Horizon Data comparison

[Davide Maglione]

Hello forum users,

i am doing some research on the various Horizon sources available. i have chosen some meteo data i have used for previous projects and i would like to do a series of analysis for each one of them using:

1 - SolarGIS Horizon profile

2 - Meteonorm Horizon (from PVsyst web sources)

3 - PVGIS Horizon (from PVsyst web sources)

my goal is to: compare the "far shading losses" and the Horizon line drawings considering the different Horizon data so to have a general overview of the differences between the Horizon datasets; of course my research will also include the study of the manuals and theory behind the definition of the Horizon profile provided by each one of the three Meteo database.

i would like to raise a couple of questions:

1. how far has to be an obstacle to be actually considered in the far shadings?
2. considering the high dependency of the horizon from the exact geographical coordinates, are there any suggestion on the horizon profile choice regarding big PV plants? 
3. has someone of you undergone some similar studies? have you found some interesting results?

also, considering point 2, a question for the software developers:

• are you considering a possible software update that would allow the user to set different horizon profiles, perhaps linking a single profile to a single subfield?

thank you very much in advance for your contribution!

Davide

[Michele Oliosi]

Hi,
Just a small comment on points 1 and 2: as a rule of thumb in PVsyst we consider that whatever obstacle that is at a distance of about 10 times the size of your system (or more) can be included in the far shadings. Since this depends on the size of the system, you can see how this also will make the point (2) less problematic.

[Davide Maglione]

Hi Michele,

thank you very much for the answer. can you clarify what you mean when referring to 10 times the size of the system? 

also, when you say that "can be included" you mean that the software will automatically recalculate the horizon profile? 

[Michele Oliosi]

10 times the size of the system: it is an approximation of course. By size of the system you can take the distance between the two PV tables that are furthest apart.

No sorry there is no tool to calculate the horizon profile. You would need to consider these objects in your horizon line manually, for example after having made some measurements. Example: there is a high rise building but quite in the distance, you can calculate the height above the horizon that the buildng would appear at.

[Davide Maglione]

Ok thank you!

Talking about the data import i noticed 2 things:

• The Sun heigth value is always cut to the 1st decimal number
• The coordinates for the horizon point are always cut to the 4th decimal number

does this affect the calculation or is just the way the software lets the user visualize the import data?

i know that in the "batch simulation" feature we can create a batch of simulation even modifying the climate data. is it possible to modify the Horizon too? 

Edited March 28, 2023 by Davide Maglione

[Michele Oliosi]

No the batch mode doesn't allow to modify the horizon at the moment. It could be implemented at some point.

The horizon line values are indeed truncated to the 1st decimal number. We consider this precision sufficient: note indeed that the sun moves of about 15° per hour.
For the horizon: the 4th decimal number is what is used for all latitude/longitude evaluations. This means a localization of your PV plant to about 10 m, which is small enough for most systems.

[Davide Maglione]

thank you Michele, 

as always very helpful answers ?

i just noticed that the latest patch addedd some modifications on the Horizon losses, in fact i am noticing substantial differences with the former far shading losses figures, independently from the Horizon provider.

can you explain to me what did the patch modified?

[Michele Oliosi]

Glad I coud help !

In the release notes:

Quote

- Horizon (far shadings): the calculation of the horizon shading has been corrected for times the sun passes the horizon line. This may slightly change the simulation results

Basically there was a correction to the way the horizon is accounted for, in the hours in which the sun passes behind the horizon line. Consider hypothetically a tall mountain peak, behind which the sun passes in less than one hour.
Before the update, the corresponding hour would have had no far shading losses. Indeed one considered the beginning and the end of the hour to verify whether the sun passed behind the horizon. We now account for the time span behind the horizon. even though it is only a fraction of the hour.

[laurahin]

We have compared the effect of Meteonorm and PVGIS horizons and found that the far shading loss was typically greater for the PVGIS version.  In making the comparisons, we noticed that the profiles as shown in the PVsyst report "Sun Paths" diagram have different characters.  Even though the number of points used to define the profiles is similar, the PVGIS profile always appears smooth whereas the Meteonorm profile is blocky.  Is there any particular reason for that?

I would expect the Meteonorm and PVGIS horizons to be the same, since they use the same spatial resolution (90m) and likely both rely on data from the Space Shuttle Radar Topography Mission, although I haven't been able to confirm this.