F R Fontaine | Université de la Réunion (original) (raw)

Papers by F R Fontaine

Nature Geoscience

Volcanic hotspot islands are thought to be surface manifestations of mantle plumes that rise from... more Volcanic hotspot islands are thought to be surface manifestations of mantle plumes that rise from the core-mantle boundary. When mantle plumes approach the surface, their mostly vertical rise must be deflected into near-horizontal flow beneath tectonic plates. This creates an opportunity to constrain their dynamics and their interactions with lithospheric plates and mid-ocean ridges. Seafloor observations have been used to propose that a focused flow in the asthenosphere transports plume heat to the nearest mid-ocean ridge, where it efficiently dissipates through formation of lithosphere. Here we present imaging results from a seismological survey of a proposed plume-to-ridge flow channel between the Réunion hotspot and the Central Indian Ridge. Rayleigh-wave tomography and shear-wave splitting confirm the presence of a channelized flow of shallow asthenosphere, eastward from the hotspot to the spreading ridge. At a larger scale, a deeper reservoir of hot asthenosphere fills vast tracts of the Indian Ocean basin east and north of Réunion Island. Its flows, decoupled from overlying lithospheres, are also directed towards the Central Indian Ridge but extend well beyond, tapped but not significantly depleted by the spreading ridge. Based on seismic and geochemical observations, we suggest that this hidden heat reservoir is generated and driven by the mantle plume, which buffers more heat near the surface than expected.

Tectonophysics, 2013

A broad-band seismic network of 28 three-component seismometers was deployed in southeastern Aust... more A broad-band seismic network of 28 three-component seismometers was deployed in southeastern Australia to examine variations in crustal thickness across the transition between Precambrian and Phanerozoic lithosphere. Receiver function observations and modelling of P-to-S conversions at the Mohorovičić discontinuity (Moho) have been employed to investigate: (i) the variations in the Moho depth across southeastern Australia, and (ii) the nature of the transition between crust and mantle. Data from temporary deployments were used together with data from the few permanent broad-band stations in the region. The extraction of P-receiver functions from high-quality seismic data recorded on these stations has enabled the determination of the crustal thickness across the region. The crustal thicknesses lie in the range 28–48 km. The Moho depth is generally well correlated with the Earth surface elevation in the southeastern Australia. The Moho estimates from receiver functions are in good agreement with results from reflection profiling. The average crustal thickness is found to be around 39 km beneath the Precambrian area in the west and even thicker beneath the Lachlan Orogen in the east (~ 43 km). The average crustal thickness in between, beneath the Murray Basin is thinner ~ 32 km. Interestingly, the crust in the Mount Gambier volcanic area is rather thick ~ 41 km, suggesting that the limit between the Delamerian and western Lachlan orogens is located east of Mount Gambier. Our results favour a position for the Tasman Line generally consistent with the interpretation by Direen and Crawford (2003) and thus to the east of the location favoured by many authors. The broader crust–mantle transition and thicker crust beneath the Lachlan Orogen suggest the presence of magmatic underplating at the base of the lower crust. The intermediate nature of the crust–mantle transition also suggests magmatic underplating beneath the Gawler Craton and the Curnamona Province.

