1

Continuous Tilt Monitoring: Lesson Learned From 20 Years Experience at Mt. Etna

Alessandro Bonaccorso, Orazio Campisi, Giuseppe Falzone and Salvatore Gambino

Istituto Nazionale di Geofisica e Vulcanologia (INGV), Catania, Italy

In this paper we describe the experiences and results gained in the use of tiltmeters atMt. Etna during the last twenty years. The tilt data represent a fundamental contribution tounderstanding of volcanic phenomena. Moreover, the tilt played an important role in ana-lytic modeling, of the sources linked to the most important recent lateral eruptions.

The tilt data also provided evidence of co-seismic variations for higher energy and shal-lower seismic events. We also discuss the marked variations that often occur over theentire volcano edifice without being associated with seismicity or eruptive activity.

This work also analyses and discusses experiments conducted to verify the limitationsof shallow borehole sensors. The main causes of noise are the thermo-elastic effects on theground and the spurious effects on the sensor due to temperature variations. To this end,results of recent experiments to verify the theoretical filtering of thermal noise at depth andto examine the real thermal coefficients of the sensors through a patented instrument arediscussed.

Finally, results obtained from signal recordings, applied theoretical studies, experimentsperformed and technological advances have recently suggested new horizons for tilt mon-itoring through the development of innovative instrumentation. We describe the applica-tion of new technology for the detection of tilt by a long base fluid tiltmeter, installed in1997, which is able to record very stable, high precision signals with very low noise, thusallowing more detailed investigation of the different phases of volcanic activity.

1. INTRODUCTION

Shallow borehole tiltmeters are used for ground deforma-tion monitoring in many volcanic areas such as Mt St. Helens[Dvorak et al.1981; Dzurisin et al., 1983; Dzurisin, 1992],Kilauea [Okamura et al., 1988], Long Valley Caldera[Mortensen & Hopkins, 1987], Piton de la Fournaise [Toutainet. al. 1992], Montserrat [Voight et. al., 1998], GunungMerapi [Rebscher et al. 2000]. Continuous tilt measurementsare usually used to record middle-short term eruption precur-

sors. In general, slow tilt variations (from weeks to months)could indicate inflation caused by rising magma prior to theeruption or deflation linked to energy release followingeruptions, while fast tilt variations (from hours to days) areoften related to rapid rise of magma and propagation ofdikes and eruptive fissures [e.g., Dzurisin et al., 1983;Okamura et al., 1988; Mellors et al., 1991; Toutain et al.,1992; Bonaccorso, 1998].

At Mt. Etna ground deformation studies began in the1970s with small leveling and EDM networks located in thesummit area (e.g., Wadge, 1976; Murray and Guest, 1982].At the end of 1970s EDM networks and a leveling routewere installed around the flanks the volcano [e.g.,Bonaccorso et al., 1995], and by the end of 1980s the GPSmeasurements covered the entire volcano edifice [e.g.,

Nunnari and Puglisi, 1994]. All these measurements wereusually repeated once per year. Continuous ground defor-mation monitoring was first attempted during the seven-ties using a few tiltmeter stations (2-3) for short periods,usually a few months. [Wadge et al., 1975; Henbest et al.,1980; Davis, 1981]. During the late eighties a network ofpendulum tiltmeters with local remote recording was

installed around the volcano [Briole et al., 1992] thatshowed a constant deformation before the 1989 eruption[Briole et al., 1990].

The Istituto Internazionale di Vulcanologia, nowIstituto Nazionale di Geofisica e Vulcanologia - Sezionedi Catania (INGV-CT), began using borehole tiltmeterswith electrolytic bubble sensors on Mt. Etna in 1977 [e.g.,

2 CONTINUOUS TILT MONITORING

Figure 1. Examples of tilt signals from single stations recorded during the propagation of the eruptive fracture of the1983 eruption (a) and non-eruptive fracture of the 1989 eruption (b).

Figure 2. Mt Etna permanent tilt network.

Bonaccorso et al. 1995]. Until 1990 a few automatic tiltstations, which transmitted data to Catania via radio fre-quency (RF), gave the first real time information associ-ated with eruptions. In particular the 1981, 1983, and1989 eruptions caused rapid tilt changes at stationslocated near the propagated fractures [Villari, 1982;Bonaccorso et al., 1990] (Fig. 1). From the end of 1990to October 1991 the network was expanded, reaching aconfiguration of nine bi-axial, shallow bore-hole instru-ments (Fig. 2), that were RF linked to Catania for real-time monitoring. During the last decade a more densenetwork has guaranteed an adequate monitoring of Mt.Etna furnishing strong contribution to understandingthe deformation response during different volcanicphases. We focused our attention on different aspectsregarding tiltmeters:

- coherence and stability of shallow borehole electronictiltmeters;

- temperature effects on shallow borehole sensors;- sensor calibration techniques;- coseismic and aseismic tilt variations;- tilt changes related to recent volcanic phenomena and

contribution to analytic source modeling;- realization of high precision long base fluid tiltmeters.The aim of this paper is to give a picture of the different

experiences and results using tiltmeters at Mt. Etna and toindicate possible future developments and advances.

