Discrepancy between calculated TSI in solar_radiation.py and ERA5 TISR

[IncubatorShokuhou]

Hello,
I hope this message finds you well. While working with the graphcast/solar_radiation.py script, there appears to be a difference between the global Total Solar Irradiance (TSI) calculated by graphcast/solar_radiation.py and the Top of Atmosphere Incident Solar Radiation (TISR) from ERA5 reanalysis data.

On January 1, 2007, at 13:00 UTC, I compared the following:

• Global TSI calculated by graphcast/solar_radiation.py
  
• TISR value from ERA5 for the corresponding moment
  
• The difference between these two values
  

Is this discrepancy expected due to differences in calculation methods or data sources? Or could there potentially be an error or inaccuracy in the calculation method used in solar_radiation.py?