VideoPoints

• ••• • •• • ••••• • •
GEOL 6390 3D Seismic Exploration - Day4 11-04-23 AM
Help Tour
© Distribution of this video is restricted by its owner
Transcript ×
Auto highlight
Font-size
00:00 Hm. Good. Morgan. a man and a fag. He
00:13 not be there. OK. All right, good. Um So
00:23 a couple of tricks to, of horizons if you've done surveying. And
00:33 well, I did my master's degree mining geophysics. So we had a
00:41 and of course, surveying underground in mine is kind of kind of tricky
00:46 it's a three dimensional survey. But you survey, you always wanna tie
00:50 loop, that's the trick. You always wanna tie the loop and
00:55 gonna make small loops, fix them then build out. So you're gonna
01:01 the same thing, picking horizons. . Go pick that first horizon.
01:06 Here we got a big, uh know, uh seamount in the
01:15 We got a big seamount of volcano the middle of the survey. You
01:19 pick right across the top. Because those horizons no longer exist if
01:26 volcano occurred later or if the volcano before and it was a seamount,
01:33 wasn't a combination space for that horizon those layers were laid down. So
01:40 can't go across the top of OK. So there's gonna be a
01:43 in the middle of your uh horizon . Then the um So you're going
01:51 be best picking away from the OK? If you can, you
02:01 try to make a big square around whole edge. But that's, that's
02:05 of tricky. OK? But that at least goes around the volcano
02:09 avoids it. What you don't want do is pick all the end lines
02:14 , then all the cross lines. , you don't wanna pick all the
02:17 lines first and all the end lines they will not tie. It's gonna
02:22 like you're gonna have cycle skips because are moving up and down. And
02:28 way a horizon looks on the in direction is gonna look different in the
02:33 line direction. OK? Just because eye is really lining things up.
02:39 , when you look at hard do old people, they'll take their
02:49 and they'll stick it on the table look at that horizon. OK?
02:55 put their head down sideways on the . And what that does? It
03:01 the data by one over cosine of where the is the angle between my
03:06 and the and the table and makes more continuous. So you can see
03:13 that then looking at them straight So um so data are looking,
03:20 going to look different when you look them in different directions. So the
03:24 is uh make little squares, 100 100. Now another thing is horizons
03:36 be easiest to pick perpendicular to the . OK. So you when you're
03:46 , default. Yeah, I have correlate across that fault. But the
03:50 are gonna look kind of sort of along the fault. If I am
03:56 parallel to that book, it's real to miss Cory. You may not
04:03 the fault if you're going sub parallel it and you're just gonna go right
04:08 . And that's the main reason your lines and cross lines don't tie
04:12 So it's a good practice to pick arbitrary line perpendicular to the fault.
04:17 , in this, this example, lines you pick to fall with were
04:22 kind of, I can't remember if were in line or cross the
04:24 They were, the fault was kind perpendicular to one of the major
04:29 Um So it's OK to pick a crazy shape, line perpendicular fault to
04:38 things to correlate. Then that gives the seed points to pick on a
04:46 . Now, the last thing is value of picking on a grid.
04:51 , the grid can be arbitrary line . OK? So you pick one
04:56 line and then you go 2020 2020 that's cool, you can do
05:02 What you wanna do is want to able to fix your picks. So
05:09 say I pick in line 1000 and 22,030. Fine. Then I'm gonna
05:19 in line 1020 7-Eleven 57. Uh . If I wanna fix those picks
05:28 they're not tying. I gotta find line they're on and that's really
05:33 So you spend a lot of time that. So once you get
05:37 try to pick on every 20th it doesn't matter whether they end in
05:41 or ones or threes, doesn't Just kind of know. Oh,
05:45 need to go through every 10th every 20th line on a grid and
05:49 you can fix things. Ok. there we are. I'm glad I'm
05:58 you're making progress and what we start this morning while Utah is making coffee
06:14 we are res we're gonna go to you make your attributes that math uh
06:29 configuration? OK. That's showing up there and, and make sure I
06:37 the screen shared. Utah is Where is my Zoom? This might
06:46 the Zoom. That's chad. Why I run a store? Who is
07:03 ? Wow. Ok. Where is ? Am I in zoom here?
07:13 not. Ok. So I got . Ok. That sounds good.
07:28 , I suppose not. Oh, it. Ah OK. I am
07:44 sharing XO A and I can look the camera. No, that's where's
07:49 camera cameras over there? Ok. . Dang. And there was a
08:00 . Hey. OK, good. let's see, it's still muted.
08:09 don't think so. OK. So want to talk about geometric attributes that
08:22 uh map deflector configuration. So this lecture eight and this is gonna be
08:28 like dip aas you uh curvature, of that nature. Uh can we
08:35 pinch outs? OK. So measuring . So here's, here's the ones
08:42 gonna talk about uh in this dip magnitude dip as curvature and the
08:48 virgins. And what? So dip and dip as we you're gonna
08:56 actually, you just always wanna calculate wrong a horizon. OK. Along
09:06 field yesterday, when you did the , you want to filter a long
09:16 not across. And what's amusing is to make it run faster. Their
09:23 is not to filter across structure. uh one or two of you yesterday
09:29 , well, why does this look ? You have all these like shadowy
09:33 on there? And that was because you use the default curvature is gonna
09:40 folds and flexures. It'll map differential across channels, it'll map uh cars
09:49 , it'll map carbonate, build it maps shapes, convergence is going
09:54 measure angular un conformity and coif forms rotation can map Wrench fault syn tectonic
10:02 , things of that nature. So what we wanna do is evaluate
10:07 algorithms to calculate volumetric dip and ASM terms of accuracy and lateral resolution.
10:15 uh a later lab, you'll probably it next week. Uh pare has
10:20 ways of computing. Dip, two which are good and two of
10:26 which are total trash. Now, what? Which is the default 11
10:31 the total trash ones because that's the they developed first, like 20 years
10:38 . And I don't know, they , actually, they should get rid
10:43 the bad ones. But sometimes computer vendors need to keep something they know
10:55 , they know is bad, let's to calculate dip because uh interpreters can
11:01 really creative and, and Bob over it has, he may say,
11:07 , I've got a workflow where I the gradient dip calculation to map the
11:15 of digenetic alteration in my reservoir, blah blah blah. So sometimes people
11:22 things in ways that the person who the algorithm never envisioned. So it's
11:29 difficult to take things away from a but they shouldn't make it a
11:36 That's stupid. OK? Because then gonna say, oh this doesn't work
11:40 you're not gonna use difficult. You'll that when you start running it,
11:43 will, you'll look at it and what the heck is this?
11:47 Then uh shaded relief mapping is something apply to surfaces. But we
11:54 when we have a volumetric dip we can approximate that same process by
12:01 at a parent dip in different OK? And then we can generate
12:07 di Asmus seismic image to determine how given reflector dips in and out of
12:11 plane. So when you're initially looking the data, you want to get
12:17 , get a feel for what's going just to animate through these things co
12:22 and right away, you got a idea. OK. So there's five
12:27 ways of computing uh biometric dip. First one, let's see if I
12:34 words. No. OK. So got complex trace analysis. That's one
12:38 in landmark a lot gradient structure That's one that's in, in
12:44 It's probably the most common one. the in portrayal, they'll call it
12:49 principal component calculation of this. But using something called a gradient structure tensor
12:56 wave destructor is more of a researchy . I haven't seen it in commercial
13:03 . The street scans for the dip the most coherent reflector, that's the
13:07 one to understand. And it is in a in landmark software and one
13:17 I developed a long time ago and correlation in four directions if there's a
13:23 out there called open detect by uh Dutch company called D GB Go
13:30 And they'll look at the apparent dip the north direction east, south west
13:36 then try to patch that together to a point dip, a dipping point
13:43 patrol. The, the dip calculation most proud of is they call it
13:49 dip. So they do the same in every box. So they estimate
13:52 dip in the four directions and then try to tie loops, OK to
13:57 sure that that dip estimation gets me to where my starting point was.
14:02 if I follow the dip go down around, I come back up,
14:05 they write in a little optimization algorithm says I want to have dips that
14:11 consistent with being able to tie a . OK. So you've got uh
14:16 of these and in uh portrayal and you got two pieces of garbage that
14:22 don't even put there because I don't they measure dip. OK. So
14:28 we measure a dipping plane, there's guy. OK. Um As a
14:35 , you would probably define the strike intersection with the horizontal plane, let's
14:42 sea level of the dipping plain and dip magnitude the angle from the
14:52 Then that looks kind of cool on map and you'll put little kind of
14:59 things on the side of the So people know which direction is
15:03 Um But to store that in the , how do you store the rectangle
15:09 unclear. So the EIC people or will tend to use being skip a
15:20 the avenue tells you which direction is . Yeah. OK. Good
15:28 We got a breathing time. It's not oriented north, south,
15:35 angle with the player that well, components of it I two compound combine
15:47 to get the magnitude and the F . OK. So that,
15:54 that's pretty straightforward. And so all those are equivalent and can be converted
16:02 to the other using simple trigonometry. . So here's the first one uh
16:09 go through and it's a complex trace . So last week we talked about
16:15 Hilbert transform and the um Hilbert transform the data that's been rotated by 90
16:23 . I think it all watch. if I take the angle between the
16:32 transform and the original data, I the instantaneous face. And then we
16:40 the, if I take the derivative the phase with time, how many
16:46 per second? OK. That would or how many cycles per second.
