In this paper, we find trajectory planning and power allocation for a vehicular network in which an unmanned-aerial- vehicle (UAV) is considered as a relay to extend coverage for two disconnected far vehicles. We show that in a two-user network with an amplify-and-forward (AF) relay, non-orthogonal- multiple-access (NOMA) always has better or equal sum-rate performance in comparison to orthogonal-multiple-access (OMA) at high signal-to-noise-ratio (SNR) regime. However, for the cases where i) base station (BS)-to-relay link is weak, or ii) two users have similar links, or iii) BS-to-relay link is similar to relay-to-weak user link, applying NOMA has negligible sum-rate gain. Hence, due to the complexity of successive-interference- cancellation (SIC) decoding in NOMA, we propose a dynamic NOMA/OMA scheme in which the OMA mode is selected for transmission when applying NOMA has only negligible gain. Further, we formulate an optimization problem that maximizes the sum-rate of the two vehicles. This problem is non-convex, and hence we propose an iterative algorithm based on alternating- optimization (AO) method which solves trajectory and power allocation sub-problems by successive-convex-approximation (SCA) and difference-of-convex (DC) methods, respectively. Finally, the above-mentioned performance is confirmed by simulations.