The existing Electromagnetic Time Reversal (EMTR)-based fault location methods generally leverage the features of the signals simulated in the EMTR reversed-time stage (relative to the direct-time stage wherein a fault occurs and its originated transient signals are measured). This paper presents for the first time an integrated time-frequency domain analysis proving a similarity between the signals respectively present in the direct time and in the reversed time. It is demonstrated that the fault current observed at the real fault location exclusively behaves as a quasi-scaled copy of the time-reversed back-injected transient current, such that, the fault location can be identified through a similarity assessment. This paper proposes calculating a cross-correlation metric to quantitatively represent the level of similarity at guessed fault locations. The similarity characteristic and the fault location performance of the time-domain metric are validated first through simulation case studies using the IEEE 34-bus test feeder and then making reference to a pilot test conducted on a real medium voltage distribution network.