16:53 there's two high radiance per cycle that's have the units of frequency. And
16:58 gonna call that the instantaneous frequency. you can't take a derivative when you
17:03 from 180 to minus 180. So I asked uh um uh Jack about
17:13 uh how do you calculate the derivative the angle? And so Anthony remembers
17:19 do you calculate the derivative of theta a phase? Remember it's got a
17:29 of minus 180 to plus 180. it's gonna blow up at that
17:36 I'm picking on you because you're as to, to Zach as we
17:40 So you are a Zach still Yeah, you gotta, you gotta
17:45 the um the art tangent of the in meters per meter of the
17:54 you take the arc tangent of the , then you go, that gives
17:58 theta or, or Phi, The uh what do I have
18:02 Phi? Okha, same thing you do for uh or as you take
18:10 art tangent, then you go back your calculus book or Google, what's
18:16 derivative of an art tangent? And gonna give you these things that have
18:21 um this is the denominator. Those funny terms. OK.
18:30 if I can take a derivative in vertically, why not take the dude
18:38 ? And that will give me cycles meter. All right. And then
18:47 five per meter will be the wave and the wave number is two pi
18:53 the wavelength. So that's the instantaneous number in the in line direction.
18:58 do it with Y that Stans instantaneous number and the cross line direction.
19:07 if I take the ratio cycles per and radiance per meter, radiance per
19:21 , then I end up with seconds meter. I get a apparent
19:27 So P is gonna be the apparent line dip two will be the apparent
19:32 dip. OK. Measured in uh per meter seconds per kilometer seconds per
19:40 foot. What do you want Now that P couple the thinking of
19:47 um If it's in, if the are in depth, then we're gonna
19:54 about meter. So we're gonna, gonna be like a roofer.
19:58 So the guys on the roof, think I mentioned this last, last
20:02 , you know, they're gonna have little triangle to set the pitch of
20:05 roof and they're gonna say if I 1 m across, I maybe want
20:09 drop a half a meter. They're don't give these guys angles and
20:14 like that. Just give them a that measures it. So how many
20:18 down do I go for every meter ? If it's time data, how
20:23 seconds down do I go for every across? So that's what the P
20:28 . Some of you are processors. is my processing there? You are
20:37 ever hear of cow pea processing? . It's the same pea some of
20:43 are earthquake people. Yes, you earthquake gal Stephanie earthquake gal,
20:52 Who you are? OK. So earthquakes, we talk about the emergence
21:02 . So an earthquake is coming from and it comes up to the surface
21:06 the earth. And you're saying what's that emergence angle? And they're
21:10 use P for that as well? . Uh oh There's a P right
21:18 down at the bottom. Yeah. . So those are the apparent
21:27 Now we can calculate the dip I got a point here. I'm
21:34 , we can calculate the dip magnitude that's gonna be the sum of the
21:40 and it'll be in again in seconds meter, some of the square square
21:44 and the diaz there's gonna be the tangent two between the cross line and
21:50 N line. OK? And that give you the uh the um
21:56 the Asmus from the in line So then you got the survey is
22:02 , you gotta fix that. Remember we talked about the analytic trace
22:07 week and we had these wavelet interferences the minima of the envelope and they
22:20 us he's really strong uh frequency They had phase changes there. So
22:33 a question yesterday that maybe Hayden brought like no, it wasn't paid
22:46 One if you brought up, my distribution ah with reed to distribute
22:57 histogram of my instantaneous frequency that she looks different than what I have in
23:04 notes. And last night with all traffic, I'm stay, I was
23:10 downtown noisy. And last night I , oh I said let's not crop
23:19 data to minus 25,000, 25,000, it eight bit, let's do 32
23:24 . So with that 32 bit there were these really strong amplitudes or
23:33 in the data than what they And those made the instantaneous frequency give
23:41 of like 250,000 Hertz. OK. ridiculous numbers. And then when she
23:51 to generate a histogram of, I the limit on the histogram when you
23:56 the data, it has a limit 400 boxes of cells. OK.
24:07 all of her data fell in one but she saw nothing. OK.
24:12 that's, that's what happened to OK. So we got, we
24:16 these spikes in the instantaneous frequency. the way we can fix it is
24:23 waiting the instantaneous frequency by the envelope biases to the strong parts and weakens
24:33 uh the weight of the weaker OK? And that gives us green
24:39 . So that's what we did. in 3D, here's a happen to
24:44 depth migrated data. I've got a slice through the amplitude data, a
24:50 slice through the amplitude data. And can see, yeah, there's a
24:54 dome down in here and here are reflectors on the plank. Here's the
25:00 dip magnitude first looks OK? Yellow high dip, green and blue are
25:11 dip, but look at all this and pepper up in here. This
25:17 be pretty smooth and flat. I've a lot of yellow in there.
25:22 morning. We used your name in and instead of picking on you,
25:26 picked on Anton and, and ask the same question I did you last
25:32 and he didn't know the answer Just make you feel good.
25:36 So you're well represented. OK. we got all these spikes in
25:43 All right. Now these these let's go in and smooth it.
25:48 we're gonna smooth the in line wave gonna smooth the temporal frequency, gonna
25:55 the crossline wave number. That's what bar is. OK. And we
25:59 come up then with a smooth So you look at this picture ah
26:04 of blues maybe some green so moderate , steep dip, see the difference
26:12 stabilize. We're using the same trick . And there's a depth slice and
26:19 is what those weights look like. is for the weighted uh temporal
26:24 And you would do the same thing the two wave numbers. OK.
26:28 bas mute, same thing, really noisy going in all directions.
26:38 ? And then smooth, fine. when I talked to Art Barnes,
26:44 asked him how he did and he five in lines, five cross
26:46 seven samples. OK. Now let's about the gradient structure temperature. And
26:56 say I've got a dipping reflector and got peaks and drops in my
27:09 Nothing has picked up. Who tempting pick on Zach? Zach. I'm
27:13 gonna pick on you. OK? have a pencil. No, too
27:33 . All right. All right. . They're going to hold that.
27:40 here is my reflector. I'm going measure the change in the data.
27:46 that picture there, you see where have DUDX, that's the change in
27:51 amplitude in the X direction, the to change in amplitude in the Y
27:57 , the UDZ change in amplitude in vertical direction. OK. Then I'm
28:03 cross correlate those changes with itself and other. So I have three different
28:11 . So I got a change in axis. We're gonna cross quarterly.
28:15 one. I have a three by cross correlation matrix on the diagonal happened
28:21 be an auto correlation. OK. we're gonna call that the gradient structure
28:29 . Now got that three by three . Which direction here, here's my
28:36 reflector which direction shows the most change amplitude vertical, hold it up
28:46 So everybody else can see. So she claims vertical. Now she's
28:54 often into off into some other space OK, perpendicular to the reflector,
29:03 . So what you're saying come on it confidently. Yes. She says
29:08 . OK. That three by three is similar to covariance matrix. They
29:14 it a uh it's a correlation OK. And the first eigenvector best
29:26 in this case, the change in data, OK? Why the change
29:31 the data? Because we're measuring changes data. That's why OK. Best
29:36 the change in the data. Last , we talked a little bit about
29:41 . This eigenvector best represented the change dip. OK. So all I
29:50 is I formed this three by three and then I compute mhm eigenvectors and
30:01 . Now, well, here I'll on Zach because Zach square root of
30:10 , how do you do that? you now? OK. I'm a
30:21 that's a square of five. But would you calculate square root of
30:26 Where? Yeah, how would you it? That's why I'm asking
30:32 Ok. He's gonna pick out his . He's gonna pick out his little
30:37 app he's gonna put on right. the square, you know the square
30:43 of five is between what and what's the square root of four?
30:54 were the report square to nine? . So you know the somewhere between
31:04 and three now, is there a man inside there? Oh,
31:09 I'm sorry, that's sexist. Is a little woman inside of there?
31:15 do all the work. No, got a little algorithm in there.
31:19 it happens to calculate. You remember first year calculus, remember Taylor series
31:27 series are your friend? I'm sure teacher said that Taylor series are
31:31 remember Taylor series expansion. So you the square root expanded in the Taylor
31:37 , use the first two terms and Newton raf and method iteratively to get
31:44 and closer to the square root of or seven or whatever. OK.
31:51 need to know that. No, , you just need to know which
31:54 to push. OK. The same is true for eigenvalues and eigenvectors problem
32:01 solved in 1960 you can find it over, you know, Google wherever
32:08 the computer MATLAB has it. It's the libraries for FORTRAN and C++ and
32:15 and everything. Like that. You need to know that it uses a
32:20 really quotient method and you go on loop and iterate et cetera, et
32:25 , you just need to know physically it means. So the first eigenvector
32:29 represents the variability of the data. what it does. OK. And
32:35 here, the most variability of the is a direction. It's a
32:39 it is the normal to the How proud are we of that
32:46 That's the eigenvalue. So it says much energy in that window of
32:53 let's say five by five by five . How much of that energy is
32:58 change is represented by that normal. it's 100% we got a perfect
33:06 If it's 33.3% there's three eigenvalues and , uh it's random data.
33:15 And if it's in between, it's kind of sort of pointer.
