|dc.description.abstract||Large-scale, high-resolution, photometrically calibrated images are key for many astrophysical problems. The INT Photometric Hα Survey has imaged the entire northern Galactic Plane in r, i and Hα filters. However, these images suffer from a number of common imaging problems, including, most critically, large-scale gradients due to scattered moonlight. The objective of this work is to produce an automated method for cleaning this data so that it can be used to produce large-scale and reliable Hα mosaics for scientific use.
We created dark-time templates to account for airglow, fringing, and other sources of dark-time counts in the images and then used a Markov Chain
Monte Carlo method to fit a linear, 2-dimensional model to the scattered moonlight. Bright stars in the images are censored from the fitted images so they do not influence the fit. Other types of model were explored, as well as a method that employed Fourier transforms to clean the data, but without fruition. The method to fit the model to the moonlight background was originally tested in the i-band, before moving onto the r-band, subtracting scaled Hα images to remove nebulosity. An empirical scaling factor was then used to translate the model fit from the r-band to the Hα band, necessary because of varying atmospheric conditions.
Finally, the cleaned data were shifted onto a common zero point before mosaicking into large scale images. The result is a strong groundwork for cleaning astronomical images by accounting for the various components to sky background but preserving features of interest. The results of this process applied to images that cover supernova remnant Simeis 147 show a substantial improvement over uncleaned imaging data. We also illustrate the versatility of this process by applying it, unprepared, to other regions in the Galactic Plane.||en_US