About this ArticleS. Zonak (szonak[at]stsci.edu)
The key is not spending time, but in investing it.
–Stephen R. Covey
The James Webb Space Telescope (JWST) has only a short 5–10 years to observe our Universe and assist us as we work to unravel its mysteries. We have a responsibility to extract as much science as possible from every moment of JWST's active mission.
NASA and STScI are no strangers to developing methodologies for maximizing science observing efficiency. Over the past three decades, the Hubble Space Telescope (HST) has provided the astronomical community with a treasure trove of data. By precisely modeling every onboard activity, like slews and mechanism moves, the ground system can carefully construct a time-ordered list of commands that supports a particular type of observation. Each command, also called an Instruction, is associated with a timestamp that denotes the exact time it is to be executed. This design, called Absolute Time Commanding, minimizes down time between commands and supports a continuous cadence of observations.
Although this approach is a standard method for transforming science templates into a series of flight software commands, onboard failures can negate some of the benefits of this fastidious scheduling method. Over the course of the mission, the quantity of lost time will accumulate, resulting in non-negligible overall observing inefficiencies. Take, for example, the situation where the HST Fine Guidance Sensor is unable to successfully acquire a guide star before an observation. Without a star to guide on, the entire observation, including all its subsequent orbits, must be skipped. Because each individual command must wait until its designated time, this renders the entire observatory idle until the spacecraft time matches the timestamp for the beginning of the next new observation.
Early on in the development of the James Webb Space Telescope (JWST), there was an opportunity to address some of these inefficiencies that limit HST. Unlike Hubble, Webb will orbit at the second Lagrange point (L2), freeing it from significant low-earth orbit scheduling constraints, like passage through the Earth's shadow and weekly ephemeris updates. Despite this freedom, JWST is limited to only eight hours a day for real-time communication.
Each visit file uploaded to the spacecraft contains the critical timing window for which the observation is valid, momentum management data, dither pointing information and a series of activities with corresponding user-friendly parameter/value pairs. Each activity invokes the series of low-level flight software commands needed to implement a particular science observation feature.
What makes this approach so spectacular is that, with full access to real-time telemetry, we are now also free to embed logic into these Onboard Operations Scripts, allowing them to make informed, autonomous decisions based on real-time conditions aboard the spacecraft. A command can commence as soon as telemetry reports that the previous command completed successfully, eliminating the absolute timing restrictions. An isolated, non-fatal error will no longer bring all observatory activities to a halt.
In this new paradigm, if the JWST Fine Guidance Sensor fails to acquire a guide star before a science exposure, OSS can make the decision on the fly to skip the current visit and immediately move on to the next observation set. If telemetry indicates that there is a critical error isolated to only a particular science instrument, the scripts can decide to cease commanding and move directly onto the next observation set. The OPE can also determine whether or not to prevent further observations involving the afflicted instrument until ground intervention deems it is okay to resume.
Supplying the spacecraft with a steady stream of new observations is also critical for keeping science observations running continuously. About every 10 days, the ground system will upload a new set of observations. The OPE is responsible for "listening" for ground requests to add new observation plan segments. Once the request is received, the OPE places it in the queue. In addition to listening for ground requests, the OPE is also responsible for immediately launching a new observation when the previous one completes. Before initiating the next visit, the OPE ensures that the current spacecraft time falls within the valid time window provided in the visit file and that the space remaining on the solid-state recorder is above a certain threshold. If any of these constraints are violated, the OPE can trigger a response that will result in all instruments currently in use to return to a known and benign state so that they are ready to resume normal science operations.
Designing, developing and implementing a brand-new Event-Driven Operations concept was a courageous, yet decisive step in shifting the observing model for NASA spacecraft orbiting at L2. The time invested in making such an advancement allowed not only Webb to reduce onboard inefficiencies, but also the Nancy Grace Roman Space Telescope, which has chosen to adopt the Event-Driven Operations model.
I would like to extend my gratitude to S. Barrow, J. Stys, and A. Welty for providing information regarding HST, as well as the history of the Event-Driven Operations concept. I would also like to thank D. Zak for providing additional context to the history of OSS and his development of the OPE.