Australian Journal of Earth Sciences, 2013

There is an ongoing debate about the tectonic evolution of southeast Australia, particularly abou... more There is an ongoing debate about the tectonic evolution of southeast Australia, particularly about the causes and nature of its accretion to a much older Precambrian core to the west. Seismic imaging of the crust can provide useful clues to address this issue. Seismic tomography imaging is a powerful tool often employed to map elastic properties of the Earth's lithosphere, but in most cases does not constrain well the depth of discontinuities such as the Mohorovičić (Moho). In this study, an alternative imaging technique known as receiver function (RF) has been employed for seismic stations near Canberra in the Lachlan Orogen to investigate: (i) the shear-wave-velocity profile in the crust and uppermost mantle, (ii) variations in the Moho depth beneath the Lachlan Orogen, and (iii) the nature of the transition between the crust and mantle. A number of styles of RF analyses were conducted: H-K stacking to obtain the best compressional–shear velocity (V P/V S) ratio and crustal thickness; nonlinear inversion for the shear-wave-velocity structure and inversion of the observed variations in RFs with back-azimuth to investigate potential dipping of the crustal layers and anisotropy. The thick crust (up to 48 km) and the mostly intermediate nature of the crust−mantle transition in the Lachlan Orogen could be due to the presence of underplating at the base of the crust, and possibly to the existing thick piles of Ordovician mafic rocks present in the mid and lower crust. Results from numerical modelling of RFs at three seismic stations (CAN, CNB and YNG) suggest that the observed variations with back-azimuth could be related to a complex structure beneath these stations with the likelihood of both a dipping Moho and crustal anisotropy. Our analysis reveals crustal thickening to the west beneath CAN station which could be due to slab convergence. The crustal thickening may also be related to the broad Macquarie volcanic arc, which is rooted to the Moho. The crustal anisotropy may arise from a strong N–S structural trend in the eastern Lachlan Orogen and to the preferred crystallographic orientation of seismically anisotropic minerals in the lower and middle crust related to the paleo-Pacific plate convergence.

Physics of the Earth and Planetary Interiors, 2008

The characteristic frequencies at which two different melt-related attenuation mechanisms occur i... more The characteristic frequencies at which two different melt-related attenuation mechanisms occur in partially molten gabbronorite are computed as a function of aspect ratios of the melt-filled cracks, and the melt viscosity. The computations were done for basaltic and andesitic melts. The melt viscosity is constrained by (i) laboratory measurements performed in the range 10^8-10^14 Pa s with a creep apparatus and in the range 10^-1 to 10^5 Pa s with a rotational Couette viscorneter, and (ii) modeling of viscosity at high temperatures. The results of the characteristic frequency calculations suggest that melt squirt flow is a viable attenuation mechanism at seismic frequencies for aspect ratio of melt inclusions in the range 10^-1 to 10^-2 for the andesite, and 10^-4 to 10^-3 for the two basalts.

Scientific Reports, 2019

Early detection of the onset of a caldera collapse can provide crucial information to understand ... more Early detection of the onset of a caldera collapse can provide crucial information to understand their formation and thus to minimize risks for the nearby population and visitors. Here, we analysed the 2007 caldera collapse of Piton de la Fournaise on La Réunion Island recorded by a broadband seismic station. We show that this instrument recorded ultra-long period (ULP) signals with frequencies in the range (0.003-0.01 Hz) accompanied by very-long period (VLP) signals (between 0.02 and 0.50 Hz) prior to and during the caldera formation suggesting it is possible to detect the beginning of the collapse at depth and anticipate its surface formation. Interestingly, VLP wave packets with a similar duration of 20 s are identified prior to and during the caldera formation. We propose that these events could result from repeating piston-like successive collapses occurring through a ring-fault structure surrounding a magma reservoir from the following arguments: the source mechanism from the main collapse, the observations of slow source processes as well as observations from the field and the characteristic ring-fault seismicity.