2. TILTMETERS

2.1.Instrument and installations

At present the Mt. Etna permanent tilt network comprisesnine bi-axial instruments (Fig. 2) installed in shallow bore-holes at about 3 m depth and one long baseline instrument(Fig. 3). The borehole instruments use a high precision elec-trolytic bubble sensor to measures the angular movement[Westphal et al., 1983]. Eight stations are equipped withApplied Geomechanics model 510 tiltmeters with a precisionof 0.01 µrad (DAM, MDZ, MEG, MMT) or 722 model with aprecision of 0.1 µrad. (CDV, MNR, MSC, PDN). The SPCstation is a Kinemetrics instrument with a precision of 0.025µrad. A tilt component is directed toward the crater, namedradial component, and a positive signal variation means craterup. The second component, the tangential one, is orthogonalto the radial and a positive signal variation means uplift in theanticlockwise direction. Air temperature and ground temper-atures, at different depths in the hole, are also recorded. Thecontrol datalogger (model Campbell CR10) is programmedfor 48 data/day sampling (1 sample every 30 minutes) andincludes acquisition of the two tilt components, air and

ground temperatures, and instrumental control parameters,such as power supply and dc/dc converter voltage. The dataare transmitted to Catania via radio-link. The data set collect-ed during the period 1990-2001 at Mt. Etna exhibit a fewshort-lived interruptions often due to electronic damage orpanel theft.

The stations SPC, PDN and CDV were installed between1987 and 1989 and the tilt network was expanded by the addi-tion of six other stations (DAM, MDZ, MGT, MMT, MNR,MSC) between November 1990 and October 1991. PDN, thehighest active station, suffered technical problems and severalinterruptions, and was replaced in 1997 with a high precisionlong base fluid tiltmeter. Finally, the EC10 station wasinstalled on the eastern flank at 9 m depth in 2000.

BONACCORSO ET AL 3

Figure 3. General scheme of the typical borehole tiltmeterinstallation at Mt. Etna. "Term" are temperature sensors atdifferent depths.

2.2.Coherence and temperature effects

Bubble sensor borehole tiltmeters are affected by varioussurface and temperature-related phenomena [e.g., Wood andKing, 1977; Wyatt et al., 1988; Dzurisin, 1992]. The noiseeffects are related to distortion effects of the topography andnear-surface heterogeneity, but mainly to the temperature fluc-tuations that cause thermoelastic strain. Theoretical studies onthermoelastic effects were conducted by Berger [1975] andHarrison and Herbst [1977]; examples of thermoelastic tiltexpected on varying the depth are shown in figure 4. The tem-perature variations could also affect the electrolyte bubble sen-sor producing (up to 3-5 µrad/°C) [Mortensen and Hopkins,1987; Wyatt et al., 1988]. The temperature effects on the sen-sor are instrumental noise while the ground thermoelasticeffects are real movements that can also be considered asground noise masking the small and slow deformation associ-ated with other geophysical and /or volcanic effects. The man-ufacturer does not provide a precise temperature coefficientbut suggests an approximate value of 1-2 µrad/°C [AGI, 1993].

Experiments conducted in the past on identical shallowborehole tiltmeters installed in close proximity [e.g., Wyattand Berger, 1980] showed no significant coherence betweensignals confirming that the noise, possibly due to thermiceffects, can mask the signal of tectonic interest.

The amplitude of the temperature wave traveling in theground decays exponentially while it propagates in depthand its variations are strongly attenuated in the first fewmeters. The thermoelastic strain effects decrease theoreti-cally with deeper installations, and this behavior should bevery accentuated in the first few meters (Fig. 4). We con-ducted an experiment [Bonaccorso et al., 1999a], installingthree tiltmeters in the same hole at different depths (3.5, 5,7.6 m), close to another hole where a single tiltmeter wasalready working, and recording three years of data. Themain aims of this investigation were to analyze the signalcoherence of the four instruments installed at the same placeat different vertical positions in order to study the tempera-ture effects on tilt recording and to verify how these effectsreduce with depth. The shallowest instrument (2 m) showedthat very limited instabilities could affect the upper superfi-cial layer without being detected in the deeper parts. The sig-nals of three instruments are well correlated in the long-termvariation. They showed a good coherency evidencing com-mon slow variations of 3-4 µrad/year. The signal of thedeepest instrument is the most stable and appears able to dis-criminate possible small variations (2-5 µrad), which couldoccur in a few months, from the seasonal fluctuations. The

4 CONTINUOUS TILT MONITORING

Figure 4. Theoretical maximum amplitude of the long-termthermoelastic tilt with depth as expected from the Harrison andHerbst (1977) equation, which also considers the surface slope.The fixed parameters are: Poisson ratio 0.3, thermal expansion 10-5, temperature amplitude of the thermal wave 12.5 °C. The twotheoretical curves are related to the two cases with and slope a =4° and a = 15°, respectively. The first case represents a desirablestable situation, while the second is similar to the possible realcondition in a volcano area, i.e. strong temperature fluctuations, noflat topographic surface. (Redrawn from Bonaccorso et al., 1999a).

