VideoPoints

• ••• • •• • ••••• • •
GEOL 6390 3D Seismic Exploration - Day5 11-10-23
Help Tour
© Distribution of this video is restricted by its owner
Transcript ×
Auto highlight
Font-size
00:02 So a question came up in the and it had to do with apparent
00:07 and true dip. And in the velocity the dips are gonna be
00:13 in angle in theta. So it's go from, you know, 0
00:18 90 degrees for dip magnitude to go time to depth. You need to
00:26 a velocity. If you wanted to accurate dips, you would need to
00:32 your data to. Yeah, there's ways of doing that. Oh You
00:38 spend a lot of money for a survey. A million dollars, $2
00:42 in prestack depth to migrate your data probably not gonna happen. OK.
00:51 , if you have some whale you've already picked he horizons. If
00:59 horizons are major velocity contrast, let's between plastics and carbonate top of salt
01:08 of salt, you can then compute interval velocity for each layer using the
01:16 control you have and kind of interpolate across and then convert it's not a
01:22 way of doing it. OK? patrol has a way of converting the
01:28 time structure map to a depth structure . You then time depth pairs from
01:40 control and the way they do it's an interpolation technique called Krey.
01:45 it's a GEOS statistic technique and it at the, well, I'm gonna
01:51 that time deformation that time velocity pair give me the true depth in
01:58 If I have a fault that I on the horizon and my uh my
02:06 drops 500 milliseconds. I'm gonna keep same discontinuity when I convert to
02:12 that's kind of what rigging does. . Says I'm going to interpolate and
02:18 the wells when I'm really close to wells. And when I'm far from
02:22 wells, I'm gonna use more of trends from the other data volume,
02:26 in this case is a two way . So you need to realize that
02:33 of these numbers in patrol are somebody a number out of a hat,
02:40 it in there for all time and called the patrol people. And
02:44 well, what number do you use ? Velocity? Well, that guy
02:47 no longer working here. They don't . Ok. Hopefully there's a different
02:53 for English units and metric units, we don't know. Curvature is the
03:00 thing. Curvature needs to have things , in depth. So a common
03:06 , let's pick a reason like if working Barnett shale, which is a
03:11 five Aleo Zoic rock in North New in the Fort Worth Basin.
03:17 velocities are 4500 m per second to m per second. They're very fast
03:25 I can use that number to convert curvature volumes approximately and things work
03:33 OK. Works fine. Um There another point there but I forgot
03:42 So let's go to the next Did I answer your question a
03:47 Y pardon? Oh, yes, fine. Then I can, I
03:58 how to get it from one project the next. OK. So you
04:02 a question, ma'am. You figured out? And they, this guy
04:08 help you. He helped. All . All right. OK. Mhm
04:23 I oh You mean my dad Those are the faults. Yeah.
04:29 . Oh So to finish that, need to pick some, some control
04:37 here. So you need to add horizon fixed. Yeah. No,
04:45 , I think you're looking at this . This is your time structure,
04:49 . This is this picture here. looking at a figure on page.
04:59 . We're looking at horizon based attributes with and now what, what we've
05:08 here is this is your horizon. now we're gonna take that horizon and
05:14 through the variant cube and you haven't that step yet or if you
05:21 you're not displaying it. Did you through? Did you, it'll say
05:27 that here. No, put it and stuff. What do you got
05:32 ? But I put it and uh this one go away. You got
05:39 one, right? It might be shallow, it's taken that long.
05:51 here this goes away. Oh, this so valuable. Ok.
06:02 there you are. That's the And now, uh, well,
06:17 got something else turned on. Thats ? Ok. Ok. I take
06:27 off. Ok. Now where you generated your surface yet, have
06:37 but go back to where you were here, it was at the
06:42 Ok. Let's put a check in of T W-2 to a travel.
06:50 OK. You don't have a surface you have to generate the surface from
06:56 e makes sense. OK. So patrol, you're gonna pick. So
07:03 pick in lines and cross lines or just have well tops, you have
07:08 seismic, you just have well Those are your, their time picks
07:14 in patrol and they jargon and their , they call that a horizon
07:22 Then you wanna put a surface through and it's easiest to think of well
07:27 . I've got 28 wells and I put a nice smooth surface through.
07:32 they're gonna put a beast line through . OK? So like you would
07:41 with but nobody traces anymore except So you put this nice smooth curve
07:46 two dimensional, in two dimensions to those um well depths that they call
07:53 surf, right? So then to , to make a horizon slice through
08:02 volume, you have to first generate surface. And then in the surface
08:08 , one of them is to compete actually extract a data volume off of
08:17 surface. And most people would call a horizon slice through the volume of
08:24 data, right? So once you a surface, then we can look
08:29 the extraction. OK. So with , we can go through the
08:34 Let me pull it up. I get my, oh I gave my
08:50 drive the cars and he's copying horizons me. Ok. Ok.
08:57 that's ok. We'll just, we'll wait a couple of seconds. It
09:00 be long Africa. Yeah, it time. Yes ma'am. Ok.
09:27 what? Ok, so let's look uh your interpretation line through there.
09:32 So, ok, go win maybe . Ok. And did you pick
09:40 every 10, something like that? did you pick? Ok, so
09:43 sure you got a line. Give me one nice restaurant.
09:47 here it is down here. All right. And now if you
09:54 go over a little. Ok. is one of these films you see
09:59 it's saucer shaped? Ok. That's it looks like. Um go over
10:05 where you got the blank space Mm. Ok. Are you tracking
10:15 mouse? I'm not seeing it Yeah, you gotta go to the
10:27 like this is a two D Um is the interpretation window and then
10:35 to that to the interpretation. It'll under you don't wanna add a new
10:40 . There's gonna be a window over someplace. Yeah. Click one of
10:45 and you're gonna whoop. There you . Now, now if I go
10:51 that, hey, thanks. That shows you where you're missing your
10:55 . So, OK, keep Looks like you have a pitch for
11:04 . Oh, I'm just, did tell you? I can't see magenta
11:10 red. OK. You've got two chromosomes, ma'am. I can't see
11:17 . Ok. So you need to pick that red there,
11:21 go pick it right across there. mean, that's what you gotta
11:24 OK. Now if the choice is you want to take it and come
11:29 with? Yeah, I think this what's happening or do you want the
11:34 just willy nilly going through it. mean, you're better off making a
11:37 . That's right. Yeah, I why I couldn't see it because it's
11:43 and red that when your eyes, know it's Oh yeah, it
12:03 It's on the floor again. Let me squeeze it. Poor Jessica
12:12 listening to his bounce on the floor uh here's my backups. Good or
12:34 in my screen sharing? OK. is it showing up there?
12:40 showing there too? OK. So gave you um 10 questions. Three
12:46 questions. OK. Here's the first and we have these uh LTs funny
12:55 things deeper in the data. They're of red blue red. So high
13:02 amplitude isolated, there's no good illumination reflectors nearby. We know we got
13:12 volcano in the middle of the They're kind of sort of s are
13:17 . This is what a carbonate, a uh a basal basaltic sill looks
13:22 an igneous still. OK. up here, if you look at
13:28 , up here, it's blue blue, here at blue,
13:32 blue in your shallow data. If made a time slice around 0 400
13:39 . And I think I have you that as one of the exercises.
13:42 some shallow gas which would be a hazard that has strong amplitude and negative
13:52 . OK. So high amplitude isolated doesn't necessarily mean gas. In this
13:59 , uh you could be, if didn't know the polarity in your
14:02 you might think this is this is , but it's, it's igneous
14:09 OK. This is uh, was practice question. And you know,
14:16 I looked at the uh grades from time, I said, all
14:20 you guys need to learn this. I'm gonna give it to you
14:23 All right. And uh the one the last test was uh angular and
14:29 . So I've got a neg sorry, I got a negative reflection
14:34 the top. And from that, know that I'm going from high MP
14:40 to low MP S. So I a positive source wavel. And then
14:46 I'm going from OK, two times times three. So I should get
14:55 here. No reflection on the No reflection on the left because it's
15:01 thickness. So clearly this one is . OK? I shouldn't get any
15:07 here. So one of the other are correct. I'm going to
15:13 I'm gonna go from low impedance to impedance. So I'm gonna get a
15:19 reflection coefficient. That's good, positive coefficient. That's OK. Uh This
15:27 we already eliminated um positive reflection coefficient OK. And then on the
15:34 it's gonna go from high and back low. So we should have a
15:39 reflection. Well, that's not Ah This one looks good. This
15:45 has the right polarity but it's a phase wavelength that means the drop should
15:51 centered at the interface and it's not . So it's the one in the
15:56 , right. Hopefully everybody got that . I this one you will have
16:05 interpret your data as you as you through the faults at different angles.
16:10 probably saw that. Hm They look when you're perpendicular to the vaults than
16:15 you're parallel. So this, this probably the hardest question in the
16:20 Um So in this image, I'm perpendicular to a fall and I picked
16:28 stick and this stick and this stick this stick and now I wanna look
16:35 kind of some parallel default. So the right direction. So if I
16:43 this curved feature and I'm gonna, holding AAA piece of uh what am
16:52 holding a mouse pad in my And I've got like a U kind
16:57 the gentle U shaped vault. And if I intersected with a vertical
17:06 I'm gonna get the letter U out it. OK. That's what it's
17:09 look like. So this, you have to visualize geometry. We got
17:16 examples. Here is the uh vertical through the amplitude. And yeah,
17:25 see Bolton here, I kind of something in here. It's pretty hard
17:30 see. Here's the coherence image I that you think, OK, that
17:37 the fault and this, that's some of a fall but it's a different
17:45 and this is some kind of a , but it's another fault and then
17:50 one, well, probably mixing So the moral of the story is
17:57 , don't try to pick your faults the strike directions. You're gonna be
18:01 , very unhappy. A quality I think I might have asked you
18:07 the lab if not for fun, your fault, picks all the little
18:16 sticks and then draw a virtual line what they call an arbitrary line parallel
18:25 the fall. And guess what? guarantee you there is not one person
18:30 here where I have a nice smooth , you probably go zigzagging up and
18:36 . It's just really hard to get smooth surface in the right direction,
18:41 it should be smooth. Thanks. right. Then in this one,
18:51 didn't give you a couple of papers look at. I think I
18:56 I gave you a powerpoint slide of of the features and I know I
18:59 in the class. But what we this, we had the basin and
19:05 the core volcano came up, And we had a volcano and then
19:11 basin, the basin sank, but had a seamount. So I was
19:18 out to uh Cardinals today that this there's a new volcanic island in the
19:24 off of Okinawa in Japan. So , maybe 50 acres or something
19:31 It may or may not be You know, sometimes they come up
19:34 then because of ISOS tasy, they to sink right. Then you also
19:40 from carbonates, they gotta be cooled . You can have super hot if
19:45 can have coral reefs grow around the as it sinks. The coral
19:50 keep growing up, growing up, up and that'll give you an atoll
19:55 . Ok. So uh what we then is on the right there was
20:05 accommodation space around the co of There's no horizon over the top
20:11 That horizon was never deposited. if you didn't know anything about the
20:18 of the coal volcano, you might say that the coal volcano came up
20:28 blew everything the kingdom comes and, you know, obliterated the overlying
20:36 but on the sides, then you see them not pinching out nicely.
20:40 would see, you know, they're actually getting a little thinner uh
20:47 look at this horizon here. So it's getting thinner uh as you get
20:55 the co volcano. So that, implies that no, it wasn't pushed
21:00 , that was actually there before OK. Now, I'm assuming all
21:10 you have made this kind of mistake your data. I mean, because
21:13 an easy one to make you make picks, then you generate a surface
21:19 we're gonna interpolate those bad picks and gonna have glitches in your horizon.
