Application of SP in Mexico

8/2/2019 Application of SP in Mexico

1/5

Recent reservoir studies involvingspectral decomposition on various data

sets from the Burgos and Macuspanabasins of Mexico document the useful-ness of this method as another way touncover the effects of hydrocarbonaccumulations on seismic data. Threesuch effects are illustrated in this arti-cle; attenuation of seismic waves pass-ing through the reservoir, preferentialreservoir illumination, and differentialreservoir reflectivity.

Such phenomena can be clearly seenin these seismic studies only because ofthe use of a new method of spectraldecomposition, the wavelet transform.

Traditionally, spectral decompositionhas used various techniques thatrequire the use of a time window (e.g.Fast Fourier transform, discrete Fouriertransform, maximum entropy, etc.).These windows produce serious dis-tortion of true spectra and/or severelylimit the vertical resolution; discretespectral energy from individual reflec-tion events that fall inside the windoware mixed together, and when the win-dow length is shortened to minimizethis problem, frequency discriminationis compromised. Wavelet transforms,on the other hand, have no such win-

dowing requirements and thereforeavoid this distortion. Note that spectraldecomposition, like inversion, isnonunique, but the wavelet transformgenerally does a vastly superior job ofrepresenting the individual waveletspectra than windowed methods. Thisis the primary reason why analysesusing the traditional techniques have

been limited to map extractions. Thewindowed techniques do not preservethe individual reflection events andintroduce distortions and artifacts. Theability to analyze the spectra of indi-vidual reflections is a major leap for-ward in the effectiveness of spectraldecomposition.

Burgos Basin studies. Two studies innortheastern Mexico illustrate the phe-nomena of differential reservoir reflec-tivity and attenuation of seismic wavespassing through the reservoir. Figure 1shows the location of Burgos Basin andAlondra and Numerador fields.Alondra produces from clean Midway-age sands with thickness of approxi-mately 20 m and with porosities of

1130 T HE LEADING EDGE NOVEMBER 2003

Application of spectral decomposition to gas basins in Mexico

MICHAEL DEANBURNETTand JOHNPATRICKCASTAGNA, Fusion Geophysical, Norman, Oklahoma, U.S.EFRANMNDEZ-HERNNDEZ, GENAROZIGA RODRGUEZ, LEONEL FIGNGARCA, JOSTRINIDADMARTNEZVZQUEZ, MARIANOTLLEZAVILS, and RAL VILA

VILLASEOR, Petrleos Mexicanos, Ciudad Pemex, Mexico

Figure 1. The Burgos Basin and the locations of Alondra and Numerador studies.

Figure 2.Midway gas reservoir showing differential reflectivity. The reservoir is not anomalousat 20 Hz but exhibits anomalous reflectivity at 40 Hz.

Downloaded 11 Jul 2011 to 119.46.218.1. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/


2/5

13-18% (low impedance) at a depth ofapproximately 2000 m. Figure 2 showstwo isofrequency panels and the inlineused as input. At 20 Hz the reservoirdoes not exhibit an anomalous reflec-tivity compared with the surroundinggeology. But at 40 Hz, the gas chargedsand clearly stands out. This behaviorcan be partially explained by the factthat the time-thickness of this sand is

close to tuning; it will appear as adipole. When filled with gas, the reflec-tivity of the sand is higher and theamplitude at all frequencies isincreased over the adjacent brine-filledgeology and the center frequency ofthe reflection from the reservoir isincreased slightly. This same effect can

be duplicated in synthetic models con-structed using local well logs.

This effect becomes spectacularwhen viewed in frequency maps as thatfrom Numerador Field (Figure 3). Thediscovery well produces gas from a

high porosity, low-impedance Jacksonsandstone which is 17 m thick at adepth of 500 m. This field is trapped tothe south and southwest by a promi-nent fault and by dip to the north andeast. The line in Figure 3 crosses theNumerador well. The sands here arelocally conformable and the phenom-ena of attenuation can be demonstratedwith this reservoir.

NOVEMBER 2003 T HE LEADING EDGE 1

Figure 3.Arbitrary line across Numerador Field showing a gas accumulation in a shallowJackson sandstone.



3/5

Since this gas accumulation can beobviously identified via an amplitudemap, it is instructive to compare thefull-spectrum conventional amplitudemap with its equivalent at discrete fre-quencies. Figure 4 shows the reservoirfault block map of conventional ampli-tudes of the trough after residual phaseremoval, along with two frequencymaps of the same horizon. Spectral sta-

bilization has been applied to removethe wavelet spectra from these maps;without this step, overall map ampli-

tudes between frequencies are domi-nated by wavelet spectra and discern-ing true spectral behavior of the geologyis difficult.

