Chapter 1 - Introduction to Seismic Attributes

01-01 - Why Utilize Seismic Attributes? (19 min.) Sample Lesson
01-02 - Review of Seismic Fundamentals (23 min.)
01-03 - Tuning & Resolution (17 min.)
01-04 - Instantaneous and Complex Trace Attributes (17 min.)
01-05 - Lab: Software Installation and Seismic Data Loading (9 min.) Quiz: 01-05 - Lab: Software Installation and Seismic Data Loading

Chapter 2 - Fundamental Seismic Attributes

02-01 - Geometric Attributes and Their Applications (25 min.)
02-02 - Frequency Domain Attribute Analysis (17 min.)
02-03 - Multispectral Coherence (17 min.)
02-04 - Enhancement and Filtering of Attribute Data (14 min.)
02-05 - Lab: Computing Spectral & Geometric Attributes (38 min.)
02-06 - Lab: Seismic Amplitude Data Conditioning (5 min.) Quiz: 02-06 - Lab: Seismic Amplitude Data Conditioning

Chapter 3 - Multi-Attribute Analysis

03-01 - Visualization Techniques for Attribute Analysis (12 min.)
03-02 - Multiattribute Integration and Interpretation (10 min.)
03-03 - Interactive Multi-Attribute Analysis Using Visualization (5 min.) Quiz: 03-03 - Interactive Multi-Attribute Analysis Using Visualization

Chapter 4 - Machine Learning with Seismic Attributes

04-01 - Dimension Reduction for Multi-Attribute Analysis (24 min.)
04-02 - Unsupervised Clustering of Seismic Attributes (28 min.)
04-03 - Lab: Unsupervised Clustering Using K-means (6 min.)
04-04 - Lab: Unsupervised Clustering Using SOM and GTM (7 min.) Quiz: 04-04 - Lab: Unsupervised Clustering Using SOM and GTM

Seismic Attribute Analysis for Reservoir Characterization / Chapter 1 - Introduction to Seismic Attributes

Lesson 01-01 - Why Utilize Seismic Attributes?

Access All SAGA Wisdom Training content Subscribe

Already a member? Sign in

Access All SAGA Wisdom Training content Subscribe

Already a member? Sign in

Transcript

01. Lesson 1.01: Why Utilize Seismic Attributes?02. Reference Books and Acknowledgements 03. Why Use Seismic Attributes?04. Seismic Facies Recognition 05. A Suite of Seismic Facies: Volcanic MTC 06. Human-Driven Interpretation 07. Interpretation Bias 08. Confirmation Bias Driven Interpretation 09. Attribute Taxonomy Based on What They Measure 10. Attribute Taxonomy Based on the Input Data 11. 6 Common Uses for Seismic Attributes 12. To Validate Your Subjective Interpretation 13. To Intelligently Extrapolate Sparse Well Measurements 14. To Reconstruct the Tectonic, Depositional and Diagenetic History 15. To Communicate Your Play in a Timely Manner to Non-Geoscientists 16. To Evaluate the Effectiveness of Alternative Processing Parameters 17. As Input to Machine Learning 18. Seismic Attributes - Key Concepts

01. Lesson 1.01: Why Utilize Seismic Attributes?

Welcome to Attributes. I'm going to get going by talking about why you might want to utilize seismic attributes, and then I'll go over some seismic fundamentals just to make sure we're kind of on the same page, and then we'll dive right into talking about themselves in a later module.
So I wanted to point out here that this course has some optional labs that you can do if you want to explore seismic attributes more. If you want to access the software, the AASPI software, all you have to do is send me an email at hbedle@ou.edu and I'll send you a downloadable file so that you can use the AASPI software and follow along in the labs. These labs have very detailed instructions. You'll be able to explore the attributes more than we can in this short course, around with parameters, and you can even use the software on your own seismic data if you want to. So let's get started.

02. Reference Books and Acknowledgements

So one of the things I want to also start off mentioning is that this is a fairly short course, and I've spent my career studying attributes and related algorithms. So there's a lot of other information out there should any topic interest you. There's a lot of information on my website, which I have on the left-hand side of the page. We have a lot documentation on every seismic attribute from the fundamentals, details about the math, and then seismic use cases. And if you want, you can also look at some textbooks that Kurt Marfurt and Satinder Chopra have written over the last decade or so that also cover and their uses for .