21:24 . And this is the way you control your picks or if you're a
21:29 working with a Geophysical partner, you they look like bad picks to
21:35 you know, they're in a maybe two or three lines at her
21:40 . So strikes football. I kind doubt it. Gulf of Mexico.
21:47 It could, it wouldn't be a fault because I don't have displacement the
21:51 end in the later question, I'll you about this where we go
21:58 Uh when we go from green to and blue to purple, those are
22:04 books. They're the surfaces dropping But here these are local bumps and
22:11 a suite of carbonate build ups, carbonate build ups will have those kind
22:16 lumpy things, but they're not gonna in a straight line. The most
22:20 thing is that hicks since everybody hopefully everybody's made them. Um,
22:26 the right answer. OK. And is one of those definition things I
22:31 about week one, we've got different of slices and in lab seven,
22:38 generated some of them. If I , I'll pick the top and the
22:44 . If I slice through the that would be a horizon slice.
22:50 I slice through the base, that be a horizon slice through the
22:54 If I drop down uh 50 milliseconds a quite parallel at any time.
23:03 If I drop down, this would a Phantom horizon twice and the yellow
23:08 would be a phantom horizon twice. and B would be another Phantom
23:14 So CC is going to be a twice or proportional. This this
23:22 it might be 60% down from the and 40% from the bottom,
23:29 And if my sedimentation that if my is subsiding at a constant rate,
23:41 the right hand side of this picture subsiding faster. So there's more accommodation
23:46 . If that approximate, if that is correct, then that straddle slice
23:53 a co evil or constant time in . OK. That would be a
24:01 mind. And that's why we like . There's more sophisticated ways of doing
24:08 today with more expensive software uh using scan and open detect. And
24:16 they'll actually try, you pick the and the base and then you'll come
24:21 a vertical line and you'll pick, will automatically pick every zero, crossing
24:28 peak, every drop. And it to correlate between those two guys keeping
24:34 in order not allowing them to truncating them when necessary. OK.
24:40 then we would call those geo chrono horizon. So they are, they
24:49 jo Chrono aic not really, they're with a logic boundaries. So here
24:57 in Houston just to review with logic . We're in Houston as I go
25:03 2023 from Houston into the Gulf of . I go to Clear Lake and
25:14 which is kind of sandy and Galveston and then which is Mud Galveston
25:23 which is maybe a coarser sand or sand anyhow. And then you go
25:31 into the Gulf of Mexico and the starts to get finer and finer and
25:37 it shall again. So the constant slice would actually, so sand,
25:45 sand, finer sand, still OK? It would show that.
25:51 what we're doing is we're, we're the lithology within seismic data.
26:00 All right. This one, I'm hoping that everybody's made these kind of
26:06 , but I'm assuming you have not Alicia because she can't see
26:13 She's picking pink on red. So can't, well, you could probably
26:17 him as it goes vertically. Um So here, this is why
26:22 don't want to pick all your in and then do the cross lines because
26:26 you go, oh man, I fix that now. OK. So
26:31 did this picked, this is what , west and east west line,
26:38 they pick the north, south lines , then they're gonna pick an East
26:41 line and guess what their correlation across two of them are, are totally
26:47 up. And you can see here the fall, this might actually jump
26:54 . It would be much better interpreting pick that fault first. So you
26:59 say, OK, I'll go from to here. That's cool. I
27:02 go from here to here. That's . And then here to here is
27:06 . If I turn my auto my auto picker on like uh like
27:11 and I did on Wednesday, they go right across there unless you have
27:17 fault displayed. OK? And probably , you've noticed that like,
27:22 darn, it's going right across the . It's just correlating. OK?
27:28 those are bad picks. OK? from Mexico survey. Um I used
27:40 use this in class but it's you know, there was a time
27:43 it was very, very difficult to public domain data. Uh So this
27:48 Schlumberger had acquired and let a Um It's poor quality, hard enough
27:57 it is, right? They don't bad quality data. To deal with
28:01 well. But here is a, , a surface time, a structure
28:08 and this fault here, I got 1340 milliseconds to the north kind of
28:16 cyanne, let's say 1400 milliseconds depth the south. That's what a normal
28:22 is gonna look like. OK. right. This is the one I
28:32 you a hint. Like, go look at Z's paper. All
28:36 . So I figured, well, you guys didn't do it. And
28:38 you hadn't gotten to lab seven, not everybody's up to lab seven
28:43 you're gonna get there right where you the straddle slicing. We look at
28:50 paper and what he has here, got uh, four different time slices
28:58 you can see that what he calls that's a channel, a little piece
29:02 a channel, a little piece of channel, a little piece of the
29:05 . So my channel is going along surface, OK? And my surface
29:11 dipping and if I cut that dipping with horizontal planes, I'm just gonna
29:21 a piece of the channel and guess ? You can go in and you've
29:27 seen in the, uh, when picking how you can paint,
29:34 You make your, your little rectangle the mouse a little bigger, you
29:40 have an auto pick and expand and start to paint your interpolation instead
29:46 having, oh, let's do the thing. Let's go carefully. Paint
29:51 so you could paint the time show them up in 3d and you
29:55 a pretty good image of the, that channel, but that's pretty
30:03 And on the right is the straddle . OK. So the straddle slice
30:07 are two picked horizons. I'm gonna proportional between them. And that straddle
30:13 is where the channel is. I have asked you, but I didn't
30:17 you why not just pick the horizon the channel. Well, when you
30:23 a bunch of channels, like in one, it's really hard to pick
30:28 surface. I mean, it's nice use auto trackers, right? Or
30:32 if it's a guided auto tracker. if I'm picking the top of
30:37 a carbonate, the top of a , if I'm picking a maximum flooding
30:43 , then my way p if I'm the peak along that horizon, top
30:50 the carbonate, it's always gonna be peak base of the carbonate, maybe
30:55 always gonna be a trough. So can start to make my horizons without
31:00 . And very quickly if I have share uh a flood plain of compacted
31:13 compacted to a little higher impedance, incise a channel, sea level comes
31:20 , fills it with mud, that's be lower impedes. Now, I
31:25 pick the top of that surface. , positive reflection coefficient across the
31:30 negative reflection coefficient oh across the point making zero reflection coefficient. It gets
31:36 be real tedious to pick it. you're better off picking something above and
31:40 that less error prone and then slicing proportion. Then you get to some
31:47 like Carlos, I I know some the data from the Giannos Basin
31:51 in Colombia and Venezuela and Bohai Basin China. If you've ever seen that
31:58 like spaghetti, there's so many channels this, you can't pick anything and
32:03 got differential compaction on top of it , but you can slice it and
32:07 it. So in this one stradle more accurately approximate a fixed geologic
32:16 they provide a more complete image of depositional environment. Here we see this
32:20 meandering channel with the loops and maybe oxo um can they be picked,
32:27 you pick a straddle site with an track or? Definitely not? But
32:31 really hard to pick them by a given Strat democratic feature along structurally
32:36 horizons appears across multiple times slightly. . So here's that channel and we
32:44 it on different the same channel on time slices and then time sizes don't
32:50 channels and inside valley. Well, I can see them, I just
32:53 pieces of them. OK. So is a real common workflow. All
33:01 now, hopefully nobody did this. when I was teaching the engineering
33:12 they just wanted to get things They didn't care about peaks and froths
33:16 what the color bar was and I some of you weren't careful about the
33:21 bar. So you go in and my, my auto tracker is not
33:27 . You know, I set the could be a, a peak.
33:31 auto tracker is not working. So gonna pick my griddle line by
33:36 I'm pretty patient, do that. then I say let's do a 3D
33:41 or a 3d auto tracker, but still got you, what you were
33:45 was picking crops and the auto tracker the peaks and you get this waffle
33:51 , right? So I would always 10% of my engineering students come up
33:55 a waffle pattern because they didn't look what's a peak in black here and
34:03 a trough in white. And so gotta look at that and then explicitly
34:10 it. Now pare tries to be user friendly and the default is peak
34:18 drop like OK, make up your . And I think the way it
34:23 the first pick you have, it it. Well, what if you
34:28 know what you were doing that first or it was a stray pick?
34:32 , now you're lost, right? you need to know what the polarity
34:37 your data. So for this it's the uh answer is the interpreter
34:42 a conflicting definition of polarity between the picks and the auto track picks.
34:52 ? This one's an erosional surface happened be from quarter to 3d again.
34:57 one of these turbinates. Is it structural fault? No, because nothing's
35:02 below it. So it can't be fall. I mean, if it
35:08 , it could have been a structural if stuff below it were folded and
35:12 it was filled in after the You know, that, that's
35:16 But no, there's no folding below . Uh, it's not a volcanic
35:22 and it's not a salt withdrawal There's nothing withdrawing. They're gonna be
35:27 . So it's just the channel and others here. I mean, there's
35:32 here, here's another edge. See , see these reflectors here just
35:36 they're eroded away. There's a nice bowl, see the base of
35:46 OK. This was one of your and uh input data, filter data
35:53 here's the difference. So I think asked you uh what's the difference?
35:59 it rejecting? It's rejecting noise. a lot of random noise. So
36:03 got rid of that and also the frequency components of the data.
36:09 So this is nice and smooth, I've lost all my resolution. Thank
36:15 . And you did this as an . OK. This one I threw
36:20 is probably the last one I was look for some seismic data. And
36:24 said, all right, you argue me and that gets a little bit
36:28 into processing but alias seem really, common with photography and with scanning.
36:38 . So probably many of you have scanner at home and then they
36:42 ok, we only won 400 dots inch. No, maybe, maybe
36:48 dots per inch. Ok. You 80 dots per inch and I take
36:52 , a $20 bill out of my . No, I'm gonna take $100
36:55 out of my pocket, but I have one of them. So I'm
36:59 have a $20.20 dollar bill with uh Jackson. I OK. And I'm
37:05 scan it and try to make kind good money, go try it.
37:10 if you go to don't try to it along, OK? You go
37:13 prison but the, the dollar bill designed to have these little bitty patterns
37:20 it that can't be sampled correctly unless have very, very expensive equipment.
37:28 ? So here a normal photograph like would take with uh analog camera
37:36 so an analog camera doesn't have this . If I took an analog picture
37:42 then scanned it, you would have property. OK. And here it
37:49 just reducing the number of pixels per . And you can see and you
37:54 these kind of patterns in there. was like a big cross. You
37:59 , those actually have a name, called Noire Fringes. You guys ever
38:03 about Noire Fringes? Oh OK. Again, what's happened in photocopy machines
38:16 photographs? So this is an example aliasing. You tell me about all
38:28 uh 11. Sure. Let's look 11. Then I'll show you ma
38:35 try to find so on this OK. What's the question? This
38:40 to be God was sponsored president instead one year appropriate. Well, the
38:51 channel is going to occur at the bottom, gonna erode sediment. We
38:59 or may not have a reflection at bottom. It depends on what,
39:03 fills it. And then once, once we fill that channel, everything
39:07 gonna be nice and flat above it . So a turbide channel only takes
39:16 sediments that were there before the turbid . Now, not everybody here has
39:23 classes in sediment technology. We'll go that in the lectures today and
39:27 But for a turbinate gonna be most at what we call a low
39:38 So sea level is rising and falling the ice. OK? Right
39:44 sea level is rising because the atmosphere warmer. So I'm living in
39:50 I used to live in Queer like ft above sea level. The Gulf
39:55 Mexico is transgressing on my property, ? So we call it transgression.
40:01 right. And, and when it's when we have a lot of ice
40:08 , oh, now I go the land can prograde. I can build
40:15 the shelf out. 0, 100 off of Galveston Island. OK.