Attention is directed toward thearea immediately north of theNumerador well. While both the con-ventional amplitude map and the fre-quency maps reveal the reservoirsorientation, the frequency maps showa different distribution of amplitudeswithin the reservoir. One reason for thismight be reservoir thickness.


Figure 4. (a) Standard amplitude map of pay. (b) 30-Hz frequency map. (c) 60-Hz frequency

map. Note the subtle differences between the frequency maps, especially the hole in amplitudeimmediately north of the Numerador at 50 Hz. This hole is probably due to a local thickening ofthe reservoir.

Figure 5. Tuning chart for extracted wavelet and individual frequencies at 30 Hz and 60 Hzversus theoretical tuning frequencies and observed spectral behavior at Numerador. Comparethese to the maps in Figure 4.

a b c



4/5

In order to investigate this possibil-ity, Figure 5 contains the full wavelettuning characteristics and those of theindividual frequencies shown in Figure

4, along with the theoretical tuning fre-quency chart for a reflector with thick-ness equal to that observed atNumerador. The apparent seismic timethickness at the Numerador well(trough-peak) is about 12 ms (close tomaximum tuning for the full-spectrumwavelet); theoretical maximum ampli-tude for this thickness should occur atabout 21 Hz. Observed amplitudes ateach frequency were extracted from themaps at Numerador and then normal-ized and plotted to compare withexpected results. This curve shows thatmaximum amplitude at Numerador

occurs at about 28 Hz. The rate of ampli-tude decay above that frequency doesnot match the expected spectral behav-ior from this known reservoir thickness.Therefore, other factors are also con-tributing to the spectral behavior of thesedata. In all three maps, for instance,acquisition/processing artifacts canclearly be observed. Most importantly,however, the reservoir is not a simple iso-lated layer as assumed in the tuningcharts. Variable reflectivity structuresrequire more sophisticated methods forthickness determination. Thus, the spec-

tral behavior is due to a combination offactors at Numerador. Work continueson this example to try to better under-stand these results. Such investigationsare impossible with techniques that usewindows in the calculations.

Figure 6 shows a map of averageamplitudes in a 50-ms window approx-imately 10 ms below the reservoir at 20Hz. This low amplitude (which is alsopresent at all other frequencies) is evi-dence of the attenuation of the reser-

NOVEMBER 2003 T HE LEADING EDGE 1

Figure 7.Location of

MacuspanaBasin.

Figure 8. Seismic section showing structural/stratigraphic features of the Cobo/Bitzal area.

Figure 6.Average amplitudes under the reser-voir at 20 Hz.



5/5

voir of seismic waves passing throughthe gas column. Note that the low-ampli-tude area corresponds to the high-ampli-tude area in Figure 4 and is consequentlyinterpreted to be caused by gas-relatedattenuation or transmission loss.

Macuspana Basin example. Figure 7shows that Macuspana Basin is justsouthwest of the Yucatan peninsula.Production here is from thin Plio-Plistocene sands thought to be fluvialin origin. These sands can be quite dis-

continuous, with reservoir pressureand water levels indicating complexreservoir geometry (Figure 8). Spectraldecomposition was successfully usedhere to indicate lateral heterogeneity.Figure 9 shows two isofrequency pan-els with the reservoir indicated. Thereservoir appears continuous on the25-Hz panel but is discontinuous on the35-Hz panel. In fact, the reservoir has a

different pressure in these two wells,indicating that the discontinuity evidentonly on the 35-Hz panel is likely to bereal. Bitzal 11 is now plugged but Bitzal9 is still producing. The fact that thereservoir amplitude extends further tothe left on the 35-Hz panel suggests thatthe reservoir is thinning in that direction.The idea here is that the gas in the reser-voir changes the resonant frequency(shifting this frequency upward asshown in the first example). This makesthe reservoir (and gaps between sand

bodies) easier to see when illuminated

at the resonant frequency. This type ofanalysis would not be possible withoutthe ability to resolve individual reflec-tion events, as the wavelet transformallows us to do.

Conclusion.We have shown examplesof spectral decomposition usingwavelet transforms applied to prob-lems in two basins in Mexico. This tech-nique has been shown to be a valuableaid in the search for hydrocarbons inthese gas-prone areas as in the directdetection of hydrocarbons and as anindicator of stratigraphic variability. TLE

Ac kn owl ed gm en ts : We th an k Pe tr ol eo sMexicanos for permission to show these results.Thanks also to Carrie Decker and Shengjie Sun

for their contributions.

Corresponding author: [email protected]


Figure 9. Isofrequency panels showing the reservoir to be more anomalous and resolved better at35 Hz (bottom) than at 25 Hz (top); the reservoir pressures in the two wells are different. Noticein particular how much farther to the left than the reservoir extends on the 35-Hz section.

Application of SP in Mexico

Documents

Transcript of Application of SP in Mexico