FCI L2 reader returns categorical data as float #3338
Description
Describe the bug
Reading categorical data from FCI L2, such as cloud type, results in a float32.
This should be int8 or uint8.
To Reproduce
# Your code here
from satpy import Scene
from satpy.utils import debug_on; debug_on()
sc = Scene(
filenames=["/media/nas/x23352/MTG/FCI/L2-cases/202602121200/W_XX-EUMETSAT-Darmstadt,IMG+SAT,MTI1+FCI-2-CT--FD------NC4E_C_EUMT_20260212121354_L2PF_OPE_20260212120000_20260212121000_N__C_0073_0000.nc"],
reader="fci_l2_nc")
sc.load(["cloud_type"])
print(sc["cloud_type"].dtype)Expected behavior
I expect an integer dtype.
Actual results
[DEBUG: 2026-02-13 10:31:47 : satpy.readers.core.yaml_reader] Reading ('/home/gholl/checkouts/satpy/satpy/etc/readers/fci_l2_nc.yaml',)
[DEBUG: 2026-02-13 10:31:47 : satpy.readers.core.yaml_reader] Assigning to fci_l2_nc: ['/media/nas/x23352/MTG/FCI/L2-cases/202602121200/W_XX-EUMETSAT-Darmstadt,IMG+SAT,MTI1+FCI-2-CT--FD------NC4E_C_EUMT_20260212121354_L2PF_OPE_20260212120000_20260212121000_N__C_0073_0000.nc']
[DEBUG: 2026-02-13 10:31:47 : rasterio.session] Could not import boto3, continuing with reduced functionality.
/home/gholl/checkouts/satpy/satpy/readers/fci_l2_nc.py:185: UserWarning: The specified chunks separate the stored chunks along dimension "number_of_rows" starting at index 4096. This could degrade performance. Instead, consider rechunking after loading.
self.nc = xr.open_dataset(
/home/gholl/checkouts/satpy/satpy/readers/fci_l2_nc.py:185: UserWarning: The specified chunks separate the stored chunks along dimension "number_of_columns" starting at index 4096. This could degrade performance. Instead, consider rechunking after loading.
self.nc = xr.open_dataset(
[DEBUG: 2026-02-13 10:31:47 : satpy.composites.config_loader] Looking for composites config file fci.yaml
[DEBUG: 2026-02-13 10:31:47 : pyorbital.tlefile] Path to the Pyorbital configuration (where e.g. platforms.txt is found): /home/gholl/miniforge3/envs/py313/lib/python3.13/site-packages/pyorbital/etc
[DEBUG: 2026-02-13 10:31:47 : satpy.composites.config_loader] Looking for composites config file visir.yaml
[DEBUG: 2026-02-13 10:31:47 : satpy.readers.fci_l2_nc] Reading in file to get dataset with key cloud_type.
[DEBUG: 2026-02-13 10:31:47 : satpy.readers.core.yaml_reader] Requested orientation for Dataset cloud_type is 'native' (default). No flipping is applied.
float32
Environment Info:
- OS: openSUSE Leap 15.6
- Satpy Version: v0.59.0-117-g5e5e33cbf
Additional context
When writing the output as mode L geotiff, trollimage will scale the pixel values. This results in pixel values that are wrong. This wouldn't happen if the data were int.