Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.eso.org/~bleibund/papers/Landskrona.ps Äàòà èçìåíåíèÿ: Sun Jan 7 21:58:59 2001 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 05:20:39 2012 Êîäèðîâêà:

Operational Concepts of Large Telescopes \Lambda
Bruno Leibundgut
European Southern Observatory
Karl­Schwarzschild­Strasse 2
D­85748 Garching
Germany
August 22, 1996
ABSTRACT
The singular way of scheduling at major observatories has favored certain types of astronomy. This has led
to the discrimination of certain types of observations which could not be accommodated. It is the goal of future
operations to open possibilities for some additional types of observational astronomy. Together with improved
observatory operations the new observing modes can provide significant progress in the acquisition of astronomical
data. Specifically, the capability to optimally match the schedule to observations should prove a major advantage
of service observing.
The VLT data flow project is designed to accommodate these new observing modes together with conventional
observing. Simulations and first experience with the ESO NTT will prepare and refine the concepts and procedures
foreseen for the VLT. The acceptance of service observing by the astronomical community is of critical importance
for this new operational mode to succeed.
Keywords: telescope operations, operational models, service observing
1. INTRODUCTION
The efficiency of the new large telescopes to increase our knowledge of the universe and its constituents will
depend significantly on how astronomers can utilize them. Only creatively­used and well­operated telescopes will
be able to contribute their share to the development of astronomy.
Operational concepts currently discussed by the various large telescope projects can be divided into two main
groups which differ mainly by the composition of their user communities. On the one side are the privately­
owned observatories with access restricted to a small community of astronomers. For many of these observatories
it is most economical and also easiest to continue operation of their telescopes as in the past. The astronomers
perform the observations themselves and are fully responsible for the acquired data. Normally they are assisted
by experienced telescope operators. In the following this situation will be referred to as conventional observing,
often also referred to as classical observing.
\Lambda To appear in the Proceedings of the SPIE Conference ``Optical Telescopes of Today and Tomorrow,'' A. Arneberg (ed.), held
May 29 ­ June 2, 1996, in Landskrona/Hven, Sweden

National and international observatories on the other hand are actively experimenting with service, or queue,
observing modes 1;2;3 . The attempt is to take better advantage of the varying conditions by combining the best­
suited observations independent of individual astronomical projects (cf. [4]). Active interaction with the scientist,
who proposed the observations and will analyze the data, during the observing process becomes nearly impossible
in this case. The break in the observational chain has to be recovered by operational procedures which enable
the astronomers to maintain control over their observations. At the same time interesting new astronomical
possibilities emerge with such a mode.
The next generation instrumentation which comes with the new telescopes will also deliver unprecedented
data at the cost of increased complexity. Astronomers will need every assistance in preparation and execution of
their observations possible. Often when observations are defined in sufficient detail the actual data acquisition can
be delegated to the observatory. Many aspects of the advantages of the various modes and the changes involved
have been collected in [4].
In the following we will discuss some of the advantages of the two observational modes and develop criteria when
they are suited best (x2). The concepts of the VLT specific data flow project which is designed to accommodate
the needs of service observing are presented in section 3. Open issues and future directions are discussed in the
conclusions.
2. ADVANTAGES --- ONE WAY OR ANOTHER
The way current observatories are operated has been developed to efficiently distribute the scarce resource of
observing time. Major guidelines are scientific merit of the proposed observations, fair distribution among the
subfields of astronomy and -- to a certain degree -- democracy. Each successful applicant is awarded a certain
amount of time at fixed calendar dates to perform the experiment. All the observatory provides is a functioning
telescope and instruments with no or little guidance on how to best perform the observations. The observatory's
r“ole is essentially to administer the resources while developing and maintaining the infrastructure.
With the astronomers guiding the observations at the telescope a quick scientific assessment of the data in
respect to their suitability for the project can be done. The vital link of the astronomical observing experience
and the researcher is maintained which assures that the data can be analyzed properly. Psychological aspects like
the personal involvement of the astronomers with the data acquisition or the astronomers' detachment from their
regular work during observing trips can also contribute to the creativity of the scientific process. Advantages of this
conventional mode for the observatory are the interaction of observatory staff with the astronomers, the experience
brought in by the external users to the observing process, and the possibility to transfer the responsibility of the
data acquisition to the astronomers directly. This is particularly important in cases of specialized observations,
the outcome of which can not be predicted.
Typical research favored by these operations is based on observations which can be obtained under ``regular''
environmental conditions delivered by the site and the telescope. Projects which require special conditions are
clearly discriminated, unless there is easy access to the facilities as is often the case at observatories with small user
communities. There, astronomers typically obtain a larger share of observing time and can mix their own projects
to make adequate use of particular meteorological circumstances. Other projects which are typically disfavored
in this mode are surveys since the normally large time demands can not be readily allocated at observatories
which serve large communities. Another type of project which normally suffers from this conventional scheduling
mode are targets of opportunity and general time series observations of objects with time scales of weeks or
months. Smaller communities, where the data exchange is organized informally, have been able to arrange for
such occasions, but the larger observatories had to introduce formal and often awkward rules to handle these
situations.

