Image Simulation with Shapelets

Abstract

We present a method to simulate deep sky images, including realistic galaxy morphologies and telescope characteristics. To achieve a wide diversity of simulated galaxy morphologies, we first use the shapelets formalism to parametrize the shapes of all objects in the Hubble Deep Fields. We measure this distribution of real galaxy morphologies in shapelet parameter space, then resample it to generate a new population of objects. These simulated galaxies can contain spiral arms, bars, discs, arbitrary radial profiles and even dust lanes or knots. To create a final image, we also model observational effects, including noise, pixellisation, astrometric distortions and a Point-Spread Function. We demonstrate that they are realistic by showing that simulated and real data have consistent distributions of morphology diagnostics: including galaxy size, ellipticity, concentration and asymmetry statistics. Sample images are made available on the world wide web. These simulations are useful to develop and calibrate precision image analysis techniques for photometry, astrometry, and shape measurement. They can also be used to assess the sensitivity of future telescopes and surveys for applications such as supernova searches, microlensing, proper motions, and weak gravitational lensing.