40:19 you've seen maps of what the coastline like during the ice age. So
40:25 could walk from the Netherlands to Great and you could walk from Australia all
40:35 way to Thailand and you could um, from Siberia to Alaska and
40:45 level was lower. Right. That a low stand. And now at
40:51 low stand, what happens? I've got all of these deltas that
40:55 originally sub aqueous. Now they're exposed I'm gonna have some storm, I'm
41:01 have an earthquake and that sand is sticking above the air. It's gonna
41:07 and down. I'll probably show a of pictures here tomorrow. Um,
41:17 Magdalena River and of Colombia where, , I know, uh Caros has
41:26 he's worked some of uh Magdalena but this is where turbos were first
41:32 detected two places. Magdalena Basin in River Outlet in Columbia and then in
41:41 , in both places, they discover because you have these massive floods and
41:47 of a sudden the telephone communications, underwater cables would he taken out?
41:53 disappeared and they said, well, going on? Well, got all
41:58 sediment moving and, and eroding the . Any other questions? Ok.
42:07 just gonna see if I can find picture of Moire fringes surgery test and
42:18 you Oh, yeah. Yeah. . Yeah. Thank you. Ok
42:37 dot dot There's my images here we . Hey, here's the pictures I
42:57 . I mean, it's so these the kind of patterns these, these
43:06 of interference patterns. Those are bore from under sampling the data and why
43:13 looking at buying stuff. I don't . Let me go. Monterey
43:19 uh shirt. Ok. I I've seen pictures, ah, here
43:31 are. You'll see this online, think a big, there. You
43:47 the picture on the left and the , but the same picture, it's
43:54 this very subtle pattern on the Quite pretty here you can see the
43:58 on the sleeves and on the pocket then here's more then to load it
44:04 the web. They're trying to make little, a little picture because it
44:08 money. They've under sampled it. , it looks like a camo
44:13 you know, camouflage shirt. That's an alias. OK? So
44:18 see it a lot when you look things on your cell phone and stuff
44:21 that, it's, it's pretty OK? Tell you what, let's
44:30 a 10 minute break and then I'll up about spectral decomposition from last
44:40 OK. So where we left off uh Saturday was uh talking about spectral
44:52 . Now, when we do a transform, we take in our,
44:57 let's say our data are four seconds , we're going to take a four
45:03 long sine wave of 10 Hertz. gonna multiply every sample times every sample
45:09 our seismic data. We call that correlation. The resulting answer when we
45:14 them all up, that's the 10 coefficient and we do 20 Hertz sign
45:23 and cross cor away add up all cross correlation values. That's the 20
45:31 component keep going 30 40 50 So that's a fourier transform that we
45:38 in seismic processing day in and day . Then the short window fourier transform
45:45 something that we do within a fixed . So we try to localize it
45:52 a window, let's say of 100 . So we wanna analyze what's the
45:57 within 100 millisecond window. OK. that's called a short window for a
46:03 , pure, pure and straight Um The continuous wave we have to
46:10 the edges or we get Stephanie, get what if we don't taper the
46:23 . OK. Da la. Good. At least you know the
46:27 . OK. The gives phenomena, ? Did you listen to Coolio?
46:34 , it's pretty good. It's It's a classic. OK. So
46:39 have to taper it a bit. my taper is a Gaussian, we'll
46:43 that a Gaussian window transform. And continuous wavel transform uses a Gaussian
46:52 But the way people implement it is say let's always make that window the
46:58 number of cycles wide. So if period is 1/10 Hertz, I'm gonna
47:07 the standard deviation of my Gaussian uh milliseconds. If it's 20 Hertz,
47:15 gonna make it. I say 10 gonna be 100 milliseconds if it's 20
47:21 , it'll be 50 milliseconds. If 50 Hertz, it'll be 20
47:25 So the window changes of the Matching Pursuit. I started talking about
47:32 last Saturday and everybody was tired. figured. Ok, so, so
47:40 got two methods. Now, we're talk about a third method and the
47:45 method is called matching pursuit. What do is read our seismic tray.
47:52 we're going to compute the data and over transform. We can think of
47:58 as a complex choice where the real is measuring the kinetic energy and the
48:07 part that ho transformed is measuring the that OK. And so that's easy
48:15 to do. You, you've already that. You've computed Hilbert transforms.
48:19 then I'm also going to compute the envelope and the instantaneous frequency. The
48:26 gonna go across the room. I I did this last week and Stephanie
48:30 uh 0.6. So I got the values are going from 25,000 to minus
48:37 . So my envelope is going from to 25,000. She picked 0.60 0.6
48:44 25,000 is 100 and uh 15,000. all the envelope peaks above 100 and
48:55 15,000. I'm gonna pick and I've 13 of them down that trace.
49:01 have an instantaneous frequency there. My frequency is gonna be stable at the
49:06 peaks. It's at the envelope the the the troughs that it's gonna
49:12 bad. I'm gonna take that I'm gonna go to a pre computed
49:24 of wavelets of complex wavelets. Then going to try to fit those 13
49:31 wavelets at different times to my seismic . I fit a complex number to
49:40 . If I got 13 complex wavelengths I'm trying to fit them to the
49:44 trace, I'm going to have 13 coefficients. One will measure the
49:50 the magnitude and the other will measure face. OK. So I can
49:54 90 degree phase or 35 degree phase whatever. I subtract that from the
50:03 . And that gives me a something that's left over. OK?
50:08 if the residual is less than some , let's say 1% of the
50:15 half a percent of my energy, gonna go and then I'm gonna calculate
50:20 complex spectra of the residual trace. still use 0.6 because we were happy
50:27 that the last time we don't have , but we, we can and
50:32 we'll keep subtracting. Yes, We are in number 10, number
50:42 slide 55 which you'd have to stand the desk to see the numbers.
50:49 ? It is 10, right. . Good. Um So we're gonna
50:56 and we're gonna keep going in a and all these little wavelets, these
51:01 wavelets, they have a center frequency they have a little spectrum. I'm
51:07 accumulate that and that's gonna build a varying spectrum. Thanks. So here's
51:15 example from uh John Will, John did his phd here at uh data
51:24 West Texas original data. What's the event I used 0.8? Oh There's
51:32 strongest event we square fitted. Um of it looks well, if you
51:38 see the, the phase of this and there's one way this has got
51:41 different phase, this may be 90 phase. This is zero degree
51:45 I think the minus 90 degree That the first part and then what
51:53 the residual? OK. Now I'm more and after four iterations some
51:59 after eight, yet more. And 16, basically, I've reconstructed
52:07 the data. OK? I've modeled with all these little wavel its.
52:12 when you do processing, you always see what's left over. So
52:16 what's left over? What's the residual the first iteration? And here it
52:22 . And I think I might have this far last week. And I
52:26 , oh, I'm gonna take uh baritone out here and then underneath that
52:32 voice is a high frequency soprano. . So they're different voices at the
52:39 time. And then that one then two and four, six and
52:50 16 iterations, everything is plotted the scale and you can see, all
52:55 , it's mopping the data very, nicely. I can then accumulate the
53:03 components. Here's the 40 Hertz component I have to accumulate them as complex
53:09 . That's a detail, but that's you have to do. And you
53:13 this vertical volume or the vertical slice the 40 Hertz spectral magnitude volume.
53:23 . And so here is a time with that same data volume. And
53:31 got a little channel in here, little channel in here. There happens
53:36 be an erosional un conformity up A reverse vault over here. A
53:41 of car stain over here. And is the uh 10 Hertz component.
53:47 you're gonna look, maybe just look this channel here and can hurt.
53:54 20 30 40 5060 70 80. you notice the channel I it gets
54:07 and weaker depending on the frequency. that's because the thickness and the feel
54:11 that channel is T at maybe 40 or 60 Hertz. I can color
54:18 them with a two D color So in this case, I have
54:21 as well and red is high and the channels are showing them mainly as
54:26 . So they are tuned at about 3540 Hertz got a little channel
54:33 It's tuned in red at about 80 . OK. So here's a
54:41 Uh Now I've displayed it with So it's a prettier picture Judy color
54:46 . Again, I'm using red for and blue for high in this
54:51 And we'll just go look through and see this channel now, I'm going
54:55 into the middle invasion on the You see the channel going down in
55:01 . I've got an angular in uh reverse fault down here, like
55:08 fault over there. So let's go at this channel. What's the spectrum
55:15 like? Here is the channel? magnitude tuned at 40 Hertz?
55:26 Inter inter flu lower magnitude tuned at Hertz. And then, well,
55:35 the heck is going on here? got, got two peaks peaks,
55:40 peaks in the spectrum. That's kind weird. Not a nice Gaus and
55:46 . So let's go look at Here's my picture horizon. I went
55:52 the picture horizon, the Phantom horizon . See I could pick this
55:57 And this one here, well, a channel here and a channel there
56:00 a channel here and a channel here , that would be harder to
56:05 So I picked this Atoka Horizon and what that high amplitude moderate frequency guy
56:13 like right here. That should And then this is a lower
56:20 lower frequency, longer wavelength. That's guy. So, so far,
56:25 good. And this was the peculiar . Oh Well, if I look
56:29 this zone, I've got very high and low frequencies. So it's an
56:36 and conformity and my spectrum is more . So what I'm measuring with the
56:41 decomposition is in the data, it not be hard to understand. It
56:45 not be easy to understand, but in the data. OK. So
56:51 a fandom horizon slice through in size from Anadarko Basin of Oklahoma. And
57:00 what we have here is uh oh, we picked a carbonate reflector
57:08 dropped out. OK? Because you , I've got peaks and troughs and
57:12 crossings in this slide. Uh This is uh is no permit area or
57:20 . Uh Actually, it's a no area. OK. So when CGG
57:26 newer surveys to make this mega um they didn't realize they didn't have
57:33 that covered it. So there's a in the middle of the servant.
57:41 . Here's the coherence image looks Here is co rendered 2035 50 Hertz
57:49 . Hey, these channels show up nice. Uh where there's strong
57:55 red, green and blue are all uh the threshold of my scale
58:01 Uh It shows up as white. ? But here this guy where it's
58:06 kind of a cyan colored. that's between green and blue. So
58:12 turned to maybe 47 Hertz and where yellow? Well, that's between red
58:18 green. So that's tuned at maybe Hertz. So you have a,
58:23 idea of relative thickness. OK? let's go compute coherence. Now,
58:32 have, at the end I asked to do 10 exercises, 10 labs
58:36 there might have been two or three I put in there. Um
58:42 Stephanie. And you don't have to that. Right. I just said
58:47 gonna grade you on the 1st But um you know, Carlos,
58:51 like, he's got time on his . So he's, he's gonna do
58:55 coherence calculations on the spectral components. , guess what? Because the wavel
59:03 changes with tuning at the before, the tuning frequency that means different spectral
59:14 are going to see those channel edges . And when we get to f
59:22 our auto tracker wants to correlate across fault because like my fingers, uh
59:29 got a fault in between my fingers everything has been shifted. So that
59:34 the peaks of one horizon line up the peaks of the next horizon.
59:38 , my auto trackers go right across and coherence. If I'm looking at
59:43 change in wavel across there, it's gonna see that fault either. But
59:48 I change the frequencies that changes the of my fingers, now, I
59:54 see it. OK? So what got plotted here is the coherence computed
60:02 20 Hertz, 35 Hertz and 50 and where they are all lined
60:10 I get black. But that means magenta guy here. Well, that's
60:18 between, let's see, red and green. So that means its
60:27 to maybe at 17 Hertz or So you get this extra understanding of
60:33 going on. OK? If I that with the broadband coherence. Now
60:40 can see where do I get a bit of uplift? Where do I
60:43 extra information by computing coherence at different ? Well, this channel seems to
60:50 up, I see some little bitty features here. I see a little
60:55 um down here. OK. Where edges are showing up at one or
61:02 frequencies. OK. So we've got methods. There's, there's one or
61:10 others out there. But let's say got three methods or spectral decomposition and
61:17 short time fourier transform, continuous wavelength and matching pu OK. So uh
61:25 kind of uh synthetic was designed by Castano and you might see him sometimes
61:33 in and out of this office into next office. What he did is
61:37 took a bunch of wavel its, knows the spectrum of the wavel.
