摘要 |
A method of full waveform inversion of seismic data comprises generating at least one convolutional filter, each comprising three or more non-zero filter coefficients, and generating a convolved observed data set, or a convolved predicted data set, by convolving the or each convolutional filter with the observed seismic data set or a predicted data set based on a subsurface model. One or more primary objective functions are generated to measure the similarity/mismatch between said convolved observed dataset and said predicted dataset, or between said convolved predicted data set and said observed data set. The at least one primary objective function is maximised/minimised by modifying at least one filter coefficient of the or each convolutional filter. One or more pre-determined reference filters are generated, comprising at least three reference coefficients. One or more secondary objective functions are generated to measure the similarity/mismatch between the filter coefficients for the convolutional filter(s) and the reference coefficients for the reference filter(s). The secondary objective function(s) are minimised/maximised by modifying at least one model coefficient of the subsurface model to produce an updated subsurface model. |