Application of Non Local Means Filter and Sparse Spike Deconvolution for Seismic Processing. Reappraisal of Rotliegend and Dinantian formations in the Netherlands

Abstract

Conventional seismic reprocessing and new unconventional methods, the Non Local Means algorithm and Sparse Spiked Deconvolution are applied to a real 2D Seismic Dataset. The aim is to get better images on the Rotliegend and Dinantian formations which are the target of Geothermal Exploration in the Netherlands. Conventional seismic reprocessing gives improvement over the seismic data, particularly due to the superior Pre-Stack Kirchhoff Time Migration used. Non Local Means Algorithm proven to increase signal to noise ratio of the seismic data while preserving fault edges. We propose different NLM workflow with different parameters for different processing target. Sparse Spike deconvolution with combined l1 norm and l2 norm regularization for maximum sparsity and smoothness constraints is found to increase vertical resolution and preserves reflection amplitudes and structures, reveals new reflection packages and small geological features emerge from the data. Although NLM and Sparse Spike deconvolution are found to improve the signal to noise ratio and vertical resolution in the mid depth section of Rotliegend reservoir, the offset limitation factor still hampers the imaging of deep section of Dinantian reservoir. We suggest for new seismic acquisition to use larger offset to get the maximum results on the Dinantian reservoir imaging