This Chapter is intended to help the observer getting started with
LWS data. It gives some hints and ideas how to proceed with the
processing steps that are not performed in the standard pipeline
processing, but are useful to make a spectrum that can be published
out of the data products. It is by no means intended to give a full
guide of all possible ways to do certain processing steps.
The sections in this chapter only give an outline of the
different steps needed for the individual post processing steps.
Most spectral data analysis
packages will have routines that perform the tasks given below.
All processes described below can be performed using the
ISO Spectral Analysis Package (ISAP). This package has been
written in collaboration by the LWS and SWS consortia and IPAC
in the USA. This analysis package is available for
observers from MPE
(see www page: http://www.mpe.garching.mpg.de/iso/isosdc.html),
and IPAC (http://www.ipac.caltech.edu/iso/iso.html).
Photocurrent values set to zero in the LSPD file indicate that the photocurrent for this ramp was not calculated or was discarded. Photocurrents may not be calculated because all of the readouts in the ramp have been discarded because of glitches, etc. Photocurrents may also be discarded because of glitches in the previous ramps. The LSPD status word can be used to confirm that this is an invalid point and give extra information about what happened during the processing of this point. See section 8.2.5 for a description of the LSPD status word.
Flux values in the LSAN file will also be set to zero when no value could be calculated. This is usually because the value read from the LSPD file was zero, but it can also be caused by a failure to find the spectral responsivity value for this point. The LSAN status word contains a flag which indicates if the point is valid or not. It also contains other flags which give more information about what has happened to this point. See section 8.2.8 for a description of the LSAN status word.
Note that the LSPD file also contains a copy of the photocurrents with no deglitching applied.
As described in section 7.4 the non-linearity of detectors produces saggy shaped sub-spectra. You should look at your LSAN data in ISAP to see if any of it is saggy: this is best done by examining the overlapping regions of detectors LW1-4 with their neighbours to see if their spectral shapes agree. LW3 is the most non-linear detector so it is best to check that one first. If some of the detectors do not agree on spectral shape (and hence are saggy) then those detectors' sub-spectra are affected by non-linearity.
If any of your detector sub-spectra look saggy the strong source
correction needs to be applied to the data (the calibration of these
saggy data is wrong, and cannot be trusted). This is done using the
LIA routine ss_corr which is being prepared (see section 9.1
for info on LIA; or contact Sunil Sidher on email:
s.d.sidher@rl.ac.uk).
The routine takes your LSPD file and
applies the coefficients to the photocurrents of detectors LW1-4. It
also tells you the coefficients and displays the photocurrents from
before and after the correction. There is a button for writing
out the new LSPD file which contains the updated photocurrents (as
well as the original ones for the other detectors).
Ideally the observations of strong sources were carried out using 1/4 second ramps,
however a warning will be displayed (in ss_corr) if they are 1/2
second ramps as the correction should not be applied directly to these
length ramps. You should have your ramps processed as 1/4 second
ramps. This is done by using
only the first quarter second of each ramp to produce a photocurrent.
It is required because for half second ramp there is even more
non-linearity of the detectors. Because the correction is found from
quarter second ramps it does not correct half second ones
precisely. Also half second ramps may hit the maximum voltage level of
the detector circuit and hence the end of the ramp no longer correctly
represents the input signal and therefore should not be used.
If you do have 1/2 second ramps (and require the strong source
correction) you should contact Sunil Sidher
(s.d.sidher@rl.ac.uk) to have the special 1/2 to 1/4 second processing
applied to your data. You will then be given the LSPD file from this
processing and you can then proceed to use ss_corr to apply the
correction, or on request you can have ss_corr run for you and the
corrected data given to you.
Once you have the LSPD file with 1/4 second ramps and the strong
source correction applied to them the second part of the pipeline
needs to be run on that data to get a new LSAN file. This is done
using short_aal, one of the LIA routines.
The new LSAN file can be read into ISAP and you should see that the data are no longer saggy. You should not rely on the absolute fluxes of this LSAN file. The absolute calibration of the non-saggy detectors can be trusted (ref 5.4.2) and hence all of the sub-spectra can be scaled to produce a relative calibration. In doing this it should be seen that now the detectors agree on the shape of the spectrum and also there is no saggyness. Often an `off source' pointing will not require correction even when the `on source' does but scaling both spectra to one of the non-linear detectors will also result in a relative calibration between the two spectra.
This process would involve the following steps:
After rebinning the repeated scans to the same wavelength scale, the scans can be co-added together, using standard co-adding routines.
The fitting of individual subspectra into one spectrum would involve the following processing steps:
As was stated in the LWS observers manual, the LWS instrument does not observe the sky background automatically. It is up to the observer to take a source and an off-source measurement if he wants to be able to subtract the background. The background radiation can be especially bright in the 157 micron line with long exposures (e.g. faint sources like galaxies). For subtraction of the astronomical background the most important step is to rebin the background spectrum (which is in a separate data product if the user asked for a background spectrum) to the same wavelength scale as the source spectrum. After that the background spectrum can be simply subtracted from the source spectrum. The ISAP software package does not have a function to perform this operation.
The LSNR file contains a copy of the data in the LSAN file, but without the absolute responsivity correction and responsivity drift corrections applied. In some cases the drift correction does not work sucessfully. In these cases you may want to apply the absolute responsivity correction to the LSNR data.
In order to do the correction you need to look at the contents of the LGIF file produced by Auto-Analysis. Each record in the LGIF file identifies a time ordered group of data in the LSNR file which will have different correction factors. For each record in the LGIF file, extract the ITK start and end times of the group (fields LGIFITKS and LGIFITKE respectively). Locate all of the data in the LSNR file which have ITK values within this range using the ITK time key in field LSNRITK. For each detector within this time range, divide the flux in the LSNR file by the appropriate absolute responsivity correction factor from the LGIF file. The correction factors are stored as an array of ten numbers in field LGIFABS in the LGIF file. If the LGIF file contains a single record then the absolute responsivity correction factors can be applied to the whole LSNR file.
For the generation of maps from raster scans it is necessary that the data are deconvolved with the LWS antenna pattern. However at the time this manual was written the procedure for the removal of the LWS beam pattern was still under investigation. Therefore at this time no further information is available.