Integration preconditioning for ocean bottom node data

SUMMARY

We investigate options to perform integration preconditioning for Ocean bottom node (OBN) data acquired with accelerometers.

When amplitude perturbations due to sensor saturation are present, integration exaggerates them through convolution with the integration filter; therefore, their correction is desirable. Furthermore, the affected samples may impact PZ calibration, wavefield separation and multi-dimensional deconvolution, therefore a high fidelity correction is needed.

We found that two algorithms, a 5D version of POCS (Projection into Convex Sets) and an adaptive version of PINNs (Physics Informed Neural Networks) are effective at interpolating the few affected data samples. We present examples of their application to field data acquired with either a standard source array or a bandwidth-controlled source array, as the latter reduces sensor saturation and complements the proposed processing solution.