.png){kind=icon}
.png){kind=icon}
Return
Prediction of Fault and Fracture Density From Seismic Attributes and Machine Learning Models Calibrated With Borehole Image Logs
.png){kind=icon}
OA
AI
Y
R
A
Y
DOI：10.1111/bre.70104.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Faults and fractures play a key role in the productivity of shale gas reservoirs by affecting porosity, permeability, and gas flow paths. However, direct characterization of these small-scale features using traditional seismic methods faces significant challenges due to resolution limitations. Using a post-stack 3D seismic data set from the Jiaoshiba area in the southeastern Sichuan Basin, multiple post-stack seismic attributes and a set of convolutional neural network (CNN)-based methods from Geoteric are applied to the shale gas reservoir of Wufeng-Longmaxi Formation for deformation characterization. Wellbore fault and fracture counts were derived from Formation Micro Image (FMI) log interpretations and core sample observations, providing a reliable data set for cross-validation. This analysis involves cross-plotting seismic characterization results with borehole deformation counts to measure how these predictions correlate with wellbore data. Results indicate a near-linear correlation between borehole deformation count and seismic characterization methods, with the Combined AI network and Birch model, as well as dip illumination attribute, showing the strongest correlation. These findings suggest that seismic methods can be effective as a proxy for predicting sub-seismic scale deformation. Furthermore, pressure coefficients, porosity and gas contents have been cross-plotted with wellbore deformation counts, indicating these important factors for shale gas exploration are correlated with deformations and seismic deformation characterization methods. A prediction of porosity, pressure coefficients, and gas contents was conducted with their correlation with these seismic characterization methods. This research highlights the potential of integrated seismic and machine learning approaches in enhancing the predictability and understanding of geological deformations at scales not directly observable by direct seismic techniques and proposes an alternative method for predicting key parameters for shale gas exploration.
Keywords:
convolutional neural network (CNN)
faults and fractures
formation micro image (FMI)
Jiaoshiba area
seismic attributes
shale
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
2.6
Papers:
106
Citations:
4.2K
Organization
Cited Papers