Abstract. Subsurface heterogeneity is characterized with a method based on facies delineation and process modeling. A geologic process model is formulated to simulate the development of four facies: channel fill, levee, splay, and floodplain, which commonly form in a meandering river and its floodplain. The model accounts for meandering, cutoff, flooding, and crevasse-splay processes and links these formational processes to the statistical properties of the resulting facies. The model is utilized in a case study to simulate the horizontal pattern of deposition and facies distributions of the Stratton field, part of the Middle Frio Formation of southeast Texas, using known or estimated relationships and historical information for the Stratton field region. The simulated distributions of longitudinal and transverse facies dimensions and corresponding transition probabilities among facies are compared to those determined from sampling an observed section of the field, yielding similar results though some differences. Similarities and differences are related to the model processes and sampling procedures. Model sensitivity is investigated by changing selected model variables and analyzing the change in the facies distributions. The application of the geologic process model to the Stratton field demonstrates the viability of linking process understanding and modeling of complex meandering stream processes to characterization of subsurface heterogeneity.
IntroductionEffective characterization of subsurface heterogeneity is essential for studies in a number of disciplines, including site investigation and remediation at hazardous waste sites, evaluation of groundwater flow and transport in regional and subregional aquifers, and assessment of hydrocarbon reserves and recovery. Two principal approaches to characterizing heterogeneity include stochastic and geologic methods [Koltermann and Gorelick, 1996]. Stochastic methods describe the continuous variation of sediment or rock characteristics, such as porosity and permeability, as spatial random functions with assumed or estimated statistical properties that can vary as a function of scale [Sudicky, 1986;Neuman, 1990