The U.S. Optical and Infrared (OIR) System is the term that has been adopted since the 2000 decadal survey, Astronomy and Astrophysics in the New Millennium1 (AANM), for the joint set of astronomical capabilities, public and private, that is available to members of the U.S. astronomical community. The suite of facilities is strongly connected by means of the researchers who successfully make use of these capabilities to conduct world-leading astrophysical research. The telescopes that today constitute the OIR System range up to 10 meters in aperture and 11.8 meters in effective aperture, with current activities aimed at producing two new facilities in the 24- to 30-meter-aperture range. Smaller telescopes are an integral part of the OIR System, since astronomical research often depends on the use of telescopes of different sizes to address different aspects of a question most effectively. An additional component of the System is archival and survey data, the use of which is becoming increasingly important as large surveys continue.
Telescopes in the System
Within the OIR System, the National Science Foundation’s (NSF’s) assets are the National Optical Astronomy Observatory (NOAO) and the U.S. share of
1 National Research Council (NRC), 2000, Astronomy and Astrophysics in the New Millennium, The National Academies Press, Washington, D.C.
the Gemini Observatory. NOAO is the U.S. national OIR observatory, operated as a federally funded research and development center by NSF, and includes the Kitt Peak National Observatory (KPNO) in Arizona and the Cerro Tololo Inter-American Observatory (CTIO) in Chile. NOAO currently provides open access to 4-meter-class telescopes, the Mayall 4-meter and the Wisconsin-Indiana-Yale-NOAO Consortium (WIYN) 3.5-meter in the north and the Blanco 4-meter and Southern Astrophysical Research (SOAR) 4.2-meter in the south. The Dark Energy Survey (DES), a project funded by the Department of Energy (DOE), NSF, and U.S. and foreign institutions, utilizes 30 percent of the time on the Blanco telescope for 5 years in exchange for having built the Dark Energy Camera (DECam). DOE is planning to build and operate the Dark Energy Spectroscopic Instrument (DESI) on the Mayall telescope starting late in this decade.
Beyond operating telescopes, NOAO has a System Science Center, with activities in software and archiving, support for U.S. users of the Gemini telescopes, the NOAO Time Allocation Committee (TAC), and various other activities associated with the OIR System and its evolution. Among these, NOAO ran the Telescope System Instrumentation Program (TSIP) for NSF during the decade of its activity, including soliciting and reviewing proposals from non-federal observatories to build instruments and provide open-access time and providing program management and reporting for those activities. The NOAO TAC routinely reviews proposals for open-access time on many different telescopes, including the U.S. fraction of time on Gemini. NOAO has organized a number of community-wide workshops and studies to understand what capabilities would be needed to carry out decadal survey science and how those capabilities could be provided. These include three public workshops on the entire OIR System as well as the more focused NOAO studies on Renewing Small Telescopes for Astronomical Research2 (ReSTAR) and Access to Large Telescopes for Astronomical Instruction and Research3 (ALTAIR), which looked at subsets of the telescopes in the OIR System. NOAO also held several community-wide workshops to plan instrumentation for Gemini and convened an OIR System Roadmap Committee,4 which examined the demographics and dynamics of the OIR System.
The Gemini Observatory is an international OIR observatory comprising two 8.1-meter telescopes, one on Maunakea, Hawaii, and one on Cerro Pachon in Chile. NSF is now a 65.5 percent partner in Gemini and also serves as the executive agency for the Gemini partnership. The Gemini telescopes are modern
2 National Optical Astronomy Observatory (NOAO), 2007, Renewing Small Telescopes for Astronomical Research, https://www.noao.edu/system/restar/files/ReSTAR_final_14jan08.pdf.
3 NOAO, 2009, Final Report of the Committee on Access to Large Telescopes for Astronomical Instruction and Research (ALTAIR), https://www.noao.edu/system/altair/files/ALTAIR_Report_Final.pdf.
4 NOAO, “Ground-based O/IR System Roadmap Committee,” http://ast.noao.edu/about/committees/system-roadmap, accessed February 1, 2015.
telescopes optimized for infrared (IR) observations and are used predominantly in a queue-scheduled mode. The international Gemini Board makes the selection of instrumental capabilities with input from the international Gemini Science and Technology Advisory Committee, and instruments are designed and built by instrumentation groups in the Gemini partner countries.
The accounting of the contents of the U.S. OIR System outside the federal observatories is not trivial. Table 3.1 lists the telescopes that were considered for this study (2.0-meter or larger aperture), with the fraction of each that are included in the U.S. OIR System (available to proposers at U.S. institutions), as well as the fraction that is available for open-access, peer-reviewed proposals. Some of the open access is restricted to particular science goals. For example, NASA provides support for the Infrared Telescope Facility (IRTF) and the Keck Observatory, which have been important for planetary science observations. Inclusion in the category “open access” is limited to time granted through regular, long-term policies; specifically, residual promised time from the now-defunct TSIP and impromptu or occasional time trades are not included. It is also worth noting that several private telescopes, including Keck, Sloan Digital Sky Survey (SDSS), and the Las Cumbres Observatory Global Telescope (LCOGT), make some or all of their data public after a proprietary period. Many telescopes are operated on behalf of consortia that include foreign partners, and telescopes on foreign soil often must make a certain fraction of their time available to host country astronomers. The effective number of telescopes in each size range is listed. Note that most of them are at non-federal facilities.
These telescopes serve a large and diverse community of privileged users affiliated with institutions that have contributed to their construction, operation, or instrumentation. However, currently only a small fraction of the OIR System is open to observing proposals from all astronomers in the community regardless of their institutional affiliation (19%, 33%, and 8% for large, medium, and small telescopes, respectively). This access is associated with facilities funded by federal agencies, typically NSF but also including NASA and DOE. Figure 3.1 shows the cumulative number of telescopes as a function of aperture, based on the numbers from Table 3.1. While Table 3.1 and Figure 3.1 consider only telescopes with aperture of 2.0 meters or greater, there are also a number of smaller telescopes, some of which provide open access, such as the Small and Moderate Aperture Research Telescope System (SMARTS) telescopes at CTIO. In the Northern Hemisphere, the only open-access small telescope is IRTF, which has near- and mid-IR instrumentation available to the community.
