ASCII Import - Clearness index high with respect to clear day model

[ocrighton]

Hello,

I am have used the "Import ASCII meteo file" tool to import some meteo files that I downloaded from the NREL Data Viewer. I believe I have imported the files correctly, but for some the 'Check data quality' tab displays the message "The Clearness index of the best clear days seems high with respect to the clear day model. Check the values with the "Best clear day Ktcc" graph. I have attached images showing this message, and the "Best clear day Ktcc" graph.

Again, I believe that these files were imported correctly, and that the data (GlobHor, Diffhor, etc.) in the PVsyst meteo file is correct. I am able to perform a simulation using the meteo file.

I was wondering what the message I am seeing might mean. If the data in the meteo file seems like it is correct, and I am able to perform a simulation using the meteo file, should I disregard the message? Should I just consider that the data might be flawed, since it is resulting in unusually high clearness index values?

Thanks very much.

[André Mermoud]

PVsyst tries to put benchmarks to the imported Meteo data. The aim is to avoid bad data due, for example, to solarimeter calibrations or erroneous imputs.

Now this is not quite easy, for some situations the data may be out of our criteria. For e very clear sky locations (low aerosols, low humidity), the data may overcome the clear day model.

In your case, the over-irradiance with respect to the model seems rather high (we rarely see more than 10%). But this only concerns winter days. Perhapes there is something unaccurate in the model of the meteo data provider. You can try to find other meteo data for your site (from another source) for comparison.

By the way this doesn't prevent you to perform the simulation.

[sbchapman]

We are seeing similar issues with a more recent SolarGIS dataset with again values exceeding 10%. Apart from just adjusting the thresholds in the software to mask the messages is there any qualitative analysis we can do of the data to assess the "correctness"?