Am I Blue? Finding the Right (Spectral) Balance

Seismic interpreters have always desired to extract as much vertical resolution from their data as possible – and that desire has only increased with the need to accurately land horizontal wells within target lithologies that fall at or below the limits of seismic resolution.

Although we often think of increasing the higher frequencies, resolution should be measured in the number of octaves, whereby halving the lowest frequency measured doubles the resolution.

There are several reasons why seismic data are band-limited.

First, if a vibrator sweep ranges between 8 and 120 Hz, the only “signal” outside of this range is in difficult to process (and usually undesirable) harmonics.

Dynamite and airgun sources may have higher frequencies, but conversion of elastic to heat energy (intrinsic attenuation), scattering from rugose surfaces and thin bed reverberations (geometric attenuation) attenuate the higher frequency signal to a level where they fall below the noise threshold. Geophone and source arrays attenuate short wavelength events where individual array elements experience different statics. Processing also attenuates frequencies. Processors often need to filter out the lowest frequencies to attenuate ground roll and ocean swell noise. Small errors in statics and velocities result in misaligned traces that when stacked preserve the lower frequencies but attenuate the higher frequencies.


Currently there are two approaches to spectral enhancement.

More modern innovations that have been given names such as “bandwidth extension,” “spectral broadening” and “spectral enhancement,” are based on a model similar to deconvolution, which assumes the earth is composed of discrete, piecewise constant impedance layers. Such a “sparse spike” assumption allows one to replace a wavelet with a spike, which is then replaced with a broader band wavelet that often exceeds the bandwidth of the seismic source.

Model-based processing is common to reflection seismology and often provides excellent results – however, the legitimacy of the model needs to be validated, such as tying the broader band product to a well not used in the processing workflow.

We have found bandwidth extension algorithms to work well in lithified Paleozoic shale resource plays and carbonate reservoirs.

In contrast, bandwidth extension can work poorly in Tertiary Basins where the reflectivity sequence is not sparse, but rather represented by upward fining and coarsening patterns.

In this article, we review the more classical workflow of spectral balancing, constrained to fall within the source bandwidth of the data.

Spectral balancing was introduced early in digital processing during the 1970s and is now relatively common in the workstation environment.


As summarized in figure 1, the interpreter decomposes each seismic trace into a suite of 5-10 overlapping pass band filtered copies of the data. Each band-passed filtered version of the trace is then scaled such that the energy within a long (e.g. 1,000 ms) window is similar down the trace.

This latter process is called automatic gain control, or AGC.

Once all the components are scaled to the same target value they are then added back together, providing a spectrally balanced output.

A more recent innovation introduced about 10 years ago is to add “bluing” to the output. In this latter case, one stretches the well logs to time, generates the reflectivity sequence from the sonic and density log and then computes its spectrum. Statistically, such spectra are rarely “white,” with the same values at 10 Hz and 100 Hz, but rather “blue,” with larger magnitude spectral components at higher (bluer) frequencies than at lower (redder) frequencies.

The objective in spectrally balancing then is to modify the seismic trace spectrum so that it approximates the well log reflectivity spectrum within the measured seismic bandwidth.

Such balancing is achieved by simply multiply each band-pass filtered and AGC’d component by exp(+βf), where f is the center frequency of the filter and β is the parameter that is obtained from the well logs that varies between 0.0 and 0.5 (black boxes in figure 1).

There are several limitations to this classic workflow:

First, one balances the measured seismic data, which is the sum of the signal plus noise. Ideally, we want to balance the signal.

Second, since the filters are applied trace by trace, the process as a whole is not amplitude friendly and inappropriate as input to more quantitative amplitude-sensitive analysis such as AVO and post-stack or prestack inversion.

Third, if the AGC window is too small or the statistics of the reflectivity sequence insufficiently smooth (an end member example would be coal bed cyclothems and sabkha sequencies), then reflectors of interest can be suppressed and artifacts created.

A fairly common means of estimating the spectrum of the signal is to cross-correlate adjacent traces to differentiate that part of the signal that is consistent (signal) and that part that is inconsistent (random noise). One then designs the spectral balancing parameters (AGC coefficients) on the consistent part of the data.

