Problems with first real time MTG FCI data

[lobsiger...@gmail.com]

Dear developers,

with first real time data of MTG-I1 now available on EUMETCast I made the 'airmass' image attached below without any problem. Then I wanted to do a composite 'natural_color' with the same raw data from 202311021030. I got an error message and consequently conda updated satpy. Unfortunately I don't remember the satpy version I started with but now I have:


# Name                    Version                   Build    Channel
satpy                        0.44.0                     pyhd8ed1ab_0    conda-forge
pyresample             1.27.1                    py311h320fe9a_0    conda-forge


But I constantly get the same error with different MTG slots, areas and composites. Even composite 'airmass' does not work anymore and now also has this very same error.

I attach the full debug output as a text file. Any ideas?

Best regards,

Ernst

[lobsiger...@gmail.com]

Dear Developers,

I wonder whether my problem is related to this FCI reader issue

https://github.com/pytroll/satpy/issues/2588

From the discussion on github it seems to me that the 40 data
segments are sorted by time but all have the same slot time?!

Actually if I load the global scene with  "pad_data=False"  I can
again make composite 'airmass' and also composite 'natural_color'
though that takes extremely long processing times (Intel XEON
32GB RAM, moderately loaded with other satpy processing tasks).

'airmass'
real    4m24.796s
user    6m51.223s
sys     0m16.339s

'natural_color'
real    7m22.727s
user    22m8.219s
sys     1m17.345s

I can also make composite 'convection' and 'true_color_raw'.
But when I want to make composite 'true_color' it fails with

an error that might again point to a block sorting problem?!

(Debug output from composite 'true_color' attached below)

'convection'
real    9m16.715s
user    20m55.170s
sys     0m57.754s

'true_color_raw'
real    8m52.891s
user    23m6.892s
sys     1m9.177s


Now that "manufacturers" get  first real time MTG FCI data via
EUMETCast I expect more users will have/report similar problems.

Best regards,
Ernst

[lobsiger...@gmail.com]

Dear All,

I have sorted the problem and sorry it was all my fault. I made my choice of files with a glob() of the processing time instead of the start time.

While this has worked with the EUMETSAT FCI L1C test data (except for enhanced composites) I expected it to work for real time data as well.

Rereading issue 2588 I finally found the problem. All is working now with "pad_data=True" and my processing times are below 1 minute as well.

Have a nice WE,

Ernst

[Martin Raspaud]

Dear Ernst,

I'm glad you got it to work in the end :)

Best regards,
Martin

> --
> 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/7e458785-a0ec-43bd-813e-10bca931325dn%40googlegroups.com <https://groups.google.com/d/msgid/pytroll/7e458785-a0ec-43bd-813e-10bca931325dn%40googlegroups.com?utm_medium=email&utm_source=footer>.