[Password Notification] About the prediction of porosity

[侯建勇]

=============================================================
添付ファイルダウンロードセンターパスワードのお知らせ
Notification of Download Center Password of Attachment
=============================================================

先程送りましたメールの添付ファイルのダウンロード場所のパスワードをお知らせします。
Password of downloading the attachment is as follows.

[Subject] About the prediction of porosity
[password] te_PxM2Mjx<=


※ダウンロードセンターのURLは元のメールをご覧ください。
*The URL of download center please refer to the original e-mail.

[侯建勇]

===========================================================================
     Notification of Attachment File Separation

 本メールの添付ファイルは予め分離されております。
お手数ですが、こちらからアクセスして取得して下さい。

 The attachment file of this e-mail was separated.
 Please get the attachment from the following URL.

   [URL] https://gqj26020.activegate-ss.jp/d/?02f987572c1112185f0839efe416d0afSIV
   [e-mail address] Us...@opendtect.org

 * If the URL is inserted line feeds, please join them and access.

ダウンロードセンターのパスワードは別メールで配送致します。 しばらくお待ち下さい。 
Please wait for a while, as the download center password is sent by another e-mail.

===========================================================================

Dear users,

I wound like to do a lateral prediction of porosity with a 3D Impedance volume and some porosity data from real wells. But, I don't find a good correlation between impedance attributes and porosity in crossplots (please see the attached file).

Who can tell me HOW to improve the correlation and What kind of attributes are better for the lateral prediction?

Many thanks

Jianyong Hou

[Paul de Groot]

Dear Jianyong Hou,

I cannot see the image you sent hence this comment may not be applicable. 

The relationship between seismic attributes and well log properties is very sensitive to the well ties. Small shifts and small errors in time-depth conversion have a detrimental effect on the correlations. As Acoustic Impedance can be calculated from sonic and density logs I recommend cross-plotting Well log AI vs Well log porosity instead of Seismic AI vs Well log porosity. Also if you train a neural network along well tracks to predict well log properties you should use well log inputs if you can. An added advantage of using only well log inputs is that you have many more samples to train on as the input is then extracted in depth at well log sampling rates as opposed to two-way time at seismic sample rates.

I hope this helps. Best regards,

Paul.

-- 

Paul de Groot

CEO

______________________________

dGB Earth Sciences
Phone:	+31 53 4315155
E-mail:	paul.d...@dgbes.com
Internet:	dgbes.com

 SEG Distinguished Achievement Award 2016

______________________________

--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+unsubscribe@opendtect.org.
To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.
To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/20180821T095548%2B0900-30560-0004.N0027055%40mail.gmail.com.

For more options, visit https://groups.google.com/a/opendtect.org/d/optout.

[Blacker, Kieran J.]

Agree with Paul, and suggest perhaps getting a copy of:

Simm, Rob, Mike Bacon, and Michael Bacon. Seismic Amplitude: An interpreter's handbook. Cambridge University Press, 2014.

Kieran John Blacker

Postgraduate Research Student

Geophysics and Borehole

 

School of Geography, Geology & the Environment

Bennett Building, University of Leicester,

University Road, LEICESTER, LE1 7RH, UK

 

+44 (0) 116 252 2523

 

Alternative email: black...@gmail.com

---

From: Paul de Groot <paul.d...@dgbes.com>
Sent: 21 August 2018 10:41:30
To: us...@opendtect.org
Subject: Re: [OpendTect_Users] [Password Notification] About the prediction of porosity

 

To unsubscribe from this group and stop receiving emails from it, send an email to users+un...@opendtect.org.

To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CABuvx8Es-UG_mCBQ_x5q_Rx-o02xuk8geWeG%3D4ywCT6qWs-dLQ%40mail.gmail.com.

[侯建勇]

Dear Paul

Thank you very much for your advice. I will try to calculate the AI from sonic and density of real wells.

Actually, I have 60 wells with sonic, density and porosity calculated from density log. I want to train a Neural Network using the real well logs and seismic attributes extracted from well location to predict porosity dirtribution in a 3d seismic volume. 

I have an another question. If I use the AI calculated from sonic and density and porosity extracted from real wells to train Neural network, Can the NN be applied to whole 3D Impedance volume to predict the property, not only along well tracks?

Best regards,

Jianyong Hou

To unsubscribe from this group and stop receiving emails from it, send an email to users+un...@opendtect.org.

To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/20180821T095548%2B0900-30560-0004.N0027055%40mail.gmail.com.

For more options, visit https://groups.google.com/a/opendtect.org/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+unsubscribe@opendtect.org.
To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CABuvx8Es-UG_mCBQ_x5q_Rx-o02xuk8geWeG%3D4ywCT6qWs-dLQ%40mail.gmail.com.

[Paul de Groot]

Dear Jianyong Hou,

If I use the AI calculated from sonic and density and porosity extracted from real wells to train Neural network, Can the NN be applied to whole 3D Impedance volume to predict the property, not only along well tracks?

Absolutely. In fact this is our standard workflow for NN seismic predictions from wells. The only thing to watch out for is that AI from logs and AI from seismic have the same units of measure. If not, apply a linear scaling function to scale the AI from seismic to the AI from logs range before passing the AI from seismic through the NN.

Best regards,

Paul.  

