4. Data reduction
Our spectra were corrected for the dark and the gain matrices, and the superposed straylight aureole was subtracted. The thus obtained time series of Ca 8542 emission spectra were displayed as a movie. Fig. 2, containing a sample of 12, i.e. each tenth, of the total of 120 spectra of prominence-A, gives an impression of what the movie clearly shows: a 'snake like winding' of the emission streak with time, but also 'lateral' motions at the boundaries where structures separate from the main prominence body.
For a quantitative analysis, Gaussian profiles were fitted to the emissions. This yields for each spectrum the spatial variation of central wavelength and maximum line intensity along the slit. The temporal and spatial variation of the Doppler shifts given in Fig. 3 shows the oscillations indicated in Fig. 2 and well visible in the movie. Fig. 3 also shows that some oscillations do not last over the total time interval and cover restricted spatial areas of a few arcsec.
Finally, the power as a function of the time frequency was determined for locations of pronounced emission after suitable spatial averaging (Fig. 4). Instead of the commonly applied Fourier analysis, we computed the Power spectra by means of a Lomb normalized periodogram (see e.g. Press et al. 1988). This algorithm also works for unevenly sampled data with time interruptions (as introduced by small cirrus clouds for prominence-C and by hardware failures for prominence-D), and furthermore allows for a much improved frequency resolution.
© European Southern Observatory (ESO) 1997
Online publication: June 30, 1998