HexaShrink, an exact scalable framework for hexahedral meshes with attributes and discontinuities: multiresolution rendering and storage of geoscience models

Abstract

With huge data acquisition progresses realized in the past decades and acquisition systems now able to produce high resolution grids and point clouds, the digitization of physical terrains becomes increasingly more precise. Such extreme quantities of generated and modeled data greatly impact computational performances on many levels of high-performance computing (HPC): storage media, memory requirements, transfer capability, and finally simulation interactivity, necessary to exploit this instance of big data. Efficient representations and storage are thus becoming This work was partly presented in [43]. This is a post-peerreview, pre-copyedit version of an article published in Computational Geosciences. The final authenticated version is available online at: http://dx.

Lossless Progressive Compression of Meshes for Upscaling and Upgridding in Reservoir Simulation with HexaShrink

“…Continuous properties such as porosity and permeability are decomposed with the separable 3D wavelet used by the image compression standard JPEG2000 (Acharya and Tsai, 2005). More details are provided in (Peyrot et al, 2019(Peyrot et al, , 2016. Note that such discrete wavelets have already been used to scale continuous properties up (Pazhoohesh et al, 2006;Rasaei and Sahimi, 2008), with promising results.…”

“…This study proposes an alternative to such trending "big data" accumulation problems that impact computational performances in experimental and simulation sciences. It relies on HexaShrink, an original decomposition scheme for structured hexahedral volume meshes (Peyrot et al, 2019(Peyrot et al, , 2016. The latter provides a consistent and fully reversible dyadic multiscale representation of mesh geometry and properties.…”

Real-Time Visualization of Large-Scale Geological Models With Nonlinear Feature-Preserving Levels of Detail

“…State-of-the-art geological model representations often focus on simplicity, data compression, or facilitating fast simulation without supporting interactive visualization of large-scale data. The corner-point grid data structure is a widely used industry standard for encoding geological models for reservoir and basin modeling and simulation [8], [9]. It stores the attributes and mesh vertices as a set of labeled one-dimensional lists of values encoded either in a standard ASCII file, e.g., the .GRDECL format, or in a binary file, e.g., the .EGRID format [9].…”

“…A recent multiresolution representation for geological models, HexaShrink [8], is able to better maintain geometric discontinuities across resolution levels by adaptively subsampling mesh vertices, i.e., prioritizing vertices that belong to geological faults. However, its wavelet-based encoding coupled with its choice of using low-frequency subbands or approximation coefficients as low-resolution versions of the attribute data oversmoothes attribute edges, as shown in Fig.…”

“…Corner-point geological models containing attribute discontinuities (top) and geological faults (bottom). From the original data (a,e) a low resolution version is computed using different methods on corner-point data: (b,f) box filtering followed by subsampling introduces incorrect attributes along discontinuities (b) and removes faults (f); (c,g) subsampling only leads to aliasing artifacts such as skipped faults (g); (d,h) HexaShrink [8] reduces mesh artifacts (h) but introduces attribute artifacts (d), similar to those introduced by box filtering (b).…”