Introduction

Both the magma feeding system geometry and the total volume of magma injected in the volcanic edifices still remain poorly known. This constitutes one of the main limitations for a better understanding and prediction of volcanic eruptive events. However, while the existence of a magma chamber is still debated on several volcanoes, it is often observed on paleovolcanoes around the world (Gudmundsson, 2002). The tracking of magma in motion within the volcanic feeding system is thus a key challenge of modern volcano-seismology. The Réunion island was created by the Réunion hotspot and is the most recent island of the Mascarene chain. Piton de la Fournaise is one of the two stratovolcanoes located on the eastern part of Réunion Island (France). Following the quiescent period between 1992 and 1998, the volcano has been quasi continuously active since the March 1998 event. The quality of the erupted basalts has been constant during the last two centuries and was described as ``steady state basalt'' (Albarède et al., 1997). The geometry of the magma feeding system is still debated. Some authors suggest that the magma feeding system is complex and composed of small magma reservoirs (Lénat and Bachèlery, 1990) but the large deformations of the whole volcano cannot be explained by such small subsurface sources (Houlié, 2005). It is now generally agreed that there is an upper magma chamber located (we will refer to this magma chamber as P2 (Aki and Ferrazzini, 2001)) at sea level (Nercessian et al., 1996; Sigmundsson et al., 1999; Aki and Ferrazzini, 2001). The volume of the upper magma chamber is estimated to be $5\cdot 10^{-8}$ $m^{3}$. Gravimetry measurements made along an East- West profile across the volcano have been used to locate the upper magma chamber (Lesquer, 1990). Its location is coincident with the observed seismicity (Nercessian et al., 1996; Battaglia et al., 2005). The proposed volume of the magma chamber is large enough (radius$\sim 500$ m) to deform the whole volcano far away from the summit and west of the Enclos Fouqué, as is observed (Houlié, 2005). On the other hand, while the proposed size for the magma chamber is in agreement with geochemical measurements, it would be undetectable to seismic imaging. The migration of fluid coming out of this upper magma chamber and circulating inside the edifice can be detected by deformation at the surface. The use of long period (LP) or very long period (VLP) seismic events ( $0.2 Hz\leq f \leq0.5 Hz$) has been successfully applied to several volcanoes (Chouet, 1988) in order to investigate fluid circulation inside several volcano edifices. We present seismological evidence for a long-term response of the volcano to the deformation induced by changes in pressure inside the magma chamber located at sea level (Houlié and Montagner, 2007). The seismic signals are associated with the main eruptive events and recorded in the $10^{-3}-10^{-2} Hz$ frequency range.

Figure 2.45: March 27th 2001. Left: GEOSCOPE data, in counts (the data are filtered by using a lowpass filter of $10^{-2} Hz$). The three components of the LH channel (the sampling rate is equal to 1 s) are plotted. A deformation is detected by the broadband seismometer on the two horizontal components of the seismogram. Right: After deconvolving the instrument response, the displacement on the Northern component is equal to $\sim 65 mm$. Right column is displacement (in meters). The time of the eruptive event is indicated by a grey vertical line.