CFHT is a classical telescope operated in a traditional manner. A comparison of CFHT and new telescopes such as Gemini, the VLT or WIYN reveals striking differences, not only in the construction, but also in the way they are operated. We explore some of the operational limitations present with CFHT and discuss some alternative operational modes which CFHT could explore. Finally, a model for how CFHT might be operated in quite a different manner in 2003 is presented.
The new telescopes are also exploring new operating modes in order to maximize the scientific return from the large investment represented by the telescope and instruments. While CFHT led the way in demonstrating that excellent image quality could be achieved by paying attention to the dome thermal effects, it also showed that there is a distribution of seeing on Mauna Kea. Not everyone gets 0.5 arcseconds! The new operational modes which will be used by the new 8-m class telescopes are designed to provide flexibility so that the observations carried out are matched to the current weather conditions.
In this paper we will explore some of the limitations encountered when considering new operational modes for CFHT. Before discussing new operational modes we will look at the current scheduling practices and how they could be improved. We will then discuss some of the new operational modes and how they may be implemented at CFHT. Finally, some operational improvements for CFHT are proposed and a schedule for CFHT in 2003 is presented based upon the ideas presented in the rest of the paper.
Many telescopes have the ability to switch easily between instruments but we will look at Gemini as a good example. The Gemini telescope has a single focus and an instrument cube which remains permanently mounted at this Cassegrain focus. Up to five instruments, including the adaptive optics, can be mounted simultaneously, with the ability to switch between them in only a few minutes. Of course, this implies the instrument is configured and ready to use but this work will be easier since the instruments will only be removed occasionally. Thus Gemini will be able to switch from the GMOS spectrograph to the near-IR imager when the moon rises for example.
Each time CFHT needs to switch to a new instrument the current instrument is removed from the telescope and the new one mounted. One may also need to change the top-end of the telescope. CFHT is even more limited at the moment since instruments do not have dedicated detectors. Thus a switch from MOS to OSIS-V requires that all the filters to be used in OSIS be focussed and the CCD moved and aligned during the day of the run. Not only does changing instruments take time, it takes considerable time to configure and prepare an instrument each time it goes on the telescope. This continual mounting and dismounting of instruments also leads to more operational problems as connectors, etc. are put under additional use.
CFHT is developing a fiber feed from f/8 to Coude which will allow us to easily switch to Gecko whenever the telescope is used at f/8. This is a small but significant step in the right direction.
The way the telescope is currently scheduled also limits the scientific effectiveness of CFHT. This is explored further in a later section.
Any new operational modes for CFHT must increase the scientific output of the telescope. How this is measured is a subject far too complex to discuss here. Another goal of new operational modes must be to increase the scientific utilization of observing time. One way of achieving this is to make more effective use of dark time. Being able to switch to a bright-time instrument when the moon rose would mean that partial dark nights could be utilized by instruments requiring dark time. The same applies to seeing. Scientific output would increase if CFHT could switch to Adaptive Optics whenever the median or better and the moon was up.
Other areas which affect the scientific efficiency of CFHT include:
Another limitation on the scientific output of CFHT is not related to the telescope, instruments or dome. The current approach of scheduling the telescope in 6 month blocks, in conjunction with rules on the number of top end changes and length of runs leads to many inefficiencies.
Projects are scheduled at times not suited to their targets. Currently eight nights are required for an instrument to be scheduled. If there are only three programs for a total of eight nights approved which use a given instrument it is quite likely that one of those projects will be scheduled at a less than optimal time.
The current scheduling system also limits our ability to respond to new ideas or discoveries which may require faster access to the telescope. It can be almost 18 months from the time that an idea is spawned until the project is on the telescope.
Highly ranked programs suffer from the weather at the same rate as lower ranked projects. At JCMT lower ranked projects are scheduled in blocks and the actual observations are performed in service mode. Higher ranked proposals are scheduled normally but if they suffer from bad weather or instrument problems, they are immediately rescheduled during the blocks where lower ranked proposals are scheduled. This effectively bumps one or more of the lower ranked programs in favor of guaranteeing time to higher ranked projects.
