This paper addresses the persistent sensing problem of moving ground targets of interest using a group of fixed wing UAVs. Especially, we aim to overcome the challenge of physical obscuration in complex mission environments. To this end, the persistent sensing problem is formulated under an optimal control framework, i.e. deploying and managing UAVs in a way maximising the visibility to the non-cooperative target.The main issue with such a persistent sensing problem is that it generally requires the knowledge of future target positions, which is uncertain. To mitigate this issue, a probabilistic map of the future target position is widely utilised. However, most of the probabilistic models use only limited information of the target. This paper proposes an innovative framework that can make the best use of all available information, not only limited information. For the validation of the feasibility, the performance of the proposed framework is tested in a Manhattan-type controlled urban environment. All the simulation tests use the same framework proposed, but utilise different level of information. The simulation results confirm that the performance of the persistent sensing significantly improves, up to 30%, when incorporating all available target information.