Help to understand weather data from PVGIS

[Asger Kappel Skau]

Hi,

I have faced a strange pattern when using the  PVGIS (2005-2020, PVGIS-SARAH2/ERA5) weather file. 

As you can see on the attected picture. I achieve a higher kWh/kWp at SW orientated PV panels than S. Furtheremore, the E values are very low and the W is very high. This makes me suspestion how that is possible. I suspect it could be because of the data maybe doesn't take into account we do the summertime/wintertime switch, and there for looking better than it is. 

Anyone there has some good explaniation on this? it is the exact same location, just two apperantly very different meteofiles. the laocation is Zealand in Denmark.

Hope anyone have a good solution or explanation.

BR Asger

[developer_bh]

Hello Asger,


thank you for your feedback and for taking the time to report this anomaly in the climate data to us!

I agree with you that a higher value of Southwest values compared to South values lowers the confidence in the climate file.

I am guessing that you are using a PVGIS file version 5.1 or earlier?

Specifically, your problem is most likely an unnatural distribution of global radiation values towards the end of the day (sunset), which is why the southwest orientation has elevated values. PVGIS is aware of the problem and claims to have solved it with version 5.2.

In the following graph you can see such an unnatural peak at the end of the day.



If you could send me your climate file I could investigate the problem further, which would be very helpful in proving the thesis and taking preventive measures if necessary. You can send me the file via PN or upload it in this thread.

Until further clarification, I would recommend that you use alternative climate data for this location, for example from Meteonorm.


Best regards,

Ben

[Asger Kappel Skau]

Hi,

Thank you for the answer. I'm currently using PVsol premium 2023, i can't see if it is PVGis 5.1 or 5.2 anywhere. I have assumed it was 5.2 because of the name PVGIS-SARAH2/ERA5, but maybe i'm wrong. The peak looks very unnatturel. 

It would be of great help if you could take a look at the weather file. 

There is two weather files attached in this message. The one called "Slagelse_2005_2005.wbv" is the PVgis and the "Flakkebjerg_1996_2015.wbv" is meteonorm 8.1. It is same location and the Flakkebjerg file is original from the PVsol database, guess it's a weather station. 

Hope you are able to help and this is the correct file format. 

Br Asger

Flakkeberg_1996_2015.wbv Slagelse_2005_2020.wbv

[developer_bh]

Hello Asger,

thank you for your answer!

• PVGIS 5.1 has a timerange from 2005 to 2016 and is called "PVGIS-SARAH/ERA-Interim" in the source.
• PVGIS 5.2 has a timerange from 2005 to 2020 and is called "PVGIS-SARAH2/ERA5" in the source.

So your file is PVGIS 5.2.

That means that my first theory of wrong data towards the day-end does not apply here.

The daily course of irradiation data for Slagelse looks plausible :

It would be helpful at this point to have a deeper look into your project. You may send it to me via PN if you wish.

 

Best regards,

Ben

 

[Asger Kappel Skau]

Okay, thank you. 

Could be of great interest if you want to look deeper into the project. 

How do i send in via PN? 

Best regards asger

[developer_bh]

Yes would be interesting indeed. You can click on my profile name and then you will find a button "Message".

[developer_bh]

Dear Asger,

 

thanks for sending the project!

The weatherfile used in the project is different from what you sent previously.

The weatherfile in the project has at least two flaws : it has missing data in may and october.

I guess this is the reason your project has these irregular results when using PVGIS.

Hope this could help you?

 

Best,

Ben

 

[Asger Kappel Skau]

Yea, that is correct, is another location not far from the other one. But the pattern is the same.

But why does that end up with a high west value? How does PVsol tackle missing weather data?

[developer_bh]

Quote

But why does that end up with a high west value?

The high west value most probably comes from PVGIS data with irradiation peaks at the end of the day. Although this is no pattern which is as obvious as in my first response, you can still find those peaks in the PVGIS data. For example on october third you have higher global-radiation at the end of the day (black square) resulting in higher irradiation on the south-west pvmodule (red) in comparison to the south-pvmodules (blue).

I did not investigate the whole year and did some maths on it due to limited time, but I'm pretty sure this is the reason.

[developer_bh]

Quote

 

How does PVsol tackle missing weather data?

 

You have a notice in the MeteoSyn-Location-Details view about missing periods of GHI data.

If you use this location this will be treated as no GHI for this period.

 

Best regards,

Ben

[Asger Kappel Skau]

Thank you very much. 

[developer_bh]

Dear Asger,

today I further investigated your issue.

The problem is that PVGIS has an offset in their timestamps, that the sun-course over the day does not match with the GHI data.

You can see it for example on 14th of may in the picture below.

That leads to an higher, unnatural GHI peak towards the end of the day.

We try to further improve our precheck when fetching the data to prevent downloading these locations.

 

Best,

Ben

[Asger Kappel Skau]

Interesting. 

Do you think this a general thing when using PVGIS (2005-2020, PVGIS-SARAH2/ERA5) for this syntetic data? 

How is it possible to dive deeper into the weather files other than using the diagram as above? I just gets symbols when opening the .Wbv files? 

 

[developer_bh]

Let's put it that way : It is not uncommon for PVGIS. But not all locations have flaws fortunately. They try to improve so it is still good to look at each location for anomalies and to reevaluate when a new version is out (f.e. PVGIS 5.3). Main things to consider are:

• missing GHI data
• unnatural GHI data
• missing temperature data
• offset in sun-course with GHI data

The best way to dive into wbv data before the simulation started is the MeteoSyn Details view at the moment.

The sun position is calculated afterwards and not included in the weatherfile. However it is based on timestamps in the weatherfile.

You can also do a database-sanity-check in MeteoSyn and put Phase I to mode "Expert". In the following dialogue you'll see "findings" of the weatherfiles in your database. The current problem you mentioned here will not be detected yet.

You also have the possibility to copy the simulation-results data into a table, f.e. paste it into Excel.

[Asger Kappel Skau]

Thanks Good to know, thank you for the help. i will try tio dig deeoer into the different weather files. 

[developer_bh]

If you want to keep the PVGIS file, I highly recommend doing minutewise simulation. There is an auto-correction of most values and you will most likely see south-oriented have higher values than southwest-oriented panels again.