REMOTE OBSERVING OPERATION

  We now outline a night from a typical remote observing run at Caltech, to demonstrate the advantages and problems associated with such operation.

 

Figure 12: The LRIS instrument control interface at Keck Observatory. This window provides a schematic of the instrument and indicates its current configuration.  

ACTS scheduling

For the testbed ACTS experiments such as ours, the satellite must be scheduled 1-2 weeks ahead of time. This is not a problem in most cases, since observing schedules at Keck are established 6 months at a time. The primary difficulty in scheduling ACTS sessions lies in the 5-hour time difference (6 hours during daylight savings time) between Hawaii (HST) and the East coast (EST). Since the satellite experiences higher demand during daylight hours, it is often difficult to run an observing session remotely during the second half of the Hawaiian night. We therefore often restrict remote runs to the first half of the night. But even this is not a critical problem, as the University of California actually allocates its Keck time in half nights, to provide more astronomers with an opportunity to use the telescopes. And in many full-night runs, the first half of an observing night is the most complex and demanding, so eavesdropping and collaborative use of the remote capabilities are very useful in that capacity.

ACTS setup

When the time comes for the ACTS session to begin, operators at JPL and at the Tripler Army Medical Center in Honolulu will turn on the transmitters at the HDR ground stations. We are fortunate that both HDR units are located at facilities which are in continuous 24-hour operation every day, so that staff are available to turn on/off the ground station transmitters. (As mentioned previously, this arrangement has been arrived at through some negotiation for our experiment, but could prove more difficult for very remote observatory locations.) The ACTS control personnel at NASA/Lewis in Cleveland then connect the satellite with each HDR, and verify that the signal strength is sufficient. Barring hardware complications and rain at either HDR location, the network is generally available within a few minutes of the scheduled time.

 

Figure 13: The LRIS instrument control interface at Keck Observatory. This window allows the user to modify the configuration of the instrument.  

Contact Keck

Once the network has been established, the observer customarily contacts the observing assistant (OA) at the telescope to indicate that they are ready, to check the weather at the telescope, etc. Also at this stage, an audio link is established over the ATM network, in order to alleviate the need to manipulate (and pay for!) an all-night phone call between the astronomers and the OA. For these purposes, we use a TCP-based audio tool called rat . Based on an earlier tool (vat), rat was developed for the MBone (Multicast Backbone), but contains full support for point-to-point audio connections. Any of the Internet telephony products commonly available could be used for this purpose. Many include useful features such as echo suppression and voice-activated microphones.

LRIS hardware setup

In order to use the LRIS instrument remotely, control must be transferred from the normal instrument control computer at Keck to the duplicate machine which is connected to the ATM network. This involves a 5-minute procedure during which the VME crates which directly control the instrument are instructed to use the alternate machine, and are then rebooted. As described in the previous section, in principle this step is unnecessary, but in practice, security concerns, processing load distribution issues, high-speed networking complications, and the desire to not affect standard Keck observing techniques has led us to establish a separate remote observing control computer. In the future, this rather awkward step should disappear from the remote observing procedure.

 

Figure 14: The LRIS instrument control interface at Keck Observatory. This window controls exposures with the CCD camera.  

LRIS software setup

While the OA initializes the telescope control systems, the observer should then start the LRIS instrument control environment. Just as when observing on-site, a single command/menu option starts up all of the necessary tools, with the display optionally redirected to a remote machine. The observer can then verify any requested configuration changes, such as special slitmasks or filters, and set up personalized instrument configuration files.

LRIS operation

When both the OA and observer are prepared, the observing session runs in the normal fashion: telescope moves and guiding are handled by the OA, while the observer controls the instrument configuration and CCD exposures. The LRIS instrument is controlled via a graphical interface which provides a schematic of the instrument and its current configuration (Figures 12), along with standard graphical elements (e.g., buttons, lists, etc.) for changing the configuration [3] (Figure 13). The CCD camera is controlled through a simple interface which allows the user to set and monitor exposure times (Figure 14). Details such as the number of output CCD amplifiers and the image save directory can be specified as well, of course. A real-time image display is provided, based on the FIGDISP software from FIGARO (Figure 15). Should questions arise, documentation for LRIS and its software are available to the observer via the World Wide Web . The figure on the cover of this report illustrates one of the first images taken remotely with Keck using the ACTS high-speed network.

Should any errors be indicated by either the instrument or detector control systems, the OA and/or engineer can be called upon to examine the problem, as with on-site observing. In such situations, we often use a collaborative whiteboard tool, wb , which allows text and graphics to be transmitted in real-time to both parties. This is useful for indicating error messages, describing image characteristics, transmitting numerical values, etc. (As an aside, we note that wb and the rest of the software used for this project is available for free, with the exception of the TCP-LFN software from Sun Consulting.) Should problems arise with the network, personnel may be contacted at the ACTS control center and/or the HDR sites.

Finally, in addition to the instrument control software, all of the usual observing software tools are available remotely: telescope pointing and UT meters, guider window eavesdropping images, etc. Of course, standard TCP tools such as telnet and ftp are used regularly to retrieve images to the local system, where any of several data reduction packages commonly used in astronomical image processing are available. As mentioned above, one of the outstanding features of remote observing is the wealth of familiar software and hardware facilities that are available at the user's home institution: printers, personal workstations and software, libraries, etc.

 

Figure 15: The LRIS instrument control interface at Keck Observatory. This window provides a real-time graphical interface for the observations. Images are transferred directly to this display as the CCD chip is read out.  

System shutdown

Following the end of the observing session, control of the LRIS instrument is returned to the primary computer. This leaves the instrument ready for the next day's engineering and non-remote observing. Finally, a phone call is usually made to the ACTS control center to verify that the run is completed.

The remote observing process was developed and optimized over the lifetime of the project, with problem solution procedures being the slowest to crystallize. The final several observing runs of the project ran smoothly from a procedural standpoint, as both the astronomers and the OAs became more familiar with this mode of operation. User interface improvements, such as menu selections for commonly used remote tools, have been added to minimize the impact of the remote connection on the observer and the OA. With the possible exception of the re-boot of the instrument control crates, it is our conclusion that the average astronomer could observe remotely on Keck with at least as much ease as actually traveling to the observatory. (In either case it would be recommended that first-time users collaborate with more experienced ones, as with any new instrument or telescope.)