Figure 5. Example of raw tilt signal recorded at DAM station(radial component) during 1991-2001 (a); 2 meter depth groundtemperature (b); tilt signal filtered by a linear correlation with tem-perature. The filtered tilt put in evidence a deflation trend during1991-93 eruption and an inflation phase starting from 1995-96.

amplitudes of the tilt along the same direction of the sensorsbetween 2 and 5 meters do not decay as theoretically expect-ed for pure thermoelastic tilt. This aspect implies that in thefirst few meters, where the annual temperature oscillationsare still evident, the amplitudes could not attenuate asexpected because of the presence of thermic noise in the sen-sors superimposed on the thermoelastic tilt. However sea-sonal tilt variations induced by temperature can be filteredby a linear correlation with temperature signals. Selectingproper time intervals, we found a better correlation withground temperature (i.e., the Term2 in figure 3). The filteredsignals (fig. 5) allow a better identification of long-termtrends (from several months to some years) also showed byGPS and EDM periodical surveys (Bonaccorso, 1996).

2.3. Sensor calibration

Laboratory evaluation of the exact noise caused on thesensor by temperature variations would allow a more ade-quate evaluation of procedures for filtering the false tilt pro-duced by the sensor, thereby fully exploiting the potential ofthe instrument even at shallow depths. A stable calibrationdevice should be free from eventual noise induced by tem-perature and surrounding movements in order to check highprecision tiltmeters and obtain a correct calibration of theseinstruments with precision of less than 0.1 µm/m. Most cal-ibration platforms use adjustment systems based onmicrometers with good mechanical characteristics, but witha resolution that is at least an order of magnitude poorer.Furthermore, the fundamental problems with usual calibra-tion platforms are: i) during the calibration time they cannotdistinguish undesirable external movements, i.e. cannot dis-criminate possible fictitious effects, induced by the sur-rounding environment, which can affect the real tilt inducedby the real adjustment of the platform; ii) they are alsounable to discriminate how much the temperature variationsaffect the sensor as noise or provoke possible undesiredmovements of the plate. In order to eliminate the abovedrawbacks, a patented calibration device has been devisedand realized that use no of mechanical apparatus to producethe inclination [Bonaccorso et al., 1999b]. It is based on thedifferent inclination that a floating object assumes as itsbaricenter varies. The device was built with a highly rigidplanar base body (granite) set to float on the mercury placedinside a tank. The base body contains a pair of micrometers,positioned near two orthogonal external borders, able totranslate a known weight in order to vary the baricenter and

BONACCORSO ET AL 5

Figure 6. Example of temperature noise detection on tilt sensorsusing the calibration device patented by Bonaccorso et al., (1999).(a) Temperature variation, and (b) - (c) "false" tilt output recordedat the two orthogonal components X and Y of an electrolytic sen-sor used in borehole type instrument (AGI, serial n. 1946). Thesampling rate is 1 data / minute. The resemblance of the "false" tiltto the temperature is well evident. In this case the calculated tem-perature coefficients, by using a linear fit, were 0.7 and -1.7 mrad/ °C for the X and Y tilt components, respectively. The calibrationdevice is essentially composed of a granite base body, floating ona mercury pool, which can be tilted on the mercury with high pre-cision by adjusting a pair of symmetric micrometers. The devicehas the advantages of being insensitive both to variations in thesurrounding environment and temperature, and it always ensures aperfectly horizontal position even if temperature changes. Thus,by provoking forced temperature variations, one can discriminatetemperature noise on the sensor by the correlation of real temper-ature variations with false tilt output induced by temperature vari-ation in the tiltmeter, as shown in this example.

inclination on the mercury of the base body by a precisevalue determined by the variation of the applied moment.Using, as in our platform, micrometers with an instrumentalerror of about 1 notch (10 µm), then it follows that it is the-oretically possible to appreciate variations of about 0.002µrad [Bonaccorso et al., 1999b].

We tested several instruments with electrolytic bubblesensor and verified that the sensor temperature coefficientsare different for each tilt component analyzed, both in mag-nitude and sign, and are comprised within a range of about± (1 to 4) µrad /°C. In figure 6 the results of a test made ona very good and stable sensor 722 model borehole type isshown. We can observe the perfect resemblance of the tiltsignal, which in our device is forced to remain without vari-ations, with the temperature. In this case the temperaturecoefficient is 0.77 and -1.7 µrad /°C for the X and Y orthog-onal tilt axes, respectively.

The checks performed with this plate on different tilt-meters with electrolytic sensors highlight how noiseinduced by temperature variations, measurable with goodprecision with this device, represents the primary source ofnoise for this category of instrument. For shallow installa-tion (2-5 m depths) the instruments coefficient should nec-essarily be evaluated before positioning the instrument. Toavoid temperature noise on the sensor the installationsshould be made at depths greater than 10 m, where the tem-

perature is stable, thus removing the problem of the falsesensor output induced by temperature variations.

3. TILT ASSOCIATED WITH RECENT ERUPTIONSAND ITS CONTRIBUTION TO MODELING

Tilt recording plays a fundamental role in real time moni-toring because it provides a signal able to detect grounddeformation changes with high precision. At Etna, tiltmeterdetected slow inflation/deflation phases but they excelled atdetection of rapid and sharp variation such as the onesrecorded during intrusions and eruptive fissure propagation.The tilt data proved useful for the control of the phenomenaevolution as well as for understanding the magma sourcemechanisms and transport process. The data were used inmodeling in combination with other types of deformationdata. Below, several examples of eruptive phenomena andassociated tilt are reported and described.

