4D Signature Estimation Improvement for Time-Lapse Seismic
by E.A. Al-Hemyari1, M. Jervis1, and R. Burnstad1
A new seismic amplitude intersection attribute is introduced for better 4D signature estimation using time-lapse seismic surveys. The new attribute is defined as the trace having amplitudes common to any two input traces where non-repeatable amplitudes on either trace are rejected. Hence, this attribute reduces the effect of incoherent 4D noise and provides an improved 4D signal estimate compared to straight differencing of time-lapse surveys. The key aspect in the estimation approach is in the use of multiple time lapse surveys to suppress the 4D noise and better estimate the 4D signal. For a given time-lapse survey, first we start by generating several intersections with other surveys, each representing a different signal estimate of the given survey. Then, all the signal estimates are stacked to produce an ideal signal model of the given survey. This estimation approach has been applied to simple synthetic and field data cases to evaluate this new attribute for suppressing 4D noise and improving the estimation of the 4D signal. A simple synthetic dataset comprising 11 CDPs was generated to simulate time-lapse acquisition with consecutive reservoir changes of 2%. A comparison is shown between using straight trace differencing and using the intersection attribute with the stacking estimation approach, to estimate reservoir changes. At noise levels comparable to the 2% reservoir change, the subtraction approach fails to resolve the reservoir changes. Using the approach described here gives less noisy estimates of the 4D signal. For the field data test, six 2-D time-lapse surveys acquired in 2011 in Eastern Saudi Arabia have been used to test the same approach. In this field dataset, there was no expected reservoir change. Hence, the suppression of 4D noise is tested. Six CDPs were extracted and a comparison is shown between using subtraction and using the stacked intersections to estimate 4D reservoir change. Stacked intersections show a decrease in NRMS between two consecutive surveys by almost a factor of two, compared to straight subtraction. This reduction is due to a reduction in 4D noise, which should allow detection of more subtle 4D signals than conventional methods of analysis.
AAPG Search and Discovery Article #90188 ©GEO-2014, 11th Middle East Geosciences Conference and Exhibition, 10-12 March 2014, Manama, Bahrain