The coronagraphic flat fields differ from normal HRC flats in two ways. The Lyot stop alters both the large scale flat-field illumination pattern and the diffraction patterns from dust specks and filter pinholes. The occulting spots also cast shadows in the center and the upper left corner of the field of view. Because the spots are in the aberrated HST
beam, the shadows are vignetted up to 0".5 beyond the nominal edges of the spots. This vignetting makes the spots appear diffuse. The occulting spots wander by several pixels over weekly time scales, so their effects must be treated separately from the static features in the flat field. Consequently, there are two distinct flat-field reference images that must be applied during calibration of coronagraphic science images. (See ACS ISR 2004-16
for detailed information.)
The first reference flat is a static pixel-to-pixel flat (header keyword PFLTFILE
) that contains the dust specks, detector response patterns, etc. For the five supported coronagraphic filters, static flats have been created using either ground-based images (F606W, F814W) or on-orbit “Earth flats” (F330W, F435W, F475W). In the static flats, the occulting spot shadows have been replaced with the corresponding regions from the standard direct-imaging flats for the same filters. These modified static flats are accurate to better than 2% over large spatial scales, but they may be less accurate for pixel-to-pixel response. For all other filters, the standard direct-imaging HRC flats are used instead of static flats, which may cause local errors of up to 10% at the locations of dust spots. Observers should examine the flat fields used to calibrate their data to determine whether any features in their science images are caused by these local flat-field errors.
The second reference flat is a spot flat field (header keyword CFLTFILE
) which contains the vignetted shadows of the occulting spots. There is a different spot flat for each fully supported filter (F330W, F435W, F475W, F606W, F814W). These spot flats must be shifted so that the locations of the shadows match those at the time of the science observations. STScI regularly monitored the location of the small occulting spot using Earth flats. A table of spot position versus time is available as a calibration reference files (header keyword SPOTTAB
). The ACS calibration pipeline finds the spot location in the table that is closest to the observation date and then shifts the spot flat by the required offset. The coronagraphic science image is then divided by the product of the static reference flat and the shifted spot flat.
Although the coronagraph suppresses diffracted light from the occulted source, it does reduce the halo of scattered light created by the polishing errors in HST
’s optics. Scattered light must be subtracted using an image of an isolated star whose color is similar to that of the science target. Observers should observe a PSF reference star immediately before or after their science observations.
Higher-order interpolation should be used because of the high frequency structure in the scattered light halo (Bi-linear interpolation is usually insufficient.) IDL’s INTERPOLATE routine with cubic convolution interpolation and IRAF
using sinc interpolation produce good results. It may be necessary to interleave shift and normalization adjustments. Experience shows that iterative subtraction yields scale factors that are accurate to within 5% and alignments that are accurate to about 0.05 pixels.