Service Mode Rules and Recommendations for Observation Blocks

Preparing Observation Blocks

Observations at all ESO telescopes are carried out by executing Observation Blocks (OBs) provided by the users. OBs for Service Mode runs with Paranal Instruments must be made with p2. For (designated) Visitor Mode observation preparation, please follow dedicated Visitor Mode Guidelines.

Please refer to the Phase 2 step-by-step preparation with p2 page and to the User Manuals of the different instruments for more specific information on the structure and content of OBs, and how to build OBs for different instruments. A number of tutorials describing step-by-step the construction of OBs for different instruments is available.

Service Mode OBs: rules and advices

It is important to keep in mind the Service Mode policies and the following rules and guidelines when designing a Service Mode programme or when preparing a Phase 2 package:

  • Some observing strategies cannot be supported in Service Mode; in particular, real-time decisions about the sequencing of OBs, complex OB sequencing, or decisions based on the outcome of previously executed OBs (like adjustment of integration times or execution of some OBs instead of others).
  • OBs are only executed once. If you want to repeat an identical observation multiple times, you must submit multiple OBs. This requirement applies to standard stars as well.
  • OBs are normally executed non-contiguously. Since efficient Service Mode operations require continuous flexibility to best match the OB constraints with actual observing conditions, OBs for a given programme may be scheduled non-contiguously. Therefore, users should not expect their OBs to be executed in a specific sequence or in a linked way. Usuers who do require contigous observations might use concatenation scheduling container (see definition below) by follwing the rules applied to this case.
  • Multi-mode, multi-configuration OBs are normally not permitted in Service Mode. Although multiple configurations (e.g. combining imaging and spectroscopy) within one OB may sometimes reduce overheads, scheduling and calibrating such OBs is inefficient and can increase the calibration load to an unsustainable level. Multi-configuration OBs are accepted only if duly justified and authorized by means of a Phase 2 Waiver Request.
  • OB execution times must be below 1 hour. Long OBs are more difficult to schedule and execute within the specified constraints because of the unpredictable evolution of the observing conditions. For this reason, OBs taking more than one hour to execute are accepted by ESO only in exceptional cases and provided that a Phase 2 Waiver Request is submitted and approved. In such cases, ESO will consider the OB successfully executed if the constraints were fulfilled during the first hour of execution, even if conditions degrade after that time.
  • Concatenation scheduling container execution time must be below 1 hour. Only in exceptional cases, and provided that a Phase 2 Waiver Request is submitted and approved, longer concatenations may be submitted. In such cases, ESO will consider the concatenated OBs successfully executed if the constraints were fulfilled during the first hour of execution, even if conditions degrade after that time.
  • User-provided calibration OBs that need to be executed contiguously with science OBs need to be specified via concatenation scheduling containers.
  • Time constraints must be indicated in the OBs. If a user intends to observe time-critical events or monitor a target at specific time windows, this must be indicated under the Time Intervals tab of the OBs. Please note that absolute (UT) time constraints refer to the interval in which the OB can be started, whereas for Local Sidereal Time (LST) time intervals, the time interval refers to the entire duration of the OB. For monitoring observations it is often more appropriate to put OBs in a time-link container. Specifying time windows as broad as possible will reduce the possibilities that your OBs are not executed because of higher priority programmes or because the observing conditions did not allow the observations during the interval that you specified.  Usage of absolute time intervals must be scientifically justified in the README file. Please read carefully the time-critial OB execution policy.
  • Specify the weakest (most relaxed) possible Constraint Set values. OBs that can be executed under a broad range of conditions are easier to schedule. In particular, for photometric calibration it is normally sufficient to obtain a short integration under photometric conditions (transparency = PHO) and carry out the rest of the integration with OBs having a transparency = CLR constraint.
  • Nested scheduling containers (see definition below) should be used only if strictly necessary as they increase complexity for observations and scheduling.

Time-linking of OBs

Some OBs must be executed within precise time windows for scientific reasons, rather than any time when the external conditions (moon, seeing, transparency...) would allow the execution. The following types of time-dependencies can be recognized:

  • Absolute time constraints, meaning that an OB must be executed at specific dates that can be predetermined. An example is the observation of a binary star at a precise phase of its period or a planetary transit observation.
  • Relative time links, implying that an OB must be executed within a time interval after the execution of a previous OB, but not necessarily at a fixed date. Examples of this are monitoring observations of a variable source at some pre-defined intervals.

Both types of time-dependency are implemented within p2. Whereas absolute time constraints are available at the level of single OBs, the relative time links are implemented within the new "Time Link" container.

Within a Time Link container, the user can define a series of OBs, having the earliest and latest time when a given OB in the series must be executed with respect to the preceding OB. The time-related information is stored in a database, from where it is retrieved by scheduling tools available to the operator on the mountain in order to build up a short-term schedule that properly takes these constraints into account.

