Due to the continuous influence of crosswind disturbance during the longitudinal landing process of the drone, the distribution of its pose points becomes asymmetric, and the navigation trajectory deviates laterally, making it difficult for traditional single navigation methods to achieve high-precision pose state estimation by mapping the drone"s motion coordinate system, thereby reducing the overall effectiveness of landing monitoring. Propose a longitudinal landing monitoring method for unmanned aerial vehicles against crosswind based on visual inertial integrated navigation. Estimating the relative pose of a drone based on visual navigation principles can determine the lateral offset of the navigation trajectory, thereby moderately offsetting the asymmetric distribution of pose points caused by crosswinds. Map the visually estimated pose to an inertial coordinate system to describe the motion of the drone and construct a kinematic coordinate system for the drone. Integrating the relative pose estimation provided by visual navigation, using observation data from visual and inertial sensors, mapping the UAV motion coordinate system output by the inertial unit to the visually defined navigation area, and completing the construction of the visual inertial integrated navigation model. On this basis, combined with landing trajectory and velocity analysis, the crosswind deflection angle is calculated and the glide point is dynamically adjusted to establish an anti crosswind control system, which offsets the crosswind effect with the longitudinal component of the lift provided by the wing. Research on the motion characteristics of unmanned aerial vehicles (UAVs) for anti crosswind longitudinal landing from five aspects: entry stage, straight-line descent stage, leveling stage, drifting stage, and ground sliding stage. Based on the two main longitudinal landing stages of trajectory descent and trajectory leveling, monitoring is implemented to complete the design of a monitoring method for UAV anti crosswind longitudinal landing under visual inertial combination navigation. The experimental results show that the velocity curve obtained by applying the proposed method under crosswind interference shows a significant downward trend at around 14 minutes, and its overall mean level and fluctuation pattern are basically consistent with the real velocity curve. The average deviation between the landing point coordinates obtained and the real coordinates is only 0.8 meters, and the deviation between the longitudinal roll angle and the actual landing pose does not exceed 6 degrees. It is not affected by crosswinds in velocity monitoring, landing point monitoring, and roll angle monitoring, and the longitudinal landing monitoring effect is good.