The first lateral eruption monitored by a tiltmeter was the1981 event. The long-term trend, characterized by a yearlyseasonal cycle, was interrupted by 7-8 µrad inflation whichlasted 2-3 months and preceded the NNW propagating finaleruptive fracture (Fig. 7). The eruption onset was accompa-nied by a sharp deflation associated with a summit area low-ering [Murray, 1982; Villari, 1983]. The geometry and open-ing of the deep vertical dike, which crossed the volcano pilebefore opening the final shallower eruptive fissure, wasmodeled using EDM and leveling data collected about oneyear before and after the eruption [Bonaccorso, 1999]. Theexpected tilt of this source is coherent with the tilt recordedduring the 2-3 months before the eruption. Therefore, the tilt,besides detecting the eruption onset, also was able to suggestthe time action of the vertical dike formation.

The 1989 eruption was accompanied by a 7 km long frac-ture which propagated from the crater area along the SSEdirection without magma output. Tilt variations were evi-denced by the pendulum tiltmeters, and in particular amarked tilt variation was recorded during the fracture prop-agation at a station located only 300 m to the east of the finalpart of the fracture field [Briole et al., 1990]. The shallowborehole continuous tilt station SPC, located in the southernflank, recorded a sharp signal variation during the final partof the fracture (Fig. 8). The data sent via RF to Cataniaallowed a real time evaluation of the process. The tilt signalwas used together with the EDM horizontal changes, record-ed across the fracture in its final part, to infer the geometryof the source, which was modeled and interpreted as a later-al dike propagation [Bonaccorso and Davis, 1993].

During the 1991-93 eruption the tilt of the full workingnetwork provided information both during the opening oferuptive fissure and during the eruption course. In fact, for

6 CONTINUOUS TILT MONITORING

Figure 7. Radial tilt recorded at INT bore-hole station from May1978 (48 data per day) with the interpretative scheme: a) long termdeflation with a seasonal behavior probably due to the temperaturenoise evidenced by the grey tendency line overplotted; b) inter-ruption of the long term deflation and small slow term inflation 3-4 months before the eruption; c) eruptive fissure opening.(Redrawn after Bonaccorso, 1999).

this eruption the tilt was able to reveal both the modest vari-ation produced by the final shallow lateral fissure feedingthe eruption and the slow and long term deflation accompa-nying the eruption (Fig. 9). The tilt was used in an inversionin a dynamic model which took into consideration the twosources acting in time, i.e. the inversion took into consider-ation the sequence of the events and constrained the sources(intrusion, fracture, deflation) by the chronology of the data[Bonaccorso, 1996]. The tilt significantly contributed toconstraining the source and making the model solution veryrobust.

During the 2001 eruption the tilt signals provided a furtherhelpful contribution to understanding eruption mechanisms.The three tiltmeters closest to the fracture field and locatedaround it (MDZ on the SSW flank, CDV on the southernflank, PDN on the summit area) recorded strong and sharpvariations during 13-14 July before the onset of the eruptionon 18 July (Fig. 10). The tilt signals helped to define the areawhere the dike was intruding. Moreover, the tilt changerecorded before the eruption (12-14 July) was used togetherwith the length variations from the GPS permanent networkstations to model the emplacement geometry of the verticaldike which intruded across the volcano pile [Bonaccorso etal., 2002].

4. SEISMIC AND ASEISMIC IMPULSIVE TILT

In the last twelve years (1990-2001), 25 discrete varia-tions of continuous tilt signal have been recorded on MountEtna, among which only two episodes were caused by theopening of the eruptive fractures (1991, 2001). The remain-ing 23 anomalies can be classified into two categories: thefirst comprises 7 “instantaneous” coseismic tilt variationsrecorded in concomitantly to strong local earthquakes(Ml≥3.3) characterized by amplitude generally not higherthan a few µrad ; the second consists of transient anomaliesranging from some hours to 1-2 days, observed with differ-ent arrival times at the various tilt stations and characterizedby variations of several µrad (from 3-4 to 30-40 µrad) (Fig.11) without any correlation to seismic events, i.e. aseismicvariations, or to other evident volcanic episodes.

BONACCORSO ET AL 7

Figure 8. Tilt recorded at SPC station during the final part (last 2km) of the 7 km long fracture of September 27 - October 3, 1989propagating from the summit craters area downwards along theSSE direction. The tilt variation was recorded during the last 2 kmof the fracture. The signal was used by Bonaccorso and Davis(1993), in combination with EDM changes, for inferring and mod-eling a shallow intruding lateral dike.

Figure 9. Radial tilt components recorded during 1992-93 from 1January 1992. The data have been filtered to remove the seasonaltemperature effects. The tilt clearly highlights the deflation accom-panying the volcano edifice during the eruption course.

8 CONTINUOUS TILT MONITORING

Figure 10. (a) Tilt recorded on the southern flank at PDN long-base mercury tiltmeter station (a) and at the shallow borehole sta-tions MDZ (b) and CDV (c) during the 2001 intrusion. CDV andMDZ stations are located close to the eruptive fissure field andabove two shallow clustered zones of the seismic swarm thatoccurred during the dike intrusion. The signals show that thedeformation caused by the dike emplacement five days before theeruption onset.

