Pure and Applied Geophysics

, Volume 175, Issue 5, pp 1765–1781 | Cite as

Cansiglio Karst Plateau: 10 Years of Geodetic–Hydrological Observations in Seismically Active Northeast Italy

  • Barbara Grillo
  • Carla Braitenberg
  • Ildikó Nagy
  • Roberto Devoti
  • David Zuliani
  • Paolo Fabris


Ten years’ geodetic observations (2006–2016) in a natural cave of the Cansiglio Plateau (Bus de la Genziana), a limestone karstic area in northeastern Italy, are discussed. The area is of medium–high seismic risk: a strong earthquake in 1936 below the plateau (Mm = 6.2) and the 1976 disastrous Friuli earthquake (Mm = 6.5) are recent events. At the foothills of the karstic massif, three springs emerge, with average flow from 5 to 10 m3/s, and which are the sources of a river. The tiltmeter station is set in a natural cavity that is part of a karstic system. From March 2013, a multiparametric logger (temperature, stage, electrical conductivity) was installed in the siphon at the bottom of the cave to discover the underground hydrodynamics. The tilt records include signals induced by hydrologic and tectonic effects. The tiltmeter signals have a clear correlation to the rainfall, the discharge series of the river and the data recorded by multiparametric loggers. Additionally, the data of a permanent GPS station located on the southern slopes of the Cansiglio Massif (CANV) show also a clear correspondence with the river level. The fast water infiltration into the epikarst, closely related to daily rainfall, is distinguished in the tilt records from the characteristic time evolution of the karstic springs, which have an impulsive level increase with successive exponential decay. It demonstrates the usefulness of geodetic measurements to reveal the hydrological response of the karst. One outcome of the work is that the tiltmeters can be used as proxies for the presence of flow channels and the pressure that builds up due to the water flow. With 10 years of data, a new multidisciplinary frontier was opened between the geodetic studies and the karstic hydrogeology to obtain a more complete geologic description of the karst plateau.


Geodesy karst hydrogeology GPS tilt measurements 

1 Introduction

The study region is located in northeastern Italy, in the seismically active area of the karstic Cansiglio Plateau. The present seismicity of NE Italy is well manifested toward the Friuli region, whereas toward the western sector a relative calmness is found. This picture emerges when considering the local seismicity recorded since the 1976 disastrous Friuli earthquake, certainly biased by the post-seismic sequence of this event. The western sector was hit in 1936 by the destructive Cansiglio earthquake, showing that the seismic potential is high in the entire region, reaching also farther west to the eastern Venetian sector. For this reason, 10 years ago it was decided to monitor the deformation of the area, installing two Zöllner-type Marussi tiltmeters in a natural cavity at 25 m depth (Bus de la Genziana). They are operating continuously since 2005 (Braitenberg et al. 2007).

During this period, we proposed an interdisciplinary study of karstic aquifers using hydrogeological data, tiltmeters and GPS observations (Grillo et al. 2011). During the year 2010, two data acquisition campaigns have been carried out to integrate the research started with the tiltmeter recordings: hydrogeological flow measurements (level, conductivity and temperature) in two principal springs and the installation of a small geodetic GPS network. The geodetic campaign extended from May to October 2010 measuring two GPS benchmarks in the neighborhood of the FReDNet permanent station CANV (950 m a.s.l; Zuliani 2003; Zuliani et al. 2009), the results of which demonstrated the presence of an observable hydrologic signal in the GPS time series (Devoti et al. 2015).

To monitor the underground hydrodynamics of Bus de la Genziana and to correlate it with the geodetic recordings, for the first time a multiparametric logger (temperature, stage, electrical conductivity) was installed in the siphon located at the bottom of the Genziana cave (587 m deep) from March 2013 to December 2016 (Grillo and Braitenberg 2015).

The decade-long tilt and GPS observations in the karstic area allows us to univocally characterize the expected deformation signal in relation to the rainfall and spring discharge from the karstic plateau. The results can be of relevance for other karstic areas worldwide (see, e.g., Longuevergne et al. 2009; Tenze et al. 2012; Braitenberg 1999b, c), because hydrologic karst systems have often common characteristics, such as prominent epikarst, a well-developed network of deep channels in which water is efficiently drained toward the base level, and a less important matrix-flow component. The knowledge of the geometry of the cave network is important, because boreholes to the channels produce drinking water provision. An improved knowledge of the relation between the karstic water flows and the geodetic signal will allow the use of the geodetic measurement as a hydrologic investigation tool in the future. The main features of the geodetic measurements are furthermore generally comparable to the signals of analogous instrumentation operating in non-karstic environs affected by fissures and faults. The understanding of the physical relations between extensometer, tilt and gravity observations and the hydrology is fulfilled also elsewhere, e.g., in the station Moxa, Germany (Jahr 2017). Here, induced pore pressure variations through pumping and injection experiments were made to test the poroelastic deformation models. The aim is to use the physical relations to plan hydro-deformation experiments to recover in situ rock physical properties (Wang and Kümpel 2003; Jahr et al. 2008).

