2021 Volume 79 Issue 2 Pages 71-84
This paper reviews the basic concept, workflow and methodological variation of geostatistical sediment body modeling. Geostatistics was originally developed in the mining industry during 1960’ and 1970’ for the purpose of mathematical estimation of high-quality lode distributions, and since then, has widely been used for multiple purposes, including petroleum exploration and production. Geostatistical sediment body modeling is a quantitative and stochastic modeling method of sediment body basically using geostatistics. The basic data for the geostatistical sediment body modeling include well petrophysics as hard data and 2D/3D seismic attribute data as soft data, and they are integrated using various geostatistical algorithms to conduct the geostatistical stochastic simulation. To obtain realistic modeling results, geological and sedimentological deterministic/qualitative model information should be provided as the constraints for the geostatistical stochastic simulation. Various geostatistical simulation methods have been developed in response to the demand for the geological/sedimentological information input, as the geological/sedimentological factors are crucial for simulating the precise facies and property distributions. In addition to the past standard method of two-point pixel-based geostatistical simulation, new methodology has been proposed after the mid 1990’, including object-based modeling, multiple-point pixel-based modeling, process-aided modeling, surface-based modeling and process model-based modeling. The object-based modeling is a stochastic simulation based on a sediment body object such as a lobe and meandering channel, which size, dimension and orientation trends are set using known information. The multiple-point pixel-based modeling is a stochastic simulation using a pixel-based training image of a sediment body, and utilizes data points as hard data. The process-aided modeling is a modeling using sedimentary process algorithm and experimental results. Surface-based modeling is a simulation based on a depositional surface, which dimension is determined on the basis of known information. The depositional process-based modeling is a simulation based on the forward process model using reasonable algorithm matching actual sedimentary processes.
In addition to the new methodology developments, selection of appropriate modeling workflow and methodology is crucial for successful modeling, in consideration of the purpose of the modeling, depositional system of the target, and data condition in density, quality and distribution.