Synthetic model for PSDM

Oil and natural gas exploration -- geology and geophysics
Post Reply
jefry123
Silver Member
Posts: 16
Joined: Wed Oct 05, 2016 4:25 am

Synthetic model for PSDM

Post by jefry123 » Thu Sep 07, 2017 3:46 pm

I am going to show the superior of PSDM in relative to time migration such as PSTM using a simple synthetic model. I want to forward model it using ray tracing method in GXII. However, I want to know your suggestions. In your opinion, what should it be look like?

GuyM
VIP Member
Posts: 513
Joined: Sat Mar 24, 2012 11:35 pm

Re: Synthetic model for PSDM

Post by GuyM » Thu Sep 07, 2017 6:19 pm

Why not use the various SEG benchmark datasets?

If you are looking towards a publication, then these would be familiar to an audience.

http://wiki.seg.org/wiki/Open_data

if you are looking for simpler models then I'd suggest:

a) a water layer down to perhaps 300-400m
b) below that a "sediment" layer, with a velocity starting at 1600 m/s and increasing at 0.7m/s per metre
- use gardner's relation to develop the density model
c) below that, a dipping interface representing an unconformity with a big velocity/density contrast, and a nice reverse fault in there
d) below that, a plane layer

If you wanted to get funky you could through some channels into the seafloor as well.

The main thing with this approach is that you get ray path bending *within* the upper layer, which in turn impacts on the angle of incidence on the dipping layer. The fault will be hard to image accurately with even anisotropic PreSTM, and the plane layer at the bottom will show up residual effects nicely. Especially if you include channels.

You could ramp up to this, of course!

jefry123
Silver Member
Posts: 16
Joined: Wed Oct 05, 2016 4:25 am

Re: Synthetic model for PSDM

Post by jefry123 » Fri Sep 08, 2017 5:13 am

The SEG models are somehow complicated. I want to do this from the beginning, RMS velocity picking, interval velocity model building, ... .
I know this is too much, but could you please draw what you suggest?

GuyM
VIP Member
Posts: 513
Joined: Sat Mar 24, 2012 11:35 pm

Re: Synthetic model for PSDM

Post by GuyM » Fri Sep 08, 2017 6:02 pm

Sure!

This was the kind of thing I meant.

It's not super geological, but it has some key elements.

a) pushdown effect from the channel
=> this will give you near and far ray-path differences; make sure your maximum offset is greater than the channel width by ~50% or so to get the real "dog-legged gather" effect in TWT where the near and far offsets need a different NMO correction.

b) fault shadow effects
=> getting the foot-wall block to be flat in depth with no edge effects will be a challenge

c) deep, horizontal layer.
=> anything you get incorrect in the near surface will appear here, magnified

The velocities I've picked out are pretty reasonable for water, muddy channel fill, seafloor sediments with a compaction gradient, something a bit more solid (limestone, carbonates, compacted shales) and then a harder - but still sedimentary - basement with a lot of diagenasis.

I'd pick densities from the literature if you are doing full modelling; Gardener's relations good for clastics, and then assume (say) limestone and highly cemented old red sandstones for the other layers. Top is seawater....

You could work up to this, of course; the near surface channel with just a plane layer under it (and a vertical gradient) is one challenge.
The reverse fault (without the channel) is another.

While you could go for some complex salt model the challenge there is its just as likely to expose issues with the forward ray tracing and synthetic as it is with the imaging side of things. Much the same logic as why the original "stealth fighter" had angular panels - much easier to be sure the model matches a possible reality...

Hope this is useful.

GuyM
Attachments
model.jpg

jefry123
Silver Member
Posts: 16
Joined: Wed Oct 05, 2016 4:25 am

Re: Synthetic model for PSDM

Post by jefry123 » Fri Feb 09, 2018 8:08 am

What do you suggest for an anisotropic PSDM?

Something that shows its preference over isotropic PSDM.

GuyM
VIP Member
Posts: 513
Joined: Sat Mar 24, 2012 11:35 pm

Re: Synthetic model for PSDM

Post by GuyM » Fri Feb 09, 2018 3:12 pm

Well - you are going to need a more complex model.

With anisotropic effects we are saying the V(horizontal) is different to V(vertical) as a result of some physical property; perhaps fractures and their orientation, or how the layers within a formation are configured.

So - maybe a nice thick layer over a salt dome with a flat target underneath (in depth) - and a nice high anisotropic factor representing fractures in that upper layer?

Salt domes are great for this kind of thing because there's a lot of sensitivity to the angle of incidence *and* the location of the incident point when it comes to how the rays are refracted through the salt. The "top salt" changes rapidly in curvature, and hence so does the normal to that surface.

That is assuming you want to show the *difference* between isotropic and anisotropic preSDM?

jefry123
Silver Member
Posts: 16
Joined: Wed Oct 05, 2016 4:25 am

Re: Synthetic model for PSDM

Post by jefry123 » Sun Feb 11, 2018 4:05 am

Yes, I am going to show that in some cases of geological structures anisotropic PSDM is a better tool than the isotropic PSDM.

GuyM
VIP Member
Posts: 513
Joined: Sat Mar 24, 2012 11:35 pm

Re: Synthetic model for PSDM

Post by GuyM » Sun Feb 11, 2018 2:12 pm

The challenge there is to make a model that demonstrates an anisotropic effect of significant size.

For the most part it tends to show up at the far offsets (as a hockey stick) which means that simply muting out the anisotropic effect and focsuing on the smaller-angle ray paths produces an adequate structural image. Its when you want to look at stratigrapic inversion or rock physics that the wider angles tend to matter more.

There's also the fact that "imaging depth" is only "true vertical depth" if you take into account anisotropy, but that again depends on the anisotropic parameters being accurate, rather than "good enough to meet the imaging criteria"

A thick, highly anisotropic layer over a complex structure with a flat layer (in depth) under it all is the way to go, I think.
Plus wide angles (>45 degrees) which means long offsets - 2-3 times the target depth should do it.

jefry123
Silver Member
Posts: 16
Joined: Wed Oct 05, 2016 4:25 am

Re: Synthetic model for PSDM

Post by jefry123 » Mon Feb 12, 2018 1:56 pm

Thank you for your suggestion.
Apparently the software I was using for the forward modelling do not take into account the anisotopic parameters in demo mode (despite that it asks me to import the thomson parameters!). Do you know any other software I could use?

GuyM
VIP Member
Posts: 513
Joined: Sat Mar 24, 2012 11:35 pm

Re: Synthetic model for PSDM

Post by GuyM » Tue Feb 13, 2018 5:42 am

Sorry no; will check with the R+D teams here but we tend to go straight to the complex model for migration algorithm development or those of clients rather than our own synthetics.

Post Reply