Preparing Your Observations
The terminology used in WFCAM observations can be confusing. It is an area where different people use different words to describe the same concepts, and different people use the same words to describe different concepts. Thus we present here a reference set of terminology. Please stick to these words when describing WFCAM observations. If you believe there is a concept that needs a word, that is not described here, please let us know.
An MSB is simply an ordered list of observations.
An Observation is a slew to a given position on the sky and the acquisition of a single guide star. No other guide stars will be used during the course of the observation. The observation then proceeds with a sequence that may include WFCAM iterators, repeat iterators, offset iterators, microstep iterators, and integrations.
A set of 4 observations, with target positions such that the gaps between the arrays in one observation are filled by another observation. The result is reasonably uniform coverage of a square of the sky with an area of 0.75 square degrees.
Because of the necessary separation between the target positions of the 4 separate observations, a different guide-star will need to be used for each.
An Offset is an accurate shift in the telescope pointing. Note that you cannot change guide-stars by doing an offset. The guide-star is specified in the target component of the observation and remains fixed for the duration of the observation. This means that if the guide-star moves off the autoguider CCD because of the offset, then observations will be unguided until you offset in such a way that brings the same guide-star back onto the autoguider CCD, this can be seen in Figure 1.1(a) of Description of WFCAM.
Because of the way the autoguider works, offsets between autoguided positions are essentially quantized to the grid of possible guide locations. This is visualized in Figure 1.4 of Description of WFCAM.
Offsets are currently specified in arcseconds of RA and Dec. The guide-star will be pulled in to the nearest guide location that corresponds to the end-point of the offset.
Where possible, the system will do a “fast offset”. In a fast offset, the secondary mirror of the telescope quickly tilts to move the guide star to its new location on the guide CCD, at the same time as the guide software moves to the new guide location on the CCD. Essentially, this moves the image of the sky on the focal plane by exactly the correct amount and allows autoguiding to continue in the new offset position, without actually moving the telescope structure. This can happen very quickly as the heavy telescope structure is not required to move. After guiding in the new offset position has been confirmed and the integration started, the telescope mount and secondary mirror will slowly move in unison to bring the beam back to the optical axis of the telescope. This process is transparent, and because autoguiding is maintained throughout, no image quality degradation results.
The formal conditions that allow a fast offset are somewhat complex, but in general, offsets less than 10″ will be fast offsets, and thus will be much quicker and more efficient than larger offsets.
Because of distortion over the WFCAM field, two integrations that are offset more than a certain distance will not be able to be mosaiced together without resampling. Typically the limiting case is approximately a 10″ offset.
Simply an ordered list of offsets. Each offset is from the telescope base position (which is generally the position specified in the target component of the observation), rather than the endpoint of the previous offset.
A microstep is a special case of an offset that results in offsets that are a particular number (N+1/2, N+1/3, N+2/3) of pixels on the IR arrays. This process is explained more below.
When a microstep pattern is done at each point in an offset pattern, the microstep offsets are relative to the current offset position, not the telescope base position.
There is a pre-defined set of microstep sequences in the OT microstep component to choose from. These have the exact appropriate offsets specified within them and are not user-editable.
An Integration is a sequence of one or more exposures. The Data Acquisition system averages the exposures together (using the arithmetic mean) so that an integration results in a single image. This image is stored in the raw data file.
Thus, an integration on one of the arrays results in a single image in a data file. Actually, each integration results in a single NDF data structure inside the raw data HDS file. It should be noted at this point that there are 4 separate data flow channels in WFCAM, one for each array. This continues to the point where there are 4 seperate raw data disks, each containing data from an individual array.
It is possible for there to be several integrations within the observation, although, with WFCAM, this only happens for focus measurements. In this case, the raw data HDS container file contains several NDF components, one for each integration.
The number of coadds is simply the number of exposures per integration.
To “do coadds” is an informal expression implying that one’s integrations consist of more than one exposure
An Exposure is a single complete cycle of the array readout mode, generally, an array reset followed by one or more array reads over a time specified as the Exposure Time. Simplistically, it is the atomic unit of observation.
For details of what happens during an exposure, including why the elapsed time for exposure is greater than the exposure time, see “Readout Modes”.
The data values that result from an exposure represent the number of electrons detected by each pixel over a time period equal to the exposure time.
The data resulting from each individual exposure is not necessarily saved to disc.
Guidelines for Scientific Survey MSB Design
These guidelines are written primarily for those preparing MSBs for the UKIDSS. In general, individual proposal users should follow them too, and consult with the WFCAM support scientist well in advance if they believe they need to deviate significantly from these guidelines.
The following unconfirmed assumptions are notable on the basis of these guidelines.
- Darks are stable for a timescale of >24 hours. A dark image is a function of exposure time only over this stability period.
- Bad / Hot pixels can be derived from darks and either daytime (dome) flats or twilight flats, and are stable over timescales >24 hours.
