Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
2000-06-12
2003-03-18
McElheny, Jr., Donald E. (Department: 2862)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
Reexamination Certificate
active
06535818
ABSTRACT:
The present invention relates to a method of stacking seismic traces, in particular a method of stacking seismic traces which results in an improved signal-to-noise ratio.
The basic principles of seismic exploration are well known. Sound waves are reflected by features in the sub-surface of the earth and the reflected signals (or traces) are picked up by sensitive transducers and recorded for later analysis. On land, the seismic source is typically provided by an explosion and the seismic traces are detected by transducers known as geophones. Over water, an impulse of compressed air is provided by a so-called “air-gun” and the seismic traces are detected by hydrophones. (Other sources are possible, however). The recorded traces are then processed by a number of known techniques to provide a representation of the sub-surface. Geologists then can use this representation to determine the likelihood of hydrocarbon deposits. Many techniques are available for analysing the seismic traces and the generation of the representation of the sub-surface from these traces is beyond the scope of the present description.
The present invention is rather concerned with one of the earliest stages in the seismic analysis known as “stacking”. It will be appreciated that the energy level of the seismic traces is generally very low, even when a very high amplitude source is used. In addition, a number of generally unavoidable noise sources are present while the seismic traces are being sampled. The noise of vehicles on roads, aeroplanes passing overhead, factories, drilling and natural phenomena such as rain and wind have the potential to add a large amount of noise to the sampled traces. Because of the low level of the seismic traces, the signal-to-noise ratio of the sampled traces can easily be so poor as to prevent, or at least seriously hinder, the further analysis of these signals. Indeed, it is possible in some circumstances that the amplitude of the noise component of these traces is greater than the amplitude of the signal component. The process of stacking seeks to increase the signal-to-noise ratio of the seismic traces by combining those from a number of different transducers and/or samples taken at different times (or over time) from the same transducer. Broadly speaking, the desired components of the seismic traces have been regarded as having a high degree of correlation while the noise elements of the seismic traces have been regarded as being uncorrelated. Thus, simple summation of the same seismic trace, as received at a number of different transducers, will generally provide an improvement in the signal-to-noise ratio.
However, this crude technique takes no account of the relative noise levels on the different traces. In U.S. Pat. No. 3,398,396 to P. Embree entitled “Diversity Seismic Record Stacking Method and System” the noise energy of each of the seismic traces is determined and a stacked trace is derived from a weighted sum of the seismic traces from each of the transducers. The weight applied to each seismic trace is in inverse proportion to the noise energy of that particular seismic trace. Thus, a contribution to the weighted sum is derived from each of the seismic traces but the contribution from those traces having a comparatively poor signal-to-noise ratio is less than that from traces having a comparatively good signal-to-noise ratio.
The diversity stacking technique has a major disadvantage as follows. The technique relies on the assumption that the noise component of each of the seismic traces is uncorrelated with the noise component in any of the other traces. However, in many practical situations, this assumption is not actually valid. It has been found that, as a result, diversity stacking can often result in rather poor signal-to-noise ratios. The correlation of noise components may be due to having originated at the same source (for example a factory) or, since the correlation is a mathematical concept, coincidentally from separate sources.
It is an object of the present invention to provide a method of stacking seismic traces which can provide an improved signal-to-noise ratio.
According to the present invention, there is provided a method of stacking a plurality of seismic traces comprising deriving a representation of the noise component of each of the seismic traces, deriving an indication of the correlation between the noise components from the representations of the noise components and producing a stacked seismic trace by combining the plurality of seismic traces in proportions derived from the correlation indications.
Because this method does not rely on the assumption that the noise on each trace is independent, correlated noise series can also be attenuated. This has been found to provide a stacked trace having a signal-to-noise ratio of between 2 dB and 7 dB better than the prior art techniques.
The traces may be derived from a number of seismic sensors but could also be derived from a single sensor (eg. geophone) over two or more sweeps. In this case the wanted signal is the same, or very nearly so, on subsequent sweeps and this makes the present invention particularly applicable as discussed in more detail below.
It will be shown that the optimum weighting factors for a combination of the seismic traces are related to the cross-energy density of the various noise signals. In a preferred embodiment the representation of the noise component of each of the seismic traces used to derive the proportions is the energy of the noise components. Thus, the embodiment uses not only zero-lag autocorrelations (equivalent to energy) as does diversity stacking but also zero-lag cross-correlations between the noise components.
Preferably, the covariance of the noise components of each of the seismic traces Is used to provide a set of weights for the combination of the plurality of seismic traces.
The invention may further comprise breaking the samples up into a plurality of time windows. Generally, the signal-to-noise ratio of the combined seismic trace will improve as the length of the windows is reduced. However, this becomes increasingly computationally intensive with a lower limit of one to two wavelengths of the information signal. A window duration of 400 milliseconds has been found to provide a good compromise with typical seismic traces in the region of 10 to 80 Hz.
The invention may be applied in the time domain or in the frequency domain. The advantage of performing the calculations in the time domain is that pre-processing of the seismic traces is reduced. However, processing the signals in the frequency domain has the advantage that noise reduction can be optimised. This is true of both full frequency domain analysis or separating the signal into frequency bands prior to time domain analysis.
Even in the time domain, the technique will usually benefit from separating each of the seismic traces into a number of different frequency bands. The previously referenced U.S. Pat. No. 3,398,396 to P. Embree illustrates separating seismic data into frequency bands but only in conjunction with diversity stacking. Preferably, though, the signals are separated into separate frequency bands using a quadrature mirror filter (QMF). See WO 9620451 or GB9614148-A0 for further details. The advantage to doing this lies in the fact that in seismic traces certain repetitive noise signals are found only in certain frequency bands while another frequency band or bands contain predominantly purely random noise. By splitting up the frequency spectrum of the seismic traces the stacking technique can be selected or tailored to the type of noise present.
For example, if the variation in the wanted signal between traces is large, covanant stacking may actually remove some signal. This may happen with first breaks when the signal amplitude is large. Diversity stacking may be applied in preference in these particular circumstances.
REFERENCES:
patent: 5572483 (1996-11-01), Chambers et al.
patent: 5850622 (1998-12-01), Vassiliou et al.
patent: 5978314 (1999-11-01), Pham
Figatner David S.
McElheny Jr. Donald E.
Moser, Patterson & Sheridan L.L.P.
Schlumberger Technology Corporation
LandOfFree
Method of stacking seismic traces does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of stacking seismic traces, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of stacking seismic traces will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3082538