Mean hazard curves

[Nitin Gupta]

Hi,
I am using openquake engine to generate hazard curves from SHARE data. Is it possible to generate arithmetic mean hazard curve instead of default where hazard curve for each logic tree branch is generated? I am using number_of_logic_tree_samples =0.

Thanks,

 Nitin

[Nitin Gupta]

Hi,
I have been still trying to figure out the answer to the question. Using the SHARE data provided at http://www.efehr.org:8080/jetspeed/portal/hazard.psml how can I generate mean(arithmetic) hazard curves/maps? Or do I need to create a script that will compute mean after combining hazard curves from each logic tree branch.

I believe SHARE Euro hazard curves/maps available from http://www.efehr.org:8080/jetspeed/portal/HazardMaps.psml are generated using teh OpenQuake NRML data given on the website above? Once hazard maps are computed is there a way I can download the actual hazard for each grid location in ascii format ?

Thanks

 Nitin

​

[Marco Pagani]

Hi Nitin,

The example you find at the link below explains how to configure the
parameters in the 'output' of the OQ-engine config file.

https://github.com/gem/oq-engine/blob/master/demos/hazard/LogicTreeCase2ClassicalPSHA/job.ini#L41

I don't think the SHARE model as available on their website is organised
in a way to get mean results directly from the OQ-engine.

You can either reorganise the input or post-process the results of
independent calculations in order to get mean mean hazard.

Regards,

Marco

> --
> You received this message because you are subscribed to the Google
> Groups "OpenQuake Users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openquake-use...@googlegroups.com
> <mailto:openquake-use...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout.

[Nitin Gupta]

Thanks Marco. I was able to reorganize the input using OQ-engine. I will update my config file using the example you provided and re-run (once my OATS account is enabled) my computations.

Thanks

Nitin

To unsubscribe from this group and stop receiving emails from it, send an email to openquake-use...@googlegroups.com.

[Marco Pagani]

Hi Nitin, the OATS account shouldn't be used for heavy calculations but just for appraising the capabilities of the OQ-engine. I suggest to install the OQ-engine locally. 

Regards,

Marco

[Laurentiu]

Dear Nitin, 

The mean hazard curves obtained from the EFEHR () portal represent the weighted(arithmetic) mean.

Each hazard curve resulted as an aggregation of thee independent source models:

- area source model (As Model)

- faults + background source model (fsFb Model)

- smoothed seismicity (SEIFA Model)

The computation was performed per tectonic regions (i.e. Active Shallow Crust, Oceanic Crust, Stable Continental, Shield, Volcanic, Subduction Inslab, Subduction Interface, Deep Seismicity -Vrancea Region ); where each GMPE of the corresponding ground motion logic tree represents and individual end-branch. 

These individual end-branches were summed up to form the tectonic-mean hazard curves, in turn these tectonic-mean hazard curves were summed up to generate the source mean-hazard curves; finally the three source mean-hazard curves were weighted to generate the final mean model (available online at efehr.org) 

The hazard curve aggregation was done by operating on the exceedance rates rather then probabilities. This is different than the OQ mean hazard curves, that is based on probabilities summation. 

Also, if you want to replicate the calculation, you need to represent each source model into individual tectonic source models. Depending on your hardware capability you might want to try a single site computation with the entire SHARE source model. If you use this option make sure, you consider the appropriate option of the 

new configuration setting of OQ (ver1.0 and above)

individual_curves = false

If this is set to false, OQ will not output the individual end-branches results. This option was not implemented at the time of the SHARE computation (Spring 2013).

I hope this helps, Cheers Laurentiu

 

[Michele Simionato]

On Tuesday, January 13, 2015 at 8:40:44 AM UTC+1, Laurentiu Danciu wrote:

Dear Nitin, If you use this option make sure, you consider the appropriate option of the 

new configuration setting of OQ (ver1.0 and above)

individual_curves = false

There is typo here, the version is 1.2 and above. Version 1.0 is now obsolete (and does not contain that feature). 

[Nitin Gupta]

Thanks Marco. Knowing the constraints of OATS account I did get the OQ 1.2 installed locally.