Instruments in the System
The ground-based astronomical landscape includes a mix of facilities with varying sensitivity, field of view, spatial resolution, spectral resolution, wavelength
TABLE 3.1 Telescopes Considered by the Committee
|Observatory/Site||Aperture||U.S. Fraction||Open Fraction|
|Large Telescopes (6-12 meters)|
|Large Binocular Telescope (LBT)||Mt. Graham, Arizona||11.8a||0.50||0.00|
|Keck 1||Maunakea, Hawaii||10.0||1.00||0.17|
|Keck 2||Maunakea, Hawaii||10.0||1.00||0.17|
|Hobby-Eberly Telescope (HET)||McDonald Observatory, Texas||9.2||0.89||0.00|
|South African Large Telescope||South African Astronomical Observatory, Sutherland, South Africa||9.2||0.40||0.00|
|Gemini N (Gillette)||Maunakea, Hawaii||7.9||0.69||0.60|
|Gemini S||Cerro Pachon, Chile||7.8||0.59||0.60|
|Magellan (Baade)||Las Campanas, Chile||6.5||0.90||0.00|
|Magellan (Clay)||Las Campanas, Chile||6.5||0.90||0.00|
|MMT||Mt. Hopkins, Arizona||6.5||1.00||0.00|
|Effective fractional number of telescopes||7.97||1.54|
|Medium Telescopes (3.5-5 meters)|
|Hale Telescope||Palomar Observatory, California||5.1||1.00||0.00|
|Discovery Channel Telescope||Happy Jack, Arizona||4.3||1.00||0.00|
|SOAR||Cerro Pachon, Chile||4.2||0.70||0.30|
|Blanco Telescope||Cerro Tololo, Chile||4.0||0.90||0.90b|
|Mayall Telescope||Kitt Peak, Arizona||4.0||1.00||1.00|
|ARC 3.5 m||Apache Point, New Mexico||3.5||1.00||0.00|
|WIYN||Kitt Peak, Arizona||3.5||1.00||0.40|
|Effective fractional number of telescopes||7.80||2.60|
|Small Telescopes (2-3 meters)|
|Shane||Lick Observatory, Mt. Hamilton, California||3.0||1.00||0.00|
|Harlan Smith||McDonald Observatory, Texas||2.7||1.00||0.00|
|DuPont||Las Campanas, Chile||2.5||0.90||0.00|
|Sloan Foundation (SDSS)||Apache Point, New Mexico||2.5||1.00||0.00c|
|Hiltner||Kitt Peak, Arizona||2.4||1.00||0.00|
|WIRO||Jelm Mtn., Wyoming||2.2||1.00||0.00|
|Bok||Kitt Peak, Arizona||2.2||1.00||0.00|
|UH 88-inch||Maunakea, Hawaii||2.2||1.00||0.00|
|Otto Struve||McDonald Observatory, Texas||2.1||1.00||0.00|
|KPNO 2.1 m||Kitt Peak, Arizona||2.1||1.00||0.00|
|LCOGT||Siding Spring, Australia||2.0||0.75||0.00|
|Effective fractional number of telescopes||12.40||1.00|
a LBT is two coupled 8.4-meter telescopes with the equivalent area of a single 11.8-meter telescope.
b This is for pre- and post-DES.
c Sloan is a survey instrument; all Sloan Digital Sky Survey (SDSS) I-III data are now public.
NOTE: Acronyms are defined in Appendix C.
FIGURE 3.1 Number of telescopes with aperture greater than a given size, as a function of aperture. The black line shows the number in the system, including public and private facilities, while the red line shows the number publicly available.
coverage, and quality. Some instruments and even telescopes are quite specialized, while others are more general purpose. The demand for time on any particular telescope is a function of several of the above factors, as well as the special capabilities of the instrumentation and the match of the instrumentation to timely science. New instruments can make an otherwise obsolete telescope relevant again. State-of-the-art instrumentation is expensive, however, and limited resources must balance new needed capabilities against redundancy to address the level of demand.
The largest telescopes all carry sizable and diverse instrument complements, typically a mix of general-purpose and special-purpose capabilities. Optical and near-IR imaging and spectroscopy capabilities exist at almost every facility. Multi-object spectroscopy (MOS), integral field unit (IFU) spectroscopy, or both are often available. While there is duplication, there is also significant demand for the generic capabilities. Most of the largest telescopes have adaptive optics (AO) systems. Table 3.2 shows the distribution of these instrumental capabilities, classified into general categories for optical, near-IR, and mid-IR wavelength regimes, based on a survey sent to the individual observatories.
The medium- and smaller-aperture telescopes have become, in many cases, more specialized than in the past and relative to larger telescopes. They tend to have fewer instruments, tailored toward the unique strengths of each facility. The smaller telescopes are in some cases used for larger projects or surveys, but they also provide testbeds for new instrumentation and for more innovative operations strategies, such as remote or robotic operation. Smaller telescopes are also used for time domain studies, so are likely to be of value in the Large Synoptic Survey Telescope (LSST) era. For example, the SMARTS 1.5-m and the 1.3-m telescopes, accessible through the NOAO peer-review proposal process, are useful for transient studies and for synoptic observations.
A critical aspect of a System view of the facilities is the extent to which their development is planned and carried out in a way that acknowledges and takes advantage of the relationship among them. The past 20 years have seen sev-
TABLE 3.2 Instrumentation on 6- to 12-meter Telescopes
NOTE: Acronyms are defined in Appendix C.
eral efforts to provide some coordination and integration to the OIR System. The McCray report, A Strategy for Ground-Based Optical and Infrared Astronomy5 first recommended an NSF-funded program aimed at enlarging open access and simultaneously providing new resources for instrumentation at the non-federal telescopes. This approach was subsequently endorsed by AANM and ultimately implemented as TSIP. Over a period of 10 years, TSIP provided 453 new, open-access nights on large telescopes and helped to enable the construction of 13 new instruments for these telescopes.
Between 1996 and 2010, a group called ACCORD (the AURA [Association of Universities for Research in Astronomy] Coordinating Council of Observatory Research Directors) met periodically to discuss matters of common interest among the operators of medium- and large-aperture telescopes, including both federal and non-federal facilities.6 This group operated without any specific authority outside of the control of the individual observatories. The goal of ACCORD was to develop consensus perspectives that could be presented to funding agencies and to policy-making and strategic planning committees. In addition, ACCORD supported some activities of broad benefit to the community, such as formulating an adaptive optics development roadmap.7 Such community-wide planning is an important way to develop the U.S. OIR System.
See Section 6.3 for a discussion and recommendation about future system planning activities.
Impact of Resources through Diverse Usage
The combined resources represented by the U.S. OIR System are considerable. A justification for characterizing these facilities as a single system rather than a collection of individual telescopes with overlapping capabilities is the way that they are used to carry out scientific research. Unlike many academic pursuits where the primary resources are privately owned and reside in personal laboratories, libraries, or computers, the field of observational astronomy necessarily relies on shared resources in the form of multipurpose telescopes and data archives. The larger the number of researchers having access to these principal tools and their products, the more diverse are both the questions pursued and the science that is accomplished.