33:18 that's what we have. So one the attributes you'll play with next week
33:23 called chaos that uses the eigenvalues. the dip is the eigenvectors. I
33:30 this from somebody us from here. . Oh Just to show you this
33:40 done uh by the TJ guy back 24th, they're different. Hey,
33:53 we want to talk about it's easy understand, but computationally more expensive,
34:00 we're going to do is take a bunch of data, a window of
34:04 . In this example, I got two D image because I didn't,
34:08 can draw too deep and I'm gonna look at 20 degrees uh 1510 uh
34:18 minus five minus 10 minus 15 minus . And I'm gonna calculate some measure
34:24 similarity. OK. What do we in velocity analysis? We use a
34:29 called semblance. So semblance is basically take the data, I average
34:37 I square that average I take the of each sample. OK? So
34:46 it, I take its average pick ratio of the two that happens to
34:50 semblance. OK? So I find well, which one's most alike,
34:55 similar, that's the winner. And I look at adjacent ones. I
34:59 a little interpolation and I compute the dip just like we would calculate
35:06 velocities, you know, you, get these little, they call them
35:10 ups where the semblance is highest and , I picked that guy. All
35:14 . But here we're gonna do it two dimensions. Gotta do in line
35:19 cross line. So at the um , they'll search in the in line
35:28 , then they'll search in the crossline and they'll put the two together the
35:32 I like to do it. I to search really in three dimensions.
35:35 instead of looking, let's say 11 in the in line and 11 searches
35:41 , I'm gonna do 100 and 21 because then you don't suffer from
35:46 OK? So the alias. If data are alias, you get the
35:49 gifts. OK? Now, another and this isn't in, I don't
35:57 it's in any of the commercial Um It's not computationally more expensive.
36:04 just takes a little skill in data and programming. And now uh I'll
36:12 my uh mouse to point here. got a fault here. OK?
36:22 if I wanna calculate the dip at location, the red dot this is
36:28 most coherent reflection. So I'm going get a dip that says it's uh
36:34 dipping to the right, almost And that is an apparent dip.
36:39 not the true dip or the It's some smeared dip. If on
36:45 other hand, I use three traces calculate the dip. Well, the
36:51 to the left is pretty good and one to the right is pretty
36:55 either of those are better than the that's centered. So that means what's
37:02 overlapping windows that um provide a means computing dip, we're gonna pick the
37:13 one. OK. So it's gonna called a Kuwahara window. And in
37:19 , here's a centered window looking down above. I got one my target
37:25 and nine uh or a total of traces. So eight surrounding traces,
37:33 gonna calculate its depth and its However, you like to consider
37:40 let's say you but this green point is also part of this amount,
37:50 one Now this one now like nine is that point. No,
37:56 no. So I'm gonna find out one is the best. Um I
38:06 going to break out just a second go back to the lecture that I
38:23 with the idea that you guys are look at it and this picture here
38:41 OK. So this is in uh seven and I voiced over so you
38:45 hear my melodious voice if you miss on Wednesday. OK. Um And
38:52 is a picture from the Luau. so Al Dori so did his phd
38:57 at U A. He's the part a guy called A L I
39:04 So here is their signal and so they're going from 0 to 1
39:11 then here's signal with 1 to 1 . Now, if I just do
39:17 smoothing filter, so here's my filter I'm gonna run that filter across.
39:25 you. There's my filter if I my filter wrong here. Oh,
39:31 got a nice result here and I a nice result there, but I
39:37 my edge. So think of it a channel edge and I'm going from
39:43 channel which maybe has a high amplitude across the cut bank and maybe I
39:49 a lower amplitude on the side of cut bank. And now I'm smearing
39:55 out. OK. So, and the filter but fine and so do
40:15 the others and then this little he has 21 different samples and figure
40:23 what the, the mean and the deviation within that black zone of 21
40:31 of this, the zone of the 21 windows that overlap the output
40:38 which one has the lowest standard Well, from this picture, you
40:44 see it's gonna be either the one the right or the, the anomaly
40:49 the one to the left. If straddles it, it's gonna have a
40:52 standard deviation, that's the winner. . That I'm gonna take the mean
41:00 that winning window and apply it to outlook point. So this technique was
41:10 in medical imaging by a Japanese scientist Kuwahara. So it's called a Kuwahara
41:16 edge preserving filtering. Uh We don't that in uh patrol and um
41:32 I didn't know I had all that there. OK. Then um here's
41:38 app application Saudi Arabia is not ugly . Their structures are pretty subtle,
41:48 carbonate structures and their surface has two . One, they have sand dunes
41:55 the sand dunes up to be, know, 50 m high,
41:59 their velocity changes during the day or kind of hard to correct for
42:05 And then underneath the sand dunes, have these um teachers they call
42:11 So Shaka is an Arabic word and kind of like tidal flats, but
42:18 a desert country in a desert So you have a lot of an
42:23 , carbonate shale and hydrate carbonate And what happens when it does
42:30 the an hydrate dissolves. So you'll high velocity an hydrate here and then
42:35 have a hole, no, an and then you have an hydrate
42:39 So their statics are really ugly. here is uh amplitude time slice and
42:46 they have an edge texture. Uh filter type there, we'll talk about
42:55 like variants like you've been using. . And that looks OK. That's
42:59 to the original data kind of Let's do structure oriented filtering with edge
43:10 and they get this image. So here's without filtering and then here it
43:21 that is the amplitude data filter and it. So, um Anthony,
43:30 do you see there? What's your there? Pardon da here.
43:44 OK. See you channel. He sees a channel. All
43:53 Something like that, right? Be more proud of your interpretation.
43:57 it out. I hear this stuff . OK. Did I tell you
44:03 get old hearing aids? You hear loudest thing in the room and that's
44:09 . OK. All right. I've got a different working hypothesis.
44:17 were talking earlier with uh Rob you know, mixing vol volcanic,
44:23 up with uh uh carbonate build They kind of look alike.
44:28 So there's different, all kind of hypothesis. So I, I know
44:32 channel, I see something different client I see Elvis in the data.
44:43 don't know who Elvis is either. no, the king living in the
44:52 that to you. OK? It's , look seven there. OK?
45:01 that's edge preserving, smoothing or edge structure oriented filtering. Now back to
45:10 and au we're gonna use the same window concept except now instead of taking
45:18 average, we're gonna find out which is best. OK? Gonna use
45:24 to measure which window is best semblance variance if you wish which window is
45:30 . And then I'm going to take dip that winning window as the
45:37 OK. And apply that to the , do the same thing vertically.
45:42 this example from West Texas here, biased vertically. Here's my analysis point
45:48 how everything is parallel. Ah That's have nice semblance here. I've got
45:54 angular un conformity. This one's gonna low assembling. OK. So this
46:01 be the winner and I'm gonna take step here and put it to this
46:05 . And what that does? It me a nice sharp angular and
46:08 discontinuity. All right. So here amplitude instantaneous in line dip, smooth
46:20 in line depth using the technique. then here's a multi window scan using
46:25 Kuwahara filter, right? Let's look the time slice amplitude. Um
46:37 So a bunch of you live right? 70 you were poor,
46:45 ? OK. So you probably go Vend Dome and try to improve,
46:54 go. Do you take that bus goes to vent? Louisiana? Is
46:57 still running? It's running? you take the bus to vent
47:03 and you start gambling. That's the gambling place to here. Ok.
47:08 . 0, bus is $10. might lose 100. But, so
47:13 is Vinton Dome underneath the cap. . And this project, uh,
47:18 and I worked on over 15 years . Um There's a a wake in
47:23 middle of it above it. And so here is the amplitude slice.
47:28 see your onion ring kind of Here is the instantaneous in line
47:35 So to the right is positive to right, that's good. But you
47:40 if it's white, but you see the black food, these are all
47:43 artifacts in there. If we average using the technique, art Barnes talked
47:49 , hey, this is pretty This is geologically reasonable. This is
47:55 . And then here is this multi dip scan I talked about and comparable
48:02 this except higher resolution and more You can actually see the edges for
48:10 et cetera. OK. So here's vertical data set, North Texas.
48:18 uh cattle limestone is the strongest easiest pick horizon in the volume. The
48:25 is painful to pick but also pretty . And then we've got cars collapsed
48:31 here, cars to cars to your steer. So think of a a
48:37 collapse at uh the home has that do with it. Or you throw
48:43 in there, then it catches on and uh kind of a big trash
48:49 , right? Ok. So here's kind of horizon and here's the magnitude
49:03 depth horizon. Thank you. And showed you this picture yesterday uh 1995
49:10 , 9799. And this was a ad for this survey. So this
49:16 thrown the pick here is vol Yeah. So it's nothing. The
49:27 is kept picked biometrically. I took horizon and sliced through that biometric
49:33 OK. So I'm extracting it if wish and notice that. Oh,
49:39 sorry, this is still picked from horizon. Correct? Maggie is in
49:54 ? 669? Now I'm gonna go and do you make are here?
50:07 very good here, pretty good What do I see? I,
50:09 happen to have a strike slip fault . I happen to have we
50:16 Cheers. Who's my structural geologist This is my structural G.
50:25 you're my environmental gal, right? . Yeah. Yeah. OK.
50:35 , you my structural geologist. you are. Oh Newton, you
50:45 geologist. OK. So read o as I, I'll show you some
50:53 of this next week. OK. I talk about structure and so if
50:57 do strike slip in a call it rigid basement, think of the basement
51:06 very rigid and then above it. have more plastic sediments. I start
51:11 simple strikes with pain. Then two happen. One, it starts to
51:17 and they'll call it a helio coal geology. You have to use big
51:22 , right? Helio coal deformation. you get these redo shears coming
51:28 Oh, and then you'll get conjugate shears clear. You guys have no
51:38 what I'm talking about. So I , well, I guess that's gonna
51:52 me someplace. This 1 may Uh I got images. OK.