-- 

Paul de Groot

CEO

______________________________

dGB Earth Sciences
Phone:	+31 53 4315155
E-mail:	paul.d...@dgbes.com
Internet:	dgbes.com

 SEG Distinguished Achievement Award 2016

______________________________

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CABuvx8Es-UG_mCBQ_x5q_Rx-o02xuk8geWeG%3D4ywCT6qWs-dLQ%40mail.gmail.com.
For more options, visit https://groups.google.com/a/opendtect.org/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+unsubscribe@opendtect.org.
To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CAOjrzPgN%3DR4%2B_NKrCbUfORY8dZ6Y%3D_z66mhGeDHdp5%3DKfopjYA%40mail.gmail.com.

[侯建勇]

=============================================================
添付ファイルダウンロードセンターパスワードのお知らせ
Notification of Download Center Password of Attachment
=============================================================

先程送りましたメールの添付ファイルのダウンロード場所のパスワードをお知らせします。
Password of downloading the attachment is as follows.

[Subject] Re: [OpendTect_Users] [Password Notification] About the prediction of porosity
[password] 3QRV7-L5dZsJ

[侯建勇]

===========================================================================
     Notification of Attachment File Separation

 本メールの添付ファイルは予め分離されております。
お手数ですが、こちらからアクセスして取得して下さい。

 The attachment file of this e-mail was separated.
 Please get the attachment from the following URL.

   [URL] https://gqj26020.activegate-ss.jp/d/?18ca39200533f756b7b112417acfad28SIX
   [e-mail address] us...@opendtect.org

 * If the URL is inserted line feeds, please join them and access.

ダウンロードセンターのパスワードは別メールで配送致します。 しばらくお待ち下さい。 
Please wait for a while, as the download center password is sent by another e-mail.

===========================================================================

Dear Paul

Thanks a lot!

I calculated the impedance with sonic and density from 60 real wells and created a log cube of the impedance. And then, I extacted the attribute set from the log cube and trained a Neural Network with the impedance and porosity from real wells (please see the attached file). But, when I applied the trained NN to whole 3D seismic impedance volume, a error message "No valid data found" is shown. I don't How to solve the problem. Could you give me more helps, please?

Best regards,

Jianyong Hou

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CAOjrzPgN%3DR4%2B_NKrCbUfORY8dZ6Y%3D_z66mhGeDHdp5%3DKfopjYA%40mail.gmail.com.

For more options, visit https://groups.google.com/a/opendtect.org/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+unsubscribe@opendtect.org.
To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CABuvx8ECuBjmU3NuDNnZvggefj7nPZ3dP1ke2f7NQTi62KKPrA%40mail.gmail.com.

[Paul de Groot]

Dear Jianyong Hou,

Yes, sorry about that. My mistake. I should have told you that this workflow involves a small hack. You need to edit the Neural Network file manually with a text editor. The NN must know where each input attribute comes from. You trained the network on a data set with AI log cubes but you want to apply it on the inverted AI cube. 

So, please go to the NLAs directory in the OpendTect project directory (in my case: C:\Surveys\F3_Demo_2015_intro\NLAs).

Now open the neural network file with a text editor (I tested it with vi on Linux and with Notepad++ and Wordpad on Windows. vi and Notepad++ worked but Wordpad did not. The NN file could not be opened any more after editing with Wordpad!).

Near the end of the Neural Network file you find the reference to the seismic file from which the attributes are computed. You need to change the Strorage d and the UserRef (file name). You can get the id from the seismic file manager. The file name can be copied and pasted if you use the "Rename" function in the manager.

In my case I changed:

1.Definition: Storage id=100010.155 output=0

1.UserRef: P-Impedance log cube

into:

1.Definition: Storage id=100010.7 output=0

1.UserRef: 7a AI Cube Std

Finally, to apply this network in OpendTect you must ensure that the original network is flushed from memory. The safest way to do this is to restart OpendTect or to change to another survey and back.

I hope this helps. Best regards,

Paul.

-- 

Paul de Groot

CEO

______________________________

dGB Earth Sciences
Phone:	+31 53 4315155
E-mail:	paul.d...@dgbes.com
Internet:	dgbes.com

 SEG Distinguished Achievement Award 2016

______________________________

On Thu, Aug 23, 2018 at 2:46 PM, Paul de Groot <paul.d...@dgbes.com> wrote:

-- 

Paul de Groot

CEO

______________________________

dGB Earth Sciences
Phone:	+31 53 4315155
E-mail:	paul.d...@dgbes.com
Internet:	dgbes.com

 SEG Distinguished Achievement Award 2016

______________________________

--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+un...@opendtect.org.
To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/20180823T115208%2B0900-8428-0003.N0032439%40mail.gmail.com.

[侯建勇]

Dear Paul

I have edited the NN file and applied it to a seismic AI volume successfully.

Thank you very much!

Jianyong Hou

To unsubscribe from this group and stop receiving emails from it, send an email to users+unsubscribe@opendtect.org.

To post to this group, send email to us...@opendtect.org.
Visit this group at https://groups.google.com/a/opendtect.org/group/users/.

To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CABuvx8HjHnA-qXfoSPXap445P31T53wJCVJaBmdaAz6CK5ge5w%40mail.gmail.com.