Efficient Multiple Prediction In Two and Three Dimensions

A method for prospecting for hydrocarbons, comprising:

• (a) obtaining seismic survey data;
• (b) using the seismic data, reconstructing shot gathers of data at points on an (x,y) grid, said grid being selected with dense mesh spacing sufficient to distinguish potential locations of downward reflection points (“bounce” points) of multiple reflections;
• (c) selecting a reconstructed gather corresponding to a bounce point, and a data trace in the survey data, then extracting two traces from the selected reconstructed gather corresponding to the source and receiver positions of the survey data trace, and convolving them, then repeating for each survey data trace within a selected aperture, thus performing pair-wise convolution of traces from a single reconstructed gather centered at a bounce-point location;
• (d) repeating step (c) for each remaining bounce point, and summing the convolution result for each processed trace in the survey data;
• (e) predicting multiple reflections in the survey data using the results of the preceding step;
• (f) adaptively subtracting the predicted multiple reflections from the survey data;
• and
• (g) predicting whether an accumulation of hydrocarbons is present using the seismic data after correction for multiple reflection noise.

A method for producing hydrocarbons, comprising:

• (a) conducting a seismic survey;
• (b) obtaining a prediction of whether data from the seismic survey indicates an accumulation of hydrocarbons, said prediction having been made by a method comprising:
• (i) using the seismic data, reconstructing shot gathers of data at points on an (x,y) grid, said grid being selected with dense mesh spacing sufficient to distinguish potential locations of downward reflection points (“bounce” points) of multiple reflections;
• (ii) selecting a reconstructed gather corresponding to a bounce point, and a data trace in the survey data, then extracting two traces from the selected reconstructed gather corresponding to the source and receiver positions of the survey data trace, and convolving them, then repeating for each survey data trace within a selected aperture, thus performing pair-wise convolution of traces from a single reconstructed gather centered at a bounce-point location;
• (iii) repeating step (c) for each remaining bounce point, and summing the convolution result for each processed trace in the survey data;
• (iv) predicting multiple reflections in the survey data using the results of the preceding step;
• (v) adaptively subtracting the predicted multiple reflections from the survey data;
• (vi) predicting whether an accumulation of hydrocarbons is present using the seismic data after correction for multiple reflection noise;
• and
• (c) in response to a positive prediction, drilling a well and producing hydrocarbons.

A method for eliminating multiple reflection noise in data from a seismic survey, comprising:

• (a) reconstructing densely sampled gathers at a plurality of potential locations of downward reflection points of the multiple reflections;
• (b) for each reconstructed gather and each output (survey) trace, extracting two traces corresponding to the survey source and receiver positions of the output trace, then convolving the pair of extracted traces and adding the convolution results obtained for all reconstructed gathers;
• and
• (c) making multiple predictions for the survey based on the summed convolution results, and correcting the survey's data to eliminate multiple reflection noise.