4. P-modes widths and widths error bars
We studied the p-modes linewidths (FWMH) for the 4 years of IRIS data from 1898 to 1992, with duty ranging from 40 to 57 %. We used the same technique as in the frequency determination (Gelly et al., 1997). However, we forced the central frequency of any given peak to our tabulated value, and the splitting value to (Lazrek et al., 1996). The widths were free parameters, and were different within even and odd groups, even though they were determined together. The linewidth is related to the lifetime of the mode by . When the linewidth becomes of the same magnitude as the distance to the daily sidelobes of the peak, our measurement is biased and the former relation no longer applies. This happens above .
As mentionned in 2, the linewidth in the spectral density is increased because of the convolution by the window function. The given values are corrected to first order, i.e. by substracting the amount by which a given linewidth is increased after the convolution, although it has been checked that this correction is well inside the error bars. Moreover, those small corrections do not exhibit a clear relationship with the linewidth in the frequency range that we considered, justifying this linear correction.
We have checked for linewidths variation among the years 1989 to 1992, and we did not find any, at our current level of precision. No dependency with the duty-cycle has been noticed. We present in Table 1 the mean widths for l=0, 1 and 2 for those 4 years.
Table 1. IRIS p-mode widths for 4 years of dataset (1989-92)
Fig. 8 shows the IRIS results and a comparison with the IPHIR (Toutain et al. , 1992) and LOI measurements (Appourchaux et al., 1997). Our result also shows a flat level of width, which was first described by Libbrecht (1988), showing that the mode damping may be due to another processes than turbulent or radiative viscosity, but we do not observe such a dip as presented for LOI results, which is interpreted as a resonance effect with convection (Frohlich et al., 1997). There is only a hint of such a feature for the mode.
4.2. Width error bars
The theoretical sigma (Toutain and Appourchaux, 1994) is given by Eq. (8) and corresponds to a fit with an unconstrained frequency on one Lorentz profile.
The difference of free parameters in the maximum likelihood fitting procedure modifies the uncertainties values. An imposed frequency fit decreases the error bars of the width (Fig. 9) and of the amplitude, and we believe that this is a real improvement in the FWHM determination of complicated cases involving a multiplet mode and a sidelobe.
© European Southern Observatory (ESO) 1998
Online publication: April 15, 1998