MAGMA Data Release 3:
Inclusion of 2012-2013 Data

Data use policy (30 AUG 2017)

Release Notes (2 MAR 2015)

Additional observations from 2012 and 2013 have been incorporated that include Mopra projects M624 (PI: Roman-Duval), M625 (PI: Bernard), M633 (PI: Wong), and M1022 (PI: Hughes). This expands the coverage of the LMC by approximately 20% compared to DR1.

Data cubes have been generated in the heliocentric velocity frame. In Livedata, a fix for an error in the application of the calibration factors and improved baseline estimation have been implemented, although a residual baseline still had to be subtracted. The re-baselining is now done iteratively by first generating a signal mask and then fitting a linear baseline to each spectrum, excluding the channels where signal was found. The signal mask is based on a smoothed cube so should include some "padding" around the actual emission.

A new cloud catalogue has not yet been generated. See Data Release 1 for the the cloud catalogue presented in the ApJS paper.

Basic Data Products

These are gzipped FITS images constructed using the SIN projection. Note that regions which were not observed have been assigned a blank value. The brightness temperature scale corresponds to the main beam brightness temperature assuming a 45" FWHM Gaussian beam. The pixel scale is 15" x 15" x 0.526 km/s.

Moment-type Images

The data and noise cubes have been processed using an IDL script that implements a dilated masking technique.

Images produced include the first 3 velocity moments and their estimated errors, the peak intensity within the mask, and a mask cube with values of 1 for good pixels and 0 for excluded pixels.

For the 'sgm' series the mask is derived by first smoothing the cube spatially, to reach a resolution of 67.5", and spectrally, by a convolving with a Gaussian with a FWHM of 3 channels. Then the mask was generated by expanding from the 5-sigma contour of the smoothed cube out to the 3 sigma contour. (Both initial and expanded mask are required to be at least 2 channels wide at every spatial pixel.) This mask is then applied to the original data cube before taking moments along the velocity axis.

For the 'gm' series the mask is derived without smoothing, by expanding from the 4-sigma contour of the original cube out to the 2 sigma contour. This yields a smaller mask, as can be seen from the mask cube below.