Basic Research Needs for Geosciences - Energetics Meetings and ...

GRAND CHALLENGE:INTEGRATED CHARACTERIZATION, MODELING, AND MONITORING OF GEOLOGIC SYSTEMSgeostatistics (Strebelle 2002), provide important improvements for incorporation of hard and softgeologic information, but these are still “structure imitating” techniques (Koltermann andGorelick 1996) and are therefore limited in their ability to account for important geologicprocesses which, for example, can give rise to unanticipated preferential flow paths.The geological architecture (distribution of rock permeability, connectivity of high and lowpermeability media, fractures and faults) determines flow pathways. Modern geology, includingknowledge of depositional systems, sequence stratigraphy, structural history, and diagenesis,illustrates that the distribution of architectural features is not random. Rock mechanicsdocuments that fracture distribution relates to strain from stretching, bending or tearing, and tothe distribution of anisotropy elements (e.g., bed thickness). One challenge is to link the hydrodynamicsof sediment deposition, erosion and transport to the spatial dimensions and reservoirproperties of the resulting deposits, predict reservoir body stacking patterns in response to baselevelchanges, and document their degree of connectivity at successively larger scales (from poreto basin scale).Early attempts at such 3D geologic process modeling can be found in Tetzlaff and Harbaugh(1989), Koltermann and Gorelick (1992), and more recently Teles et al. (2001) and Euzen et al.(2004). The newer models are improved in part by enhanced computer hardware and governingequations. Much work, however, remains to be done to make quantitative geologic depositionalmodels theoretically rigorous and to reach their potential in subsurface characterization. Becausethe boundary conditions (including sediment supply, tectonic history, and climate) will never beknown precisely, these geologic models may be most useful as hypothesis-testing, rather thanpredictive, tools. One important byproduct might be improved geostatistical models or processgeostatisticalhybrids. Another might be improved understanding of which geologic features aremost important for certain fluid flow and transport processes, providing a means of estimatingoptimal levels of geologic detail (discussed in the prior section).Incorporating large and diverse datasets into modelsHill and Tiedeman (2007) describe the current state of inverse modeling in groundwater flow andtransport. Current practice in inverse modeling tends to decouple processes, to aggregateparameters across scales, and to include only a limited amount of the available data. It isanticipated that advances in characterization methods will produce increasingly larger datasets asthe spatial extent and the spatial and temporal resolutions of measurements increase. In addition,with the development of new measurement devices, data types not currently commonplace maybecome routinely available. The scientific challenge is to incorporate these larger data volumesand disparate data types (see Figure 30) into inverse modeling while simultaneously satisfyingdiverse physical and chemical constraints. In particular, in the context of monitoring perturbedgeological systems, models will have to handle streams of different data types measured over avariety of spatial and temporal resolutions, including data indirectly related to system statevariables and model parameters. The computational burden of such comprehensive inversemodeling requires fundamental advances in mathematical methods and numerical algorithms.Revolutionary advances in the modeling process are required to achieve more automated, realtimeincorporation of diverse datasets into models.Basic Research Needs for Geosciences: Facilitating 21 st Century Energy Systems 83