Productivity and Impact of Space-based Astronomical Facilities

. In 2001, 18 journals published about 1270 astronomical papers that reported and/or analyzed data gathered by space-based observatories and missions. These papers were cited 24,460 times in papers published in 2002–2004, an average of 19.26 citations per paper or 6.42 citations per paper per year (sometimes called impact or impact factor). About 60 satellites, rockets, balloons, and planetary missions were represented, including six ground-based Cerenkov detectors for ultra-high energy gamma rays, because we didn’t know where else to put them. Of these facilities, 21 provided the data for at least ﬁve papers, when credit was divided equally among all contributing facilities. We analyze here distributions of papers, citations, and impact factors among the facilities and among subject areas and compare the results with studies of optical and radio telescopes (Trimble et al. and Trimble & Zaich). Some similarities include the rarity of completely uncited papers (only 41 of 1274, or 3.2%) and the concentrations of the most highly cited papers toward popular topics, high-proﬁle journals, and the most successful telescopes of the year. Some important differences arise because many space-based observatories have lifetimes shorter than the typical time required to think of an interesting astronomical observation, propose for it, get the data, write the paper and publish it (including the ﬁght with the referee), and have citations accumulate. The result is superstar status in citation numbers for XMM-Newton (whose ﬁrst-light package appeared in 2001) and in paper numbers for Chandra (launched 5 months earlier), while aging satellites ( RXTE , BeppoSAX , ASCA ) and the archival-only ROSAT , ISO , IRAS , etc., were still important contributors, but with fewer papers and less highly cited papers. The impact factor of 6.42 for the totality of these gamma-ray, X-ray, ultraviolet, space infrared and optical, and planetary mission papers (6.42) was larger than the corresponding radio (4.52) and optical (5.47) numbers. Notice that HST is included with “optical” but Hipparcos with “space.” A contemplated fourth paper will divide credit for papers and citations among every observatory of any sort that contributed to each published paper.


INTRODUCTION
Bean counting of papers and citations in optical astronomy began in 1981 (Abt 1981). Paper I of this series (Trimble et al. 2005) provided a synoptic view of the papers resulting from optical telescopes and the citations to them by comparing data from 2001 to 2003 with data from 1990 to 1993 (Trimble 1996). An extension to the literature of radio astronomy was fairly straightforward (Trimble & Zaich 2006, Paper II), because radio telescopes, like optical ones, generally stay put for a number of years after their commissioning. The situation for X-ray, ultraviolet, gamma-ray, and other space-based astronomy is by no means analogous, because most missions have lifetimes that are shorter than the "cycle time" for proposals to be written, data acquired, papers written and published, and citations to accumulate.
We nevertheless attempt here a compilation of paper num-bers and citation rates for everything that is not ground-based optical and infrared astronomy, plus HST (Paper I) and not ground-based or satellite radio, millimeter, and submillimeter astronomy (Paper II). This turns out to include the obvious wave bands, but also one optical space-based mission (Hipparcos) and ground-based observations of very high energy gamma rays, because they had not appeared anywhere else. After some cogitation, the decision was made to also include data from solar system missions (Apollo to Galileo) that had contributed to the astronomical literature of 2001, most often through studies of the objects they were aimed at, but sometimes through use of ultraviolet or particle monitors aimed at non-solar-system objects.
No previous analogous studies seem to exist for either radio or space astronomy to provide either guidance or a basis for examining changes. We have made the analyses of Paper II and the present work as nearly identical as possible to that of the optical telescope study in Paper I.

METHODS
Keeping in mind the goal of including all observational astronomy not represented in Papers I and II, V. T. went page by page through all the issues of 19 journals published in 2001 and identified all the papers that reported or analyzed data from any space-based astronomical facility (including balloon, rocket, and shuttle-borne detectors, and Hipparcos, but excluding HST from Paper I and the satellite radio and millimeter missions-HALCA, COBE, etc.-from Paper II). There were 1274 such papers, 217 of which also included optical data and 55 of which also included radio (but no optical) data. Get out your abacus to verify that the total of non-optical papers is therefore 2055, compared to about 2100 with optical data. Half of all observational astronomy, in other words, remained in the wave band of its birth, at least until 2001.
The journals and paper yields were A&A (including Letters, For each paper, the following information was recorded: name of first author, number of additional authors, volume and page number, total number of pages, subject matter (using the same categories as in Paper I), and the identity of all the satellites, missions, detectors, and so forth contributing data to the paper in the order they were mentioned by the authors. For a few papers, it was not possible to determine which facilities were used, and for a few others the subject was unclear. These do not appear in Tables 2 and 3. Assignment of subject was based on what the authors said they had in mind. For instance, a measurement of the D/H ratio might have been aimed at constraining big bang nucleosynthesis ("cosmology") or chemical evolution in our Galaxy ("Milky Way") or fractionation in the interstellar medium ("ISM"). P. Z. then went to the online version (Web of Science) of the Science Citation Index (SCI) and recorded the number of citations to each paper from 2002 to 2004. That we find fewer completely uncited papers than have been recorded in other studies (e.g., Meylan et al. 2004) suggests that the SCI database is somewhat more complete than the Astrophysics Data System (ADS) version. T. B. did some of the citation counting for Paper I and collected the information on birth and death dates that were important to Table 3 and some of the results in § 3.
The most difficult decisions were how to apportion papers and citations among the facilities used for a single paper and which facilities to report individually. The decisions made were those of Papers I and II. That is, equal credit was given to all satellites (etc.) used for a paper, according to the authors. The maximum number was seven (fewer than in the optical and radio cases) for several studies of long-term variability of X-ray sources, with memories stretching back to Tenma (1983Tenma ( -1985 and Ariel V (1974Ariel V ( -1980 in the X-ray and ANS (1974)(1975)(1976) in the ultraviolet, but not to Uhuru (1970( -1975( ) or OSO 7 (1971( -1973. Citations were similarly divided equally, except that even the chief bean counter drew the line at assigning one-seventh of a citation to an ultraviolet balloon. Division was as equal as integers could make it, with one extra given to each of the telescopes mentioned first in the paper to make up the total. Thus, 14 references to three facilities were divided as 5, 5, 4, and so forth. About 80 facilities contributed at least a fraction of a paper to the 2001 literature. Of these, 21 were credited with at least five papers (often made up of assorted fractions) and appear individually in Table 3.

