The Scheduling Guru
By Linda Copman, based on an interview with Barbara Schaefer
Photo: Barbara Schaefer ponders proposals from prospective observers who are hoping to reserve time on the Keck Telescopes. Photo by Gloria Martin, Keck Observatory.
Barbara Schaefer is arguably Keck Observatory's longest-serving employee. Schaefer started working on Keck-related projects three years before the project was funded by the W. M. Keck Foundation.
Schaefer was recruited by Jerry Nelson, principal designer of the Keck Telescopes, because she had experience operating large telescopes and because of her experience on Mauna Kea, one of the proposed sites for the telescope. After graduating from the University of Wisconsin with a degree in Astronomy and Physics in 1975, Schaefer spent four years operating the 4-meter and 2.1-meter telescopes at Kitt Peak National Observatory in Arizona. In 1979 she made her first move to Hawai`i Island, where she worked for the NASA Infra-Red Telescope Facility (IRTF), one of a handful of observatories on Mauna Kea at that time (now there are thirteen).
With the opening of the Keck II Telescope in 1996, it soon became apparent that a coordinator was needed to manage the expanded team of Observing Assistants (OAs) who were needed to operate both the Keck I and Keck II Telescopes, as well as to interact with the visiting astronomers before, during, and after their runs. The position of Observing Support Coordinator was created, and Barbara Schaefer was promoted to fill this role. Schaefer has remained in this position for the past 12 years.
Schaefer's duties can be divided roughly into thirds. One third of her time is spent hiring, scheduling, and supervising the OAs. To her credit, during Schaefer's 12-year tenure as Coordinator, only a handful of OAs have moved on to new positions, some of them within the Observatory. Another third of Schaefer's time is spent interfacing between the night crew (OAs) and the day crew to make sure that issues that arise during nightly observing runs are clearly communicated to and addressed by the daytime operations crew. And the remaining third of Schaefer's time is spent scheduling observing time on the Keck Telescopes and interacting with the observers.
Schaefer uses multiple spreadsheets and strategies to complete the daunting task of scheduling, which can take anywhere from 7 to 12 months from start to publication of the final six-month observing schedule. The two weeks prior to distribution of the draft schedule is the most intense for Schaefer, and her colleagues know to steer clear of her during this time. Schaefer jokes that she hangs a sign on her door which reads, "If you knock, you better have chocolate." This year, her colleagues added a corollary, "The darker the chocolate, the darker the night." Dark nights, with less than 32 percent of ambient moonlight, are at a premium, especially for astronomers who study the faint objects at the fringes of the universe.
"In reality," says Schaefer, "I can't be bribed." If a night of observing at Keck Observatory were ever put out to bid on E-bay, Schaefer believes it would attract high-stakes bidders. But, of course, this is not how the system works.
The Keck Observatory scheduling system runs more like a well-oiled clock, with all stakeholders knowing well in advance what to expect and few last-minute surprises. Planning for the second semester of 2008, which runs from August 2008 through January 2009, begins in January 2008. The first step is to send out an internal memo to Keck department heads and administrators, in order to determine engineering needs for the upcoming semester. In the early days, from 1994-96, only about half of the available nights were allocated for science, while half the nights were reserved for building and testing the telescopes and instruments. The number of engineering nights has decreased substantially, so that now only about fourteen to sixteen nights per semester are set aside for commissioning new instruments, making upgrades to operations, and repairs and maintenance, such as the rotation of mirror segments every two years so that individual segments can be recoated. This means that roughly ten percent of nights per year are allocated for telescope and instrument engineering: the remaining 90 percent of the nights at Keck Observatory are dedicated to producing science.
"I like solving puzzles and this schedule is a huge puzzle. You have to put everything in the right places. For example, what instrument configurations cause problems? I enjoy the challenge of solving these problems." - Barbara Schaefer
The next step in the scheduling process is to establish a chart of available time indicating the percentage of "dark" time for each night. "Dark" time is measured by determining what percentage of each night has no moonlight. The coordinates of the moon between the evening and morning twilights are used to determine the relative darkness of each night. The University of California publishes a Keck Observing Calendar indicating the percentage of darkness for each night. The nights are then grouped into "dark" (less than 68 percemt of the night has no moonlight), "grey" (between 32 - 68 percent of the night has no moonlight), or "bright" (less than 32 percent of the night has no moonlight) categories.
Next, the right ascension and declination of the sky over Mauna Kea are plotted. At different times of year, different stars are visible in the sky, depending on their east/west (right ascension) and north/south (declination) coordinates. "Imagine the Earth is a glass sphere with a light bulb in the center," explains Schaefer. "If the light projected the longitude and latitude lines of the Earth out onto the sky, these lines would define the right ascension and declination of celestial objects. As the Earth rotates on its axis and revolves around the Sun, different right ascension and declination coordinates are visible from our location on the planet."
This is of key importance to astronomers who wish to observe scientific targets that may only be visible during certain months of the year. For example, the galactic center is only visible from Keck Observatory from late April through July 2008. Another important consideration is the local sidereal time (LST), or the time according to the stars. The LST is four minutes per night different than Earth's 24-hour clock. By calculating the LST midway through each night, one can determine which stars will be directly overhead midway through that night. All of these factors - the LST, right ascension and declination coordinates, and darkness rating - are critically important to astronomers in determining when the optimum times to schedule their observations will be.