The previous paragraph lists a few astronomical reasons why it would be advantageous to change scheduling
procedures. The exploration of parameter space mostly inaccessible so far can add significantly to astronomical
progress. It is this widening of the observational options which is scientifically most interesting and leads to the
discussion of service observing. The possibility to select observations matching the prevailing conditions best will
enable projects depending on special circumstances. This observational edge may speed up results which otherwise
would rely on the meteorological luck of the draw. Surveys can be carried out mixed with other observations. It
has to be noted that most massive surveys have been obtained in service mode as the example of the Palomar
Observatory and the ESO/SERC Sky Surveys, the current near­IR surveys (DENIS and 2MASS), or the searches
for massive halo objects (MACHO, EROS, OGLE) with their scientific spin­offs show. Another observational
niche difficult to access at large observatories is the monitoring of variable objects and observations of targets
of opportunity. Such synoptic observations can very easily be fitted into a flexible scheduling process, which
is a prerequisite for service observing. A small number of objects spaced around the sky requiring just a few
observations can easily be accommodated in service mode, while they may not reach critical program size for time
allocation in the conventional case. Another aspect more difficult to quantify may be the more efficient use of the
available time as programs will be performed to the exact amount of exposures needed. Finer adjustment to the
moon phases gains some dark hours which are often lost when programs are scheduled conventionally in blocks
of complete nights 5;6 . Observing projects may improve by the needed preparations which entail a complete road
maps of how the data will be analyzed. This should happen even though there will be no formal requirement
to do so. A further possible spinoff is the availability of an extensive archive of observations. Archival research,
although not yet a primary resource, may contribute significantly to projects where the combination of data from
different wavelength regimes is advantageous.
While there remain many astronomical projects which clearly are best served in conventional mode, the above
examples show that there are a few reasons why it may be interesting to explore some other operational modes
with new telescopes. The selection of the most appropriate observing mode for each project will be an important
decision which will need careful evaluation by the astronomer.
The move to new operational schemes must be driven by improvements in the scientific process. Many
discussions on operational issues have focussed on advantages for observatories, which are not negligible, but can
not by themselves justify the proposed, stringent changes. The astronomical community must embrace the new
opportunities for success. It should also be noted that all observatories plan to offer both observing modes leaving
all options open.
3. INFORMATION EXCHANGE IN A COMPLEX OBSERVATORY
THE VLT DATA FLOW
Removing the astronomers from the actual observations at the telescope implies that other means must be
provided to retain their control over the observational process. The development of the procedures to guarantee
the astronomer's participation has started at ESO within the On­Line Data Flow project (see also [1]). Its basis
was developed in the VLT Science Operations Plan 5 and a document which defines the astronomical requirements
on the observatory information chain 7 . The link between the astronomer and the observatory should be close and
transparent. It also should remain flexible.
There are a few very clearly separated stages in the astronomical information and data cycle 1;7 . Each phase
has demands and services which have to be identified and carefully combined. The definition of the observing
program and its scientific evaluation by peers as well as the definition of the individual observations should be
provided by the astronomy community. The observatory solely supports and administers this process. Scheduling
of programs in conventional mode and the observations in service mode, however, are performed by the observa­
tory. The demand for exact observation definition drives the requirement for the second phase of the astronomer
-- observatory interaction. To built a schedule which optimally matches the prevailing conditions with the obser­

