Fringe correction in the PREREPIX package
110 likes | 235 Vues
This document details the implementation of a comprehensive fringe correction and preprocessing pipeline within the PREREPIX package applied to CFH12K data. Key steps include the generation of master frames such as BIAS and FLATS, and the application of robust fitting methods for scaling factors. Challenges addressed include masking issues and the necessity for quality assessments of the fringe correction results. The empirical methods demonstrated effectiveness even in non-optimal conditions, although further evaluation of residuals is required for quality assurance.
Fringe correction in the PREREPIX package
E N D
Presentation Transcript
Fringe correction in the PREREPIX package M. Radovich (OAC, Naples) Y. Mellier, E. Bertin, G. Missonnier (IAP)
Prereduction • Implementation: PERL + PDL [+ pTK] • Status of the pipeline: all the prereduction steps are implemented • Main problems: fringe correction and masking • To be done: quality assessment (e.g. fringe correction); the generation of masks should be improved
The Preprocessing steps Generation of master frames: • BIAS • DARK • DOME & TWILIGHT FLATS, • FRINGE • SUPERFLAT Science frames: preprocessing + generation of • Weight maps • Previews
Fringe correction • The method: for each science frame we assume that • A smoothed background is subtracted from each image, the so obtained frames are combined to build a master fringe frame • Objects cannot be masked since “detections” may include fringes • For each science frame, the scaling factor is computed using a robust fitting method • >5 frames and large dither patterns are required !! • Is the fringe pattern stable during one night/run ? PCA analysis could be used. • To be done: evaluate fringe residuals
Rescaling of the master fringe frame: counts in the master fringe and the image frames are displayed. The red line shows the results of the robust fit
Example of fringe subtraction with and without rescaling K=0.83 K=1.0 Raw image (section)
Fringe frame built from a run with: narrow dither sequence, only two separate fields Raw image (first field) Corrected image, first field Corrected image, second field
Superflat • One master frame for each run, obtained from the combination of fringe corrected science frames. • To minimize the impact of bright stars, before combination science frames are “smoothed” using SExtractor
Speed Example: processing of 300 files for each step
Conclusions • The empirical method for fringe subtraction (combination of frames + computation of the scaling factor by a robust fit) works for CFH12K data even in non optimal conditions (narrow dither, few different pointings, one fringe frame per run ) • The algorithm is still slow • Need to evaluate residuals (how much the correction is good ?)