The efficiency and the noises are different for each pixel.
To estimate realistically our errors we must take this effect into account.
In the modified code the pixel properties can be define individually with map of quantum efficiency, dark current and read noise.
The detector properties are define in the aXeSIM configuration file with the keywords:
RDNOISE _float/string_ DC _float/string_ QE _float/string_ NPIXX _integer_ NPIXY _integer_
In the case where RDNOISE, DC or QE are define with a float, all the detector pixels have the same properties.
In the case where RDNOISE, DC or QE are define with a string, the value have to be the path to the RDNOISE, DC or QE map.
The map format have to be a fits image with the same shape of the simulated detector (NPIXX,NPIXY).
Noise distribution model¶
- Composite model :
- one gaussian component centered one the mean value
- one exponential component to simulate the queue of the noise distribution
- Mask array : define criteria for bad pixel and mask it in the product using the DQ array
We should used the same convention than HST: