Apply bias, dark, and flat correction

Calibrate a synthetic CCD image with supplied masters and verify the result.

Use ccdred with prepared bias, dark, and flat masters. This companion to the preprocessing reference shows subtraction, dark-exposure scaling, and flat normalization.

Prerequisite: complete the FITS quickstart and use an environment with astroimred[full]. All arrays below have the same shape.

1. Prepare a science image and three masters

import numpy as np
from astropy.nddata import CCDData

from astroimred.reduction.preproc import ccdred

science = CCDData(np.full((8, 8), 124.0), unit="adu")
master_bias = CCDData(np.full((8, 8), 10.0), unit="adu")
master_dark = CCDData(np.full((8, 8), 2.0), unit="adu")
master_flat = CCDData(np.full((8, 8), 2.0), unit="adu")

The dark master here is already bias-corrected. Its exposure is 15 seconds; the science exposure is 30 seconds. The flat’s constant value makes its normalization easy to verify.

2. Apply the calibration explicitly

calibrated = ccdred(
    science,
    mbias=master_bias,
    mdark=master_dark,
    mflat=master_flat,
    dark_scale=True,
    exptime_data=30.0,
    exptime_dark=15.0,
    flat_norm_value=None,
    do_crrej=False,
    verbose_bdf=0,
)

The calculation is (124 - 10 - 2 * (30 / 15)) / (2 / 2) = 110 ADU. Here flat_norm_value=None divides the flat by its mean before correction. The default flat_norm_value=1 instead uses the supplied flat normalization.

np.testing.assert_allclose(calibrated.data, 110.0)
np.testing.assert_allclose(science.data, 124.0)
calibrated.shape, str(calibrated.unit), float(calibrated.data.mean())

((8, 8), 'adu', 110.0)

ccdred returns a reduced copy. Omitting both a master and its path skips that calibration stage; the routine does not discover suitable masters.

3. Use FITS files for an observation

Replace the paths below with matching science and master files. This example is illustrative and is not executed by the documentation build.

from astroimred.fitsmgmt.io import load_ccd

science = load_ccd("science.fits")
calibrated = ccdred(
    science,
    mbiaspath="master-bias.fits",
    mdarkpath="master-dark.fits",
    mflatpath="master-flat.fits",
    dark_scale=True,
    exptime_data=30.0,
    exptime_dark=15.0,
    flat_norm_value=None,
    output="science-calibrated.fits",
    overwrite=False,
)

Supply exposure times in the same unit. If you use header lookup instead, check exptime_key and both exposure values first: missing exposure metadata can leave the dark unscaled. overwrite=False protects an existing output.

4. Check uncertainty before measuring

ccdred does not calculate or propagate calibration uncertainties. This example verifies calibrated values only; prepare the uncertainty model required by your measurement separately.

Combine matching calibrated exposures with the combination tutorial. For fringe subtraction, masks, or cosmic-ray rejection, start with the ccdred reference and crrej reference.