Strategic trajectory planning (STP) is critical for improving flight efficiency and ensuring operational safety, particularly in large-scale flight operations. Given the long lead time of STP, accurately analysing wind forecast uncertainty is essential to enhancing the quality of planned trajectories. However, most existing research overlooks the time-variant nature of wind forecast uncertainty. This may lead to significant discrepancies between planned and actual flight trajectories, increasing operational costs and conflict risks. Therefore, this paper proposes a novel bilevel STP framework for large-scale flights that explicitly accounts for time-variant wind forecast uncertainty. The upper-level model optimises trajectories across multiple flights to minimise total flight time, based on the departure times determined by the lower-level model. The lower-level model mitigates potential conflicts by adjusting the departure times according to the trajectories selected by the upper level. To solve this problem efficiently, a time-variant A* algorithm (TVA*) and a multi-objective cooperative co-evolution algorithm (MOCCEA) are developed, supported by static expectation (SE) and dynamic equilibrium grouping (DEG) strategies to accelerate computation. Experimental results confirm that the proposed method yields consistently dominant Pareto fronts, significantly enhancing flight efficiency while ensuring operational safety and fairness.