03. Why Use Seismic Attributes?

So getting started, I want to kind of start off this course to hopefully incentivize you to finish the course. So let me tell you about some of the reasons that you may want to use attributes.

04. Seismic Facies Recognition

And so in order to understand why you may wanna use them, let's just start off by looking at a cross-section. So here we have a vertical seismic cross-section. We're looking at a seismic amplitude plotted in a gray scale. This is the Kora 3D dataset, so we're looking the Kora volcano, kind of here in this transparently seismic area. We have a few well logs in the area to tie our seismic wiggles to actual and rocks. And as a seismic interpreter, when we look at it, we're often looking for different seismic so we can understand what's going on in the subsurface. So in this figure, we have a lot of different interesting (which is why I like to show it), fromwhich give a bright reflection, to a lot of sedimentary kind of induced features from small-scale slope channels to this progradation that's happening and kind of prograding out into the Taranaki on the left-hand side of the screen. We've got basin floor fans that are stacked on top of each other, mass transport complex kind of here. We've also got marine shale which looks fairly seismically transparent. Here we've got an intermix of some tuffs and some clastics. As well as a lot of igneous features. So we've got some bright reflectors that are likely sills. We've got lava flows. We've got this very hard to image magma conduit, and even the basement rocks.

05. A Suite of Seismic Facies: Volcanic MTC

So as we're kind of thinking about, OK, well, how do we interpret these, let's zoom into one of these just as an example. So here we've a volcanic mass transport complex. We've got well control, so we know that that is what the is showing us here. So we may look at how do we describe it in seismic data to kind of validate or have confidence in our interpretation. So we might look at the internal reflection configuration. So there's no continuous reflections within the mass transport complex. If we look at the external shape of it, it's very chaotic looking; it's a flow feature; there's perhaps some coarser inclusions within the MTC. We could talk about it in terms of its amplitude. So it's got a very variable amplitude. We've got some areas that are very high amplitude, other areas that are lower amplitude. And then we wanna start thinking about, OK, well, how can I get more out of the data? How can I see more patterns so that I can interpret them? And this is where seismic attributes come into play. So some of the we may look at (and that I'll be talking about later on in this course) are attributes like. So how similar is one waveform to the one next to it? We also have attributes that can help us see the patterns regarding the geologic dip more or convergence of reflectors. There's also curvature, so understanding how curved that reflector is. We can look at the spectrum, and even kind of like the , so the variability of the values in the seismic data.

06. Human-Driven Interpretation

And so when we think about this, we're trying to relate the attributes to what we're seeing geologically. We have to think about, OK, in our brains, in our human brains (or perhaps in Kurt Murfurt's brains here) we have to think about, OK, well, what are we seeing? What are our contextual clues that we're seeing as humans in terms of interpreting this as a geological feature? So with a mass transport deposit, often we see these at the base of some slump scars. Here we've got an example from a landside. And then we would have this related mass transport failure slope complex down here, so it's very jumbled. We're kind of picking up on lots of aspects of it.
And this is where the attributes can help us because we can think about the geologic expression. So this mass transport complex, it's very broken up, it's jumbled, so using the attribute we would expect it to be very low coherence because one waveform wouldn't be like the one next to it because of how mixed up it is.
We can think about . So we can use a texture attribute called . And we wouldn't expect the seismic to be, again, similar to its neighbours. We would expect it to have a high level of entropy. So that's another attribute we might use.
And then when you have a failure like this, the seismic reflectors aren't going to be continuous. So we can use different to help us with parallelism or reflector convergence. And so when all of these different aspects combine, that center point would be where we might identify it as a mass transport deposit in seismic data. And that is kind of like a way of clustering.

07. Interpretation Bias

One of the things I wanna mention as we go into this course is interpretation bias. So we always wanna be aware of when we're seeing things and we're interpreting things, there could be a lot of possibilities. So interpretation is a very non-unique problem because there are many different changes from to porosity that can change the acoustic , change the reflection coefficient, and we need to be aware of these, and especially what we're seeing.

