Access to additional variance components of GMM standard deviation models

[mmoschetti-usgs]

I don't think there is a straightforward way to access additional variance components (specifically, phiSS and phiS2S) from the GMM standard deviation models that compute these values. Are the OQ developers considering ways to evaluate and return these values in the GMM calculations?

[micheles]

Not at the moment. Saving such quantities would be hard and would require a lot of data storage, possibly slowing down continental scale calculations. If you are interested only in single site calculations, then there are no performance constraints; but then, you can just implement what you want at the level of hazardlib, without using the engine proper.
Maybe you can share some light on the use case you have in mind?