Statistical facies classification from multiple seismic attributes: comparison between Bayesian classification and expectation–maximization method and application in petrophysical inversion

Grana, Dario; Lang, Xiaozheng; Wu, Wenting

doi:10.1111/1365-2478.12428

Cited by 26 publications

(3 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These models were classified into distinct water masses. Bayesian classification (e.g., Avseth et al, 2005;Grana et al, 2017) was trained based on the existing direct measurements of temperature and salinity profiles of spatially located nearby Argo floats taken during the acquisition of the seismic oceanography data. N w different types of water were identified including the Central Atlantic Water, the Mediterranean Outflow and the Subarctic Intermediate Waters, as described in Comas-Rodríguez et al (2011).…”

Section: Number Of Channels 636mentioning

confidence: 99%

Geostatistical Seismic Inversion for Temperature and Salinity in the Madeira Abyssal Plain

et al. 2021

View full text Add to dashboard Cite

A two-dimensional multichannel seismic reflection profile acquired in the Madeira Abyssal Plain during June 2016 was used in a modeling workflow comprising seismic oceanography processing, geostatistical inversion and Bayesian classification to predict the probability of occurrence of distinct water masses. The seismic section was processed to image in detail the fine scale structure of the water column using seismic oceanography. The processing sequence was developed to preserve, as much as possible, the relative seismic amplitudes of the data and enhance the shallow structure of the water column by effectively suppressing the direct arrival. The migrated seismic oceanography section shows an eddy at the expected Mediterranean Outflow Water depths, steeply dipping reflectors, which indicate the possible presence of frontal activity or secondary dipping eddy structures, and strong horizontal reflections between intermediate water masses suggestive of double diffuse processes. We then developed and applied an iterative geostatistical seismic oceanography inversion methodology to predict the spatial distribution of temperature and salinity. Due to the lack of contemporaneous direct measurements of temperature and salinity we used a global ocean model as spatial constraint during the inversion and nearby contemporaneous ARGO data to infer the expected statistical properties of both model parameters. After the inversion, Bayesian classification was applied to all temperature and salinity models inverted during the last iteration to predict the spatial distribution of three distinct water masses. A preliminary interpretation of these probabilistic models agrees with the expected ocean dynamics of the region.

show abstract

Section: Number Of Channels 636mentioning

confidence: 99%

Geostatistical Seismic Inversion for Temperature and Salinity in the Madeira Abyssal Plain

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Some studies use the aid of seismic attributes (e.g. P wave impedance, S wave impedance and density) from inversion to generate the facies model in the reservoir through classification (Roncarolo & Grana, 2010;Grana, 2016;Tellez et al, 2017). Since elastic inversions provide only information about the seismic bandwidth, some studies avoid this problem by combining the inversion with the low frequency model, in addition to petrophysical information, for the construction of the facies model (Sams & Saussus, 2013;Zabihi Naeini & Exley, 2017).…”

Section: Introductionmentioning

confidence: 99%

3D lithofluid facies classification by machine learning algorithms

Aguiar¹,

Oliveira²,

Matos³

et al. 2021

Proceeding of the 17th International Congress of the Brazilian Geophysical

View full text Add to dashboard Cite

There must be an adequate integration of well and seismic data to create a reliable facies model. However, heterogeneous rocks, lack of data and inaccurate information accumulate uncertainties for model characterization. Here we seek to simulate a scenario where there are few wells and only one seismic data to create the facies model. The main goal is to estimate the predominant lithology distribution with fluid information. This strategy can help in new wells' locations and reducing the risk of false discoveries. The Norne field located on the Norwegian Sea will be used as an example for this methodology. Machine learning techniques have been commonly used for automation and task optimization processes. We show that these methods can achieve high performance when receiving proper processed data. The summarized workflow can be described in two stages: electrofacies classification on eight different wells and use of these facies as labels for classification on the field's seismic volume. The results indicate accuracy close to 90% both in well and seismic scenarios. The facies model with lithofluid information is still compared with the reservoir simulator from the same period to show the high similarity.

show abstract

“…Cluster analysis methods attempt to group various attributes of interest within some generally multi-dimensional feature-(or attribute-) space, based on some distance measure between each data point from the centres of identified groups of data which are commonly referred to as clusters. The probabilities that each data point belongs to each of the identified clusters can then be estimated by assigning a probability distribution to each of the clusters, for example using a multivariate Gaussian distribution in a Gaussian mixture model (GMM) which is a term used to describe a sum of Gaussians, e.g., Grana et al (2016). A typical problem with such cluster analysis methods is that they estimate probabilities with high entropy (uncertainty) for those data points that fall equidistant from cluster centres, or in cases where there is a significant overlap between different cluster distributions.…”

Section: Introductionmentioning

confidence: 99%

Variational Bayesian inversion (VBI) of quasi-localized seismic attributes for the spatial distribution of geological facies

Nawaz

Curtis

2018

Geophysical Journal International

View full text Add to dashboard Cite

We introduce a new Bayesian inversion method that estimates the spatial distribution of geological facies from attributes of seismic data, by showing how the usual probabilistic inverse problem can be solved using an optimization framework while still providing full probabilistic results. Our mathematical model consists of seismic attributes as observed data, which are assumed to have been generated by the geological facies. The method infers the post-inversion (posterior) probability density of the facies plus some other unknown model parameters, from both the seismic attributes and geological prior information. Most previous research in this domain is based on the localized likelihoods assumption, whereby the seismic attributes at a location are assumed to depend on the facies only at that location. Such an assumption is unrealistic because of imperfect seismic data acquisition and processing, and fundamental limitations of seismic imaging methods. In this paper, we relax this assumption: we allow probabilistic dependence between seismic attributes at a location and the facies in any neighbourhood of that location through a spatial filter. We term such likelihoods quasi-localized. Exact Bayesian inference is impractical because it requires normalization of the posterior distribution which is intractable for large models and must be approximated. Stochastic sampling (e.g., by using Markov chain Monte-Carlo-McMC) is the most commonly used approximate inference method but it is computationally expensive and detection of its convergence is often subjective and unreliable. We use the variational Bayes method which is a more efficient alternative that offers reliable detection of convergence. It achieves this by replacing the intractable posterior distribution by a tractable approximation. Inference can then be performed using the approximate distribution in an optimization framework, thus circumventing the need for sampling, while still providing probabilistic results. We show in a noisy synthetic example that the new method recovered the coefficients of the spatial filter with reasonable accuracy, and recovered the correct facies distribution. We also show that our method is robust against weak prior information and non-localized likelihoods, and that it outperforms previous methods which require likelihoods to be localized. Our method is computationally efficient, and is expected to be applicable to 3D models of realistic size on modern computers without incurring any significant computational limitations.

show abstract

Statistical facies classification from multiple seismic attributes: comparison between Bayesian classification and expectation–maximization method and application in petrophysical inversion

Cited by 26 publications

References 35 publications

Geostatistical Seismic Inversion for Temperature and Salinity in the Madeira Abyssal Plain

Geostatistical Seismic Inversion for Temperature and Salinity in the Madeira Abyssal Plain

3D lithofluid facies classification by machine learning algorithms

Variational Bayesian inversion (VBI) of quasi-localized seismic attributes for the spatial distribution of geological facies

Contact Info

Product

Resources

About