We present two independent, automated methods for estimating the absolute horizontal misori-entat... more We present two independent, automated methods for estimating the absolute horizontal misori-entation of seismic sensors. We apply both methods to 44 free-fall ocean-bottom seismometers (OBSs) of the RHUM-RUM experiment (http://www.rhum-rum.net/). The techniques measure the 3-D directions of particle motion of (1) P-waves and (2) Rayleigh waves of earthquake recordings. For P-waves, we used a principal component analysis to determine the directions of particle motions (polarizations) in multiple frequency passbands. We correct for polarization deviations due to seismic anisotropy and dipping discontinuities using a simple fit equation, which yields significantly more accurate OBS orientations. For Rayleigh waves, we evaluated the degree of elliptical polarization in the vertical plane in the time and frequency domain. The results obtained for the RHUM-RUM OBS stations differed, on average, by 3.1 • and 3.7 • between the methods, using circular mean and median statistics, which is within the methods' estimate uncertainties. Using P-waves, we obtained orientation estimates for 31 ocean-bottom seismometers with an average uncertainty (95 per cent confidence interval) of 11 • per station. For 7 of these OBS, data coverage was sufficient to correct polarization measurements for underlying seismic anisotropy and dipping discontinuities, improving their average orientation uncertainty from 11 • to 6 • per station. Using Rayleigh waves, we obtained misorientation estimates for 40 OBS, with an average uncertainty of 16 • per station. The good agreement of results obtained using the two methods indicates that they should also be useful for detecting misorientations of terrestrial seismic stations.

S U M M A R Y Ocean wave activity excites seismic waves that propagate through the solid earth, k... more S U M M A R Y Ocean wave activity excites seismic waves that propagate through the solid earth, known as microseisms, which, once recorded on oceanic islands, can be used to analyse the swell. Here, we analyse the microseismic noise recorded in different period ranges by the permanent seismic station RER on La Réunion Island and by a temporary network of 10 broad-band seismic stations deployed on the island to analyse extreme swell events. We perform a comparative analysis of cyclonic and austral swell events by analysing not only the primary (PM, ∼10– 20 s period) and secondary (SM, ∼3–10 s) microseisms but also the long-period secondary microseisms (LPSMs, ∼ 7–10 s), which may result from the interaction between incident ocean waves and the reflected waves off the coast. We compare the microseismic observations with buoy data when available and with hindcasts from numerical ocean wave models. We show that each cyclone is characterized by its own individual signature in the SM, which depends not only on its distance and intensity but also on its dynamics and trajectory. Thus, the SM contains relevant information for cyclone detection and monitoring. Analysing the PM and the LPSM, and comparing it to direct buoy observations and/or wave numerical models allows characterizing the local impact of the swell with the island in terms of amplitude, period, and sometimes, direction of propagation, making possible to use a seismic station as an ocean wave gauge. The microseisms, which link the atmosphere, the ocean and the solid Earth, can thus provide valuable observations on extreme swell events, in addition to oceanic and meteorological data.

Shear-wave splitting measurements in the Galápagos archipelago show a rapid change from consisten... more Shear-wave splitting measurements in the Galápagos archipelago show a rapid change from consistently oriented anisotropy to no measurable anisotropy. At the western edge of the archipelago delay times are 0.4–0.9 s and fast polarization directions are 81–109°E. These directions are consistent with anisotropy resulting from shear of the asthenosphere by the overlying plate; there is no indication of fossil lithospheric anisotropy in the plate spreading direction. In contrast, beneath the center of the archipelago the upper mantle is isotropic or weakly anisotropic. The isotropic region coincides approximately with a volume of anomalously low upper mantle velocities, suggesting that the presence of melt may weaken the effects of fabric on anisotropy or that melt preferred orientation generates a vertical fast polarization direction. Alternatively, the complex flow field associated with a near-ridge hotspot may result in apparent isotropy.

