**NEW RESULTS** (from Feb/08) There are 4 final cubes. FINAL_CUBE.RDVFITS is the cube at the original resolution. RDVFITS means the cube is in Right Ascension - Declination - Velocity order and FITS is (obviously) the format. FINAL_CUBE_VSMOTH.RDVFITS is the cube smoothed in velocity. FINAL_CUBE_RDSMTH.RDVFITS is the cube smoothed spatially (i.e. in RA-DEC). And FINAL_CUBE_RDVSMT.RDVFITS is the cube smoothed both spatially and in velocity. Smoothing details are in the paper. Files designated .VRDFITS are the same except with velocity as first axis. Note that these are the files used for the preliminary paper.
All final cubes can be found HERE.
All signal-to-noise cubes can be also be found HERE.
Zeroth and 1st moment maps can be also be found HERE.
Beam maps (including downloadable fits) at all azimuths HERE
Ignore the rest of this page as it included preliminary results only.
OLDER RESULTS (from Sept/07) Cleaned cubes/maps HERE
Preliminary HI Map from Nov 28 - Dec 6, 2004 Observing Run: Preliminary HI Map from Nov 28 - Dec 6, 2004 Observing Run.
Preliminary HI Velocity Field Map from Nov 28 - Dec 6, 2004 Observing Run: Preliminary HI Velocity Map from Nov 28 - Dec 6, 2004 Observing Run.
Preliminary Continuum Map from Nov 28 - Dec 6, 2004 Observing Run: Preliminary Continuum Map from Nov 28 - Dec 6, 2004 Observing Run.
At Arecibo in area /share/ngc2903/data, the directories are: 04.11.28/ 04.12.06/ 04.12.19/ 05.02.10/ 05.03.01/ 05.03.07/ 05.03.26/ 04.12.01/ 04.12.14/ 04.12.20/ 05.02.11/ 05.03.02/ 05.03.21/ procs/ 04.12.02/ 04.12.15/ 04.12.21/ 05.02.12/ 05.03.03/ 05.03.22/ raw/ 04.12.03/ 04.12.16/ 04.12.22/ 05.02.13/ 05.03.04/ 05.03.23/ work/ 04.12.04/ 04.12.17/ 04.12.23/ 05.02.14/ 05.03.05/ 05.03.24/ 04.12.05/ 04.12.18/ 04.12.26/ 05.02.28/ 05.03.06/ 05.03.25/
Total disk-space usage is 76,015,914 in units of 512 bytes. Therefore total disk space for unprocessed IDL-style files is about 39 Gbytes. Each calibrated scan (starting with j) takes 75.3 Mbytes. Assume that the semi-calibrated scans (starting with d) takes the same. If there are 12 scans per night (more if calibrators but they are also smaller), then each night needs 1.8 Gbyte. Estimate a total of 37 nights so total for calibrated scans is 66.7 Gbyte. Now assume need half again as much for the fits files (say 35 Gbyte). So total is (raw=39 G, calibrated=67 G, fits = 35 G). The total is 141 Gbyte without any working room. A good estimate would be 200 Gbyte for data alone.
NEW BEAMS AS OF MAY 20, 2005
BEAM 0 Azimuth = 197 BEAM 0 Azimuth = 215
The following are beams with contours at the 0.5%, 1% and 50% levels. Left to right is Beam 0 through Beam 6 for Azimuth = 197 degrees.
The following are beams with contours at the 0.5%, 1% and 50% levels. Left to right is Beam 0 through Beam 6 for Azimuth = 215 degrees.
The following are the difference between the beams at the two azimuths, Left to right is Beam 0 through Beam 6. Note that the peak flux of the calibrator is 1.85 Jy/beam, the rms noise on the beam0 map (taking the 197 az) is 1.6 mJy/beam and the rms noise on the difference map (taking beam 0) is 2.8 mJy/beam.
September 22, 2005
Results from azimuth=104. These beams were made by interpolating
the 103/106 beams with weighting, setting the peak to 1, setting
the center to the center of a 'power of 2' map, eliminating negative
values and, for cases in which there was a cutoff, creating a
'smooth' gaussian edge so as not to produce artifacts when cleaning.
Map size = 1024 x 256 pixels. Pixel size = -13.959 X 35.677 arcsec.
This is to match the map scale.
The fits beams can be found here, FINALBMn_AZ104.FITS, n=0..6
For the DATA CUBE for AZ104, (same size, same pixel size, but 1809 channels, go to the same directory as above but look for FINALCUBE_AZ104.FITS. (It is large). The dec spacing has been chosen so that the positional rms in dec is minimized and yet any beams that scan the same place have already been averaged together in this cube. About 20 channels at either end of the frequency band have been discarded since they are noisy and/or bad. The Milky Way Galaxy is still there though.
Finally, for a list of which beam corresponds to which declination row, go to this directory and get the file, BEAMLIST_AZ104.FITS. A value of Zero means 'no data', a value of 1 means 'beam 0', 2 means 'beam 1' ... and 7 means 'beam 6'.
N.B. There are 12 azimuths altogether so there are 11 more beams and cubes to deal with and clean. Once they are all done, then the cleaned cubes can be averaged. Finally, the maps will have to be interpolated onto a regular grid so that the ra and dec spacing are equal.
1) The first plot is the dirty map (one channel at az104).
All negatives greater than 3 sigma are first removed.
2) The middle plot is after cleaning once. i.e. the map is cleaned with each of 7 beams, then a new map is reconstrcted from the cleaned maps so that only the rows corresponding to the correctly cleaned beams are put back, i.e. if row one was actually obtained with beam 6 then only the cleaned row from the map cleaned with beam 6 is put back into the map. The cleaning is to the first negative which for all beams gave about 40 clean components. All negatives greater than 3sigma are again removed.
3) The right plot is after cleaning twice. Same defaults.
NEW RESULTS These new results are as above *except* 1. An error in the positions were first corrected. 2. No negatives are removed before the cleaning -- since the cleaning is to the first negative, this step does not seem to be necessary. 3. Where there was a row with no data, instead of substituting back the beam zero values, a blank was put back and then an interpolation was made over the blanked region. This seemed to improve the results. 4. Lyle's correction for the ratio of the beam sidelobes to total beam, based on milky way values, has been applied.
Only one clean is necessary as the following shows. There are of order 20 or so ccs. The second clean (3rd map from left ) puts in some negatives, possibly because the beam is not perfectly characterized by the continuum mapping. The difference between the original (dirty at left) and first cleaned (2nd from left) maps is significant, as can be revealed by subtracting the two. The subtraction can be seen on the far right. The following is channel 1159 for az104.
Here are the velocity slices for beam 5 (from aips). To the
right of the image is the idl scan from which the data are taken.
When reading into aips, both pols of the idl file are first
averaged and, if the pol was all zeros, then only one of the
pols values was put into the aips files. To the right of
the idl scan number is the value of the mean within a box
centered on the emission (the emission as shown in the 11th
file down). The box size is BLC=1091, 187; TRC=1337,308.
scan c092811+220056.506401382.sav, Mean= 3.1795E-03
c092811+220056.506400708.sav, Mean= 2.7712E-03
Last revised Sept. 14, 2007
Judith A. Irwin