Figure 11. (a) Radial components of an aseismic tilt anomalyobserved at different stations. The arrival times of the tangentialcomponents are coincident with the radial ones. Unit scale is 4mrad for MMT, SPC and DAM, unit scale 20 mrad for the othercomponents. (b) Etna map with the epicenters calculated for themain aseismic tilt episodes.

4.1.Coseismic variations

Impulsive tilt variations correlated with local earthquakesare characterized by simultaneous recordings at differentstations and irregular values ranging between 1 and 10 µrad.

Mt. Etna offsets are generally correlated to seismic eventslocated on the high western sector of the volcano and con-

fined to a depth not greater than 7 km. The deeper seismicevents and the seismic events located in other sectors pro-duced no observed deformation. Coseismic tilt amplitudedecays with the inverse square of distance from the earth-quake source [Aki and Richards, 1980], so tilt stations farfrom earthquake foci will not detect a significant signal.

We observed that coseismic changes on short-base-line

BONACCORSO ET AL 9

Figure 12. Pizzi Deneri observatory (PDN) at Mt. Etna in the high north-eastern flank, with a scheme of the under-ground tunnel in which the monitoring instrumentation is installed, and a scheme with the specifications of the lasersensor used to detect the mercury level vertical changes.

Mt. Etna instrument may have large amplitudes and differ-ent directions compared to theoretical ones calculated byusing dislocation fault models [Bonaccorso et al. 1996].Discrepancies between coseismic tilt variation and predict-ed surface displacement [Press, 1965; King et al., 1977;Allen, 1978] have been noted for other earthquakes both atlocal and teleseismic distances and may be due to move-ments on cracks, fractures, and minor faults near the instru-ment site [McHugh and Johnston, 1977], instability ofinstruments [Takemoto, 1995] or to acceleration, all ofwhich influence the shallow tilt sensors [Wyatt, 1988].

4.2. Aseismic variations

Between 1990 and 2001, the Mt. Etna tilt network record-ed a total of 16 variations not directly correlated to seismicactivity or evident volcanic episodes. Propagation of crustaldeformation without dominant seismicity (slow-movingdeformation) has been observed by tiltmeters in tectonicand volcanic areas. Different mechanisms have been

involved: tilt recorded before earthquakes at San AndreasFault [Mortensen and Johnston, 1976] and CentralApennines (Italy) [Bella et al. 1986], intrusion event at Itovolcano [Okada and Yamamoto, 1991], lateral dike propa-gation at Stromboli [Bonaccorso, 1998].

The anomalies recorded on Mt. Etna have common char-acteristics: a mean duration of 10-20 hours, considerableamplitude variation in most cases, arrival times not coinci-dent at the different stations and similar signal patterns. Thetilt anomaly delay observed at different stations is caused bythe location of the station with respect to the source anddepends on the propagation velocity of crustal deformation.Using measurements of the delays recorded and assuming apoint-source deformation buried in a constant velocity half-space, an analysis procedure based on the Lee and Valdes[1985] localization program has been developed to deter-mine velocity and source localization [Gambino, 2002]. Apropagation velocity ranging between 4.5 and 6 km/day anda common source area located 3-6 km SSW from summit(error 0.7-1.7 Km) at a depth of 5-10 Km (error 3.0-5.7 Km)have been estimated for the four aseismic events character-ized by large variations and clear first arrivals recorded byat least six stations (Fig. 11). Etna aseismic tilt variationshave preliminarily been interpreted as the result of the prop-agating tensile stress induced by rising magma [Bonaccorsoand Gambino, 1997; Gambino, 2002]. It is noteworthy howthe location area of aseismic variations has been affected,during the 1990-2001 period, by strong seismic releases at adepth of 7-12 km. The relation between the source of thisaseismic tilt (low velocity stress discharge) and the seismic-ity (fast velocity stress discharge) is still an open point to bebetter investigated.

5. NEW TECHNOLOGY FOR HIGH PRECISIONTILTMETERS

Long-base devices should be preferable for grounddeformation monitoring because of their capacity to meas-ure a broad deformation field rather than possible localinstabilities or noise [Horsfall and King, 1978; Agnew,1986]. In addition, in the volcanic environment it is impor-tant to use reliable and high precision instrumentation. Wedevised a new long-base instrument and installed it at thevolcanological observatory of Pizzi Deneri (PDN), whichis located 2850 m a.s.l. on the north-eastern flank of MountEtna volcano (3340 m a.s.l.), about 2 km distant from thesummit craters (Fig. 12). The observatory has two 80 mlong artificial underground orthogonal tunnels where thetiltmeter arms were positioned The three points (the centralcommon and the two extremities) are disengaged each withrespect to the other and free to move independently. A tube

10 CONTINUOUS TILT MONITORING

Figure 13. Tilt radial components at bore-hole MSC and long-basePDN stations. A clear variation of the radial component (tilt up) isevident during the initial part of the January 9-14, 1998 seismicswarm as indicated by the dark area. This tilt was used byBonaccorso and Patanè (2001) for modeling a tensile dislocationincreasing upward in time interpreted as a shallow dike emplacement.

filled with mercury, positioned along the tunnels, has beenutilized to measure the tilt. At the two extremities and in thecentral part, the tube is connected to three beakers. An opti-cal laser sensor, fixed in the top of each beaker, emits radi-ation that is reflected by a target inside the beaker floatingon the mercury. This kind of sensor (see inset of figure 12for characteristics) has a resolution of the order of µm,which in the tube (80 m long) corresponds to tilt variationsin the order of 0.01 - 0.05 µrad [Bonaccorso et al., 1998].

