--> Queue Planning and Execution | Gemini Observatory

Change page style: 

Queue Planning and Execution

You are here

This page contains the following sections:

 

 

Queue Management and Planning

At each telescope a team of queue coordinators (QC) is scheduled to plan the queue nights. Three of these at each site also serve as the core-QCs, who have the responsibility to oversee the queue execution over the longer term and ensure priorities are set consistently and decisions are taken correctly to optimize the program completion. The queue planning is distributed between about 10 staff members, who also have many other duties. About 20-25% of their time is currently used on planning and managing the queue.

The QCs rely on the content of the Observing Database (ODB) to ensure they have a full overview of the queue content. On a nightly basis the scheduled QC uses the Queue Planning Tool (QPT) shown below to construct the plan. Several variant plans are made for different sets of observing conditions, e.g. photometric & good seeing (CC=50% IQ=70%), cirrus & good seeing (CC=70% IQ=70%), cirrus & poor seeing (CC=70% IQ=85%), etc. Thus, the night is pre-planned for not only the most likely conditions but also changing conditions should they happen. The QC uses available weather forecasting and satellite images of cloud patterns to aid the decisions about how to optimize the plan for the night.

The top-level principles guiding the queue management and planning are (1) use the largest possible fraction of the time on completing band 1 and 2 programs, and (2) make the best use of all telescope time. These principles translate to the following concrete guidelines for the queue coordinators:

 

  • Within a ranking band all programs have equal priority.
  • Aim to complete started programs.
  • Do not lose early targets in band 1 and band 2 programs. This can override band ranking if needed to ensure Band 2 targets are observed.
  • Limit the execution in better than requested conditions, in order to enable completion of the more demanding band 1 and 2 programs.
  • Select Band 3 programs to start based on: probability of reaching the PI-defined minimum useful time, probability of completion, complementary to band 1 and 2 in conditions and RA, and connection to a thesis program.

 

In addition to these principles, the queue coordinators plan with the completion rate goals in mind. The goals are endorsed by the Gemini Board and focus on reaching high completion rates in band 1 and 2, and to obtain publishable datasets for started band 3 programs. The completion rate goals are detailed on the operations statistics page.


Queue Planning Tool

Figure 1: Screenshot of the main view of the Queue Planning Tool. Top left lists the plan variants for the night, matching different observing conditions. Bottom left lists observations available to be scheduled. Top right shows the elevation plot for the observations scheduled in the selected plan variant. Below the elevation plot is a listing of those observations, followed by detail for the selected observation.

Execution of the queue – A night at Gemini

A queue night at Gemini usually has only two staff members present, a Science Operations Specialist (SOS) who operates the telescope and provides technical expertise to assist the Observer with the instruments, and an Observer (either another SOS or a staff astronomer) who operates the instruments and exectutes the planned queue observations. The team will assess the observing conditions using available monitoring (seeing monitors, all sky cameras, or by just plain looking at the sky). They will then choose the appropriate plan variant put together by the QC and execute the observations as laid out in the plan. A typical plan contains observations from three to five different queue programs in order to optimize the use of the night.

In the event of changing conditions, the Observer will switch plan variant as needed to avoid taking data in poorer conditions than required and to minimize the time spent taking data in better than requested conditions. Switching between the three or four mounted instruments is accomplished by moving a fold mirror and happens in a few minutes during the slew to the new target. Currently 75-80% of all queue nights use two or more instruments. Due to the large demand for GMOS at both sites there are nights in dark time and good conditions during which only GMOS is used.

If technical problems occur, the Observer and SOS can contact both engineering and science staff for assistance. In case technical problems are confined to one of the mounted instruments, the troubleshooting can be deferred to the next day and the queue execution will simply switch to another instrument, limiting the technical time loss at night.

As the data are taken, the Observer will perform a preliminary assessment of the data quality in order to make decisions about whether the observation is completed correctly and also to detect any technical problems. All data are transferred directly to the Gemini Archive and are usually accessible to the users within minutes of being obtained. This is essential for the timely delivery of data from Rapid Target-of-Opportunity (ToO) observations. Final data quality assessment is done the day after the observations by staff members working at sea-level. This is a more thorough assessment to verify: (1) the correctness of the headers; (2) the observing condition constraints are met; (3) the necessary calibrations have been obtained; and (4) that the data have successfully arrived at the archive. If needed, header corrections will be made and the data re-ingested. Although not required to obtain the data, program data are grouped into “packages” on a regular basis (nominally weekly) and provided to the PI for convenience in downloading the data.

Current Modifications to the Gemini Queue Operations

Gemini is in the process of assessing how to make the management, planning and execution of the queue less demanding of staff effort. In this process, staff effort will be used for development and testing of improved science operations software. In order to have effort available for this work and to optimize where effort is spent, the Observatory will put in place some changes to the queue execution, that are intended to save effort and have minimal impact on the science productivity of Gemini. The first changes took effect in early 2010B. The changes have been endorsed by the Operations Working Group and/or the Gemini Board, and will remain in effect until improved science operations software become available, or beyond that if the savings in effort are still necessary and/or if the impact on the science productivity of Gemini is found to be negligible. Additional changes may be required and will in that case be described on this page. The current changes and their expected impact on Gemini users are summarized in the following sections.

The changes to the queue operations, as well as the work on the improved science operations software, are part the ongoing process of changes to transition the Observatory operations to the funding level expected following the UK withdrawal from the partnership. See the Transition Plan for more information about the transition in general.