Torsion oscillatory deformation experiments have been performed at high temperatures (600-1170 • ... more Torsion oscillatory deformation experiments have been performed at high temperatures (600-1170 • C) and over a wide range of low frequencies (20-2.10 −3 Hz) on fine-grained gabbronorite samples from the Oman ophiolite in order to determine the shear wave attenuation as a function of temperature and melt fraction. The specimens have a small and uniform grain size (0.25-0.3 mm) and do not contain secondary, hydrated minerals. Measurements of internal friction (Q −1 ) were performed using a forced oscillatory torsion apparatus at small strains (∼10 −7 ), and with increasing small temperature steps to reduce thermal microcracking. The general dependence of Q −1 to frequency is Q −1 ∝ ω −α , where ω is the angular velocity of forced oscillations and α is an empirical exponent. Below the melting temperature (∼1050 • C), α has average values of ∼0.15 at low frequency (≤0.5 Hz) and 0.06 at higher frequency. Above the melting temperature, α has average values of ∼0.22 at low frequency and −0.02 at higher frequency. This frequency dependence of Q −1 is attributed to a viscoelastic behaviour due to the diffusion controlled grain boundary sliding, and partially to the squirt flow of the meltphase wetting grain boundaries. The onset of melting is associated with a markedly higher Q −1 and a stronger dependence of Q −1 on temperature. The melt-related mechanical dissipation process could be a melt squirt flow. The characteristic frequency for the melt squirt flow is ω m ∼ 0.15-300 Hz when the melt pocket aspect ratio is ∼10 −3 -10 −2 . Around the melting temperature the internal friction can be approximated by an experimental power law Q −1 = A · [ω −1 · d −1 · exp(−E a /RT)] α with α ∼ 0.08, A = 34.72 s −α µm −α and E a ∼ 873 kJ mol −1 .

Seismic records from La Réunion Island very broadband Geoscope station are investigated to constr... more Seismic records from La Réunion Island very broadband Geoscope station are investigated to constrain the link between the 2007 eruptive sequence and the related caldera collapse of the Piton de la Fournaise volcano. Tilt estimated from seismic records reveals that the three 2007 eruptions belong to a single inflation-deflation cycle. Tilt trend indicates that the small-volume summit eruption of 18 February occurred during a phase of continuous inflation that started in January 2007. Inflation decelerated 24 days before a second short-lived, small-volume eruption on 30 March, almost simultaneous with a sudden, large-scale deflation of the volcano. Deflation rate, which had stabilized at relatively low level, increased anew on 1 April while no magma was erupted, followed on 2 April by a major distal eruption and on 5 April by a summit caldera collapse. Long-term tilt variation suggests that the 2007 eruptive succession was triggered by a deep magma input.

S U M M A R Y A temporary network of 10 broad-band seismic stations has been installed in French ... more S U M M A R Y A temporary network of 10 broad-band seismic stations has been installed in French Polynesia for the Polynesian Lithosphere and Upper Mantle Experiment (PLUME). All the seismic stations were installed either on volcanic islands or on atolls of the various archipelagos of French Polynesia in a manner which complements the geographic coverage provided by the regional permanent stations. The primary aim of PLUME is to image the upper mantle structures related to plate motion and hotspot activity. However, because of its proximity to all sites, the ocean is responsible for a high level of noise in the seismic data and we show that these data can also be used to analyse ocean wave activity. The power spectral density (PSD) analyses of the seismic data recorded in French Polynesia show clear peaks in the 0.05– 0.10 Hz band (periods between 10 and 20 s), which corresponds to swell frequencies. Clear peaks in this frequency band are also observed in infrasonic data recorded on Tahiti. Ground motion analysis shows that the swell-related seismic noise (SRSN) is linearly polarized in the horizontal plane and its amplitude decreases rapidly with the distance from the shore. The microseismic and the infrasonic 'noise' amplitudes show very similar variations from station to station and both are strongly correlated with the swell amplitudes predicted by the National Oceanic and Atmospheric Administration (NOAA), wind-forced, 'WaveWatch' models. The swell direction can be estimated from SRSN polarization analysis but this has to be done with care since, for some cases, the ground motions are strongly controlled by the islands' anisometric shapes and by swell refraction processes. We find cases, however, such as Tahiti or roughly circular Tuamotu atolls, where the azimuth of the swell is in good agreement with the seismic estimates. We, therefore, demonstrate that the SRSN and the infrasonic signal observed in French Polynesia can be used in such cases as a proxy for swell amplitude and azimuth. From the continuous analysis of the data recorded in 2003 at the permanent seismic station PPTL in Tahiti, transfer functions have been obtained. This could provide a way to quantify the swell activity during the last two decades and, therefore, assist in the investigation of climate changes.