2 Geographical and Geological Setting

The Cansiglio–Cavallo Plateau is a karstic massif situated in the Forealps (Prealpi Carniche) (Fig. 1), which stretches forward as a mountainous block toward the plain. It is divided between two regions, Veneto on the west and Friuli-Venezia-Giulia on the east and three provinces Pordenone, Treviso and Belluno (towns outside the orientation map of Fig. 1). Its maximum height above mean sea level is 2200 m, and it has two plateaus of medium height of 1000 m, the Cansiglio and the Piancavallo Plateau.
Fig. 1

Localization of the study area. Red triangles: location of GPS stations; white arrows show horizontal movement of GPS after strong rainfall (details in Devoti et al. 2015). Genziana cave: tiltmeter station. Green dots: the locations of the three karstic springs forming the Livenza River. Red and blue: dominant regional thrusts. Also shown is the geological setting of the study area with the principal structural lineaments (Design by A. Riva). Surface projection of the Genziana cave sketched in black

A geological description of Cansiglio Plateau is discussed in Cancian and Ghetti (1989). The outcropping rocks range in age from Upper Jurassic to Paleocene and are mainly composed of carbonates (Fig. 1). The eastern area is characterized by a thick succession of Cretaceous peritidal carbonates (Cellina limestones), while the central western part is characterized by slope breccia deposits (Fadalto Formation and Mount Cavallo Formation), all capped by basinal marly carbonates (Scaglia Formation).

Notoriously, the zone is of medium–high seismic risk: in recent history, we recall the strong earthquake of 1936, which according to the magnitude scale used was quantified as having Ms = 5.8 (magnitude of surface wave) or Mm = 6.2 (macroseismic magnitude) (Pettenati and Sirovich 2003). 65 km to the northeast the Friuli earthquake hit on 6 May 1976 with M = 6.4, with relevant aftershocks (Pondrelli et al. 2001). The present local seismicity (OGS-RSC Working Group 2012) is shown in Fig. 2. The seismicity is monitored with a local network due to the presence of a natural gas storage concession. Details on the relation between gas pumping and seismicity can be found in Priolo et al. (2015). The magnitudes range from − 1.8 to 4.5, with average magnitude of 0.6 in the period 2012–2017. Clockwise from upper left, the figure displays the seismicity for the period 1 January 2012–31 March 2017. Then a depth profile along the Caneva line and two parallel profiles orthogonal to the thrust are traced. On the seismicity map, the gray star shows the position of the Genziana station, the smaller black dots show locations of local towns, and the three red dots show the three springs marked in Fig. 1. A distinct fault plane (brown line overlain on seismicity depth plot in Fig. 2c) is delineated by the seismicity in the southern profile. In the northern profile, this fault is no longer seen (Fig. 2). The Cansiglio Plateau has very few seismic events, displaying a seismic gap, which could be interpreted as due to the fact that it is a rigid block with little seismicity. The plateau is bounded at the southwest and southeast by regional thrust faults (Cavallin 1980). The Caneva Line thrust faulting resulted in a relatively wide asymmetric anticline (Cansiglio Anticline) in the hanging wall, with steep dipping beds near the main fault. The main thrust plane is also associated with minor faults developing a quite wide cataclastic zone, about 500 m in width. The main karst springs emerge in the lower limb of the anticline, where the Mesozoic limestones (Upper Jurassic Polcenigo limestone and Cretaceous Cellina limestones) are in tectonic contact with the Cenozoic and Quaternary impermeable units of the footwall.
Fig. 2

Present local seismicity (OGS-RSC Working Group 2012) for the period 1 January 2012 to 31 March 2017. a Epicenters with colors representing depth; following three graphs. bd Seismic events along the different profiles, with depth distribution for the profile shown below the map. The distance along the section extends from arrowtail to arrowhead of the profile

The River Livenza flows down from the southeastern slope of the carbonatic Massif of Cansiglio–Cavallo. It is supplied by three main springs: the Gorgazzo, which has a recharge basin of 170 km2, the Santissima of 500 km2 and the Molinetto of 230 km2. All three have an average flow from 5 to 10 m3/s (Cucchi et al. 1999).

The massif is characterized by a markedly deep karst, with about 200 caves and clear karst surface morphology. Although the annual mean precipitation is about 1800 mm, the Cansiglio Plateau for the time being has no surface runoff, but acts like an endorheic basin with a pronounced system of underground drainage through caves. Essentially, two noteworthy caves are considered: the Bus de la Genziana (Fig. 1) with a maximum depth of 587 m and a development of 8 km, and the Abisso Col de la Rizza, the deepest cavity of the area, reaching 800 m below the surface. Morphologically, all caves have a complex tunnel system, including shafts, halls, canyons, meanders and sometimes are also well decorated (Grillo 2007).

The hydrologic connection between the Cansiglio Plateau and two of the three main Friulian sources, the Santissima and Molinetto origins of the Livenza River, have been demonstrated by recent tracing examinations (Filippini et al. 2016).

3 Geodetic and Hydrologic/Environmental Data

3.1 Climate Data

Rainfall, snow, atmospheric pressure and air–temperature data were obtained from the ARPA Veneto (regional Meteorological Service and Snow and Avalanches Service), considering the weather station Cansiglio–Tramedere (1028 m a.s.l.) for hourly sampling. It is placed at about 8 km north from the CANV station and 2.5 km from the tiltmeter location.

3.2 Hydrological Monitoring

Multiparametric loggers (CTD-Diver, Schlumberger Water Services) have been installed in 2010 in the springs Santissima and Gorgazzo located in Polcenigo at the foothills of the massif, while a barometer (Baro-Diver, Schlumberger Water Services) was placed close to one of the springs for barometric correction of the stage data. The pressure is used to reduce the stage observations for atmospheric pressure variations using the inverse barometric hydrologic response. These instruments record variations in water level (h), temperature (T) and specific electrical conductivity (EC) every hour.