Unfortunately, this approach is still not amplitude friendly and can remove geology if the spectra are not smooth.


Figure 2 illustrates a more modern approach that can be applied to both post-stack and prestack migrated data volumes.

First, we suppress crosscutting noise using a structure-oriented filtering algorithm, leaving mostly signal in the data.

Next, the data are decomposed into time-frequency spectral components.

Finally, we compute a smoothed average spectrum.

If the survey has sufficient geologic variability within the smoothing window (i.e. no perfect “railroad tracks”), this spectrum will represent the time-varying source wavelet.

This single average spectrum is used to design a single time-varying spectral scaling factor that is applied to each and every trace. Geologic tuning features and amplitudes are thus preserved.

We apply this workflow to a legacy volume acquired in the Gulf of Mexico:

  • Figures 3a and b show the average spectrum before and after spectral balancing.
  • Figures 3c and d show a representative segment of the seismic data where we see the vertical resolution has been enhanced.

 

Comments (0)

 

Geophysical Corner

Geophysical Corner - Satinder Chopra
Satinder Chopra, award-winning chief geophysicist (reservoir), at Arcis Seismic Solutions, Calgary, Canada, and a past AAPG-SEG Joint Distinguished Lecturer began serving as the editor of the Geophysical Corner column in 2012.

Marcílio Matos is a research scientist for Signal Processing Research, Training and Consulting, and co-investigator for the Attribute Assisted Seismic Processing and Interpretation Consortium at the University of Oklahoma, Norman.

Geophysical Corner - Kurt Marfurt
AAPG member Kurt J. Marfurt is with the University of Oklahoma, Norman, Okla.

Geophysical Corner

The Geophysical Corner is a regular column in the EXPLORER that features geophysical case studies, techniques and application to the petroleum industry.

VIEW COLUMN ARCHIVES

Image Gallery

See Also: Bulletin Article

In this study, seismic models and a Starfak and Tiger Shoal fields data set in the Gulf of Mexico Basin are used to investigate uncertainties caused by the frequency dependence of seismic data and solutions for avoiding pitfalls in seismic-stratigraphic and facies interpretation. Seismic amplitude and instantaneous attributes, along with stratigraphic interpretation of these attributes, are controlled by seismic interference, or tuning, between thin geologic units. Seismic-tuning effects include thickness tuning and frequency tuning, which cause nonlinear variations of reflection amplitude and instantaneous seismic attributes with thickness and/or data frequency. Seismic modeling shows that, whereas thickness tuning determines seismic-interference patterns and, therefore, occurrence of seismic events and seismic facies in layered rock, frequency tuning may further influence the nature of the correlation of seismic data and geologic time and modify seismic facies. Frequency dependence offers a new dimension of seismic data, which has not been fully used in seismic interpretation of geology.

Field-data examples demonstrate that a stratigraphic formation is typically composed of lithofacies of varying thicknesses, and a broadband, stacked seismic data set is not necessarily optimal for stratigraphic and facies interpretation. Although it is difficult to predict correct frequency components for interpretation of not-yet-known geologic targets, local geologic models and well data can be used to optimize the frequency components of seismic data to a certain degree and intentionally modify seismic-interference patterns and seismic facies for better seismic interpretation of geologic surfaces, sediment-dispersal patterns, geomorphology, and sequence stratigraphy.

Desktop /Portals/0/PackFlashItemImages/WebReady/frequency-dependent-seismic.jpg?width=50&h=50&mode=crop&anchor=middlecenter&quality=90amp;encoder=freeimage&progressive=true 3609 Bulletin Article

Oil degradation in the Gullfaks field led to hydrogeochemical processes that caused high CO2 partial pressure and a massive release of sodium into the formation water. Hydrogeochemical modeling of the inorganic equilibrium reactions of water-rock-gas interactions allows us to quantitatively analyze the pathways and consequences of these complex interconnected reactions. This approach considers interactions among mineral assemblages (anorthite, albite, K-feldspar, quartz, kaolinite, goethite, calcite, dolomite, siderite, dawsonite, and nahcolite), various aqueous solutions, and a multicomponent fixed-pressure gas phase (CO2, CH4, and H2) at 4496-psi (31-mPa) reservoir pressure. The modeling concept is based on the anoxic degradation of crude oil (irreversible conversion of n-alkanes to CO2, CH4, H2, and acetic acid) at oil-water contacts. These water-soluble degradation products are the driving forces for inorganic reactions among mineral assemblages, components dissolved in the formation water, and a coexisting gas at equilibrium conditions.

