How to interpret Solar Estimate Energy Today?

For me, pretty much every day looks like:

I really don’t understand the reason behind and how this even makes sense. My interpretation of a changing forecast through the day would be that weather forecast just wasn’t perfect. But that would never ever warrant a 10kWh to 50kWh swing … and these huge spikes!
Especially in my area where the weather is ridiculously predictable. A random cloud here and there wouldn’t make a difference.

Next, I would expect the curve to smoothly converge to a final value as more and more gets certain, the more we reach the end of the day.

And lastly, is the interpretation of the last datapoint that this is the “certain” estimate? Certain in the sense that no unknowns are left in the model: System parameters are known, the course of the sun (and hence angles) are known and the only unknown part, the weather is now also known. Again, why the heck this big spike from 17kWh to 45kWh and then back a the time when the sun has pretty much set and the bulk of the production would have happened already? So even if the weather turns 180 degrees, that would never ever create such a spike.

Anyone getting reasonable data out of this?
Or is this more pseudorandomly generated bogus data?

PS: I know these are not measurements but first order models are trivial, see PVWatts Calculator . I am NOT asking for accuracy to actual production .

You don’t mention which solar forecast integration you are using.

I am only aware of this one: Forecast.Solar - Home Assistant

Are there more?

There’s also Solcast (3rd party integration), GitHub - oziee/ha-solcast-solar: Solcast Integration for Home Assistant

There may be more.