High-sensitivity strain measurements from underground interferometric stations: geodynamic phenomena at Gran Sasso and first records from Canfranc

Abstract

Earth's surface and interior continuously deform as a result of geological and geophys- ical processes. To study these phenomena and to understand better the rheological properties of the Earth, measurements of Earth's deformation become of fundamental importance, providing a critical link between Earth's structure and dynamics, also in order to optimize the response to natural hazards and identify potential risk areas. The study of crustal deformation is a complex but very important research topic that encompasses several scienti c disciplines, including di erential geometry, theory of elas- ticity, geodynamics and physics in general. The use of di erent kind of geodetic data to study geodynamic phenomena became increasingly important, playing a key role in the knowledge of their temporal and mag- nitude variations at many di erent spatial and time scales. These measurements provide signi cant constraints on the changes in Earth's lithosphere and processes that cause them, like for example movements of magma, changes in strain before, during, and after earthquakes, motion of ice sheets. Yet even today, large portions of the Earth are infrequently monitored, or not at all. Deformation can be measured in several ways, as relative movement of points on the Earth's surface, through measurements of strain and tilt, or by GPS (Global Position- ing System), VLBI (Very Long Base Interferometry) and SLR (Satellite Laser Ranging) measurements. Among the di erent types of instruments, the laser strainmeters (or in- terferometers), measuring the displacement between two points away from a few meters to over a kilometer, are characterized by very high accuracy and long-term stability, necessary to investigate processes of crustal deformation. The analysis of interferomet- ric data allows to study both local and global geodynamic phenomena in a broad band of frequencies. This thesis introduces results related to some analysis of data recorded by two laser interferometers installed at Gran Sasso (Italy) Underground Laboratories and describes the installation of two new laser interferometers in the Canfranc (Spain) Underground Laboratory, at the end of August 2011, with the analysis of their rst sequences. In the rst chapter some general concepts about strain, crustal deformation and their measurements are introduced. The study of the deformation on the Earth's surface improved in the last fty years, changing from mostly descriptive and qualitative to more quantitative. The state and magnitude of the stress in the Earth's lithosphere, and thus of the deformation, play an important role on various geophysical problems, such as the plate mechanisms, energy budget of the Earth, earthquake mechanism and crustal movements. In Chapter 2 there is a description of the Earth's tidal deformation. The body tides, due to a direct e ect of gravitational attraction from the Sun, Moon and other objects, can be modeled very accurately. In addition, there is a part of deformation, known as ocean loading, arising from the mass uctuations of the oceans. These last is also rather well understood, but the modeling of its e ects still needs to be improved. This phenomenon is very signi cant for the interferometric strain sequences because they are clearly dominated by the semidiurnal and diurnal strain tides. Chapter 3 describes the laser interferometry and, in particular, the operating principle of the Gran Sasso (Italy) laser interferometers, which provide very high-sensitivity strain data, by comparing the optical length of a longer measurement arm (about 90 meters in length) and a shorter xed reference arm (about 20 cm in length). Although the interferometers measure strain directly, the presence of cavities, topography, and local inhomogeneities of the crust can modify the strain measurement considerably. Also environmental and anthropic e ects appear as anomalous or noise signals in a broad frequency range, which includes for example the Earth tides. It is necessary to take some or all these possible e ects into account, depending on the phenomenon studied. In Chapter 4 data produced by Gran Sasso interferometers, rst alone and then to- gether with those produced by a third laser interferometer installed in Baksan (Russia) Underground Laboratory, are used to estimate the Free Core Nutation (FCN) param- eters. Even if tidal signal-to-noise ratio for strain is usually lower than for gravity, the analysis of strain data is promising, because relative perturbations in strain tides are about 10 times larger than in gravity tides. The inversion of realistic synthetic tidal parameters (obtained from observed amplitude and phase of eight diurnal tidal components) shows