From mito5ra@gmail.com Fri May  4 02:12:24 2007
Date: Thu, 3 May 2007 23:11:49 -0700
From: Masahiro Ito 
To: Keith Riles 
Subject: Re: Question on 99%ile seismic intervals

Hi Keith,

I did change the time interval based on the new result.

The time stamps in the new list most likely have +60 sec from the old list.
The time offset I used in the old list was determined by the time shifted
coincidence analysis between EY_SEISZ 3-10Hz and AS_Q 5-7Hz minute trend
data (http://physics.uoregon.edu/~mito/S5/060907/
eyseis_vs_h1asq_timeshift/eyseis_vs_h1asq_timeshift.html). The scatter plots
showed that -60sec time offset was slightly coincided better than the data
with no offset.

I did the similar analysis using EY_SEISZ 3-10Hz minute trend and DAMR_ERR
5-7Hz triggers for entire S5 last March (and yesterday). The reason for
using 5-7Hz triggers was that they were available for entire S5 and the
frequency band represented the truck traffic event. The set of scatter plots
(see the following links) showed that 0 offset was better choice than -60
sec. I used this new result for the updated list.

EY_SEISZ vs DARM_ERR 5-7Hz trigger:


EY_SEISZ minute trend vs AS_Q 70-110Hz minute trend:
http://physics.uoregon.edu/~mito/S5/070501/070501.html

EY_SEISZ minute trend vs AS_Q 4-12Hz minute trend:
http://physics.uoregon.edu/~mito/S5/070502/070502.html

Masahiro Ito
University of Oregon


On May 2, 2007, at 8:10 AM, Keith Riles wrote:

> Hi Masahiro,
>
> I'm afraid I've been very slow about inserting the
> seismic intervals you sent me into the DQ database,
> but my semester finally ended, and I'm catching up
> on old tasks.
>
> I started to insert your 99%ile EX 3-10 Hz intervals
> today, but I was troubled to see that the overlap
> between the new list of intervals and the old list
> of intevals for early S5 data is not very high. Specifically,
> in many cases the end-times for the original list
> from early October 2006 coincide with the start-times
> from March 2007. For example, in October 2006 you
> has these first three intervals:
>
> 815412840 815412960
> 815494680 815494740
> 815495340 815495400
>
> But now you have:
>
> 815412900 815413020 120
> 815494740 815494800 60
> 815495400 815495460 60
>
> There is some overlap for the first interval, but none for
> the next two, which seem typical of other intervals in the
> files. I'm attaching the old and new versions of your interval
> files to this e-mail.
>
> I wondered if you had discovered and fixed a problem
> with the definition of the start time that had been off
> by one minute previously?
>
> I assume I should use your latest list and have it supersede
> the previous list, going back to the start of S5, but wanted
> to double-check with you first.
>
> You and I exchanged e-mails about a similar issue on October 24, 2006
> concerning some incorrect durations in which (I think) you confirmed
> the initial set of intervals above.
>
> Could you double-check to make sure a bug hasn't entered into the
> latest calculations of the intervals?
>
> thanks very much,
> Keith
> 
>