52:22 That's not very much fine. And a cartoon of it. Let's see
52:41 we got it done. I know seen pictures of sidewalk with reel
52:50 Oh, here's wet clay. So slip and here's the sheer, here's
52:55 main strikes swp, then they got on perpendicular to it and I will
53:01 one when we get a break. . All right. Why do I
53:12 that the way you do interpretation, look at these little things, architectural
53:23 . Deep voice. OK. And I see a main fault and I
53:27 Redo shares with it, I ah, strike slip regime.
53:33 you know, that's, that's how gonna form. OK. I have
53:37 pop up block. You do pop blocks. You didn't do structure.
53:43 . So you got a pop up , right? Sweat Rambo Castle.
53:55 , it's not XRD. Ok. casts, you know what a rhomb
54:01 is. Strike slip. A Falls down Rambo chasm in the United
54:09 , Salton Sea in southern California. you know where Salton Sea is dropped
54:14 . Ok. See you south of A, Los Angeles. Uh,
54:20 Dead Sea in the whole way than . Rambo casts. Ok. Lowest
54:28 in that part of the world. , strike slip. Uh, the
54:32 River is along a stretch football both the Gulf of a, up to
54:38 sea of Galilee. Another Rambo OK. I'll try to get some
54:45 like that. All right. So is the north south dip and here
54:49 is volumetrically from the good horizon extract . Or why is that statistics?
54:59 I computed volumetric gift, I used by three traces and 11 sample.
55:05 I computed horizon dip, I use by three traces, one sample the
55:12 of my pick horizon. If I cross cutting noise, that cross cutting
55:18 is gonna move my pick up and . Like I said the other
55:20 remember we wanna pick zero crossings. we can, it's gonna move it
55:24 and down. If I'm computing the in a window, vertical window of
55:32 that cross cutting noise cancels out. the East West apparent dip again,
55:40 in the yellow square. OK. the red square. Not very
55:44 I mean it's, it's fine outside . And then here it is biometrically
55:52 apparent dip. And you might see got dimples in here. Let's go
56:00 the apparent dip. So now I'm to look the north-south apparent dip and
56:08 I'm gonna just calculate the apparent dip to be so deeply 1591 21
56:22 And then I go back to 180 is the same as zero. So
56:28 looking at different avenues, I can speakers that are perpendicular to my apparent
56:37 like shaded relief. Now I'm gonna through the uh Ellenberg Don looks a
56:48 bit like his cheese. This is I have the collapse features.
56:55 And then if I go the apparent , you'll see a shaded looks like
57:00 relief like that. You can see those collapsed hatred. The uh left
57:08 side of this image is bad or fracturing because the horizontal well will complete
57:20 the harsh collapsed features which are connected the aquifer down deeper and your then
57:26 horizontal well just produces walk. So you wanna map these guys before you
57:32 your horizontal, the northeast part is ? But the, the west and
57:37 south part are nonproductive but they just water. OK? Uh We did
57:44 as a lab exercise. We got AU. Here is the uh a
57:50 Sliced Dish Avenue with the interpolation turned . So I see plenty of
57:58 OK? And here's the dip magnitude then I'm gonna co render them using
58:03 monochrome gray color bar. You'll do again next week. And then uh
58:09 can put co OK. Now here's parent tip at arbitrary angle. Same
58:18 I just sent you. So di is estimated using the vertical window.
58:25 it's in general provides more robust estimates those based on picked horizons. The
58:31 volumes form the bases of curvature which talk about next coherence, amplified
58:38 seismic textures structure or F train, you did yesterday or last week in
58:43 lab. Hip anatomy are key components completely to the uh think of uh
58:55 a software package out there in paleo where we're generating geo chrono stratigraphic horizons
59:05 to pick them and where they pinch on lap off lap with map those
59:15 . So it's a routinely used to build uh tilted transverse and isotropic.
59:23 , I know how a Joe talked anisotropy and migration, right? So
59:30 what we measure with the seismic method the surface, we measure the horizontal
59:40 of velocity. And did he go that attack? Why we're just measuring
59:47 ? OK. He probably used the slowness, right? Yeah.
59:52 So I can think of the slowness one over velocity and you can think
59:58 it as a vector. So I a horizontal slowness and the vertical
60:02 So here's my source, here's my , I go down and up.
60:08 I have two times the vertical slowness the Yeah. So S sub Z
60:16 Z and then I have my offset , let's say if I'm looking at
60:22 midpoint, so two X, two times s of X as I move
60:28 out, I changed the X and X so I can get the coefficient
60:34 S of X which is S sub , the horizontal component of slowness,
60:40 have no leverage over the work. . So I can measure one component
60:45 velocity. I have a vertical. , I can do a couple of
60:52 . I might have a sonic clog gonna measure the vertical component of
60:57 If it's a vertical well, or be at the wrong frequency. That's
61:02 issue. Or I can have time depth conversions by tying a synthetic to
61:12 seismic data. So then I have time pairs and that gives me velocity
61:19 there in a nice smooth way. consistent with my seismic. All
61:24 So now I hit go down, I go through a horizontal shale.
61:30 know that vertical velocity from my, , I know my horizontal velocity from
61:38 seismic velocity analysis. I can measure anisotropy. Then it's a shale.
61:44 know. Yeah, shale is laid flat, flat, boring,
61:48 laid down flat. Then later it structurally deformed. And now the shale
61:52 against the salt plan is maybe 6080 depth. OK? I'm willing to
62:00 that the anisotropy hasn't changed, but I have is a rotation. So
62:06 , oh I can measure the I'll use one of these four or
62:09 techniques. I got the dip. know what the anisotropy is here.
62:13 just gonna rotate that dip matrix and remigrate it. So that's what people
62:17 they call it Hilton transverse isotropy. Thompson will talk a lot about
62:24 OK. And what do we have ? Let's have coffee. Oh,
62:33 third floor. Oh, ok. . OK. I'm gonna turn my
62:40 off. I'm gonna Oh, yes, sir. Newton. Newton's
62:45 a question. No question. All right. Let me turn my
62:57 off. They um you talk. right. So we're still in the
98:10 lecture number eight you want to talk ? So we're gonna be oh
98:24 A lot. So I could get for that. OK. I think
98:43 something you call escaping that might have about that. Wait and COVID which
98:52 the key to rooted and the shape they, you know GEO bodies that
99:00 all collapse a build up other kinds shapes and then there's something a behr
99:06 , but there's no ay in the petrol software to define axions of wes
99:12 can be aligned with faults and exhibit that fall below the limits of seismic
99:20 . So we got different measures of . We can take 3d derivatives of
99:28 vector dip estimates. So we can the dip. OK. You actually
99:43 wow, I'm gonna take a frequency . I can take a derivative by
99:48 into the from time domain to temporal multiplying by I omega and then then
99:57 back. And that gives me the derivative I can do derivatives that as
100:02 . OK. Or I can do two D derivative of surfaces fit to
100:09 local hip. And the long wavelength are obtained by fitting a quadratic surface
100:15 nine more distant points. This is go and maybe Petrel does this.
100:25 then you can also fit the quadratic , uh two points on a larger
100:30 . OK. So I showed you picture last week. Again, I
100:38 on a I think, I mean morning instead of Zack Zack, I
100:44 him about the derivative of an Art . Thank. And uh that's why
100:49 formula has uh three halves in the . OK. So uh this is
100:56 slope in meters per meter and this the second derivative. So curvature is
101:02 a simple second derivative, but the derivative with a correction term that measures
101:08 rotation of the surface here uh you to they take their picture, they're
101:16 biometric identification. There's a lot of of doing it. One of the
101:21 ways of measuring the shape in your . OK. So if you look
101:25 my face, I got a bothered look at the most positive curvature
101:32 perpendicular to my nose, no other , they're more Antin most. If
101:42 look at my nose here. this is for um Bria and uh
101:52 , OK. You see it's still in this direction. So the most
101:56 curvature can have a positive value. just means there's no other direction,
102:01 negative. OK? Then for my , I also have S an
102:06 So I look at different directions. ? And the most negative curvature is
102:12 be this way. That's the most . And then if I search in
102:16 directions, ah most positive is this . It's still syal, but it's
102:23 the least inclined of all the OK? So the positive curvature can
102:28 a negative value as well. So have like a bowl then for my
102:32 , I have what we call uh jaw. That's like your, you're
102:38 a lantern in the dark. So positive curvature goes along my jaw,
102:45 ? And most negative uh kind of . And that means I have a
102:50 here. Now, regardless of how paced the security camera, however,
102:57 look at it, the shape of face is the same. It's really
103:01 to change. And here's picture when was here at uh 2006, here
103:05 1 2018. Keep my face is same now. OK. Put on
103:12 couple of pounds here, that's But this is the same.
103:16 So it's a good thing for We use it for molecular docking.
103:20 molecular docking in pharmaceuticals, you've seen that or Coronavirus guy? Ok.
103:30 I come up with a, a that goes in and locks into that
103:39 is spikes and keeps it from Ok. So you've got to define
103:45 shape in three dimensions. Why? it's in your blood going in all
103:52 sections, you know, as you're around. So we use it for
103:56 a few different things. No. , one of my, uh
104:02 uh how I uh he drew this of a saddle. I've got a
104:07 here to the saddle and I wanna the smallest circle. So there's this
104:15 to that point and that is the , that's the maximum curvature.
104:26 And then the biggest circle, So the specific radius, OK?
104:39 notice in this example that the the, the, the maximum curvature
104:48 a negative value. It's in clal the minimum curvature is anti.