If an OB with absolute time constraint or time-linked OB that acquired an absolute time constraint following execution of a previous OB in sequence (i.e. OB to be observed after earliest from and before latest from time of the previous OB in the sequence) is not successfully completed within the specified time interval, it will expire and get status F(ailed). Such an OB is not observable any more and policy for time-critical OB execution applies.

If the time-linked OB expired in the middle of a time-link sequence, the sequence execution continues as follows:

  • If the failed OB is not the very first OB of the time link, it had the absolute time window corresponding to delay from the previously executed OB. After it expired the next OB acquires an absolute time window by adding the relative minimum and maximum time delays to an assumed hypothetical execution for the failed OB in the middle of its constraint window. 
  • If the failed OB is the very first OB of the time link, the failure can only occur if this OB has one or more absolute time constraints defined and all of them have expired. In this case the next OB acquires an absolute time window by adding its relative minimum and maximum time delays to an assumed hypothetical execution of the failed OB at the end of its last absolute time interval. 

It should be noticed that, depending on the length of the relative time intervals, and the delays between them, a failure of an OB in a sequence may result in a cascade of failing OBs.

Concatenation of OBs

In some cases it may be desired to execute the OBs consecutively, with no other observations in between. This has been implemented in p2 within the "Concatenation" container. The Concatenation container consists of two or more OBs that must be executed "back-to-back" without breaks. The sequence of the execution of OBs in a Concatenation typically follows the sequence as they are listed in the p2 window. 

Definition of groups of OBs

Groups of OBs are used to express the preference to complete observations of a given group of OBs before continuing with other OBs (or groups of OBs) within the same observing run. This is the most loose scheduling container concept, and the priority for execution of the group with respect to other groups within the same run is defined though group priority that has values 1-10 (1 top, 10 lowest priority) as for the user priority for loose OBs. The priority for execution of OBs within the given group is regulated through the OB group contribution. 

If OBs within the group, whose observation started, are not observable (constraints are not fulfilled), it is possible to start observations of another group. After that group score defines which groupo will be given priority in case both groups of OBs are observable again. 

Definition of Nested Containers

For Service Mode observations that use VLT or VLTI instruments on Paranal it is now possible to design more complex observing strategies in the p2 tool with nested scheduling containers. For example a science case that requires time-monitoring of a set of concatenations of science+telluric OBs can be expressed as time-links of concatenations. For the VLTI imaging observations use of groups of pairs of science+calibrator is mandatory, such that the group defines the set of concatenations that contribute to the same image or uv plane. 


Additional Service Mode Requirements for KMOS

Selected DIT Values

DIT values can be freely chosen. For short exposures, however, the minimum DIT is 2.47s. For long exposures on faint targets with no risk of saturation we recommend to work with NDIT=1 and a DIT depending on the frequency of sky frames needed. A DIT >1000s is only allowed for the IZ band.

Note that the background limited performance between the OH lines is reached in 300s.

Saturation and persistence limits

It is important to avoid saturation of the detectors because the persistence artefacts can last up to several hours and thus affect subesequent observations.

Therefore, targets should not be brighter than 6th magnitude in J,H,K bands.

There are also persistence effects for illuminations of >40.000 ADUs/DIT/pixel. Please make sure, with the help of the ETC, that such fluxes are avoided for your brightest targets, even for better conditions than foreseen (e.g. using seeing 0.4" in the ETC).

Since telluric standard stars, as part of our calibration plan, are taken with three arms, one per detector, it is advisable to avoid placing faint, high priority targets on those arms. At the moment Arms #3, #13 and #21 are allocated to standards, but this might change depending on arm failures (see the KMOS news for the most recent information).  

Absolute offsets

Please note that, unlike as for all other VLT instruments, the offsets (e.g. dither, etc.) for KMOS are defined in an absolute sense, referring to the initial pointing position. I.e. the offsets are not cumulative, relative to the previous position.

Rotator optimization and priority 1 targets

Rotator optimization is applied if one or more arms are locked for a certain period due to technical problems. In this case, the instrument is rotated in such a way that the maximum number of priority 1 and 2 targets at the Telescope at Science position are covered by the available arms. To make this work it is important that the science targets have a range of priorities from 1 to 3. Note that science targets that are defined at the Telescope at Sky position are not considered for rotator optimization. Priority 1 and 2 targets in this position can thus be lost if the corresponding arm is locked.

Some KMOS acquisition templates allow to suppress the rotator optimization. This might be useful if you want to enforce that your main target is on a certain IFU or if you need to make sure that 3 IFUs allocated on sky, i.e. one per detector, keep their positions (but look at the KMOS news whether these arms are indeed available). Note, however, that in this case you accept that more than 10% of your priority 1 targets can get lost.

Instrument selector

On this page: