Abstract: Although unmanned aerial vehicles （UAVs） have become a critical component of low-altitude communication networks， the non-stationarity of their air-to-ground （A2G） channels remains a significant challenge. Existing research has primarily focused on characterizing non-stationarity induced by high-speed mobility， whereas channel properties during low-speed or hovering states are often overlooked. In this study， we characterized a non-stationarity mechanism that is dominant during the hovering state through field measurements. We found that attitude instability induced by environmental factors such as wind couples with the non-isotropic radiation characteristics of the airborne antenna. Specifically， attitude fluctuations alter the antenna’s spatial orientation， which in turn modulates the directional gain perceived at the receiver. This coupling effect ultimately leads to severe fluctuations in the received power of the line-of-sight （LoS） path. To empirically validate this mechanism， we developed a high-precision A2G channel sounding platform. This platform integrates a self-developed wireless transceiver， a high-precision inertial measurement unit （IMU）， and GPS to enable the synchronous acquisition of wideband channel impulse response （CIR） snapshots， high-frequency UAV attitude data （pitch， roll， yaw）， and precise positioning data. Leveraging this platform， we conducted channel measurement experiments for a UAV hovering at an altitude of 40 m and compared the channel characteristics under two conditions， including calm （stable attitude） and windy （disturbed attitude） atmospheres. The measurement results validate our hypothesis. 1） Attitude variations are minimal in calm conditions， and the A2G channel exhibited high stationarity. 2） In windy conditions， the UAV executed drastic attitude adjustments to maintain its position. This attitude jitter causes the LoS path to rapidly sweep across the antenna’s gain null and steep-gain-slope regions， which induced significant and rapid signal fading. Consequently， the channel stationary time as quantified using the average power delay profile （APDP） correlation method decreased sharply from over 40 ms to as low as 1 ms. Thus， our results confirm that the attitude-antenna coupling effect is the dominant mechanism for channel non-stationarity during UAV hovering. This finding provides critical guidance for the design and optimization of high-reliability A2G communication systems， particularly regarding antenna selection and channel tracking algorithms.