104:59 How did that happen? Oh, , you got your eigenvector again,
105:08 direction shows the most change in All right, I'm going to be
105:14 . So the mathematical definition and that of geophysicists use is the maximum curvature
105:22 the one with the greatest magnitude and minimum curvature is one with the least
105:29 . But a lot of people well, maximum, shouldn't that be
105:32 in terms of sign value than So you're gonna find half of the
105:38 say one way half of the software one Petrell says, maximum curvature is
105:45 than minimum curvature. So you gotta with it. That's part of your
105:51 as interpreter. You just gotta deal inconsistencies out there and make sure things
105:57 defined. How do you figure it ? Go apply it to your
106:01 Look at a feature that's clearly anti and synch over plot it, you
106:07 , co render it and say, , I know what this software is
106:11 and it's gonna do it until at the next release, but probably for
106:14 time, right? But don't assume what you're calling maximum curvature is what
106:21 people are in my work. I use K one and K two.
106:25 want positive and K two most negative confuses those two. OK? Now
106:33 have like OK, there's an it will be, this gets a
106:43 complicated, they will be perpendicular. I have your eigenvector again? I
106:51 another eigenvector. All right. These perpendicular but now in a dipping plane
107:04 I projected onto a horizontal plane, not perpendicular anymore. But anyhow,
107:13 some complications there, but I wouldn't too much. Most of your data
107:18 20 degree data. It's not too . OK. So here, here's
107:24 nose, positive curvature has a positive . Negative curvature has a positive
107:31 I have a don't OK. Here's eyeball, uh negative curvature has a
107:39 value. Positive curvature has a negative , have a ball Here's my
107:46 Positive curvature has a positive value. curvature is zero. Anticline, negative
107:53 less than zero, positive is Here's my positive curvature accent. I
108:00 a valley and you can see I get an elongated ball, I can
108:04 an elongated dome. I can get a saddle and then stretch saddles.
108:11 if both curvatures are zero, I a pointer event, it can be
108:14 but it's still pointer. OK. what about the strike? What does
108:18 do rotates it? And I did . So they go back in 2005
108:47 2000 F. Um Here's what we in the curvature com computation. We
109:01 to take a second root of obstruction the in line direction. Well,
109:08 the first derivative of a parent dip the in line direction. We take
109:13 second derivative structure in the cross line . That's the first derivative of crossline
109:20 in the crossline direction. And then have to take the second derivative of
109:24 mixed directions. So I take the of the in line dip in the
109:28 line and the crossline dip in the line and I average the two.
109:33 those are the equate, those are values I need to compute curvature.
109:38 you can do this two ways, can compute, you can calculate a
109:45 and compute second derivatives. That's what do with horizon based curvature. But
109:50 you have a volumetric estimate of I don't depict anything. I take
109:55 volumetric estimates and dip and in line crossline and off I go. All
110:02 . Well, there I am. , here's my um when I take
110:08 derivative, first derivative Hayden, remember you take the first derivative DDX.
110:18 do I do? I got three . No, no, no.
110:27 a, that's for these two If I ask any question, these
110:31 , it has something to do with Arcania this side of the room.
110:34 first derivative or your calculus find a calculation. I'll give you an
110:43 You give me a delta. You about that too. OK. So
110:49 got, I got three values. wanna calculate the, I got three
110:53 of a curve. I wanna calculate derivative at the center point. Do
110:59 remember how do you do that? view of businesses? Why?
111:09 So tell me, what am I ? What am I doing? Um
111:19 . You got a two delta as . That's cool because I'm delta
111:23 What's an enumerator? Pardon? Three ? What are the coefficients? I'm
111:34 repeating the words you're saying? So you're mumbling something and I'm mumbling it
111:38 , we're going down the toilet. . Let's go buy, you can
111:45 a bottle from Tess Yo, Tess derivative. Remember I got discrete
111:54 I know how we talked about Oh We joke. Nobody remembers you
112:02 Lily Party for Sunday tangent. well, it's gonna be the tangent
112:09 the curve. But how, what would you do? I've got three
112:13 . I got uh Y at minus , Y at plus one, I
112:17 Y at zero. I wanna calculate . OK. OK. OK.
112:25 you're gonna take Y at plus one Y at minus one over to delta
112:30 to delta X. Remember that and as delta X gets smaller and
112:35 it becomes a Yeah, just gonna the derivative. That's how you calculate
112:43 derivative. Just take those two OK? Then you remember,
112:48 then you've forgotten from calculus that oh can be a little more accurate instead
112:54 using two points, I can use points. Oh And then I can
112:59 six points so I can get more more accurate estimates, all right of
113:06 derivative. So that's what we're doing . So now I'm gonna use whole
113:14 of points but here is the, plus one here is the minus one
113:22 then there's a correction term here and correction term there and a correction term
113:27 and a correction term there. So has got at least 66 points.
113:32 . Maybe a couple more that are to see when I take the derivative
113:38 the frequency domain. Now this is signal analysis. And I know you've
113:43 some of this Howie Joe, he talked about wave equation migration.
113:51 Yeah. Yeah. Yeah. OK. Good. You know we
113:53 this, we got the wave You remember the wave equation,
114:06 OK. You remember the heat equation the place and you remember it?
114:20 dimension you're gonna take the second derivative , let's say pressure in X is
114:27 to M over the velocity, second of pressure. With time one dimension
114:36 three dimensions you're gonna take D squared , DX squared, D squared,
114:43 , dy squared, D squared pressure squared, we're gonna use that upside
114:48 triangle. OK. Well, plus del squared pressure equals minus one over
114:54 squared, DP squared, D times or D squared, pressure times
115:02 That's the equation. And then one of solving that, that how I
115:06 how we went through it is oh go take a 48 transform in
115:12 And then in KX and Ky and I have a one dimensional ordinary differential
115:18 which was clearly your second least favorite . Given your response to my
115:28 Now, the least favorite being partial equations. Did you guys take
115:34 You, you took it? The had to. OK. So maybe
115:39 was your least favorite math class? right. Uh But then if you
115:44 it in a one dimensional one by transform, then you go down and
115:47 solved all of that. So I he did all that. So you
115:53 not like this, but you should comfortable that you didn't like it
116:02 Kind of like when I drive for , I get on center street in
116:08 . Oh, man, where the am I? But I'm comfortable because
116:12 been lost there before. You ever on center street? Some place you
116:16 take Alabama over here and you wander and the names of the roads change
116:22 you. Right. We did. complain about that. Yeah, I
116:24 things to complain about like what, is this Lakewood or whatever Rockwood becoming
116:33 ? What the heck is going on ? OK. So now I take
116:39 derivative, the derivative in the time ddudt in the frequency domain is I
116:49 U capital U the spectrum, the , the derivative in the space domain
116:59 dudx is in the frequency domain is , the wave number times the spectrum
117:08 of KX fine. So we're gonna this KX in there. I'm pointing
117:14 the upper right. So if I at this operator and the correct,
117:20 , I'm sorry, this is the derivative. I can, oh I'm
117:31 get rid of the high frequencies. gonna put a little siler on
117:36 Yeah. Yeah. To have the wave numbers or wavelength rather the highest
117:43 numbers. So those will be food were traced. So if my bin
117:52 is 25 m, what's the smallest we can have draw a picture?
118:08 , no, it's not in a NA NA NA NA just draw a
118:12 . Just put a, I've got a grid. What's the smallest wavelength
118:20 can have? I got to define digitally. Remember aliasing. OK.
118:28 with, we're talking about sampling the in the in the frequency domain.
118:34 . So we have the data are a grid. This data set happens
118:41 be 25 m by 25 m. my X RD gal. How many
118:51 per wavelength do I need to have least talked about it last evening?
119:01 the helicopter? Yeah, two points wavelength, right? If you have
119:10 than two points per wavelength, you're have aliasing. OK. But now
119:17 can go ahead and happily create these . But if I 12,
119:23 I wanna have the derivatives applied to that are well sampled, not poorly
119:29 . I don't know if I have little clock thing here. Let's
119:33 I I have it on my 70. No, I don't have
119:41 here. OK. OK. So you guys are gonna a hamster
119:49 think about this again. Yeah, ask you a question on aliasing next
119:55 . That's what I'll do. All . That's the best way I'm gonna
119:58 that down. Oh OK. Area question because that will really mess up
120:25 interpretation. OK. So what we do, I have two points per
120:33 in wind. Great things are What if I go with 45
120:41 Oh, now my space is no 25 m. Now it's like 27
120:46 no, 33 some meters. So take 25 squared plus 25 squared sum
120:51 square root. I know like 0, I need to get rid
120:56 more of those. So a lot the early people who did curvature,
121:00 just had total crap. OK? they would do it on just calculating
121:07 just have noise, just have noise emphasize, footprint would have all kinds
121:13 a oas and so you need to it a little bit. That's the
121:19 we're applying. I hate it. most positive curvature, anti climate.
121:28 looks all right. Oh Most Wow. I see a lot of
121:34 pretty organized corren that you'll do that week. So I'm using um binary
121:44 bar. Now notice the red and are hot, it's fine. We'll
121:52 through that. I might why? then it's going to have a stripe
121:58 the most negative will have a Then we have a oiler in his
122:05 . Why are we staying? Oiler. OK. Yeah.
122:16 Oiler equation E to the I five one minus one equals zero.
122:21 That's Oiler. He was also big um conic sections. So we would
122:27 a bottle like this, cut it or cut it horizontally and he's got
122:35 uh circle, cut it at an . You get in the whips,
122:39 do it with a cone or you're have hyperbole. OK. So you
122:46 all the conic sections out of He is also a cop.
