This is one technique of writing arrays to fits files and correspondingly accessing those arrays. I create a fits file where each column of data corresponds to an array. In this example I have three ndarrays (wv_full_test,  fx_full_test, and (fx_full_test*.1).

(1)  Writing table data arrays to a fits file:
$import pyfits

$col1 = pyfits.Column(name = ‘wave’, format = ‘D’, array = wv_full_test)
$col2 = pyfits.Column(name = ‘fx’, format = ‘D’, array = fx_full_test)
$col3 = pyfits.Column(name = ‘fx_error’, format = ‘D’, array = (wv_full_test*.1))

$cols = pyfits.ColDefs([col1, col2, col3])
$tbhdu = pyfits.new_table(cols)
$hdu = pyfits.PrimaryHDU()
$thdulist = pyfits.HDUList([hdu, tbhdu])
$tbhdu.writeto(‘filename.fits’)

(2) Reading fits files into table data arrays:
$input = pyfits.core.getdata(‘filename.fits’, 1)

(3) To access each individual array:
$wavelength_array = input.field(‘wave’)
$fx_array = input.field(‘fx’)
$fx_error_array = input.field(‘fx’)

For more information take a look at the Pyfits Handbook