ELASTIC-WAVE EQUATION TRAVEL-TIME AND WAVE-FORM INVERSION OF CROSSWELL DATA

Citation
Cx. Zhou et al., ELASTIC-WAVE EQUATION TRAVEL-TIME AND WAVE-FORM INVERSION OF CROSSWELL DATA, Geophysics, 62(3), 1997, pp. 853-868
Citations number
25
Categorie Soggetti
Geochemitry & Geophysics
Journal title
ISSN journal
00168033
Volume
62
Issue
3
Year of publication
1997
Pages
853 - 868
Database
ISI
SICI code
0016-8033(1997)62:3<853:EETAWI>2.0.ZU;2-Z
Abstract
A method is presented for reconstructing P- and S-velocity distributio ns from elastic traveltimes and waveforms. The input data consist of c rosswell hydrophone records generated by a piezoelectric borehole sour ce. Borehole effects are partially accounted for by using a low-freque ncy Green's function to simulate the pressure generated in the fluid-f illed receiver well. The tube waves in the borehole are ignored, on th e assumption that they can be removed from the field data by median fi ltering. In addition, the source-radiation pattern is partially taken into account by inverting for the equivalent stress components acting on the earth at the source location. The elastic wave equation travelt ime and waveform inversion (WTW) method is applied to both synthetic c rosswell data and the McElroy field crosswell data. As predicted by th eory, results show that elastic WTW tomograms provide a sharper interf ace image than delineated in the traveltime tomograms. The spatial res olution of the McElroy traveltime tomogram is about 20 m compared to a bout 3 m and 1.5 m, respectively, for the associated P- and S-velocity WTW tomograms. From these tomograms, detailed porosity maps of the in terwell geology are constructed. There is a very good correlation betw een the P-velocity tomograms and the P-velocity log profiles, and ther e is a good correlation between the smooth parts of the S-velocity tom ogram and the S-velocity logs. Unfortunately, the high-wavenumber part s of the S-velocity tomograms do not correlate well with the high-wave number parts of the S-velocity logs. We believe this problem is partly caused by not taking into account attenuation effects in the WTW algo rithm.