A Relational Database for the Digitization of Fluvial Architecture: Toward Quantitative Synthetic Depositional Models

Facies models for fluvial depositional systems aim to summarize the sedimentological features of a specific fluvial type (e.g. braided, ephemeral) through a process of distillation of several real-world examples, in order to provide conceptual frameworks that are straightforwardly applicable to subsurface prediction problems. However, such models are often based on few case studies and are qualitative in nature, thereby resulting in poor predictive power. Our aim is to generate quantitative depositional models for fluvial systems that are based on the synthesis of many different case histories and continuously refined by adding data when they become available.

A relational database for the storage of data relating to fluvial architecture has been devised, developed and populated with literature- and field-derived data from studies of both modern rivers and their ancient counterparts preserved in the stratigraphic record. The database scheme characterizes fluvial architecture at three different scales of observation, corresponding to many genetic-unit types (large-scale depositional elements, architectural elements and facies units), recording all the essential architectural features, including style of internal organization, geometries, spatial distribution and reciprocal relationships of genetic units. The database classifies datasets - either in whole or in part - according to both controlling factors (e.g. climate type, tectonic setting) and context-descriptive characteristics (e.g. river pattern, dominant transport mechanism). The data can therefore be filtered on the parameters according to which they are classified, allowing the exclusive selection of data relevant for the model.

To demonstrate the value of the approach, an example synthetic depositional model for braided fluvial systems in arid/semiarid basins is presented here, and some of its features are compared with analogous data from other settings. Resultant models are based on outcrop studies of the Permian Organ Rock Fm., Triassic Moenkopi Fm., Jurassic Kayenta Fm.
(all from Utah), the Chester Pebble Beds Fm. and Helsby Fm.
(both Cheshire Basin, UK), together with literature-derived data. In comparison to traditional facies models, the improved usefulness of synthetic models derived from this database approach to subsurface predictions is evident, as their quantitative content is particularly suitable to inform well-to-well correlations and to constrain stochastic reservoir models.

AAPG Search and Discovery Article #90142 © 2012 AAPG Annual Convention and Exhibition, April 22-25, 2012, Long Beach, California