61:41 he knows this complex spectrum of the Hertz wavel, the 30 Hertz,
61:45 20 Hertz, the 60 Hertz He knows the spectra. So he
61:50 add those specter up the complex spectra the magnitude. Here's the true
61:58 Then he can add the wavelets up the time domain and here's the
62:04 So which of these three methods that discussed best represent the true spectrum?
62:13 . The problem is a little bit and non unique because what I've got
62:19 measurement. So let's say 1000 samples I got and two millisecond data means
62:26 have 501 measurements. And now I've um 100. So I've got 20
62:36 501 I have 20 times as much as input. Um That's, that's
62:43 of non unique, right? So I add those output up and I
62:47 the input, that's what we That's the goal. So here's the
62:53 time for you to transform, probably 100 millisecond window. And you can
62:57 what is done this anomaly here where have two strong reflectors of different
63:04 it's blurred and continues wave transform does little better and matching pursuit good better
63:14 still. And then here is uh events, five events and here's the
63:22 specter, here's what I get out the short time fourier transform blurred,
63:27 too much at high frequency, something wavelength transforms better. And then here's
63:34 pursuit a better still. And then event here a short wavelength and a
63:41 wavelength or low high frequency and a frequency. Here's what the true specter
63:46 be short time point to transform. It's actually a little too low um
63:54 when transformed better. And then here have distinct measures. So for this
64:00 , anyhow matching Pursuit has higher higher spectral resolution. OK. So
64:08 are two things in this picture. want you to know one, if
64:13 work with other companies and different people are gonna change the names on
64:18 . Ok. So this particular paper a guy called Weer, he worked
64:24 Norse Heathrow. Well, even Norse changed its name now it's Aquino.
64:31 . Um Leopard didn't change his He's still Leopard. And uh he
64:36 using a software called Geo Teric from company called Foster Findlay Associates in Great
64:44 . Uh And it's, it's one the, you know, it's
64:49 it's a nice piece of software with color displays and so forth. So
64:55 them, they have these three methods they're calling them constant bandwidth, constant
65:02 matching pursuit. And what they really is charge window or you transform continuously
65:11 transform in that six. So why they have different names? Not?
65:16 don't, there's no copyright here. were the, the methods were all
65:20 by academic types. I think it's marketing. They want to make themselves
65:26 different then another person or it could some of their scientists came from an
65:34 engineering background. So constant Q is we use in circuit theory that doesn't
65:40 to geophysicists just expect confusion and try figure out what the methods really
65:48 OK. So sometimes you'll hear about new method and it's something you already
65:53 what it is. It's just got different name. Now, what you
65:56 to notice here is the difference in resolution. Hey, smeared vertically bad
66:07 vertically a lot sharper vertical. So less vertical mixing on the matching
66:16 then the short time fourier transform or continuous wayward transform. OK. Now
66:22 look at some horizon slices. So got a big horizon slice uh on
66:27 top and then some little windows A B below. So here uh is
66:33 short time fourier transform continuous wavelength transform pursuit. And oh look at,
66:41 see this channel right here and I pretty good over here that better
66:50 there. But OK there, then I've lost it. And then I've
66:57 a dark, you know, so amplitude channel here with a stronger
67:05 I see it. Well, here see it. OK. Here and
67:08 watched it and then here I've got nice oxo uh weaker, losing it
67:16 . I've lost it. So I'll on TESS only because she's in the
67:21 and I never picked on her. one do you like best? She
67:33 the middle one? OK. How like the left hand one? You
67:42 , it's a trick question. Alicia, which, what would you
67:45 ? You see more stuff? Right the left one, we're seeing
67:50 Stop. This one is actually the . So the more stuff we're seeing
67:57 due to that mixing I showed in previous picture. So this little Oxo
68:05 either it might be 20 milliseconds above 10 milliseconds below. If I animated
68:10 , I'd find it. But what doing is we're, we're mixing.
68:16 this is a game you have to with seismic attributes, you know,
68:21 got defaults like with their uh their volume. A defaults 15 samples.
68:31 , if my data are sampled oh, maybe it's uh data from
68:40 Faja down there in the Orinoco Basin Venezuela, which is some of the
68:47 biggest. Our sands, they sample data at a half a millisecond up
68:53 the Athabasca sands in Alberta. They a half a millisecond and other data
69:00 see especially older data with ocean bottom where you have limited space, maybe
69:06 milliseconds. Well, 15 samples is a bit different for the different kinds
69:10 data. Oh And then some of may apply these techniques into patrol because
69:16 do that uh 2 3D radar data that's got that's got sample interments in
69:23 . So you have to think about and for coherence, I don't know
69:29 I talked about this the other Um For coherence, the best a
69:35 window is you look at the dominant in your data. So you're gonna
69:40 the dominant period, go to your of interest, look at the peak
69:44 peak distance or the trough the trough . That's the dominant period.
69:50 As my window gets my, my period is already mixing Strat gray.
69:57 ? As my window gets bigger and . Well, I'm kind of looking
70:00 the same stratigraphy, a little bit mixing more mixing, but I'm definitely
70:04 the signal to noise now as I greater than the dominant period, now
70:11 adding more Strat gray. So, know, start with the dominant period
70:16 then see maybe I can make it , sometimes you can go to two
70:20 three samples that, you know, good quality data. Um So that's
70:25 rule. So we have Strat gra is one of the issues we have
70:31 um seismic attributes in particular. So and destructive interference from the top and
70:40 of the thin bed did rise. changes in the seismic amplitude and phase
70:45 components of the specter can be used detect lateral changes in layer thickness and
70:52 well below the limits of water So you can map it, people
70:58 this all the time. You can't how thick it is, but you
71:02 map it the peak spectral frequency. mode of the spectrum is a is
71:07 good zero order representation. The seismic response. There's other measures you can
71:14 that might look at be sensitive to binding upward, coarsening attenuation. So
71:22 of you may be interested in. , co2 sequestration, I'm putting carbon
71:32 into the subsurface that's going to fill pores, it's going to scatter the
71:40 data. It might have this squirt where I'll change mechanical energy into heat
71:47 . So my spectrum is going to , how am I gonna measure those
71:51 in spectrum. How many spectra decomposition exploration? Uh Common use of spectral
72:01 is with gas hydrates. I'll probably some pictures of gas hydrates tomorrow
72:06 but that's where the methane and water into ice that you can write a
72:10 to. OK. Uh And so gonna be a attenuation, not in
72:16 hydrate but where you have the free underneath the frozen hydrate. So a
72:21 of use for spectral decomposition and of three methods we talked about the matching
72:28 provides less vertical mixing of Strat gray that based on, let's say discrete
72:33 transformation. A short term. Test question. I'll give that's another
72:46 . What's the answer? You're not like a geologist? Six.
73:07 OK. Let's take another 10 minute . Continue working on your stuff.
73:13 is volunteering to be user help bless his heart. And by the
73:22 , that's the handicap for the folks are remote because you learn by the
73:27 next to you. You say, , I don't wanna do what that
73:31 just did sometimes it's that way or , that's the way you do
73:35 Then you wanna copy the next either way you learn. OK.
73:53 we had a question on lab What page was it on ma'am?
73:59 five. OK. So we just talking about spectral decomposition. I mentioned
74:06 Geo Teric does they have three Landmark has one method of discrete short
74:16 48 transform. And then they have on things from John Castano's company like
74:23 Pursuit and, and another one called entropy. Um and uh GEO terror
74:30 GEO Teric uh paleo scan guys, find continuous wavelength transform and vel geo
74:42 continuous wave with transform. So those people were out there pare does
74:48 it doesn't do any of those It does something much simpler. I
74:57 call it. And a fellow called Lean Gao would call it uh a
75:03 probe. So what you do is out of your hat, you pick
75:11 frequency you're interested in and then you either a cosign or a sign and
75:19 cross correlate. So you're getting one correlation coefficient and if it's a 20
75:26 wavel uh and you can display the , they don't look too bad,
75:31 tapered. So you're gonna have a Hertz uh sign wavelength that's going to
75:36 you the 20 Hertz cross correlation coefficient the sign. Now everybody else in
75:45 world uses spectral magnitudes. So we two ways of computing no, we
75:53 three ways of computing spectral magnitudes giving the patrol software. One, I
76:01 the 20 Hertz of the sign and compute cross correlation coefficient, take the
76:06 Hertz of the cosine cross correlation Then in the calculator, I add
76:13 square of the one volume plus the of the other volume and then take
76:19 square root and that gives you the or you think of it. It
76:27 , then you say, isn't that we did with the envelope? We
76:31 the data and the Tober transform the two of them added and took
76:36 square root. Yes. So you take the cosine or let's say the
76:43 transform and compute the envelope curing And then the third way I could
76:55 the sine transform computers quadrature which is be the cosine transform sum those two
77:03 up, you get the square So the simplest way to do uh
77:09 spectra magnitude is going to be let's say the cosine transform or the
77:14 transform one of them and then compute envelope of that. Now do I
77:22 the in the lab? Do I about virtual calculations in patrol?
77:35 I did. OK. So here where they're very cle very, very
77:41 , right? And also is why of their software seems to be computed
77:51 a peculiar way. So most people here's my volume, I'm gonna give
77:56 a volume. Well, if I a volume from Trinidad offshore Trinidad,
78:06 100 gigabytes, that's pretty common 100 . And then you're gonna ask me
78:12 give 20 spectral components. Now I 20 100 gigabyte volumes. I start
78:17 up all my disk space. So I wanna check things out. So
78:24 patrol, you can compute the any let's say the 20 Hertz sign
78:40 OK? But don't realize it. ? Then I'm gonna take the envelope
78:50 that 20 Hertz component. And then I compute them, if I just
78:58 to look at a time slice through envelope, the software is smart enough
79:03 use 15 samples for the envelope. think that there maybe it's 21 whatever
79:08 default is, they're gonna compute 21 and to compute 21 samples, they
79:14 to do that sign transform, not the whole trace but just for 21
79:21 which uses maybe 21 plus or minus samples up and down. So they're
79:26 computing a small part of the data give you a time source. If
79:31 compute the whole volume, this is inefficient. But if you just wanna
79:35 at a time slice or horizon slice a virtual line or in line or
79:40 cross line, it's really, really . So what uh Arles can do
79:50 while he was working at us, gonna do this with a 20 Hertz
79:56 , fine component and then he's going compute variants from it. So instead
80:05 generating a volume of the 20 he's just gonna not realize that.
80:10 compute variants and you may or may choose to realize it. You get
80:15 , a 20 Hertz volume or 30 volume or 40 Hertz volume.
80:19 how different do they look? They're to be different, we're gonna be
80:27 . Does that help a bit? instead of giving you the answer I
80:32 you three which looking at the expression your face you were unhappy about.
80:49 , we're gonna do Yeah. Let's do them. Yeah. You
80:56 do the spec mag. Yeah, gotta calculate it and there's three ways
80:59 calculating it. Do the way you it's simple or the way that's queer
81:06 you, maybe that's better. Yeah, because I I've shown you
81:14 bunch of pictures of co rendering with magnitude. Everybody co renders with spectral
81:20 . Like what does negative red Uh uh It's a little complicated.
