Interval Velocity Estimation Using Bayesian Inversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating Dix interval velocities from RMS velocity data are uncertain due to inaccuracies in identifying primary reflection events and lack of consideration for geological knowledge, leading to inconsistent velocity data and difficulty in interpreting seismic images for hydrocarbon exploration.
Innovation Solution
A method that provides an initial model with associated uncertainties for interval velocities and RMS velocities, using Bayesian inversion or constrained least-squares techniques to estimate a second model with uncertainties, incorporating geological knowledge and statistical uncertainties to improve the accuracy of interval velocity estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional Dix inversion methods are used to estimate interval velocities from RMS velocity data, then the process is simple and quick, but the results have high uncertainty and inconsistency due to inaccuracies in identifying primary reflection events
Solution Approach 1:
The method performs preliminary identification and selection of primary reflection events from seismic data before conducting the Dix inversion. By pre-processing the seismic data to distinguish primary reflections from multiples and noise, the method ensures that only reliable reflection events are used in the velocity inversion, thereby reducing uncertainty in the interval velocity estimates without significantly increasing overall complexity
Solution Approach 2:
The method incorporates iterative feedback mechanisms where the inversion process uses uncertainty estimates from initial velocity models to guide subsequent refinement steps. The uncertainty information feeds back into the inversion algorithm to adjust weighting and constraints, progressively improving the accuracy of interval velocity estimates while managing computational complexity through adaptive refinement
2Reliability
If RMS velocity data is used directly without considering geological knowledge, then the processing is straightforward, but the velocity models lack consistency with actual subsurface structures
Solution Approach 1:
The method transforms the inversion process by incorporating geological knowledge as constraints on velocity parameters. By modifying the parameter space to include geologically realistic bounds, trends, and relationships (such as velocity-depth trends and lateral continuity), the method ensures that resulting velocity models are consistent with subsurface geological structures while maintaining processing feasibility through parameter regularization
Solution Approach 2:
The method introduces geological knowledge as an intermediary layer between the raw RMS velocity data and the final interval velocity model. This intermediary incorporates domain expertise in the form of prior geological models, expected velocity trends, and structural constraints that mediate the inversion process, guiding the solution toward geologically realistic results while filtering out artifacts from noisy or ambiguous seismic data
3Loss of information
If conventional velocity inversion is performed without uncertainty analysis, then the computation is faster, but the resulting velocity data cannot be reliably interpreted for hydrocarbon exploration
Solution Approach 1:
The method implements uncertainty analysis at critical stages of the inversion process rather than throughout the entire workflow. By applying uncertainty propagation and analysis selectively to key steps such as primary reflection identification and final velocity model generation, the method retains essential uncertainty information while avoiding the computational burden of full uncertainty propagation through every processing stage, thus balancing information retention with processing efficiency
Data Source
AI summary
A method of estimating a velocity of a geological layer includes a. providing a first, initial model including an interval velocity associated with a subsurface location and an uncertainty associated with the interval velocity; b. providing data including an actual or approximated root-mean-square (RMS) velocity associated with a subsurface location and an uncertainty associated with the RMS velocity; and c. estimating a second model including an interval velocity associated with a subsurface location and an uncertainty associated with the interval velocity, based on the interval velocity and the uncertainty of the first model, and the RMS velocity and the uncertainty of the data.