Since the year 2013, in the cave Bus de la Genziana another Diver CTD has been installed to record the water flow dynamics below the tiltmeters, with hourly sampling, at the depth of 587 m from the surface, or 433 m above sea level (Fig. 3a, b).
Fig. 3

The Genziana cave and instrumentation: a Simplified geological section of the Genziana cave and b picture of Diver instrument located at the bottom of the cave. c Schematic design of the Marussi tiltmeter. The digital acquisition occurs by means of a magnetic induction transducer. d The pair of tiltmeters in the Bus de la Genziana–Pian Cansiglio, located 25 m below the surface (Photo: Barbara Grillo)

The hydrometric station of the Livenza River is located at the foothills of the massif in the city of Sacile, about 10 km south from the springs area. The data consist in the river level measurement with hourly sampling and are available through the monitoring network of the Civil Protection of Friuli Venezia Giulia.

3.3 Marussi Tiltmeters

Since 2005, the University of Trieste runs a tiltmeter station in the natural cave “Bus de la Genziana” 25 m below the entrance (Grillo et al. 2011). The Genziana tiltmeters are horizontal pendulums with Zöllner-type suspension and described in detail (Braitenberg 1999a; Zadro and Braitenberg 1999). They are sturdy instruments due to their relatively big size (0.5 m tall) and stable mount, inside a cast iron bell resting on compact rock. The iron bell is sustained by three supports placed on the solid rock. The horizontal arm rotates in the horizontal plane around a subvertical axis, the rotation angle being picked up by a magnetic transducer. The pendulum is sensitive to a tilting of the ground at 90° with respect to the off-vertical angle of the pendulum’s rotation axis (see Fig. 3c) plus a mass attraction effect. The digital data acquisition has a sampling rate of 1 h and uses an inductive transducer. The resolution of the tiltmeter is near 5 nrad, the value corresponding to the unit value of the digitizing process. Sign convention is positive for east- and northward tilting of the pendulum (tilting down).

3.4 GPS Observations

Two permanent GPS stations are available on the Cansiglio Massif, and five temporary stations had been installed to outline the area affected by the hydrologic deformation. The permanent stations are the following: the Caneva station (CANV), belonging to the FReDNet (; Battaglia et al. 2003), and the Tambre station (TAMB), owned by the Regione Veneto authority ( The CANV station is located at 800 m a.s.l. at the southern margin of the Cansiglio–Cavallo Plateau at a distance of about 8 km from the Genziana cave and is established on a reinforced concrete pillar anchored on bedrock. The TAMB station is located on the plateau at about 400 m from the entrance of the Genziana cave, settled on the roof of a stone-made house. Five monitoring stations were set up along the southern margin of the plateau and on top of Mt. Pizzoc in the southwestern edge of the plateau and were measured occasionally in the last years (CN01, CN03 and CN04 in Fig. 1).

4 Description of Geodetic Observations in Relation to Hydrologic Data

4.1 Characteristic Signals of Tiltmeter Observations

Interpretation of the tilt signal must take into account both environmental factors and deformation, since the tilting can be provoked by thermal expansion, deformation due to loading or pressure variations, as well as being sensitive to earth tides, tectonic stresses and tectonic deformation (Jahr 2017). The environmental generative sources are temperature, superficial and underground water flow and snow (Jahr et al. 2008), and atmospheric pressure gradient. Our objective is to be able to identify the different signals and quantify the source. The Cansiglio station started recording in December 2005, including an instrumental adjustment and testing period. The useful recordings for representing the tectonic movement start from 13 February 2006. The long-term tilting direction is principally southward, with a slowdown in 2008 (Grillo et al. 2011), relative stability in 2009, an episode of northward movement until 2012 and a return to the southward movement which is still ongoing (Fig. 4a, b). In all graphs, increase of tilt numbers corresponds to downward tilting of the platform on which the tiltmeter stands toward north and east, respectively. The amount of tilting is such that the mass effect on the deviation of the vertical, which would also generate a tilt signal, is negligible.
Fig. 4

The 10 years of tilt in Genziana cave, hourly rainfall and Livenza River stage. a The tilting signal of the two components of the tiltmeter compared with the stage data of Livenza River and the rainfall measured at the Cansiglio weather station. Please notice the different scale factors used for representing the EW and NS components. We note a small eastward movement and a nearly seasonal signal in the component EW, and a southward movement in the component NS, which could be of tectonic origin. The hydrologic signal manifests itself as spikes corresponding to rain events, and also as slow movement in correlation with the runoff curves of the aquifer (Grillo et al. 2011). b The hydrologic, temperature and electrical conductivity data of the two springs (Gorgazzo and Santissima springs, and the measurements below Genziana cave. Also shown is the stage of the River Livenza and hourly rainfall

The contribution of the barometric effect is negligible in comparison to the hydrologic effect, and correlation to atmospheric pressure gradient changes could not be found (Kroner et al. 2005; Boy et al. 2009), probably masked by the hydrologic effect.

The tiltmeter time series are highly correlated with the rainfall data of the Cansiglio weather station (station Tremedere ARPA Veneto) and to the stage data of River Livenza, as can be seen in Figs. 4 and 6. Both EW and NS components have a seasonal periodical variation, as can be seen after detrending and low pass filtering the time series. By least squares adjustment, a third-order polynomial is fitted to the NS and EW components separately and subtracted; then, the outcome of this operation is low pass filtered by a running average of 10 days. The final residual is the signal that most clearly identifies the hydrologically induced tilting (Fig. 5). In another tilt station (Grotta Gigante station) installed in a karst plateau (Classical Karst, straddling the Italian–Slovenian border; outside the map of Fig. 1), the spectral analysis had demonstrated that the yearly and seasonal signal has both thermal (365, 183 day period) and rainfall (365, 183, 171 days) spectral components, so it is generated by the superposition of both effects (Tenze et al. 2012). Also in Genziana, an annual and semi-annual signal can be identified, although not regularly repeating itself (Fig. 5). The different amplitudes of the long-term movement of the NS and EW components cannot be due to the mountings or instability of the location, because the two instruments are identical and there are no visible differences in the site on which the instruments stand (Fig. 3d). Mutual swap of the position of the two instruments confirmed the southward polarized movement. The tilt effect of barometric changes has been shown to be induced by the horizontal gradient of air pressure (Kroner et al. 2005; Boy et al. 2009).
Fig. 5

