Most existing wind-aware UAV routing methods assume a fixed flight altitude from the outset. By doing so, these methods overlook how dramatically wind speed and direction can change with height inside the atmospheric boundary layer (ABL), where even modest shifts in altitude can expose the UAV to very different wind conditions. In this experiment, the first 3D path-planning framework is presented that jointly optimizes horizontal trajectory and altitude by coupling an empirically fitted Hellmann wind profile (α=0.427,R2>0.95) with a 3D-extended CVaR risk metric, solved via a hybrid ACO-PSO optimizer. Ant Colony Optimization explores a discrete 3D grid using altitudeaware heuristics; Particle Swarm Optimization refines continuous waypoints. Validation on a 1.94 km Sundarbans (Bangladesh) logistics mission with 33 obstacles and a DJI FlyCart 30 shows 1.8 % energy savings, 5.9 % CVaR risk reduction, and 1.9 % flighttime improvement over the best fixed-altitude baseline. Critically, under 90° crosswind the adaptive policy selects 0−47 m altitudewell below forecast altitudes up to 180m-because crosswind drag scales quadratically with wind speed (P∝∥W∥2sin2θwind ), inverting conventional altitude-optimization intuitions. The validated Hellmann framework generalises to other coastal and wetland environments.