A_95 Parameter Calculation issue

03-Unexpected behaviour/errors
Post Reply
Wael.Aloqaily
Posts: 1
Joined: 25 Sep 2020, 16:28

A_95 Parameter Calculation issue

Post by Wael.Aloqaily »

I will appreciate your help on the following issue regarding A_95 parameter calculation.

I am wondering how SesimoSignal calculates A_95 parameter. As I tried to verify it with the following simple step-signal example:
a = 0 for t = [0,1)
a = 1g for t =[1,2)
a = 0 for t = 2
where "[" means including, "(" means excluding, a is acceleration and t is the time.

According to my simple hand calculation, and considering the area under a^2 since pi and 2g are constants in Arias intensity formula:
The area under a^2 is = 1.0.
To achieve 95% of that area, the acceleration threshold required = sqrt(95)*1 = 0.974679434g.

However using Seismsignal for this example, the value reported for A_95 is 0.99750g.
If we calculate the corresponding area = (A_95)^2*t = (0.9975)^2*(1) = 0.99501, which correspond to 99.5% of Arias intensity.

Thank you in advance!
User avatar
seismosoft
Posts: 1012
Joined: 06 Jul 2007, 04:55

Re: A_95 Parameter Calculation issue

Post by seismosoft »

The small error (0.974679434 vs 0.975) is probably due to the fact that the arias intensity is checked in steps of 1%. If you check the table in the Ground Motion Parameters with the Arias Intensity build-up at t=1.95 the Arias Intensity is 95.04950%, which gives this small difference with your hand calculations.
In any case note that your simple example is not ideal for checking the A95 parameter, due to the abrupt change in the acceleration. A triangular shape for the acceleration (with a gradation of the acceleration values) is probably better suited for these kind of checks.
Seismosoft Support
Clavin
Posts: 1
Joined: 04 Dec 2021, 16:53

Re: A_95 Parameter Calculation issue

Post by Clavin »

Hello.

I have doubts regarding the calculation of the A95, I was comparing some acelograms used in the study by Sarma and Yang (1987) that I could find in the PEER. But the A95 parameter in the SeismoSignal is different from the one shown in the Sarma and Yang article database.

Taking as an example the Ferndale earthquake (S44W) of 12-21-54 in position 22 of the Sarma and Yang article (page 130), and in the PEER in position 20. Its A95 is 0.1278 g, but the A95 value of SeismoSignal is 0.16133 g, presenting a difference of 26.24%.

As you can see, the difference is noticeable, and I want to know if this could be caused by the lack of accuracy of the program or I was missing some configuration.
User avatar
seismosoft
Posts: 1012
Joined: 06 Jul 2007, 04:55

Re: A_95 Parameter Calculation issue

Post by seismosoft »

The A95 parameter has been included in SeismoSignal for almost 20 years now and it has been tested quite well for several users (including ourselves). Without obviously being 100% sure, we would be rather surprised to be honest, if there was a bug in its calculation (after all the definition of the parameter is quite simple).
Furthermore, we cannot check the specific calculations presented in the paper with just the value of the A95 parameter, we are unaware of the exact records that have been used and how these had been processed (e.g. if they have been filtered or baseline corrected etc).
In any case, if you have some doubts regarding the calculations of a specific record (e.g. you get different values according to your calculations), please send it to support@seismosoft.com so that we can have a look at it.
Seismosoft Support
User avatar
seismosoft
Posts: 1012
Joined: 06 Jul 2007, 04:55

Re: A_95 Parameter Calculation issue

Post by seismosoft »

From the file that you sent us we have been able to spot the bug and fix it in SeismoSignal v2022 Release-1 build-20. Note that existing v2022 licenses are valid for this new build.
Thanks for helping us fix the bug, and apologies for any trouble that this caused to you and your work.
Seismosoft Support
Post Reply

Return to “03-Unexpected behaviour/errors”