The following sections give examples of common use cases for the GeRT. More examples can be found in the Spitzer Data Analysis Cookbook.
2.1 Second-Pass Filtering
The online filtering of MIPS-Ge products can be affected by bright sources and can yield negative "side-lobes" around bright emission regions (e.g., see examples in MIPS Data Handbook). If you see negative side-lobes around the source of interest in the *mfilt.fits archive mosaic product, you should use the unfiltered *msaic.fits image or re-filter your data offline using one of the procedures below. Filtering is particularly useful at 70 micron to remove streaking. Users need to first identify the location of bright emission regions to be masked from an initial first pass data reduction (and/or the online data products). In the second pass filtering, bright emission regions are masked to avoid biasing the filtering corrections by sources.
2.1.1 Bright Point Source Filtering at 70 micron
Performs column filtering followed by a high-pass time median filter of the original BCDs, ignoring pixels near the location of a bright point source(s).
Make input lists of BCDs, uncertainties, and bmasks., e.g.:
unix% ls *_bcd.fits > bcd70.lis
unix% ls *_bunc.fits > unc70.lis
unix% ls *_bmask.fits > mask70.lis
Create a file, source.tbl, to identify the positions of bright emission to be masked. The input source table needs to be in IPAC table format (use the output of APEX or make by hand or via other software). The values within the IPAC table must be located between the vertical ``|'' to be read properly. e.g., source.tbl:
|srcid |RA |Dec |
|i |d |d |
1 XXX.xxxxx XX.xxxxx
RA,Dec is source position in decimal degrees.
Copy source.tbl into your working directory.
Edit mask_pointsource.nl to point to the proper source list name, e.g.,
PointSourceList = 'source.tbl',
(NOTE: the comma at the end of namelist lines is required)
Run cleanup70.tcsh. Performs column filtering followed by a high-pass time median filter of the original BCDs, ignoring pixels near the location of a bright point source(s).
where bcd70.lis is the list of original BCDs, mask70.lis is the bmask list, and unc70.lis is the list of uncertainties. The updated filtered BCDs are saved in the output cc$OUTID directory. The filtering is controlled by caler_cleanup70.nl. The median_count parameter is the high-pass filtering width in DCEs (16 for online processing). “colfilt = 1” means apply column filter and “colfiltfirst = 1” means apply column filter before high-pass filter. Users may want to test the quality of the filtering corrections on their own data by switching the order of the filters (colfiltfirst = 1/0) and modifying the high-pass filter width (median_count = 10 -- 50 DCEs).
Coadd the updated BCDs using MOPEX or the software of your choice.
2.1.2 Extended Source Filtering at 70 micron (with no column filtering)
Performs high-pass time median filter of the original BCDs. Does not do column filtering since the sizes of extended sources represent a significant fraction of the array.
Make input lists of BCDs, uncertainties, and bmasks (see §2.1.1).
Make source.tbl to identify the positions of bright emission to be masked. The input source table needs to be in IPAC table format (see §2.1.1).
Copy source.tbl into your working directory.
Edit mask_pointsource_extended.nl to point to the proper source list name and set the size of the mask regions to cover all of the extended area to be masked. e.g., PointSourceList = 'source.tbl',
#Mask_Radius takes precedence over MaskBox_X(Y)Size,
# Mask_Radius = 3,
MaskBox_Xsize = 20,
MaskBox_Ysize = 20,
Sizes in original pixels, 5x5 box works for point sources, but a larger MaskBox_*size should be used to cover extended sources as needed.
Run cleanup70_extended.tcsh. This does a time median filter to remove streaking. Depending on the size-scale of your region of interest, you should modify the median filter window size in caler_cleanup70_extended.nl. e.g.,
median_count = 30,
The online processing median_count = 16 (at 70 um). For large sources/extended regions you may want to use a larger window size.
where bcd70.lis is the list of original BCDs, mask70.lis is the bmask list, and unc70.lis is the list of uncertainties. The updated filtered BCDs are saved in the output cc$OUTID directory.
Coadd the updated BCDs using MOPEX or software of your choice.
2.2 Bright Source Filtering at 160 micron
Performs high-pass time median filter of the original BCDs, ignoring source pixels. Filtering generally works best at 160um for scan data, given that photometry (small-field) does not have enough off-source data to derive good corrections.
