MOIRCS Information for Imaging Mode

Optical Distortion and Mosaicking

The distortion of the MOIRCS optics is expressed well by the third-order polynomial as a function of the distance from the optcal center (xc=858, yc=1034 for chip 1, xc=1178, yc=1012 for chip 2). The distortion coefficients will change every time when we execute the warm-up / cool-down process for engineering or when we move the internal focus adjustment system. The date of occurence after the open use is as follows.

15 December, 2005
14 Feb, 2006
24 July, 2006
13 October, 2006
13 January, 2007
09 April, 2007
29 June, 2007
03 Oct, 2007 (Engineering chip)
17 July, 2008 (Scientific chip)

If you need the geotran database file (for IRAF) for correcting distortion, please contact to the SS.

As MOIRCS is the twin-optics system, the affection of the gravitational flexure is independent with each channel. Therefore, at low elevation the rule mosaicking each channel image will slightly change from the one at high elevation (typically 2-6 pixels in EL=30 deg: this may also be affected by the differential atomospheric dispersion).

Please note that the mosaicking rule will also change by the thermal cycle. The detector replace that occured on Oct 2007 and July 2008 has also changed the rule for mosaicking very much.

A Photometric Offset in Partial-Read Mode (!! Important !!)

Previously we announced a possible systematic offset in the photometry between the whole-read and a partial-read data. Through the following engineering tests in late 2008 we confirmed the existence of the systematics. In brief, the apparent change of gain is a function of the partial-read area size and input flux. The raw count (ADU) become larger if the read-out area is smaller or the input light flux is larger.

To evaluate this, we took several datasets of a constant-flux thermal light source with changing the size of partial-read area. To change the flux coming to the detector, we took the data with changing filters to standard-K filter (brightest), Ks filter(medium), and K-contnuum filter (smallest). All data were taken in 21-second exposure (nominally the minimum exposure by whole read under NDUMMYREAD=2 option).

The result is summarized below. In the figure, vertical axis shows the degree of apparent increase of counts with respect to the counts in whole-read mode. Horizontal axis is the size of the partial-read area. The count level in whole-read mode is shown in each label.

As expected, the systematics are a clear function of the partial-read size. But a more strikingly, it shows that the level of the systematics is a strong function of the input flux, in the sense that if the count become higher (more flux), more affection is observed. Usually we use partial-read mode for very bright standard stars to avoid saturation. Therefore, in the real situation during standard-star observation the input flux should be much stronger (x 5-10?) than that of the light source we used here. If the relation between input flux and the level of the sytematics keeps its trend we see here, the maximum offset of the photometry may become >10% level (see the figure below for the case of 512x512 partial-read mode). We will check this by observing several standard stars under photometric condition in the future.

In the end, we recommend the observers to select standard stars that are as faint as possible (K >13 mag at least). If it is difficult, try defocusing to avoid saturation.

Stray Light from Nearby Bright Stars

If your observing field is close to very bright (JHK < 1 mag) stars, the data may suffer from a significant contamination by that stars, like the examples below. It is caused by the interferred stellar lights that come through the side of the secondary mirror. As the secondary mirror does not have a blocking buffle to suppress it (or, we cannot put it as it emits strong thermal emission), it is currently very difficult to eliminate it. The light tends to contaminate the field when the bright star is located between 2.5 to 0.5 degrees from the target. As the pattern is localized in the focal plane, it is not always happen even if the star is located in that range.

We strongly recommend to check whether there are such bright stars or not before the observation. If there is such a star within 2.5-0.5 degree from your target, please prepare the backup targets beforehand.

The Website below (VizieR) may be useful for the check.

Affection of Moon for Imaging Mode

The affection of th Moon to the background level in near infrared is generally smaller than in optical wavelength. But the sky tend to be brighter than dark night a bit, especially in bluer band. The affection is large when there is a cirrus in the sky.

On bright night we checked how the background level go up with the distance from the Moon. The figure below is the result of our experiment. In the figure, plus marks at >30deg is the reference sky magnitude that are measured far away from the moon. The rise of sky level is almost negligible at >15 degrees, while at 10 degree a clear affection is observed in all bands measured. We also saw some pattern by the scattered light in background at <10 degree.

As the autoguider (AG) is operated in R-band wavelength, the affection of the moon is serious. The AG software will fail the automatic sky estimation roughly at <15 degrees (in some cases, the affection occurs even at 25 degrees away). If the observation uses the autoguider, <30 degree from the moon may be with high risk of AG failure. Of course the affection of the scattered moonlight will be much stronger if the condition of the sky is not clear. Especially the halo from the moon (at ~22deg) will affect the AG background level much.

In conclusion, the affection of the moon on the MOIRCS data may be negligible if your targets are away from the moon by >20 degrees (on clear condition). If you plan to use the AG (for Y, NB, and the spectroscopic observation), the separation must be at least >30 degrees. You should check the separation between the moon and your targets when submitting the proposals, and should explisitly state the dates you want to avoid the assignment on the "Scheduling Requirement" section of the proposal.

Notes on the Autoguider Users

You may require the use of the Autoguider (AG) for more accurate tracking if the single exposure is longer than 3 minutes. The users of Narroubands or Y-band filters will have to take it into account. We strongly recommend such users to check the availability of the suitable stars for Autoguider before you finalize the observing coordinates. You can find the more detailed description about the Autoguider Issue below.

Bad Data During Oct 30th to Nov 2nd 2007

Some data taken during Oct 30th to Nov 2nd may have suffered from the affection of trouble in channel-1 filter turret. Please do not use the channel-1 data during the period.

Flat Fielding

The dome flat in J band shows a tilt along x direction with a level of ~6%. In H-band a slimilar level of tilt is also suspected. These tilt pattern should be removed from the raw dome data using the sky flat. On the other hand, the sky flat generally contain some level (< a few %) of affection by the fringe pattern caused by filter substrate.

Unfortunately, the dome flat in Ks-band is usually very difficult to take because the thermal-background light (without dome lamp on) from the surface of the dome screen is already too bright and variable. Currently we have never succeeded to take high-quality dome flats in Ks band.

The sky under the narrowband imaging observation is usually not "flat". This is because the central wavelength is the function of the angle of incidence to the filter, or approximately, the distance from the pointing center. The use of dome flats for NB data is recommended, especially the observation is aimed the high-accuracy photometry.

If the sky during an exposure varies rapidly due to cloulds etc, you will see a tilt pattern at each detector quadrant on the image. It is the artificial pattern caused by the CDS readout method. If such pattern appears frequently and strongly on your data, the accuracy of the sky flat by these dataset will become poor, though it depends on the level of tilt patterns on your dataset (low-level pattern is usually seen). We recommend to use the domeflat as well as the sky flat if the sky changes significantly during your observation.

Data Reduction Packages