next up previous
Next: NETWORK ARCHITECTURE Up: Remote observing with the Previous: Remote observing with the

INTRODUCTION

  Remote use of astronomical telescopes has been a topic of interest for many years, since space-based observing platforms (e.g., IUE) began to demonstrate total remote operation out of sheer necessity. Initially, a number of ground-based radio and optical telescopes (e.g., the WIYN Telescope[1]) introduced queue-based scheduling, a mixture of remote and interactive observing modes. Only very recently are optical telescopes beginning to realize the benefits of true remote observing: for example, observations with modest size detectors at Apache Point Observatory are being carried out remotely using the Internet[2]. In this project, we have established remote interactive observing capabilities for Keck Observatory on Mauna Kea for observers at Caltech, in Pasadena, California. In undertaking this project, we were motivated by several operational and scientific advantages that remote observing would offer.

One primary concern is the high altitude of the Keck Observatory. At 13,600 feet of elevation, the summit of Mauna Kea is a demanding location for both mental and physical exertion. In spite of the requirement that all astronomers spend a night at Hale Pohaku (altitude $\approx$9000 feet) for acclimatization before proceeding to the summit for a night of observing, about 15% of the people who do not observe often at Mauna Kea become sufficiently ill during the course of a 3 night run that they have to leave the summit for at least 12 hours. Approximately 75% of the people coming to the summit to observe for a full night experience some discomfort such as a mild headache, and almost all experience some loss of judgment, irritability, etc. Remote observing provides an environment for all observers that is free of these difficulties, and also provides an opportunity for people who cannot tolerate high altitudes (e.g., pregnant women, those with heart conditions, etc.) to observe with the Keck Telescopes.

Another logistic motivation for remote observing involves issues common to large telescopes located at sites distant from the home institutions: In general, the larger the telescope, the more heavily over-subscribed it is. Runs are therefore often only 1 or 2 nights in duration. Since 2-3 observers come to each run, this means substantial sums of money are spent on travel and related expenses. The additional night of acclimatization for high-altitude sites such as Mauna Kea increases the cost further. Finally, the salary cost for ``wasted time'' during these runs is quite large. An excellent example of the potential savings is that of the European Southern Observatory (ESO ): There are 19 telescopes near the Atacama desert in Chile, including two 3.6-meter telescopes and a set of four 8-meter giants currently under construction (the Very Large Telescope, or VLT). The observing site is widely regarding as one of the very best in the world, yet it is half-way around the globe from most of its large European user population. Understandably, remote observing is gaining popularity among European astronomers[3]. Remote diagnosis of hardware and software problems also becomes more feasible with an operational remote observing system. In our case, the teams that built the instrument hardware and software for the Keck Telescopes are located at Caltech or a campus of the University of California. Just their presence in the same buildings as the remote observers can be extremely helpful when problems arise in the operation of the telescope. The establishment of remote observing from California implies the presence of a network connection, which can allow engineers and programmers to analyze the remote systems essentially instantaneously. Again, both travel and time are saved, and effective help from highly skilled and experienced people in California can be obtained quickly when necessary.

In addition to these operational advantages, there are strong scientific advantages to remote observing as well. With remote observing, every member of a large collaboration can participate in obtaining the data. It is possible for one part of the team to concentrate on obtaining the observations, while other team members can be analyzing the scientific results from the last observation, checking the instrumental performance to make sure everything is working correctly (particularly the detector), or browsing the literature or catalogs of objects as necessary to prepare for the next set of observations. The inclusion of students in the observing session becomes much easier, cheaper, and more routine when no travel is required, i.e., they don't need to miss classes. The facilities available at remote observing sites (e.g., Caltech) usually far exceed those available at the observatory site, whether it be computer hardware, office and library supplies, or a pizza delivery. Recall also that the remote site may be located such that the night hours of observing overlap, or even coincide with standard business hours at the remote site.

Given these strong motivations for remote observing, there are then several issues which must be examined in order to make remote observing feasible, from both technical and operational viewpoints. The primary issue involves the large size of astronomical images. The optical instruments in use at the Keck Telescope have frames that are currently 8 Megabytes in size, soon to become a factor of 4 larger. Although actual integration times depend on the scientific program, and range from less than a second to an hour, the quality of ground based optical and infrared astronomy observations is very sensitive to weather conditions, including clouds and atmospheric turbulence. Hence, even though observing sessions are planned in detail in advance, careful ``quick look'' analysis of each image is important in defining what to do next, how long the next exposure should be, whether to switch to brighter objects due to poor sky conditions, how to modify the program to cope with unexpected failures of non-critical telescope or instrument components, etc. An operating mode such as this clearly requires a means of viewing or retrieving the images at the remote observing site with a minimal amount of delay. The public Internet connection between Hawaii and California was in 1995 (and still is today) insufficient for these purposes.

We therefore submitted a proposal to NASA as part of its Advanced Communications Technology Satellite (ACTS) Gigabit Satellite Network (GSN) testbed program. We received funding to establish a high-speed ATM network running from the dome of the 10-meter Keck Telescopes on the summit of Mauna Kea in Hawaii to the Caltech campus in Pasadena, California, using the ACTS satellite as the network link across the Pacific Ocean. This network has been used to support remote observing, remote diagnosis of problems, remote software development, and other related tasks. In the sections that follow, we will outline the network architecture and topology, describe the role of the ACTS satellite in our system, demonstrate the capabilities and remote operation of a specific instrument on the Keck II Telescope, and summarize the benefits and difficulties which we have encountered during the course of this ACTS demonstration project.


next up previous
Next: NETWORK ARCHITECTURE Up: Remote observing with the Previous: Remote observing with the
Patrick Shopbell
8/11/1997