Make input lists of BCDs, uncertainties, and bmasks (see §2.1.1).
Make source.tbl to identify the positions of bright emission to be masked (see §2.1.1).
Copy source.tbl into your working directory.
Edit mask_pointsource160.nl to point to the proper source list name and set the size of the mask regions appropriately, e.g.:
PointSourceList = 'source.tbl',
#Mask_Radius takes precedence over MaskBox_X(Y)Size,
# Mask_Radius = 3,
MaskBox_Xsize = 5,
MaskBox_Ysize = 5,
(sizes in original pixels)
Run cleanup160.tcsh. This does a time median filter to remove any residual pixel response variations as a function time. Depending on the size-scale of your region of interest, you should modify the median filter window size in caler_cleanup160.nl. e.g.,
median_count = 20,
A small median_count (=16) is ok for point sources, while larger windows (30-40) can be used for large sources/extended regions. The default online median_count = 20 for 160 micron. How the data is taken (e.g., fast scan vs photometry) may affect the proper choice of the 160 micron window filter size (median_count).
where bcd160.lis is the list of original BCDs, mask160.lis is the bmask list and unc160.lis is the list of uncertainties. The updated "two-pass" cleaned-up BCDs are saved in the output c2c$OUTID directory.
Coadd the updated BCDs with MOPEX or the software of your choice.
2.3 Correcting Bad Stim Solutions for Bright Regions
In cases of bright regions, the stim-minus-background solutions may be corrupted, yielding data jumps near the sources (see the MIPS Instrument Handbook for examples). If only one (or a few stims frames) is affected, you can simply delete the "bad" stim frames from the input list and re-run the GeRT to interpolate over this region. You may want to zero-out the affected pixels in the raw stim and/or background ramp (to save most of the stim frame). Online we currently do a simple cubic spline, which smoothly connects all stim measurements. Users may want to play around with the different methods of stim interpolation to see what works best for their data (see the INTERP module). Expert users could also clean-up artifacts in the interpstim.fits file offline (remove large jumps and/or smooth the stim solutions) and then re-run the SLOPECAL module.
2.4 Recovering Saturated Data
2.4.1 Saturated Stim (Stim+Sky) Data
Several projects push the saturation limits of MIPS-Ge. In some cases, you may find that your non-stim data are not saturated, but you get NaN's in the BCDs since the stim+sky data did not have enough reads to derive a stim solution. One can get more information out of such data sets with offline reprocessing. The online system ignores the first few reads [4] in the stim ramp due to the slight stim warm-up nonlinearity. Offline, the user could change the parameter StimLo = 4, to StimLo = 1,2,3 to save more stim reads at the beginning of the ramp (within the &CVTIN namelist block in the MIPS*_SLOPE_0.nl file). However, this effectively changes the calibration of the data (e.g., stim slopes are lower so BCD values are higher). To constrain the effects on calibration, compare the stim response solutions (interpstim.fits) as a function of the number of the initial stim reads ignored, and make the appropriate calibration correction to your data (few--10%).
2.4.2 Saturated Source Data
In cases where the source of interest is strongly saturated, you may try the following. Change DataLo = 0, (within &CVTIN namelist block in the MIPS*_SLOPE_0.nl file) and change Min_Num_Samples = 2, within the &SLOPE namelist block. For online processing, we reject the first read in the data ramp (DataLo = 1) and require a ramp segment of 4 good samples for calculating the slope (Min_Num_Samples = 4). It is possible to calculate a slope with only 2 reads; however, if you change Min_Num_Samples < 4, the RADHIT module will NOT search for radhits within the segment (RADHIT requires 4 reads for checking the end points and for checking for positive vs negative jumps). This technique does not work well for extended regions with low coverage (will show too many radhits). The technique has proven successfully for recovering the core of a saturation source with good redundancy. For 160 micron fast-reset data users could also try saving the reads after the reset in the middle of the DCE, i.e., change number_pixels_to_ignor within the RESETIN namelist block. Users changing namelist values should check the effects on calibration. In the case of recovering a saturated core, scale the unsaturated wings to match the default processing.