Ocean waves activity is a major source of microvibrations that travel through the solid Earth, kn... more Ocean waves activity is a major source of microvibrations that travel through the solid Earth, known as microseismic noise and recorded worldwide by broadband seismometers. Analysis of microseismic noise in continuous seismic records can be used to investigate noise sources in the oceans such as storms, and their variations in space and time, making possible the regional and global-scale monitoring of the wave climate. In order to complete the knowledge of the Atlantic and Pacific oceans microseismic noise sources, we analyse 1 yr of continuous data recorded by permanent seismic stations located in the Indian Ocean basin.We primarily focus on secondary microseisms (SM) that are dominated by Rayleigh waves between 6 and 11 s of period. Continuous polarization analyses in this frequency band at 15 individual seismic stations allow us to quantify the number of polarized signal corresponding to Rayleigh waves, and to retrieve their backazimuths (BAZ) in the time–frequency domain. We observe clear seasonal variations in the number of polarized signals and in their frequencies, but not in their BAZ that consistently point towards the Southern part of the basin throughout the year. This property is very peculiar to the Indian Ocean that is closed on its Northern side, and therefore not affected by large ocean storms during Northern Hemisphere winters. We show that the noise amplitude seasonal variations and the backazimuth directions are consistent with the source areas computed from ocean wave models.

Ocean wave activity excites seismic waves that propagate through the solid earth, known as micros... more Ocean wave activity excites seismic waves that propagate through the solid earth, known as microseismic noise. Here we use a network of 57 ocean bottom seismometers (OBS) deployed around La Réunion Island in the southwest Indian Ocean to investigate the noise generated in the secondary microseismic band as a tropical cyclone moved over the network. Spectral and polarization analyses show that microseisms strongly increase in the 0.1–0.35 Hz frequency band as the cyclone approaches and that this noise is composed of both compressional and surface waves, confirming theoretical predictions. We infer the location of maximum noise amplitude in space and time and show that it roughly coincides with the location of maximum ocean wave interactions. Although this analysis was retrospectively performed, microseisms recorded on the seafloor can be considered a novel source of information for future real-time tracking and monitoring of major storms, complementing atmospheric, oceanographic, and satellite observations.

Acta Oecologica, 2015

We deployed five broadband three-components seismic stations in the Iles Eparses in the south-wes... more We deployed five broadband three-components seismic stations in the Iles Eparses in the south-west Indian Ocean and on Mayotte Island, between April 2011 and January 2014. These small and remote oceanic islands suffer the effects of strong ocean swells that affect their coastal environments but most islands are not instrumented by wave gauges to characterize the swells. However, wave action on the coast causes high levels of ground vibrations in the solid earth, so-called microseismic noise. We use this link between the solid earth and ocean wave activity to quantify the swells locally. Spectral analyses of the continuous seismic data show clear peaks in the 0.05e0.10 Hz frequency band (periods between 10 and 20 s), corresponding to the ocean wave periods of the local swells. We analyze an example of austral swell occurring in August 2013 and a cyclonic event (Felleng) that developed in January 2013, and quantify the ground motion at each station induced by these events. In both cases, we find a linear polarization in the horizontal plane with microseismic amplitude directly correlated to the swell height (as predicted by the global swell model WaveWatchIII), and a direction of polarization close to the predicted swell propagation direction. Although this analysis has not been performed in real time, it demonstrates that terrestrial seismic stations can be efficiently used as wave gauges, and are particularly well suited for quantifying extreme swell events. This approach may therefore provide useful and cheaper alternatives to wave buoys for monitoring swells and the related environmental processes such as beach erosion or coral reef damages.

