The command CUBE builds a 3-D image from the current index. It is assumed that the current index defines such an image (i.e. corresponds to a single line, observed with a single resolution, and towards a single source). The first axis of the cube corresponds to the current X unit, the second to RA (or L) and the third to DEC (or B). Projection information is written so that the coordinate system is automatically recognized when processed by GILDAS .
Please refer to the GILDAS documentation for later processing. The recommended first steps are i) a transposition (task TRANSPOSE to put the axes in RA,DEC,Velocity ordering and ii) the tasks FILL_CUBE (for undersampled data) or MAKE_CUBE (for oversampled data) to resample the data cube on a finer grid for nice contouring.
An alternative processing technique is to use command GRID , which creates a table suited for map making using the task GRID_CUBE. This route is recommended for oversampled data, while the use of MAP and FILL_CUBE is probably better for undersampled data.
Finally GRID /IMAGE will directly produce data cubes from oversampled or undersampled data, using a Gaussian convolution kernel for interpolation, and reasonable defaults for the cube size and pixel size in both axes.
CUBE and GRID can also produce images or tables of several mathematical functions of the input spectra rather than simple channel maps. Please refer to the internal help for details.