MTI1 FCI l1c test data problem

[lobsiger...@gmail.com]

Dear developers,

I downloaded 1 repeat cycle (RC0072, noon) of MTI1 FCI l1c data from:

https://sftp.eumetsat.int/public/folder/UsCVknVOOkSyCdgpMimJNQ/User-Materials/Test-Data/MTG/MTG_FCI_L1C_SpectrallyRepresentative_TD-360_May2022/

I noted that many enhancement failed probably because solar angles
and the like are not yet available in the test data or cannot be
read by the current FCI reader. Nevertheless I could make a couple
of full disk composites that I resampled with area 'fci_0deg_4km':

https://we.tl/t-3MO8OF53DI         wetransfer, link valid 1 week, 65MB

While coastlines fit well the data does currently not seem to fully
fill the (GEOS) FCI area definition as distributed with SatPy 0.36.
My 'fci_0deg_4km' is the SatPy distributed 'fci_0deg_2km' where
I simply set the pixel sizes from (5568 x 5568) to (2784 x 2784).

What is the problem? Is this a FCI reader problem or are all pixels
in the test data towards the limb empty or maybe flagged unusable?


Best regards,
Ernst

P.S: reader='fci_l1c_nc' , three 20%x20% reduced size examples attached

[lobsiger...@gmail.com]

Hi again,

with the same MTI test data I made an area='full_scan' image which means in my script using resampler='native' resulting in a 5568x5568 pixel image.

When overlaying the grid and coastlines with Pycoast I see the same problem at the limb as before. Attached a reduced picture 10%x10% + legend at left.

Does this result again point to a problem in the EUMETSAT distibuted test data? Any easy possible tests I could still do to find out? Any other ideas?

Regards,

Ernst

[lobsiger...@gmail.com]

It's me once more,

After finally reading this document

https://sftp.eumetsat.int/public/folder/UsCVknVOOkSyCdgpMimJNQ/User-Materials/Test-Data/MTG/MTG_FCI_L1C_SpectrallyRepresentative_TD-360_May2022/MTGTD-360%20Spectrally%20Representative%20FCI%20L1C%20Test%20Products%20-%20Package%20Description%20(1290888%20V1).pdf

chapter "4.2.7 Missing Pixels at High Satellite Zenith Angles" explained it all. Sorry for not doing my homework. But at least I did RTFM before asking silly questions to OPS.

As I suspected before the missing pixels are flagged unusable <cite> with “missing_warning” in the pixel quality variable. </cite>. So everything falls into place now :-).

Cheers,

Ernst

[martin....@smhi.se]

Hi Ernst!

Sorry for not answering earlier, many of us are on holiday at the moment, and I myself have been quite busy.

But, yeah, as you found out this is a "feature" of the test dataset. Happy you got your answer already.

Best regards,
Martin

On 04/08/2022 19:45, "lobsiger...@gmail.com" <lobsiger...@gmail.com> wrote:
> It's me once more,
>
> After finally reading this document
>
> https://sftp.eumetsat.int/public/folder/UsCVknVOOkSyCdgpMimJNQ/User-Materials/Test-Data/MTG/MTG_FCI_L1C_SpectrallyRepresentative_TD-360_May2022/MTGTD-360%20Spectrally%20Representative%20FCI%20L1C%20Test%20Products%20-%20Package%20Description%20(1290888%20V1).pdf
>

> chapter "4.2.7Missing Pixels at High Satellite Zenith Angles" explained

> https://we.tl/t-3MO8OF53DI <https://we.tl/t-3MO8OF53DI>        

> wetransfer, link valid 1 week, 65MB
>
> While coastlines fit well the data does currently not seem to fully
> fill the (GEOS) FCI area definition as distributed with SatPy 0.36.
> My 'fci_0deg_4km' is the SatPy distributed 'fci_0deg_2km' where
> I simply set the pixel sizes from (5568 x 5568) to (2784 x 2784).
>
> What is the problem? Is this a FCI reader problem or are all pixels
> in the test data towards the limb empty or maybe flagged unusable?
>
>
> Best regards,
> Ernst
>
> P.S: reader='fci_l1c_nc' , three 20%x20% reduced size examples
> attached
>

> --
> You received this message because you are subscribed to the Google
> Groups "pytroll" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to pytroll+u...@googlegroups.com
> <mailto:pytroll+u...@googlegroups.com>.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/pytroll/4624bdc1-ab71-44b6-9a32-4f5f38ed3c34n%40googlegroups.com
> <https://groups.google.com/d/msgid/pytroll/4624bdc1-ab71-44b6-9a32-4f5f38ed3c34n%40googlegroups.com?utm_medium=email&utm_source=footer>.
>

[lobsiger...@gmail.com]

Hi Martin,

this was my bad. Quite embarrassing as I often insist that people should RTFM. Another "feature", as I have suspected, is the reason for most enhancements not working:

<cite>

In particular, the following variables under the /state/celestial group are not currently computed
by the simulation code:
• solar_elevation
• solar_azimuth
• orbit_phase
• sun_eclipse_by_earth
• sun_eclipse_by_moon
Note that the solar_elevation and solar_azimuth variables are intended to describe the solar
angles in the instrument frame (as seen by the satellite), rather than for each pixel on the ground.

</cite>

Have a nice day,

Ernst