This paper presents an approach to planning long distance trajectories for small unmanned aerial vehicles. The feasibility of such long-distance flights is dependent on exploiting atmospheric energy, which is possible in regions where there is a strong vertical component of wind. Here a treebased approach is used to find a feasible trajectory between the start position and a distant (i.e. beyond simple gliding range) goal. To improve computation time a set of branches is pre-computed from the space of allowable inputs allowing fast expansion of a node. Nodes are selected for expansion based on a weighted random approach. Simulation results show that feasible paths can be found using this method.