The reduction of spectra collected using the modern CCD and IPCS detectors often require complex procedures in order to ensure that the best possible results are obtained. In many cases, the dataset which the observer brings back from the Observatory is incomplete in the sense that some of the data required for an ideal reduction is missing. It is therefore vital for the observer to ensure that equal effort is paid to obtaining a comprehensive set of calibration frames, to that which was expended on the ‘Object’ frames. Different types of object will produce different requirements in terms of calibration and the following are merely guidelines.
The detectors are frequently aligned on the Spectrograph such that the central order falls (mainly) parallel to the detector rows or columns. The orders themselves however, are often not parallel to each other. The algorithms used for tracing orders assume the orders run approximately in the horizontal direction. It is therefore necessary to rotate any column-parallel frames before reduction (through 90 degrees). echomop is very flexible as to the exact order orientation and highly sloped and severely distorted orders can usually be extracted without difficulty provided that a trace frame of sufficient quality is provided. This will generally be in the form of a bright star exposure, or a template trace frame such as an exposure of a continuum lamp (with a narrow dekker preferably). If the target object under study is sufficiently bright, then the object frame itself can of course be used as the trace template frame. echomop is capable of tracing very poor object frames, but with the result that the modelled order positions will be less well determined, impacting on all subsequent phases of a reduction. You are encouraged to provide the best order-tracing template possible. The Figure below shows an example of the order trace paths plot.
Images not conforming to this format will need to be rotated through 90 degrees.
A bias current is routinely applied to CCD detectors to ensure that, as near as possible they are operating in a linear manner. This current has the effect that a non-zero count is recorded in all pixels. The observer should ensure that a number of ‘bias frames’ are obtained to adequately estimate the effect. The average of the bias frames should be taken, and the result subtracted from all other frames before using them in a reduction. Failure to account for the bias may have a detrimental effect on the signal-to-noise-ratio (S/N) in the final spectrum.
echomop does not provide any special facilities for the subtraction of the bias current from CCD frames. However, it does require that such subtraction be performed before processing takes place. The bias subtraction may often be done automatically at the observatory by the data recording software but it is up to you to check if this is the case. KAPPA or fiGARO can be used to subtract the bias if necessary.
For example:
will subtract bias /nnn/ from data /inframe/ and put the resulting data in /outframe/.
For cases where the bias subtraction represents a significant source of error it may be desirable to determine the error due to bias subtraction on a pixel-by-pixel basis using many frames. The resulting error estimates must then be copied into the object data frames error arrays. echomop will incorporate these errors when calculating the variances on the extracted spectra.
CCD frames typically include a number of rows/columns not exposed to the light. The pixels in these rows often contain high counts and can affect some of the automatic reduction procedures. It is prudent to clip off these rows/columns before reduction proceeds. They will be the same for all frames and can thus be removed using a simple looping command procedure.
Dark frames are essential for some applications. They are advisable when using on-chip binning. For CCDs, the dark current may be high if it has been exposed to too much light, or if power has been interrupted to a cold chip. A number of long exposure (30 to 60 minutes) dark frames should be combined using a median filter, and the result scaled according to the object frame exposure time. This frame may then be subtracted from the object frame before reduction begins. If you wish to ensure that the error properties are accounted for then the dark frame should be copied into the object frame variance array.
Again for CCDs only, a further advantage of collecting long exposure dark frames is the option of deriving an independent estimate of the rate of cosmic-ray events as a function of their intensity. This can provide a useful comparison with long exposure object frames and a check on how effective cosmic-ray removal has been.
A bright quartz lamp should be used with a dekker size a few steps larger than that used for the object observations, but order overlapping should of course be avoided. It is important to ensure that all orders are as bright as possible. It may be necessary to use different flat-field exposures to obtain good frames when observing in a spectral region where the quartz lamp intensity or detector efficiency changes rapidly. Each flat-field frame should be constructed by calculating the median of a large number of runs where possible. If you wish to include the errors on the flat-field then they should be placed in the flat-field variance array. If no variance estimates are provided then root-N statistics will be assumed. echomop can utilise different flat-field frames for different orders if required.
Many of the algorithms used assume that photon counting statistics apply. However, it is unnecessary to correct raw frames since the conversion factor (ADUs/photon) is prompted for when required.
In general arc frames should be obtained to bracket each object exposure in time, allowing us to check that no shift has occurred during the exposure. UCLES is extremely stable over relatively long periods and in practice shifts are not detectable. The advantage of using a single frame (possibly generated by co-adding the before and after arc frames) is that this allows the resolution of the final spectrum to be easily investigated. This is because the arc spectrum is always extracted in an identical manner to that of the object; using the first (and in this case only) supplied arc frame. As with the flat-field frame it may also be advantageous to take multiple arc frames, each exposed correctly to ensure high S/N in particular orders. echomop can use a different arc frame for each order if required.
All raw data should be pre-processed in the following ways before using echomop:
If a cosmic-ray-contaminated frame is to be used for tracing the paths of the orders across the frame then:
ech_decos1 task may be used stand alone,
or it may be invoked from the main task by specifying the parameters TUNE_CRTRC=YES
and TUNE_FCHECK=YES at the start of a reduction.
provides this facility and should be the preferred method when multiple frames are available. The procedure calculates a median image and then rejects pixels which deviate by more than n sigma from the median value (note that this means that all the frames must be of equal exposure time). Bad pixels are flagged by placing values in the quality array of the frame concerned.