| 摘要 |
1. Method for filtering noise from discrete noisy seismic signals in a time interval ([1,..., T]), said method comprising the steps of: - using said noisy seismic signals for determining at least one reference channel (x (t)) for said time interval as an estimate of said noise; - determining coefficients for M temporally local filters (w(i, t)), said filters forming a filter bank and M being a number equal to or larger than two, by minimizing a cost function (J(t)) representing a measure of the error of the output of said filter bank; and - applying said filter bank to said at least one estimate (x (t)) to determine M filtered estimates of said noise. 2. A method according to Claim 1, wherein the cost function (J (t)) is temporally global, representing a measure of the error of the total output of said filter bank in the time interval ([1,..., T]). 3. A method according to Claim 1, including the step of multiplying M filtered estimates with temporal window functions ((h(i, t)). 4. A method according to Claim 3, wherein the temporal window functions are characterized by the requirement that only adjoining windows overlap. 5. The method according to Claim 3, wherein the cost function (J (t)) is minimized using the approximation that the sum, weighted by window functions, of the output of adjacent filters of the M filters (w(i, t)) is equal when applied to the same signal in time regions where said window functions overlap. 6. The method of Claim 1, wherein the discrete noisy signals used as input are recordings from individual seismic sensors prior to any group forming techniques. 7. A method according to any of Claims 1 to 5, wherein global optimization of the filtered output signal is solved by the method of principal components. 8. A method according to any of Claims 1 to 4, wherein the global optimization of the filtered output signal is solved by a damped least squares approach. 9. A method according to Claim 7, wherein the number of principal components is adjusted to vary the degree of filtering. 10. A method according to any of the preceding Claims, wherein adaptive filtering is achieved by use of a filtering means, comprising a plurality of local filters. 11. A method according to Claim 10, wherein each local filter is a multichannel filter. 12. A method according to any of Claims, wherein the filtered output signal is reprocessed in an iterative manner to further filters the noise. 13. A method according to any of the preceding Claims, wherein the reference signal is obtained by moveout differentiation of data signals. 14. A method according to any of Claims 1 to 12, wherein the reference signal is obtained by median stacking. 15. A method according to Claim 14, wherein the median stacking is followed by moveout differentiation of the stacked signals. 16. A method according to any of Claims 1 to 12, wherein the seismic data is obtained from multicomponent sensing means, and the reference signal is obtained by polarization filtering of each component sensed by the sensing means. 17. A method according to Claim 15, wherein polarization filtering is combined with moveout differentiation. 18. Iterative use of a method in accordance with Claim 1, wherein the filtered noise estimate determined in a previous application of the filtering method is used as the noise estimate in a subsequent filtering step. 19. Use of the method of Claim 1 in a seismic 2D or 3D survey of land, marine, or transitional zone. |
