To address the issue of low signal-to-noise ratio in disturbance localization on power communication optical cables caused by environmental noise in phase sensitive-optical time domain reflectometry (Φ-OTDR) systems, the Φ-OTDR localization technique is investigated based on moving variance averaging and spectral subtraction. A coherent detection Φ-OTDR system is built to connect optical fibers and optical cables to carry out galloping simulation experiments. The backward Rayleigh scattering signal is acquired by using the system and the scattering signal is demodulated to obtain the amplitude signal curve. By selecting an appropriate number of curves and an optimal step size, the amplitude signal curves are processed using moving variance average to generate a preliminary disturbance localization map. To suppress ambient noise outside the localization region, noise reduction using spectral subtraction is performed to achieve enhanced localization. A comparative analysis is conducted between spectral subtraction and two alternative noise reduction methods-wavelet decomposition and empirical mode decomposition. The results demonstrate that spectral subtraction significantly outperforms the other two algorithms in localization effectiveness, achieving an SNR improvement exceeding 20 dB while maintaining a computational time as low as 0.58 s.