- Reset Anomaly is predictable (i.e. either stable or can be simply modeled if necessary)
- Linearity Measures are available and are stable over timescales of >1 week
- Cross-talk matrices are available and are stable over timescales of >1 week
- Image Persistence is negligible and can be ignored. Although lab tests at ATC look promising, we should bear in mind the possibility that this might not be strictly true, and keep our options open in that respect.
- Calibration MSBs are separate, will include darks and photometric standard fields, and will be observed every hour.
Constraints on MSBs
Maximum Duration: Each MSB should take no more than 1 hour of elapsed time. This is to allow hourly observation of Calibration MSBs.
Minimum Duration: In general, MSBs should not take less than 10 minutes of elapsed time.
Number of different positions on the sky: Each MSB should contain integrations at 16 or more different positions on the sky in each filter used. This is to ensure good sky estimation and as a fallback, in case the processing has to rely on self-calibrating MSBs for instrumental signature removal.
Offset distances: We recommend keeping offsets to <10″. This enables fast-offsetting for efficiency and is well below the distance at which differential distortion requires you to re-sample the data in order to form a mosaic.
Frequency of filter changes: Survey MSBs should not change filters more often than once every 10 minutes. This allows the full set (YZJHK) of filters to be covered in a 1 hour MSB. We recommend minimizing filter changes as much as possible. This both increases operational efficiency and facilitates better sky subtraction and fringe removal.
Positions of observations within an MSB: Multiple science observations within an MSB should be grouped as closely as possible on the sky, to improve observational efficiency and sky estimation.
Offset sky integrations: If an MSB requires offset sky integrations for sky subtraction, then all necessary offset sky observations will be included in that MSB. This ensures both that they are carried out, and that they are carried out within 1 hour (the maximum MSB duration given in these guidelines), which we believe to be the longest recomendable interval for sky subtraction.
Exposure Times and Darks: WFCAM darks are observed at the start of every night; only certain combinations of darks and coadds are observed. The WFCAM FAQ page lists the exposure times and coadds routinely observed. Please stick to one of these combinations for your observations. If you need to use any other exposure time or coadd, please make sure to discuss with your support astronomer.
The WFCAM IR array pixel scale is approximately 0.4″/pixel (see Introduction for more details). This allows a large survey area, though does not critically sample the UKIRT/WFCAM point spread function under most observing conditions. In order to obtain a critical sampling of the PSF, a technique known as microstepping is used.
Basically, microstepping involves taking several separate integrations at precisely offset telescope positions, with the precise offsets corresponding to precise fractions of a pixel on the IR arrays along each axis. The options available are “2×2” and “3×3” microstepping. “2×2” microstepping involves taking 4 integrations, offset in 1/2 pixel steps, “3×3” microstepping takes 9 integrations, offset by 1/3 pixel steps.
Because the autoguider can only allow offsets quantized to its grid of possible guide locations (see Figure 1.4 in Description of WFCAM), the microstep offsets are selected such that the offset corresponds to both an integer number of autoguider offset steps, and N+1/2, N+1/3, or N+2/3 pixel steps on the IR arrays, where N is an arbitrary integer.
The actual step sizes involved are given in Table 4.1. “1×1” microstepping is essentially “no microstepping”.
|Microstep Sequence||Number of Poisitions||Offset Size in Arcseconds||Offset size in IR Pixels||Offset Size in Autoguider Steps|
|3×3 Tiny||9||1.07, 2.13||2.67, 5.33||3, 6|
|3×3 Small||9||2.13, 4.26||5.33, 10.65||6, 12|
|3×3 Medium||9||4.26, 5.33||10.65, 13.33||12, 15|
|3×3 Large||9||5.33, 7.46||13.33, 18.65||15, 21|
All WFCAM observations are required to use a standard, observatory defined, calibration strategy. This ensures uniformity of data in the science archive.
A selection of dark frames is carried out before sunset at the start of the observing night. These contain darks of standard exposure time, coadds, and readout mode combinations. A table of these combinations is given in Table 4.2. If your program requires exposure times not listed in this table, then you must include darks in your own MSBs.
|Exposure time [seconds]||Coadds||Readmode|
Flatfields are obtained from twilight sky. We will also be investigating a tungsten-halogen illuminated screen in the dome. There is no need for an individual observer or PI to prepare flatfield observations, the details are taken care of by the observatory, though observers may of course find that they are required to take (pre-prepared) flat field observations on one or more of their observing nights. The observations are done in twilight when it would be too bright to carry out science observing, so there in no time impact on the project in doing this.
Standard Star Fields
The initial calibration plan for WFCAM involved taking a standard star field every hour throughout the night. In light of analysis done on the data between comissioning in late 2004 and survey data in late 2005, it has been decided that it is sufficient to take standard star field observations every 2 hours during photometric conditions, and in non-photometric conditions to take them at the start of the night and at midnight.
A set of observations of a standard field in all 5 broadband filters takes about 5 minutes. This overhead should generally be included in your telescope time request.
The standard procedure adopted for the calibration is to observe a single standard star in only one of the cameras (currently camera 3), with a 3-point jitter. This assumes that relative zero-points between the four cameras are stable. You might want to center the standard on a different camera if you are not using mainly camera 3.