[OpendTect_Users] MPSI_Plugin

[Renaud Meunier]

Dear Arnaud,

> Thanks for the information. I wonder what change in deterministic
> inversion parameters (scaler, LN error correction, well constraints,
> block size, inversion Z range) forces the user to redo the pre-processing?

As you know in MPSI you have two pre-processings, one for the
Deterministic Inversion (DI) and one for Stochastic Inversion (SI). You
need to re-compute the DI one each time the scaler, well constraints,
block size and inversion range changes. Now I would like to add that the
DI pre-processing (entirely optional) is designed to gain computation
time, however it is not always the case. It takes time to compute by
itself so can slow down things during the testing phase where you are very
likely to change your inversion settings. My advice to user will be not to
use this option at all when testing (on the fly computation when you are
testing the various inversion settings). When you are happy with the
settings and are very unlikely to change any of them (scaler, well
constraints, block size and inversion range) you might consider using this
option. However in this case also you need to take into account the
advantages disadvantages of doing so. Basically it will be useful if the
inversions matrices are long to compute (proportional to the number of
samples per trace so generally when you have more than 500 samples per
trace) as reading the matrix’ values from file is quicker than to compute
the matrices (for 500 samples traces there is no time gain).

I hope it clarifies things for you and our users.

Kind regards,

Renaud

-- 
Renaud Meunier
Geoscientist
Earthworks Environment and Resources Ltd Prestex House, Edison Road
Salisbury, Wiltshire. SP2 7NU

Direct Line: 01722 343999
Fax: 01722 343991
www.sorviodvnvm.co.uk

>   Dear Renaud,
>
> Thanks for the information. I wonder what change in deterministic
> inversion parameters (scaler, LN error correction, well constraints,
> block size, inversion Z range) forces the user to redo the pre-processing?
>
> Best regards,
> Arnaud Huck.
>
> -- Senior Geoscientist
> -- dGB Earth Sciences
> -- Nijverheidstraat 11-2, 7511 JM Enschede, The Netherlands
> -- mailto: arnaud.huck at dgbes.com, http://www.dgbes.com
> -- Tel: +31 53 4315155 , Fax: +31 53 4315104
>
>
> On 07/16/2010 10:07 AM, Renaud Meunier wrote:
>> Dear E. Falebita,
>>
>> One thing I forgot to mention is that when the wavelet scalar has been
>> properly estimated you can use the same value everywhere (using
>> different
>> well constraints, using a filtered initial model, doing a stochastic
>> inversion
). Basically you only have to do the job once for a given
>> seismic and wavelet.
>>
>> I hope it clarifies things.
>>
>> Best regards,
>>
>> Renaud.
>>
>
>
> _______________________________________________
> Users mailing list
> Users at opendtect.org
> http://lists.opendtect.org/mailman/listinfo/users
>