Manual Despiking

The way we despike jiggle maps depends on whether we have done short or long integrations and on whether our source is compact or extended. In the above example we only have 3 integrations and despike will miss most spikes. We can still use scuclip but if we want to go deep, we have to make sure that we don't clip the source as well. We know that IRC$+$10216 is relatively compact with a faint `halo' type emission surrounding the core. It may therefore enough to use scuclip, but go deeper than in our initial despike effort. To be on the safe side, i.e. to make sure that we do not clip the source, we set the central bolometer to bad before we start and reset it back to good afterwards using change_quality.

% change_quality 'i86_lon_sky{b19}'
SURF: run 86 was a MAP observation of IRC+10216
SURF: file has data for 37 bolometers, measured at 192 positions.
 - there are data for 4 exposure(s) in 3 integration(s) in 1 
measurements.
BAD_QUALITY - Set quality to bad? (No will set quality to good) /YES/ 
> 
% scuclip
IN - Name of input file containing demodulated map data 
/@i86_sho_sky/ > 
SURF: run 86 was a MAP observation with JIGGLE sampling of object 
IRC+10216
OUT - Name of output file /'i86_lon_clip'/ > 
NSIGMA - How many sigma to despike bolometers /5/ > 4
SURF: Removed 10 spikes
% change_quality 'i86_lon_clip{b19}'
SURF: run 86 was a MAP observation of IRC+10216
SURF: file has data for 37 bolometers, measured at 192 positions.
 - there are data for 4 exposure(s) in 3 integration(s) in 1 
measurements.
BAD_QUALITY - Set quality to bad? (No will set quality to good) /YES/ 
> no

Compared to our first 5 sigma we now found 10 additional spikes. We could probably have used 3 sigma, but with a small data set it is better to be conservative, For the short array, we have to be really careful when using scuclip, if we work with 64-point jiggle maps. If we are looking at a point source centered in the array, the jiggle steps are so large, that the first ring of bolometers will pick up the source in part of the jiggle pattern. scuclip can in this case interpret the source as being noise, since it only shows up in a just a couple of points, and flag them. The end result can easily be that when we regrid the map, we get an artificially narrow image, because we despiked the ``flanks'' of the source. If we want to run scuclip on the short array, we should not only blank the central bolometer, but also the first bolometer ring (seven bolometers), and then reset the flagged bolometers afterwards and despike them manually, which becomes somewhat tedious.

For short maps on strong or extended objects we therefore mostly end up doing manual despiking. We can either use the interactive shell script dspbol or the FIGARO routine sclean.

% sclean
IMAGE - (IMage) Name of image to be cleaned /o86_lon_sky/ > 
o86_lon_sky
OUTPUT - (OUTput) Name of resulting image /o86_lon_clip/ > 
o86_lon_clip
IDEV - Device for image display /'xwindows'/ >

which will give you an image similar to Fig. . sclean is an interactive application and if you select the xwindow, and type `b' while the cursor is over a selected bolometer you will get the display shown in Fig. . Here the left hand side shows all the bolometers, and the right hand side shows just the bolometer you selected. The two most useful commands are probably `a' - delete the pixel selected, and `y' - delete the indicated area and fix by interpolation along the y-axis. Manual despiking like this is usually only needed for short integration maps of strong sources or when taking out more extended glitches in scan maps, which are not picked up by automatic despiking.

Figure: The display from the FIGARO package sclean. sclean allows you to easily inspect the time series data and despike data.