An improved electro-thermal model is proposed considering the dependency of parameters not only on temperature and SoC (state of charge), but also on current rate. All the impedance parameters involved in the model are extracted from the direct current internal resistance (DCIR) tests, in which more than four hundred data sets are obtained in order to investigate the dependency of parameters on temperature, SoC, and current comprehensively. All dependency relationships are formulated by a semi-empirical approach based on the derivation of Butler-Volmer equation and Arrhenius empirical equation with other mathematical analysis. Verification results show that the improved model complemented by current dependent parameters can provide good prediction both in voltage and temperature responses for wide ranges of applied current rates and temperatures. Furthermore, in order to extend the engineering application of the proposed model, a nested loop program invoking the improved electro-thermal model is presented to predict the power performance of the battery. The effects of temperature and SoC on the available maximum cell output power are illustrated with a series of simulated contours.