Add clearsky index function to pvlib.irradiance

[kevinsa5]

The clearness index kt (the ratio of surface-level GHI to ET irradiance) can be calculated via pvlib.irradiance.clearness_index and is used in the DIRINT and DISC models. Though other metrics in the same vein are not necessary for the irradiance models implemented in pvlib, including them in pvlib could be a convenience to users. Is there interest in implementing other irradiance metrics for users to access? Some examples:

• clear-sky index: the ratio of surface GHI to modeled clear-sky GHI
• Variability Scores as described in this Sandia paper
• Variability Index as described in this Sandia paper, coauthored by Cliff Hansen

[cwhanse]

I think clearsky_index is needed. Would we pass the clear-sky GHI as an argument, or the name of a clear-sky model? We'd have to agree how to deal with low irradiance conditions, e.g., an upper bound on clearsky_index.

I think the variability calculations are interesting and would be worth adding as functions to the library.

[kevinsa5]

Considering clearsky_index as a convenience function, it seems appropriate for it to accept an optional clear-sky model name rather than clear-sky GHI data -- if the user already has both measured and modeled data, then there's not much left to do besides the division anyways. The way that clearness_index handles low-light conditions seems appropriate here as well.

If there is no available bandwidth from the core team for this, I am happy to work up some PRs. Though I cannot promise they would be ready any time soon... free time is always in short supply.

[wholmgren]

I am all for including these functions. Usually the best approach is to implement very simple functions that implement the core algorithm, then add convenience functions/methods around them. So, for clearsky index, I'd want a function that doesn't do any more than the division and limits (much like clearness_index). That API would be straightforward and can be integrated into many modeling workflows. Consider something like Location.clearsky_index if you want the convenience of computing clear sky data within the function/method call.

Please do submit PRs!

[cwhanse]

Closed via #571. If we want to add the variability index, let's open a new issue.