|Stargazing Across an Ocean: the Keck Remote Observing Facility|
|Contact: Paul Preuss, firstname.lastname@example.org|
Hawaii, February 21, 2007, 3:30 in the morning: Nao Suzuki and Rahman Amanullah, two Berkeley Lab astrophysics postdocs, are chasing supernovae with one of the twin Keck telescopes on the summit of Mauna Kea. In the telescope's control room an observing assistant (OA) stands by to help partly by occasionally dashing outside into the frigid night air, 4,205 meters above sea level, to guess at the changing weather.
All night the sky has been obscured by broken cirrus, a translucent veil that shrouds the distant supernovae that were the night's original targets. Suzuki and Amanullah have turned to brighter objects, the host galaxies of supernovae recently found by the Nearby Supernova Factory, which are visible through the clouds.
Finally, Amanullah says, "just as we're running out of substitutes," the clouds begin to clear. He reaches for the finder chart for candidate 07D3cc, the night's single highest priority a supernova found just days earlier by the Canada-France-Hawaii Telescope's SuperNova Legacy Survey (SNLS). The goal is to use the star's spectrum to determine its redshift and identify its type, in hopes that it's a Type Ia, a standard candle for measuring the expansion history of the universe.
The Keck's OA confirms that the telescope has locked onto the guide star indicated by the finder chart. Suzuki says, "Let's start with a half-hour exposure."
Half an hour later, Amanullah checks the spectrum. "It's awfully faint. We need a second exposure." By now there's only an hour left until dawn.
"We're going to keep going until the last minute," says Suzuki, "although I still cannot see the Milky Way, so extinction must be at least two magnitudes," a significant dimming of the target.
Sure enough, the next exposure plus a third exposure the observers squeeze in before dawn brightens the sky is fainter than the first. Did they get enough data to identify the type and redshift of 07D3cc? The answer will have to wait until graduate student Joshua Meyers analyzes the data back at Berkeley Lab.
It was a typical midwinter night of observing, with one wrinkle: neither observer was at the telescope. Suzuki was sitting in Keck Headquarters in the town of Waimea, 32 kilometers from the summit, and Amanullah was some 3,800 kilometers away, in Berkeley Lab's Keck Remote Observing Facility, a windowless room deep in the bowels of Bldg. 50B, the size of a walk-in closet, crowded with computer monitors. The videoconferencing monitor has three windows: Berkeley Lab, Waimea, and the telescope control room.
Virtual Network Computing (VNC) windows, running simultaneously in all three locations, controlled the settings of the telescope and its instruments. Monitors show where the telescope is pointing, what it's seeing, the settings of the spectrograph, data plots, and other information. With a few clicks on the keyboard, either Suzuki in Waimea or Amanullah in Berkeley can change the spectrograph's filters, set exposure times, and open the shutters; with the OA's cooperation they can move the eight-story-tall, 300-ton telescope to pinpoint a distant supernova in a crowded field of stars.
The Physics Division's Tony Spadafora is scientific coordinator of the remote observing facility, which saw its first use in October, 2005. "Not so long ago we'd have to send two or three people to Hawaii for one night's observing, an expert and trainees," he says. "Now we're sending one, which saves a lot of time and money."
Early on, Deb Agarwal of the Computational Research Division met with potential users and Keck representatives, including remote-operation coordinator Bob Kibrick at UC Santa Cruz, to lay out the system. The remote observing room had to be used for no other purpose, and communications with the telescope had to be over a dedicated subnet, firewalled from possible interference.
Agarwal says, "Craig Leres and I threw the system together with spare parts" which included Sun high-resolution monitors and a graphics card that could support three monitors at once "and we evolved from there. If we'd bought a system before we'd actually tried to use it, it would have been wrong." Despite inevitable glitches it took Agarwal and Leres only two weeks to get the system up and running.
