TOA reflectance not between 0 and 1

[zxdawn]

• enpt version: 0.19.2
• Python version: 3.10.10
• Operating System: Mac OS M2

Description

The converted TOA reflectance is not between 0 and 1.

What I Did

config_minimal = dict(path_l1b_enmap_image=f_enmap)
CTR = EnPT_Controller(**config_minimal)
CTR.read_L1B_data()

# radiance
print(CTR.L1_obj.vnir.data[:, :, 0])

# reflectance
CTR.run_toaRad2toaRef()
print(CTR.L1_obj.vnir.data[:, :, 0])

Output

array([[75.735275, 75.68156 , 74.43002 , ..., 73.790054, 73.466194,
        74.22445 ],
       [76.41141 , 75.18983 , 74.571556, ..., 73.85979 , 73.63975 ,
        73.25236 ],
       [76.01936 , 75.75181 , 75.10099 , ..., 74.102554, 73.778694,
        73.7038  ],
       ...,
       [66.28344 , 68.94043 , 70.51789 , ..., 72.46776 , 73.046265,
        74.46463 ],
       [67.42289 , 67.8516  , 68.39808 , ..., 73.19657 , 73.11548 ,
        73.56175 ],
       [67.921844, 68.80148 , 67.12692 , ..., 73.925385, 73.25494 ,
        73.73582 ]], dtype=float32)


array([[2145, 2143, 2108, ..., 2090, 2081, 2102],
       [2164, 2130, 2112, ..., 2092, 2086, 2075],
       [2153, 2145, 2127, ..., 2099, 2090, 2087],
       ...,
       [1877, 1952, 1997, ..., 2052, 2069, 2109],
       [1909, 1922, 1937, ..., 2073, 2071, 2083],
       [1924, 1949, 1901, ..., 2094, 2075, 2088]], dtype=int16)

[danschef]

The TOA reflectance is internally scaled between 0 and 10000 as this allows to store the data in integer values, which in turn reduces the memory footprint of the processing pipeline. Although there is a parameter to adjust the TOA reflectance scale factor (EnPT_Controller(scale_factor_toa_ref=1), documented here), I just noted that setting it to 1 does not change the data type of the TOA reflectance values to float. If needed, I could fix that. But you also just divide the TOA reflectance values by the scale factor.

[zxdawn]

Thanks for the info. If I add the scale_factor_toa_ref , the results are all 0. I guess that's the issue with data type (int)

import numpy as np
from enpt.execution.controller import EnPT_Controller

config_minimal = dict(path_l1b_enmap_image=filename, scale_factor_toa_ref=1)
CTR = EnPT_Controller(**config_minimal)
CTR.read_L1B_data()

CTR.run_toaRad2toaRef()
print(np.any(CTR.L1_obj.vnir.data))
# output: False

Anyway, dividing the TOA reflectance values by the default scale factor (10000) works well.