Probing atmospheric turbulence with VLBI residuals: when theory meets observations

Geodetic very long baseline interferometry (VLBI) is affected by stochastic fluctuations in atmospheric water vapor, which degrade delay measurements and limit the precision of derived geodetic parameters. The high quality and data rate of modern VLBI observations allow us to analyze high-rate residuals on two baselines (ONSA13NE–WETTZ13S, 920 km; MACGO12M–WESTFORD, 3138 km) to quantify turbulence-induced variations. Using power spectral density analysis and debiased Whittle maximum likelihood estimation, we estimate two turbulence parameters from 31 VGOS sessions: the cutoff frequency λ (linked to the outer scale of turbulence) and the variance σ ² (reflecting turbulence intensity). Both parameters exhibit clear seasonal patterns consistent with atmospheric dynamics: higher λ in winter indicates stable stratification with smaller eddies, while lower λ in summer reflects buoyancy-driven convection with larger turbulent structures. Cross-validation using co-located Onsala telescopes yields correlations exceeding 0.82 for σ ² and 0.62 for λ. These results provide a first step toward refining stochastic VLBI models and open perspectives for climatological applications through comparisons with satellite-derived atmospheric data.