Furthermore, the vast majority of U.S. researchers depend on access to multiple facilities, whether open or proprietary. Ground-based data are often combined with
5 NRC, 1995, A Strategy for Ground-Based Optical and Infrared Astronomy, National Academy Press, Washington, D.C.
6 ACCORD became defunct in 2010 following the release of NWNH and the transition to the current NSF AST director.
7 “A Roadmap for the Development of United States Astronomical Adaptive Optics,” 2008, http://www.aura-astronomy.org/news/AO_Roadmap2008_Final.pdf, accessed February 1, 2015.
space-based data, and many studies are conducted across a range of wavelengths, as noted in Chapter 2. Large projects are carried out by large teams, in which some of the participants bring access to private facilities to the collaboration. The NOAO Ground-Based OIR System Roadmap Committee Survey8 conducted in 2011 states that 74 percent of the survey respondents were satisfied with the capabilities that they could currently access through these dynamic collaborative mechanisms. Note that this statement predates the decision by NSF to divest several of the federal telescope assets in the Northern Hemisphere as recommended by the PRC.9
This NOAO Roadmap Survey revealed that most U.S. researchers obtain and use the data they need from multiple available sources (see Figure 3.2; use of multiple telescopes by individuals is indicated by connecting lines and their emphasis). The most heavily used facilities (used by more than 20% of respondents), were, in order of usage, Keck 2 10-meter, SDSS 2.5-meter, Gemini North 8-meter, Keck 1 10-meter, NOAO/KPNO Mayall 4-meter, Gemini South 8-meter, MMT 6.5-meter, NOAO/CTIO Blanco 4-meter, and Magellan-Baade 6.5-meter.
The small, medium, and large telescopes listed in Table 3.1 account for more than 9,000 nights of telescope time yearly available to U.S. proposers and annually generate approximately 2,000 refereed publications (corrected by the U.S. fraction). The number of individual principal investigators (PIs) from U.S. institutions, summing the lists from these observatories, exceeds 1,500 per year. The number of people—scientific, technical, and administrative—employed to support the subset of these observatories that are primarily U.S. organizations is more than 700. Appendix B provides demographic data from observatories that responded to a request by the National Research Council for input on the numbers of employees in science, technical, and engineering positions; the numbers of proposals accepted; total users; and resulting publications.
Through broad participation of the professional astronomical community comes an increased chance for high-impact discovery. Survey data such as images, photometry, or limited spectroscopy from the Digitized Palomar Observatory Sky Survey (DPOSS), Two Micron All Sky Survey (2MASS), SDSS, and Wide-Field Infrared Survey Explorer (WISE) have been available to all researchers and lead to the high impact of these projects (Appendix B shows 600 papers per year based on SDSS data). Among PI-generated observations at pointed (as opposed to survey) telescopes, the productivity (papers published per telescope) and the impact of the
8 NOAO, 2012, Ground-Based O/IR System Roadmap Committee Community Survey Summary of Results from U.S. Based/Sponsored Respondents, http://ast.noao.edu/sites/default/files/SummaryDocumentSystemRoadmapCommunitySurveyV1.5.pdf.
9 National Science Foundation (NSF), 2012, Advancing Astronomy in the Coming Decade: Opportunities and Challenges. Report of the National Science Foundation Division of Astronomical Sciences Portfolio Review Committee, http://www.nsf.gov/mps/ast/portfolioreview/reports/ast_portfolio_review_report.pdf.
FIGURE 3.2 This figure, from the National Optical Astronomy Observatory Ground-Based Optical and Infrared (OIR) System Roadmap Committee report submitted to the Portfolio Review Committee, shows the relative number of common users of different telescopes through the thickness of the lines (red or black) connecting the facilities. The thicker the line, the more common users. For example, many researchers used both Gemini North and Keck 1 and 2 during the period covered by the survey (2008-2011). The relative number of U.S. users of each telescope in the U.S. OIR System from 2008 to 2011 is indicated by the size of the ellipse. The U.S. OIR community makes extensive use of all the U.S. OIR system facilities, independent of their access status, through scientific collaboration, time exchanges, and open-access programs. Facilities colored yellow received significant federal funding in support of either instrumentation or operations or other related activities of the facility. There were 1,178 respondents to the survey. Complete survey results are available at http://ast.noao.edu/about/committees/system-roadmap. SOURCE: Courtesy of B. Jannuzi (University of Arizona) and the National Optical Astronomy Observatory/Association of Universities for Research in Astronomy/ National Science Foundation.
papers (defined as the ratio of the number of citations for a paper relative to the mean or median citation count among all papers that year) have been tracked for all the major (above 3.5 meters in size) OIR ground-based telescopes worldwide.10 Considering data between 2008 and 2012, Keck, the Very Large Telescope (VLT), the United Kingdom Infrared Telescope (UKIRT), the Canada France Hawaii Telescope (CFHT), and Subaru are consistently at or near the top of the productivity
10 D. Crabtree, 2014, A bibliometric analysis of observatory publications for the period 2008-2012, Proceedings of the SPIE 9149, Observatory Operations: Strategies, Processes, and Systems V, 91490A.
rankings, while HET, Keck, UKIRT, CFHT, and Magellan lead rankings in impact per paper. The total impact per telescope (defined as the average impact per paper times the number of papers from that telescope) is led by Keck, UKIRT, CFHT, VLT, and Subaru. For Keck, over a 3-year period about 75 percent of the time on the two 10-meter telescopes went to PIs at the university partners Caltech, University of California, and University of Hawaii, and the other 25 percent went to 238 programs with PIs at 57 different institutions; in 2012, 39 percent of the lead authors of Keck AO papers were from non-partner institutions, illustrating the diverse community associated with the high productivity and large impact of this facility.11 Keck, CFHT, Magellan, and Gemini each have a large number of unique PIs per year (see Appendix B), while the lowest impact facilities had a factor of 10 fewer.
CONCLUSION: Interest from and telescope usage by a large, diverse, and active community of high-quality researchers is correlated with high-impact scientific output.
Archives are increasingly important in both space-based and ground-based astronomy for the long-term science return of facilities. There are now more papers published based on Hubble Space Telescope (HST) archival data than papers based on the original observations.12 There are many archives in the United States associated with different observatories and long-term projects, with different levels of output. While space-based archives are a routine part of NASA missions, archives for ground-based data are not uniform.13 Ground-based public archives are the most well developed for large surveys such as SDSS.14 For the most part, OIR facilities with individual PI science programs have not provided public archives (with the exception of Gemini, Keck, and, more recently, NOAO). Chapter 5 of New
11 See white paper by Cohen and Martin; note that as indicated in Table 1, public access to Keck is now reduced from 25 to 17% with the cancellation of TSIP. (J. Cohen and C. Martin, 2014, “The Crucial Role of W.M. Keck Observatory in the U.S. Astronomical System,” white paper submitted to the committee).
12 NASA, Hubble Space Telescope, 2011, “Hubble Racks up 10,000 Science Papers,” December 6, http://www.nasa.gov/mission_pages/hubble/science/10k-papers.html.
13 NSF, 2012, Advancing Astronomy in the Coming Decade.
14 The SDSS project had approximately three times the publication rate of the most productive of the facilities mentioned above, the pair of Keck telescopes. SDSS led the citation rankings of all facilities in 2009 (J.P. Madrid and D. Macchetto, 2009, High-impact astronomical observatories, Bulletin of the American Astronomical Society 41:913-914).