81:29 . What you think of it? . I'll stop again. So we're
81:46 go and uh talk, not sure attributes this is gonna be I think
81:57 would be pretty easy to understand and everything else we've done up to present
82:04 start to make them. All So, so this is gonna be
82:15 11. OK. Um At the , so attribute expression of tectonic
82:27 So today and tomorrow we're basically gonna about how do seismic attributes help us
82:37 tectonic features, Strat graphic features or depositional environment, carbonate depositional environment.
82:47 then hopefully we'll have time to talk little bit about hats. Thanks.
82:54 we're gonna use coherence to accelerate the of faults on 3d volumes. Use
83:00 attributes to provide a preliminary interpretation across surveys that have different amplitude and
83:09 We can identify the appearance and structural of salt and shale dye appears on
83:15 attributes. Use curvature to define an planes and use coherence and curvature as
83:22 aid to predicting right. Things are most common, most important faults if
83:30 in a tensile terrain, so I the pull apart stresses, I'll have
83:36 faults. If I have a compressive , I'll have reverse faults. If
83:42 have uh sheer stresses, then as look across the fall, if things
83:49 moving to the left, that's left , and if I look across the
83:53 and they're moving to the right, right lateral. And so here is
83:59 normal fault kind of review. And my house, it's always my
84:07 That's what's normal. Ok. So terrains, there's a little block with
84:17 fractures in it. On seismic we can see offset quite a
84:24 You know, the faults you you could see some offset on
84:28 the fractures. Nah, we're not see that, you know, usually
84:33 don't see fractures on seismic data and seismic attributes rather what you're going to
84:40 , you're going to infer fractures by the lithology and having structural theology
84:50 Ok. So you're going to well, if my fault is not
84:58 , you know, my fault is of crooked in places and I'm grinding
85:04 , the hanging wall is grinding against foot wall and my structural geology colleagues
85:11 me that a quartz rich rock is to fracture more easily than a quay
85:19 rock, right? So we got the geophysicists, we're providing the shape
85:25 the fault. Oh And even by the offset between the different horizons when
85:31 fault was moving, ok, by difference in accommodation space. Uh But
85:38 we need to use a structural geology to predict fractures. We're not gonna
85:42 fractures directly. When you talk, Leon Thompson talks to you,
85:49 he'll uh present a lot on anisotropy that, that map fractures and
85:58 That's a, that's a direct But from post act dating, you
86:03 . OK. Here's some growth falls uh Gulf of Mexico. I happen
86:07 be in, in Mexico uh northeast Mexico, south of uh Texas and
86:17 have to think of San Jose Las and I forget what basin it's
86:22 But anyhow, just taking the seismic co rendered dip in as you and
86:30 can animate through it. So, this picture and you're probably comfortable now
86:34 most of you are plotting dip. the uh we're looking at a west
86:38 east line. If I'm dipping to right, it's kind of reddish colors
86:42 I'm dipping to the left kind of colors into the plane is blue in
86:49 me, uh yellow. OK? then where you can make a little
86:59 and you look at this, you them color code and you start to
87:02 a bunch of district fault blocks that this way, they're rotated the next
87:08 . Um It's pretty complicated but you a quick feel of what the tectonic
87:17 in this particular base. OK. we can look at a time
87:24 So we're gonna look at coherence co with volumetric dip and uh the magnitude
87:30 dip avenue up shallow. We see . Uh land data are usually
87:36 you know, collected on some kind a grid. This is rectilinear,
87:39 very often they'll have brick formats and they'll have parallel parallelogram type uh
87:49 So you'll see different kinds of but it'll always be a regular grid
87:54 that's the economic way to collect the . So I've got a, a
87:58 animation group of this as well. you know, if you look at
88:05 level, OK. Well, these fall blocks are dipping to the
88:11 . Here's my north arrow, these blocks, they're dipping to the
88:18 these fall blocks, they're dipping to southeast. This one is kind of
88:25 of flat with a light slight dip the north and you keep going through
88:31 you get an idea and the coherence on there as well. So you
88:35 all the faulty. So very quickly get a quick understanding of what the
88:45 style is and then if you were to pick faults, maybe you'd
88:50 well, which faults are the most to pick which ones I should I
88:53 first. Right? Ok. Here's faults in Alberta. Uh And I've
89:03 a slice through seismic amplitude and you think, well, maybe that's a
89:09 and I'm gonna pick every 20th line then I'll join that fall. And
89:15 we ran coherence, we realized, , this fault is on echelon like
89:22 step when you looking and it's not to be a good seal.
89:28 So what the coherence maps do is allow you to see kind of the
89:33 of the fault. And without they can help you name faults and
89:39 them. Here is a polygonal faulting the North Sea, the seismic uh
89:46 slice of the seismic camp, it's complicated. And here is a semblance
89:53 without dip steering. So the same you would get with variants without dip
89:58 . And here is a coherent uh structure technique but with hip steering and
90:04 we see are all of these um faults. OK? And this is
90:10 the Seismic Amplitude data look like. I think I showed you this
90:15 I showed you the vertical data on on one of our, you
90:18 practice questions. Bye. So running variants along structural depth helps us understand
90:30 style. I mean, it would forever to pick those individual form in
90:34 conventional sense of. And there I've them tied with the right scale and
90:41 are co rendered. OK? Um , excuse me. And we're gonna
90:55 curvature and we've got the, these anomalies uh positive curvature in red.
91:02 is the most positive curvature. There a little bit of blue. That
91:05 it's like my eyeball. It's uh know, a structural ho OK?
91:11 where it's white is it's not deformed the positive direction. So what I
91:18 you to notice as we animate through we're matching. I want to reuse
91:27 word uh we're ma mapping curvature, on that vertical seismic data. Sometimes
91:37 the edges of folks, sometimes they're . OK? But it's definitely mapping
91:45 and there's no no argument that it mapping the positive curvature in that data
91:53 as you uh animated across, then up to you to decide.
92:01 is that a fall or is that football of a normal th right.
92:14 cook a bit. We can do same thing with the most negative
92:17 And this is gonna measure the features are more syco so in many
92:22 they are the hanging wall next to fall and in other areas, probably
92:31 real sin, you know, like F but the point is it's mapping
92:41 features in the data and you've been curvature, it just, you
92:45 just run it. It's a lot than picking all these forms and
92:49 I mean that would take a long time and you'd be be
92:54 OK. Here's a example. Another from Alberta show coherence and it's got
93:05 bunch of an echelon faults. Most curvature. Gosh, that looks
93:15 Most negative curvature. Oh Let's put together. Like I showed you in
93:20 one using transparency in powerpoint. In powerpoint? Because all we had
93:28 the PDF of the Polish paper from , right? We're just gonna put
93:32 together there we are. So what have now? Well, I've got
93:40 anomaly, coherent anomaly and a structural in between. Wow, that's probably
93:49 horse. Now, if you're comfortable that being a horse and you should
93:54 comfortable with this and that being a and another horse. OK. Then
94:03 you like horse, what if I a negative anomaly in between the two
94:07 ? Well, then I gotta grab and there's another and another. And
94:15 I've got the drained. So nobody admitted to being a structural geologist
94:19 right? Because you knew I'd pick you, Zach. Are you my
94:24 geologist? No. Hell no, said. Right. Hell no.
94:30 . Cho, I got a fault . I got a fault system
94:35 I, I have to somehow, that fault ends in this fall,
94:41 I've got to get that strain from fault walk to the next. So
94:47 it gonna do it? There's something a relay ramp. There's gonna be
94:51 kind of a flexure in between. . So that's what we're mapping
94:58 That's probably a relay ramp connecting And if you're comfortable with that,
95:02 , there's another one and some here's another relay ramp here and then
95:06 might be some antic fault. So can see how the curvature and the
95:12 together shows you the, the real faults with displacement across them. Uh
95:20 subtle faults where you don't have a coherence anomaly, you just have a
95:25 anomaly and then all the folds that associated with with as well.
95:34 So this picture is from Beckham and near San Antonio. And uh you're
95:42 with this uh right in here. Stephanie, where have you seen that
95:54 from, from the quarry here? , Anthony, right. Where have
96:09 seen it? Yeah. Which Yeah. Which building on campus?
96:17 this, not this one pretty ugly . How about Fleming? That's where
96:22 comes from. That's where the, stone in the ou library comes
96:27 This is probably the second most popular building stone in, in the United
96:33 . The first is uh Indiana limestone is just kind of creamy and nothing
96:38 it just produces real smooth. That's it comes from. OK. And
96:43 kind of the uh Yano uplift. these two guys there at Southwest Research
96:51 in San Antonio, they were able get permission to go to this
96:55 which is kind of cool. And neat about a quarry? Well,
96:57 go map it and then you come the next week and see what's behind
97:01 you just mapped and come back again map it again. So, kind
97:08 point out. Notice this fault face straight. Ok. Geophysicist. We
97:18 to, I'd like to think that faults always look like this.
97:21 Because they're easier to pick. And they're like this, well, they're
97:27 to pick. OK? If they like this, this is where you're
97:33 have more strain, especially as you a block against it, they're gonna
97:38 of grind against each other. There'll be a little tensile here.
97:41 be more compressive, we're gonna generate here or there again. What the
97:46 geologists do? Another thing? Notice fault here. This isn't straight
97:54 It's crooked. So you remember from geology, how many years ago?
98:05 Wester, three years, couple of ago. So everybody remembers intro
98:11 You remember Moore Circle and then Coon's . So each material is gonna have
98:21 coefficient of internal friction. And when put um compressive stresses on it or
98:30 stresses, tensile, usually rocks are , tensely, they're pretty strong,
98:37 supplies. OK? So, so the rock, if it's under
98:44 stress is going to break according to law and, and, and more
98:52 . So the angle that it breaks is gonna be different for different
98:59 So on this picture, you'll notice you're going through what uh their own
99:08 called, you know, with a unit one, it's got this
99:13 then it's got a little different then a different angle and then almost
99:18 . So you've got these different Now hang on, I got this
99:23 and is moving along a different There's gonna be gaps, there's gonna
99:27 gaps as I move along it. am I gonna accommodate those gaps?
99:35 ? How am I gonna accommodate those ? You know, moving, moving
99:40 along a crooked surface, gonna have have fractures like I'm, I'm I
99:48 like give and start to break So now a good structural geologist can
99:54 this information plus knowledge of the What is it clay rich? Is
100:02 quartz rich? Is it carbonate rich predict fractures from the shape of the
100:07 ? Right. So that's their role this whole thing and fracture prediction.
100:13 all model. We're not measuring the from the seismic, we're inferring them
100:18 a geologic model. Ok. So here is um I've got three
100:26 members or Farrell and Morris have three members and this is in the Edwards
100:31 . So, uh in South Texas North Mexico, uh the, the
100:39 . Uh Well, we have the limestone, I'll, I'll come back
100:42 the other one. but they produced the Permian basin and so forth.
100:47 this is a very competent rock. now let's think, what would this
100:51 like with seismic attributes? Well, have a coherence anomaly, a strong
100:56 anomaly with a lot of offset. then it kind of tails off nothing
101:02 a moderate coherence anomaly that gets When we look at it from
101:11 we're gonna call that a horse tail pattern, no interpreters, call it
101:16 tail type pattern. So it looks the tail on a horse.
101:21 Now I gotta get the strain from system over to this block. I'm
101:26 do that with a re a relay . So here's the relay ramp.
