Is this the next opportunity for this integration?

[BJReplay]

Down near the bottom of this article is an interesting comment:

The progress of a cloud over a city could be tracked by successive drops in generation from the rooftop panels far below.

This high-resolution data could be used to improve solar forecasts.

"That's definitely on the road map," Dr de Hoog said.

This integration has around 20,000 users.

I can see a scenario where a user of the integration could opt in to reporting real time solar isolation back to Solcast (perhaps in exchange for an increase in the hobbyists quota from say 10 to 20 calls per day).

What do you think, @billy-solcast?

[billy-solcast]

It's interesting what they say, and I can definitely see the benefits. We've also had some home users reach out to us about it as well.

We've had this discussion internally in the past, and I believe (just before my time) they did a small trial to assess and review the results. Unfortunately I believe the results were a net negative. Now this was a while ago, so improvements of hardware and models, and a lot more solar rooftops likely changes things.

I'll add this to the list of improvements items for discussion. I won't be able to provide any guarantees beyond that unfortunately.

[BJReplay]

I won't be able to provide any guarantees beyond that unfortunately.

Understood, of course.

You've got my direct email, so reach out if necessary, or via this repo if it is of interest.

When combined with auto-dampening (where the integration learns about local shading, and adjusts the local forecast for discovered shading), then the returned data may be of more use; if the returned data was less helpful than expected, but that was partly because local rooftop data was shaded by trees rather than clouds, then you're not getting a real-time cloud signal, you're getting a fixed shading (dampening signal).

Given that we can return data such as worst case and applied dampening - indicating what the system is detecting as local shading (or, in my case, right now "mud" on the panels as a result of a dust storm followed by light rain, and I haven't yet cleaned them), then you're looking for two things: short term drops in output which indicate a cloud, and that moving over your "array" of sites to provide real time cloud information, as well as output below the dampened forecast, which indicates heavier than forecast cloud.