Jan. 14, 2021 — Before DESI, the Dark Energy Spectroscopic Instrument, can begin its 5-year mission from an Arizona mountaintop to produce the largest 3D sky map yet, researchers first needed an even bigger 2D map of the universe.
The 2D map, pieced together from 200,000 telescope images and several years of satellite data, lacks information about galaxy distances, and DESI will supply this and provide other useful details by measuring the color signatures and “redshift” of galaxies and quasars in its survey. Objects’ redder colors provide telltale information about their distance from Earth and about how quickly they are moving away from us – and this phenomenon is known as redshift.
In the end, this 2D map of the universe is the largest ever, based on the area of sky covered, its depth in imaging faint objects, and its more than 1 billion galaxy images.
The ambitious, 6-year effort to capture images and stitch them together for this 2D map – which involved 1,405 observing nights at three telescopes on two continents and years of data from a space satellite, an upgraded camera to image incredibly faint and distant galaxies, 150 observers and 50 other researchers from around the world. The effort also required 1 petabyte of data – enough to store 1 million movies – and 100 million CPU hours at Berkeley Lab’s National Energy Research Scientific Computing Center (NERSC).
2D map sets the stage for DESI observations, with a goal to solve dark energy mystery
“This is the biggest map by almost any measure,” said David Schlegel, co-project scientist for DESI who led the imaging project, known as the DESI Legacy Imaging Surveys. Schlegel is a cosmologist at the U.S. Department of Energy’s Lawrence Berkeley National Laboratory (Berkeley Lab), which is the lead institution for the international DESI collaboration.
The map covers half of the sky, and digitally sprawls over 10 trillion pixels, which is equivalent to a mosaic of 833,000 high-res smartphone photos. The DESI collaboration has about 600 participating scientists from 54 institutions around the world.
Publicly viewable at legacysurvey.org/viewer, the Sky Viewer map includes 2 billion objects – more than half of which are galaxies – and numerous clickable filters to select from specific object types or surveys. Some of the objects are individually labeled, and viewers can choose to display constellations, for example, and galaxies and quasars that will be imaged by DESI. Quasars are among the brightest objects in the universe, with supermassive black holes at their center that emit powerful jets of matter.
DESI is equipped with an array of 5,000 swiveling, automated robots, each toting a thin fiber-optic cable that will be pointed at individual objects. These cables will gather the light from 35 million galaxies and 2.4 million quasars during the five years of DESI observations.
DESI will collect and transmit data from these measurements to Berkeley Lab’s NERSC from Kitt Peak. Researchers at NERSC have already prepared for this incoming data by identifying which data-processing tasks would take up the most computing time and improving the code to speed up these tasks on the center’s current- and next-generation supercomputers. “In the end, we increased processing throughput five to seven times, which was a big accomplishment – bigger than I expected,” said Laurie Stephey, a data analytics engineer at NERSC who played a key role in the effort.

The primary purpose of compiling the 2D map data is to identify these galaxy and quasar targets for DESI, which will measure their light to pinpoint their redshift and distance. This will ultimately provide new details about mysterious dark energy that is driving the universe’s accelerating expansion.
Software guides observing plan, and standardizes and stitches imaging data
Piecing together all of the DESI surveys’ images to create a seamless sky map was no trivial task, Schlegel explained. “One of the goals is to get a really uniform image by stitching together multiple observations,” he said. “We started out scattershot. And cameras do have gaps – they miss stuff. Part of the challenge here was planning the observing program so that you could fill in all of the gaps – that was a huge logistical challenge. You have to make sure it is as homogeneous as possible.”
The three surveys that comprise the DESI Legacy Imaging Surveys conducted imaging in three different colors, and each survey took three separate images of the same sky areas to ensure complete coverage. This new, ground-based imaging data was also supplemented by imaging data from NASA’s Wide-field Infrared Survey Explorer (WISE) satellite mission, which collected space images in four bands of infrared light.
For the Legacy Imaging Surveys’ data-taking effort, Schlegel designed a code, improved over time, that helped to calculate the best approach and timing for capturing the best images to completely cover half of the sky, considering hours of darkness, weather, exposure time, planetary and satellite paths, and moon brightness and location, among other variables.
Dustin Lang, DESI imaging scientist at the Perimeter Institute for Theoretical Physics in Canada, played a key role in standardizing all of the imaging data from ground- and sky-based surveys and stitching it together.
In some images, Lang noted, “the sky might be really stable and calm,” while on another night “we might have light clouds or just a turbulent atmosphere that causes blurring in the images.” His challenge: to develop software that recognized the good data without diluting it with the bad data. “What we wanted to think about is what the stars and galaxies looked like above the atmosphere,” he said, and to make sure the images matched up even when they were taken under different conditions.
Lang created “The Tractor,” a so-called “inference-based” model of the sky, to compare with data for the shape and brightness of objects imaged by different surveys and to select the best fit. The Tractor drew heavily upon supercomputer resources at Berkeley Lab’s NERSC to process the Legacy Imaging Surveys’ data and ensure its quality and consistency.
Imaging data from 3 surveys seeds other science research
Arjun Dey, the DESI project scientist for the National Science Foundation’s NOIRLab, which includes the Kitt Peak National Observatory site where DESI is situated, was a major contributor to two of the three imaging surveys, serving as the lead scientist for the Mayall z-band Legacy Survey (MzLS) carried out at Kitt Peak, and as co-lead scientist with Schlegel for the Dark Energy Camera Legacy Survey (DECaLS) carried out at a NOIRLab site in Chile.
The third DESI-preparatory survey, known as the Beijing-Arizona Sky Survey or (BASS), was conducted at Kitt Peak and supported by an international collaboration including the Chinese Academy of Sciences and the University of Arizona.
Researchers from China made more than 90 trips to Kitt Peak to carry out observations for BASS, which was supported by an international collaboration including the National Astronomical Observatories of China (NAOC) and the University of Arizona. “A joint research team of more than 40 people from 11 institutes in China and the U.S. participated in BASS and contributed to the success of this data release,” said Hu Zou, an astrophysicist at the Key Laboratory of Optical Astronomy in Beijing and a co-lead investigator for BASS. “This team will also play an important role in the future of the DESI survey and related sciences,” he added.
The collective effort of the three surveys, Dey said, “was one of the most uniform, deep surveys of the sky that has ever been undertaken. It was really exciting to participate.”
All of the raw data from the imaging surveys has been released to the scientific community and public. This final data release, known as Data Release 9 or DR9, has been preceded by eight other data releases. The data have already spawned several disparate research projects, including citizen science efforts that utilize the wisdom of crowds.
The Mayall Telescope at Kitt Peak National Observatory and the Blanco Telescope at Cerro Tololo Inter-American Observatory are operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation. The Bok Telescope is located on Kitt Peak and operated by Steward Observatory, University of Arizona. The authors are honored to be permitted to conduct astronomical research on Iolkam Du’ag (Kitt Peak), a mountain with particular significance to the Tohono O’odham.
This research used resources of the National Energy Research Scientific Computing Center, which is supported by the U.S. Department of Energy Office of Science under Contract No. DE-AC02-05CH11231.
For more details, please see https://www.legacysurvey.org/acknowledgment/.
Click here to read the full article.
Source: Glenn Roberts Jr., LBL