Of the five remote observing sites established on the mainland so far, all in California, some are designated "mainland only," meaning observations can be done entirely from the remote site nobody needs to go to Hawaii. Berkeley Lab operates in "eavesdrop" mode, however: an expert observer must be present in Waimea even though the observer in Berkeley can also control the instrument.
Mainland-only status will require new equipment, including back-up communications and control systems. But even in eavesdrop mode, the remote facility has conferred major advantages on astronomers and astrophysicists at Berkeley Lab and UC Berkeley.
"Most important has been training students," says Spadafora. "Another enormously helpful aspect has been the availability to campus astronomers. A night on the world's biggest telescope is extremely valuable, and a whole semester's schedule has to be established in advance. We share with the campus, and occasionally a night scheduled for telescope engineering becomes available for observing at the last minute; we are in a position to take quick advantage. Some of the star astronomers on campus have made use of our convenient access to the Keck."
There may be no such thing as a routine night on a telescope in midwinter, however, especially not when using a long-distance connection. Setting up observations for the night of February 20-21 began at 2:00 in the afternoon Hawaii time, 4:00 o'clock in Berkeley, and not only the weather failed to cooperate; the Virtual Network Computing windows were not responding to instructions.
In the first months of remote operation Agarwal had gotten used to waking up at 2:00 in the morning to answer questions from observers with problems, but that time was past. Today she was at the keyboard again, on the phone with UCSC's Kibrick while Spadafora and Physics postdoc Kyle Dawson looked on.
Dawson, trained on the Keck and the most experienced observer with the remote facility, was there to train Amanullah in its operation. Also on hand was Joshua Meyers, a graduate student Dawson had recruited into the Supernova Cosmology Project, who would reduce whatever supernova data was acquired that night.
"There are various ways to start the VNC servers. When they started that night, you could connect to the VNCs but they were not working," says Agarwal. "Bob Kibrick saw that the servers had to be killed and restarted. He sent us the script, so that if this happens again, we can fix it ourselves. Over the past two and half years we've eliminated a lot of problems this way."
Once the problem was fixed, Dawson in Berkeley and Suzuki in Waimea discussed options for targets, given the iffy weather. Then Dawson showed Amanullah how to use the VNC to calibrate the spectrograph's red and blue cameras, using lamps in the telescope that emit standard elemental lines, and perform other operations. When everything that could be done in advance had been done, it was time for a dinner break.
In Hawaii, the sky grew darker. "Winter nights are longer, but on Mauna Kea February is the worst month for weather," says Suzuki. "Even with high-tech instruments, ground-based telescopes are always fighting the weather." By midnight in Berkeley, Amanullah was on his own, except for the videoconference connection. SNLS supernova 07D3cc was their early morning prize, but was the data good enough?
Four days later, Suzuki was back in Berkeley, and he and Meyers had determined that indeed 07D3cc's faint spectrum was sharp enough to reveal both its type and its redshift. To convert the raw data into a spectrum, Meyers first had to remove defects peculiar to the spectrograph's CCD and the telescope itself. The resulting spectrum mixed light from the supernova with light from all the stars in its host galaxy, including the hydrogen lines (which Type Ia supernovae lack) that are one of the easiest ways to determine redshift.
"Eighty percent of the analysis isn't hard; the software does it for you," Meyers says. "The last 20 percent I did by hand." By comparing the spectrum to templates of other galaxies, Meyers determined that the host galaxy's red shift, or z, was 0.7, corresponding to roughly nine and a half billion light years away. "Then I cycled through every supernova template I had" some 350 individual templates "and found the best match. It was close enough for me to determine that 07D3cc was a Type Ia supernova, and that we had caught it probably a day or two before its maximum brightness."
Suzuki says the supernova will go into a Hubble diagram being constructed by SNLS, a graph showing the history of the expansion of the universe. "Eventually they want to include 500 Type Ia supernovae in the Hubble diagram," he says. "This hard-won supernova is going to be one data point among those 500."