Make SEVIRI calibration more transparent to the user

[sfinkens]

Feature Request

Is your feature request related to a problem? Please describe.
In some situations SEVIRI users request GSICS calibration but get Nominal calibration. Reason: If GISCS coefficients are not available for a certain channel, the SEVIRI calibration handler silently falls back to nominal coefficients. For users it's hard to know which coefficient has been applied. They would have to compare both calibration methods or check the header contents.

Example: MSG3-SEVI-MSG15-0100-NA-20130302231244.078000000Z-NA.nat has GSICS coefficients only for IR_087 and IR_120.

Describe the solution you'd like

• Add dataset attributes calib_method describing the calibration method and calib_coefs specifying the applied coefficients
• Update documentation to describe this behaviour. Also, add link to GSICS website for latest coefficients.
• Log a warning when falling back to Nominal coefficients

Describe any changes to existing user workflow
None

Additional context
None

[pdebuyl]

Question: if the user requests a specific set of calibration, shouldn't satpy raise an actual error?

[mraspaud]

Question: if the user requests a specific set of calibration, shouldn't satpy raise an actual error?

I agree actually... (that this should be raising an error)

[sfinkens]

Good point! Then we would need an additional reader kwarg to allow mixing of coefficients.