Thanks Laurentiu for the detailed explanation. I don't understand what do you mean by representing each source model into individual tectonic source model? If I understand correctly EFEHR does provide each individual source models in OQ format (NRML files) where each individual source has a defined tectonic region type , based on which correct set of GMPEs, with appropriate weight, can be applied. I mean the hazard maps on EFEHR website not produced from OQ formatted files, downloaded from http://www.efehr.org:8080/jetspeed/portal/hazard.psml.

Also in regards to exceedance rates - What formula were you using convert the exceedance prob to exceedance rates ? You were not only applying source model weights but GMPE weights as well on the exceedance rates ?

Thanks

 Nitin

--

[Laurentiu]

Hi Nitin, 

That's right, the hazard maps on the EFEHR website were not produced by OQ. We operated mainly on the hazard curves, as obtained per individual end-branches and aggregated in a post-processing. The end branches were mainly corresponding to a mean source model, and the corresponding GMPE linked via the tectonic settings. If you have enough computational power, then you might be able to re-run the entire model. I would expect to a slight difference of 10% when compared with the OQ output, for maps at very short return periods (<100Yrs) due to the rates/probabilities conversion. For this conversion I have use the following formulas: 

To convert probability to rate 

r = - [ln (1 – P)] / t

To convert rate back to probability 

p = 1 – exp {-r*t} 

where rate (r) , time period (t)i.e 50years, probability of exceedance [P].

I hope this helps, cheers. Laurentiu

[Nitin Gupta]

Laurentiu,

Thank you for your clarification on the SHARE OQ data. Since I am working with OQ data available from EFEHR website and would like to re-produce SHARE hazard curve/map results, if I use OQ v1.2 with new configuration settings to generate single mean hazard curve (exceedance prob) and convert to rate, results should be similar.  

Thanks

 Nitin

[Nitin Gupta]

I want to make sure it is OK with OpenQuake group and SHARE group if I share hazard curves and maps comparison between SHARE results downloaded from http://www.efehr.org:8080/jetspeed/portal/hazard.psml and ones produced from most recent OQ1.2 engine using SHARE OQ inpute files ?

Thanks

 Nitin

[Marco Pagani]

Hi Nitin,

Thank you for your email. If you think that discussing this comparison
can help the community - from our perspective - you can certainly share
your results.

Thank you and best wishes,

Marco

> /*r = - [ln (1 – P)] / t*/

> To convert rate back to probability

> */p = 1 – exp {-r*t}/*

> <mailto:michele....@gmail.com>> wrote:
>
>
>
> On Tuesday, January 13, 2015 at 8:40:44 AM UTC+1,
> Laurentiu Danciu wrote:
>
> Dear Nitin, If you use this option make sure, you
> consider the appropriate option of the
> new configuration setting of OQ (ver1.0 and above)
> individual_curves = false
>
>
> There is typo here, the version is 1.2 and above.
> Version 1.0 is now obsolete (and does not contain that
> feature).
>
> --
> You received this message because you are subscribed to
> the Google Groups "OpenQuake Users" group.
> To unsubscribe from this group and stop receiving emails
> from it, send an email to
> openquake-use...@googlegroups.com

> <mailto:openquake-use...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> You received this message because you are subscribed to the
> Google Groups "OpenQuake Users" group.
> To unsubscribe from this group and stop receiving emails
> from it, send an email to
> openquake-use...@googlegroups.com

> <mailto:openquake-use...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> You received this message because you are subscribed to the
> Google Groups "OpenQuake Users" group.
> To unsubscribe from this group and stop receiving emails from
> it, send an email to
> openquake-use...@googlegroups.com

> <mailto:openquake-use...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> You received this message because you are subscribed to the Google
> Groups "OpenQuake Users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openquake-use...@googlegroups.com

> <mailto:openquake-use...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.

--

*MARCO PAGANI* | Seismic hazard team coordinator | skype mm.pagani | g+
marco....@globalquakemodel.org | +39-0382-5169863

*GLOBAL EARTHQUAKE MODEL * working together to assess risk

*GEM -* globalquakemodel.org <http://www.globalquakemodel.org> | *T -*
@GEMwrld <http://twitter.com/GEMwrld> | *F -* GEMwrld
<http://www.facebook.com/GEMwrld>