The modeling results quantitatively reproduce the proven alteration of mineral assemblages in the reservoir triggered by oil degradation, showing (1) nearly complete dissolution of plagioclase; (2) stability of K-feldspar; (3) massive precipitation of kaolinite and, to a lesser degree, of Ca-Mg-Fe carbonate; and (4) observed uncommonly high CO2 partial pressure (61 psi [0.42 mPa] at maximum). The evolving composition of coexisting formation water is strongly influenced by the uptake of carbonate carbon from oil degradation and sodium released from dissolving albitic plagioclase. This causes supersaturation with regard to thermodynamically stable dawsonite. The modeling results also indicate that nahcolite may form as a CO2-sequestering sodium carbonate instead of dawsonite, likely controlling CO2 partial pressure.

Desktop /Portals/0/PackFlashItemImages/WebReady/controls-on-co2-fate-and-behavior.jpg?width=50&h=50&mode=crop&anchor=middlecenter&quality=90amp;encoder=freeimage&progressive=true 5727 Bulletin Article

Analog outcrops are commonly used to develop predictive reservoir models and provide quantitative parameters that describe the architecture and facies distribution of sedimentary deposits at a subseismic scale, all of which aids exploration and production strategies. The focus of this study is to create a detailed geological model that contains realistic reservoir parameters and to apply nonlinear acoustic full-waveform prestack seismic inversion to this model to investigate whether this information can be recovered and to examine which geological features can be resolved by this process.

Outcrop data from the fluviodeltaic sequence of the Book Cliffs (Utah) are used for the geological and petrophysical two-dimensional model. Eight depositional environments are populated with average petrophysical reservoir properties adopted from a North Sea field. These units are termed lithotypes here. Synthetic acoustic prestack seismic data are then generated with the help of an algorithm that includes all internal multiples and transmission effects. A nonlinear acoustic full-waveform inversion is then applied to the synthetic data, and two media parameters, compressibility (inversely related to the square of the compressional wave velocity vP) and bulk density, ρ, are recovered at a resolution higher than the shortest wavelength in the data. This is possible because the inversion exploits the nonlinear nature of the relationship between the recorded data and the medium contrast properties. In conventional linear inversion, these details remain masked by the noise caused by the nonlinear effects in the data. Random noise added to the data is rejected by the nonlinear inversion, contributing to improved spatial resolution. The results show that the eight lithotypes can be successfully recovered at a subseismic scale and with a low degree of processing artifacts. This technique can provide a useful basis for more accurate reservoir modeling and field development planning, allowing targeting of smaller reservoir units such as distributary channels and lower shoreface sands.

Desktop /Portals/0/PackFlashItemImages/WebReady/acoustic-nonlinear-full-waveform-inversion-book-cliffs-utah.jpg?width=50&h=50&mode=crop&anchor=middlecenter&quality=90amp;encoder=freeimage&progressive=true 3253 Bulletin Article

See Also: CD DVD

Desktop /Portals/0/images/_site/AAPG-newlogo-vertical-morepadding.jpg?width=50&h=50&mode=crop&anchor=middlecenter&quality=90amp;encoder=freeimage&progressive=true 4071 CD-DVD

See Also: Energy Policy Blog

The Organization of Petroleum Exporting Countries (OPEC, including Saudi Arabia, UAE, Venezuela, Algeria,  Indonesia, Iraq, Iran, Kuwait, Libya, Nigeria and Qatar) seems less of a menace now that the United States oil and natural gas production is booming. But OPEC continues to be a major exporter that can influence global oil supply and prices. However, OPEC countries are very dependent on export revenues to provide the jobs and services demanded by their populations, so a major shift in production volumes and prices is unlikely.

Desktop /Portals/0/PackFlashItemImages/WebReady/remember-opec-2014-05may-07.jpg?width=50&h=50&mode=crop&anchor=middlecenter&quality=90amp;encoder=freeimage&progressive=true 12897 Energy Policy Blog