Mutiple-point Geostatistics to Estimate Sandstone Distribution

Abstract

The familiar geostatistical methods, like Kriging and Sequential Gaussian Simulation, borrow a two-point statistics information known as variogram. A variogram itself is a function that defines the value of covariance and distance between two points which then used to estimate unknown value across the spatial map. The use of variogram as an estimation parameter is what makes the result of geostatistical simulation relatively smooth. Which is why this simulation is not able to model complex geology features such as fracture, channel, fluvial, delta, etc. which tends to have a certain distribution pattern that can not be well reconstructed only using a simple Gaussian simulation. The limitations of this two-point statistics are what motivate the introduction of multi-point geostatistics in the study of geological modeling. Like its name, multi-point statistics is a statistical methods that borrows information from multi-points, replacing the role of variogram with the so- called training image. An algorithm called Single Normal Equation Simulation (SNESIM) is introduced for the multi-point statistics simulation. This paper presents how to estimate sandstone distribution with using multiple-point geostatistic.