atlantic associations with physical wind summaries, indicating temporal-support sensitivity. Climate-circulation associations found with six-month heteroskedasticity- and autocorrelation-consistent covariance do not survive a fitted long-memory covariance sensitivity, whereas circulation-surface associations are more stable. Exactly 200 shuffled and 200 Fourier phase-randomized simulations per station support dominant linear temporal dependence with metric-specific residual contrasts. A further 200 periodic controls per station isolate daily-cycle contributions. The results support conditional observational associations and finite-scale diagnostics, while explicitly separating periodicity, persistence, and uncertainty from causal mediation or demonstrated long-term wind predictability.
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
SDG: Climate action
Open Access status: hybrid
Estimated APC paid: $2,890
OpenAlex ID: W7237224875