The mercury was chosen as fluid due its properties suchas the low freezing point (ca. –30 °C), low evaporationbetween –10 and 10 °C , high density, low and constantthermal expansion.

The tunnels are not an absolute requirement for this kindof instrument and the mercury filled tube could be alsopositioned at very shallow depth below the ground surface.

Several tests and calibrations were performed on theinstrumentation to verify this sensitivity [Bonaccorso et al.,1998, Bonaccorso et al., 1999b]. One of the major advan-tages of this long-base tilt is that it is not affected by tem-perature changes. The instrumentation was able to detectco-seismic (i.e., main volcano earthquakes and seismicswarms), eruptive (i.e., intrusion preceding the eruptions);and co-explosive events (i.e., summit crater paroxysticphases). The seismic swarm of January 1998 was well stud-ied through the combination of seismic and tilt data[Bonaccorso and Patanè, 2001]. In particular, the small tiltvariations recorded at two summit stations (MSC and PDN)(Fig. 13) allowed us to infer a shallow intrusive episodeinterpreted as the dike emplacement which then fed theFebruary-November 1999 eruption. The sensitivity of thetiltmeter recorded precursor tilt before the onset of both the

BONACCORSO ET AL 11

Figure 14. Four years of tilt recorded at the mercury long-base instrument at Pizzi Deneri observatory. The tilt anom-alies preceding the February 1999 summit eruption and the July 2001 lateral eruption have been evidentiated. In theinset a zoom of the July 1998 period shows the ability of the instrumentation to detect a 0.1 - 0.15 mrad co-explosivechange recorded in coincidence with the lava fountain from SE crater on June 24, 2000. During this paroxystic phasethe acquisition sampling rate was 1 sample/minute.

1999 summit eruption and the 2001 lateral eruptions (Fig.14). The 1999 summit eruption was preceded by very shal-low seismicity, and the long-base tiltmeter recorded avery small but clear inflation precursor. In the case of the2001 lateral eruption, the 13-15 July 2001 tilt variation isthe clearest signal accompanying the emplacement of adike during the days before the eruption onset and in pro-viding helpful data for source mechanism modeling[Bonaccorso et al., 2002].

Finally, in order to fully show the potential of thisinstrument, we give an example of a co-explosive tiltrecorded during the paroxystic phase (lava fountainevent) at the SE crater on July 22, 1998 (Fig. 14). The SEcrater had several lava fountain events during 1996, 1997and 1998 [La Delfa et al., 2001]. During 1998 the lavafountains lasted only 10-30 minutes (high pressure dis-charge), and in particular during summer 1998, due to thegood availability of energy from the solar panels, we didsome tests with sampling rate acquisition of 1data/minute. Under these conditions, the instrumentationrecorded even the co-eruptive event showing a clear 0.1 –0.15 µrad deflation unlikely detectable with other instru-mentation.

6. CONCLUSIONS

Tilt is a powerful tool for Mt. Etna monitoring and sur-veillance; tilt is a “simple” signal that doesn’t need treat-ments and is of immediate understanding.

This paper summarizes the activities conducted byINGV-CT in using tiltmeters for real time continuousmonitoring at Mt. Etna during the last twenty years.

The improvements and the developments made in usingthis technology are described.

Tilt changes have accompanied the main eruptive phas-es during this time interval, and the continuous tilt moni-toring has allowed us to identify eruption precursors gen-erally related to rapid rise of magma and formation ofdikes and eruptive fissures. Tilt data also provided animportant contribution to the analytical modeling of thevolcano sources acting during the several eruptionsoccurred at Mt. Etna volcano.

Therefore, we highlighted how continuous tilt monitor-ing plays a fundamental role for surveillance purposes,and how it also represents a powerful tool to infer the vol-canic phenomena in time and space, especially during thefinal stages of an intrusion.

Acknowledgements. We thank M. Lisowski and D. Dzurisinfor their constructive reviews. We are indebted with L. Villariwho promoted the tilt networks and encouraged our work.

REFERENCES

AGI - Applied Geomechanics Incorporated, Notes on tempera-ture coefficients and long-term drift of Applied Geomechanicstiltmeters, Report n. A-93-1001, AGI, Santa Cruz, California,1993.

Aki, K., and P.G. Richards, Quantitative seismology, pp. 799-849, W. H. Freeman, San Francisco, 1980.

Allen, R. V., Tiltmeter observations near a large earthquake, Bull.Seism. Soc. Am. 68, 855-857, 1978.

Agnew, D.C., Strainmeters and tiltmeters. Rev. Geophys. 24, 579-624, 1986.

Bella, F., R. Bella, A. Ermini, V. Sgrigna, and P.F. Biagi, Possibleprecursory tilts preceding some earthquakes (3.0≤ M ≤ 3.8)occurred in Central Italy between February 1981 and June1983, Earthq. Pred. Res. 4, 147-154, 1986.

Berger, J., A note on thermoelastic strains and tilts, J. Geophys.Res., 80, 274-277, 1975.

