Record Length Extension

04-Future developments & requests
Post Reply
huffte
Posts: 832
Joined: 22 Jul 2011, 10:19
Location: Cookeville, Tennessee, USA
Contact:

Record Length Extension

Post by huffte » 15 Oct 2011, 18:48

I have been unable to load some Japanese records - ones with 60,000 points - into SeismoSpect. I get a 'range check error' and I think it s a limitation on the number of points which SeismoSpect is able to handle in a single record.

I am not certain if this is a processor-based limitation or a SeismoSpect coding-based limitation.
Tim Huff

User avatar
seismosoft
Posts: 567
Joined: 06 Jul 2007, 04:55

Re: Record Length Extension

Post by seismosoft » 16 Oct 2011, 10:35

Thanks for this note.

Yes, the internal algorithm of SeismoSpect (and SeismoSignal) does currently feature a 32000 data points limitation. We hope to be able to overcome such constraint in future versions, though it is not as straightforward as one would have imagined or desired.

Having said the above, the 'range check error' message provided to the users should certainly be replaced with something more informative and user-friendly.

Thanks again,

Seismosoft Support

huffte
Posts: 832
Joined: 22 Jul 2011, 10:19
Location: Cookeville, Tennessee, USA
Contact:

Re: Record Length Extension

Post by huffte » 09 Nov 2011, 03:51

Is there a time-frame planned for extending the record length limitation in either of the programs? I am in immediate need of having the capability to analyze records having 37500 points and cannot find anything comparable to SeismoSoft.
Tim Huff

User avatar
seismosoft
Posts: 567
Joined: 06 Jul 2007, 04:55

Re: Record Length Extension

Post by seismosoft » 13 Nov 2011, 23:47

Dear huffte,

We cannot implement this before mid 2012, since it requires the complete re-structuring of the entire memory management of the program, and we currently do not have the resources to do it, we are afraid.

Seismosoft Support


huffte
Posts: 832
Joined: 22 Jul 2011, 10:19
Location: Cookeville, Tennessee, USA
Contact:

Re: Record Length Extension

Post by huffte » 14 Nov 2011, 05:39

Thank you for the information seismosoft.

For others having this issue, you may be able to either:
1. truncate the files at the beginning and/or end, or
2. eliminate every other data point and double the time step.

Read in the first 32000 points and observe the initial portion which may be negligible. Then read in the last 32000 points to see which time frame might be negligible at the end. You may be able to zero in on the significant shaking this way.

And manipulating the file with a text editor to eliminate every other point - then doubling the time step - is an option as well.

So this is how I have handled it and will contine to do so for the near future.

Thanks again.
Tim Huff

User avatar
seismosoft
Posts: 567
Joined: 06 Jul 2007, 04:55

Re: Record Length Extension

Post by seismosoft » 02 Apr 2012, 11:46

Dear huffte,

Whilst the forthcoming release will now cater for the introduction of records with up to 260.000 data points, the use of such large accelerograms may give rise to very slow analyses and system memory management difficulties, as discussed in http://www.seismosoft.com/forum/topic.asp?TOPIC_ID=562 .

Hence, we would still suggest for users to follow your wise suggestion and instructions for truncating such long time-histories, removing their typically negligible initial and closing stages.

We note also that doubling of the time-step can be carried out automatically by the software, avoiding the need for manual operations using a text editor of spreadsheet.

Many thanks for all the good input,

Seismosoft Support

Post Reply