The seasonal and hydrological signal in tilt. The figure shows the detrended, seasonal (low pass filtered) and hydrological (high pass filtered) tilt signal for station Bus de la Genziana. Time window, years 2012–2017

4.2 Discussion of Hydrogeological Campaign Compared to the Tiltmeter Observations

The Cansiglio karst aquifer is a mature pre-alpine deep karst, characterized by high permeability and rapid runoff through enlarged fractures and caves, although mitigated by the existence of a base flow component through a network of smaller channels (Grillo 2007). The underground karst phenomenon is mainly developed in Monte Cavallo limestone with a complex of caves 600–800 m deep, controlled by the geological–structural setting. The aquifer has a high conductivity and high vulnerability (Cucchi et al. 1999).

Starting in the year 2010, a few hydrological campaigns limited in time were made studying the Gorgazzo (50 m a.s.l.) and Santissima (30 m a.s.l.) springs (see data in Fig. 4b). These studies confirmed what previous studies had shown: the Gorgazzo spring is characterized by a drainage through a well-developed fracture network, because it has highly variable conductivity and sudden and abundant rate of flow, which normalizes after a few hours and thus falls within the classic case of upward siphons (electrical conductivity between 150 and 370 μS/cm and temperature values of 10–11.5 °C). The Santissima spring has electrical conductivity values between 220 and 300 μS/cm and temperature values of 8–9.5 °C, typical of a less well-developed karst network draining with mixing of waters (Grillo 2007). The system of this spring is different because it has smaller water level fluctuations with longer times of decay. This, in terms of chemical and physical properties, is in support of a water reservoir of significant extensions feeding the Santissima spring (A.R.P.A. F.V.G. 2006; Filippini et al. 2016). For a limited number of months, the hydrological studies of the two springs overlap with the tiltmeter recordings. As seen in Fig. 6, the tiltmeter signal is associated with the stage signals of the Gorgazzo and Santissima springs, which confirms that the deformation monitored with the tiltmeters is induced mainly by the incoming rainwater and the drainage to the karst conduct system. The impulsive deformation at rainfall onset is highly amplified, and the subsequent slow decay has a similar time evolution to the Gorgazzo although shorter, but the amplitude of the spring is small compared to the onset of deformation. Similar relations hold for the level variations of the Santissima spring.
Fig. 6

Comparing the river level, the spring level (Santissima 30 m a.s.l., Gorgazzo 50 m a.s.l.) and the tiltmeter signals at 1000 m a.s.l. (EW, NS) from 15 August to 30 November 2010: the spring levels correlate very well with the tiltmeter data. The tilt transients last some days like the groundwater runoff signals

The level in the siphon, recorded far below the tiltmeters with the CTD data logger, gives definitive clues about the dynamic source of the tilting. Data are available for 4 years, starting 2013, and only partly overlap with the data available for the two springs (Fig. 7a). In contrast to the records of the two springs and River Livenza, the water level is zero, except after rainfalls. This is because the water channels always filled with water are supposed to be at a lower level than the siphon. The water level in the siphon rises abruptly by 4–5 m after a few millimeters of rain: a statistical analysis has shown that when the rainfall is above 12 mm/h, the water level will rise by more than 5 m. In three extreme cases, the monitored water level had reached 27 m (16 May 2013, 31 mm of rainfall in 5 h, and 25 h for level decay) and 50 m (31 January 2014 and 26 December 2013) with heavy rainfall, lasting for 43 and 79 h, respectively. The velocity of maximum rise was 7 m/h. The characteristic response is an impulsive signal and near to linearly decaying discharge (Fig. 7b). The complete recession of the event lasts up to 10 days.
Fig. 7

Comparison between tiltmeter data and the CTD multiparametric station below the cave, next to other ambient and hydrologic parameters. a Entire time series. Tiltmeter data, atmospheric pressure, hourly rainfall and river series, the recordings of the underground Genziana flow monitoring, conductivity and water temperature. The water level below the cave Genziana rises quickly to 4–5 m with a few mm of rain: it had reached 27 m in one event and had risen over 50 m two times (compare text). b Significant time window cut-out from (a), period 7 March to 29 May 2013. On tilt, we note the earth tides and the impulsive signals marking the onset of the underground water level rise. Event A shows the time of 25 m water level rise below the tiltmeter station (compare text)

The electrical conductivity values are on average of 230 μS/cm, while the parameter reduces down to 150 μS/cm when the karstic system fills due to rainfall. The water temperature variations are well correlated to the conductivity and also have the characteristic impulsive increase with slow recovery. The medium temperature is 8 °C, with variations limited to − 0.15° to +0.30°. Increase of temperature correlates with reduction of electrical conductivity and with onset of rainfall (Grillo and Braitenberg 2015).

These parameters show that the hydrologic system is affected by mixing of new infiltration water, which is very fast in the inflow, but slower in the outflow, so the water level abruptly increases and then slowly decays.

It is to be expected that the tilting correlates to the filling of the siphon below the instruments. This is most convincingly seen in Fig. 7b, where in a time window of 2.5 months data for the tilt, siphon water height, level of River Livenza and rainfall are shown. The river level must be at least 3.3 m high to flood the siphon, and at all times that the river rises above this threshold level the siphon is filled. Every time the siphon is filled, there is a tilting event, with 2 µrad tilting for a rise of 5 m in the siphon. The tilting has a short duration and marks the filling stage of the siphon. The marking of the filling stage is demonstrated best by analyzing the time derivative of the siphon, as shown in the next figure. Only in rare cases in which the siphon is filled above the 5 m mark do the tiltmeters also have a slow recovery of deformation, as at the event A in Fig. 7b.

