JWST Observing Overheads and Time Accounting Policy
JWST observing program time allocation requests must account for the total time required to achieve science goals, including on-source integration time, direct overhead activities for a observing program, and indirect overheads associated with observations including calibrations.
When proposing for any observing program, astronomers must specify the time required to execute observations that will achieve the science goals of the program. In addition to on-source integration time, the time associated with various overhead activities (e.g., needed calibrations and set-up for observations) must be accounted for. This report outlines the methodology to account for, and to enable observers to compute the time required for JWST overhead activities associated with science observations as part of the proposal preparation process. Further information on overheads is provided at JWST Observing Overheads and Time Accounting Overview and NASA-SMD Policy 8: Definition of Observing Time.
This report is a response to PPS SDR3 RFA 9, for a decision on how overheads will be applied to science program allocations. August 14, 2014
General types of JWST overheads
Observing overheads (time not spent collecting science photons) associated with JWST science programs can be divided into three broad categories:
- Time due to overheads directly associated with, and largely determined by, specific activities of an observer’s program (hereafter OVERDIR1). These include activities such as guide star acquisition, filter and detector configuration changes, and small angle maneuvers (SAMs) to position or reposition the science target (e.g., target acquisitions, dithers). These are also known as deterministic direct overheads.
- Time due to overheads directly associated with, but which are generally not solely determined by, the specifications of an observer’s program (hereafter OVERDIR2). The time spent executing major slews falls into this category since visits in an observer’s program may be interspersed with visits from other programs to minimize angular momentum unloads and maximize the efficiency of the observatory. Since JWST operations will be event-driven, the exact slew time to an observer’s target will not be known until the observing schedule is finalized, and the observation is executed. These are also known as statistical direct overheads.
- Time due to overheads indirectly incurred in general support of science observations (hereafter OVERIND). These include activities such as angular momentum unloads, wavefront sensing and control, instrument calibration, and observatory maintenance. These are also known as indirect overheads.
An early efficiency benchmark recommended that the sum of all overhead activities for programs in the Design Reference Mission (DRM) should not exceed ~30% of wall-clock time (NGST Monograph 5). A more recent, comprehensive description of all known overhead activities involving each of the four current JWST science instruments, the observatory, and the ground segment is given in the Science Observatory and Instrument Overheads Technical Report (Gordon et al. 2012a). There, indirect overhead activities (OVERIND) are estimated to require ~16% of wall-clock time. A preliminary study of the time required for direct overheads (OVERDIR1, OVERDIR2), based on the latest revision of the Science Operation Design Reference Mission (Gordon et al. 2012b), is described in the JWST Observing Efficiency Report Technical Report (Tumlinson et al. 2014). A key difference between the early DRM and the SODRM 2012 is the inclusion of a larger number of moderate and short duration exposures (with attendant increase in the number of visits and dithers), which is more likely to reflect actual usage in JWST’s first year of science operations. The next step toward more realistic estimates of the overhead times involves scheduling studies based on the SODRM, and the inclusion of the latest results from flight hardware testing. Such work is currently in progress. The most recent Observation Efficiency Allocations Report (JWST-RPT-004166, Revision H, 2014) finds that “The usage and overhead estimates for the science use cases developed in conjunction with the Science Instrument teams and STScI SO-DRM show system-level efficiencies of 64% and 67%, respectively.”
Policies for determining overhead times
In the initial 2001 Announcement of Opportunity for Next Generation Space Telescope Flight Investigations (AO 01-OSS-05), observing time was defined for GTOs to include all three types of overheads described above. Specifically, the AO states: “When estimating the duration of their GTO investigations, proposers should assume the NGST Level 2 requirement of 70% observing efficiency on the entire system (observatory plus ground system).”
Thus, to support GTO observation planning, STScI must establish clear policies to account for the time needed for overhead activities (OVERDIR1, OVERDIR2, OVERIND) by a given science program, and to enable GTOs to accurately estimate this time. Since it is preferable to adopt a single system to support all users when possible, STScI will use the same overhead accounting process and proposal planning tools for GOs. The overhead accounting policy for both GOs and GTOs will be as follows. For each science program:
OVERDIR1 times for a specific GO or GTO observing proposal will be calculated based on the number of OVERDIR1 events required by that proposal and the estimated duration of those events. For Cycle 1 proposals, OVERDIR1 time estimates will be based on measurements taken during cryo-testing, and from simulations of activities that cannot be measured before launch.
OVERDIR2 times will be calculated based on the expected number of OVERDIR2 events (e.g., major slews) required by the observing proposal and an average duration of those events. For Cycle 1 proposals, the average will be based on simulations, and for proposals for subsequent cycles, it will be computed from previously executed programs.
