Abstract: Random noise attenuation is an important step in seismic data processing；however，most random noise attenuation
methods have some problems such as poor de-noising effect and damaging effective signals to a certain degree. Empirical
Mode Decomposition（EMD）can self-adaptively decompose the signal into multi-scale Intrinsic Mode Function（IFM）and
wavelet transform modulus maxima de-noising method is available for low S/N signals. We construct a new random noise
attenuation algorithm by combining EMD with wavelet transform modulus maxima de-noising method，which can effectively
separate signals form random noise and well improve signal to noise ratio（S/N）of seismic data. The algorithm is applied on
numerical simulation and field data for random noise attenuation. The results reveals that the de-noising effect of new algorithm
is obviously better than that of conventional EMD.

Key words: empirical mode decomposition, wavelet transform modulus maxima, random noise attenuation, signal to noiseratio（S/N）