Worlds, New Horizons in Astronomy and Astrophysics15 (NWNH) included a section “Data and Software,” subdivided into “Data Archives” and “Data Reduction and Analysis Software.” The “Data Archives” section emphasized the high science return from the public archives for HST, SDSS, and 2MASS, and the high expected return from future large archives such as that for LSST. NWNH noted that the Virtual Observatory (VO) has established data archiving standards that should enhance the value of archival data sets. It also noted that NASA includes data handling and archiving as an integral component of its space missions. Here is a brief summary of representative current large archives from U.S. ground-based OIR facilities.
Examples of Current Ground-Based OIR Data Archives
2MASS was the first digital, OIR all-sky survey from the ground, started in 1997 and completed in 2001 and conducted at near-IR J, H, Ks wavelengths. Its data products include a digital sky atlas with 4 million 8-arcminute x 16-arcminute fields, a point source catalog with positions and fluxes for 300 million stars and other unresolved sources, and an extended source catalog with positions and magnitudes for more than 1 million galaxies and nebulae. The data are publicly available through the Infrared Processing and Analysis Center (IPAC).
All OIR raw data and metadata such as observing logs are publicly available after 12 months via the Canadian Astronomy Data Centre (CADC); no catalogs are published.
In collaboration with NASA, Keck makes OIR raw data from DEIMOS, ESI, HIRES, KI, LRIS, MOSFIRE, NIRC, NIRC2, NIRSPEC, and OSIRIS instruments publicly available after a default proprietary period of 18 months via the Keck Observatory Archive through the NASA Exoplanet Science Institute at IPAC. Pipe-lined “quick-look” data products are also available for some instruments. No catalogs are published.
15 NRC, 2010, New Worlds, New Horizons in Astronomy and Astrophysics, The National Academies Press, Washington, D.C.
NOAO Science Archive
The NOAO Science Archive provides access to two types of data products. Relatively high-level data products (typically uniformly reduced, calibrated, reprojected, and stacked images but not catalogs) from most of the surveys carried out with NOAO facilities are available for public use. The larger fraction of the holdings includes all raw data taken by every NOAO instrument since 2004. Individual reduced images are included for the few instruments (wide-field imagers) with data reduction pipelines. Proprietary access is provided immediately to investigators with ongoing programs; public access is provided for all data after a default proprietary period of 18 months.
Sloan Digital Sky Survey
SDSS includes ugriz data and spectra. The last SDSS-III release, DR12,16 includes over a billion catalogued objects and 5 million spectra from four surveys (BOSS [Baryon Oscillations], MARVELS [precision radial velocities], and SEGUE and APOGEE [galactic structure in the optical and H-band, respectively]). SDSS-IV is now under way. The raw data, the pipeline-reduced data, and science-ready catalogs are all made publicly available after a 12-month proprietary period through a web-enabled database hosted at Johns Hopkins University via a 5-year, but possibly renewable, grant.
The fully automated private Palomar Transient Factory (PTF) and Intermediate Palomar Transient Factory (iPTF) optical g-band and R-band wide-field survey obtained with the Palomar Oschin 48-inch Schmidt telescope is archived at IPAC. There was a small public data release in the spring of 2014, with the rest of the PTF data to follow in 2015 and iPTF in 2016.
Las Cumbres Observatory Global Telescope
LCOGT provides optical image and source catalogs through an agreement with IPAC; currently 870,000 images and 2 billion photometric measurements are available. Data are proprietary for the initial 12 months.
There are also a number of European-led survey projects whose data are world-public. These include UKIDSS (in Y, J, H, and K filters, currently with a catalog of 84 million objects measured in the Large Area Survey, 63 million in the Galactic
16 S. Alam, F.D. Albareti, C.A. Prieto, F. Anders, S.F. Anderson, B.H. Andrews, E. Armengaud, É. Aubourg, S. Bailey, J.E. Bautista, R.L. Beaton, T.C. Beers, et al., 2015, The eleventh and twelfth data releases of the Sloan Digital Sky Survey: Final data from SDSS-III, arXiv:1501.00963 [astro-ph.IM].
Clusters Survey, and 700 million objects in the Galactic Plane Survey), iPHAS (r, i, and H-alpha data currently for 200 million galactic plane sources), and UVEX (U, g, r filter data, currently for 200 square degrees). There are six public infrared surveys from the VISTA 4-meter telescope that are also slowly becoming available to the U.S. community. VLT data become publicly accessible after a proprietary period.
The Virtual Observatory
The International Virtual Observatory Alliance (IVOA) has defined a number of standards and protocols to facilitate the interoperation of data centers and to permit the exploitation of heterogeneous, distributed collections of astronomical data; it is not a data repository in its own right. The U.S. Virtual Astronomical Observatory (USVAO; formerly the National Virtual Observatory) was a nationally funded project to use these technologies; however, USVAO has been terminated, and HEASARC, MAST, and IPAC have taken responsibility for maintaining its core functions. LSST has a baseline plan of using VOEvent to broadcast its alerts. An unfunded consortium of interested organizations has formed the USVAO to maintain a U.S. representation at the IVOA, and the American Astronomical Society Working Group on Astronomical Software is forming a VO special interest group for discussions and collaborations within the U.S. astronomical community on VO standards.
NWNH recommended that “Proposals for new major ground-based facilities and instruments with significant federal funding should be required as a matter of agency policy to include a plan and if necessary a budget for ensuring appropriate data acquisition, processing, archiving, and public access after a suitable proprietary period.”17 This is a worthy goal for all major surveys from both public and private facilities. Starting in 2011, NSF required applicants for funding to include a data management plan in their grant proposals.
A second issue raised in NWNH and reiterated by the PRC report18 was long-term data curation, with the following recommendation: “NSF, NASA, and DOE should plan for effective long-term curation of, and access to, large astronomical data sets after completion of the missions or projects that produced these data, given the likely future scientific benefit of the data. NASA currently supports widely used curated data archives, and similar data curation models could be adopted by NSF and DOE.”19 While the archiving of data is a long-established part of the culture of astronomy, continuing to enhance the curation and accessibility of data is
17 NRC, 2010, New Worlds, New Horizons, The National Academies Press, Washington, D.C., p. 31.
18 NSF, 2012, Advancing Astronomy in the Coming Decade.
19 NRC, 2010, New Worlds, New Horizons, p. 147.
consistent with new and pending federal mandates concerning access to the results of federally funded research. NSF released its public-access plan for availability of published papers in March 2015.20
CONCLUSION: Consistent with NWNH recommendations and federal mandates, a data archive that is publicly accessible and well curated is a commendable central goal for every major survey from a public or private facility.
Finally, NWNH noted that general-purpose community analysis software packages for data reduction and processing have aged and were not designed for the era of big data. There is thus a concomitant need for development of and investment in new, flexible, modular packages. New general-purpose software toolkits will be needed, especially when LSST comes online.21 The adoption of standard VO interfaces and metadata by different archives would help facilitate searches and analyses involving multiple databases.22
These issues are important to address in the coming era of even larger databases.23 Listed below are examples of some of the many planned archives, culminating in a discussion of the LSST archive.
