The Many Lives of Data: New Science from the Hubble Archive

Rachel Osten,

A good astronomical archive is like a well-stocked wine cellar: the datasets age like fine wines. Far from being relegated to a dusty electronic retirement after fulfilling their primary purpose, archival data can often be repurposed to accomplish different science, which brings new life to the data. As the holdings at the Mikulski Archive for Space Telescopes have grown, such opportunities have increased. These days, the science that can be done with archival data is limited only by the creativity of the astronomer. In many cases, the new science might not have been fully justified in its own right if it had been proposed earlier. Now, however, each new application adds value to both the dataset and the archive. One recent paper, Osten (2012), illustrates the expanding research opportunities with archival datasets. In this example, the original observations were a week-long stare at the Galactic Bulge, intended to discover transiting extrasolar planets. The new use is to study magnetic activity in old stars.

With a large haystack, there are many needles

The Advanced Camera for Surveys (ACS) on board the Hubble Space Telescope has a Wide Field Channel (WFC) with a large field of view (202 × 202 arcsec). With a full complement of filters, ACS/WFC was primarily designed for imaging surveys. The high stability of the instrument further enables precise relative photometry. Coupled with Hubble’s high spatial resolution, these performance factors led Kailash Sahu to devise a proposal for Cycle 12 to use ACS/WFC to search for transits of extrasolar planets in the old stellar population of the Galactic Bulge. The program searched for periodically dimming sources in a sample of 180,000 stars brighter than V = 27. The observations continued for a week, interrupted only by Earth occultations of the spacecraft. The exposures used two broadband filters, corresponding roughly to V and I  bands. The exposures lasted about five minutes, six per Hubble orbit, and the typical spacing between exposures was about eight minutes.

For planet hunting, Sahu and colleagues formed photometric time series from subtracted images, and they searched them for evidence of transiting planets with orbital periods in the range 0.42–4.2 days. Because the sightline towards the Bulge lies in the constellation Sagittarius, they named this dataset SWEEPS, for “Sagittarius Window Eclipsing Extrasolar Planet Search.” Sahu et al. (2006) detailed the initial results: the discovery of 16 planetary candidates, about half of which they thought were genuine.

Only two of the host stars were bright enough to follow-up with the ground-based spectroscopy needed to estimate the stellar mass and place limits on the planetary mass.

Sahu’s team found a population of ultra-short-period planets (periods of less than one day), which occur preferentially around lower-mass stars (less than about 0.9 Msun). The absence of this type of planets around higher-mass stars could be due to irradiated evaporation on timescales less than 10 GY, the assumed age of the Bulge stars in the field. The frequency of planets around stars in the Bulge appears to be similar to that in the solar neighborhood, implying similar mechanisms of planetary formation and migration. The Sahu study also provides an important constraint on the frequency of exoplanets in a second Galactic region—the Bulge—which is quite different from the solar neighborhood.

From early on, other astronomers recognized that the SWEEPS dataset was ripe for spinoff science. The stellar field is rich, containing approximately 245,000 stars brighter than V = 30, with a broad distribution of metallicity. The crowded field facilitates the study of systematic effects due to increased stellar density. Also, the short exposure times, coupled with the relatively long duration of the observations, opens up a wide range of parameter space for time-domain science on the stars themselves. Indeed, the abstract of the original proposal (GO 9750) suggests “a variety of spinoff projects, including a census of variable stars and of hot white dwarfs in the bulge, and the metallicity distribution of bulge dwarfs.” In addition, the SWEEPS dataset enables searches for interesting and rare objects. With such a large haystack, there must be many scientific “needles.”

An immediate offshoot of the SWEEPS dataset was a poster by Sahu at the American Astronomical Society (AAS) meeting in 2006. It described the identification and study of 165 short-period eclipsing binaries with periods in the range 0.2–5 days.

