Importing/exporting GMF

[Clotaire Michel]

Dear developers,

I am currently running scenario damage computations. 

The sequence:

openquake --run-hazard job_hazard.ini

openquake --run-risk job_risk.ini --hazard-output-id ID

openquake --export-risk ID ./results

leads to completely different results (order of magnitudes) from 

openquake --run-hazard job_hazard.ini

openquake --export-hazard ID ./results

openquake --load-gmf ./results/calc_ID/gmf_scenario/gmf_scenario.xml

openquake --run-risk job_risk.ini --hazard-output-id ID2

openquake --export-risk ID ./results

The results of the second are clearly not those expected.

Am I doing something wrong when importing the GMF into openquake?

Bonus question: what are the units of the produced GMFs (GMPE=ChiouYoungs2008 in SA(T))? I guess g, but I could not find this information anywhere.

Thanks for the answer

[Clotaire Michel]

Update:

Trying to re-export the re-imported GMF leads to the error message:

openquake@ubuntu-64:/media/sf_vbox/scenario_damage_test2$ openquake --export-hazard 130 ./results

Traceback (most recent call last):

  File "/usr/bin/openquake", line 9, in <module>

    load_entry_point('openquake.engine==1.0.0', 'console_scripts', 'openquake')()

  File "/usr/lib/python2.7/dist-packages/openquake/engine/bin/openquake_cli.py", line 473, in main

    export_hazard(output_id, expanduser(target_dir), args.export_type)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/bin/openquake_cli.py", line 313, in export_hazard

    export(hazard_export.export, haz_output_id, target_dir, export_type)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/bin/openquake_cli.py", line 350, in export

    the_file = fn(output_id, target_dir, export_type)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/export/hazard.py", line 65, in export

    return export_fn(output, target)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/export/core.py", line 55, in wrapped

    return fn(output, target, **kwargs)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/export/hazard.py", line 354, in export_gmf_scenario_xml

    writer.serialize(gmfs)

  File "/usr/lib/python2.7/dist-packages/openquake/nrmllib/hazard/writers.py", line 884, in serialize

    node.node_to_nrml(gmfset, dest)

  File "/usr/lib/python2.7/dist-packages/openquake/nrmllib/node.py", line 668, in node_to_nrml

    node_to_xml(root, output)

  File "/usr/lib/python2.7/dist-packages/openquake/nrmllib/node.py", line 630, in node_to_xml

    w.serialize(node)

  File "/usr/lib/python2.7/dist-packages/openquake/nrmllib/writers.py", line 85, in serialize

    self.serialize(subnode)

  File "/usr/lib/python2.7/dist-packages/openquake/nrmllib/writers.py", line 84, in serialize

    for subnode in node:

  File "/usr/lib/python2.7/dist-packages/openquake/nrmllib/hazard/writers.py", line 324, in gen_gmfs

    for gmf in gmf_set:

  File "/usr/lib/python2.7/dist-packages/openquake/engine/db/models.py", line 2300, in get_gmfs_scenario

    imts = distinct(from_string(x) for x in hc.intensity_measure_types)

TypeError: 'NoneType' object is not iterable

[Michele Simionato]

We are aware of the problem and working to solve. I will let you know when we have a fix.

[Michele Simionato]

I have a branch where the problem is fixed. However the fix was rather nontrivial and involved a change in the database.

I need to write a migration script and then there will be a scrupulous code review. Please expect the fix to be released next week.

[Michele Simionato]

Sorry for the late reply. The problem was actually fixed one week ago and it has been available in a nightly builds since them.

[Clotaire Michel]

Dear Michele,

thanks for your reply and update of the code. I uploaded the Virtual machine from October 20th. However I got the following error when I tried to upload an exported GMF. I understood that some attributes of the xml file gmf_scenario have been added, but is seems that the attribute ruptureId is not correctly filled during the export (always the same value), which leads to the error when it is uploaded again. My file has actually 1 GMF but 5 different intensity measures.

thanks

clotaire

openquake@ubuntu-64:/media/sf_vbox/scenario_risk_test2$ oq-engine --load-gmf ./results/calc_2/gmf_scenario/gmf_scenario.xml

Traceback (most recent call last):

  File "/usr/bin/oq-engine", line 9, in <module>

    load_entry_point('openquake.engine==1.2.0-1-dev1413752954-8ed039d', 'console_scripts', 'oq-engine')()

  File "/usr/lib/python2.7/dist-packages/openquake/engine/bin/openquake_cli.py", line 517, in main

    out = import_gmf_scenario(f)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/tools/import_gmf_scenario.py", line 161, in import_gmf_scenario

    imts, tags, rows = read_data(fileobj)

  File "/usr/lib/python2.7/dist-packages/openquake/engine/tools/import_gmf_scenario.py", line 83, in read_data

    raise DuplicatedTag(tag)

openquake.engine.tools.import_gmf_scenario.DuplicatedTag: scenario-0000000000

[Michele Simionato]

Ah, this is a stupid bug. As you understand, the problem is that you have more than one IMT in your file while in my test file I have a single IMT. The fix is trivial and I will improve the test too. The fix will be available for tomorrow.