.png){kind=icon}
.png){kind=icon}
Return
Time-Lapse Seismic Data Matching by a Two-Step Framework
.png){kind=icon}
PRE
AI
DOI：10.1109/TGRS.2026.3655441.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Time-lapse seismic technology is a vital tool for reservoir monitoring. In onshore time-lapse seismic applications, achieving repeatable data acquisition poses significant challenges. Processing time-lapse seismic data to suppress nonrepeatable noise is crucial for accurate reservoir monitoring. A common approach to achieve repeatability is to match the baseline and monitor data, and we decompose the matching problem into two key part: time registration and wavelet correction. For time registration, we employ the time-shift analysis and have provided a novel noniterative local wavelet similarity (LWS) method to generate an efficient and stable time-shift analysis map, which can effectively avoid cycle skipping in conventional methods. We use the signal peaks to adaptively construct molecular windows that divide original signal into local wavelet segmentations. Then, the similarity map is calculated using those local segmentations. For wavelet correction, we develop a reservoir-preserved matching filter (MF) method to compensate for the residual time, amplitude, phase, and frequency distortions. We use the prior well-logging information to delineate the reservoir section by predictive painting. Then, the filter coefficients are obtained by a block-wise joint inversion which use different data fitting terms in reservoir and nonreservoir sections, and is designed to incorporate the reservoir information to avoid the damage to the reservoir image. In synthetic and field data tests, the proposed similarity measurement illustrates higher robustness and accuracy. The high consistency between the matched result and the theoretical images also indicates that the proposed method can effectively eliminate the nonrepeatable noise and enhance the repeatability of time-lapse seismic data. Moreover, compared with conventional nonstationary MF, the proposed method can offer a more reasonable result.
Keywords:
Local similarity
matching filtering
seismic image registration
time-lapse seismic
Journal
IF:
8.6
Papers:
2.1W
Citations:
10.7W