Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
2000-06-28
2002-04-09
McElheny, Jr., Donald E. (Department: 2862)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
C367S074000
Reexamination Certificate
active
06370477
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to compression methods and apparatus for seismic data.
BACKGROUND OF THE INVENTION
Data compression (or reduction) is a digital signal processing technique for reducing the amount of data to be dealt with without losing essential information in the process. This is essentially done by the removal of redundancy in the data and may involve the discarding of uninteresting parts of the data. Such compression can result in some loss of data accuracy. Data compression that allows the exact reconstruction of the original data is often referred to in the literature as lossless. Data compression that involves some reduction in accuracy is known as lossy. Common examples of data compression are “rounding” and “down sampling”; both methods are usually lossy.
Seismic data acquisition requires a large number of seismic experiments to be conducted in order to obtain a reliable image of the Earth's subsurface. Each experiment involves the generation of a sound wave using an appropriate source and measuring the earth's response by a large number of receivers. A large scale seismic survey thus produces an enormous amount of data which will normally be in digital format, which has to be transmitted, stored and processed. To facilitate the handling of such large volumes of data, data compression can be utilized.
A lossy data compression technique that is routinely used in seismic data acquisition is group forming. This involves the retention and transmission and processing of the sum of neighboring receivers within fixed-sized groups, instead of the individual measurements.
Group forming is not used primarily for data compression. Group forming suppresses random ambient noise and suppresses waves with low apparent velocities, such as groundroll in land seismics. Thus group forming attenuates the high spatial frequency content of the data. However, the attenuation is performed in a crude way as it only partially suppresses apparently slowly propagating waves and alters the rest of the data. Consequently there is a good reason to omit group forming from the acquisition stage and to record the output of every receiver individually. This then permits the application of more sophisticated methods for reducing random and coherent noise. However the abolition of group forming at the acquisition stage greatly increases the amount of data to be handled downstream.
In IEEE Int. SYM. Circuits & Systems, New Orleans, La., May 1-3, 1990, Vol. 2, 1573-6, A. Spanias, S. Johnson et al. describe several transform based methods for seismic data compression. The methods include the Discrete Fourier Transformation (DFT), the Discrete Cosine Transformation (DCT), the Walsh-Hardamard Transform (WHT), and the Karhunen-Loeve Transform (KLT). However the DCT in the form described in the publication and applied to a sliding frame of N data points can be used for a relative comparison between several different transformations. When applied as data compression method, the sliding frame produces a large amount of redundant data in the transform domain.
It is therefore an object of the present invention to provide a method for compressing seismic data. It is another object of the invention to provide a method for compressing seismic data without using group forming.
SUMMARY OF THE INVENTION
The invention provides a first level of compression in which local spatial or temporal discrete trigonometric (i.e. either sine or cosine) transformations of type IV are applied to seismic data signals (
FIG. 4
, step
314
). Discrete sine/cosine transformations of type IV are known as such. A general description is given for example by H. S. Malvar in: “Lapped transforms for efficient transform/subband coding”, IEEE ASSP, vol. 38, no. 6, June 1990. The local spatial or temporal discrete sine/cosine transformation results in transform coefficients which are more compact and less correlated that the original data. Both of these properties can be advantageously exploited in subsequent data processing steps.
The compactness of the transform coefficient is exploited in a processing step, which can be described as a requantization or round-off step. The purpose of this step is to retain selected coefficients at high accuracy and other coefficients at lesser accuracy so as to reduce the quantity of data needed to describe the coefficients and thereby achieve further data compression.
The reduced correlation of the transform coefficients provides an opportunity to apply encoding schemes so as to further reduce the amount of data to be stored or transmitted. Applicable coding schemes are known as such, e.g., Huffmann coding or Amplitude coding.
The seismic data signals to which the method is applied are traces typically obtained from a number of receivers, e.g. geophones or hydrophones (
FIG. 4
, step
310
). These may be arranged in combinations all of which are well known in the prior art. One of these is, for example, a conventional 3-D land seismic layout of linear arrays of geophones arranged in a number of parallel lines. The use of local transformations in the method permits the compression of the data over a certain numbers of receivers contained within each line. A local transformation is one in which the transformation is applied over defined windows of traces, as is well understood in the art. Thus the number of traces over which the local transformation is applied at each successive stage of the transformation is referred to as a spatial window and the window may be varied according to which type of transformation is applied.
The windows of the local transformations are defined by a window function, the window function being chosen so that the transformation is orthonormal and invertible. The window function is chosen so that the transformations are applied over a central window overlapping the adjoining windows (
FIG. 4
, step
312
), most preferably overlapping half of those windows.
The transformation may be performed in two steps, the first step comprising a folding step in which the central window is combined with adjacent half windows to produce a folded signal and the second step comprising compressing a cosine transformation which is performed on the folded signal.
In addition to the local spatial transformation, preferably a local temporal transformation is applied to the data. The combination of both local transformations achieves a better compression ratio. The local temporal transformation is preferably a local temporal discrete sine/cosine transformation of type IV. However other signal transformations and decompositions may be used, such as an ordinary local discrete cosine transformation, and a local fourier transformation. The local spatial trigonometric transformation and local temporal trigonometric transformation may be applied in any order.
The transform coefficients, as representing the original data after the transformation, form a set of data to which different compression methods can be applied (
FIG. 4
, step
316
). These compression methods may be collectively referred to as (re-) quantization (
FIG. 5
, step
332
) and encoding (
FIG. 5
, step
340
).
The quantization process when used for compressing data usually includes a scaling step and a round-off. The quantization process is designed to reduce the high-frequency components or coefficients while maintaining the low-frequency components with higher accuracy.
The scaling is preferably achieved through dividing by first scalar coefficients representing low frequencies and dividing by second scalar coefficients representing high frequencies. The first scalar is chosen to be less than the second scalar, since the larger the scalar, the greater the compression which will be achieved. In this way the coefficients representing the low frequencies which are of particular interest in seismic analysis will not be compressed so much as these representing high frequencies, so that the accuracy of the former is preserved.
The scaling may be achieved by uniform quantizing (
FIG. 5
, step
334
) using a nearest
Batzer William B.
McElheny Jr. Donald E.
Schlumberger Technology Corporation
Wang William L.
LandOfFree
Compression method and apparatus for seismic data does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Compression method and apparatus for seismic data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Compression method and apparatus for seismic data will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2919250