Geophysical Research Letters, 2009

Geophysical Journal International, 2007

Upper-mantle flow beneath the South Pacific is investigated by analysing shear wave splitting par... more Upper-mantle flow beneath the South Pacific is investigated by analysing shear wave splitting parameters at eight permanent long-period and broad-band seismic stations and 10 broad-band stations deployed in French Polynesia from 2001 to 2005 in the framework of the Polynesian Lithosphere and Upper Mantle Experiment (PLUME). Despite the small number of events and the rather poor backazimuthal coverage due to the geographical distribution of the natural seismicity, upper-mantle seismic anisotropy has been detected at all stations except at Tahiti where two permanent stations with 15 yr of data show an apparent isotropy. The median value of fast polarization azimuths (N67.5 • W) is parallel to the present Pacific absolute plate motion direction in French Polynesia (APM: N67 • W). This suggests that the observed SKS fast polarization directions result mainly from olivine crystal preferred orientations produced by deformation in the sublithospheric mantle due to viscous entrainment by the moving Pacific Plate and preserved in the lithosphere as the plate cools. However, analysis of individual measurements highlights variations of splitting parameters with event backazimuth that imply an actual upper-mantle structure more complex than a single anisotropic layer with horizontal fast axis. A forward approach shows that a two-layer structure of anisotropy beneath French Polynesia better explains the splitting observations than a single anisotropic layer. Secondorder variations in the measurements may also indicate the presence of small-scale lateral heterogeneities. The influence of plumes or fracture zones within the studied area does not appear to dominate the large-scale anisotropy pattern but may explain these second-order splitting variations across the network.

Journal of Geophysical Research, 2009

1] We report measurements of long-period P wave polarization (P pol ) in Australia and Tahiti mad... more 1] We report measurements of long-period P wave polarization (P pol ) in Australia and Tahiti made by combining modeling of the polarization deviation and harmonic analysis. The analysis of the deviation of the horizontal polarization of the P wave as a function of event back azimuth may be used to obtain information about (1) sensor misorientation, (2) dipping discontinuities, (3) seismic anisotropy, and (4) velocity heterogeneities beneath a seismic station. The results from harmonic analysis and a grid search using Snell's law suggest the presence of a dipping seismic discontinuity beneath stations CTAO and CAN in Australia. These results are consistent with published receiver function studies for these stations. The P pol fast axis orientation is close to the N-S absolute plate motion direction at station TAU (Tasmania), which may be due to plate-motion-driven alignment of olivine crystals in the asthenosphere. Interestingly, measurements of SKS splitting at Tahiti (French Polynesia) show an apparent isotropy, whereas an inversion of P pol observations at PPTL seismic station located in Tahiti suggests the presence of two anisotropic layers. The fast axis azimuth is oriented E-W in the upper layer, and it is close to the NW-SE orientation in the lower layer. Since P pol orientations are used for real-time earthquake locations, especially in poorly instrumented areas such as the South Pacific, we show that the bias from anisotropy and sensor misorientation determined here can be corrected to improve the location accuracy, which yields fundamental data for rapid location necessary for effective tsunami warning.