Planned Data Archives
DECam (including DES and Open Use time)
There are two classes of data taken with the DECam on the CTIO Blanco 4-meter, both of which will ultimately end up in the NOAO science archive: DES data and community data resulting from NOAO time allocations. Raw and calibrated pixels in g, r, i, z, and Y filters are made available after a proprietary period for each (the proprietary period for DES images is 12 months). In addition, there are two planned releases of co-added, calibrated images and catalogs for the DES
20 NSF, 2015, Public Access Plan: Today’s Data, Tomorrow’s Discoveries: Increasing Access to the Results of Research Funded by the National Science Foundation, March 15, http://www.nsf.gov/pubs/2015/nsf15052/nsf15052.pdf.
21 S. Oey, P. Price, L. Hartmann, J.U. Monnier, and C.U. Miller, 2014, “Enabling Science: OIR System Software Tools,” white paper submitted to the committee.
22 See the white paper by Drory et al. discussing integration of archiving practices (N. Drory, M. Shetrone, and N. Gaffney, 2014, “Software and the US OIR System,” white paper submitted to the committee).
23 See the following presentations to the committee on October 12, 2014, available at http://sites.nationalacademies.org/BPA/BPA_087934#presentations: David Silva, National Optical Astronomy Observatory, “NOAO Today and Tomorrow”; Richard White, Space Telescope Science Institute, “Data Panel Discussion”; George Helou, California Institute of Technology, “Archiving Ground-Based Data: Perspective from Space”; Mario Juric, Large Synoptic Survey Telescope.
data, one based on the first two observing seasons and the second following the full five seasons of data. It is expected that these will be archived and served in the short term by the National Center for Supercomputing Applications (NCSA), which leads DES Data Management. For the longer term, these DES data releases will be curated by NOAO.
The Panoramic Survey Telescope and Rapid Response System (Pan-STARRS) survey from the PS1 telescope in Hawaii has repeatedly imaged 75 percent of the observable Northern Hemisphere in g, r, i, z, y passbands since 2010. Its full catalog will be publicly released in April 201524 through an interface at the MAST. The current PS1 database includes 35 billion measurements for 6.5 billion objects.
Zwicky Transient Facility
The Zwicky Transient Facility (ZTF), the follow-up to PTF and iPTF, will produce synoptic 976 megapixel images in R-band and g-band obtained with the Palomar Oschin 48-inch Schmidt telescope. Raw and calibrated pixels will be archived. Photometry will be cataloged for all objects seen in the frames. IPAC at Caltech will process and host the data, with an interface through the Infrared Science Archive (IRSA). Thanks to NSF Mid-Scale Innovations Program (MSIP) funding, approximately half of the ZTF data are expected to be public, starting in 2018, with staged data releases thereafter. Transient alerts will be broadcast publicly in near-real time beginning in the third year of operation.
The LSST archive, to be housed at the NCSA for user access, will accumulate approximately 20 TB of ugrizy data per night, generating for the archive 60 PB of raw data and 15 PB of catalog data over the 10-year lifetime.25,26 Level 1 data27 and
24 See K. Chambers, 2015, “Pan-STARRS and the Future of Optical/IR Sky Surveys in the Northern Hemisphere,” white paper submitted to the committee.
25 PB = petabyte = 1015 byte = 1 million billion bytes = 1 million gigabytes; TB = terabyte = 1012 bytes = 1 thousand gigabytes.
27 In the context of LSST, the term Level 1 is used to mean the following: Nightly processing resulting in catalogs of objects whose positions or fluxes have changed, resulting in alerts broadcast worldwide after 60 seconds. This includes solving for orbits of solar system objects (Large Synoptic Survey Telescope, 2013, “LSST Data Products Definition Document,” http://ls.st/dpdd).
Level 2 data28 are project deliverables. Data products will include calibrated and co-added images, per-epoch object catalogs, and static-sky object catalogs. These catalogs will contain stellar and galaxy photometry, positions (and parallaxes and proper motions for stars), and shapes measured with enough fidelity to be used to measure cosmic shear. LSST will generate alerts for transients that will be made public within 1 minute, along with enough information for the recipients to act on the alert. Although LSST will not provide a sophisticated alert broker (see Section 5.2), the system is expected to provide information to help external brokers classify alerts (e.g., ugrizy postage stamps and the complete light-curve of any object detected at that position). Once a year (twice in the first year) there is expected to be a release of all the raw and processed data and the transient and static-sky catalogs. There will be approximately a 1-year delay from the time that the last data are taken until the release. This data release will be made available to everyone in the United States and Chile and to LSST’s international partners. The release will be in the form of a sophisticated database, although the project will also support access using traditional formats. The LSST project will generate well-calibrated databases that can be used for a broad range of science without further data processing.
CONCLUSION: LSST will accumulate more than 20 TB of data per night during an anticipated 10-year lifetime. The LSST project will generate sophisticated, well-calibrated databases that will enable many projects without further data processing. Generating higher-level data and algorithms is not part of the LSST project charge.
LSST will provide a data center to serve alerts, images, and catalogs. In addition, 10 percent of the center’s resources (CPU cycles and database storage) will be available to the community to support additional activities, such as computing and saving additional parameters for a subset of brighter galaxies. Additionally, the LSST data center at NCSA will be colocated with publicly available petascale computing facilities, and it should be possible to apply for time to carry out extensive analyses that require more than the 10 percent of its resources that LSST will make available.29 This will be important for the community, since downloading
28 Level 2: Yearly processing of all the data taken to date, including de-blending sources and optimal processing of multi-epoch data, resulting in calibrated catalogs of positions, fluxes, and shapes for objects of sufficient quality to enable a wide range of science without returning to the pixels. The Level 2 catalogs will include characterization of object variability (parallaxes, proper motions, summaries of Lomb-Scargle periodograms) as well as flux measurements optimized for various purposes (aperture, psf, Petrosian, model), and structural parameters for simple galaxy models (e.g., 2-compo-nent constrained Sersic indices) and some derived quantities such as photo-z catalogs.
29 This need for computer center access is particularly crucial for researchers at smaller institutions (C.T. Liu, B. Willman, J. Pepper, M. Rutkowski, D. Norman, K. Cruz, J. Bochanski, H. Lee, J. Isler, J. Gizis, J.A. Smith, et al., 2014, “Maximizing LSST’s Scientific Return: Ensuring Participation from Smaller Institutions,” white paper submitted to the committee).
large data sets to local computing resources is not practical. Generating higher-level algorithms and data products for science beyond the primary mission is not part of the LSST project’s scope. These Level 3 data30 activities are important for extracting the maximum science from NSF’s large investment in this forefront instrument; DOE is already funding efforts to prepare for Level 3 processing aimed at its scientific goals.
CONCLUSION: LSST will provide a data center to serve alerts, images, and catalogs, with 10 percent of the center’s resources (CPU cycles and database storage) reserved for the community. The data center will be colocated with publicly available petascale computing facilities at NCSA.