101:31 a little bit curved to fit I'm gonna have a big radius
101:37 So a small negative curvature anomaly and a big radius of curvature, a
101:45 radius circle appeared. So a small curvature now, OK. So that's
101:53 pattern we're gonna see. Now we'll to the Glen Rose formation. So
101:58 what's on Fleming Hall and the library on campus, et cetera. And
102:05 what are we seeing? This is little less competent rock. So instead
102:08 a big bulk, we have smaller . Here's what coherence would see.
102:16 I look at a map view, would call those an echelon fall.
102:21 stair step looking, OK? And , I'm not gonna see these conjugate
102:32 on seismic data. They're gonna fall seismic resolution. They're going to be
102:38 know, maybe 1/10 of a wavelength something like that. So, what
102:42 going to see seismically is a curved . So I'm gonna have a positive
102:47 anomaly on the foot wall and a curvature anomaly a little weaker on the
102:55 . And then we've got a relay camp except here it's a steeper relay
103:00 more, more conformed. Then we to the Eagle Ford. So the
103:06 Ford is one of the shale It's the second or third biggest shale
103:13 in the United States. Uh First the Permian basin which has got like
103:19 or seven levels. And then the is either number two or the Eagle
103:22 number two and they switch back and the um, so the, uh
103:31 not the Eagle Ford shale. We it a shale, but it's really
103:35 50% calcium carbonate. So it's really black limestone, right? But we
103:42 fracture and, and produce oil out it. Now, in this
103:47 it's a little bit more plastic, little more tile because it's got some
103:53 in it, quite a bit of in it. And we're not gonna
103:58 a coherent anomaly. They can see fall in the outcrop. So they
104:04 this is a pretty big fault, it just looks like it draped over
104:11 . So we're gonna have a positive anomaly and a negative curvature anomaly and
104:16 gonna bracket the fault but we're not have a coherent sly. So here
104:22 the three in members they competent call doctor or brittle if you
104:29 Now, the first application of curvature mapping fracture enhanced production was in the
104:34 formation back in 1968. And uh of the, the the structure map
104:42 made mainly from well tops. They have had a little two D seismic
104:46 3d seismic. And so they calculated structure and then computed curvature of that
104:54 structure map and found a ha where curvature was highest in the yellow
105:00 Those were the best producing wells. I take a relatively brutal rock and
105:09 bend it, the strain is greatest the bends are most and where it's
105:18 dipping plainer, there's no strain. where the fractures occur is where the
105:23 is greatest. OK. So that's correlation of curvature with fractures. Curvature
105:29 see fractures, it sees and measure strain. Then we use a geologic
105:36 says ah sprain is one of the things we need to have fracture.
105:41 . So we got different fold. a fold with a positive curvature,
105:45 and negative curvature anomaly. And here's reverse fault positive on the uh hanging
105:52 , negative on the foot wall, fault real common positive on the foot
105:58 negative on the hanging wall. Ciconte P down in uh Mexico.
106:08 . This ha here happens to be volcanic Sill. You guys are probably
106:13 with this, the name Ciconte Peck uh a means seven volcanoes. So
106:20 seven volcanoes in the survey of pretty geology. I got a pop up
106:30 . You remember what a pop up is? We should remember what a
106:35 up block is. OK. We strike swift system. Compressive.
106:45 a little bit compressive. It's right? You're a geologist,
106:49 So when you know, you wanna people, you know, really unsure
106:53 what you do. Oh Let's put the words together and we're gonna call
106:58 trans expression. Mm We also have tension. I might have an
107:07 So we're strike slit and then some the blocks are gonna, I thought
107:15 , so we have a fault on side and because we're squeezing them together
107:19 that, it just comes out So that, that's a pop up
107:22 . So here's one here a point the mouse so that other folks can
107:29 . So here's a pop up see the fault on both sides.
107:33 then here's another pop up block. A third one. There might be
107:41 . My volcanic seal, a volcanics up here and here's what coherence
107:52 . Uh It's OK but you not really talking to me. No
108:01 , false, low coherent radiographic Oh Render it. It's OK.
108:07 helping a lot. Let's look at curvature now. Hey, that maps
108:11 pretty nice. OK. Now, just mapping two things right here.
108:20 mapping the fold axis of that pop club. So it's map on the
108:28 on the side. It's mapping the of her fall here. It's mapping
108:35 edge of a fall. Here it mapping the edge of the fall.
108:39 . It's mapping the fall of actions . OK. And here mapping the
108:44 of a fall here is mapping a axis. You're mapping of whole
108:50 So you know, one of the that you don't wanna, you
108:55 some people will say look at that coherent. I see our cur
109:03 I see fractures, I see Well, step back and think a
109:09 bit is everything anomaly you're seeing. it a fault or could it be
109:16 a fall? You know, in case, I've got bored.
109:27 The top of block. OK. negative curvature it's showing the valley shaped
109:32 of features again. It's looking, seeing the edges of the pop up
109:39 edges of the pop up block. fall in fall. Oh And in
109:44 areas, I might see a little structure uh I have to search
109:51 Most of them are are false. this case, like even here,
109:55 is actually part of a a Um There's my top up block.
110:07 It might have shown this picture before this picture. My tectonics is such
110:15 my lithology is such that my reflectors wrote uh I've just slid them along
110:23 fall. So what do I I have a coherent anomaly, the
110:27 it changes across the fall and maybe amplitude changes across the fall here.
110:34 got a f with Conjugate fing work imaging both will do the same
110:41 And now I have a coherent anomaly the middle and on the, I
110:47 , well, for a normal I'm gonna have a negative curvature
110:52 And on the foot, I have positive curvature anomaly. So I'm gonna
110:55 positive curvature, coherent negative coverage. if my fault offset is below seismic
111:05 or maybe it's just been a fog Fold kind of affair, then I
111:11 have a coherence anomaly at all. I have this red blue pattern,
111:17 curvature, negative curvature anomaly that's bracketing ball. OK. Trump. In
111:22 cases, I would be comfortable saying a fault. And then if I
111:27 a bey as well, which we about last week and I won't ask
111:31 about that on the test because that's little bit more uh advanced and it's
111:36 in patrol. So you're not comfortable it. There, we can map
111:41 called location. OK. Here's an growth hall where uh I'm sliding along
111:54 . I'm gonna see that fault on because the dip on this side is
112:00 than the dip on the right. . So I have a change in
112:04 and then the wavel is gonna be too because they're not a line.
112:07 I'll see that falls on both curvature coherence. Good morning. Uh I'll
112:18 a coherence anomaly. But if my haven't then rotated, I'm just looking
112:26 flat tip on the left, flat on the right. I have no
112:30 in enclave and in this one, things are pretty well aligned. I
112:36 have this little bit of offset I'll have a, a small curvature
112:41 because I'll be able to map that change. But I won't see a
112:49 coherence in normal. Here's a maybe fault down here and it becomes progressively
112:58 of a lecture. So I'll see fault if that's how you want to
113:05 it on curvature, but I'm not see it on coherent. And then
113:09 I have infill of grabbing, I have offset on the fall down
113:16 But then as I fill it in there's compaction, all I end up
113:21 is approach anomaly and down here, may not have a curva now.
113:27 you can see how you, you to put all these measurements together in
113:30 head in the context of a geologic . OK. Uh This data
113:39 uh another data set from New it's from Southern New Zealand. So
113:43 the Great South Basin and it's got bunch of fall paint in it.
113:47 has other features as well, but shell, it's got pretty complicated fall
113:53 . And here I plotted um coherence from yellow for high coherence to
114:02 OK. And then I've got my and I have my amplitude. So
114:08 got positive and negative curvature against this binary color bar and I have coherence
114:14 this. And the, if I high coherence, I'm gonna make it
114:18 to the background grade. All So on this Timewise, I have
114:28 racketed or the curvature brackets, the seen un coherent. So if you
114:35 carefully at this, I have like , a red, yellow blue
114:39 red, yellow blue pattern, a , yellow blue pattern. And on
114:44 vertical, you might, on the slides, you might be able to
114:47 that all. Then if I go little deeper, I've got some areas
114:55 I don't see any yellow anomaly, coherent anomaly. And there my my
115:03 on the faults are small or I a lot of conjugate faulting. So
115:08 just looks curved, but I can be comfortable calling these faults because it's
115:13 the same pattern and it's the same as this right over here. All
115:20 have is a coherence anomaly. So of these attributes is not better than
115:26 other, they're complementary. I've got couple of animation loops. This one
115:33 go, I think vertical, I took 20 slices. Uh We're better
115:43 doing it this way we do This way. OK. So down
115:50 , I just see basically blue a little bit of coherence anomalies over
115:55 go up a little higher. You see how the here I have good
116:02 . No curvature. Here, I all three attributes method, mostly
116:13 anomaly, a little bit of coherence here. More of a coherence anomaly
116:19 then go here up here and the one, I'm going to show you
116:24 vertical slices what it looks like going vertically. So you can see,
116:31 , I got, I've got those , blue pattern that I've been talking
116:39 the red on the foot wall, on the hanging wall, red
116:46 on the football glue on the hanging . And in fact, I'm
116:55 thinking pare, sorry. And now was my Yeah. OK. But
117:15 starts the program. OK. what can I? Welcome back a
117:44 . All right. So one thing need to um let's go to the
118:03 one. Oh no, let's go . You see this blue smear and
118:07 red smear like what the heck is on there? Oh, no,
118:10 on now, don't do that. is the program? I'll try it
118:23 more time. OK. In the . OK. OK. So this
119:07 smear or the red smear, it's be obvious to some of you but
119:11 all. So the faults are going different directions that vertical line in the
119:22 . If it's parallel to one of folks, I'm gonna see maybe just
119:27 hanging wall and I'll get a red . If it's a little further
119:33 I'm gonna see, I'm see, see the foot wall, I'll get
119:38 red blur and if it's on the wall, I'm get a blue
119:41 Ok? And when the fault is , I get a nice sharp
119:46 So that's what's happening here. So there's faults there, some parallel
119:56 this line. It's as simple as . Whereas if I go up
120:02 see the faults are perpendicular to the here, there's some parallel again.
120:06 that's an appearance uh When you slice data and come on now don't crash
120:16 me again. Oh oh A reach in it doesn't, is it dying
120:32 any of your guys? I don't recover. Oh Let me.
120:54 Now I'm on slide 38 same survey , sliced through amplitude tho through
121:04 So I've got a, a red slice, a green thyme slice.
121:11 here I'm going to see some pro there. So I got coal faults
121:16 between. I've got some CESA going and here is a coherent image at
121:26 shower level, seeing the faults and this deeper level seeing channels cutting through
121:36 platform. I got wi thoughts I've these stair steps. Do you see
121:43 vertical stair step? That's pretty And you're gonna see that in all
121:47 coherent calculations and the variance doesn't Now, there's a technique out there
121:56 pro probability. There's a couple of of doing it. And,
122:03 one of them is you think of coherent anomaly as a cloud of points
122:12 then you're going to take that crowd points and come up with and
122:24 So here's my cloud of coherence And uh what am I gonna ask
122:30 about? Yes. Yes. Not pence or what is it? It's
122:40 eigenvector, right? OK. OK. Which direction best represents this
122:50 of points? Hey, well, me on that. Well,
122:57 that's gonna, it's gonna be it was gonna go through it.
123:02 . So if I have this here's my, my fault.
123:08 the first eigenvector will be uh parallel the second eigenvector is going to look
123:14 this, right? And that's kind you got this diffuse thing with all
123:19 little wiggly. Well, you can find the two eigenvectors that best represent
123:25 . And well, let's go smooth this way and sharp and perpendicular and
123:31 with the help of your Eigen OK, allows us to go from
123:37 image on the bottom to uh this . OK. So you can clean
123:43 up and there's a couple of other like that and it's more of the
123:54 OK. Well, that those ice vectors on my plane, they're a
124:00 , they tell you a direction the eigenvector least represents the data, that's
124:07 normal, the, the coherent anomaly FT point. So now I can
124:14 the, the magnitude of fall point the AU I can code them.
