Turbulence profiles and GLAO systems: comparative analysis of Paranal and Mauna Kea for wide-field near-infrared imaging

Ground Layer Adaptive Optics (GLAO) performance depends critically on the structure of atmospheric turbulence in the first kilometer above the telescope. We compare Paranal and Mauna Kea using HAWK-I+GRAAL and ULTIMATE-Subaru as representative instruments spanning 7^′ – 14^′ diameter fields, combining analytical modeling with Monte Carlo AO simulations to predict delivered PSF quality at K-band (2.2 μm). The two sites exhibit dramatically different ground layer structure: Mauna Kea confines 57% of turbulence to ≤80 m altitude, while Paranal distributes 76% across 0–1200 m. Monte Carlo simulations predict on-axis FWHM of 104 mas for HAWK-I+GRAAL and 176 mas for ULTIMATE-Subaru under good seeing. On-sky validation using archival HAWK-I+GRAAL K_s-band imaging (N = 369 Gaia-matched stars; ϵ0=0.″66 ) confirms the predicted misconjugation signature: FWHM increases monotonically with field radius at +0.″00494arcmin⁻¹, yielding a ∼11% center-to-corner gradient consistent with simulations. We discover a strong directional asymmetry in the PSF field: the south-east quadrant degrades ∼21× faster than the north-west, driven by frozen-flow transport in the stable SSW ambient wind (218.°5, 4.1 m s⁻¹), demonstrating that wind-vector telemetry is a necessary ingredient for individual-exposure GLAO PSF reconstruction. For compact fields (≲2^′), Paranal’s higher turbulence removal fraction provides superior absolute image quality. For wide-field applications (≳5^′), Mauna Kea’s confined ground layer is decisive: ULTIMATE-Subaru maintains <8% FWHM variation over a 14^′ field. The on-sky results demonstrate that azimuthally symmetric misconjugation models are insufficient for individual-exposure PSF reconstruction, motivating the incorporation of wind-vector telemetry into future GLAO PSF frameworks.