Achieving Stable Observing Schedules in an Unstable World

Space Telescope Science Institute

3700 San Martin Drive

Baltimore, MD 21218

USA

Abstract

Operations of the Hubble Space Telescope (HST) require the creation of stable and efficient observation schedules in an environment where inputs to the plan can change daily. Operations must allow observers to adjust observation parameters after submitting the proposal. PIs must also be informed well in advance the approximate date of an observation so they can plan for coordinated observations and data analysis. Scheduling is complicated due to ongoing changes in the HST operational parameters and because the precise ephemeris for HST is not known in advance. Given these constraints, it is not possible to create a single static schedule of observations. Instead scheduling should be considered an ongoing process which creates and refines schedules as required. Unlike other applications of replanning, the HST problem places a premium on ensuring that a replan minimally disturbs the existing plan. A process and architecture is presented which achieves these goals by dividing scheduling into long term and short term components. The long term scheduler, the main focus of this paper, provides approximate 4-8 week plan windows for observations. A plan window is a subset of an observation's constraint windows, and represents a best effort commitment to schedule in the window. The long range planner ensures plan stability, balances resources, and provides the short term scheduler with the proper mixture of visits to create week long schedules. The short term scheduler builds efficient week long observation schedules by selecting observations who have plan windows open within the week.

The long term scheduler as implemented within the Spike software system provides support for achieving stable observations schedules. Spike models the planning process as a function which takes as input a previous plan, a set of proposals, and some search criteria and produces as output a new plan. Stability is ensured by using the input plan to guide the creation of a new plan. Through this mechanism Spike can handle instabilities such as changed observation specifications, out of date observation products, and errors in loading observation specifications. Special routines are provided for planning and ensuring stability for observations linked by timing requirements (e.g. Observation 2 after observation 1 by 6-8 days). Spike provides a combination heuristic and stochastic search engine with user defined weights for finding near optimal plans.