With 10 million time domain events per night, orbits for 6 million solar system bodies, 3 billion galaxies, and over 10 billion stars, the LSST archive will be a treasure trove for research. For many astronomers, analyzing these data will be a new mode of doing science. The astronomical community needs to develop algorithms and procedures for data processing and analysis for specific scientific programs. Useful activities to build this competence include developing training networks (see Section 3.4 for discussion and a recommendation in this regard), establishing data challenges, and applying LSST algorithms to existing and forthcoming large data sets such as DES, PS1, and PTF/ZTF. This practical work will help to standardize data protocols,31 equip the community to use LSST data, and produce substantial scientific results.
CONCLUSION: LSST will use standard protocols to serve data where available (e.g., VOEvent) and will work with the community to evolve and establish future standards.
CONCLUSION: Making effective use of petabyte-scale databases (“big data”) requires new skills, and the astronomical community working in this area needs to continue to develop algorithms and procedures for data processing and analysis to take advantage of the next generation of data sets.
CONCLUSION: The scientific return from large surveys (both ground- and space-based) would be maximized if their data and catalogs were made
30 Level 3: Further processing of the Level 1 or 2 catalogs or the pixels, enabling more specialized studies. For example, GALFIT-style modeling of bright galaxies, generating catalogs of interacting galaxies, or co-processing LSST images with near-IR data.
31 From the LSST page: “The content of the alert packet itself will be formatted to confirm to VOEvent (or other relevant) standard. Where the existing standard is inadequate for LSST needs, LSST will propose extensions and work with the community to reach a common solution” (Large Synoptic Survey Telescope, 2013, “LSST Data Products Definition Document,” http://ls.st/dpdd).
widely available using standard protocols, with appropriate data products made available for copying or downloading when possible. Because of the volumes of data involved, the centers serving the data would be most useful if appropriate public computing cycles and storage were available to users to take data-intensive analysis to the data instead of requiring redundant copies of the data on local computing resources.
Training in instrumentation, software, observing, and data analysis is an essential aspect of the astronomical enterprise.32 Classroom training in basic instrumentation and data acquisition and analysis is often part of the undergraduate curriculum for astronomy and physics majors. More specialized training occurs both at the undergraduate and graduate levels within research groups, either at universities or national laboratories such as NOAO and the NASA centers, or as part of an NSF Research Experiences for Undergraduates (REU) program. Many students may not know where their true interests and talents lie without trying several different kinds of projects, thus broadening their experience and skill sets. As a result, advanced technical training may be somewhat haphazard and often occurs toward the end of the student experience. As astronomical instrumentation and software grow increasingly complex, there is a greater need for specialization in these fields and with it, a need for earlier and more systematic training.
Training in Instrumentation
Many graduate programs look favorably upon applicants with expressed interests in instrumentation. There are currently 40 departments in 38 institutions in the United States offering a Ph.D. in astronomy, astrophysics, or planetary sciences.33 Of these, half have instrumentation programs in OIR; radio, millimeter,
32 Training has been discussed in several of the recent reports that consider the OIR System: NSF, 2012, Advancing Astronomy in the Coming Decade: Opportunities and Challenges, report of the National Science Foundation Division of Astronomical Sciences Portfolio Review Committee, August 12, http://www.nsf.gov/mps/ast/portfolioreview/reports/ast_portfolio_review_report.pdf; NOAO, 2009, Final Report of the Committee on Access to Large Telescopes for Astronomical Instruction and Research (ALTAIR), https://www.noao.edu/system/altair/files/ALTAIR_Report_Final.pdf; NOAO, 2007, Renewing Small Telescopes for Astronomical Research, https://www.noao.edu/system/restar/files/ReSTAR_final_14jan08.pdf; NRC, 1995, A Strategy for Ground-Based Optical and Infrared Astronomy, National Academy Press, Washington, D.C.
33 S. Nicholson and P.J. Mulvey, 2014, Roster of the Physics Departments with Enrollment and Degree Data, 2013, Focus On, August, American Institute of Physics Statistical Research Center, http://www.aip.org/statistics/reports/roster-physics-2013.
and submillimeter (RMS); or spaceflight instrumentation (including rocket-borne instruments). Several departments create a holistic training experience for their students by engaging them in instrument testing and technology development, observing, and analysis,34 and have used observatory access to train future instrument builders.35 Smaller telescopes play an important role in providing testbeds for instrumentation and training for instrument builders, somewhat analogous to that of university-scale experiments within NASA’s suborbital program. By way of example, the Texas Robotic Optical Transient Search Experiment telescope and the Palomar 60-inch have been used effectively as testbeds for robotic operation.
The NSF Portfolio Review Committee warned that “it will be difficult to attract and retain the next generation of instrument builders.”36 and NWNH noted “the opportunities for training students in instrumentation have declined precipitously over the past 20 years. Training for the next generation of instrumentalists is most efficient when there is a steady state hierarchy of project sizes, so that people can progress from relatively smaller, simpler, and faster projects to responsibilities in larger and more complex activities.”37 As the complexity and the timescale for developing instruments for major telescopes grows, it is increasingly difficult to find projects that are sufficiently self-contained yet also offer substantial opportunities to develop students’ and postdocs’ skills. The beginning of new instrument projects does not always dovetail or overlap with the completion of old ones, so appropriate projects are not always available for students.
There are a number of actions that could help nurture and train young instrumentalists. These include offering training networks (see the end of this section for a recommendation in this regard), maintaining small instrument programs for undergraduate departments to develop talent for graduate programs, running instrumentation workshops and NSF REU programs concentrating on instrumentation, and fostering diverse populations in instrument groups.38
Training in Software
Astronomical software involving observations falls into two broad categories: that needed to run the hardware and take raw data, and that needed to render raw data astronomically useful. The former category requires not only software
34 S.E. Tuttle, H. Lee, C. Froning, and M. Montgomery, 2014, “Builders Instead of Consumers: Training Astronomers in Instrumentation and Observation,” white paper submitted to the committee.
35 J. Cohen and C. Martin, 2014, “The Crucial Role of W.M. Keck Observatory in the U.S. Astronomical System,” white paper submitted to the committee.
36 NSF, 2012, Advancing Astronomy in the Coming Decade, p. 71.
37 NRC, 2010, New Worlds, New Horizons, p. 149.
engineering expertise, but also adequate understanding of the hardware and the ability to interface with the instrument builders. The latter category characterizes and removes (in most cases) instrumental signatures and proceeds to produce tables of astronomical quantities—calibrated spectra, fluxes, positions, and for extended objects, model parameters and shapes; this is often referred to as pipeline processing. Pipeline processing requires a sophisticated understanding of the characteristics of the raw data, the scientific problems that it can be used to elucidate, and the trade-offs involved in algorithmic and implementation choices. There is a third category of software that takes pipeline products and produces quantities of interest for one or another specific astronomical study, sometimes involving outputs from multiple pipelines.
There are unsolved problems at all stages in the processing. Most of the algorithmic innovation in these pipelines comes from people trained as astronomers. Building a successful production pipeline requires software engineering as well as astronomical expertise, and pipelines are typically masterminded by astronomers who have chosen to specialize in algorithms, often working within software groups or large collaborations.