OVERIND times will be computed as a rate -- a statistical assessment to account for the aggregate duration of those activities over a given cycle. It will be calculated by multiplying the sum of the integration time and OVERDIR1+OVERDIR2 by 1/(1- f_OVERIND) where f_OVERIND is the fraction of wall-clock time previously spent on OVERIND activities. For Cycle 1 proposals, f_OVERIND will be based on simulations.
The most significant OVERDIR2 activity will be due to major slews (e.g., Tumlinson et al. 2014). To calculate the number of major slews needed by a program, the Astronomers' Proposal Tool (APT) will analyze the requirements of all of the visits in that program (e.g., positions, durations, guide star availability constraints, and determine which are likely to be scheduled together. The visits grouped in this way will be referred to as “Same Scheduling Sets” (SSS). The statistical cost of the major slews will be assessed upon these Same Scheduling Sets. The algorithms to be used to group visits into SSS in known as "Smart Accounting" and is built into the APT.
Implementation of overhead accounting policy
Implementation of this accounting policy will include the following elements:
- The policies above, as well as the indirect overhead rate for the cycle (1/(1- f_OVERIND)), will be communicated in the Call for Proposals.
- The Astronomer's Proposal Tool (APT) will calculate the time required by the overhead activities associated with an observing program. APT will provide a breakdown of the overhead times as well as the total time to be requested during the proposal process.
- All times will be computed to the nearest second, and total observing time requests will be expressed and communicated to the proposer in units of hours to two decimal places. For GTOs, the total time requested by their programs will be deducted from their global allocation (as specified in the AO), after the programs are checked for duplications and finalized. For GOs, the total time requested by their programs will be provided to the TAC for review. The total time requested by TAC-approved GO programs will be deducted from the GO time available in that cycle.
- Time will be allocated on a per program basis, but STScI Time Allocation Management will track the usage for all programs on the visit level. Programs will be tracked through execution. STScI Time Allocation Management will enable any differences from the approved allocation to be identified.
- APT will adopt the best estimates of the overhead durations needed to compute OVERDIR1, OVERDIR2 and OVERIND, which are available at the time of the writing of the Call for Proposals for each cycle. For Cycle 1, the overhead models will be initially developed using testing of the flight hardware together with scheduling studies of the JWST Science Operations Design Reference Mission. These parameters will be stored in the JWST Project Reference Database and kept under configuration control.
- Updates to the overhead model will be revised as needed for the JWST scheduling system as in-orbit data becomes available, but updates will be released to GOs and GTOs only at appropriate times in the proposing and observing life-cycle. The number of releases will be minimized.
- Integration time requests of approved proposals will be generally protected against modest changes in the overhead rates through STScI Time Allocation Management. Changes to the overhead model that increase the total observing time required by an approved program by up to 10% will routinely be allowed to execute, with no change to the allocated integration time. If the overhead time requirements are lower than estimated, the programs will still execute with the allocated target integration time. For larger increases in the overhead requirements, program execution will be at the Director’s discretion, which will include consideration of observatory and instrument status, and may involve changes to the approved program.
- GTO and GO program overhead charges may need to be recalculated when the time required for overhead activities are properly established, and to compute the remaining time available in subsequent cycles to reflect actual usage. GO and GTO programs may need to be revised to account for changes in instrumental sensitivity and/or background levels. Policies in that regard will be discussed in a separate document.
Differences from customary policies for observers
JWST time allocation policies explicitly attribute the time required by indirect overhead activities to individual observing programs. The more usual policy of space- and ground- based observatories is to make such costs invisible to the user, by reducing, ab initio, the total time available for science by the time required for overhead activities such as instrument calibration and observatory maintenance. Exposing the time needed for indirect overhead activities provides total cost accounting that will allow the overall observatory efficiency to be more transparent to users, and improves general accountability. However, since this policy is not one that observers are accustomed to, STScI will take extra care to ensure that the policy and the potential impact on approved observing programs are communicated clearly.
Lee, J., Reid, I.N., & Lotz, J.M., 2014, Overhead Accounting Policy for JWST Science Users,
Gordon, K., et al. 2012a, JWST Observatory and Instrument Overheads, JWST-STScI002564
Gordon, K., et al. 2012b, Science Operations Design Reference Mission, JWST-STScI-000045, Rev. C
Johns, A. et al. 2011 Proposal Preparation System, System Design Review 3, Request for Action 9 (PPS SDR3 RFA 9)
Mather, J. et al., NGST Monograph No. 5: System Level Requirements, Recommendations, and Guidelines, (2000).
Mitchell, L., Observation Efficiency Allocations Report, JWST-RPT-004166, (2010).
Tumlinson, J., et al. 2014, JWST Observing Efficiency: Baseline Results from SODRM 2012, JWST- STScI-003350
2001 Announcement of Opportunity for Next Generation Space Telescope Flight Investigations (AO 01-OSS-05)