To maximize the amount of data gathering and balance the energy consumption of the nodes, a scheme of mobile data gathering in cluster based wireless sensor networks (WSNs) is considered, where the data rates of the nodes are adjustable. The problem of maximizing the data gathering is formulated as a nonlinear programming, the methods of dual decomposition, layered optimization and sub-gradient descent are employed to derive the optimal allocations of data rates, flow rates and the sojourn time of the mobile sink. The simulation results demonstrate that the method proposed in this paper can obtain a better performance in balancing the energy consumption and maximizing the amount of data gathering.