## 7. ConclusionsPrecise p-mode resonance frequencies are required if we aim to improve the precision of the inversions for the interior sound speed. With this in mind, we have investigated the importance of asymmetry in the p-mode resonance profiles. It is known that these effects will in general be small for low- global resonances and also reversed for velocity measurements compare for intensities one. A detailed study, using GOLF data, simulated GOLF data and Monte Carlo simulations has shown that meaningful measurements of asymmetric effects can be performed. When the GOLF data is fitted using asymmetric fitting based upon the Nigam formulation, a number of anomalies, previously encountered with Lorentzian fitting, tend to go away. Such anomalies include the variation of resonance widths amongst members of a multiplet. The dispersion of many of the fitting parameters is also reduced when using asymmetric fitting. We have also investigated the optimum choice of spectral window, when using these fittings. This is a compromise between large windows, for which the fitting procedure may become formally less valid, and small windows, which increase the statistical uncertainty due to stochastic excitation effects. The breakdown in validity of the fitting procedure for large windows is manifest in our studies. This may be due to a limitation in the approximation used in the profile expression adopted, but may also be due to a formal error in assuming that the profiles are additive on a scale of spectral energy. For this reason we exclude the idea of fitting together the entire p-mode spectrum, when using asymmetric profiles. The asymmetry parameters derived are claimed in the literature to be due to two distinct physical processes: the existence of direct signals from excitation sources and the interaction of the resonances with correlated solar noise. These two processes make contributions to the asymmetry parameter with the same functional form, within the approximation used here. It is therefore difficult to see how we can hope to separate these effects, in order to determine the height of the excitation region, unless we make the arbitrary assumption that one or other of the effects is negligible. Work is now in hand to use these improved GOLF frequencies for inversions. This is being reported in a second paper (Basu et al. 1999). As always, it is important to have good frequencies also for medium-, in order to carry out this inversion. At the time of writing, MDI medium- frequencies are only available based on Lorentzian fitting. Although we are using these, there is an obvious inconsistency with our improved asymmetric frequencies. Improved inversions must therefore await the availability of asymmetric data at both low and medium-. © European Southern Observatory (ESO) 2000 Online publication: March 9, 2000 |