 

  • No queue planning for poor observing conditions (effective Sept 2010 - present)

    QCs will no longer produce plan variants for thick clouds (CC=80 or worse combined with any other observing conditions) and very poor seeing (IQ=Any combined with any other conditions). The QCs will ensure that programs are activated/deactivated in the ODB to optimize completion of band 1-3 programs using these conditions. The Observer will select appropriate observations from the active programs in the ODB for observation in these conditions. Designating programs as "active" and "inactive" is a dynamic process and the selected programs will likely change frequently throughout the semester. If necessary, Band 1 and 2 programs using poor conditions may still explicitly be scheduled to ensure completion.

    This change may lead to somewhat lower efficiency on the sky in these poor observing conditions, which in turn may affect the completion rates for band 3 and poor weather programs.


  • Simplified detailed time accounting (effective Sept 2010 - July 2011)

    The principles of the telescope time accounting are summarized on the time accounting page. To save effort previously used in achieving accuracy on the few-minutes level, QCs will be asked to apply a “too small to worry about” criterion to the nightly time accounting consistency. Anything that affects time accounting with less than about 15min/night (technical faults, weather etc.) or about 5min/prg on a given night is considered to be in this category. This change may result in small “overcharges” (or “undercharges”) in some programs. It is not expected that these will affect the partner imbalances. Below a total of 30min or 10% of a program length (whichever is smallest) QCs will not follow up on PI/NGO questions about time accounting unless requested by local Head of Science Operations. The QCs will allow small overruns of allocated time to complete programs as planned. For reference, in semesters 2009A-2010A the average time usage to complete a program was 104% of the allocated program time. This change is superseded by the following change as of August 2011.


  • Suspension of time accounting corrections less than one hour (effective from August 2011)

    Effective August 2011 (at the start of 2011B) Gemini will suspend telescope time accounting corrections, as well as corrections to internal night-logs and fault reports for all corrections less than 1 hour of duration. The change was approved by the Board in early June 2011. Statistics from four previous semesters show that the effect on the partner imbalances is minimal as the corrections are approximately distributed as the nominal partner shares. The effect is in all cases expected to be less than 3.5% of a partner's share in a given semester.

    As a consequence of the change, the program time accounting is less accurate and programs may appear to run out of allocated time before they are completed. To avoid stopping executions of programs prematurely, NGO support and Contact scientists must ensure that each program contains a top level note with the title “Phase II-filling” that states how the program time was filled, choosing from one of these cases

     

    1. Phase II was filled to the allocated time, execute all science observations.
    2. Phase II was overfilled, of the N science observations execute only M of them (this option is used when the PI overfills the program to provide flexibility in the queue execution).
    3. Phase II was not filled as targets are not known (use for ToO programs).
    4. The phase II was filled to only N hours, as the PI wants to assess data from the first observations before defining observations for the remainder of the allocation.

    The note should be dated and carry the initials of the NGO and/or CS who added the note. In the last two cases, the NGO/CS edits the note as new observations are defined. The phase II definitions use the average overhead times (acquisitions and readout) to calculate the planned time. Thus, on average the planned time for an observation is a good approximation of the actual time needed for that observation.

    The QCs have been given instructions in how to avoid stopping execution of a program before it is complete. The QCs depend on the notes described above being present in the programs. The PI and NGO for a program can request a review of the time accounting for a given program, if they believe that the execution of a program has been stopped before completion due to missing time accounting corrections. The Head of Science Operations or core-QC can decide to approve execution of the remaining observations without spending resources on an extensive review of the time accounting. The Head of Science Operations and core-QCs will respond to requests for review within one week.


  • No band 4 (poor weather) time accounting, program revisions and data quality assessment (effective Sept 2010-present)

    QCs will not adjust time accounting for band 4 (poor weather) programs or revise phase IIs for the programs to enable re-observation of targets. Quality assessment (QA) checks will not be done for band 4 data. All data retain the QA state the Observer may have set, or left undefined if the Observer did not set the state. It is possible that some QA states may be incorrect. The Observer will still add comments to the e-obslog as relevant for these observations, and daytime staff members will still ensure that all data taken arrive at the GSA.

    If re-observations of targets are needed, or calibrations are missing, for band 4 (poor weather) programs it will be the responsibility of the PIs to revise phaseIIs to add the required observations as copies of the existing observations. The NGOs can assist as needed. The observations should be clearly marked as a copy in the title and will be set to “ready” by CS/QC without further checks. Re-observations will only be executed if the PI defines them. It is recommended that Band 4 PIs fetch their program from the ODB and download the data from the GSA on a regular basis to monitor the progress of their programs.


  • Closing the telescope based on forecast for very poor conditions (effective from Sept 2010-present)

    The local Head of Science Operations may choose to close the telescope or call off use of the laser guide star system based on early weather forecasts if these are sufficiently poor. At Gemini North it is expected that the decision can be made based on the morning forecast, thereby saving staff effort for the day/night. In addition, the night crew will be given more leeway to close the telescope if observations in poor conditions are extremely inefficient or are putting unnecessary risk on the telescope or instruments. The change may result in fewer poor weather programs being executed.

 


Gemini Observatory Participants


Queue Planning and Execution | Gemini Observatory

Error

The website encountered an unexpected error. Please try again later.