Hello Jens,

Today, I would like to show that I cannot find a way to
generate weights combining both
1. Phase Picking weights
3. Bisquare weighting
"Weights are multiplied together" according to the U.G.

I have used a real case but where picking weights were
set to 0 or 3. In order to use different combinations of
parameters, I join STATION 1,2 and 3 .HYP
where the only difference is the value of the bisquare
residual parameter Test(36).

>From hyp.out:

Case   Test(36)    Event RMS   Obs W    Bisquare W

1           0.00            0.7             Yes            No
2           0.80            0.6              No            Yes
3           0.50            0.7             Yes            No

This table shows that from the supplied parameters,
the picking weights are only taken into account
when bisquare weighting is not active.

In other words, hypocenter never multiplies Phase
Picking weights (1.) with Bisquare Weights (3.) as stated
in the User's Guide.

If there is a way to do it, I would be grateful to know how.

Best regards and many thanks for the help.

Freddy.


________________________________________
From: Jens Havskov [[email protected]] On Behalf Of Jens Havskov 
[[email protected]]
Sent: Sunday, October 07, 2012 10:20 PM
To: Dr.Freddy Aldersons
Cc: [email protected]
Subject: Re: [seisan] Local events: How does the weighting really work in 
Hypocenter ?

freddy

i am not sure about bisquare weighting, seem to remember that barry
lienert did not think it should be used for local events. however weights
0-4 and 9 and distance weighting works according to my exprience, maybe
you could give an exmaple ?

jens

On Thu, 4 Oct 2012, Dr.Freddy Aldersons wrote:

> Hi,
>
> in addition to my question of yesterday regarding
> the combined use of bisquare weighting and distance weighting,
> I'd like to ask some clarification regarding how weights are applied
> in Hypocenter for local events.
>
> In the Hypocenter main User's Guide (pp. 26-27), three weighting
> variables are potentially combined together:
> 1. User specified weights (observational weights 0-4 Hypo71 style)
> 2. Distance weighting
> 3. Bisquare weighting
> "Weights 1.-3. are multiplied together ..."
>
> What I see from my use of Hypocenter for local events is that phase
> observational weights (1. above) are never used, and that distance
> weighting does not work properly together with bisquare weighting.
>
> The difference between observed and predicted times seems to largely
> dominate the process. This is fine but I would like also to see the 
> possibility
> that phases with a weight of 0 would be given a higher weight than
> those with a weight of 3 and this is something I just can't obtain.
>
> In other words, the statement of the U.G. "Weights 1.-3. are multiplied 
> together ..."
> is not what I observe in practice.
>
> Is is just a question of switches or is there something else ?
>
> Best regards,
>
>                      Freddy.
>
>
> _______________________________________________
> seisan mailing list
> [email protected]
> http://mailman.uib.no/listinfo/seisan
>

Attachment: hyp.out
Description: hyp.out

Attachment: STATION1.HYP.0
Description: STATION1.HYP.0

Attachment: STATION2.HYP.0
Description: STATION2.HYP.0

Attachment: STATION3.HYP.0
Description: STATION3.HYP.0

_______________________________________________
seisan mailing list
[email protected]
http://mailman.uib.no/listinfo/seisan

Reply via email to