Frequency-domain correlation of nonuniform time series: Coherence across distant wind stations

We analyze multi-year wind velocity records from six U.S. airports to examine spatial and temporal correlations across the national airspace system. The velocity time series, characterized by uneven sampling and differing time bases, is evaluated using a frequency-domain approach based on the nonuniform discrete Fourier transform. This Fourier-domain based approach enables direct estimation of cross-covariance and cross-spectral relationships without interpolation or resampling in the time-domain, preserving variance and phase integrity. Applied to airport wind data, we aim to separate and interpret contributions from different temporal scales, including externally forced periodicities, such as the diurnal cycle. The analysis reveals scale-dependent and spatially organized correlations. We demonstrate that coherence in the inertial subrange decays rapidly with distance, while diurnal and annual components persist over much broader separations. The results show the scale-dependent organization of atmospheric variability across geographically distributed sites and highlight the utility of frequency-domain analysis for irregular environmental datasets where uniform sampling cannot be assumed.