Learning the art of astronomical software often occurs within the context of research. Students involved in instrumentation projects often participate in the design and development of control and acquisition software as well as hardware. Because pipeline expertise is not widespread in the astronomical community, and because the complexity of modern algorithms and codes is not widely appreciated, many students are self-taught, and many never acquire more than the rudiments of software engineering. Therefore, a more systematic approach to algorithm training could be valuable.
Training in Observation
Large surveys such as 2MASS, SDSS, and Gaia (and eventually LSST) provide products of great value to astronomers, but surveys are not the whole story in astronomy. As practitioners in the field—at all levels—are becoming less connected to raw data acquisition and more reliant on data products available in archives, the need for training in observational techniques and associated data uncertainties is less obvious but remains important. Such training will assure not only appropriate use and acquisition of data today, but also help in the design of next-generation instruments and facilities.
Small OIR telescopes, equipped with basic instrumentation, are available on many campuses.39 Learning observing techniques on these is adequate for pro-
39 The telescope maker DFM, for example, has sold over 100 telescopes with apertures 0.4-1.0 meter to colleges and other organizations and individuals (per DFM Engineering, Inc.).
grams such as basic imaging or spectroscopy, which benefit from the proliferation of well-designed hardware and software systems, including efficient observing scripts, data reduction pipelines, and open-access analysis suites.
The most demanding observational programs still require the presence of the observer making real-time observations with the telescope. Even the most detailed specifications planned in advance of an observing run account only partially for the variations in observing conditions and instrument state that are typical of ground-based observing. Students therefore have much to learn from active participation in real-time decision making and data acquisition at the telescope for tasks such as the alignment of multi-slit masks and coronagraphs, altering priorities on a target list in response to quick-look analysis of incoming data, deciding after a first observation whether an object warrants a larger investment of telescope time, or changing programs entirely based on seeing or weather conditions.
Many premier research telescopes are accessible to Ph.D. students through competitive proposals. Some students also have privileged access via their institutional affiliations. Despite this, student access to the largest telescopes, equipped with more complex instrumentation requiring more sophisticated experimental design, is often limited. This limitation is in part because such telescopes are heavily oversubscribed and also because they are logistically more complicated to use. Insofar as students are increasingly removed from instrument operation and data acquisition, they are at increasing risk of failing to appreciate the shortcomings of the data with which they may work and less well positioned to write persuasive and effective proposals for telescope time.
Cross-Institutional Training in Critical Areas
Coursework and on-the-job apprenticeships offered by individual institutions are sometimes supplemented by cross-institutional workshops and “summer schools” (which are not necessarily held in the summer). These are particularly well suited to more advanced training for which teaching resources may not be available, either because of small enrollments or limited expertise. Radio astronomers have long used such workshops and schools to teach aperture synthesis techniques. European countries have integrated these methods into their astronomical culture via long-term multi-institutional collaborations (known as training networks) involving graduate students, postdocs, and faculty members intended to provide methodological instruction through research; examples in astronomy include networks aimed at preparing for weak lensing40 and Gaia41 data. An important feature
of these networks is that their activities are spread out over the course of 2 or more years, giving a cohort time to develop and solidify.
The U.S. OIR community has made relatively little use of training networks, although there are some well-established annual series (e.g., the adaptive optics workshops at Santa Cruz,42 the astroinformatics and astrostatistics workshops run by Penn State and others,43 the Santa Barbara Modules for Experiments in Stellar Astrophysics (MESA) workshops,44 NASA’s Sagan Exoplanet Summer Workshops,)45 and recently, LSST community workshops;46 however, none aim to train a generation of students in the science and technology of emerging fields. NOAO has sponsored many workshops for the community.47All of these workshops are specialized one-time meetings that do not develop a cohort of experts the way training networks can.
The small numbers of builders of astronomical instruments and writers of astronomical software make these segments of the OIR ecosystem particularly vulnerable to major changes in the astronomical landscape. There is concern that the current downturn in funding is squeezing out a future generation of instrument builders. The writers of astronomical software face a potential embarrassment of riches, with the prospect of more data available than people able to render it useful to the astronomical community. It is essential to provide adequate training in order to fully exploit future data sets. One solution to help train future generations would be a connected sequence of schools specifically aimed at training students in, for example, the data reduction and analysis skills relevant to LSST48 and, more generally, big data science. As astronomy moves to more research based on
43 See submitted comments by Borne (K. Borne, “Comments on Data Science Methods,” white paper submitted to the committee). The white paper by Loredo et al. from the LSST Informatics and Statistics Science Collaboration emphasizes the need to have broader educational efforts in addition to summer schools and the creation of interdisciplinary funding programs; the coauthors are a mix of astronomers and information scientists (T.J. Loredo, J. Babu, K.D. Borne, E. Feigelson, P. Freeman, J. Hilbe, Z. Ivezic, C. Schafer, and A. Siemiginowska, 2014, “Astronomical Information Sciences for O/IR Synoptic Survey Astronomy,” white paper submitted to the committee).
46 Large Synoptic Survey Telescope, “LSST Project and Community Workshop,” https://project.lsst.org/meetings/lsst2014/, accessed March 1, 2015; Large Synoptic Survey Telescope, “LSST and NOAO Observing Cadences Workshop,” https://project.lsst.org/meetings/ocw/, accessed March 1, 2015.
47 NOAO, “National Optical Astronomy Observatory Affiliated Meetings,” http://ast.noao.edu/activities/meetings-colloquia/noao, accessed February 1, 2015.
48 L. Walkowicz, A. Connolly, Z. Ivezic, M. Juric, V. Kalogera, C. Lintott, P. Marshall, and M. Strauss, 2014, “Software Training Networks in the LSST Era,” white paper submitted to the committee.
pipeline-reduced data sets, it becomes even more important that the consumers understand the processing algorithms in order to understand the strengths and limitations of the catalogs. Some aspects of Level 3 LSST processing (e.g., transient alert brokers) could be structured as training networks competed by NSF, with student and postdoc development as one of the deliverables.49 Such schools might likewise be appropriate for advanced topics in instrumentation, software visualization tools, and analysis.50
CONCLUSION: Specialized training in general observing, instrumentation, software, and data analysis techniques is essential for ensuring that the next generation of astronomers has the requisite skills to accomplish the best science.
RECOMMENDATION: The National Science Foundation (NSF) should support a coordinated suite of schools, workshops, and training networks run by experts to train the future generation of astronomers and maintain instrumentation, software, and data analysis expertise. Some of this training might best be planned as a sequence, with later topics building on earlier ones. NSF should use existing instrument and research programs to support training to build instruments.
Behind every great observational scientific breakthrough there is an instrument builder. Without new tools for discovery, the ability to push the frontiers in astronomy would be significantly compromised. Increasingly, scientific instruments require an enormous investment in software to exploit the data; LSST, like the Large Hadron Collider (LHC), is as much a software project as a hardware project. It is imperative to sustain talent in these important areas in order for the U.S. astronomical community to remain competitive.
