Ocean bottom node (OBN) seismic data processing commonly uses the combination of hydrophone (P) and vertical geophone (Z) components to separate upgoing and downgoing wavefields while attenuating ghosts and multiples. Because the vertical components are often contaminated by strong converted shear waves due to the scattering in the shallow subsurface, a good Vz noise removal is crucial for better P/Z calibration, wavefield separation and deconvolution. Most importantly, a better wavefield separation will in turn improve the quality of the seismic images. This paper will present a novel method of simultaneous Vz noise attenuation and P/Z calibration in the 3D curvelet domain. Furthermore, we will demonstrate that our new 3D method outperforms a 2-pass 2D curvelet approach. We will also assess the effect and necessity of Vz attenuation prior to shot carpet regularization. Both investigations are supported by field data examples from the Utsira OBN project.