latte-data-processing

CCD calibration frames preprocessing and data reduction scripts used for NTHUAC's LATTE observations

The LICENSE only applies to .py scripts.

Please pip install astropy astroalign scikit-image if you want to use it.

Pixel math formulas

The following are the details of pixel math operations that is done on the calibration frames and data by the processing script.

Dark (and bias)

Let $D_R[i]$ denote the raw frames and $D_M$ be the resulting (combined) master frame.

$D_M = \text{mean}(D_R)$

Downsampled dark

Additionally, the 5min dark frame is downscaled (downsampled) from 4096x4096 to 2048x2048 for use in the bin 2x2 images by local mean downscaling.

Flat (Regular flats)

These are for the flat frames that does not require special handling (band Ha/OIII).

Let $F_R[i]$ denote the raw flat frames, $D_F$ be the master dark with the corresponding exposure time, and $F_M$ be the resulting (combined) master flat frame.

$F_R^\prime[i] = F_R[i] - D_F$

$F_M = \color{blue}{\text{fix}}\color{black}(\text{mean}(F_R^\prime))$

Since the bias is included in the flat dark frames, it is not removed additionally. By $\color{blue}{\text{fix}}$ we remove dead pixel column by taking mean of local pixels.

Flat (Long-short flats)

These are flat frames that are captured with an exposure less than 10 seconds (band L/B/V/R), which may leave a artifact on images due to a slow-moving shutter curtain.

Let $F_L[i]$ denote the long flat frames, $F_S[i]$ denote the short flat frames, and $F_M$ be the resulting (combined) master flat frame.

$F_M = \color{blue}{\text{fix}}\color{black}(\text{mean}(F_L) - \text{mean}(F_S))$

The flat dark is neglated due to a rather short exposure time. Since the bias is included in the short flat frames, it is not removed additionally. By $\color{blue}{\text{fix}}$ we remove dead pixel column by taking mean of local pixels.

Downsampled flat

Additionally, the master flat frame for B, V, R is downscaled (downsampled) from 4096x4096 to 2048x2048 for use in the bin 2x2 images by local mean downscaling.

Light

Let $L_R[i]$ denote the raw light frames, $F$ denote the corresponding master flat frame, $D$ denote the corresponding master dark frame, and $L_M$ be the resulting (combined) master flat frame.

Note that the $\color{red}{\text{red}}$ part is a single number!

$L_M = \text{median}(L_R^\prime)$

Since the bias is included in the dark frames, it is not removed additionally. Also, each light frame is registered (aligned) using astroalign in prior to being combined.