To further illustrate the tiltmeter response during the rising stage in the siphon, we calculate the first time derivative of the stage and compare it to the tilt records, calculating the cross-correlation function between the two. A negative lag corresponds to a delay in the siphon The cross-correlation function is maximum (cross-correlation coefficient 0.38) for zero delay between the diver stage rate and the tiltmeter records, and has a smaller but well-developed negative extreme at − 54 h delay of the diver change rate with respect to the tilting.

We further select the time of each steep rise in stage and a time window covering the successive 12 h. We draw the tilt hodograph for each of these events and calculate the azimuth (positive from EW anticlockwise) of the tilting at the time of maximum tilting in each of these windows (Fig. 8).
Fig. 8

Tilting signal induced by the rapid stage rise in the siphon below the tiltmeter. Upper left: detrended and high pass filtered NS tilt component. Middle left: time derivative of water level in Genziana cave. Lower left: cross-correlation function between tiltmeter and stage rate in siphon. Negative lag: siphon rate delayed with respect to tilt. Upper right: hodograph of tilting in the 12 h following the rapid stage rise in the Genziana cave. Lower right: polar histogram of orientations of hydrologic-induced tilt signal

4.3 Model for Explaining the Diver and Tiltmeter Observations

The physical model that explains the tilt observations must comply with the observations. We have no direct observations of the shafts, so the physical model we propose here is a conjecture. To justify the model, we summarize the responses of the diver and the tilt to variations in rainfall and the Livenza River, and then formulate a structural model that explains the observations. The conceptual model can be used for a numerical simulation of the hydraulic flows and the induced deformation, which requires a separate study based on the present one.

The characteristics of the diver observations below the tilt station are twofold: a general maximum height of 5 m water height and linear level decrease, accompanied by a tilt signal limited mainly to the fast water level increase. A possible model could be made of a shaft of which the lower part evolves to tight fissures filled with silt and mud (see Fig. 9). The shaft collects the rain water from above and fills up with water up to 5 m maximum height, above which the water drains entering a free lateral channel. The shaft quickly fills up due to the rain and slowly releases the water through the tight fissures. The permeability of the filled fissures is inversely proportional to the water height, so as to cancel the height-proportional term in the Darcy flow rule, and the outgoing flux is at constant rate, as is also the decrease of water level above the diver. Due to the presence of the fissure filling, with viscous rheology, the stress acting on the fissure walls is proportional to the first time derivative of the pressure increase due to the water level height change. Therefore, the movement of the tiltmeters is perceptible only in the impinging phase, whereas during outflow the time derivative is too low to generate an observable tilt signal. This is documented in Fig. 8, where the NS tilting is compared to the first time derivative of the diver. It can be verified that the strong peaks in the time derivative of the diver are well correlated to the peaks of the NS component. The tilting occurs in both NS and EW components for all cases in which the velocity of the increasing diver level overcomes a threshold value (here chosen as of 1 m/h, but also other values can be chosen and the results are not affected). In rare cases of strong rainfall, the flow in the channel system below the cave is such that the water pressure is increased, and the vertical shaft of the diver is filled by a lateral arm we conjecture must be connected with the channel system below the diver. The shaft of the diver acts as a pressure gauge, with a fast increase of the water level by up to 25 m (see Event A in Fig. 7) governed by the pressure pulse in the channel. The level decays after the pressure events are different from the common linear decay described above, because it no longer represents the mud-filled shafts and their permeability, but the flow in the open channel system, with a different flow dynamic. The tilt response is proportional to a superposition of viscous and elastic time evolution.
Fig. 9

Cartoon illustrating the mechanism that explains the upper limit of 5 m water level and linear level decrease in case of standard rainfalls, and the uprising to over 25 m level in case of exceptionally strong rainfalls that produce an overpressure in the lower conduit (see text)

4.4 Discussion of GPS Observations in Relation to Subsurface Hydrologic Flows

The hydrologic-induced deformation has also been detected with GPS stations installed on the karstic plateau. The results of a dedicated campaign were discussed in Devoti et al. (2015) and it was demonstrated that the horizontal movement of the stations after rainfall is outward toward the margins of the plateau and orthogonal to the prevailing fracture direction (Devoti et al. 2016). The signals were interpreted as due to filling and subsequent pressure exerted by vertical fractures.

In Fig. 4a, we show the updated time series (updated with respect to the paper Devoti et al. 2015) of the permanent station CANV, which has a consistent pattern of deformation correlated to the rainfall (Fig. 10), with a fast displacement (1–2 days) of up to 15 mm and a subsequent slow rebound following the rainfall event (lasting 2–3 weeks). The signals were interpreted as due to the percolation of rainwater filling subvertical fractures resulting in a lateral pressure stress and subsequent elastic opening of the fractures, followed later on by closure after draining.
Fig. 10

GPS, tiltmeter observations, hourly rainfall and stage of the River Livenza during flood events. During the event of the Livenza flood from 31 October to 3 November 2010 (Event B in figure), the total amount of rain was of 520 mm. The interdisciplinary comparison shows a clear correspondence in the recordings. For the GPS NS component, the shift is 16 mm to south and the tilt is 35 µrad southward; for the EW component, it is 12 mm to the east recorded by the GPS and 3 µrad to the east for the tiltmeter. The inset shows a zoom of the EW, NS tilt and rainfall for the period 29 October–10 November 2010. The colors of the curves in the inset are those of the full graph

On comparing the tiltmeter’s signal recorded at the Genziana station with the local rainfall series, the Livenza River gauge height, and the time series of the GPS stations located on the southern slopes of the Cansiglio Massif, a clear correspondence with the water runoff is seen (Fig. 10).

