Chen, W., Chen, Y. and Liu, W., 2016. Ground roll attenuation using improved complete ensemble empirical mode decomposition. Journal of Seismic Exploration, 25: 485-495. Empirical mode decomposition (EMD) is a fully adaptive signal decomposition algorithm. It has been used to attenuate both random noise and coherent ground roll by removing the first one or two decomposed components in each frequency slice, which acts as a dip filter to separate different wavenumber components. The mode-mixing problem is the biggest drawback of this decomposition technique, which refers to the phenomenon that the each decomposed component is related with multiple oscillating frequencies. Noise-assisted variations of EMDs, like ensemble empirical mode decomposition (EEMD) and complete ensemble empirical mode decomposition (CEEMD), can solve the mode-mixing problem to some extent but will cause other problems, such as strong residual noise and spurious artifacts. The newly developed improved complete ensemble empirical mode decomposition (ICEEMD) are intended to solve the two drawbacks of CEEMD. In this paper, we propose to apply the ICEEMD algorithm for removing the highly oscillating components in seismic data, mainly corresponding to the ground roll noise, by removing the first decomposed component of each frequency slice. The performance is compared with the FK-based and CEEMD-based approaches and is demonstrated to be very successful.