Bonaccorso, A., Dynamic inversion of ground deformation datafor modelling volcanic sources (Etna 1991-93), Geophys. Res.Lett., 23, 5, 451-454, 1996.

Bonaccorso, A., Evidence of a dyke-sheet intrusion at Strombolivolcano inferred through continuous tilt, Geophys. Res. Lett.,25, 22, 4225-4228, 1998.

Bonaccorso, A., The March 1981 Mt. Etna eruption inferredthrough ground deformation modeling, Phys. Earth Plan. Sc.112, 125-136, 1999.

Bonaccorso, A., and P. Davis, Dislocation modelling of the 1989dike intrusion into the flank of Mount Etna, Sicily, J. Geophys.Res., 98, 4261-4268, 1993.

Bonaccorso, A., and S. Gambino, Impulsive tilt variations onMount Etna volcano (1990-1993), Tectonoph., 270, 115-125,1997.

Bonaccorso, A., and D. Patanè, Shear response to an intrusiveepisode at Mt. Etna volcano (January 1998) inferred throughseismic and tilt data, Tectonoph., 334, 61-75, 2001.

Bonaccorso, A., O. Campisi, G. Falzone, B. Puglisi, R. VelarditaR., and L. Villari, Ground deformation: geodimeter trilatera-tion and borehole tiltmetry, in “Mt. Etna 1989 eruption”,Barberi F., Bertagnini A. & Landi P. (Eds). CNR-GNV SpecialIssue, 44-47, 1990.

Bonaccorso, A., O. Campisi, G. Falzone, S. Gambino, G.Laudani, B. Puglisi, M. Rossi, R. Velardita, and L. Villari,Ground deformation on active volcanoes in Sicily (Italy),Proceedings Workshop “New challenges for geodesy involcanoes monitoring”, Walferdange, Luxemburg, 14-16 June1993, in Cahiers du Centre Europeen de Geodynamique et deSeismologie, v.8, 349-358, 1995.

Bonaccorso, A., F. Ferrucci, D. Patanè, and L.Villari, Fast defor-mation processes and eruptive activity at Mt. Etna (Italy), J.Geophys. Res., 101, 17467-17480, 1996.

Bonaccorso, A., G. Falzone, G. Raia, and L. Villari, Applicationof new technology for ground deformation continuous detec-tion, Acta Vulcanologica, 10 (1), 7-12, 1998.

Bonaccorso, A., G., Falzone, and S., Gambino, An investigation

12 CONTINUOUS TILT MONITORING

BONACCORSO ET AL 13

into shallow borehole tiltmeters, Geophys. Res. Lett., 26, 11,1637-1649, 1999a.

Bonaccorso, A., G. Falzone, and G. Raia, Device for checkingand calibrating high precision inclinometric sensors, PCT(Patent Corporation Treaty) international application, ref. n.PCT/IT99/00188, 1999b.

Bonaccorso, A., M. Aloisi, and M. Mattia, Dike emplacementforerunning the Etna July 2001 eruption modeled through con-tinuous tilt and GPS data, Geophys. Res. Lett., 29, 13,10.1029/2001GL014397, 2002.

Briole, P., G. Nunnari, and G. Puglisi, Ground deformation: b)GPS and IPGP tilt data, in “Mt. Etna: the 1989 eruption”,Barberi F., Bertagnini A., Landi P. eds., CNR GNV SpecialIssue, Giardini, Pisa, 47-52, 1990.

Briole, P., R. Gaulon, G. Nunnari, G. Puglisi, and J. C. Ruegg,Measurements of ground movement on Mount Etna, Sicily: asystematic plan to record different temporal and spatial com-ponents of ground movement associated with active volcan-ism. IAVCEI Proc. Volc., P. Gasparini, R. Scarpa, K. Aki eds.,Springer-Verlag, Berlin, 3, 120-129, 1992.

Davis, P.M., Gravity and tilt earth tides measured on an activevolcano, Mt. Etna, Sicily, J. Volc. Geoth. Res., 11, 213-223,1981.

Dvorak, J., A. Okamura, C.E. Mortensen, and M.J. Johnston,Summary of electronic tilt studies at Mount. St. Helens, inGeological Survey Professional Paper.1250. The 1980 erup-tion of Mt. St Helens, Washington, P.W. Lipman & D. R.Mullineaux (eds), 169-174. Washington DC. United StatesGovernament Printing Office, 1981.

Dzurisin, D., J.A. Westphal, and D.J. Johnson, Eruption predic-tion aided by electronic tiltmeter data at Mount St. Helens,Science, 221, 1381-1383, 1983.

Dzurisin, D., Electronic tiltmeters for volcano monitoring : les-sons from Mount St. Helens, in Monitoring volcanoes: tech-niques and strategies by the staff of the Cascades observatory,1980-90, Ewert, J.W., and D.A. Swanson, Eds, U.S.G.S. Book,69-83, 1992.

Gambino, S., Coseismic and aseismic tilt variations on MountEtna, Pure Appl. Geophys., 159, 2751-2762, 2002.

Harrison, J.C., and K. Herbst, Thermoelastic strains and tiltsrevised, Geoph. Res. Lett., 4/11, 535, -537, 1977.

Henbest, S.N., A.A. Mills, and P. Ottey, Two tiltmeters and anintegrating seismometer for the monitoring of volcanic activi-ty, and results of some trials on Mount Etna, J. Volc. Geoth.Res., 4, 133-149, 1980.

