The observations

This is using the same MERLIN data set that was used in the previous tutorial (tutorial 1).

The data

Initial Calibration

In the previous tutorial this data set was used to demonstrate the MERLIN procedure. As the initial start point of this tutorial use the MERLIN procedure following the inputs used before to load and create a calibrated split file.

Initially set the soft link to the data and the MERLIN runfile, follow the steps you used in yesterdays tutorial.

e.g. EXPORT WORK='/home/user/BASIC'
EXPORT RUNFIL='/home/user/BASIC'

And copy the MERLIN.ehex file to the /home/user/BASIC directory remembering to change the .ehex to the appropriate extended hexidecimal number.

Or start AIPS using yesterdays aips number (if you are on the same machine).

The purpose of this tutorial will be to improve the initial results that were created by running this procedure. This tutorial will take you through phase self-calibration of the source, re-weighting these data to account for the different telescopes that make up MERLIN and then different image weighting schemes.

Step 0

As explained in the previous tutorial set up and run the MERLIN procedure, with no self-calibration (for more details, see the previous tutorial here).

The MERLIN procedure will produce a calibrated data set which will be used in the rest of the tutorial.

Improving the initial image

After you have successfully run the MERLIN procedure, with stipulating no self-calibration, will have created a calibrated split file. Using this file as a start point we will apply several rounds of phase self-calibration.

Initially running SCMAP

In order to apply the self-calibration we will initially use the AIPS task 'SCMAP'. SCMAP performs a series of imaging and then self-calibration cycles. In essence SCMAP is equivalent to running imagr to create an image and then using CALIB to use the clean components from this image to self-calibrate.

Suggested SCMAP inputs
Task 'SCMAP'
getn i $ split file.
solint 1
soltype 'L1'
niter 2500
Solmode 'p'
cell 0.04
imsize 1024
uvwtfn 'NA'
Nmaps 5
refant 8
Aparm 3 0 0 0 0 0 1 0 1 0
dotv 1 $ activate the tv
go scmap

This will apply 5 rounds of phase only self-calibration. At this initial stage we will not interactively box these data during the cleaning.

Inspect the images created and the sn table solutions that scmap has created. These SN tables are attached to the file original file that scmap was run on. These sn tables should be plotted with the task 'SNPLT'.

Further rounds of self-calibration

Now we will re-run scmap with a larger number of clean iterations (10000) plus during this SCMAP run we will guide the cleaning and self-calibration process via manually boxing the image during cleaning process.

Suggested SCMAP inputs
Task 'SCMAP'
getn i $ output scmap from the previous round.
solint 1
soltype 'L1'
niter 10000
Solmode 'p'
cell 0.04
imsize 1024
uvwtfn 'NA'
Nmaps 5
refant 8
Aparm 3 0 0 0 0 0 1 0 1 0
dotv 1 $ activate the tv

This will perform 5 cycles of phase self-cal. During the imaging process SCMAP will allow you to box the areas that you wish to clean (i.e. the areas you believe to be REAL).

Once SCMAP has finished inspect the final images and compare them to the image that the MERLIN procedure initially produced. At this stage you should begin to see a reasonable improvement in the image quality.

SCMAP will have produced a UV file (.SCMAP.2) plus is till attach the new SN tables to the original UV split file. Use the Task 'SNPLT' to inspect each of the SN tables that SCMAP has produced. This should show the phases to be converging more at each iteration.

Re-inspect the UV data and clip any further bad data points

Now also uvplt the amplitude versus uv distance. The data are of reasonable good quality now with no obvious errors. However if we now subtract the clean components, derived from our cleaned image, and re-plot that data, it will allow us to more clearly see any lower level data errors that remain in the data.

For example, note that in the latest image there is some spurious extended emission to the North-East of the brightest extended hotspot. So we still probably have some bad data.

Using the aips task UVSUB to remove the clean components.
Use UVSUB to get2n latest image

Settings for UVSUB
Task 'uvsub'
get2n [the latest cleaned image]
ncomp 10000 0
cmethod 'dft'
factor 0
go uvsub

This will create a uvsub file where the clean components from our best image have been removed.

Now UVPLT this subtracted data.
tget uvplt;getn 'new uvsub file'.
Looking at the plot we are clearly missing some extended flux but also there aere some bad data points with high fluxes. These can be clipped.

By looking at the uvplt out puts judge at what level to clip the data.

Settings for CLIP
task 'clip'
getn [the file we created with uvsub]
aparm 0.125 0 0 0 1 0 $ Flag data points in excess of 125mJy
go clip
This will clip everything above 125mJy.

Now we have re-run the Task 'uvsub' to add back the real signal we initially subtracted out.

Some Inputs
Tget uvsub
getn clipped data
factor -1
go $ add back the structure that we previously uv-subtracted.

Re-image the data

Some Inputs
tget imagr
getn uvsub data set (This is the file that you have just added back the clean components)
cell 0.04
niter 10000
minpatch 255
uvweight 'NA'
go imagr (don't window the image this time).

Re-run imagr using robust 0.
Uvweight '' ;robust 0

This will have hopefully now meant that you have produced two good images, one of which will be naturally weighted and one of which will have an intermediate data weighting.

Re-weight the data

MERLIN consists of several types of telescopes. This experiment used 8 telescopes (including the now defunct Wardle). Of these telescopes we some are notably more sensitive, in particular the 32m telescope at Cambridge and the 76m Lovell. Consequently we can alter the weights of the data from these telescopes according to their relative sensitivities and improve the noise levels in our image.

Using wtmod we will now re-weight the data according to the sensitivity of the individual telescopes.

Task wtmod
antwt 1 4 1 1 1 1 60 1
go wtmod

Now tget imagr and re-run with robust 0 for the newly re-weighted image. Again using no boxing. Compare the noise levels of this map and the previous robust 0 map.

This re-weighting of the data should produce a modest improvement the noise levels on you image.

Further imaging

These re-weighted self-calibrated data can now imaged using image weighting schemes, as well as different deconvolution algorithms, such as STDCLEAN or MEM - these were briefly described in Tom's lecture this morning.