This paper presents a new charge similarity approach to accurately detect faulty feeders, phases and sections in resonant grounded power distribution networks during high impedance faults. The proposed fault detection scheme is developed using a charge similarity-based approach where the similarity factor (which is actually a charge integral) is determined by performing the double integral action on the inner product of the zero-sequence and phase currents of the respective feeder. The variation in the similarity factor provides an indication of healthy or faulty feeders. The similarity factor for a feeder having the highest variation identifies it as a faulty feeder and vice-versa. Another important feature of the proposed fault detection scheme is its ability to work with the low sampling frequency (i.e., 2 kHz) by utilizing the existing measurement devices. A distribution network having three feeders with a resonant grounded substation is used to validate the proposed scheme where all analyses are carried out in MATLAB/Simulink. Simulation results show that the proposed scheme is capable of recognizing faulty feeders, phases and sections for fault impedances from 0.1 Ω to 1 MΩ under different fault inception angles (FIAs) scenarios.