08. Confirmation Bias Driven Interpretations

One of the most common ones we've run into with interpreters is confirmation bias. And this is where we see what we're familiar with, OK. And so here we're looking at some seismic data. I'm not giving you any contextual clues about what it is. And we've got these kind of triangular right here. And so we can easily go in, and if we don't know anything about the area, the that we're working in, we may say, hey, this looks like a pinnacle reef. I've looked at a lot of pinnacle reefs in seismic and that's what I think it is. And so if this is a pinnacle reef, here's how I would map it out: these would be carbonates; perhaps this is some sort of fault-related fold feature that I'm seeing down there; we've got some seismic artifacts beneath the pinnacle reef because of imaging problems or velocity differences.
But we can also look at it and say, OK, well, it's a very triangular cone-shaped feature. Maybe this is volcanic. And I've got a sill here. We've got some mounding above the sill, some forced folds from when it was emplaced. Here's my conduits, my dikes, and then here's the volcano itself.
And so you could see that either one of these would fit the seismic data. So that's why we don't want go in with any preconceived notion of what it is or just choosing what something is based on the shape or our experience. But rather we want to collect as much information as possible, whether it's from outcrop studies, well logs, just other folks who understand and characterized the basin, so that we know what to expect. And we call this in-context interpretation because we're using the contextual clues, we're using other sources of information besides the seismic to make the most probable interpretation.

09. Attribute Taxonomy Based on What They Measure

So jumping back to attributes, there's a lot of different ways to organize attributes, to think about them. Sometimes people organize attributes based on what they measure. So they could be things like: the spectrum, the of the seismic data; we have our attributes that help us better map, better see the reflection configuration; we have discontinuity attributes, so edge-detecting attributes; we could also have attributes that help us see more details and nuances in the texture; we have attributes, attributes, and time-lapse attributes.

10. Attribute Taxonomy Based on the Input Data

Oftentimes, people also organize and think about attributes based on the input data that you have. So you can have your horizon-based attributes. These would be things like your time structure, dip magnitude and , curvature, making shaded relief maps or bump maps that you can do in some programs. You can also have attributes that are . So you can think about if you've picked a top and a base horizon, so these might measure the thickness. You can have a RMS amplitude attribute, looking at the RMS amplitude within that entire formation or a peak amplitude within that formation. You could count the number of zero crossings between that top and that base. So those are formation or window-based attributes.
We also have attributes that are run just on a single trace (so a single wiggle) that can look at the amplitude, the that I'll be talking about in one of the next lectures. You have waveform shapes that you could kind of cluster together. Or you can run , look at on that single trace.
You also have multi-trace attributes, and these are ones where you're taking one trace (that's the trace you're analyzing) and you're comparing it to its neighbours. So how similar is it to its neighbours? That would be . Curvature, when you're comparing it to its neighbours, you can start calculating the curvature, the of that particular horizon at that point if you map it into the neighbours. You have amplitude gradient, so calculating how much the amplitude changes from one waveform compared to its neighbours.
And then you can run attributes on your prestack data, so if you have angle stacks or gatherers. You can calculate interval velocity, you can do AVO analysis, elastic , λ.⍴, μ.⍴ type analyses.
So there's a lot of attributes out there. And what an attribute is, is essentially it's just running math on the attribute, on the seismic wiggle itself, that one seismic trace, or comparing it to its neighbours. And you're just doing math in order to extract patterns in the seismic data to allow us as interpreters to see those patterns better with our eyes so that we could perform our in-context interpretation.

11. 6 Common Uses for Seismic Attributes

So getting to the mean of this mini lecture, here are 6 common uses for attributes. So these are the ones that we tend to use them most often for, and I'm just going to talk through them really quickly and just give you a sampling of attribute usage and why you may want to calculate them.

12. To Validate Your Subjective Interpretation

So one of the main reasons we use attributes is in order to validate our interpretation, in order to bring out patterns of deposition or tectonic deformation in different ways. And this may be to better communicate to our team members, to our management, to partners that we're working with on our research project. And so this is one example of a seismic attribute display. This is in the Great South of New Zealand, and I'll be using this dataset a lot for future lectures too. But we're displaying a couple of different attributes here. We've got an edge detector, which clearly shows in the black color some of the faults, as well as highlights . And then we're also showing dip and dip magnitude in here, so that we can see the different direction that that one reflector is dipping with this .

