Locating the source of water pollution is of great significance to water environment protection. According to the concentration model, the localization problem only related to source location in wireless sensor networks is proposed. The Unscented Kalman Filter (UKF) method is used to solve the problem. In the simulation part, based on the simulation data the localization algorithm is tested. The results show that the estimates of the source location vary with the initial values in numerical calculation, and compared with the UKF and least square localization methods which estimate the parameters including pollution source location, initial diffusion time and mass flow rate, the localization method in this paper has higher accuracy and stability.