A scale attribute for texture in well- and seismic data
Journal
SEG Technical Program Expanded Abstracts
Journal Volume
19
Journal Issue
1
Date Issued
2000-01-01
Author(s)
Herrmann, Felix
Abstract
For many years people have been struggling to integrate w ell and seismic data. As main reasons for this struggle one may list the inherent bandwidth limitation of seismic data; the problem of seismic amplitudes and finally the apparent inabilit y to delineate and characterize those transitions in w ells that can be linked to and held responsible for the major re ection events and their signatures. As a means of approaching these problems an alternative approach is presen ted and partly tested in this paper. The method is based on the in troduction of a new at- tribute which, at a fixed scale, characterizes the location and sharpness of the transitions in the well and the loca- tion and signature of the corresponding re ection events. As such the method entails an approach where the vari- ations in the magnitude of the well properties and re ec- tion amplitudes are characterized by their order of magni- tudes. These orders of magnitude are represented by scale exponents, whic h express the localtexture (sharpness) of the interfaces and the nature of the re ection signature. It appears that these exponents are robust and localized and truly independent of the actual point values in the log and seismic data. This independence makes these exponents righ tful candidates for a new trace attribute, complement- ing existing attributes such as instan taneousphase and frequency. Because sharpness characterizes the local tex- ture the attribute promises to serve as a key quantity for the integration of well and seismic data on the level of tex- ture, with the additional benefit of being able to generate pseudo wells and seismic data. The proposed method to obtain the scale exponents w as first in troduced in Herrmann and Stark (1999; 2000; 2000). There it w as sho wn that, as opposed to multi- ple scale wavelet methods, it is possible to estimate frac- tal scale exponents at a fixe d scale b y the monoscale fi- transform. Similar attempts (Dessing, 1997) ha vebeen made, using the instantaneous phase, which is difficult to compute from well and/or seismic data. As shown below the application of the monoscale method to m ulti-trace post-stac k migrated data may be regarded as highly suc- cessful in the sense that the attribute assigns only one value to a re ector, in a manner that is lateral consistent and insensitiv eto amplitude variations along the inter- face. This lateral consistency makes the scale attribute useful for revealing the (singularity) structure of the sub- surface. Besides the strict locality additional advan tages are phys- ical interpretation; scale-in variance; insensitivit yto the seismic w aveletand reconstruction capabilit y. The lo- cality opposessmeared attributes suc h as instantaneous frequency/phase while the scale-in variancerefers to the scale-invarian t manner in which the sharpness of the re- ectors is characterized. This sharpness characteriza- tion has the advantage that it also applies to the seismic w avelet and is additative under convolution. This latter propert yfacilitates both the interpretation and seismic w avelet deconovlution on the level of the attribute. Fi- nally, the reconstruction capability allows for the genera- tion of pseudo wells and re ectivity, based on the location and sharpness characterization of the attribute. The setup of this paper is as follows. Generalized layer transitions of varying sharpness are introduced first, fol- lowed by a brief outline of the fixed scale analysis method. Subsequently, attention is paid to the directionalit yof fractional order transitions and the construction of pseudo w ells and seismic data from the scale attributes. Finally, the method will be put to the test on the Mobil A V O dataset and some re ection traces from the Gulf of Mex- ico. The well of the first data set is "blocked", post stack migrated data is analyzed and reconstructed.
Type
journal article