vational requirements the scheduler will need accurate input from a meteorological site monitor, telescope and
instrument status, and astronomical restrictions (e.g. moon phase). The actual observations will be handled by
specific telescope, instrument, and detector software 8 which will return the data products (frames and logs) to the
data flow. The further data handling involves archiving and pipeline processing for provisional quality control.
All these processes fall into the responsibility of the observatory.
These considerations led to the adoption of a two­layered system relaying information among the various
subprocesses. The VLT could be run from the control system alone without the data flow superstructure, but
the technical description of the instrument and the interaction with the scheduling process are considered too
detailed and cumbersome for astronomers who infrequently interact with the system. The data flow acts as the
intermediary between the astronomers and the technical software. There are four fundamental agents the data
flow is connecting: the astronomer, the scheduler, the technical software, i.e. the observing facilities, and the
archive. The VLT concept does reflect their basic needs. ``Observation Blocks'' contain the complete information
relevant for an individual observation 1;9 . An observation in what follows is considered a single pointing of the
telescope with a specific instrument setup to acquire a coherent data set. Apart from obvious quantities, like
coordinates and instrumental setup, observation blocks can also contain global requirements of importance to the
scheduler, general comments of interest to the observer, and links to reduction procedures or quality control. A
specific feature is the modularity by which observation blocks can refer to other observation blocks to combine
observations. It is thus possible, e.g., to link regular observations with the acquisition of calibration data.
All information on the instrument and its operation during the observation is encapsulated in ``instrument
templates 1;9 .'' These structures define commonly used setups and are embedded in the observation blocks. The
astronomer defines the specific parameters of the setup in a template parameter file which accompanies the
template. This should ease the astronomers' interaction with the VLT system as many details of instrument
operations can be served by the templates. Some observations will not be offered with templates in which case
the option to drop to the level of the VLT technical software is still available.
Astronomers whose proposals successfully passed the selection process will hence have to prepare the obser­
vation blocks and the templates for each observation in their program. This preparatory phase is foreseen for
all proposals and guarantees the close involvement of the astronomers. Even conventional observations will be
prepared in this preparatory phase to familiarize the astronomers with the system. In this case the astronomers
will use their observation blocks at the telescope during their assigned nights. During service time all available
observation blocks are provided in a central database polled by the scheduling software. The scheduling is a very
complex process which currently is not yet fully defined for the VLT. Since the best performance is achieved only
when sufficient information is available and the detailed procedure depends on many different sources, it is im­
portant to collect as much intelligence as possible. Observations with detailed descriptions of their requirements
are more likely to achieve the requested quality as they can be scheduled accordingly.
A long­term plan for the semester or a fair fraction of it schedules conventional observing runs and defines
the requested instrument setups. This will be required even with instrument changes becoming possible at short
notice as special filters or gratings will have to be mounted ahead of time. Flexible scheduling itself will rely on
local information sources which describe the prevailing conditions of and at the observatory. Meteorological input
is provided by a site monitor. Image quality assessment (including sky background), possible forecast of critical
parameters (e.g. cloud cover, precipitable water vapor, seeing) will provide the basis of the selection process. ESO
has maintained a program to characterize prevailing meteorological conditions over the past several years 10 and
is embarking on a project to forecast some of the meteorological parameters on the basis of a few hours. Options
for operations with limited information have to be developed as well.
An important aspect of the scheduling process is the underlying criteria which govern the selection. This is
largely unexplored territory for all observatories with the notable exceptions of HST and the NOAO operations
of the WYIN telescope 2 . Important results are expected from simulations with mock projects which encompass
a large set of observations and a variety of requirements. Although it will be impossible to fully simulate the
scheduling of a semester describing the creativity of programs of real astronomers and vagaries of real­time

operations, simple strategies can be tested and compared. The effect of operational overheads, instrument changes,
decision time scales, importance of program completion, and weighing of different conditions can be explored ahead
of time.
The telescope and instrument software returns raw data frames to the data flow. Archiving and further
processing complete the cycle. The data archive captures all relevant information for a given observation. This
includes the original request contained in the observation block and template together with the actual conditions.
Other relevant observations, e.g. calibrations and standard star data, linked to the project are also stored in this
central place. The astronomers will receive (or retrieve) their data from this archive. Once data become public
it will be accessible by the whole astronomical community.
At the VLT a routine pipeline processing of all data obtained in service mode and, possibly, conventional ob­
serving will be attempted. The pipeline results are used for a quick assessment of the data, potentially influencing
the further observations of the night. They will provide preliminary removal of instrumental and detector effects.
The quality control will follow the pipeline reductions to ensure that the observations correspond to what was
requested.
To test the concepts and procedures of the data flow a set of reference proposals has been defined. These
observational projects with some scientific background have been designed to cover a large range of observational
requests and techniques. They will be used to check the interfaces and the interactions of the various parts of the
data flow. For a first check they will play the r“ole of external astronomers. The reference proposals can also be
used for simple scheduling simulations.
4. BUILDING THE EXPERIENCE
The complexity of the operations of modern large telescopes should not be underestimated. The required
information exchange between the astronomers and the observatory represents a vital link to assure that the
observatory delivers what is requested and expected by the astronomers. The future will look different for the
regular user even observing conventionally. Telescope operations have been long ago delegated to specialists and
astronomers have accepted the help provided by telescope operators. The complexity of the instrumentation and
the observational procedures will further emphasize the astronomers' understanding of technical aspects. It should
be the goal of the operations to keep the astronomers' interaction with the facilities and the staff as simple as
possible. Every possible help the observatory can provide should be available to support the astronomers in their
scientific experiments. They should be able to concentrate on the observational aspects rather than technicalities.
Nevertheless, the astronomers will have to be provided with sufficient information so that they can understand
and assess their data in all observational aspects.
To assure acceptance of these conceptual changes the collaboration of the astronomical community has to be
assured. The early involvement of future users of the system can only improve the operations. Several science
test cases for the VLT have been solicited from the European astronomical community. These test cases will be
used in addition to the internal reference proposals to test the procedures. They have the additional advantage
to be based on real science projects and the external astronomers are experts in the requested observations who
can provide helpful criticism. Since the data flow is the VLT's interface with the astronomical community it will
largely define the perception of the observatory. The input from users is essential for a successful development.
The recommissioning of ESO's New Technology Telescope (NTT) will be combined with the start of service
observing at ESO. Several programs have been approved and the data flow system will undergo a first real test
before the end of 1996. At first, the service mode will be restricted to direct imaging in the optical, some of
the operationally least demanding observations. Spectroscopic observations in the service mode will offered only
during the following semester.

