Combining CEEMD and predictive deconvolution for the suppression of multiple reflections and coherent noise in GPR signals

Complete Ensemble Empirical Mode Decomposition (CEEMD) is a novel noise assisted and adaptive technique appropriate for decomposing GPR signals. In this study we combine CEEMD and predictive deconvolution on several GPR lines to examine its capability to suppress both coherent and random noise. The effect of this approach is presented in a GPR section and depth-slices where the instantaneous envelope was additionally used. The results obtained are promising as CEEMD improves the response from deeper levels, while the predictive deconvolution removes coherent noise and partially suppresses multiple reflections.