Horsfall, J.A.C., and G.C.P. King, A new geophysical tiltmeter,Nature, 274, 675-676, 1978.

King, C.Y., R. D. Nason, and R. O. Burford, Coseismic stepsrecorded on creep meters along the San Andreas fault, J.Geophys. Res., 82, 1655-1662, 1977.

La Delfa, S., G. Patanè, R. Clocchiatti, J.L. Joron, and J.C.Tanguy, Activity preceding the February 1999 fissure eruption:inferred mechanism from seismological and geochemical data,J. Volc. Geoth. Res., 105, 121-139, 2001.

Lee, W. H. K., and C. M. Valdes, HYPO71PC: a Personal com-

puter version of the HYPO71 earthquake location program, U.S. Geol. Surv. Open File Report 85-749, 43 pp., 1985.

McHugh, S., and M.J.S., Johnston, An analysis of coseismic tiltchanges from an array in Central California, J. Geophys. Res.,82, 5692-5698, 1977.

Mellors, R., J.L Chatelain, B.L. Isacks, G. Hade, M. Bevis, andM. Prevot, A tilt and seismicity episode in the New Hebrides(Vanuatu) Island Arc, J. Geophys. Res., 96, 16535-16546,1991.

Mortensen, C.E., and D.G. Hopkins, Tiltmeter measurements inLong Valley caldera, California, J. Geophys. Res., 92, 13767-13776, 1987.

Murray, J.B., Les deformations de l’ Etna a la suite de l’ eruptionde Mars 1981. Bulletin P.I.R.P.S.E.V., n. 57, 1-30, I.P.G. –France, 1982.

Murray, J.B., and Guest, J.E., Vertical ground deformation onMount Etna 1975-1980, Geological Societies of America Bull.,93, 1160-1175, 1982.

Nunnari, G., and G. Puglisi, The Global Positioning System asuseful technique for measuring ground deformation in vol-canic areas. J. Volcanol. Geoth. Res., 61, 267-280, 1994.

Rebscher, D., M. Westerhaus, W. Welle, and I.G.M.A. Nandaka,Monitoring Ground Deformation at the Decade VolcanoGunung Merapi, Indonesia. Phys. Chem. Earth, Part A. 25, (9-11), 755-757, 2000.

Okada, Y., and E. Yamamoto, Dyke intrusion model for the 1989seismovolcanic activity off Ito, Central Japan, J. Geophys.Res., 96, 10361-10376, 1991.

Okamura, A.T., J.J. Dvorak, R.Y. Koyanagi, and W. R. Tanigawa,Surface deformation during dike propagation, U.S. GeologicalSurvey Professional Paper, 1463, 165-181, 1988.

Press, F., Displacements, strains and tilts at teleseismic distances,J. Geophys. Res, 70, 2395-2412, 1965.

Takemoto, S., Recent results obtained from continuous monitor-ing of crustal deformation, J. Phys Earth, 43, 407-420, 1995.

Toutain, J.P., P. Bachèlery, J.L. Blum, J.L. Cheminèe, H.Delorme, L. Fontaine, P. Kowalski, and P. Taochy, Real-timemonitoring of vertical ground deformations during eruptions atPiton de. la Fournaise, Geophys. Res. Lett., 19, 553-556, 1992.

Villari, L. (Editor), 1981 Etna Report, Open file report 2/83,I.I.V.-C.N.R., Catania-Italy, 1983.

Villari, L., L’ Etna, Le Scienze, 168, 1982.Voight, B, R.P. Hoblitt, A.B. Clarke, A.B. Lockhart, A.D. Miller,

L. Lynch, and J. McMahon, Remarkable cyclic ground defor-mation monitored in real-time on Montserrat, and its use ineruption forecasting, Geophys. Res. Lett., 25, 3405-3408,1998.

Wadge, G., Deformation of Mt. Etna, 1971-1974, J. Volc. Geoth.Res., 1, 237-263, 1976.

Wadge, G., J.A.C. Horsfall, and J.L. Brander, Tilt and strain mon-itoring of the 1974 eruption of Mt. Etna, Nature, 254, 21-23,1975.

Westphal, J.A., M.A. Carr, W.F. Miller, and D. Dzurisin,Expendable bubble tiltmeter for geophysical monitoring,Review of Scientific Instruments, 415-418, 1983.

Wood, M.D., and N.E. King, Relations between earthquakes,weather, and soil tilt, Science, 197, 154-156, 1977.

Wyatt, F.K., and J. Berger, Investigations of tilt measurementsusing shallow borehole tiltmeters, J. Geophys. Res., 85, 4351-4362, 1980.

Wyatt, F., S. Morrissey, and D.C. Agnew, Shallow borehole tilt :a reprise, J. Geophys. Res., 93, 9197-9201, 1988.

Wyatt, F., Measurements of coseismic deformation in SouthernCalifornia: 1972-1982, J. Geophys. Res., 93, 7923-7942, 1988.

Alessandro Bonaccorso, Orazio Campisi, Giuseppe Falzone andSalvatore Gambino, Istituto Nazionale Geofisica e Vulcanologia(INGV), Piazza Roma, 2, 95123, Catania, Italy.

14 CONTINUOUS TILT MONITORING