KeyError raised in AkkarBommer2010SWISS01 for PGV about oq-engine

[micheles]

Seems to be the intended behavior, indeed the docstring of the class says

        :raises KeyError:
            If ``imt`` is not available in the table and no interpolation
            can be done.

[g-weatherill]

I don't think this was the intended behaviour of the GMPE, but perhaps the issue raised didn't provide a full enough explanation.

AkkarBommer2010SWISS01, AkkarBommer2010SWISS04 and AkkarBommer2010SWISS08 all inherit from AkkarBommer2010, which is defined as supported for PGA, SA and PGV. The sub-classes don't change the DEFINED_FOR_INTENSITY_MEASURE_TYPES attribute, so they are also indicted as being supported for PGA, SA and PGV. However, the Swiss subclasses use the rock adjustment factors here: https://github.com/gem/oq-engine/blob/master/openquake/hazardlib/gsim/akkar_bommer_2010_swiss_coeffs.py, which are only defined for PGA and SA, but not for PGV. The outcome of this is that OQ will pass the validation checks if you try to run calculations with them using PGV but then later crash out with theKeyError that occurs when it tries to get the PGV coefficients from these tables.

The fix is just to update DEFINED_FOR_INTENSITY_MEASURE_TYPES = {PGA, SA} in the subclasses.

[micheles]

Ah, thanks Graeme, now it is clear.