Operation of CFHT as a mature facility, with many instruments used and in development and with shrinking resources, faces several constraints which become significant only in relation to the scientific productivity of the telescope. Some preliminary results of a database of publications, their citations and observing nights, which is facilitating such a comparison, are presented. The future expected operational constraints are also briefly commented on.
When SAC has proposed the title of this presentation, I was somewhat embarrassed because the purpose of CFHT Users' Meeting is essentially to open windows to new projects and new dreams and I was given a role to moderate or perhaps even limit the potential of the dreams of our scientific community. I think that the danger CFHT is facing presently in the era of 8-metre class telescopes is in acceptance of the constraints without a realistic interpretation of our ``dreams''.. We have to develop for the next ten years a plan with a competitive set of instruments and a reasonable operating budget that the agencies could afford.
In order to prepare and implement a plan that will follow the present five year plan (1996 - 2000), we need to identify clearly what are the operational constraints. One of the tasks will be to reduce the pressure and manpower overload resulting from the decrease in the number of staff positions decided in December 1995, coupled with the continuous increase in the number and complexity of instruments and modes in which they are operated. In spite of these, we should not abandon our ``dreams'' and continue acquiring the best possible set of instruments which will maintain CFHT competitive and easier to operate.
First I'll explain database tools we are developing for a more quantitative appraisal of the CFHT scientific productivity, of the use of our manpower, and of the operating costs for the different available foci. This should shed light on the basic question of how much we are paying for the complexity of our observing environment. Then I'll stress that current situation has existed since the start of CFHT: We have been trying to achieve more developments, operate more instruments with more complex functions, all that with a shrinking manpower and a progressively reduced operating budget. Based on that experience, which still seems to have permitted to maintain high standards of CFHT, I'll recommend to prepare for the future by complementing MEGACAM with a wide field IR camera (WIRCAM) at f/8 and to define a plan for the next ten years including new scheduling and observing modes, and possibly some transformations in our telescope environment which would allow easier operations and some improvement of the dome seeing (ventilation). These last items will be expanded in Dennis Crabtree's presentation.
In order to evaluate the scientific productivity of CFHT in a quantitative way, we have developed a database of all publications in refereed journals based on CFHT observations. The librarian scans regularly the major journals and publications cited in observing time requests, preprints and articles we receive them directly from observers. The database may still miss some papers at we consider it complete at the 95% level. We encourage everyone to check the list of publications, which is available here on our Web page and to inform us if it needs to be corrected or updated. A CFHT policy mentioned to each PI when the telescope time is allocated requests the observers to send CFHT any published paper if based on the CFHT observations; however, we experience problems with enforcing the policy which is not followed by all observers. We need help of all users of CFHT in completion of the database.
At present, the list of the publications covers years 1990 - 1997, but we will probably go back to earlier years once we consolidate the current database. We will consult the citation index of all these publications on the yearly basis with an intention (to be discussed with the SAC and the Board) to make the results available on the Web. The database includes the name of instrument, focus and the ``runID'', the latter in cases when we were able to clearly identify the observation runs used by the authors of the paper. Setting up of the database would be simpler if editors of the journals could agree on a standard set of key parameters to be used in presentations of observations in order to retrieve easily the data from various archives (these could include eg. telescope, date of observation, runID, instrument, focus, detector, ... ).
We are also developing a database of semester schedules and we have been able to get the statistics of the use of different instruments and focus from 1987 to 1998. Soon we will complete it to extend from the beginning of the observations at CFHT. We have only one problem to track modifications of the schedule done in various technical emergency situations. We intend to have an automatic and reliable means of keeping the data of the actual schedule, including such items as the telescope time effectively used for science, or for engineering or lost due to poor weather. Relating of the publications to the schedule databases gives meaningful statistics. We have the total number of publications per year between 1990 and 1996, the total number of citations of these papers as determined in October 1997 as a function of the year of publication and we can compare those with the number of scheduled nights for each focus or for each instrument.
 |
The average time between observations and publication amounts to three years. On average, we have 78 publications per year, 20 citations per paper five years after publication. The ``productivity'' for each instrument or focus could be crudely evaluated by the total number of citations of papers published between 1990 and 1996 divided by the total number of nights scheduled for the focus (or the instrument) which was used in a given publication. The most interesting and best established results are shown in Figures 1 and 2. We will have to check and update our database to confirm the trends visible in the figure, but we think that we have already a meaningful basis to present these as preliminary results. Clearly the prime focus, followed by f/8, appear to be the most productive foci. The prime focus is by far the most productive in term of citations per observational night. We are not presenting the similar statistics per instrument because we are not yet sure that it would be fully meaningful due to the limited number of nights scheduled, to the limited lifetime of some instruments or to the limited number of publications which have been published for some new instruments at the time we consulted the citation index.
 |
Concerning the new instruments: The arrival of MOS in 1993 contributed substantially to an increase in the rate of publications and citations for the f/8 focus and we expect the effect to be amplified with the full use of MOS, OSIS and the AOB. On the contrary, UH8k has not yet any contribution in our statistics. It should be noted that a small number of particularly significant papers are currently highly cited (>100 citations after a few years), these are mainly the survey papers such as CFRS, CNOC, or the UH redshift survey.