The hydrologic-induced deformation was particularly evident during the flood of the Livenza River between October 31 and November 3, 2010 (Fig. 10) (Grillo 2010). In those days, a total amount of 520 mm of rain fell with a peak 29 mm/h. The GPS station and the tiltmeters recorded the displacement and deformation induced by the water influx into the system instantly, and also due to the underground water runoff during the flood. The flood resulted in a GPS maximum displacement of 10 mm to the east and 15 mm to the south. The tiltmeters show a 1 µrad variation toward east and then a continuous westward drift of about 3 µrad during the rainy days, recovering the initial easterly position in the following week. The NS component provided initial complex variations during the rainy period, drifting first to the north and then slowly to the south and recovering the initial position in the following week (see inset in Fig. 10).

5 Discussion

We observed a variety of signals in and around the caves of the Cansiglio Plateau, ranging from standard hydrological signals such as rainfall, stage data, temperature and electrical conductivity, to GPS observations and measurements of tilt in the cave Bus de la Genziana. From this variety of observations, we show how the tiltmeters respond to the fast infiltration, related to the amount and duration of the rainfall, and to a lesser extent to the phreatic discharge, because of lower pressure change. In contrast, the horizontal displacement from GPS does not show any reaction from the fast infiltration, but records the slower phreatic discharge. The amplitude relation between the GPS movement and the amount of rainfall is approximately linear, accounting for 3 mm displacement every 100 mm cumulative rainfall for strong events. The prevailing movement is horizontal along a stable direction 130–160° north azimuth for stations at the eastern part of the plateau. The mutual correlation is significant, with a high correlation coefficient of 0.92 (Devoti et al. 2015). The direction of movement points to the area of the two springs mentioned above (Fig. 7) and is directed orthogonal to the frontal thrust systems of the Cansiglio Massif. The high correlation with the rainfall suggests that the deformation of the GPS, like the tiltmeters, is caused by subsurface hydrologic processes in the karstic vadose zone rather than with deep-rooted water table variations in the phreatic zone, as pointed out by Devoti et al. (2015).

A geological and structural survey revealed that the whole southern Cansiglio slope is affected pervasively by structural features related to the Cansiglio thrust and by other NNW–SSE and NNE–SSW conjugate faults and fractures perpendicular to the Cansiglio southern slope (Fig. 11). These structural features are responsible for rock mass weakening that leads to large-scale gravitational instability and favors the nucleation of karstic landforms (dolines and swallow holes) (Devoti et al. 2016).
Fig. 11

Cansiglio Plateau model: the slope is affected pervasively by structural features (faults, thrusts, fractures). The tiltmeters and GPS observations record two different movements: the epikarst fast infiltrations and the phreatic discharge (modified after Devoti et al. 2016). The long-term tilting could be due to a tilting of the slope facing the thrust, due to the growth of the anticline

The region to which the Cansiglio carbonatic massif belongs is subjected to compressional stress-oriented SSE–NNW with subhorizontal angle, as has been deduced from focal mechanisms (Bressan et al. 2003). This direction is nearly compatible with southward tilting that has been observed over the past 10 years. Different studies (Davis et al. 1983; Yeats 1986; Huang and Johnson 2016) have proposed the model in which an anticline on an active thrust fault shows growth and increase of the anticline topographic slopes. A southward tilting, such as the one documented by the tiltmeters, could be interpreted by this model as being the expression of the southward movement of the massif along the two thrust faults that mark the margin of the southern border of the massif (Fig. 1) and are responsible of the southward tilting of the slope facing the thrust fault.

6 Conclusions

We describe an active slope deformation monitored with GPS and tiltmeter stations in a karstic limestone plateau in southeastern Alps (Cansiglio Plateau), one of the most interesting karstic areas of northeastern Italy. Considering the geophysical and hydrogeologic setting, the geodetic tilt station located in Bus de la Genziana is situated in a strategic position. The tiltmeters record crustal movements: long-term tectonic movement and hydrogeological information (epikarst fast infiltration and phreatic slow unloading).

The long-term tilt has a definite southward direction, with episodes of stability and northward movement. The tilting reflects the actual tectonic situation in northeast Italy, which shows the convergence of the Adriatic and Euroasiatic plates.

The long-term movement is overprinted by fast tilting induced by rainfall and a slower movement due to the phreatic water discharge. GPS observations record the slower movement in the horizontal components as well, and movement is principally horizontal in the SSW direction and back, and can reach the order of 1 cm during the hydrologic-induced movement. The recovery of the deformation following the rain can last up to 2 weeks. The interpretation of the different time constants is a fast water pressure change in the epikarst and a slower water pressure buildup at deeper karst levels, where water fills the karstic channels, with local level increments of up to 50 m. We can approximate the behavior of the Cansiglio Massif as a network of drainage channels with dominant preferential directions, flowing in the karst vadose reticulum with different trigger times depending on the amount and duration of rainfall. The long-term deformation pattern revealed by geodetic instruments probably reflects the discharge of the karst aquifer, a first impulsive reaction due to rapid and turbulent flow in the conduit network, followed by a slow discharge in the porous matrix (pores and fissures).

The direct link between the aquifer system cycles and the induced surface deformation provides interesting insights into karst-style hydrological processes that could also be relevant in the assessment of hydrologic hazards. The GPS and the tilt observations are complementary and sensitive enough to study and monitor the effects of water infiltration in karst systems.

