For large energy storage in microgrids, vanadium redox flow batteries (VRBs) have received much attention in recent years. VRBs are promising due to the design flexibility, low manufacturing cost for large scale, indefinite lifetime and recyclable electrolytes. VRB modeling is the prerequisite for battery analysis. In previous studies, the effect of temperature on the battery performance is neglected for simplicity. This temperature independent model may lead to large modeling errors when the surrounding temperature varies in a wide range. In this paper, a thermal-dependent electrical circuit model is proposed to describe the charge/discharge characteristics of VRB. The model is validated by the experimental data. State of charge (SOC) estimation is another key problem in management of VRB since an accurate estimation method can well prevent the over-charge/discharge of battery. Therefore, it is necessary to explore an appropriate method for this novel flow battery. Extended kalman filter (EKF) is implemented in this model to achieve a robust SOC estimation. Simulation results show that EKF based estimator is accurate in SOC prediction with temperature variations. The process and measurement errors are minimized by EKF. The estimation of SOC and temperature facilitates optimal battery operation and thermal management.