Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
2003-04-14
2004-10-12
McElheny, Jr., Donald (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
C702S013000
Reexamination Certificate
active
06804609
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to methods for using seismic data to predict reservoir properties at unexplored subterranean locations. In another aspect, the invention relates to a method for predicting reservoir properties at unexplored subterranean locations based on known reservoir properties at explored subterranean locations and seismic attributes generated from a seismic survey or surveys encompassing both the explored and unexplored subterranean locations.
2. Description of the Prior Art
Seismic surveys gather important information about the subsurface of the earth. Data gathered from a seismic survey is typically manipulated to yield a pool of unique seismic attributes. Seismic attributes can be defined as analytical measurements of the seismic expression of geologic conditions and can take a variety of forms. Frequently, seismic attributes are measurements of a seismic waveform's amplitude, length, area, symmetry, frequency, or phase. In addition, seismic attributes may be discrete classifications (e.g., pattern assignments or facies), structural (time or depth horizons, isochrons, or isopachs), and spatial coordinates (e.g., X-coordinate and Y-coordinate). Each seismic attribute responds to particular reservoir conditions in a unique manner. Thus, seismic attributes can be used to predict reservoir properties (e.g., porosity, thickness, or fluid type) of the subterranean formation. By using multiple seismic attributes for reservoir property prediction, noise contamination may be reduced and accuracy of the prediction may be enhanced.
A number of conventional methods exist for using multiple seismic attributes to predict reservoir properties of a subterranean formation. When known reservoir properties (typically from well logs) are available at locations within the surveyed region, those known reservoir properties can be used to help “calibrate” the seismic attributes. A variety of methods exist for calibrating seismic attributes with known reservoir properties in an effort to more accurately predict reservoir properties at unexplored locations. One conventional calibration method performs calculations for all possible combinations of the seismic attributes. Such an exhaustive approach is very computationally intensive and can require long periods of time and expensive computers to achieve. Another conventional calibration method involves selecting a pre-identified group of the seismic attributes to use in the calculation. However, one can never be sure that the pre-selected seismic attributes provide the best solution. Still another conventional calibration method progressively adds seismic attributes to a predetermined starting attribute. This method, however, can result in a local answer which is not the optimal global solution.
Typically, the above-described conventional techniques each yield a multi-variable prediction equation that can be employed to calculate a predicted reservoir property at a certain unexplored location based on multiple seismic attributes at that location. However, there is currently no procedure for quantitatively determining the relative contribution of each seismic attribute used in the prediction equation. Such a procedure for quantitatively determining the predictive significance of each seismic attribute would be helpful for selecting which seismic attributes to extract from seismic survey data for future property predictions.
OBJECTS AND SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a computationally efficient method of predicting reservoir properties using multiple seismic attributes.
Another object of the present invention is to provide a more accurate method of predicting reservoir properties using multiple seismic attributes that avoids providing a local solution.
Still another object of the present invention is to provide a quantitative method of determining the relative significance of individual attributes used as variables in multi-variable prediction equations.
It should be understood that the above-listed objects are only exemplary, and not all the objects listed above need be accomplished by the invention described and claimed herein.
In one embodiment of the present invention, there is provided a method of predicting a reservoir property at an unexplored subterranean location based on a seismic attribute pool generated from a seismic survey and known reservoir properties at explored subterranean locations. The seismic attribute pool includes a plurality of common seismic attributes for each explored and unexplored location. The property prediction method comprises the steps of: (a) performing a first regression of the known reservoir properties and a first one of the common seismic attributes at the explored locations, with the regression yielding a first prediction equation for calculating a first predicted reservoir property; (b) calculating first residuals for the known reservoir properties and corresponding first predicted reservoir properties generated with the first prediction equation; (c) correlating the first residuals with the common seismic attributes not used in the regression of step (a); and (d) selecting the common seismic attribute with the highest correlation from step (c) as a second one of the common seismic attributes.
In another embodiment of the present invention, there is provided a stepwise regression method for predicting a reservoir property at an unexplored subterranean location based on a seismic attribute pool generated from a seismic survey and known reservoir properties at explored subterranean locations. The seismic attribute pool includes a plurality of common seismic attributes for each explored and unexplored location. The stepwise cumulative regression method comprises the steps of: (a) selecting one of the common seismic attributes as a current starting attribute and proceeding to step (b); (b) adding the selected seismic attribute from the previous step to a cumulative attribute set and proceeding to step (c); (c) performing a regression of the known reservoir properties and the seismic attribute or attributes in the cumulative attribute set, to thereby yield a current prediction equation for calculating a predicted reservoir property and proceeding to step (d); (d) calculating a current correlation value, a current prediction error and current residuals for the known reservoir properties and corresponding predicted reservoir properties generated with the current prediction equation and proceeding to step (e); (e) proceeding to step (g) if the selected seismic attribute from step (b) is the starting seismic attribute, otherwise comparing the current correlation value and prediction error to a prior correlation value and prediction error and proceeding to step (f); (f) proceeding to step (j) if the current correlation value or prediction error is worse than the prior correlation value and prediction error, otherwise proceeding to step (g); (g) correlating the residuals from step (d) with each of the common seismic attributes not currently in the cumulative attribute data set and proceeding to step (h); (h) designating the current prediction equation, correlation value, and prediction error as a prior prediction equation, correlation value, and prediction error and proceeding to step (i); (i) selecting the seismic attribute with the highest correlation from step (g) as a next seismic attribute and returning to step (b); and (j) designating the prior prediction equation as an optimum prediction equation for the current starting attribute.
In still another embodiment of the present invention, there is provided a method of predicting a reservoir property at an unexplored subterranean location based on a seismic attribute pool generated from a seismic survey and known reservoir properties at explored subterranean locations. The seismic attribute pool includes a plurality of common seismic attributes for each explored and unexplored location. The prediction method comprises the steps
ConocoPhillips Company
Cross Ryan N.
Kelly Kameron D.
McElheny Jr. Donald
LandOfFree
Property prediction using residual stepwise regression does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Property prediction using residual stepwise regression, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Property prediction using residual stepwise regression will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3277393