In wireless sensor networks (WSNs) enabled by fixed ground infrastructure, sink nodes located at the margin of the coverage area often suffer from bad communication channels. In this paper, we use UAV-carried aerial base station (ABS) to assist ground base station (GBS) for data collection from sink nodes in WSN, and focus on the co-deployment of ABS and GBS from aspects of coverage partition and bandwidth resource allocation, aiming at minimizing the total energy consumption for uplink transmission of all the sensor nodes in WSN. In particular, we first derive the estimated energy consumption with the sink nodes' density instead of their exact positions. Then, to tackle the non-convexity of the original problem, we decomposing it into two subproblems, which are alternately solved to iteratively improve the sum energy consumption. Numerical results show that the proposed design significantly reduces the energy consumption of all sink nodes as compared to benchmark schemes.