122:53 here's the carrots become perpendicular. That's be K one curvature. A two
123:03 . Cut him a long way And then you can cut them in
123:09 angle, oil or curvature. And think in patrol, they might call
123:12 a parent curvature or something like They give it a different name and
123:18 cutting at an angle. Ok. here is the apparent curvature of that
123:23 data set at like north northeast and here's northwest, southeast. So it's
123:31 shaded illumination. Again, we're looking the curve that's fine fracture vault sets
123:38 you will. Ok. What about shaping that? Now we got another
123:42 another top. It's in patrol as . Here's an image, a photo
123:47 this guy Woodward's hand and here is elevation scan, ok? I can
123:55 a surface to it, compute curvature then he's color coded them according to
124:03 shape. He's got red as normal , a yellow as ridges,
124:09 greens as saddles, blue as, , as a bowls, et
124:19 And he sticks his hand into a device. He said, ah,
124:26 , that's woodwork. And of you've probably seen this movie,
124:31 maybe 15 years old. Nobody's seen movie. You guys don't have a
124:38 . You don't listen to bad music you don't see old movies. You
124:42 , he's got Mr Yaga's eyeball and uses it to get into the uh
124:48 security and stuff like that. You it in a plastic bag. You
124:51 remember that movie anyhow. So the scans are real popular in security.
125:02 to American Airlines. They're in Oklahoma to get in. You gotta
125:07 at that and they're looking at, looking at the blood vessels in your
125:12 pretty hard to describe to disguise Unless you take Mr Yaga's eyeball out
125:18 carry it around in a plastic then then you can do that.
125:23 ? You guys got a lot of to do. You gotta, you
125:25 , watch old movies and listen to music. OK? Different shapes.
125:32 me look at your hand kind of . Carlos, hold your hand
125:42 mind, see how smooth it I do programming. I do a
125:45 of programming. When you're program, do that, it makes your hand
125:50 and smooth. So it would be to recognize me and differentiate us.
125:57 ? All right. We got the curvatures. K one is always greater
126:04 K two by definition, most positive always greater than most negative.
126:09 Can most positive be negative. then most negative has to be more
126:13 . And then the shape index is function of those two are tangent as
126:21 uh Anthony and Zach will say goes minus pi over two and plus pi
126:26 two, right? I just told that. So if I normalize,
126:32 gonna go from minus one to plus most negative curvature, most positive
126:38 OK. This was designed by OK. So don't blame your
126:46 19 twenties. Got topography. I'm calculate the curvature of the topography.
126:53 wanna know the shape when it rains inch. And South Park Colorado,
127:03 much water is gonna go in the River? You need to be able
127:06 map out ridges and valleys and Ok. It's also done in um
127:16 . The oak trees and northern hemisphere to grow on the south side of
127:27 mountain on a ridge and the pine on the north side. Then you
127:33 to Argentina, ah pine trees are the south side. Oak trees on
127:37 north side. So the uh forestry use shape indices as well to figure
127:43 . Well, where's the vegetation? kind of vegetation are you gonna have
127:46 certain places? And there's a bowl one. There's a valley saddle
127:59 don't. Ok. And then we the curved because, oh, this
128:07 a ridge, that's a rig, a rig, that's a ridge but
128:14 . So we need how to form are as well. Ok. So
128:18 have two values, urbanism and shape . There's one for gravity, those
128:24 you who love gravity, apply the index to the gravity So this is
128:29 common than you think. You you go through airport security, they're
128:33 shape industries to you, mapping your and your ears and stuff like
128:38 And, um, so I've got highs, gravity low. Ok.
128:46 data set from, uh, New , the shape index, man.
128:50 isn't talking to me. I see lot of green which is a
128:56 Ah, here's the curving, this deformed it is. At least I
129:01 something uh you know, high Let me co render them. Here's
129:08 I and check it where its Those are bowl shapes where jello are
129:15 where it's cyan our valleys. So Cyanne yellow, those are the edges
129:22 my faults. And in this I have something called uh sinesis.
129:31 remember? Ces. Yeah. It's like clay, dewatering.
129:39 OK. Shave de water. So shrinking and curling up. Ok?
129:43 form these little hexagonal patterns. But can map that and then I can
129:52 them by using coherence or variance if will. And then if I want
129:57 get components on my tape index this , I got my filter for
130:06 Here's the ball. OK, minus and a little bit out valley centered
130:12 minus 0.5 saddle around zero ridge around dome one. And if you are
130:20 with filtering, if I add all D I for everything, but they're
130:26 my bowl shaped component and I see those 10 centers and then I bring
130:33 into the 12 um the expression on . You haven't really talked in detail
130:49 coherence, but you're using variants. you, you know what it looks
130:53 if I have flat layers, not parallel layers and they're cut by a
131:01 and they haven't been deformed. I have a coherence anomaly because the wavel
131:06 is changing across there. OK. amplitude might change as well, but
131:12 the WBO is changing. Here is common situation. I have a little
131:18 of offset along the fault and then have on the hanging wall, it
131:25 like it's dragged up and on the , it looks like it's dragged.
131:35 , it could be dragged if it's very soft ductile material, more likely
131:41 have a main fault and I have bunch of conjugate fault. OK.
131:46 that conjugate faults are offsetting, you , let's say 10 m each,
131:51 main faults offsetting 200 m. I resolve that 10 m offset. So
131:57 looks curved. OK? And same uh conjugate vaults on the other side
132:06 one of you folks weren't paying attention how a Joe's class about the importance
132:10 velocity and your velocity isn't perfect. the image is smeared along the
132:17 so it could be a limit of seismic processing as well. So it's
132:24 look like it's smeared across there. in either case, that pattern is
132:28 , very common. Now, if have less offset, all I have
132:34 a flexure and I have a positive , a negative curvature anomaly and no
132:40 Anoma. So I had this bracketing like I showed you earlier and then
132:48 , no offset. What if I map the fault plane? Well,
132:52 we have something called a bear which a derivative of curvature. We have
132:59 . First derivative is dip the dip one defector. Then second derivative is
133:07 , two curvature vectors. Third derivative a guarantee. And yeah, there
133:12 to be three of them but we're just add them up. OK.
133:16 bar and she ah you don you they say no uh on,
133:26 on that day on the U right a, on a canoe Neptune they
133:35 a sh you can OK. So looking for Uranus before they, before
133:42 found it because Neptune's pattern should be ellipse. And you may remember from
133:50 that I have an acceleration as I around here. OK. As I
133:55 around that ellipse. Well, if deform that ellipse and make it instead
134:01 ellipse, I add an appar seat that ellipse, I make it apparent
134:05 not a perfect ellipse anymore because there's other planet out there right now we're
134:10 for planet X. OK? Or bunch of people are. But what
134:16 gonna call that aberrant behavior. So a change in acceleration. So I
134:20 constant acceleration on the lips. But I had to change it, that's
134:26 the word comes from and from an or all right. So we've got
134:37 velocity, the SDT acceleration D squared squared third derivative jerk and you need
134:51 t-shirt. Don't be a jerk. . That's like physics humor. So
135:00 guy here is a tiy guard where have all kinds of noise but you're
135:10 around, when you're going around a circle, it's acceleration and you will
135:17 him screaming and stuff like that and like about here and stop it.
135:32 Anyhow, you see how this, track changes, the circle changes its
135:41 . So now it's not just th you know, I think I get
135:48 momentum. It's actually changing the What do you do? You,
135:52 go, yeah, it's dirt. never been on a roller coaster.
136:02 . You've been on a roller All right. Good. Good.
136:04 to hear that. All right, . So that's jerk. OK.
136:08 , that's, that's what we're, talking about third derivatives. OK?
136:15 of arithmetic. You don't need to the arithmetic. Here's what it looks
136:20 for that data set, but it's is Fletcher's. So not only can
136:28 measure how flexed the rock is, in what direction. So here the
136:36 he has a, a magnitude which plotting against the gray scale and,
136:44 orientation, which I'm plotting against the color book. So we can compare
136:50 against coherent. And you'll say, , I see all these little bitty
136:54 here. I don't see a channel up there. And these faults I
137:03 very nicely. Then if I look coherence, oh, I don't see
137:09 faults very well. Why no I mean the offset is below a
137:17 . I see it. I do this channel but this channel, if
137:23 look at the seismic data, I a flat reflector in the channel and
137:27 to it, the floodplain is So I'm looking at the changes in
137:32 , flat to flat, no OK? So one attribute is measuring
137:38 channel but not looking into the small . And the other one is measuring
137:44 small, small font, not you , not seeing the channel and then
137:48 can co render them, you get best of both worlds. OK?
137:53 that CSIS area just plotting them up a a bey and then down
138:00 there's a bunch of uh channels and oh cutting down the the uh the
138:11 . Oh small channels over here, the small uh FTS over here.
138:16 no uh coherent some channels there. ? So one thing you can do
138:27 you guys have played with patrol So you're probably comfortable with this.
138:34 I wanna look at just the fox are oriented in this example, North
138:44 . So I've got my circular color like you've been using where I got
138:48 is blue and South is yellow and 61 22 43 20 are the other
138:57 colors. And now I'm gonna put as the background. Then I'm going
139:04 put the a gray color bar for aberrancy magnitude. So everything that's kind
139:14 uh not, not uh deformed, is gonna be pure gray. Everything
139:21 is deformed is gonna be some And then just so I know where
139:24 am. I'm gonna use a binary , white color bar for the
139:27 So you've been doing these kind of and you're gonna do more of
139:30 Ok? But I'm gonna do one thing and we're gonna go into that
139:35 bar for the ASM you, I'm say I only want to look at
139:41 north. I'm sorry, the north is zero degrees and the south which
139:48 minus 180 plus 180. And I'm gonna make transparent and just look
139:56 my background whatever I have on Oh I sent it to gray.
