End-to-End Data Provenance

Abstract Data provenance is the chronology of an object, i.e., all the transformations and actions that modified the state of the object from the time of creation to the present. End-to-end data provenance is the ability to account not only for the data provenance of a single object, but also to account recursively for the data provenance of all other objects that affected the state of the object. The purpose of data provenance is to help ensure data integrity. Modeling has become an important aspect of shared earth problems. As the role and extent of modeling of the subsurface expands and its use is taken up by larger teams, determining the provenance of data becomes ever more important. The explosion of data and the appearance of increasingly diverse teams require the establishment of better data provenance to guard the integrity of the modeling results. Such a data provenance model is presented in this paper. The economic consequences of such an approach are significant: better integrity of modeling results leads to more confidence in those results, which reduces perceived risk and, in turn, can improve the economic performance of field development. The solution to the problem of end-to-end data provenance has six elements: The ability to uniquely identify objects of the shared earth problemAn object repository that supports federation and extensibilityThe ability to associate an immutable change history with each objectThe loose coupling of metadata and actual data of an objectClient-side caching of actual dataA comprehensive security model for the object. Each of these elements solves an important aspect of ensuring data integrity. Such an approach is increasingly important as model-based approaches become ubiquitous, and the available data become more voluminous, and the teams that handle the data and modeling results become larger and more diverse.