Unusual azimuth check

[WyndStryke]

The azimuth for site f524-9f14-f613-e077 is unusual for the latitude of this northern hemisphere rooftop. Should this be -92 instead of -88? (Ignore the issue if the site is facing towards North.)

Just some random thoughts about this check. Not raising as an issue because it isn't. I'm in the Northern hemisphere so adjust accordingly.

Firstly, perhaps the Azimuth check could be a little less sensitive? East/west is a common setup for arrays, and -92 is only 2 degrees towards north, it's not facing north in any practical way. Solcast itself displays up to -68 as East, then -67 onwards as North-east. So perhaps that could be the dividing line (alternatively, the E to ENE transition at -79)?

Secondly, perhaps the pitch could also be taken into account. Very shallow northerly roofs work very well, much the same as flat roofs. Although picking a specific pitch would be arbitrary.

Thirdly, perhaps the azimuth could be translated into S/SSE/SE/ESE/etc. If they've misunderstood the way Solcast does azimuths in the first place, just seeing the number isn't going to help on it's own, a translation would be helpful.

The azimuth for site f524-9f14-f613-e077 is unusual for the latitude of this northern hemisphere rooftop. Should this be -92 (E) instead of -88 (E)? (Ignore the issue if the site is facing towards North.)

Fourthly, perhaps if they have matching arrays on both sides of the roof (S and N for example), then this might indicate that they do understand the azimuth, and it's unlikely to be a mistake. In my case it's -88 (east) and 92 (west), with a 23 degree tilt.

'Ignore' is of course the obvious button to click so this isn't a usability concern.

Just to reiterate, none of this is an issue.

[autoSteve]

All of this has happened before, and all of it will happen again, @WyndStryke.

This is a perfect opportunity for you to check something for me before ignoring the issue raised. Your timing is impeccable.

You are northern hemi. When you log in at toolkit.solcast.com.au, and mess with either of your site azimuth does this happen:

• -45 / 45 = North-East in the hint, and North-West respecively
• -135 / 135 = South-East, and South-West respectively.

i.e. 0° = North, 180° = South.

Given you are very close to the margin at almost exactly East/West orientation any misconfiguration would probably have near zero forecast impact. Do your azimuth settings pass this sanity check? Because northern hemi., is 0° considered North?

[gcoan]

There's a further discussion on the azimuth warning on #331 where the warning message was raised as an issue (now closed)

probably best to continue discussion about whether to improve the validation/change the message here

[gcoan]

Responding to @autoSteve question about how the azimuth is displayed/explained in the Solcast site

(@gcoan, in case our emails are now blocked could I get you to log in and mess with one of your azimuth settings?)

Here's my East facing array, Solcast correctly describes it as East:
(SS: Image removed. Privacy)

West:

If I increase the azimuth of the West, 111 is reported as West, 112 as South-West. Decreasing the azimuth, 67 is West and 66 is reported as North-West

Similarly on the East array, -113 is reported as South-East then -112 all the way to -68 is East, and -67 is North East.

BTW you will notice I have a fairly low efficiency factor on solcast. This is because I found I was getting PV50 forecasts that I rarely achieved with the config set per the Solcast guidance. I decreased the efficiency to compensate. Noting the other thread discussion that the auto dampening only ever reduces the Solcast forecast, never increases it, I will need to adjust the efficiency factors and let the dampening sort out the under performance (its not shading that is causing the drop). I do wonder whether there should be a small ability for the integration to bump up the forecast, e.g. by 5% in the case of over performance, but that's a different discussion topic

[autoSteve]

So the Solcast azimuth is consistent the globe over.

Thought so, and thank you, Geoff. 0° = North, 180° = South, completely the opposite of Forecast.solar.

[gcoan]

Here (from the other github issue) is my suggestion on rewording the readme:

Instead of
"If an unusual azimuth is configured then the integration will raise a warning for the issue. If the orientation is truly unusual (e.g. a northern hemisphere rooftop pointing towards North/North East/North West), then simply ignore the issue."

how about "A Solcast configuration of roof aligned North/North East/North West in the Northern hemisphere or South/South East/South West in the Southern hemisphere is considered to be possibly unusual because these orientations are not directly facing the sun at any time.
On start-up the integration will validate your Solcast azimuth setting in order to highlight a potential mis-configuration and will issue a warning message in the Home Assistant log if it detects an unusual roof alignment. If you receive this warning and have confirmed your Solcast settings are correct then the warning message can be simply ignored, the warning is there to try to catch configuration mistakes."

[WyndStryke]

Yep, as confirmed above, 0 degrees is north in the northern hemisphere too, on the Solcast website. Azimuth seems to be inconsistent on other forecasting sites too - sometimes it is 0 to 359, sometimes it is -180 to 180, sometimes north is zero, or south is zero. So plenty of scope for confusion when people are configuring it.

I actually just increased my efficiency from 90% (the default?) to 98% since it has been consistently underestimating the main forecast generation and PV90 by about that amount on good forecast days. Although the PV10 forecast often seems too optimistic on days when the overall forecast is bad (i.e., it seems that if the forecast is good, then the actual is better, but if the forecast is bad, then the actual is worse).

[autoSteve]

Discussion pinned and locked for now in the hope that others will double check azimuth, or simply ignore the issue.

If the azimuth is incorrect at a more skewed house orientation then the forecast values will likely be "shifted" left or right of actual generation.

This is why the logged warning is there, along with a raised issue, because multiple issues have been opened here for the same symptom, and have have always resulted in the same answer. "Your azimuth is wrong. Have you migrated from Forecast.solar, by any chance?"

With Solcast, 0° = North, 180° = South, and with other providers it is the opposite. West is a positive number, East is negative.

There are corner-cases. There always are. For an East/West facing property with solar panels facing in each direction the issue will be raised. Check your azimuth that is set for each face, move on and ignore the raised issue. (Don't just assume your azimuth is correct and move on. Check it per the README.md)

Whether the azimuth sanity check can be finessed is a topic for elsewhere. I think it serves an adequate purpose as it is. Those freaked out by an ignorable issue raised can be directed here.

This happens. Courtesy @robyholmes.