140:05 no one uh 30 degrees west of . See how I changed my
140:16 60 changed it again. East but I got it centered around 9270
140:26 then 120 degrees northwest and then north and all the way back again.
140:35 . So now what I have, looking at different false steps on the
140:41 line. OK. So you may a hypothesis that says, hm,
140:50 faults that are oriented with their let's say northeast, southwest, those
141:01 perpendicular to today's minimum stress. And if I inject cropping and high pressure
141:11 in it, I can open those easier. OK. So these zones
141:17 gonna be easier to open in areas I don't have those northeast,
141:21 southwest. Uh those aren't going to as easily. So I can make
141:27 steps. Also, the different faults often generated at different times in
141:34 So in some cases, you'll have group of faults that is di genetically
141:43 and cemented and another that is So one forms a flow barrier,
141:50 other flows forms a flow conduit. , you may not know that.
141:56 can we do? Let's look at different fault sets, find out what
142:01 density of the false sets are everywhere your uh serving. And now it's
142:08 production of oil gas and water to to these different fault sets. You
142:13 uh huh. These ones are really for producing water, which I'm not
142:21 in and others produce less water, oil. Uh those I am interested
142:26 . So you answer those kind of . OK. Yeah. Are they
142:35 that again? Are, are they OK. The colors here correspond to
142:45 color bar and you probably, you ask this question, but I'll answer
142:50 because I didn't hear your whole Some of the faults are smeared on
142:58 line. So here is my Here is my fault. OK?
143:04 the fog looks like this, it's be really, really sharp. If
143:08 fog looks like this, it's gonna really, really smeared. OK.
143:15 a lot of this is the way look at it. I probably have
143:18 , a picture that's not the question asked. But I'm gonna just look
143:22 it again. Notice how some of are, are this this wine I
143:27 goes kind of northwest Southeast. So is smeared and these are sharper that
143:34 sharp. Same. OK. So out your question again, I'll lean
143:40 . Was that the question? Oh . OK. Good. Thanks.
143:52 Here you bring it into the Box in patrol and now you, you
143:57 with your opacity and stuff like that I can make a volume of
144:03 OK? And the colors OK? we got long wave short wavelength and
144:11 got volumetric versus horizon data uh North Fort Worth Basin. So shale resource
144:21 in Northern Texas, a long wavelength , moderate wavelength, short wavelength,
144:29 short wavelength. OK. The way Saleh Al Gori came up with this
144:35 , he was looking at a potential data uh or potential field papers.
144:43 . And how they uh how they their data. In this case,
144:49 sun shading, right? So it of a uh thermal imagery data.
144:54 they got an airplane and they're measuring um the heat, OK. Radiative
145:03 from the surface happens to be from Africa. And they're gonna take derivative
145:11 . They're gonna take a first Well, there's one, a 0.75
145:19 while it is red and a 1.25 quoted is blue and the first derivative
145:27 green. So you're comfortable with, the? No, I have made
145:31 comfortable whether you wanted to be or with the first derivative because we just
145:35 through that. What the heck is fractional derivative? Well, the way
145:39 think about that is I'm gonna take IKX and take a power fractional
145:43 right. So here is how uh did it, the derivative DPDX take
145:51 fourier transform of the dip multiplied by , take the inverse transform. So
145:58 is the first derivative, the second of IKX square. Oh Joe did
146:03 , I know he did your So we're in the wrong Neurological
146:17 Gonna look at the cato horizon. is uh the curvature computed using an
146:26 calculator or a calculator, let's just a calculator along this picked horizon.
146:34 so I took the time of the and here's the old survey, the
146:40 1997 survey, the 99 survey and it's the same calculation. But with
146:48 dips calculated biometric for you, remember the volumetric dips were a little
146:53 So that's what the, yeah, helps a lot and then coherent
147:03 If you will. You see white in the snow. Now,
147:09 uh, you know, he did phd. There you go. White
147:16 in the snow. He didn't understand because, ah, a baby camel
147:20 the sand dune that he understood. that's, and ok, now let's
147:28 along the time. Let's look at yellow time slot. So nothing
147:33 here's the coherence image kind of ugly because we're in uh sands and,
147:40 know, kind of gravels and stuff that kind of ugly here. And
147:43 is the cattle limestone. Ok. square cut out. What's the square
147:55 ? What's the square cut out this ? What's the square cut up?
148:12 ? Yeah. Why? No It's perfect square. Why no
148:27 Ok. Farmer Smith doesn't want no geophysicist on his property. As simple
148:35 that. He owns the oil If he owns just a farm,
148:41 can still shoot under it. But in the United States, Canada,
148:48 are the main two places. The commonly on the oil rights and the
148:55 who owns it might be from 102 years ago. And the person who
149:01 the oil rights is different from the who owns the gas rights, different
149:04 the person who owns the coal the zinc rights, the gravel rights
149:08 could be all mixed up and they be, they own the oil rights
149:13 the Pennsylvanian Deeper or Pennsylvanian Shallower. . So as a seismic processing
149:26 then you process the whole thing. they image under there? Yeah,
149:31 imaging under there. But when they it to the client, they have
149:34 cut it out. That's a no area. So that's, that's the
149:39 . And if it's, you'll see real common in, in Canada,
149:43 have all the rights, you down to 3000 m and then 3000
149:48 , they gotta cut all that So you have a mute zone,
149:51 m down in some kind of square and then sometimes they'll cut it in
149:56 shallow part. That's what, that's they are. And these cause problems
150:01 the cause of artifacts in the seismic . A real real come no per
150:08 , other areas you'll see it If there's a platform in the way
150:13 like that, that's more of an where you can't really see beneath
150:18 OK? So this was coherence here the simple curvature. I'm just gonna
150:26 the first derivatives biometrically and calculated that OK. Let's do 0.75 derivative 0.5
150:43 . Well, I'm starting to see geology. I'm seeing a grid
150:48 I see it, a strike football I see these things, I'm gonna
150:53 redel shear. Then I see some of, you know, funny rounder
150:58 . So I'll go through there OK. First derivative 0.75 derivative longer
151:07 longer still very long. OK. look at this and then look at
151:14 coherent big difference one better than the . No one is measuring something different
151:23 the other. One is measuring what's to the shape of the reflectors.
151:30 the coherence is measuring variant if you is measuring the continuity of the
151:39 OK? So one's measuring how is orientation changing? And the other is
151:43 the cap. Let's go down to Ellenberger Dolemite and you see these
151:52 OK? What cars like coherent? I got kind of fill right?
152:00 the data being scattered and so So I have a poor image and
152:05 here is the curvature and they're You'll notice that the big cars features
152:14 I don't know, maybe look at one here and I'll leave my mouse
152:18 and you'll say, oh, it's of an intersection of a monument here
152:22 maybe another monument there. There are lot of the intersect. OK.
152:33 here I've got a single data I've got amplitude in line and crossline
152:39 can do that calculate coherence, there's , the same data set.
152:45 And here is a car's collapse. , the cars collapsed, there's a
152:54 collapsed. OK? Some collapses and got a fault, then another
153:03 Then I'm gonna ask, do I a quadratic surface. Sure.
153:07 So I'm gonna calculate the principal curvatures their strike, most negative curvature.
153:14 the bowls like my eyeball are gonna these blue areas. I do have
153:19 domo areas locally, but basically what was relatively flat limestone that had you
153:27 it. Oh, real common It might collapse. There's a
153:35 another fault over there. I can render that with the coherence on the
153:42 slide and the uh amplitude on Well, that's a good question.
154:03 what I think they look like. go to Arkansas. There's a little
154:05 of Devil's Den State Park with big in it. And the hiking trails
154:09 down these joints so you can walk half mile down one joint and then
154:12 turn, make a sharp left you walk a quarter mile, the
154:16 one and then you go another turn and it's a whole lot of fun
154:20 walk through. But I think these that are big like you can walk
154:24 them. OK? OK. Here's most positive curvature. Remember my
154:35 the most positive curvature has a negative . If it's a ball, all
154:40 four things are the cars collab never . OK. There's my collapse features
154:49 my fault cor render it measuring the part and the edges of these collapsed
155:02 . OK? But measuring the edges the collapse features and the edges of
155:06 fault I can put positive and negative together. Same thing. OK.
155:14 What about shape components? OK. the shape index modulated by curved.
155:20 I happen to use white or non and the blue RD wax features.
155:33 ? And then I can calculate the component. And then, so here's
155:40 bowl component in blue collapse feature collapse . And then I can go use
155:49 visualization, you see the green So I did the visualization and patrol
155:54 now I can rotate this guy and I can see these collapsed chimneys in
156:00 dimensions. That's cool. And in shallow part up here is the Atoka
156:12 . And there was a guy called who realized that, oh, I
156:18 drill in the structural ro because up that Atoka formation towards the shallow
156:26 that's where I have my gravels that good reservoirs. So instead of throwing
156:31 structural highs, he drove the structural and he made, you know,
156:37 , at least $350 million because that's Jackson school at UT, not the
156:44 School at uh not the Jackson School ou. So he gave it to
156:48 University of Texas in Austin to their school. $350 million. Not
156:55 not bad by growing woes. And they're associated with the deeper
157:00 Later on, he had more mo more money by selling it to Mitchell
157:04 down in Galveston. He was in and then later to uh Devin for
157:09 shale. Ok. Now you can William and volumes. We got the
157:17 . How deformed are the valleys and ridges? Ok. So I'm gonna
157:21 that one green. I'm gonna plot one blue north south and put that
157:26 red northeast. Southwest. Ok. my money? I'm in Dubai.