CFHT remains second after AAT in terms of the overall productivity, as was shown by Trimble tri95 (1995) on the basis of papers published in 1990 - 1991. Our database confirms that CFHT is still firmly on the second place after AAT. One can ask if there exists more scientific pressure for one particular instruments or one particular focus. The unexpected answer is NO! On the basis of the statistical analysis of the numbers of nights requested, ranked and scheduled by TAC one finds that the probability of being selected is the same for any available focus, instrument, or observing mode. This is not contradicted by the fact that there are instruments more requested than others.
The next question: Is there more scientific pressure for some specific range of right ascensions? Again, the answer is NO, the probability of being selected by TAC is independent of the position of the targets. And, again, there are positions on the sky which are more targeted than others To improve the scientific use of the telescope time, we have to implement new scheduling modes. We are currently actively thinking about the block and/or queue scheduling and service observing modes as most likely to improve the scientific return of CFHT.
We have the statistical data on the number of hours lost to weather and to technical problems. The data for 1996 and 1997 give the total losses of 36.3 and 30.8 percent, with the detailed split between the technical/shutdown/weather fractions as 0.8/3.9/31.6 and 5.6/3.4/21.8 percent. We leave these data to the reader to contemplate, but we are quite satisfied by these numbers.
Another measure of the overall efficiency is the statistics on open shutter (exposure time), in relation to the total available time. These data are accessible from the CADC archive and give the average efficiency (open shutter time)/(available time) of 72%. The average efficiency defined in this way appears lower on nights with time lost for any of the causes. In particular, we still experience more problems on first nights after instrument changes than we would like to see and the ``first-night syndrome'' is still with us. Clearly, we could increase our efficiency if we scheduled long runs with the same focus, the same instrument, and in the same observing mode.
Here, some statistics will be again very useful to illustrate the current state of affairs. The currently implemented five-year plan for 1996 - 2000 has three major effects on the CFHT manpower:
The statistics of hours worked by the technical staff for three recent semesters gives the fractions of times spent on various activities. In particular:
In terms of cost per focus/night of observation, the prime focus is the least expensive, followed by f/8, then coude and then f/35. Clearly, a reduction on complexity of our operations would reduce the overhead involved in the frequent and labor-intensive changes. Of particular significance is the fact that during the recent years and months, the load of new developments (TCS4, KIR, CFH12k, OSIS-IR and MEGAPRIME) has significantly increased, without reducing operational constraints. In this situation the CFHT staff is in a permanent danger of being over-burnt. This leads to a necessity of establishing very clear priorities for development of new projects and for strict definitions of their relation to the on-going operational issues.
Any new project affecting the planning of CFHT should follow a well established route. In particular, it should be:
This process will have to happen in the situation of further budget cuts planned for 2001. In view of this restraint, we will ask the SAC to concentrate on particularly demanded instruments for the most productive foci. We will ask for de-commissioning of f/35 and coude mirror trains (at the latest during the semester 2000II).
| Existing in 1993 | Added after 1993 |
|---|---|
| MOS, MOS-FP | MOS-ARGUS |
| SIS, SIS-FP | SIS  OSIS |
| AOB, AOB-FP, OASIS | |
| REDEYE-W | KIR |
| FOCAM | UH8k |
| FTS InSb | FTS BEAR, FTS InAsGa |
| Gecko | |
| De-commissioned | In development |
| coude f/8 | CFH12k |
| Herzberg | OSIS-IR |
| REDEYE-N | Coude fiber |
| MEGAPRIME/MEGACAM |
Our intention is to develop service observing and queue scheduling. Both modes of operations should increase the overall efficiency of the telescope, but will require more work by the CFHT staff. Effectively, this will lead to a transfer of the load from the astronomy communities in the participating countries to CFHT.
Clearly, for normal operations, CFHT must maintain new developments in detectors and software. With time, the role of the preventive maintenance will keep on becoming more important to avoid major failure of the aging infrastructure.
On the development front, CFHT should attempt to remain in the forefront of the world research. One of the attractive scientific niches, particularly well suited for the use of the bright time will be the Wide Field IR camera at the Cassegrain focus. The other developments and upgrades will require careful prioritization in view of the manpower limitations discussed above. Among many projects that have been discussed as possibly leading to implementation, but requiring a hard choice between them, we have:
Several significant developments with potential of deeply affecting the future of CFHT will take place within the next months. Among them will be the recommendations of the Next Generation CFHT (NGC) Committee which will formulate a report about a possible replacement of the 3.6 meter telescope. The report will be formulated before the next meeting of the Board in December 1998.
The Advisory Committee has issued a draft report to the agencies, based on its findings during the visit to CFHT in March 1998. The Advisory Committee was charged by the agencies with a duty to find ways of preparing CFHT to operate with $5M (the 1996 dollars) in 2001. The recommendation of the Committee has been actually to prepare a 10-year plan for the years 2001 - 2010. This is a major development whose significance is currently becoming evaluated for the purpose of the actual planning.
If the operating cost will still have to be reduced after 2001, a new strategy must be developed. Several possibilities remain to be explored. Among them would be to share the telescope time between a large set of 4-meter class telescopes by developing a plan to specialize each of them for specific scientific applications. Although such a collaboration would be clearly very effective, creation of the new organizational structure would require:
In conclusion, I may offer a variation on the proclamation of May 1998:
Soyez réalistes, demandez l'impossible! (mai 1968)
This time it would be:
Soyez irréalistes: acceptez les contraintes! (mai 1998)
