0 Assigned ID: 1854 ----- Attempting Submission 1 (Fri Mar 01 22:16:47 GMT 2013) ----- The scientific output of any instrument hinges upon the quality of its calibration. Modern astronomical experiments rely on their instruments to be extremely precisely calibrated; these demands will only increase in upcoming missions, along with demands to reduce calibration overheads. This proposal is to explore new kinds of calibration methods that make use of science data (rather than specifically designed calibration data). We propose to develop a new form of probabilistic self-calibration and apply it to all of the science data taken with the WFC3 IR instrument through the F110W and F160W filters. Our method relies on the fact that in the full corpus of MAST, many similar kinds of astronomical sources (like stars) touch every part of the detector. We can therefore construct a probabilistic model for the entire imaging data set, deviations from which are plausibly described as calibration errors. The method also relies on sensible priors from existing calibration information and astronomical knowledge. Our project will deliver a new, pixel-to-pixel flat field for the two filters, verifying the existing flat (which is based on pre-launch data), and potentially improving the current calibration. After achieving the baseline goals we will explore higher-order issues like persistence and time variations in sensitivity. RESOLVED STELLAR POPULATIONS Default Medium true false false false false SUPPORTED