that the resolving power of strain tides is comparable to that of gravity tides if tidal parameters are inverted minimizing the L2 mis t (as usually done). Both resolving powers improve if data are inverted minimizing the L1 mis t, and this improvement is particularly notable for gravity tides. The inversions of strain records have been performed after correcting measured strain for local distortion of the regional strain eld and ocean loading. For estimating the FCR parameters, eight diurnal tidal constituents (Q1, O1, P1, K1, 1, 1, J1 and OO1) have been used, by comparing mea- surements (corrected for ocean loading) and model predictions (corrected for the local strain distortion), minimizing the L1 mis t. The analysis of the only Gran Sasso strain data provides a value for the FCN period (about 429 sidereal days) robust and compa- rable to those from gravity tides, obtained from the joint inversion of data from several stations. The agreement between observations and predictions looks better than in any previous work that makes use of strain tides. The joint analysis of Gran Sasso and Bak- san strain data con rms, but does not improve, these results recently obtained. In both cases the quality factor is not well constrained because of the large uncertainty on the 1 phase; however the results are consistent with recently published values ( 20000). In Chapter 5 the new mode of faulting, discovered in the last decades and referred to as slow slip earthquakes, is examined. Many aspects of slow slip remain unexplained. Here an attempt to describe the characteristics of the rupture propagation through the analysis of strain records from three di erent slow events related to the 1978 Izu- Oshima (Japan) earthquake, the 1999 Durmid Hill (California) slow event and the 2003 Tokachi-oki (Japan) earthquake. The signals recorded during these slow events exhibit the same peculiarities observed in the strain sequences recorded at Gran Sasso during the 6 April 2009 L'Aquila earthquake, rst direct measurement of the di usive character of the rupture propagation. By using two di erent propagation mechanisms along 1D straight paths, namely constant propagation velocity and di usive processes, predicted strain history at both interferometers is fully consistent with di usive slip propagation, but inconsistent with constant velocity propagation. Not all slow earthquakes analysed are consistent with only one of the two models tested. Two of the four slow events (L'Aquila and Izu-Oshima F3 fault) are only consistent with di usive slip propagation. Constant velocity propagation gives much worst t to data, being unable to t the shape of minimum observed on signals recorded by BA interferometer and SHI borehole strainmeter for L'Aquila and Izu-Oshima slow earthquakes, respectively. For both slow events, the seismic moment density decreases about linearly with distance along the path, like the steady-state solution of 1D di usive processes. In the other cases (Durmid Hill, Tokachi-oki and Izu-Oshima F4 fault) it is not possible to discriminate the type of propagation. Observations are consistent with both types of slip propagation but for them the shape of seismic moment seems somewhat unrealistic, being a bell-shaped distribution peaked on the nodal line. These results suggest the necessity to deepen the source features, not well constrained in some cases. Moreover, also the assumption of 1D very thin path might be a source of uncertainty. In the last chapter the installation, occured in August 2011, of two new laser interferom- eters is described and their rst records analysed. These instruments, operating since November 2011, are installed in the Canfranc (Spain) Underground Laboratory (LSC). The LSC is located at depth in one of the most seismically active areas in Western Europe, at the Pyrenean chain that marks the boundary between the European plate and the Iberian microplate. The rst tests on strain data recorded by these interfer- ometers evidence the capability of producing clear records of low-frequency signals, for example relating to seismic waves, Earth free oscillations, and possible local aseismic stress release. A preliminary tidal analysis shows a good agreement between observed and predicted tides in the diurnal tidal band, suggesting that, if any, local strain dis- tortion e ects are small. In the semidiurnal tidal band, discrepancies between observed and predicted tides are noticeable; this might be a consequence of inadequate Earth and/or ocean models. These results deserve further investigation; in particular it would be interesting to deepen the local distortion e ect in the di erent frequency bands and estimate the ocean loading in more detail, especially in the Bay of Biscay which could be the main source of the discrepancies observed in the semidiurnal tidal band. [edited by author]