Overview of seismic texture and directional texture attributes in OpendTect, including dip steering and GLCM analysis for facies and geomorphology. The presentation also shows how Geo5’s FracTex plugin uses directional attributes to analyze anisotropy and fracture orientation.
#OpendTect #Geo5 #FracTex #TextureAttributes #SeismicFacies #DipSteering #Anisotropy #SeismicInterpretation
Duration: 12:10
--- Transcript ---
This presentation has a three element as the basic texture attributes as implemented in OpendTect and then there is a second group of attributes which are called directional texture attributes. Also implemented in the free version of OpendTect and these were developed by Geo5 a partner of dGB. And Geo5 is a spin-off of Joanneum Research Institute in Austria these directional attributes these directional texture attributes are the input again to FracTex.
And FracTex is a commercial plugin by Geo5 now what are texture attributes it's really already quite old they have been around since the 1970s it comes from image processing. And it's based on Haralick description of gray level co-occurrence Matrix these attributes describe the roughness. And smoothness of an image and what a gray level co-occurrence matrix is I will describe in my next slide in seismic adaptation that has been used for many years for seismic facies interpretation.
And also to highlight geomorphological features and it's still used today it's a very valuable interpretation technology attribute just today it's March 2025 in first break the latest issue I saw a presentation in which GLCM attributes were used. So how do we construct a gray level co-occurrence Matrix Let's look at this image. And the values are the gray levels here so we have pixels of zero gray and one two.
And three gray levels now each of these cells we can identify into in this Matrix so this is the cell with position 0 0. Then we have 0 1 02 on and so on what we're going to do is we're going to count how many neighbors we have of zero to zero. And this will be mapped in The Matrix in this cell so this is one occurrence of zero next to zero this is another one zero to zero.
Then we also have these two cells from zero on the other side to zero and this one. So in total we have four occurrence of zero next to zero in this particular image now we can normalize that to probabilities on how many occurrences do we have in total in this image. And then we fill in the Matrix as follows we just count everything converted to probabilities and that is our GLCM Matrix from that GLCM Matrix we compute attributes.
And we do that in three different groups one group is called the contrast group and these are measurement based on the distance from the GLCM diagonal. And the first attribute in this contrast group is called contrast and this is an example input on the left. And let me get my slide pointed again laser pointer so input here and attribute on the right contrast dissimilarity belongs to the contrast group homogeneity.
And then we come into the next group and that is called the orderliness group and these are measurements of how organized the GLCM Matrix itself is this particular attribute is called the angular second moment this one is the energy. And this one is the entropy all part of the orderliness group now then the third group we have is just statistical measurements. So on the right we see the GLCM mean the GLCM variance the GLCM standard deviation and GLCM correlation all these attributes that I've shown here are available in a group in the attribute engine under the texture attributes.
Now all these attributes can be dip steered and dip steered is a typical thing in OpendTect whereby we use a plugin to calculate the dip field we call that output the steering Cube. And the steering Cube has two component it has the inline dip and the crossline dip and we can use that dip to create small virtual Horizons at every position just by following the dip outwards. And then we can extract the information we want along the dip that structural reflection system so let let's have a look how that works this is the dip field calculated dip we're looking here at dips dipping in this direction colored in red.
And dips going to the other direction in blue and in between we have white and yellow colors. Now if I would extract a multi-trace attribute from this Central position I can go traces outwards. And I can train samples upwards and downwards and I can do that in a certain circle or certain Cube in three dimensions.
And extract the information if I don't use any dip steering this is how the information will be extracted just horizontally. And vertically and then I compute my attribute if I use dip steering I'm starting from here I look at a dip at this location. And I'm following the dip to find the next cell and then from here I can find the next cell using that dip.
And so on so I can really follow the seismic reflector everywhere if I put this input into an attribute I get a response that really belongs to the geological information that we are following. And here we see an example of that effect of the effect of dip steering on a GLCM attribute in this case correlation on the left we see non-dip steered GLCM correlation. And on the right we see dip steered here we see that the attribute is really following our seismic Reflections.
So we get a response pertaining to the geologic information and not just cross cutting all these different layers. So dip steered is always giving you a better response that's the message now let's look at directional attribute. So that is something that was computed by or developed by Geo5 and it's available in the free version of OpendTect each cell in this case the middle cell here has 26 neighbors.
And that allows for computing computations in 13 different directions so I can from this cell go into the inline Direction. And just compare what is next to these cell information I can go into the crossline direction I can go vertical. But I can also go from this cell to that cell and up there in a certain direction.
So that gives 30 different ways to analyze the information in all these different directions if we do that for texture analysis let's look at this gray image just a synthetic gray image. And here are the gray levels in this particular gray image scale image and what we're going to do is we're going to compute different attributes for this image in different directions. So first we're going to compute The GLCM Matrix just in the horizontal Direction and that gives you this Matrix from which we can compute the GLCM energy contrast homogeneity entropy cluster tendency.
And you get these values now the next thing we're doing we're going to repeat it. But now we're going to compute it in this direction and we get different values here for the attribute responses similarly here in this direction different responses as compared to this direction as compared to doing it in all directions. Now what have we learned from this analysis that if we would compute it in this direction the horizontal Direction.
Then all the attribute values tend to be average for this particular image if we go into vertical direction we find the highest cluster tendency in this direction we find the highest contrast. And the lowest homogeneity in this direction lowest energy lowest cluster tendency and in all directions the highest entropy an example in 2D this is a interpreted Horizon. And we see channels in this system this is in the Vienna Basin this is coherency and this is the interpretation we're going to compute different GLCM attributes energy homogeneity.
And so on and we find this one the energy the most interesting because it tends to pick up the channels best compared to the other attributes. So now we're going to analyze this in different directions and what we see is that some of these channels are picked up in different directions they're brightening in different directions. So there is an an isotropy in this image and what can we learn from all these first of all the texture attributes are valuable Tools in seismic interpretation that has been for the last 50 years or.
So and it's still valuable today original applications are for seismic facies analysis and visualization of geomorphological features the dip steering constraints the analysis to stratigraphic layering. And generates higher signal to noise ratios and better responses for texture attributes directional analysis reveals anisotropy. And the image and now we come to the FracTex so dip steered directional attributes have potential for analyzing anisotropy.
And rock properties and can thus be used in the analysis of fracture density stress Fields fluid flow. And paths and that is where FracTex comes in so FracTex it computes directional variations of these GLCM attributes. And that is done for mainly for fracture analysis there are two basic outputs from FracTex one is the anisotropy.
So that is a factor which ranges from one to higher and the higher it is the more anisotropic the medium is one means the medium is isotropic. And Fracture dip azimuth that is the azimuth in which we have the largest variability in variation in docm attribute response.