And here’s the sunshine_duration, which seems to be linked to the irradiance values. Anything else would be surprising:
I’m sure that cloud cover is not used directly for calculating solar radiation on the Open-Meteo side. Of course, it goes without saying that there will be some kind of correlation.
Also keep in mind that cloud cover is instantaneous and irradiance is skewed by 15mins for instantaneous
Just wanted to add that, of course, we are talking about predictions/estimations here. These are never 100% correct, sometimes far from it.
BUT: I have been using SolCast for 9 months now and its predictions were usually quite accurate. They have been that good that I don’t tend to look at weather forecasts anymore but just check the expected overall daily yield and curve to know whether it’s going to be a nice/sunny day or whether I can expect clouds or even rain. The curve over the day provides temporal information after all.
And so I am less concerned about getting 100% correct predictions but about getting predictions that are close to SolCast’s. And here I saw huge deviations with the cloud code. The latest version works better although there is still some underestimation going on.
Thanks for the feedback, I think I have a rough idea as to why there’s an underestimation. When I deal with the underestimation, I’ll begin working on improving cloud compensation code.
1 Like
Now it’s underestimating by about 4kWh for today and tomorrow.
And now I have the rate limit problem again. Strange because the IP addresse doesn’t change that often and I’ve never seen this for SolCast (which has an even stricter rate limit of only 10 API calls per day).
I have a problem setting it up. When I add the integration via the link to HACS, I am only asked for the location. The integration only shows me the weather for the next week. No data on the probable solar production. What is my mistake? Thanks for the help.
At a guess you’ve setup the Open Meteo Weather integration not this custom integration for the solar forecast…
Read back here Open-Meteo Solar Forecast, then install this integration via hacs, reboot HA then setup and you’re good to go…
You were right. I set up the wrong integration. Thanks for the quick help. Now it works but I have the next question. In the energy dashboard I can see the production for the next few days. But in the integration I can only see the production for today and tomorrow. Is there any way to display the production for the next few days as an entity? Thanks
Not yet, but adding this is fairly straightforward.
1 Like
+1
that would be very nice. Would need this for my Apex Chart 
I’m currently trying this integration too.
I have a few questions:
- What does an Api Key bring? I didn’t enter any. Does that have an impact?
- Why can I only enter full numbers for the solar power? I have 6.3 and 3.5 for my modules power
- If I have 2 configurations (East/West), can I add the power together or do I have to build my own power sensors for this?
Thanks
Add two instances of the integration, one for east and one for West, with the respective power amount in each. You need to convert your kW to W, so 6300 and 3500.
For the energy dashboard, you can then choose both and HA will sum them. If you are using for other purposes then yes you may need to create your own combined template sensors.
1 Like
Ok thanks. I will try that
As you’ve already received responses for the last two questions, I’ll answer your API key one.
Basically this for the individuals that are paying for Open-Meteo’s commercial use license: https://open-meteo.com/en/pricing. It might also make sense if you’re self hosting Open-Meteo and want to protect it from outside use by requiring an API key.
1 Like
This works for the production and forecast, but for the peaktime i have no clue how to combine to dates
ha-open-meteo-solar-forecast v0.1.3 has been released. Wind speed and cell efficiency is now taken into account for cell temperature calculation in an effort to resolve the underestimation issue.
ha-open-meteo-solar-forecast v0.1.4 has been released. A typo in cell temperature calculation was fixed so be sure to update to that, however the impact should be minor.
ha-open-meteo-solar-forecast v0.1.6 has been released. It fixes the start time for power sensors, clarifies what efficiency factor is for and adds D{2…7} energy production sensors.
Looks good, pretty good results, a bit overestimating for me today, though, for tomorrow as well.
No need to get the cloud covering back into the calculation?
You might need to adjust your efficiency factor to fit the data after this change. I’m not still not sure about how to incorporate cloud cover given that different users have different experiences with it.