Considering that the southern Alpine front accommodates a compression of a few millimeters per year and that the area is known to be in Zone 2 (defined as: Zone 2: Municipalities in this area may be affected by quite strong earthquakes, Protezione Civile 2015) at medium to high seismic risk, we are asking how and if these sudden shifts due to hydraulic load may affect the geophysical and geodynamic context of the Cansiglio area. The analysis of time series of the permanent GPS at Caneva suggests an elastic response of a hydro-structure with a drainage system directed along NW–SE, parallel to the direction of the complex headwater of Polcenigo–Caneva. Furthermore, an improved knowledge of the relation between the karstic water flows and the geodetic signal will allow the use of the geodetic measurement as a hydrologic investigation tool in the future.



We thank Alberto Casagrande for the constant engagement and for the precious collaboration; A.R.P.A. Veneto Centro Meteorologico of Teolo for providing the meteorological data; Ivano di Fant (Ufficio Idrografico del Servizio Gestione delle Risorse Idriche—Friuli Venezia Giulia) for the hydrometrical data of Livenza River; Dr. Alberto Riva for providing the geological map; OGS-RSC Working Group for publishing the local seismicity on their homepage; Dr. Paola Favero and the Comando Forestale of Pian Cansiglio for the hospitality and the collaboration; the local cavers (in particular, the speleologists of Unione Speleologica Pordenonese CAI Pordenone) for the support and collaboration in the installation and sampling of the diver in the bottom of Bus de la Genziana. We acknowledge the use of the GMT software (Wessel et al. 2013). We thank Georg Kaufmann and an anonymous reviewer for meticulous reviews.


