PSA versus SA

[emthompson-usgs]

An issue that has come up for me is that PSA and SA are not the same thing. As a community, we have been treating them as interchangeable for a long time. For small damping values, the difference is not a big deal, but it is not trivial at longer periods (~7 sec and greater, depending on the level of precision that you care about). If larger dampings are ever used, the differences are much greater and extend to shorter periods. There is no consistency among GMMs in terms of using PSA vs SA. For example, all of the NGA models use PSA, but the ESM dataset uses SA, and so any GMMs derived from the ESM dataset use SA. It would be helpful to acknowledge that these IMTs differ by adding PSA as an option and tracking its use via the DEFINED_FOR_INTENSITY_MEASURE_TYPES class attribute.

This paper derives conversion equations between PSA and SA:
https://zhangyamagata.sakura.ne.jp/wp/wp-content/uploads/2025/04/2025_7.pdf
The figures are also helpful in understanding the differences. It may even be useful to implement these conversion equations so that GMMs with different choices of PSA vs SA can be combined in the same logic tree.

[CB-quakemodel]

Hi Eric,

We will need to take some time to think about a way to pragmatically implement a means of distinguishing between these two often interchangeably-treated intensity measures.

I have not thought in detail too much about it yet, but I'm fairly confident there will be some quite substantial implications for this change that we will need to carefully accommodate within the engine. Once we find some time (currently we are focused on the 2026 Mosaic release) we will for sure look into addressing this issue. Thanks for providing the paper containing the conversion equations too.

Thanks,

Christopher