dask.array<shape=(76, 76), dtype=float32, chunksize=(76, 76)>
    crs      object +proj=utm +datum=WGS84 +ellps=WGS84 +no_defs +units=m +zone=33.0 +type=crs
  * y        (y) float64 4.006e+06 4.006e+06 4.005e+06 ... 3.951e+06 3.95e+06
  * x        (x) float64 4.094e+05 4.104e+05 4.114e+05 ... 4.832e+05 4.842e+05
    sensor:               slstr
    view:                 n
    polarization:         None
    standard_name:        toa_brightness_temperature
    coordinates:          ('longitude_in', 'latitude_in')
    resolution:           1000
    wavelength:           (10.4, 10.85, 11.3)
    file_type:            esa_l1b_ntir
    long_name:            Gridded pixel brightness temperature for channel S8...
    units:                K
    level:                None
    calibration:          brightness_temperature
    name:                 S8_in
    platform_name:        Sentinel-3A
    modifiers:            ()
    start_time:           2018-07-20 09:12:25.111891
    end_time:             2018-07-20 09:15:24.803164
    area:                 Area ID: LS8_duplicate\nDescription: Patch subset i...
    ancillary_variables:  []

I have this Xarray which resulted from reading and resampling a NetCDF file using satpy.

newscn = scn.resample(area_def, resampler='nearest')

Printing the newscn results in an xarray. How can I extract the (76, 76) array with float32 values as a numpy array?

  • Assuming your xarray array is stored in a variable named arr you should be able to access the values (as a numpy array) by running arr.values. Commented Sep 10, 2019 at 14:38

1 Answer 1


To resolve this matter after I loaded the new scene using


where filename_S8 = 'S8_in' as I want to read the the thermal band 8 from the Sentinel 3 data product. Then I used this line to access the (76, 76) image array as a numpy array:

Arr = newscn[filename_S8]
S3_BT = Arr.compute().data

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.