  1. A.R.P.A. F.V.G. (2006). Rilevamento dello stato dei corpi idrici sotterranei della Regione Friuli Venezia Giulia (Survey on the state of underground hydrologic units of the Friuli Venezia Giulia Region). Final Report, pp. 68–71. Regione Autonoma Friuli Venezia Giulia.Google Scholar
  2. Battaglia, M., Zuliani, D., Pascutti, D., Michelini, A., Marson, I., Murray, M. H., et al. (2003). Network assesses earthquake potential in Italy’s Southern Alps. EOS, 84(28), 262–264.CrossRefGoogle Scholar
  3. Boy, J.-P., Longuevergne, L., Boudin, F., Jacob, T., Lyard, F., Llubes, M., et al. (2009). Modelling atmospheric and induced non-tidal oceanic loading contributions to surface gravity and tilt measurements. Journal of Geodynamics. Scholar
  4. Braitenberg, C. (1999a). The Friuli (NE Italy) tilt/strain gauges and short term observations. Annali di Geofisica, 42, 1–28. Scholar
  5. Braitenberg, C. (1999b). Estimating the hydrologic induced signal in geodetic measurements with predicitive filtering methods. Geophysical Research Letters, 26, 775–778.CrossRefGoogle Scholar
  6. Braitenberg, C. (1999c). The hydrologic induced strain—A review. Marees Terrestres Bulletin D’Informations, 131, 1071–1081.Google Scholar
  7. Braitenberg, C., Grillo, B., Nagy, I., Zidarich, S., & Piccin, A. (2007). La stazione geodetico—geofisica ipogea del Bus de la Genziana (1000VTV)—Pian Cansiglio. Atti e Memorie della C.G.E.B., S.A.G. CAI, Trieste, Italia, 41, 105–120.Google Scholar
  8. Bressan, G., Bragato, P. L., & Venturini, C. (2003). Stress and strain tensors based on focal mechanisms in the seismotectonic framework of the Friuli-Venezia Giulia Region (Northeastern Italy). Bulletin of the Seismological Society of America, 93, 1280–1297. Scholar
  9. Cancian, G., & Ghetti, S. (1989). Stratigrafia del Bus de la Genziana (Cansiglio, Prealpi Venete). Studi Trentini di Scienze Naturali—Acta Biologica, Trento, 65, 125–140.Google Scholar
  10. Cavallin, A. (1980). Assetto strutturale del Massiccio Cansiglio—Cavallo, Prealpi Carniche Occ. Atti del 2° Convegno di Studi sul Territorio della provincia di Pordenone (Piancavallo, 19–2 ottobre 1979).Google Scholar
  11. Cucchi, F., Forti, P., Giaconi, M., & Giorgetti, F. (1999). Note idrogeologiche sulle sorgenti del Fiume Livenza. Atti della Giornata Mondiale dell’Acqua “Acque Sotterranee: Risorsa Invisibile”, Roma, 23 marzo 1998 (Pubbl. n°1955 del GNDCI, LR49), pp. 51–60.Google Scholar
  12. Davis, D., Suppe, J., & Dahlen, F. A. (1983). Mechanics of fold and thrust belts and accretionary wedges. Journal of Geophysical Research, 88, 1153–1172.CrossRefGoogle Scholar
  13. Devoti, R., Falcucci, E., Gori, S., Poli, M.E., Zanferrari, A., et al.. (2016). Karstic slope “breathing”: Morpho-structural influence and hazard implication. Poster General Assembly EGU, Vienna.Google Scholar
  14. Devoti, R., Zuliani, D., Braitenberg, C., Fabris, P., & Grillo, B. (2015). Hydrologically induced slope deformations detected by GPS and clinometric surveys in the Cansiglio Plateau, southern Alps. Earth and Planetary Science Letters, 419, 134–142. Scholar
  15. Filippini, M., Casagrande, G., Fiorucci, A., Gargini, A., Grillo, B., Riva, A., et al. (2016). Geological and hydrogeological investigations for the design of a multitracer test in a major karst aquifer (Cansiglio-Cavallo, Italian Alps). Rendiconti online della Società Geologica Italiana, 39(suppl. 1). ISSN 2035-8008.Google Scholar
  16. Grillo, B. (2007). Contributo alle conoscenze idrogeologiche dell’Altopiano del Cansiglio. Atti e Memorie della C.G.E.B., S.A.G. CAI, Trieste, Italia, 41, 5–15.Google Scholar
  17. Grillo, B. (2010). Applicazioni geodetiche allo studio dell’idrogeologia del Cansiglio. AA. 2009–2010. Master Thesis, Environmental Sciences University of Trieste. Edizioni Accademiche Italiane. ISBN 978-3-639-66321-1.Google Scholar
  18. Grillo, B., & Braitenberg, C. (2015). Monitoraggio delle acque di fondo del Bus de la Genziana (Pian Cansiglio, Nord-Est Italia). Atti e Memorie della C.G.E.B., S.A.G. CAI, Trieste, Italia, 46, 3–14.Google Scholar
  19. Grillo, B., Braitenberg, C., Devoti, R., & Nagy, I. (2011). The study of karstic aquifers by geodetic measurements in Bus de la Genziana station—Cansiglio plateau (North-Eastern Italy). Acta Carsologica. Scholar
  20. Huang, W. J., & Johnson, K. M. (2016). A boundary element model of fault-cored anticlines incorporating the combined mechanisms of fault slip and buckling. Terrestrial, Atmospheric and Oceanic Sciences (TAO). Scholar
  21. Jahr, T. (2017). Non-tidal tilt and strain signals recorded at the Geodynamic Observatory Moxa, Thuringia/Germany. Geodesy and Geodynamics,. Scholar
  22. Jahr, T., Jentzsch, G., Gebauer, A., & Lau, T. (2008). Deformation, seismicity, and fluids: Results of the 2004/2005 water injection experiment at the KTB/Germany. Journal of Geophysical Research, 113(B11), 410.CrossRefGoogle Scholar
  23. Kroner, C., Jahr, T., Kuhlmann, S., & Fischer, K. D. (2005). Pressure-induced noise on horizontal seismometer and strainmeter records evaluated by finite element modeling. Geophysical Journal International, 161(1), 167–178. Scholar
  24. Longuevergne, L., Florsch, N., Boudin, F., Oudin, L., & Camerlynck, C. (2009). Tilt and strain deformation induced by hydrologically active natural fractures: Application to the tiltmeters installed in Sainte-Croix-aux-Mines observatory (France). Geophysical Journal International, 178, 667–677. Scholar
  25. OGS-RSC Working Group. (2012). Rete Sismica di Collalto.,
  26. Pettenati, F., & Sirovich, L. (2003). Source inversion of intensity patterns of earthquakes: A destructive shock in 1936 in the northeast Italy. Journal of Geophysical Research, 109, B10309. Scholar
  27. Pondrelli, S., Ekström, G., & Morelli, A. (2001). Seismotectonic re-evaluation of the 1976 Friuli, Italy, seismic sequence. Journal of Seismology, 5, 73–83. Scholar
  28. Priolo E., Romanelli, M., Plasencia Linares, M. P., Garbin, M., Peruzza, L., Romano, M. A., et al. (2015). Seismic monitoring of an underground natural gas storage facility: The Collalto Seismic Network. Seismological Research Letters, 86(1), 109–123 + esupp.
  29. Protezione Civile. (2015). Retrieved October 10, 2017, from
  30. Tenze, D., Braitenberg, C., & Nagy, I. (2012). Karst deformations due to environmental factors: Evidences from the horizontal pendulums of Grotta Gigante, Italy. Bollettino di Geofisica Teorica ed Applicata, 53, 331–345. Scholar
  31. Wang, R., & Kümpel, H.-J. (2003). Poroelasticity: Efficient modeling of strongly coupled, slow deformation processes in multilayered half-space. Geophysics, 68(2), 1–13. Scholar
  32. Wessel, P., Smith, W. H. F., Scharroo, R., Luis, J. F., & Wobbe, F. (2013). Generic Mapping Tools: Improved version released. Eos Transactions American Geophysical Union, 94, 409–410.CrossRefGoogle Scholar
  33. Yeats, R. S. (1986). Active faults related to folding. In R. E. Wallace (Ed.), Active tectonics (pp. 63–79). Washington, DC: National Academy Press.Google Scholar
  34. Zadro, M., & Braitenberg, C. (1999). Measurements and interpretations of tilt-strain gauges in seismically active areas. Earth Science Reviews, 47, 151–187. Scholar
  35. Zuliani, D. (2003). FReDNet: una rete di ricevitori GPS per la valutazione del potenziale sismico nelle Alpi sudorientali italiane. GNGTS, Atti del 22° Convegno Nazionale, Roma, 18–20 November 2003.
  36. Zuliani, D., Priolo, E., Palmieri, F., & Fabris, P. (2009). Progetto GPS-RTK: una rete GPS per il posizionamento in tempo reale nel Friuli Venezia Giulia. Dimensione Geometra, 12, 30–36.Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  • Barbara Grillo
    • 1
  • Carla Braitenberg
    • 1
  • Ildikó Nagy
    • 1
  • Roberto Devoti
    • 2
  • David Zuliani
    • 3
  • Paolo Fabris
    • 3
  1. 1.Department of Mathematics and GeosciencesUniversity of TriesteTriesteItaly
  2. 2.Centro Nazionale TerremotiIstituto Nazionale di Geofisica e VulcanologiaRomeItaly
  3. 3.Centro di Ricerche Sismologiche - CRSOGS - Istituto Nazionale di Oceanografia e di Geofisica SperimentaleCussignacco, UdineItaly

Personalised recommendations