124:21 here are these faults. Now, are color coded according to the zip
124:27 . And the zip mat I have in here, which is that the
124:33 and then I can bring them into , a box probe which is one
124:36 the last exercises uh where you can look at those false color coded in
124:44 . And, and then problem here's from the Gulf of Mexico and you
124:50 tell all that fault is dipping to southeast that dip to the north.
124:55 ? But blue and red dipping to north because it's blue and then you
125:03 move that guy around in 3D and code that so this is the direction
125:08 interpretation is going is trying to get closer to a finished product if you
125:14 , machine learning is gonna play a role in this. OK. All
125:20 . Here's a complex walking north slope Alaska. That's uh we're looking at
125:25 Shubert formation, which is this one . I can see some faults.
125:31 . So we're looking at mine the prime. So I see this
125:35 here where the yellow arrow is. on the time structure map.
125:40 I see little dimples here. here's my fault. There's a
125:45 I see them on the time structure . Not bad. And then,
125:54 , here's that one thought what it like. Here's the other kind of
125:59 and here's a third area where I see anything. OK. Here's the
126:06 anomaly. So if I have enough , I'm gonna have a low coherence
126:16 . And now you see the same horizon, if you look at
126:20 guy and you make a horizon place coherent, OK? I got a
126:26 of, you know, conjugate B looks pretty good. And then there
126:33 the, uh, I think he's color coded by arrow. Yeah,
126:38 the purple one is. Well, kind of sees these little falls and
126:43 the blue arrow is. He doesn't that on coherence at all. There's
126:47 a little flex. OK? let's look at the super Bey guide
126:54 is the measurement of lecture. now we see these kinds of good
127:01 in great detail. OK? But actually mapping these things as well.
127:08 of course, the Aber is the derivative curvature. So I think I've
127:12 a curvature now. Well, here's abery and it has you,
127:16 So I can break them into flex sets or bulk sets, those that
127:24 green which are flexing towards the northeast those that are magenta that are flexing
127:30 the southwest, you know, those of things. And then see,
127:34 , if production a better production or production associated with proximity to one of
127:41 both sites. And then here is negative curvature, long wavelength and then
127:47 weight. OK. Excuse me, some reason, I'm not gonna put
127:59 on my test, the exam for week for Wednesday. What I like
128:05 do. So, um I've got nice, clear falling and I would
128:15 that a groin. Here's my co and I'm like, yeah, a
128:19 bit of a groin. And then question is, is this a
128:28 Let's go. Is that a Is that a horse or are they
128:35 ? So here's most positive curvature. seeing the edges of the groin and
128:43 most negative curvature co rendered in powerpoint . Ok. Let's see, Mister
128:51 . So then the question is a . Is this a groin or is
128:57 so so on, on the I'll, I'll pick on Carlos.
129:03 this for or are they horsed and psychology? Do you think it's what
129:19 ? Ok. So why do you it's bing versus bowing? It looks
129:29 folding? But what do you think is now? You gotta be the
129:34 . Thank you. OK. So do you say it's palsy? Uh
129:39 in the, in the what from could like? Ok, Robert
129:59 what do you think of it faulting folding? So that say that
130:10 Yeah. Yeah, because you're gonna or Chevron's gonna drill? Well,
130:15 first thought the one on the it would be when something that had
130:23 do with the process? Ok. so that means, so what do
130:28 , you haven't stuck your neck out though? So you say,
130:31 you might be processing the shoes. agree with that. On the
130:34 Yeah, on the left. right. Ok. So what are
130:39 saying on the left? Um, , what, yeah, what,
130:44 do you think it is folding or ? So, ok. So uh
130:58 do you guys think? Put your hat on now? Don't, don't
131:03 like a science person? OK. say, OK, why do you
131:08 it's folding? Well, I, agree. It looks like folding.
131:15 looks like folding. OK. It's same survey. So it's clearly in
131:19 same basin. It's the same What can you say about the
131:24 It's clearly it's on the writing. , what kind of stresses do we
131:29 to form a groin? You have have tensile stresses, right?
131:36 And what kind of stresses do we to have proposing as the compressor?
131:46 you're gonna use a geological reason to ? All right, it's clearly faulting
131:52 faulting on the web. Horses and . And then because of all the
131:56 that Roberto said the processing issues, conditions or maybe conjugate faulting, there's
132:02 lot of reasons, you know why it may not be as resolved.
132:07 if we're an extensional terrain, the answer is it's gonna be more extensional
132:15 . So you're not gonna have folding an extensional domain. So in the
132:20 of Mexico, it's almost all extensional for the Perdido fold belt, which
132:27 on the Mexico US border and there have a little bit of the slope
132:31 coming down and crunching and, and a little bit. But that's the
132:36 that proves the rule showing most It's gonna be one style at a
132:42 time or the other. Yes, can reactivate normal faults and make them
132:51 faults. You can do that but , you know, 2 3% of
132:56 time. Uh but normally you wanna things. Keep it simple stupid,
133:02 ? The kids, that's the acronym , right. You're good at
133:10 Acronyms are like abbreviations. Keep it stupid. A is OK? Gonna
133:18 you. Now, Roberto knows that works for big oil companies. So
133:24 . Oh, but you're working, probably speak Spanish at work.
133:27 Uh Together everyone achieves more. What's acronym? Phh, there you
133:41 You had that. You probably have holidays evening, Saturdays and Sundays.
133:46 acronym P OK. So a lot interpretation is driven by a geologic
133:55 OK. So you really integrate So the more structural geology you can
134:01 familiar with both of you who when you're employed as a um geophysicist
134:07 they say, oh, hey, , we got a training budget.
134:10 need to use it, don't. car don't tell your management. You
134:17 learn more about static from seismic Um You know, I don't,
134:22 don't know anything about turbid systems. need to see Turbos an outcrop and
134:28 go to Northeastern Spain. Uh some in Ireland, I mean, nice
134:36 , nice trips. OK. Bob Canada, again, coherence,
134:48 positive curvature, most negative curvature put together. Now you try to figure
134:53 horse Robbins and um relay ramps, ? Oh I duplicated that one.
135:06 thought I got rid of that, I'm gonna get rid of it
135:14 Thanks. Hey, here's a time through seismic amplitude. Um I'm gonna
135:25 a coherent anomaly or coherent picture on and we got some queer falls and
135:33 the fault ends, got some falls the fall ends well, do faults
135:40 end or do they kind of taper co render again in powerpoint? So
135:48 pictures aren't as pretty as you can in patrol. Most positive curvature.
135:53 gonna be on the football party. , look at this. It's
136:00 I've got a curvature anomaly on one and it continues. I got a
136:05 anomaly on one side and it continues then I have these curvature anomalies.
136:10 put the most negative curvature on. That's on the other side and that
136:14 . So now I've got red, , blue. Oh Here is just
136:17 blue. I still got a Just the displacement on that fall.
136:23 small. OK. So I'm not a coherent anomaly down here. I
136:29 this red, blue pattern, blue pattern, red blue pattern.
136:33 got faults all in here probably FTS just small offense. So some interpretation
136:43 you can use. Sometimes things are , very complicated. So in this
136:49 , uh we did this at uh , these areas were so darn
136:56 Let's just go to find kind of edge, You have a deformation
137:02 So there's amalgamated fault. So it's , very complicated. Let's not try
137:07 pick those. Let's just pick the and the western part. And then
137:11 had some other faults here, strikes fall, let's go pick those
137:15 football here, strikes football there. what happened is when I showed this
137:21 a guy in uh Midwin in the basin, he says, well,
137:26 company, we drilled this little block and we got 10 million barrels of
137:30 . Well, kind of depends you know what your objective is.
137:36 those were important. Um You might done a workflow like this, picking
137:43 the Cora volcano and you're gonna look why a, a prime, I'm
137:48 this ball kind of sort of perpendicular prime. I'm cutting this fault or
137:56 fault from the salt film and a . OK. So a, a
138:01 should see sharp fault. EB I'm gonna see uh diffuse faults.
138:08 here's a, a prime. Nice fault. Nice fault. There's
138:13 time slide I showed you fault over . So yeah, I've got a
138:18 across those faults and pick and then one, here's the strike direction.
138:28 uh What do I do here? looks like you go down here.
138:30 looks like you go across here. looks like you go up. It
138:34 confusing in the right direction. So from the right direction are confusing.
138:41 pick an arbitrary line that makes a and I made it far enough out
138:45 I didn't see any faults on Just the one fault right here.
138:51 that one talk and now, that guy, I can make
138:57 Oh, I'm sorry, I'm staying . I'm sorry. I just put
139:00 auto tracker on and went all the around and back to here.
139:06 So that's one way to accelerate your . If you co render or if
139:12 in one image you show your coherence that. Well, let me pick
139:15 line, a seismic line that doesn't across any fall. Let me get
139:19 stuff correlated in here and start to some control in your day. This
139:28 from Northeast Texas and South Northwest So it's some uh formation called the
139:36 Valley limestone. And I've got uh slice through amplitude, time, sliced
139:43 coherence. And I might have mentioned the other day that, well,
139:50 coherence on the uh on the, sorry, the coherence is computed from
139:58 just this time slice but 11 And these faults are difficult to see
140:05 the time sheet. So here's my image I go through and now I'm
140:13 to use the coherence to kind of a strike line. I'm not gonna
140:17 able to make those fault surfaces. guys saw how picking coherence on time
140:23 is problematic, but I can definitely it fault A fault B, fault
140:29 fault B and give them different OK? So now I've got a
140:34 of these faults colored that cross And I can go up uh 50
140:41 and pick them again and start cross , for correlate the faults. And
140:47 I can look at a north south and pick some of the key faults
140:54 and then go 50 lines to the , make some more. And then
140:59 you know it, I have a grid kind of name and call
141:09 OK? So now I can go at the fall. So the faults
141:14 picked in this exercise here were pretty , but it's easy to get confused
141:19 you have the main fault and the fall. You know, you're
141:24 picking down to the right down to right and then somebody knocks on the
141:29 , you say hello, then you picking down to the left, down
141:32 the left. You pick a different really easy to do. And then
141:36 gotta erase them or rename them or to go to lunch. So being
141:42 to name them is, is very and then you could drop up and
141:47 a couple 100 milliseconds and do the thing. Get a course for
141:53 Once you got that coarse grid, gonna interpret false the normal way.
141:58 an example from Trinidad and here we've complex ball pain that's difficult to see
142:06 amplitude. You really don't see much pain. And here's the coherence
142:12 Oh I've got it. This is of those trans professional things.
142:19 So here's the main strike slip fall then there's antithetic F bye. And
142:28 we'll look up at the northeast Here's the Galioto Ridge. It was
142:34 coherence image when we did this work 1999 in landmark, we had 32
142:41 . That's all we had. We to live with 32 colors. I
142:47 my microphone on again and moved it what we see here. Well,
142:56 this side, it's dipping to the because it's kind of magenta and on
143:03 side it's dipping to the south. I got a change in dip across
143:08 fault like this. OK. So got my hand pointing in different directions
143:14 the fall. That's what you see Wrench tectonics. I've got rotation about
143:18 thought. OK? And I don't it across. So here's his real
143:25 zone. This is where I've got rotation and here uh I've got faults
143:31 I don't have much rotation, same . So we look at those two
143:35 . EE prime is perpendicular and DD is parallel. So DD prime going
143:41 across the Giota ridge, nothing exciting . My normal faults bunch of
143:50 OK. Uh Kind of similar Then here I'm dipping steeply to the
143:57 and here I'm actually dipping to the . Uh That's what you see in
144:04 tectonics. You'll see a complicated for one with a change in accommodation space
144:12 Alberta. We're gonna look at reflector . So how do, how does
144:19 dip change vertically? Ok. How things pinch out and open up
144:25 Which direction do I have less accommodation ? Which area do I have more
144:31 space? So in this little I'm pinching out in the south
144:39 So I have more accommodation space to north. This one I'm pinching out
144:46 the north northwest. So I have accommodation finished on the northwest side of
144:54 fault. So the fall, the aren't flat, just falling vertically,
144:59 rotating in the plane and within the block. And now everything is done
145:08 where it's white, there's no pinch , everything is parallel. Of
145:16 you can make a animation of Ok. How about compressional terrains?
