Paleokarst reservoirs: Efficient and flexible characterization using point-spread-function-based convolution modeling

Abstract

Paleokarst originates from the collapse, degradation, and infill of karstified rock, and it typically features spatially heterogeneous elements such as breakdown products, sediment infills, and preserved open cavities on all scales. Paleokarst may further contain aquifer or hydrocarbon reservoirs and may pose a drilling hazard during exploration. Seismic characterization of paleokarst reservoirs therefore remains a challenging and important task. We have determined how the application of 2D (3D) spatial convolution operators, referred to as point-spread functions (PSFs), allows for seismic modeling of complex and heterogeneous paleokarst geology at a cost equivalent to conventional repeated 1D convolution. Unlike the latter, which only considers vertical resolution effects, PSF-based convolution modeling yields simulated prestack depth-migrated images accounting for 3D resolution effects vertically and laterally caused by acquisition geometries, frequency-band limitations, and propagation effects in the overburden. We confirm the validity of the approach by a comparison of modeled results to results obtained from a published physical modeling experiment. Finally, we present four additional separate case studies to highlight the usability and flexibility of the approach by assessing different issues and challenges pertaining to characterizing and interpreting seismic features of paleokarst. Through PSF-based convolution modeling, geoscientists working with paleokarst seismic data may be better able to understand how various acquisition and modeling parameters affect seismic images of paleokarst geology.