Slicing multi-dimensional data cubes

How to slice data cubes

APLpy supports extracting a slice from n-dimensional FITS cubes, and re-ordering dimensions. The two key arguments to FITSFigure to control this are dimensions and slices. These arguments can also be passed to show_contour().

The dimensions argument is used to specify which dimensions should be used for the x- and y-axis respectively (zero based). The default values are [0, 1] which means that the x-axis should use the first dimension in the FITS cube, and the y-axis should use the second dimension. For a 2-dimensional FITS file, this means that one can use [1, 0] to flip the axes. For a FITS cube with R.A., Declination, and Velocity, [0, 2] would make a R.A.-Velocity plot.

The slices argument gives the pixels slice to extract from the remaining dimensions, skipping the dimensions used, so slices should be a list with length n-2 where n is the number of dimensions in the FITS file. For example, if one has a FITS file with R.A., Declination, Velocity, and Time (in that order), then:

  • dimensions=[0, 1] means the plot will be an R.A-Declination plot, and slices=[33, 56] means that pixel slices 33 and 56 will be used in Velocity and Time respectively (in this case, dimensions does not need to be specified since it defaults to [0, 1])
  • dimensions=[0, 2] means the plot will be an R.A-Velocity plot, and slices=[22, 56] means that pixel slices 22 and 56 will be used in Declination and Time respectively.
  • dimensions=[3, 2] means the plot will be a Time-Velocity plot, and slices=[10, 22] means that pixel slices 10 and 22 will be used in R.A and Declination respectively.

See Arbitrary coordinate systems for information on formatting the labels when non-longitude/latitude coordinates are used.

Aspect ratio

When plotting images in sky coordinates, APLpy makes pixel square by default, but it is possible to change this. When calling show_grayscale() or show_colorscale(), simply add aspect='auto' which will override the aspect='equal' default. The aspect='auto' is demonstrated below.


The following script demonstrates this functionality in use:

import matplotlib

import aplpy

f = aplpy.FITSFigure('L1448_13CO.fits.gz', slices=[222],
f = aplpy.FITSFigure('L1448_13CO.fits.gz', slices=[222],
                     dimensions=[1, 0], figsize=(5,5))
f = aplpy.FITSFigure('L1448_13CO.fits.gz', dimensions=[2, 1],
                     slices=[50], figsize=(5,5))