RESULTS
These are divided into small subsections, titled to suggest either what you might have expected or the opposite. They are somewhat different from the subsections of Papers I and II, reflecting a rather different set of phenomena.

Friends Don't Let Friends Go Uncited
For the optical sample, only 133 of the 2100 papers (6.3%) went completely uncited in the 2 years after publication (Paper I), and this drops to 63 of 2100 (3.0%) with 3 years of citations. The radio rate was not very different, 28 zeros out of 836 papers (3.3%) after 3 years (Paper II). And in the space sample, 41 of 1274 papers (3.2%) gathered no moss of citations. The uncited space papers came from both unpopular subdisciplines (e.g., cataclysmic variables) and popular ones (e.g., active galactic nuclei) and from both high and low profile journals. But there were no uncited papers reporting data from XMM, Chandra, FUSE, EUVE, or CGRO.
Proportional uncitedness is even lower for multiwavelength papers: 5 of 220 (2.3%) for spaceϩoptical; and none of the spaceϩradio or opticalϩradio papers scored an impact factor of zero. The percentages of uncited papers represent upper limits to reality, because they include a few substantial ones by well-known authors, on hot topics pursued with major instruments, and published in high-profile journals, indicating that somehow the papers were not being retrieved correctly or were not properly entered in the SCI database. Table 1 represents the other end of the citation spectrum, listing the most-cited papers, ordered by numbers of citations and indicating the journal of publication, the topic, and the facilities used. We list 20 papers (extending down to 100 citations in 3 years) rather than the 10 of Papers I and II, because the top of the list is so heavily weighted by the reports of the instrument package on XMM-Newton. As in the optical and radio cases, these highly cited papers represent a relatively small subset of topics, journals, and facilities. It is worth noting that all the XMM ones come from the "first light" package, which filled an entire issue of Astronomy and Astrophysics with Letters.

One Wavelength's Mite Is Another Wavelength's Poisson?
Table 2 is a slightly cluttered one, dividing up the "space" papers and citations by subdiscipline, with columns for numbers of papers, numbers of citations, citations per paper, and  fraction of all papers in that subdiscipline. In addition, it shows citations per paper and fraction of all papers for the optical sample (updated from Paper I) and the radio one (Paper II). Some subjects are loved and much written about across the full electromagnetic spectrum (active galaxies for instance). On the other hand, white dwarfs and exoplanets belong largely to the optical astronomers, while radio folk are more interested in the interstellar medium, and users of space-based facilities in neutron stars and black holes. Some subjects also have much higher than average citation rates in all three samples. These include cosmology (with many papers), but also gamma-ray bursts, brown dwarfs, and exoplanets (with rather few papers each). Galaxies (normal and active) beat out stars at all wavelengths, although "optical observations of stars" is still the largest single category of paper when the three samples are summed. And binary stars of all sorts (except those with neutron star or black hole components) always do badly in citations per paper. So, curiously, do solar system topics, although less so for the papers reporting data acquired by actually going to the objects of study.

Are We Holding Our Own?
In the case of optical telescopes, bean counting goes back at least to Abt (1981), and Paper I was partly a decade-on update of a 1990-1993 study (Trimble 1996). We are not aware of any previous comparitive study of papers and citations in space-based astronomy. Given the "aging gorilla" phenomenon ( § 3.4), it might not even be very wise to try. It would, however, be possible to go back to the literature of some years when missions other than XMM and Chandra should have been at peak productivity, and parcel out credit for papers and later citations among all the missions (etc.) that contributed to papers in each year. You could do this if we don't get around to it.