Estimating the connected region is essential for node deployment and trajectory planning in UAV assisted wireless sensor networks (WSNs). However, existing estimation models are constructed for traditional ground based WSNs, which does not consider the impact of UAV mobility in the three-dimensional (3D) space. In this paper, the reason for the failure of traditional models in the 3D space is analyzed firstly and then an estimation model applicable to the 3D space is proposed by introducing the impact of radio irregularity that is common for sensor nodes. Experimental results under different flight modes clearly show that the proposed model can estimate the connected region in the 3D space much more accurately. Although this model gives a somewhat pessimistic estimate, it is effective for almost all directions in the 3D space. Its application in UAV trajectory planning for data collection is also evaluated through simulation, which shows that using the proposed model could reduce the flight time and improve the transmission efficiency simultaneously.