145:28 kind of thing we showed in that , the faults aren't necessarily straight depending
145:36 the strength of the rock on your , the fall plane is gonna change
145:44 and I'm gonna have gaps. How I gonna deal with those gaps in
145:49 rock gonna fracture the rock? So then a good structural geologist,
145:55 knew this mythology and that lithology in shape of the fall. They could
145:59 where the fractures are gonna be most . OK. Data set from Canada
146:06 a reverse fault here. OK. Se Kinder, my buddy Chopra picked
146:13 horizon at a zero crossing 123456. show on faults here they are.
146:24 got a fault here. A fault . A fault here. A fault
146:30 . Well, hold it. This just stops. This fault stops,
146:33 fault stops, this fault stops. does that happen? OK. Let's
146:40 most positive curvature on it. Oh put most negative coverage on it.
146:46 Let's look at all three of them . Now we see that this fault
146:53 stop and that fault doesn't stop. got that pattern, red, black
146:58 pattern that continues. So what I'm is I'm losing offset on those faults
147:05 they taper out to zero. Then I got all this stuff perpendicular
147:12 it. That's the um transfer F to get the strain from one block
147:22 the next. When we looked the black F we can see on
147:33 and we can see the offset on vertical data. OK? This is
147:38 clear. This one pretty clear, little less clear. This one's real
147:42 and the green are faults that we on the curvature anomalies. Well,
147:49 , up here it's definitely a And uh here I have the
147:55 That's definitely a fault here. I've a lecture again. That's a
148:01 Then down here. Well, queer offset here and then here it's just
148:05 flexure again. So the curvature is at more subtle deformation um before you
148:14 offset the ref reflectors back to Ciconte where I show you the strike slip
148:25 and the pop up blocks. I'm at the time slice through the
148:29 Here's I got an clinal features. are, some of them are pop
148:35 blocks. Some of them are any . Then I got some sync
148:40 I got lava flows. I got seal. Uh Ignacy, I've got
148:46 ses it's like a bunch of them hurt the data quality down deeper.
148:55 then uh let's go look, what's guy? Well, as I come
149:01 to, it looks, looks Andy . All right there. And is
149:08 guy a fault? There's a pop feature like I showed you earlier,
149:15 curvature in it. This guy is fault. I've got negative curvature on
149:20 side, positive curvature on that and can go down deeper. There's one
149:28 down here, positive curvature anomaly with pop up block. Here's my pop
149:36 block down here and deeper, still , coherence didn't help as much.
149:50 . Here is the, um, close to six o'clock. So to
149:59 the they call M Surface Anthony. you my structural geologist now?
150:12 So one of the field courses you're have to go on, you
150:15 I don't know much about over thrust . I think I need to go
150:20 a helicopter trip to the rocky Pretty good, huh? Yeah.
150:26 . So you get in helicopter and look at all the outcrops. So
150:30 the rocky mountains like Idaho eastern those rocks have moved 100 and 5200
150:41 from where they were and they're gonna along a de comal surface. An
150:48 surface is what we call them. on this picture, the de com
150:53 surface this is happens to be Here's my decoma surface or detachment
150:59 there might be salt in it, might be shales in it and then
151:03 rocks kind of go up above. I have repeated section, a double
151:11 section. OK. And on this , um here is down deep,
151:23 have any faults here. I do some of the faults and the folds
151:29 here. Here, I have a faulting here. I have some
151:32 right? So a coherence image I some of the false ways is so
151:39 don't see the day comal surface, I see the way it's coming up
151:44 this red horizon. OK. So see both ways and I see some
151:51 and then I have, here's where fault comes in. It's a little
151:57 here. My formation is thicker and I have fractures in this direction.
152:04 have fractures in that direction. When drill, if this is the still
152:10 stress regimes keeps the same, we want to drill and this direction,
152:19 a lot of fractures here, but under compression. We really want to
152:25 parallel to the fall blocks through the fall because those are gonna be in
152:33 tension. So you wanna go through number three then because we have
152:39 we can map the thickness of the where you have that doubling of the
152:46 , we're gonna have a thicker unit in other areas you're gonna have some
152:52 thin and so you can use all attributes to map no, back to
153:02 , this kind of mapping thirsty So how are, how are we
153:10 , what do you think you've taken geology? No, you've got like
153:17 . Well, he's not your guys's . You got four structural geology professors
153:22 , right? You haven't taken any geology. OK. Based on
153:30 OK. Yeah, that's more extensional heat flow and stuff like that.
153:36 mean, that's important too. But geology is kind of fun. All
153:41 . So back to Anthony, how I gonna push big part of the
153:48 mountains. 44 kilometer stick 200 How am I gonna do that?
153:54 mean, the same thing is trip Colombia and in Peru, et cetera
153:59 the four land basins. How do , how are you gonna push
154:03 If I push it, I'm gonna the rocks, right? And then
154:08 doing the pushing, you know, tectonics and you're gonna crush the
154:17 So, so you haven't done in of your geology classes? Have you
154:22 ? The beer can experiment? So when you go home, you're
154:30 to do two things. Why you're have to have two beers after this
154:39 one you're gonna drink right away. other one you're gonna put and get
154:43 real cold, right? You're gonna a nice table like this. You
154:49 get a little wet joke. It half a degree a degree or if
154:55 need to, you can even, can even lift it up a little
154:59 . You can put the beer can it. OK? It's empty and
155:04 room temperature. OK? Then you're take the other beer can out of
155:08 refrigerator and make the experiment work, ? You gotta drink it quick.
155:13 can is still cold. I put down upside down. Remember to put
155:18 upside down. Then when I start the table, the one that slides
155:25 and further gonna be the cold. why? Same brand of beer?
155:40 Saint Arnold. Same beer. WW. Why? So what's in
155:53 can condensation? What's gonna happen to cold can as you put it at
155:58 temperature? Mm. Warm up And then what happens to what's inside
156:06 can remember you drank the beer? error. What's gonna happen to the
156:12 ? Is it warms up? It's expand and that'll support the weight of
156:20 beer can and you'll have less friction to your first thing. Ok.
156:26 this is what you have in a called mall surface. You're gonna have
156:28 overpressure Z. So it's the poor that is really supporting rock and the
156:36 is just sliding downhill, maybe a a degree. It's not being pushed
156:40 all. It is sliding downhill like 200 kilometers and it doesn't get all
156:46 and stuff like that. So that's de com on service. And it's
156:50 poor increase in poor pressure that bears like 90% of the, the weight
156:57 that rock reduces the friction. So comes on. Ok, an hour
157:03 gonna do the experiment, right? long. Oh, you guys,
157:11 family is not here that you got here is grandma here. No,
157:19 is here, right? So, he's home, he never washes the
157:24 . They're probably sitting in the sink . But grandma's house, he's gonna
157:27 the dishes. Same thing. He's the table, kitchen table.
157:31 grandma, I'm gonna wash your Oh, for him, she brought
157:36 the nice crystal which came from her , from Germany back in 1860 Precious
157:43 heirloom. Thank you. Was rinse off cold water or actually you got
157:52 water. The air inside is got that hot glass. You put
157:57 on there. Table is wet. table with grandma's house is 100%
158:02 got quarter degree. So slope on , the air uh starts to expand
158:09 the glass is hot German crystal, ? 1868 right? Which band and
158:16 walk and walking right off the side you are no longer Grant's favorite.
158:23 remember that, I'm sure you've seen , right? You've seen glasses creep
158:30 the table after you wash them. , no, OK. You don't
158:34 , you don't wash your dishes, just leave them in the sink.
158:37 . Definitely knows what I'm talking That's the one that OK, here's
158:44 example from a French faulting then strikes faulting and it's from the UAE United
158:51 Emirates. Such a surface. And um this fellow uh Melville did,
158:57 computed uh coherence and curvature and he's a lot of things in this.
159:02 is one of the big producing uh you know, like 30 kilometers
159:07 and he's got sand bars in it he's got rainfall. So here's the
159:12 trend and here are these redel shears I talked about the other day.
159:20 maximum curvature positive and negative values. then here is the wrench faults and
159:30 these happen to be sandbars. They got two different patterns you're seeing
159:36 the data. Let's look at it vertical. Here's my wrench falls with
159:42 redo shears in it on coherence, coherence image OK. And appear if
159:53 let me try to expand it bacon . Hey, I got it this
160:09 . A lot of these things have that is geologist, interpreters put on
160:18 on it. That's not what I . Gotta go, I gotta go
160:27 here. Get those. OK. there it is. OK. Uh
160:50 see. Can do it. I do that whoever. OK. But
161:04 area here, oh, now I'm it, this area where my,
161:13 mouse, I see those vertical see how they're like pinched up.
161:19 we're doing this, we're moving a like this and we're squeezing it up
161:26 we give them a name. We them TP like native Americans, you
161:33 , Sioux Indians, Apache. apaches didn't live in but the Sioux
161:38 within the TPS. So they call TP structures. And these kind of
161:43 are common to French faults. But of the ways you would recognize uh
161:50 faults. So they're squeezing, I'm skip this one. I'll skip this
161:59 . All right. One thing you do with attributes, they are now
162:08 and frequency, they're gonna be sensitive the amplitude and frequency of the data
162:12 acquired. So you're working for a . And so here you are
162:17 at um Jevon and they bought, they bought Texaco and they bought who
162:26 they buy recently and bought a bunch companies. So they have all this
162:31 they, that they bought and now want to piece them together with all
162:36 eight groups. And you go, , I could reprocess them all,
162:41 them all if I had the original gathers and geometries and stuff. But
162:49 what you tend to lose these over years, especially, you know,
162:54 may decide the heck with this I'm gonna make more money at
162:57 Exxon, I'm not going to Chevron whatever. And then all of a
163:03 he knew where the data was because in the it part Deron buys has
163:09 the it people say to heck with , we're gonna work for Amazon or
163:14 they get washed. Ok. So can you integrate this? So this
163:19 like a 19 on the left. should have to point to it.
163:24 Jessica and see what I'm talking This is a 1995 survey from
163:31 This is a uh survey from uh Energy in 1999. This one was
163:39 BP 2006 and this one was like from Occidental and they were all shot
163:47 different times using different equipment, different think of how variants and coherence and
163:54 correlation work. I take adjacent trass the same survey, I cross correlate
163:59 . I I cross correlate in the , numerator. And then I normalize
164:04 the denominator not terribly sensitive to amplitude and frequency you're taking that out.
164:13 then I can map, I can this strike slip fault across different
164:21 I can map this reverse fall across surveys and it allows you to make
164:28 mosaic and interpret them. You still to move things up and down a
164:32 , but you don't have to reprocess all the way. Ok. This
164:41 be, you know what, it's to 6 and we'll come back and
164:48 about shale and salt diet. Appears morning at 830 with coffee. Is
164:58 ok? Tomorrow we got a football at six o'clock. Where are you
165:08 gonna
-
+