Earth and Planetary Science Letters, 2013

Keywords: SKS splitting anisotropy lithosphere asthenosphere plume La Ré union a b s t r a c t If... more Keywords: SKS splitting anisotropy lithosphere asthenosphere plume La Ré union a b s t r a c t If upper mantle anisotropy beneath fast-moving oceanic plates is expected to align the fast azimuths close to the plate motion directions, the upper mantle flow pattern beneath slow-moving oceanic plates will reflect the relative motion between the moving plate and the underlying large-scale convecting mantle. In addition to the non-correlation of the fast azimuths with the plate motion direction, the flow and anisotropy pattern may be locally perturbed by other factors such as the upwelling and the sublithospheric spreading of mantle plumes. Investigating such plume-lithosphere interaction is strongly dependent on the available seismological data, which are generally sparse in oceanic environment. In this study, we take the opportunity of recent temporary deployments of 15 seismic stations and 5 permanent stations on the Piton de la Fournaise volcano, the active locus of La Ré union hotspot and of 6 permanent stations installed along or close to its fossil track of about 3700 km in length, to analyze azimuthal anisotropy detected by SKS wave splitting and to decipher the various possible origins of anisotropy beneath the Western Indian Ocean. From about 150 good and fair splitting measurements and more than 1000 null splitting measurements, we attempt to distinguish between the influence of a local plume signature and large-scale mantle flow. The large-scale anisotropy pattern obtained at the SW-Indian Ocean island stations is well explained by plate motion relative to the deep mantle circulation. By contrast, stations on La Ré union Island show a complex signature characterized by numerous ''nulls'' and by fast split shear wave polarizations trending normal to the plate motion direction and obtained within a small backazimuthal window, that cannot be explained by either a single or two anisotropic layers. Despite the sparse spatial coverage which precludes a unique answer, we show that such pattern may be compatible with a simple model of sublithospheric spreading of La Ré union plume characterized by a conduit located at 100-200 km north of La Ré union Island. Anisotropy beneath the new GEOSCOPE station in Rodrigues Island does not appear to be influenced by La Ré union plume-spreading signature but is fully compatible with either a model of large-scale deep mantle convection pattern and/or with a channeled asthenospheric flow beneath the Rodrigues ridge.

Eos, Transactions American Geophysical Union, 2002

The French Ministere de la Recherche is funding a multidisciplinary project, the Polynesian Litho... more The French Ministere de la Recherche is funding a multidisciplinary project, the Polynesian Lithosphere and Upper Mantle Experiment (PLUME), to image the upper mantle structures beneath French Polynesia. This region of the southwestern Pacific, which is far from any plate boundary comprises oceanic lithosphere with ages varying between 30 and 100 Ma, as well as two major fracture zones. The

Geophysical Journal International, 2006

A temporary network of 10 broad-band seismic stations has been installed in French Polynesia for ... more A temporary network of 10 broad-band seismic stations has been installed in French Polynesia for the Polynesian Lithosphere and Upper Mantle Experiment (PLUME). All the seismic stations were installed either on volcanic islands or on atolls of the various archipelagos of French Polynesia in a manner which complements the geographic coverage provided by the regional permanent stations. The primary aim of PLUME is to image the upper mantle structures related to plate motion and hotspot activity. However, because of its proximity to all sites, the ocean is responsible for a high level of noise in the seismic data and we show that these data can also be used to analyse ocean wave activity. The power spectral density (PSD) analyses of the seismic data recorded in French Polynesia show clear peaks in the 0.05–0.10 Hz band (periods between 10 and 20 s), which corresponds to swell frequencies. Clear peaks in this frequency band are also observed in infrasonic data recorded on Tahiti. Ground motion analysis shows that the swell-related seismic noise (SRSN) is linearly polarized in the horizontal plane and its amplitude decreases rapidly with the distance from the shore. The microseismic and the infrasonic ‘noise’ amplitudes show very similar variations from station to station and both are strongly correlated with the swell amplitudes predicted by the National Oceanic and Atmospheric Administration (NOAA), wind-forced, ‘WaveWatch’ models. The swell direction can be estimated from SRSN polarization analysis but this has to be done with care since, for some cases, the ground motions are strongly controlled by the islands' anisometric shapes and by swell refraction processes. We find cases, however, such as Tahiti or roughly circular Tuamotu atolls, where the azimuth of the swell is in good agreement with the seismic estimates. We, therefore, demonstrate that the SRSN and the infrasonic signal observed in French Polynesia can be used in such cases as a proxy for swell amplitude and azimuth. From the continuous analysis of the data recorded in 2003 at the permanent seismic station PPTL in Tahiti, transfer functions have been obtained. This could provide a way to quantify the swell activity during the last two decades and, therefore, assist in the investigation of climate changes.