Scientific utilization of CFHT could be increased by revamping the way the telescope is scheduled.
As there is another paper devoted to this subject in these Proceedings (see paper by Tim Abbott) only a summary of queue scheduling will be presented here.
In queue scheduling projects which are accepted by the TAC submit a detailed description of the observations to be made. This is referred to as Phase II of the proposal process. These descriptions must define the conditions (seeing, sky brightness, etc.) which must be met before taking the observations. As it is impossible to predict when any given set of observations will be taken, the observations are taken in service mode. The observer selects which observations to take using a set of rules to compare the current conditions with the requirements of the observations in the queue.
Queue scheduling increases scientific output since a project which requires 0.4 arcsecond seeing will get data taken under those conditions assuming they occur during the semester. In classical scheduling, this project is scheduled on fixed dates and then enters the weather lottery.
Queue scheduling can increase the utilization of dark time as well. Even with direct imaging an observations requiring B-band images could be taken when the moon is below the horizon (even if the moon is 8 days from new) and then the observer could switch to a program doing narrow-band imaging in the red when the moon rises.
CFHT will be operating CFH12k in queue mode in Semester 99II. The real advantages of queue scheduling will be realized when CFHT has the ability to switch between instruments.
I will distinguish between Waimea-based observing and true remote observing which mean off island. Each offers its own set of benefits and challenges.
Over 90% of Keck observing is done from Keck's Waimea headquarters building. Observers find they are much more efficient at 2500 feet than at 14,000 feet.
There is no technical challenge in implementing Waimea observing for CFHT. The new 45 Mbps link to the summit has more than enough bandwidth to support a video link and the data from instruments, including CFH12k.
Unfortunately, CFHT must deal with several issues before we are in a position to offer this to our communities. First, and most significant, is the need to have two people at the summit for safety reasons. Keck operates two telescopes so they don't face this problem. There have been many discussions on how to deal with this issue and various solutions have been suggested.
There are other logistical issues such as day sleeping quarters, food and drink for nighttime observers, and heat in the observing room in Waimea which must be solved before CFHT could offer Waimea-based observing. None of these are particularly difficult to solve but they will take money and time.
True remote observing (i.e. from Canada or France) must deal with the two person summit rule as well as dealing with some technical challenges. In terms of communications between the remote site and CFHT there are three areas: control of the instrument (and possibly telescope), status information and display and interaction with the data.
Control of the instrument does not require high bandwidth and all instruments could be operated remotely by using a Netscape interface, as is used for Gecko, or possibly Netscape plus Java for more complicated instruments. Likewise, status information could be viewed with a Java applet and only requires low bandwidth.
However, the display and interaction with the data does require all of the available bandwidth; more conventional approaches. by using an intelligent display system, with an image server in Hawaii and remote clients, incorporating on-the-fly compression of data would make remote interaction with the data fairly efficient. Lossy compression would allow high compression ratios (approximately 100) and would provide enough image fidelity for many purposes. The system must be able to send subrasters with minimal loss of fidelity to allow for closer examination but these could still be compressed by five or six to one. Such a system could be built today utilizing available code and compression algorithms.
In order for CFHT to really take advantage of queue scheduling we must develop the ability to switch between instruments during the night. Ideally, this would include the ability to switch between prime focus and f/8 during the night. This is not an easy idea to implement! However, the flexibility that would be available would allow CFHT to realize some large scientific benefits. For example, being able to switch between Megacam and the proposed wide-field IR imager depending upon whether the moon is above the horizon would allow the best use of both these instruments.
If the reader will permit me to pursue a bit of fantasy I will present a mock operations schedule for 2003. I will assume that the number of instruments at CFHT has been reduced, and that there are dedicated detectors for each instrument. The instruments being operated in 2003 include:
I will also assume that CFHT in 2003 can switch between the three operating foci of the telescope, Coude, Prime and f/8. Figure 1 shows hypothetical operations schedule with instruments mounted and operational at all three foci. Note the small number of instrument changes and complete lack of top-end changes in this fantasy!
