|
|||
| Joint Astronomy Centre |
|
|
|
| Show document only |
The Infrared SkyFor a number of reasons, observing in the IR is more complicated than in the optical domain. Mainly, this is due to a much higher and variable background, and by the presence of variable absorption and emission features. Auroral, OH, and O2 lines produce strong variations shortward of 2.3 microns, while thermal emission is the main source of background emission at longer wavelengths. As a result, it is all but uncommon to be looking at sources that are thousands of times fainter than the background.For all these reasons, it is very important to devise an observation strategy that accounts for these peculiarities and that provides a precise way of estimating and subtracting the background, which in the worst case can show significant variations on a time-scale of 1 to 3 minutes. There are basically two ways of obtaining an accurate sky subtraction: (1) use a direct external sky observation, and (2) use the images of the target itself. It is nearly impossible to provide definite guidelines on the strategy to follow. This is a complex optimization problem that takes into account (1) the size of the target, (2) the overheads, (3) the required photometric accuracy. We suggest you contact your support scientist well in advance of your observing run to discuss these details. Small targets, or uncrowded fields.If the targets are very small compared to the size of the detector, or if the crowding is such that most of the detector is actually occupied by blank sky, then it is possible to use the target images themselves to estimate the sky. This is achieved by combining the target images using rejection algorithms, so that the result is a blank sky image. Ideally, the offsets should be at least 2/3 times the size of the largest object, which might conflict with the requirement of keeping the offsets below 10 arcsec to exploit the fast offsetting mode offered by the secondary mirror. Also, it is important to note that the best results are achieved when a large number of positions are obtained. Ideally, one would use a totally random sequence of offsets, but a 25-point jitter should still do the job.Careful planning of the offset sequence is needed in preparing this type of observations, especially if a frequent sampling of the sky is required. Indeed, the sky will only be monitored at end of a complicated sequence made of: (1) the number of microstepping positions (N_micro), (2) the number of coadds (N_coadd), (3) the number of offset positions (N_offset), (4) the exposure time (T_exp). As a result, we get a sky image every: T_exp x N_offset x N_coadd x N_micro. With 5 seconds, 2 coadds, 25-point jitter, and a 2x2 microstep, we get a sky measurement every 1000 seconds, which might or might not be enough for your science. Large targets, or crowded fieldsIn this case, the target images are not suitable to construct a sky image, and separate sky observations are required. Ideally, one would want to integrate on the sky using the same observations strategy as the object. Sky observations that are shorter than the target observation will result in the noise being dominated by the noise in the sky, rather than in the target. It might be obvious, but overhead are heavily reduced if the sky observations are unguided.Sky observation can be obtained by either using a dedicated MSB, or by using a survey container. Very large targets, or tiled observationsIn some very specific cases, this is the most efficient observation strategy. If WFCAM is used to map an extended region in the sky, with several subsequent pointings, and each pointing is short enough, then the observations obtained at contiguous positions can be used as "sky" for each other. Again, this is entirely specific to your program. A program aimed at mapping the core of M31, for example, would not be able to take advantage of this strategy. |