In this paper, a denoising algorithm based on local mean decomposition is proposed to improve the signal-to-noise ratio (SNR) in Brillouin optical time-domain analysis (BOTDA) sensing systems. First, the signal collected by a BOTDA sensing system is adaptively decomposed into product function (PF) components with real physical meaning. Then, the PF components containing signal energy are reconstructed to get the denoised signal after the distribution of the signal energy on each spatial scale is calculated. To further improve the denoising performance of the algorithm, we introduce a Chebyshev digital band-pass filter to filter and reconstruct the PF components in the frequency domain. The experimental results show that compared with that of the original signal, the SNR of the signal denoised by the algorithm is improved by at least 10 dB, and the algorithm provides a simple and effective denosing scheme for the sensing systems.