Reduced system inertia decreases the time available for control actions to prevent the system frequency from violating security limits after a disturbance. Therefore, as system inertia is reduced by the shift toward renewable generation that provides little or no inertia, system operators must deploy faster frequency control actions if they are to preserve security and quality of supply. Realising methods for the fast and accurate detection, localization and sizing of an active power disturbance (e.g., the loss/connection of a large generator or load) will be a crucial enabler for the successful implementation of these faster actions. This paper presents a novel method that can simultaneously estimate the time, size, and location of a disturbance using PMU measurements of the active power output of a limited number of generators and the impedance matrix of the system. The immediate power change at the remote generator terminals is combined with the synchronizing power coefficient matrix in a two-stage process. Stage one uses a decision tree to determine that a disturbance has occurred, which then initiates stage two that estimates the size and location of the disturbance. This is based on the level of agreement between the monitored generators. Case studies and sensitivity analysis for the IEEE 39 bus test network are presented to verify the accuracy of the proposed method for varied levels of measurement noise, impedance matrix errors, and topology errors for various disturbance sizes and locations.