13. To Intelligently Extrapolate Sparse Well Measurements

Another reason that we may want to use attributes is in order to extrapolate away from a well where we have seismic data. So if we think about wells, those are essentially like a 1D kind of section of the earth, and we may have information in them regarding the density,, fluids, mineralogy. And what we can do is when we correlate the wells to the seismic data, to the, we may notice, OK, well, in this case, if we have a high amplitude, it's correlating to a porous oil-rich sand. And so we can then look for that same amplitude as it moves away from the wellbore to get a sense of how large that reservoir may be. And we can do it with , we can do it with inversion attributes also. So it's just allowing us to kind of move away from the wellbore with the seismic data.

14. To Reconstruct the Tectonic, Depositional and Diagenetic History

A third reason we may use attributes is in order to in detail reconstruct the tectonic and depositional history. If we know the depositional history, we know using our geologic analogs that we've been studying for our careers, that may allow us to infer or infer rough ranges, understand what theorientation or fracture density will be. And so these attributes, like I'm showing here, help us see these patterns. So on this left-hand side we're looking at the Shublik formation in the Alaska North Slope. We've got here... looking at (I think) the most positive, the most negative curvature, as well as . And we can notice a lot of faults and fractures, and we can see the direction of them. And this would be something that might be a bit harder to pick out just in our seismic amplitude data, but we can see it and we can confidently interpret it with the seismic attributes. Over here we're looking at the Red Fork channels in Oklahoma using the coherence attribute. And so we're very clearly able to delineate some of these major channels with this attribute as well as some of the minor channels. And these are actually really hard to see, the minor ones particularly, in the raw seismic data itself.

15. To Communicate Your Play in a Timely Manner to Non-Geoscientists

A fourth reason we might use attributes is in order to communicate our play, our reasoning to non-geoscientists in a very quick period of time. So a lot of times you may end up working with folks in finance, with engineers and landmen, as well as folks in the government, and if you can put together a nice, clear display that kind of summarizes your weeks or months of work analyzing the seismic data, that helps them gain confidence in seeing what you're seeing. So just like with these examples where we've got some very small-scale channels, we can point them out better in the seismic attribute than we can with the seismic amplitude data, or these karsting over here on the left-hand side. So they're great as visual communication tools.

16. To Evaluate the Effectiveness of Alternative Processing Parameters

Another reason we may use attributes (and I'm a big fan of this) is to evaluate our processing parameters to kind of use it as a QC tool. And so is one of the best attributes for this. You can QC a seismic dataset that you're given. Or as shown in this case, we've got 2 s through the same volume. In this first volume we had just migrated the data with the original velocities, and then in the second volume we had tweaked those velocities to be a little bit more accurate. And you can see that some of these channels show up much more clearly, much more cleanly. There's a bit less noise. And so we can use coherence to give us confidence that we should go forward with this new velocity model in the second .

17. As Input to Machine Learning

And then finally, attributes are really good as inputs. So whether we're doing seismic ification or perhaps where you are trying to do some fault picking with al neural networks. So we have a lot of . We can do multi-dimensional analyses with the aid of machine learning. And so they're very valuable in that sense. And later on in Module 4, we will actually get to do that and you can around with it in the AASPI software if you want to do the optional labs and test out a lot of different machine learning algorithms running on seismic attributes that you calculate earlier on in the labs.

18. Seismic Attributes - Key Concepts

And so overall, I just want to emphasize that attributes are incredibly helpful, not just for geophysicists, but also for geologists. And their main use is to find some of those hidden patterns that don't just appear in the seismic data, like the seismic amplitude data itself. is a great one for helping us see faults, to see the edge of channels. We can run many different attributes focusing on amplitude, on , on comparing one waveform to another. We can look at them on a horizon. We can look at them in a . We can calculate . So there's a lot that can be done to really reveal the subsurface to us and help us improve our geologic interpretations and give us more confidence in our geological interpretations.