Dear All,

I am seeing startlingly different behaviour when between makecube and 
wcsmosaic (see the two attached images, which were created from the same 
data using these two routines) when I use them to create cubes from four 
files that have different spectral extents, although they are repeated 
observations of the same source. I have changed the way that I clean up 
the raw datafiles, and so one pair of files has the original spectral 
extent, while the other pair have had about 30% clipped from the ends of 
the spectra. The actual makecube command that I use is
"makecube 
in=/mraosw/data4/harp/mjlsg11/raw_cleaned/a20081111_00036_01_0001 
out=temp autogrid=false detectors=! genvar=tsys outcat=! inweight=true 
pixsize=\[7.2,7.2\] sparse=false trim=true weights=false spread=nearest 
params=\[1.0,1.33\] system=tracking usedetpos=false"

If I use makecube, then mfittrend with the following commands (from 
hislist), I get pictureA:

1: 2009 May 11 15:32:21.519 - MAKECUBE        (SMURF V0.4.1)

   Parameters: ALIGNSYS=FALSE AUTOGRID=FALSE BADMASK='OR'
      CROTA=-55.000000000042 DETECTORS=!
      FBL=[1.5135963227979,0.00037905821943821]
      FBR=[1.508831094304,0.0071842303514043]
      FLBND=[1.508806786544,0.00035475123779951,-424.20370675136]
      FUBND=[1.5193107862253,0.01119288100032,442.75726064658]
      FTL=[1.5192864791151,0.0043632445261451]
      FTR=[1.5145213640061,0.011168574035431] GENVAR='tsys' [log in to unmask]
      INWEIGHT=TRUE LBND=[-118,-99] LBOUND=[-118,-99,-512] NTILE=1
      NPOLBIN=1 OUT=@temp OUTCAT=! PARAMS=[1,1.33] PIXSIZE=[7.2,7.2]
      MSG_FILTER='NORM' REF=! REFLAT='0:19:53' REFLON='5:47:00.0'
      SPARSE=FALSE SPECBOUNDS='-423.7804 442.3339' TILEDIMS=! TRIM=TRUE
      WEIGHTS=FALSE SPREAD='nearest' SYSTEM='tracking' UBND=[120,100]
      UBOUND=[120,100,511] USEDETPOS=FALSE
   Software: /soft/star/bin/smurf/makecube

2: 2009 May 11 15:32:35.914 - MFITTREND       (KAPPA 1.9-11)

   Parameters: AUTO=TRUE AXIS=@3 CLIP=[2,2,2.5,3] FITTYPE='Polynomial'
      IN=@temp MASK=! METHOD='Single' MODIFYIN=FALSE NUMBIN=32 ORDER=1
      OUT=@ngc2071_12co_b RMSCLIP=! SUBTRACT=TRUE TITLE=! VARIANCE=TRUE
   Software: /soft/star/bin/kappa/mfittrend

If I use makecube and mfittrend on the individual files, and then 
wcsmosaic (hislist below) to put the final cubes together, I get pictureB.

1: 2009 May 11 14:44:22.095 - MAKECUBE        (SMURF V0.4.1)

   Parameters: ALIGNSYS=FALSE AUTOGRID=FALSE BADMASK='OR'
      CROTA=-55.000000000042 DETECTORS=!
      FBL=[1.5137192681308,0.00050775633035526]
      FBR=[1.5089740600772,0.0072843421205392]
      FLBND=[1.5089497522708,0.00048344931397717,-424.20370675136]
      FUBND=[1.5190906087172,0.011052734630646,442.75726064658]
      FTL=[1.5190663015624,0.0042516937035154]
      FTR=[1.5143211976496,0.011028427618018] GENVAR='tsys'
      IN=@/mraosw/data4/harp/mjlsg11/raw_cleaned/a20081111_00036_01_000?
      INWEIGHT=TRUE LBND=[-117,-94] LBOUND=[-117,-94,-512] NTILE=1
      NPOLBIN=1 OUT=@temp OUTCAT=! PARAMS=[1,1.33] PIXSIZE=[7.2,7.2]
      MSG_FILTER='NORM' REF=! REFLAT='0:19:53' REFLON='5:47:00.0'
      SPARSE=FALSE SPECBOUNDS='-423.7804 442.3339' TILEDIMS=! TRIM=TRUE
      WEIGHTS=FALSE SPREAD='nearest' SYSTEM='tracking' UBND=[120,93]
      UBOUND=[120,93,511] USEDETPOS=FALSE
   Software: /star/bin/smurf/makecube

2: 2009 May 11 14:44:36.318 - MFITTREND       (KAPPA 1.9-11)

   Parameters: AUTO=TRUE AXIS=@3 CLIP=[2,2,2.5,3] FITTYPE='Polynomial'
      IN=@temp MASK=! METHOD='Single' MODIFYIN=FALSE NUMBIN=32 ORDER=1
      OUT=@mc_20081111_00036_01 RMSCLIP=! SUBTRACT=TRUE TITLE=!
      VARIANCE=TRUE
   Software: /star/bin/kappa/mfittrend

3: 2009 May 11 14:48:29.082 - WCSMOSAIC       (KAPPA 1.9-11)

   Parameters: ACC=0.05
      FLBND=[1.5089097074157,0.00048344931397717,-424.20370675136]
      FUBND=[1.5195481106442,0.011430254274311,442.75726064658]
      GENVAR=FALSE ILEVEL=2 [log in to unmask] LBND=!
      LBOUND=[-117,-99,-512] MAXPIX=1000 METHOD='nearest'
      OUT=@test_ngc2071_12co REF=! UBND=! UBOUND=[122,104,511]
      VARIANCE=TRUE WLIM=1E-10
   Software: /star/bin/kappa/wcsmosaic


Both of these are run on the same 4 files:

/mraosw/data4/harp/mjlsg11/raw_cleaned/a20081111_00036_01_0001
/mraosw/data4/harp/mjlsg11/raw_cleaned/a20081111_00037_01_0001
/mraosw/data4/harp/mjlsg11/raw_cleaned/a20090131_00028_01_0001
/mraosw/data4/harp/mjlsg11/raw_cleaned/a20090131_00029_01_0001

the files from 20081111 have a spectral extent, from ndftrace:
      No. of dimensions:  3
      Dimension size(s):  1024 x 16 x 5488
      Pixel bounds     :  1:1024, 1:16, 1:5488
      Total pixels     :  89915392
........
        Frame title         : "3-d compound coordinate system"
        Domain              : DSBSPECTRUM-SPACEINDEX-TIME
        First pixel centre  : -423.7804, 1, 2008-11-11 12:45:27

while the files from 20090131 have:
      No. of dimensions:  3
      Dimension size(s):  716 x 12 x 5805
      Pixel bounds     :  155:870, 1:12, 1:5805
      Total pixels     :  49876560
........
        Frame title         : "3-d compound coordinate system"
        Domain              : DSBSPECTRUM-SPACEINDEX-TIME
        First pixel centre  : -293.3614, 1, 2009-01-31 08:26:41


Any clues about this? What this looks like visually is the detector 
gains problem, but it clearly is not. Is it possible that part of the 
'gains' issue was that the spectral dimension was not being identically 
written for all files, and makecube needs files that have identical 
spectral extents?

Anyway, I hope for enlightenment, and a method of reducing files of 
varying spectral extents.


Cheers,

Jane

-- 

Dr Jane V. Buckle 	
Astrophysics Group, Cavendish Lab, J J Thomson Avenue, Cambridge CB3 0HE
Tel: +44 (0)1223 337298