Read out noise

Every time a pixel is read by the electronic, an error is introduced. This is called read noise. This is the noise of the amplifier which converts the counts into a change in analog voltage for the ADC. This kind of uncertainty is represented by default by AddNormalReadNoise. This Task simulates the read noise as a normal distribution whose parameters can be defined in the configuration file.

<channel> channel
    <detector>
        <read_noise> True </read_noise>
        <read_noise_task> AddNormalReadNoise </read_noise_task>
        <read_noise_sigma unit="ct"> 10 </read_noise_sigma>
    <detector>
</channel>

A different realization of the same distribution is added to each pixel of each sub-exposure.

\[S_{meas} = S_{meas} + \mathcal{N}(\mu = 0, \sigma = \sigma_{RN})\]

Alternatively, a map of read noise measured for each pixel can be used. A default Task is provided for this scope assuming numpy array (see numpy documentation) as input: AddReadNoiseMapNumpy

<channel> channel
    <detector>
        <read_noise> True </read_noise>
        <read_noise_task> AddReadNoiseMapNumpy </read_noise_task>
        <read_noise_filename> read_noise_map.npy </read_noise_filename>
    <detector>
</channel>

Note

Other custom realizations of this Task can be developed by the user (see Custom Tasks).

Note

For reproducibility, the seed for the random generator can be set as described in Random seed and Random generators. Remember that in the case of multiple chunks used, the random seed used in any chunk is stored in the output file for reproducibility.