Beyond graduate study, the career paths for instrument builders and software specialists are less well-defined than for theorists and observers. In order to be
49 See also the white paper by Willman et al. on training and Level 3 efforts (B. Willman, K. Olsen, J. Bochanski, N. Brandt, A. Burgasser, W. Clarkson, M. Cooper, K. Covey, H. Ferguson, E. Gawiser, M. Geha, et al., 2014, “Enabling a Diverse User Community to Produce Cutting-Edge Science with LSST,” white paper submitted to the committee).
50 See white papers submitted to the committee by Liu et al. on training (C.T. Liu, B. Willman, J. Pepper, M. Rutkowski, D. Norman, K. Cruz, J. Bochanski, H. Lee, J. Isler, J. Gizis, J.A. Smith, et al., 2014, “Maximizing LSST’s Scientific Return: Ensuring Participation from Smaller Institutions”) and by Drory et al. specifically on software training (N. Drory, M. Shetrone, and N. Gaffney, 2014, “Software and the US OIR System.”
professionally competitive, students in these fields essentially complete two theses: one instrumental or software-oriented and one focusing on the analysis, or what some would consider “the science.” Many institutions make sure that instrumentation students have independent ownership of their projects and publish their results in journals specifically for instrumentation. However, the lengthy timeline for big instrument projects often leads to fewer publications for instrumentalists. This puts them at a disadvantage for prize postdocs and tenure advancement. The development timescale for big projects is much longer than a student or postdoc career, and it is more difficult for them to play central roles because groups cannot accept risk of failure or slips in schedule. These difficulties disincentivize potential instrument builders. Similarly, the increased importance of software to the astronomical enterprise, and the creativity and astronomical knowledge involved, is not always reflected in postdoctoral and faculty appointments.
CONCLUSION: Long timescales for complex projects and oversubscription of instrument funding lines discourage early career specialization in instrumentation.
It is worth noting that while there may be fewer academic tenure-track jobs for instrument builders and software experts, these graduates rarely have trouble finding employment outside of astronomy because they have outstanding technical training. While producing highly skilled individuals who leave astronomy adds high-level workers to society, that is not the primary motivation for training students in astronomy graduate programs. There is a perception inside and outside the United States that other countries are doing a better job at providing stable funding for engineers and instrumentalists. There are widespread views in the instrumentation community that many talented young instrument builders in the United States are following jobs to other countries. Because scientific productivity rests upon the success of instruments that are used at telescopes, the U.S. OIR community cannot risk losing this talent and expertise and ceding leadership to other countries.
A well-known challenge for instrumentation is that it is expensive to maintain a laboratory infrastructure that includes machine shops; project managers; and optical, mechanical, electrical, and software engineers. Often, laboratory personnel can be sustained only when universities make this a priority by allocating long-term funding; in many cases universities are either closing down labs or scaling back personnel. In addition, the community is increasingly at risk of losing small and medium telescopes (private and public) that provide testbeds for instrumentation.
There are also new challenges that are directly related to the increasing cost, size, and complexity of new instruments for 6- to 10-meter telescopes and the next generation of giant telescopes. The cost of observatory instruments now constitutes a significant fraction of the cost of the telescope. Seed money, typically needed to make a project competitive for funding through the Major Research Instrumenta-
tion (MRI) program, has become increasingly scarce. The Advanced Technologies and Instrumentation (ATI) and MRI programs are oversubscribed by greater than 5:1, and their budget lines are inadequate to support the rising costs of instruments. The MSIP program is a new line of support but is underfunded relative to the number of excellent proposals,51 both on high-priority science recommended in decadal surveys and on new avenues of exploration. There is often a lead time of several years to obtain funding for instruments, which then take several years to complete and commission along with the writing of pipeline software.
CONCLUSION: NSF ATI and MRI funding is inadequate to support the rising cost of small and medium instrument projects.
Some financial support for instrumentation projects has emerged from federal agencies outside NSF, such as DOE and NASA. Over the past 30 years, the DOE National Laboratories have become increasingly engaged in ground-based OIR projects, which provide an avenue for both instrumentalists and software experts. Major projects under way or about to be undertaken include DECam by Fermilab in concert with NOAO, the LSST camera by Stanford Linear Accelerator Center (SLAC) in concert with NSF, and DESI by Lawrence Berkeley National Laboratory (LBNL) in concert with NOAO. The principal contributions of the DOE laboratories and their associated university scientists have been in instrument development and construction and to a lesser extent data processing. The teams at the national laboratories are program-oriented and can be expected to move on to non-astronomical projects when each program is completed. Some individuals, both from the labs and from the universities, may choose to continue pursuing astronomy, bringing their expertise with them.
Because instrument development on private facilities is more driven by PIs than by agency or national programmatics, there is still an important role for universities and small labs to play, especially with respect to training for later involvement in the larger-scale projects. NSF can have a significant impact through its research grants and instrumentation programs. Review panels recognize ideas that are creative. Many good ideas do percolate up through the grants program. However, it is more difficult to identify funding sources to explore the high-risk
51 A presentation by Jim Ulvestad to the AAAC in November 2014 notes that there were 38 preproposals requesting $398 million; 12 full proposals were invited, and 2 full awards and 1 development award were funded, with about $14 million in annual funding for FY2014 and FY2015 (http://www.nsf.gov/attachments/130395/public/Ulvestad_NSF-AST-AAAC-2014Nov17.pdf); see also the NSF Mathematical and Physical Sciences Directorate’s response to NWNH recommendations: NSF, 2014, NSF Division of Astronomical Sciences (AST) Report, http://www.nsf.gov/attachments/130395/public/Ulvestad_NSF-AST-AAAC-2014Nov17.pdf; NSF, 2015, “Dear Colleague Letter: Status of MPS/ AST Response to Recommendations of New Worlds, New Horizons Decadal Survey,” March 4, http://www.nsf.gov/pubs/2015/nsf15044/nsf15044.pdf.
but potentially transformative technologies that are needed for the United States to remain competitive.52
CONCLUSION: There is inadequate funding for instrumentation programs. This is largely the result of the increasing cost and complexity of instruments for the next generation of telescopes, with funding gaps between projects. The increased complexity of instruments also requires stronger engineering and project management components than in the past, and it is rare to have this as part of the training in astronomy instrumentation programs.
CONCLUSION: The need to complete complex expensive projects means that less funding is going toward explorative technology development.
52 See the white papers submitted to the committee highlighting the need for instrument development and training as well as funding for transformational instrumentation and technologies: T.E. Armandroff, 2014, “Input from McDonald Observatory to the Committee on a Strategy to Optimize the U.S. OIR System in the Era of the LSST for Questions 3 and 10” and J.D. Monnier, J.T. Armstrong, M.J. Creech-Eakman, S.T. Ridgway, T.A. ten Brummelaar, and G.T. van Belle, 2014, “Funding Technology Development and Novel Instrumentation Today in Order to Enable Breakthrough Observing Techniques Tomorrow.”