Later, Gilliland (2008) used the data to study photometric variations in a sample of several hundred red-giant stars in the Bulge with luminosities in the range 30–350 Lsun. Most of these stars exhibit variations that can be interpreted as stellar oscillations. Even though the SWEEPS time resolution was not sufficient for asteroseismic analyses, the results established the near ubiquity of oscillations in evolved stars, and they were a harbinger of the rich asteroseismic results to come from the Kepler mission.

Sahu also used the SWEEPS dataset to search for microlensing events, which involve planetary-mass objects. During a star-star lensing event—which is due to a distant source star in the Bulge coincidentally passing behind and apparently near a lens star at intermediate distance—a planetary companion of the lens star can cause a major perturbation of the source-star brightness if the apparent position of the planet comes close to the Einstein ring. The typical timescale of a microlensing event is from hours to days, which is a good match to the SWEEPS dataset. As described in a poster from the 2009 AAS meeting, Sahu found no such events in the SWEEPS data. This result implies that planetary-mass objects from one Earth mass to 10 Jupiter masses make only a small contribution to the total mass budget.

A second epoch of observations of the SWEEPS field occurred roughly two years after the first. Although much shallower—only 2 Hubble orbits compared to 105 orbits in the 2004 program—the short baseline of the new observations, coupled with the high precision of the first-epoch data, nevertheless constrained the proper motions of thousands of stars in the Bulge. Clarkson et al. (2008) presented these results and used them to separate Bulge stars from the disk populations, and to investigate the dynamics of Bulge stars. Later, Clarkson et al. (2011) used the same dataset, together with the proper-motion constraints, to provide the first detection of blue straggler stars in the Bulge. Blue straggler stars are old, hydrogen-burning stars, which have temperatures and luminosities resembling younger stars. They are thought to result from accretion or stellar mergers. Previously, blue stragglers had been seen in a range of older stellar populations, and thus they were expected to occur in the Bulge as well. Before Clarkson et al. (2008), however, the absence of a precise and accurate method to separate Bulge stars from younger disk stars in the field had prevented a firm conclusion. Even so, the new discoveries—at least 18 genuine blue stragglers in a sample of 42 potential blue stragglers—appear to be discrepant from findings in stellar clusters (Clarkson et al. 2011).

Blips, not dips: stellar activity at old ages

Members of the original SWEEPS team led the early follow-on studies of the data. The beauty of the open archive is that, after the initial proprietary period for the data has elapsed, a wider audience can access and utilize the data, and expand the scientific applications even further. As an example, Osten et al. (2012) used the densely time-sampled data to study transient events caused by stellar flares, in a program called “Deep, Rapid Archival Flare Transient Search in the Galactic Bulge” (DRAFTS).

Stellar flares are transient increases in the luminosity of a star, usually of short duration, with timescales ranging from a few minutes to several hours. Flares occur as the result of changing magnetic configurations above the stellar surface where starlight originates. For stars with convective outer envelopes, flares are the most dramatic example of variability observed during residency on the main sequence. Flares occur on many types of cool stars, but appear more frequently in young stars, because their enhanced magnetic activity is driven by the interplay of rapid rotation and other factors of stellar youth.

While stellar flaring is expected to occur primarily in young stars, flares do arise in older stars, too (witness our middle-aged sun’s flare events). Young M dwarfs can flare many times per day, a factor that facilitates further studies using pointed observations. Meanwhile, little is known about flaring in older stars, partly because it is more difficult to observe. The decline in stellar activity at ages greater than 1 GY means that short exposures of single stars will generally not find flaring activity. The SWEEPS dataset nicely fills this gap in parameter space with a huge sample of stars.

The signatures of stellar flares are found across most of the electromagnetic spectrum, from radio wavelengths to high-energy photons. They are observed in the widest range of stellar types at high energies, due to the increasing contrast between the coronal emission from flares and the non-flaring coronal emission from stars at increasingly shorter wavelengths. For this reason, at optical wavelengths, the typically blue flares are most readily observed in M dwarfs, which are intrinsically red.