Dynamic Management of Large Seismic Cubes

in: 24th gOcad Meeting, ASGA

Abstract

Seismic surveys provide the only “continuous” high resolution 3D information on reservoirs. Of paramount importance in the Shared Earth Model design, this information is growing continuously due to considerable technical advances in the past years. From the reservoir scale, the resulting seismic cubes have moved to the basin scale, providing interpreters with considerable amounts of data. However, the interpretation scale has not evolved, interpreters usually being focused on local reservoirs in the whole basin. It is therefore necessary to find out techniques that manage such scale contrasts dynamically during interpretation. From a hardware point of view, multi-gigabyte seismic cubes do not fit in computers main memory, nor in texture memory. Beside a dynamic management of seismic scales during the interpretation, this latter point supposes a dynamic management of memory. Therefore, the notion of dynamic system for interactive interpretation of large volumes involves both a dynamic management of seismic scales, from the basin scale down to detail analysis of a potential reservoir, and a dynamic management of memory, based on hardware limitations. This paper focuses on the latter point and describes the design of a double-level cache system implemented between disk and main memory, and between main memory and texture memory, that handles each data request and provides a full control of data flows between memory levels. This makes it possible to dynamically manage data requests and make interactive roaming through large volumes possible.

Download / Links

    BibTeX Reference

    @inproceedings{CastaniéRM2004,
     abstract = { Seismic surveys provide the only “continuous” high resolution 3D information on reservoirs. Of paramount importance in the Shared Earth Model design, this information is growing continuously due to considerable technical advances in the past years. From the reservoir scale, the resulting seismic cubes have moved to the basin scale, providing interpreters with considerable amounts of data. However, the interpretation scale has not evolved, interpreters usually being focused on local reservoirs in the whole basin. It is therefore necessary to find out techniques that manage such scale contrasts dynamically during interpretation. From a hardware point of view, multi-gigabyte seismic cubes do not fit in computers main memory, nor in texture memory. Beside a dynamic management of seismic scales during the interpretation, this latter point supposes a dynamic management of memory. Therefore, the notion of dynamic system for interactive interpretation of large volumes involves both a dynamic management of seismic scales, from the basin scale down to detail analysis of a potential reservoir, and a dynamic management of memory, based on hardware limitations. This paper focuses on the latter point and describes the design of a double-level cache system implemented between disk and main memory, and between main memory and texture memory, that handles each data request and provides a full control of data flows between memory levels. This makes it possible to dynamically manage data requests and make interactive roaming through large volumes possible. },
     author = { Castanié, Laurent AND Bosquet, Fabien AND Levy, Bruno },
     booktitle = { 24th gOcad Meeting },
     month = { "june" },
     publisher = { ASGA },
     title = { Dynamic Management of Large Seismic Cubes },
     year = { 2004 }
    }