Observing Condition Constraints

All queue-mode observations must have observing condition constraints specified by the proposer that describe the minimum (i.e. poorest) conditions under which the observation should be executed. Starting in 2007B, classical programmes must also specify the minimum acceptable conditions and, optionally, a backup programme able to take advantage of poorer conditions. The observing condition constraints must be specified in the Phase I proposal to avoid loading the queue entirely with one type of conditions (e.g. best image quality).

The constraints are divided into five categories (if appropriate, values for Mauna Kea and Cerro Pachon are given separately):

• Image quality
• Sky transparency (cloud cover)
• Sky transparency (water vapour content)
• Sky and telescope background
• Air mass (zenith distance)

The specific properties corresponding to these categories usually are wavelength dependent and will not be relevant for all observations. For example, the sky background at visible wavelengths is dominated by the lunar phase and moon-to-target angle. At mid-IR wavelength the combination of cloud cover and water vapour condition define the background and its variability. For the image quality, sky transparency and background we have chosen to represent the variation in these conditions (which is deterministic in the case of visible sky brightness, statistical in the case of water vapour column, for example) by a percentile representing the frequency of occurrence of the specific property. Observing constraints are specified in terms of these percentiles (see examples below) e.g. (best) 20%-ile, 50%-ile (better than median) etc.

This page provide a translation between the frequency of occurrence and the specific value for the relevant property as well as further information and guidance on their use by observers. The emphasis is on providing observers with these constraints in meaningful units (and corresponding to those used in the integration time calculator) as well as indicating their likelihood.

Temporal constraints, e.g. for time-critical observations or periodic monitoring, and GMOS-specific constraints (such as those which affect mask cutting) are (to be) described elsewhere.

Several examples serve to illustrate how the specific scientific objectives of a program might affect the users choice of constraints:

1. example - NIRI spectroscopy of an extended object
2. example - NIRI imaging of structure within an extended object

Image Quality (non-AO) - MK and CP

Note that these values apply to the telescope pointing at zenith. The performance degradation away from the zenith can be approximated crudely as (air mass)^0.6 in the visible and short wavelength infrared, and the integration time calculators take into account the dependence of image quality on wavelength (by interpolation) and airmass when calculating signal-to-noise ratios. The exponent is lower and variable in the 10µm and 20µm windows; values being used in the integration time calculators at these wavelengths are uncertain and may be updated. If your program requires a certain absolute image quality (e.g. for resolving objects at small separations) you should consider the possible elevations at which your observations could be executed when deciding upon image quality constraints.

Explanation of table entries:

1. Numerical values in the constraint columns are the current measured delivered image quality, defined as the 50% encircled energy diameter in arcsec, in the telescope focal plane at the specified wavelengths. These values are expected to improve as thermal control of the facility is optimised. The 50% EED is equal to the full width at half maximum for a Gaussian profile.
2. "Any" means that the observation can be scheduled under any image quality conditions. At optical and near-IR wavelength the image quality distribution has a long (non-Gaussian) tail. The values quoted are typical of the poorest conditions.
3. Use of wavefront sensors for image motion compensation (fast guiding) is required. For most wavelengths, values for peripheral and on-instrument WFSs are given. In all cases the use of the PWFS for closed-loop primary mirror figure (aO) correction is assumed.
4. See the specific instrument pages for descriptions of the wavefront sensors required or available for use with each instrument. In this table it is assumed that the WFS star is slightly brighter than the 'knee' in the WFS performance curve. The location of the knee, corresponding to the point where there are insufficient guide star photons to overcome the noise sources, is critically dependent on the readout noise of the WFS detectors.
5. The guide star limiting magnitude depends principally on the seeing, cloud cover and wind speed. More details are provided in the instrument and wavefront sensor pages, but as a guide:
   • PWFS - guide star R-band magnitude limit of ~14 mag (~15 mag if guiding at lower frequency).
   • OIWFS - guide star magnitude limits in the V-band (for the optical OIWFS used with GMOS N & S) of 14-16 mag (see this table) and in the K-band (for the NIFS near-IR OIWFS) of ~14.5 mag in good conditions.
6. Image quality at Gemini South has been measured to be within ~10%(~20%)(~50%) of the theoretical diffraction limit 20%(70%)(85%) of the time in the 10 µm atmospheric window and within 10% of the diffraction limit 85% of the time in the 20 µm atmospheric window. The diffraction-limited FWHM is 0.31" at 11.7µm and 0.49" at 18.3 µm, the central wavelengths of the filters in which the percentiles have been evaluated. Analysis of Gemini North mid-IR data is in progress but image FWHM are expected to be similar. Note that there is little difference between the 20%-ile and 70%-ile image quality bins in the N band. PIs may like to take this into account before requesting IQ20 conditions given their low frequency of occurrence. This table was updated in March 2007 for PIs planning proposals for 2007B; the previous values were <0.35", 0.45", 0.55" and 1.2" in the N band (no constraints given at Q).
7. Estimates of the AO-corrected image quality are described on the Altair pages).

Note that the relevant parameter here is image quality and not simply seeing, that is, a wind speed distribution and the telescope performance (e.g. windshake, servo and wavefront sensor characteristics) have been incorporated into the analysis. The model was adapted by Mark Chun from original Mathematica calculations by Charles Jenkins (see also Jenkins 1998, MNRAS, 294, 69) with a subsequent correction (in August 2002) by Phil Puxley to the extant seeing distribution. 