For example, for cosmologists who study faint galaxies at the far reaches of our universe, dark time is crucial. The visible wavelengths emitted by these targets are so faint that they must be observed when the moon is down and the night is more than 68 percent free of moonlight. On the other hand, certain instruments that measure infrared wavelengths are not adversely affected by moonlight, so observations utilizing these instruments can be scheduled on bright nights with more moonlight. In the first semester of 2008 there are 182 nights, and roughly one third of these nights are dark (58 nights), one third are grey (56 nights), and one third are bright (68 nights).
Image: A sample table showing the percent darkness, right ascension (RA) and declination (Dec), and local sidereal time (LST) for some of the nights in the first semester of 2008. Table courtesy of Barbara Schaefer.
Schaefer compiles a "time available" chart, identifying how many of each kind of night there are in each month in the semester. After subtracting out the engineering nights, she then begins the complicated process of allocating nights among the various partner institutions. The University of Hawai`i (UH) gets 10 percent of available nights on the Keck I Telescope and 15 percent of available nights on the Keck II Telescope. Keck Observatory's Director is awarded one night per month, spread evenly between dark, grey, and bright nights over the course of the semester. NASA has one-sixth of however many nights remain after UH and the Director get their cuts. For the first semester of 2008, the Telescope System Instrumentation Program (TSIP) was awarded four nights on each telescope; this allocation varies by semester and TSIP will receive eight nights on each telescope in the second semester of 2008. Together, these four constituencies (UH, Director, NASA, and TSIP) account for roughly one-third of all the available nights. The remaining two-thirds of the available nights are divided between Caltech and the University of California. Each partner's total number of nights is carefully divided into thirds, so that everyone gets some dark, some grey, and some bright nights.
One final piece of the puzzle is the specifications for the various instruments available at the Observatory. For example, the interferometer (IF) uses both telescopes together for the entire night, and these nights are typically scheduled first to ensure that both telescopes are available on the same nights. Certain instruments occupy the same focal position on the telescope, and time must be allocated between scheduling these instruments to allow for them to be physically exchanged. Staff limitations also come into play, since special expertise is required to operate the laser guide star adaptive optic systems, for example. Other factors which must be considered include scheduled upgrades to a particular instrument, which could take that instrument out of commission for a week or more. Partner institutions are notified of any specific limitations or additions to instrument availability - along with their time allocations for the upcoming semester. The percent dark table, time available chart, and instrumentation specifications are sent to each of the partner institutions, with instructions to "spread their time as evenly as possible over the semester."
Four to five months prior to the start of the upcoming semester, the Time Allocation Committees (TACs) at each institution accept proposals for observing time from their respective constituencies. The TACs utilize a peer review process to rate the incoming proposals. Most institutions receive many times more proposals than can be accommodated in the available time, so the TACs play a critical role in reviewing and ranking the various proposals. All proposals that the TACs deem worthy of telescope time are sent to Schaefer along with a suggested time table for accommodating the highest ranking proposals.
"On May 1 and November 1, I close my door," says Schaefer.
Schaefer then compiles a draft schedule by placing all of the observers into time slots on a master spreadsheet. When conflicts arise, Schaefer often emails the affected partner institutions to get their input on adjusting the schedule to accommodate everyone's needs. About two and a half months prior to the start of the new semester, Schaefer sends the finished draft out to the partner institutions and the executive staff of the Observatory for final approval. They respond with suggested changes, if any, and two months prior to the start of the semester, the schedule is posted online. The final step is to assign support astronomers (SAs) and observing assistants (OAs) for each night - a step that is completed about one month prior to the start of the semester.
The majority of applicants are research and/or teaching staff at the partner institutions, and, of course, there is a high percentage of returning observers over the years. UC is increasingly awarding half nights to different observing teams, in order to maximize time on the Keck Telescopes, since some objects are only visible in the sky for half the night. Schaefer estimates that between 1300 and 1400 individuals have observed at Keck Observatory over the years, and she is now compiling a photographic log of each of the individual observers. The number of PIs (Principal Investigators or those scientists actually granted time for their proposals) is smaller than 1300-1400. The larger numbers apply to those who have actually come observing and include the PIs, their graduate students, and others collaborating on projects. In FY07, 397 astronomers observed on the Kecks.
For the past few semesters, UC and Caltech have traded a very few Keck Observatory nights for nights at the neighboring Gemini and Subaru telescopes. This practice allows researchers shared access to unique facilities and instruments at the neighboring observatories. In return, scientists who generally use Gemini and Subaru are able to make use of the unique instruments at Keck Observatory.
Other extraneous factors that may impact the schedule for a particular night are called "targets of opportunity." These are transitory cosmic events such as an exploding comet or a gamma ray burst that occur randomly during times the telescopes have already been scheduled for other projects. When such events occur, the scheduled astronomer may receive a phone call from an off-site colleague to request that the telescope be redirected to observe the opportunistic phenomena. "This happens a couple of nights per semester," reports Schaefer.
Photo: Barbara Schaefer celebrates the successful completion of another semester's schedule. Photo by Gloria Martin.
Because time on the Keck Telescopes is so precious, the scientists who observe at Keck almost always do their observing on-site. When scientists are on-site, they are able to make instantaneous decisions about where to look in the sky, what to focus on, and what to focus on as a backup target in case of technical difficulties. This dynamic decision-making generates robust science and optimizes the scientific output of the Keck Telescopes. Other telescopes which rely on a computerized scheduling system sacrifice the human ingenuity factor which is so significant at Keck Observatory.
"Scientists like to be able to make decisions on the fly and do their own science. At Keck, our scientists are intimately involved in doing science 100 percent of their time here." - Barbara Schaefer