Best practice for extracting PGA + 1 Sigma from Scenario Hazard (Truncation Level = 1)

[Ridho Alfi Mubarok]

Hello everyone,

I am currently modeling a Scenario Hazard. My target output is a ground shaking map that represents the PGA + 1 Sigma boundary (84th percentile).

I have recently completed the simulation with the following parameters in my job.ini:

• number_of_ground_motion_fields = 1000

• truncation_level = 1

• spatial_correlation = yes

From this calculation, OpenQuake naturally generated a large set of GMF data. I have a few questions regarding the post-processing steps to obtain a single, final PGA + 1 Sigma map:

1. Is it methodologically valid if I just randomly select a single simulation (realization) out of the 1000 GMFs to serve as the hazard map?

2. Or is there another, more recommended post-processing method to map the PGA + 1 Sigma values from these scenario results given the parameters used?

I would highly appreciate any insights or advice on the best practices for this specific case. Thank you very much in advance!

[Anirudh Rao]

Dear Ridho Alfi Barack,

You cannot obtain the mean + 1 sigma map from the outputs of the scenario calculator. You would have to call the ground motion models using the OpenQuake hazard library in a short python script yourself. Please see previous discussions on this subject here: https://groups.google.com/g/openquake-users/c/B15byzkH2SI/m/3I1dgxMgFQAJ

Best regards,

Anirudh

[Ridho Alfi Mubarok]

Thank you very much for the clear explanation! I now understand that I cannot directly extract the mean + 1 sigma map from the scenario calculator outputs.

However, I have already completed a scenario calculation run using the parameters truncation_level = 1, number_of_ground_motion_fields = 1000, and spatial_correlation = yes.

Given that I already have these outputs, what is the proper way to process them?

1. Is it methodologically valid if I just randomly select a single realization (e.g., realization #12) out of the 1000 GMFs to be used as the scenario ground shaking map?

2. If picking just one simulation is considered scientifically incorrect, how should I best process these 1000 simulations (which are already truncated at the 1 sigma limit) so that the results are appropriate?

Thank you again for your time and guidance!

Best regards,

[Ridho Alfi Mubarok]

However, suppose I set truncation_level = 3 with number_of_ground_motion_fields = 500, and then I manually calculate the 84th percentile PGA (Median + 1 Sigma) statistically from all those scenario calculator outputs (the GMFs). Would this approach be methodologically valid?