Interpretation of the table is shown in the following example. An image at K of a target at zenith with a bright guide star in the Peripheral Wavefront Sensor would be expected to have a 50% EED of no more than 0.35 arcsec 20% of the time and no more than 0.55 arcsec 70% of the time.

Sky Transparency (Cloud Cover) - MK and CP

Explanation of table entries:

1. The percentiles are based on long-term data for Mauna Kea and correspond to fractions of the usable time.
2. "Photometric"  - cloudless and capable of delivery stable flux.
3. "Cloudless" - for photometry accurate to a few percent in the mid-IR, careful attention must be paid to regular observations of suitable standard stars. The former 20%-ile ("low sky noise") bin was removed from this table on 2006 May 25.
4. "Patchy cloud" - relatively transparent patches, maybe cirrus, sometimes amongst thicker cloud resulting in some loss in transmission and variability. For the purpose of integration time calculation it is assumed that clearer patches have a transmission that is poorer by 0.3 mag than the nominal atmospheric extinction.
5. "Cloudy" - cloud cover over essentially the whole sky. For the purpose of integration time calculation it is assumed that the transmission is poorer by 2 mag than the nominal atmospheric extinction and is variable. The increase in background makes these conditions unusable at thermal infrared wavelengths.
6. "Usable" - any conditions under which the telescope is open. The increase in background makes these conditions unusable at thermal infrared wavelengths. For the purpose of integration time calculation it is assumed that the transmission is poorer by 3 mag than the nominal atmospheric extinction.

Sky Transparency (Water Vapour)

Explanation of table entries:

1. In the integration time calculator the optical transparency is derived from model transmission spectra. See "Sky Transparency (cloud cover)" for a related constraint.
2. Near- and mid-IR transparencies are characterised by the precipitable water vapour content (in mm) derived from the 225GHz zenith optical depth. The atmospheric absorption is strongly wavelength dependent as shown in model transmission spectra. The CSO 220 GHz optical depth, "tau", is used to determine the current PWV conditions on MK while a Gemini PWV 20 micron camera known as "IRMA" is used on CP.
3. Percentiles for Mauna Kea are based on long-term data. For Cerro Pachon, conditions are assumed to be similar to La Silla, where statistics indicate that the PWV is approximately twice that of Cerro Paranal. Therefore, the percentiles given are based on 1992-1994 data from the ESO/VLT site at Cerro Paranal, multiplied by 2. Note that the percentiles are based on year-round data, but the PWV during Chilean summer (Jan-Mar) is generally double that of the rest of the year, as shown in the Paranal data and the 1999 data from the future ALMA site. The Mauna Kea data do not show a strong seasonal variation. 
4. "any" means that the observation can be scheduled under any conditions.

Sky Background

Explanation of joint table entries:

1. The near-infrared  J- and H-band backgrounds are dominated by OH airglow lines. The K-band comprises both OH and thermal emission. Within the integration time calculator the background is assumed to be constant (once the sun is sufficiently below the horizon) even though the OH component is known to vary during the night (more details). Twilight, with its brighter background, may be usable if near-IR wavefront sensors are installed.
2. Near-IR 3-5um and mid-IR 8-25um background is defined by the combination of water vapour and cloud cover conditions defined above. 
3. All values pertain to the zenith.
4. Optical background values originate from a Monte Carlo simulation of the sky brightness using a model which includes scattered moonlight and zodiacal light, and pertains to high ecliptic latitude. The sky colour is different between constraint bins. Crudely speaking, the moon is below the horizon during about one half of queue-mode hours. These tabulated values are used to scale an empirical sky spectrum within the integration time calculator.
5. "any" means that the observation can be scheduled under any conditions.

Air Mass

This constraint defines the maximum air mass [= sec(zenith distance) = 1/cos(zd)] at which the target should be observed. The air mass affects the sky transparency (e.g. the general atmospheric extinction as well as the depth and breadth of specific absorption bands due to atmospheric constituents such as water vapour and CO₂), sky brightness and image quality. As a crude first approximation, the sky transparency and brightness each become poorer in proportion to the increase in air mass (e.g. sky brightness is twice as great at air mass = 2 than at air mass = 1) and the image quality degrades as (air mass)^0.6.

The airmass constraint is not used at phase I but can be entered in the integration time calculators to show how the expected signal-to-noise for an observation varies with elevation. By default at phase II there is no elevation constraint and it is not possible to edit the elevation constraint field in the observing tool since since this maximizes schedulability. When needed, Gemini staff can set the airmass or hour angle constraints. Use of these constraints is equivalent to a change to better conditions constraints than approved by the ITAC, so approval must be granted via the change request procedure before the elevation constraints can be modified. An example of an observation that would use these constraints is one using GMOS that needs to restrict the hour angles so that the position angle of the slit(s) is close to the parallactic angle. Targets with no elevation constraints will be observed at airmasses < 2.0.

Constraint sets used previously:

• Pre-August 2002
• August 2002 - August 2004

Last update July 29, 2007; Rachel Mason, Tom Geballe and Jim de Buizer