157:33 , well, that's kind of I can co render it, I
157:39 calculate roses. Well, let me back here. Well, thank
157:45 Sure. That's pretty, kind of . But come on, man,
157:52 it's blue, here's my arrow. can see this blue stuff is north
157:57 and I can see this red stuff , well, here's my north arrow
158:02 east northeast. What's the value of ? Well, if you're just going
158:09 display it, I mean, it makes pretty wallpaper. OK.
158:14 , at every Voxel in the I now have a measure of how
158:20 form the subs surfaces. And in orientation, the computer doesn't see
158:28 You're doing pretty heavy pattern recognition when lining these things up with the North
158:33 . OK. So now I can to correlate production from horizontal wells to
158:41 to the structural deformation trend. And can ask questions. Do I have
158:47 production when I cut perpendicular to them when I'm parallel to a particular of
158:56 and set? Ok. So you make all the red ones separate all
158:58 green ones separate all the blue ones in a rose ditch. OK.
159:05 convergence is yeah. So curvature we biometric depth and then we said,
159:19 the change in depth in the in direction? What's the change in depth
159:22 a cross line direction? We put in there and calculate the curvature.
159:27 about the change in dip in the direction? So if I'm dipping to
159:32 north and then I dip even more the north as I go deeper,
159:39 means I'm pinching out to the OK. So I can measure that
159:45 in depth. So this picture here the first paper I've seen published by
159:49 fellow called Art Barnes. He was landmark and he just used it.
159:55 was just looking at the convergence in dimensions and he says, oh,
160:02 , if it's Cyan, I'm converging the right and the red and
160:08 I'm converging to the left. So you can see it's pinching out to
160:11 left here it's pinching out. So got my normal from, let's say
160:20 gradient structure tensor. And then I'm take, if I took the uh
160:29 of the normal, I do all this, that's going to give me
160:34 called the mean curvature or the average . So K one plus K
160:42 And here's my in line apparent crossline, apparent dip and my parents
160:48 magnitude. So the change in how meters down do I go for every
160:57 in line. How many meters down I go for every meter cross
161:03 I need the NZ thing. So many does it is a trick
161:12 How many meters down do I go every meter down? How many meters
161:23 do you go for every meter Yeah, took me eight years to
161:32 that out. Ok. But yeah . So if you got these apparent
161:38 in line cross line and you wanna normals. Well, now I got
161:42 apparent dip, an apparent dip parent vertically is one is one.
161:47 So then you're gonna get the So P squared P square and like
161:56 does this one come from? That's it comes from. You gotta get
162:01 three apparent dips and normalize. All . So you're gonna see these and
162:05 the equations. Nobody explains that. so those are the different dips and
162:13 gyp exist. It's kind of like to Big Second. Who's been the
162:25 D so a big, do you what big ticket is? You don't
162:36 environmentalist, you've been to big right? Tess. If it bites
162:41 stings, it thrives in the big . You haven't been there. The
162:47 Texas. It's a national on a park. It's a national, what
162:57 it? That big ticket National Preserve . So it, it's,
163:04 it's dense. OK. Been the ticket anyhow. If you go
163:11 they got the towns and stuff like . Right. If it doesn't
163:17 people paint it. If they have saw, they paint it, they
163:23 a bucket, they paint it, got a hubcap, they paint
163:26 Come on, grandma's got this No, you don't. When you
163:32 visit in breakfast she doesn't have stuff that around her house. Like
163:38 maybe she wouldn't have a pain that but tainted hub down, you go
163:46 big, big business is kind of and crafts. There. You get
163:49 painted everything. If it doesn't it's painted. OK. You guys
163:55 don't have a right now. The thing is true. If you're a
164:02 , if it's a vector, what we gonna do with it? We're
164:06 take divergences. What else are we do? We pick a curl,
164:14 it is. If it's a we're gonna do that. So it's
164:19 . Oh Divergence says whether you got source or a sink, the curl
164:25 , is it going down the Right? The rotation part. So
164:30 got dip, the mean curvature is the divergence of the vector depth.
164:37 can take the curl as well. , what do I do with
164:41 Well, it's got different components. curl of a vector is a vector
164:47 then I can measure angular and OK? A little bit of mathematics
164:53 worry about it, but that's what mathematics is. I'm just taking the
164:57 , it's taking different kind of I know you had corals and he'll
165:03 us. You forgot that you remember . Not as long ago as it
165:10 been for me. Thanks. So me, it's like 53 years
165:17 54 years ago. So here we , I'm gonna make it now.
165:24 of that is measuring the re change depth with respect to the vertical.
165:32 wanna be a little more careful than , the change in depth with respect
165:38 the direction perpendicular to the average reflector . OK. You can be a
165:44 better than that. So here is same data set from West Texas.
165:48 shown a couple of times and I'm to reflect your convergence. Uh Let's
165:55 at wine a a prime. I it's purple here. That means it's
166:00 to the north northeast. It's green to the south side, my north
166:08 . Uh it's flat here. So not converging. I mean, just
166:11 the middle and let's look at line a prime. Here's a, a
166:17 . Oh Converging to the north conversion the south. We kno OK.
166:30 , well, magenta means it's converging the east northeast. There's real light
166:37 . Converging just slightly to the west blue means it's converging north south.
166:44 here we are. All right. converging in the north, north east
166:49 out. I've got a reverse talk and then gentle convergence this way and
166:54 it's white, everything's parallel So what the blue one? It's converging perpendicular
166:59 the plane. Look at that Oh yeah, it is conversion to
167:05 right. So what do I I've got a reverse fall. My
167:12 fault is coming like this up. don't have to be. That's
167:20 They can rotate a bit. So I have more accommodation space to the
167:28 than I do, actually less accommodation I do to the south. So
167:34 pinching out on the north towards the in front of that fall.
167:42 You had to pick all these It would take a long walk,
167:50 ? All right. And then I've a couple of weeks to show you
167:55 reflect the rotation that's rotation about the . So another equation we can
168:03 the positive goes down to the right up to the right. Yeah,
168:12 to the right in the middle and to the right yonder, the pictures
168:17 down, drop ground. Here's that data set with all kinds of
168:22 And remember I said we had a here or here is all cars.
168:28 are quite complicated. OK? Gonna this one. OK. Here,
168:38 got, I'm doing volume versus horizon curvature. I've got to change and
168:45 across this fall. I'm gonna see on cured. I don't have a
168:51 in dip across that fall. If were to pick the green horizon and
168:55 curvature, I'd have a fall I'd curvature anomaly. But here I just
169:00 the same dip. Then if I differential compaction, I'll see that as
169:10 a curvature in here's another differential So here the floodplain is compacting with
169:17 to the fill. Here, the is compact. I've got an amalgamated
169:22 . I'll see that on curvature. curvature I won't see. And this
169:28 becomes so complicated with the differential compaction have a lot of anomalies, but
169:35 can't untangle them. So that's, that. So any questions on curvature
169:52 two points? Yeah, alpha. So the, yeah, so I
170:03 what you're talking about. 7075. . Oh Wow. So I didn't
170:14 , keep going. Um I think know which slide you talk this
170:30 correct. OK. So in this , uh we're looking uh so the
170:36 far away looking at slide 89. if I were to take the first
170:43 in the frequency domain, I multiply IKX, if I were to take
170:48 second derivative, I multiply by IKX , so IKX squared is I
170:57 KX squared minus KX squared third derivative fourth derivative IKX to the fourth
171:05 If we wanted to take a fractional , which these Cohen and Cohen
171:09 we had to figure out what do mean by that? What the heck
171:12 a fractional derivative? Now, since time, there are some publications in
171:17 mathematical journals about what a fractional derivative , but it's more of a mathematical
171:23 . And so here, instead of this, the second power, I'm
171:27 take the one half power or the power. OK? So that's all
171:33 is. So it's trying to define a fraction. So we would take
171:37 0.75 derivative of the shaded of the shaded heat flow heat picture. And
171:45 would give me the red and the alpha equal to 1.25 would give me
171:50 blue and the green alpha is equal one. So it's a way of
171:54 fractional derivatives that helps. Now, fact that you're confused, that's why
172:02 don't like. That's how we OK? And instead the better way
172:12 you guys also didn't like, but OK. Um Back here,
172:23 here I said let me put a in the frequency domain here.
172:30 if I did a filter that instead KX, instead of walking by
172:35 I wanted to do the half then I would have a filter.
172:42 would be one over square meter that would be the curve.
172:49 And the reason we do it this is, you know, you're trying
172:53 market ideas to people in an oil or a service company and the people
173:00 are gonna make the decision whether this valid, they're gonna be mainly seismic
173:09 . If you talk to a seismic and say I'm taking a fractional
173:16 they're going to say, what the are you talking about? Right?
173:19 not gonna like it. If they , oh, I'm gonna apply a
173:23 in the wave number domain. They filters every day, every day of
173:28 week, all day long. So understand that. So the fractional
173:33 how we started the way we describe today, we apply a filbert.
173:40 . So a long wavelength curvature, show you pictures of this next
173:45 but the long wavelength curvature is gonna you the longer wavelength features, short
173:51 that's gonna show the shorter features. be noisier too. OK? So
173:58 time, right? So we'll come at one. All right. Any
174:06 on the lab right now that will your lunch if you don't resolve you
174:13 one. OK? That's
-
+