5. CONCLUSIONS
The introduction of new observing modes in combination with the improved instrumental capabilities is a
daunting task. The prospective advantages are significant and may provide an important observational edge over
other approaches. At the same time the success of the experiment almost entirely depends on the acceptance by
the community. The astronomers will have to learn to optimally use the new possibilities. Since many of the
changes have been initiated by the observatories it will be their r“ole to convince the rest of the community of the
gains. A fundamental requirement is the smooth operation of the observatory and the improved data quality has
to become an essential argument.
At the VLT the data flow will link the observatory to the astronomers and expand the interaction between
them. It presents the astronomers with all observational possibilities and lets them make best use of the facilities.
An integrated approach has yielded a system which will entail all operational aspects. It should be noted that there
exists a clear separation between the needed infrastructure, the data flow, and the operational model. A flexible
data flow system will provide the options to built and improve operational models without major limitations.
Definition of an operational model for the VLT will have to tackle open questions like the scale of the prepara­
tory phase, the exact criteria for the scheduling process, and the degree of pipeline processing. A convincing
model must include compelling reasons for the expanded preparatory phase. Flexible scheduling drives most of
the complexity of the VLT data cycle. The operational model will have to set the astronomical priorities, possibly
based on the results from simulations.
First lessons from simulating the information and data cycle with mock projects will be followed by service
operations of the NTT in early 1997. Further refinement in the VLT environment at the NTT 11 will provide a
solid basis for a successful start of service observations at the VLT itself.
Service observing will remain an experiment for the first few years. It must not be seen in isolation as it
is introduced to compliment the current observational capabilities. Observing with the VLT will be possible in
conventional as well as service modes. The selection of which mode suits the program and its observation best
should ideally be based on astronomical criteria. This can be achieved when sufficient trust has been built through
reliable delivery of high­quality data.
6. ACKNOWLEDGEMENTS
Building the operations of a complete observatory is not a small task and depends on many people. The views
expressed in this article are based on discussions with many colleagues at ESO. The data flow project is headed
by P. Quinn and includes M. Albrecht, E. Allaert, D. Baade, A. M. Chavan, P. GrosbÜl, M. Peron, G. Raffi, and
J. Spyromilio. They have been instrumental in developing some of the ideas. I am also grateful to A. Renzini,
the VLT project scientist, for many discussions regarding these issues.
7. REFERENCES
[1] Peron, M. & GrosbÜl, P. 1996, these proceedings
[2] Silva, D. 1996, NOAO Newsletter 47
[3] Puxley, P. et al. 1996, these proceedings
[4] Davies, J., Robson, I., & Boroson, T. (eds.) 1996, New Observing Modes for the Next Century, ASP Confer­
ence Series 87

[5] Baade, D. 1995, ``VLT Science Operations Plan'', VLT­SPE­ESO­10000­0441, Garching: ESO
[6] Boroson, T. 1996, in New Observing Modes for the Next Century, eds. J. Davies, I. Robson, and T. Boroson,
San Francisco: ASP Conference Series
[7] GrosbÜl, P. & Leibundgut, B. 1995, ``VLT On­Line Data Flow: Requirement Specification'', VLT­SPE­ESO­
10100­0749, Garching: ESO
[8] Wirenstrand, K. & Raffi, G. 1996, these proceedings
[9] Leibundgut B. 1996, in New Observing Modes for the Next Century, eds. J. Davies, I. Robson, and T.
Boroson, San Francisco: ASP Conference Series
[10] Sarazin, M. 1996, in New Observing Modes for the Next Century, eds. J. Davies, I. Robson, and T. Boroson,
San Francisco: ASP Conference Series
[11] Wallander, A. 1996, these proceedings