1 Introduction

The five-minute solar oscillations have proved to be a powerful tool in aiding our understanding of the solar interior. The mysterious nature of these oscillations was identified by way of theoretical arguments in 1970 and confirmed by observations in 1975. The oscillations we see on the surface are due to sound waves generated and trapped inside the sun. Sound waves, produced by pressure fluctuations, set the sun vibrating in millions of different patterns or p-modes. In the same way that geologists use seismic waves from earthquakes, helioseismology permits us to deduce the properties of the inner layers of the sun from the study of a perturbation on its surface. Despite the large impact helioseismology has produced in solar physics, the origin of the oscillations has not yet been clearly revealed, that is, the precise mechanism generating the pressure fluctuations is still uncertain. Although simulations seem to sufficiently describe the driving of the p-modes (Stein & Nordlund 2001), no observational evidence can clearly select this phenomenon as dominant in the excitation.

A first identikit of the source has been drawn by the asymmetry of the p-mode line profiles. If the p-modes are excited as randomly forced, damped harmonic oscillators, the shape of their lines in the power spectra would show a lorentzian profile. The asymmetry in these profiles (Duvall et al. 1993) was originally interpreted as the presence of a localised excitation source (Gabriel 1992; Roxburgh & Vorontsov 1993) in a thin layer near the top of the convection zone.

The presence of a solar "background'' in the $\ell -\nu$ diagram of the phase difference between the intensity and velocity signals (I-V), first discovered from ground observations by Deubner et al. (1990) as a negative phase "plateau'' and an inter-ridge regime at high-$\ell$ values and low frequency, has been associated with a possible signature of the source of the resonant oscillations.

Different scenarios have been invoked to explain the observed background. The model proposed by Deubner et al. (1996) interprets the background's characteristics in the phase spectra in terms of the response of a cavity in the atmosphere. The photospheric wave field, characterised by evanescent five-minute p-modes and high frequency running acoustic waves, gradually transforms into resonant oscillations, possibly with several eigenfrequencies. The background has been also explained in terms of the acoustic source, primarily identified with the fast cooling associated with the convective "downdrafts'' (Skartlien & Rast 2000). This latter interpretation is consistent with the observations of Goode et al. (1998), Espagnet et al. (1996) and Strous et al. (2000). In fact, seismic events have been detected following a darkening and a collapse of plasma localised in the intergranular lanes. These events last a few minutes and extend over an area of few arcseconds. These downdrafts are thought to be due to buoyant acceleration after a radiative cooling at the surface, that is, statistically triggered by convection.

Recently, a correlation between magnetic oscillations, H$\alpha$ bright points and the background locations (as revealed using a local analysis technique to filter the data) has been found (Moretti et al. 2000). Nevertheless, no direct correlation between the magnetic flux and the downflows has been found so far.

Observational results are often in disagreement when finding spatial and temporal correlation between Ca K bright points, magnetic field cancellations, UV jets etc. (Cook et al. 1996; Hoekzema et al. 1997; Lites et al. 1999). Transition region explosive events have been correlated with magnetic cancellations as a consequence of the upward expansion of relaxed magnetic ropes (Chae et al. 1998), but these cancellations are difficult to localise due to their small scales ($\simeq$1 $^{\prime \prime }$).

The evidence in the photosphere of a strong seismic downplume associated with a big flare was detected by MDI (Kosovichev & Zarkova 1998); and high $\ell$-degree mode excitation has been reported, too (Haber et al. 1988). Similar events were predicted and noted as a possible origin of free oscillations in the sun (Wolff 1972).

In this paper we investigate the possible nature of the source of the solar oscillations when they are identified by the I-V phase differences using a local analysis. The spatial and temporal characteristics of the source locations mimic those of the events related to the explosive chromospheric evaporation (Briand, private communication; Canfield & Metcalf 1987).

We invoke the downward plumes to relax their energy in the photosphere and trigger seismic events.