Studying the Atmosphere On the Way to Probing Dark Energy

May 5, 2015

Among the many opportunities in the LSST project, it necessitates a new understanding of our own atmosphere. LSST science depends on photometric redshift determination, which in turn depends on accurate measurements of the flux from celestial objects. At wavelengths where our atmosphere glows, this presents a novel challenge.

The LSST filter bands, showing total system throughput

For the current and next generation of astronomical survey to probe dark energy, the issue of photometric redshifts is increasingly crucial.  With millions of galaxies to be imaged, it is not possible to obtain time-consuming spectra to reveal line features, the traditional manner of determining redshifts.  Instead, the redshift of every galaxy must be estimated by the different flux in each of several broad filter bands.  Explorations of many of the issues with photometric redshift determination is an important research area in which a number of KIPAC scientists, including David Burke, Brian Gerke, Marina Shmakova, Jack Singal, Patrick Kelly, and others, have been involved.

For the Large Synoptic Survey Telescope (LSST) project, which needs to determine the redshifts of an unprecedented number of galaxies (on the order of a billion), to higher redshifts than have traditionally been undertaken, photometric redshift techniques are critical.  Unlike most other optical surveys, LSST will include a filter band at wavelengths longer than 950 nm, well into the near infrared range that is redder than the human eye can see.  This will allow the important spectral features from high redshift galaxies to still be within a filter band of the system, rather than being redshifted entirely out.  The band, longward of 950 nm, receives the designation "y", and is made possible by new deep depletion CCD technology which provides CCD response at those wavelengths.

However, the inclusion of a y filter band raises complicated new issues.  In that region of the spectrum, there are many emission and absorption lines originating from a variety of molecules in the atmosphere.  Unlike the traditional optical bands, which have been explored for years and contain fewer lines, there is not much data available on the magnitude, variability, and complex time dependence of the behavior of the y band atmospheric lines.  For a survey like LSST, which intends to use total measured fluxes in the y band as crucial information for determining redshifts of distant objects, the issues resulting from these lines in the atmosphere must be thoroughly understood.

There are several proposed band pass shapes for the y filter, designated y2, y3, and y4.  The figure shows the y4 band shape (along with the other LSST bands, and including the instrument and atmosphere response), while y2 and y3 have less response at the longest wavelengths.  The choice of which y filter to use involves a trade-off between increased response to celestial objects and increased level of overall atmospheric signal.

KIPAC astronomer Kirk Gilmore, along with three colleagues from Harvard and the University of Hawaii, has carried out a program of precision measurements of the sky brightness in the y band, at the future LSST site in the Andes mountains.  They used a CCD camera along with a filter and lens system to achieve a system throughput similar to that for the proposed LSST y3 filter.  Measuring the sky brightness variability due to the atmospheric lines to the levels needed is a difficult procedure because of the challenges of precisely calibrating the system response.

In a paper to appear in the Proceedings of the Astronomical Society of the Pacific, they report the detection of significant spatial and temporal variation in the atmospheric line emission.  They observe a pattern consistent with the presence of travelling and standing density perturbations in the atmosphere analogous to ocean waves.  These atmospheric waves change the relative atmospheric brightness of a given spot by around 4% as they move across the sky, while the overall brightness in the band varies by a factor of 2 during the course of a few days due to the bulk whole-sky changes in atmospheric conditions at the site.  They also conclude that the relative variability would be the same for the other proposed y band filter response shapes. 

Gilmore and colleagues suggest that LSST would benefit from a concurrent y band sky brightness monitor of the type they employed, and a strategy to observe in y band only when the sky is darkest there, so that the relative effect from the spatial variability due to the atmospheric waves is minimized.  They estimate that, because of the atmospheric waves, more than 15 individual y band images will be needed for a given field of view to achieve uniform differential flux calibration to the percent level in the field.  This information about atmospheric emission in the y band will be essential to developing a successful observing plan for LSST, and for carrying out data analysis with LSST science images.  It is a testament to the interconnectedness of science that the precision characterization of dark energy in the farthest reaches of the cosmos that will be enabled with LSST depends critically on understanding things that are happening just several miles above the ground. 

This work is based in part on a paper submitted to Proceedings of the Astronomical Society of the Pacific, and is available from astro-ph at

Science Contact:
Kirk Gilmore