Cosmic structures woven together during the tug of war between gravity and dark energy present a multi-faceted challenge for scientists, as we seek to untangle each galaxy from the luminous cacophony of filaments and clusters across large swaths of space and time.
We love staring at the beautiful images taken by the Dark Energy Camera (DECam) at the Blanco telescope. The image above shows a cluster of galaxies laid on a backdrop of even more distant galaxies. To investigate the mysteries of the accelerating expansion, Dark Energy Survey (DES) scientists need to do a bit more – we need to develop a comprehensive census of the content across the universe: how many stars and galaxies are there in a given swatch of space-time fabric?
A critical step comes in creating a high-fidelity and detailed list of the observed celestial objects: these are called “catalogs” by astrophysicists and astronomers. The most common pieces of information are the position and brightness: this is the minimum information necessary to know where a galaxy resides in spacetime.
With our hard-working scientists in the data management team and the powerful computers at National Center for Supercomputing Applications (NCSA), DES has developed new algorithms and pipelines for efficiently sifting the objects out of our images. We start with raw images straight from DECam, and then we refine them to remove artifacts, like satellite trails, cosmic rays and faulty pixels. From these “reduced” images, we must then find and characterize discrete objects, like galaxies and stars – cut the wheat from the chaff.
However, there is a limit to what we can do. For example, a very far-away object may appear extremely small and faint – so faint that it will look like a piece of the sky and get missed during the cataloging procedure. In some cases, it is not possible to tell the difference between a faint object and a noisy patch of sky. In addition, not every astronomical object is “willing” to be cataloged: it can be disguised as a part of another object. For example, near the center of today’s image, there is a very large, bright galaxy with many smaller neighbors. Discerning all the objects here is similar to the difficulty one might have in noticing a flea in a picture of an elephant.
Objects also tend to hide from the computers when a piece of the sky is full of them: spotting a small object becomes as difficult as finding Waldo (Wally) on a crowded beach!
DES takes more detailed images than previous projects, like the Sloan Digital Sky Survey (SDSS). Thus, we are more pestered by the “hiding” objects problem. We see a more tangled web. As one solution, a group of DES scientists have employed an image restoration algorithm, derived from work by computer vision scientists. This algorithm successfully eliminates the impact of close neighbors when cataloging the “hiding” objects. Upon application to DES images, they have been able to find many “Waldos,” so we can add them to DES catalogs.
For more detailed description of the method, you can find a preprint of the paper here: http://arxiv.org/abs/1409.2885.
Det.’s Yuanyuan Zhang and B. Nord
Image: Det.’s Marty Murphy and Reidar Hahn