When Wind Alignment Reveals Spatial Structure: A Falsification Study of Airport-Associated NO2 Observability at Gimpo and Jeju, South Korea

Geostationary atmospheric composition observations offer new opportunities to study sub-daily emission-related variability, but source-specific signals can be difficult to separate from heterogeneous spatial backgrounds. We examine whether wind alignment improves the observability of airport-associated tropospheric NO₂ using October 2022 GEMS observations at Gimpo (GMP) and Jeju (CJU) airports in South Korea, combined with official route-specific departure records and ERA5 winds. We compare a conventional 20-km area mean with a wind-aligned downwind–upwind NO₂ contrast and progressively test the result using paired bootstrap comparisons, temporal fixed effects, wind-direction adjustment, rotation-angle placebos, and lag decomposition. At GMP, wind alignment initially produced a much stronger association with preceding departures (β = 0.417 versus 0.073 for area averaging). However, the association was strongly attenuated by date and wind-direction adjustment, and the actual wind orientation was not clearly distinguished from alternative rotated orientations. No comparable enhancement was observed at CJU. These results show that wind alignment can amplify direction-dependent spatial variation that is suppressed by area averaging, but that a stronger activity association does not by itself establish source-specific observability.