G.G.R. Iovine1, S.L. Gariano1, P. Iaquinta1, P. Lollino2 and O.G. Terranova1
1CNR-IRPI, via cavour, 6 ÛÒ 87036 Rende (Cosenza), ITALIA [email@example.com]
2CNR-IRPI, via Amendola, 122 I ÛÒ 70126 Bari, ITALIA
Keywords: Rainfall-induced landslide, Slope stability analysis, Risk mitigation, Monitoring network, Alarm
With reference to landslides, Italy is exposed to the worst risk conditions among the industrialized countries (Guzzetti, 2000). In particular, the abundant and prolonged rains fallen in the 2008-2009 and 20092010 wet seasons triggered a great number of shallow and deep slope movements in Calabria (Southern Italy), mostly in the northernmost area. These geo-hydrological crises were so severe that the Italian government had to declare the ÛÏstate of emergencyÛ on both circumstances.
Among the landslides activated in the mentioned wet periods, a largedebris slide threatened the southern suburbs of San Benedetto Ullano (SBU), a small village in the Cosenza province. In this paper, a brief description of the considered slope movement — that first mobilized on Jan. 28, 2009, and after a short period of inactivity, re-activated on Feb. 11, 2010 — is provided. In addition, the decision support system implemented following the first phases of activation is summarized. Based also on such system, the Authority of Civil Protection could effectively manage the geo-hydrological risk during the cited emergency phases.
2. The Case Study
Calabri a is an accretionary wedge made of a series of Jurassic to Early Cretaceous ophiolite-bearing tectonic units, plus overlying Hercynian and pre-Hercynian basement nappes (Amodio-Morelli et al., 1976). In the region, average yearly rainfalls vary between 1,000 and 2,000 mm/y in mountainous and internal areas, and between 600 and 900 mm/y in coastal areas, with a mean regional value of about 1,150 mm/y. As also confirmed in a recent review of the general frame of storm conditions in Calabria, heavy rains are by far more frequent on the Jonian side of the region (Terranova, 2004). Over 70% of the yearly precipitation occurs from October to March, with negligible monthly values from June to September. Average monthly rains recorded at Montalto Uffugo (i.e. the closest rain gauge to SBU) are shown in Table 1.
San Benedetto Ullano (450 m a.s.l.) is located in Northern Calabria, along the left flank of the Crati Graben (cf. Figure 1). It lies at the base of the Coastal Chain, in a sector marked by a N-S trending normal fault, which extends for ca. 30 km. Along this fault, the metamorphic rocks of the Coastal Chain, to the West, give place to the Pliocene-Quaternary sediments of the graben, to the East (CASMEZ, 1967).
The San Rocco landslide, considered in this study, developed at the southern margin of SBU, between the historical centre and the cemetery, and threatened the San Rocco church. In Figure 2, the area mostly affected by the activations of the San Rocco landslide in 2008-09 and 2009-10 is dashed in red, whereas the overall area threatened by the slope movement is dashed in light green.
2.1 The San Rocco landslide activation of Winter 2008-2009
At the Montalto Uffugo gauge, the cumulated amounts of rain from November 2008 to March 2009 exceeded more than twice the seasonal average (cf. Table 1). During the two and three months antecedent to the landslide mobilization, cumulated rains approached the first two critical cases ever recorded since 1921 (Iovine et al., 2009).
On Jan. 28. 2009, after ca. 500 mm of cumulated rains in 10 days, the landslide started mobilizing and a set of fissures was noticed by the inhabitants along the road network. In the same area, the existence of a dormant slope movement had already been mapped in the management plan of geo-hydrologic risk of the Calabrian regional administration (PAI, 2001). Nevertheless, the sector actually affected by the mobilization resulted by far larger than the mapped one.
Repeated detailed geomorphologic field surveys allowed to map the evolution of the phenomenon, and to progressively update the extent of the affected zone. In addition, a set of datum points was selected over the landslide body and along its margins to monitor the kinematics of the phenomenon by means of recurrent hand-made measures of ground surface displacement, performed by a team of volunteers (specifically trained by the authors for this purpose). Starting from Feb. 11, 2009, a set of precision-extensometers plus a meteoric station were installed, and a real-time monitoring system of data collection, transfer via GSM, and pre-processing was implemented. The monitoring system was implemented so that the frequency of data transfer to the pre-processing unit depended on the observed values. Accordingly, distinct automated protocols were defined to timely activate the field inspections and inform the team of experts and the Authority of Civil Protection (cf. Iovine et al., 2009).
A peak of about 10 cm/day of velocity was recorded on Jan. 30, 2009, upslope of the S. Rocco church. On Jan. 31, the width of fissures along the road network reached ca. 20 cm, indicating velocities of ca. 7 cm/day at several monitoring points. Starting from Feb. 5, fractures kept on widening and lengthening, and new ruptures gradually opened even outside the sector affected by the first instability. Significant velocities (in the range 4-5.5 cm/day) were still observed between Feb. 8 and Feb. 12, in coincidence of two days of intense rains (over 120 mm). In the following months, observed velocities gradually decreased, down to few mm/month in the late spring 2009.
Observed daily displacements resulted to be generally consistent among themselves, and well related with rainfall regime, although being characterized by a quite complex behavior. Based on geomorphological evidences and on kinematic observations, the slope movement could be interpreted as a large, confined phenomenon (including a smaller pre-existing landslide), that was experiencing an incipient phase of activation.
2.2 The San Rocco landslide activation of Winter 2009-2010
In the study area, exceptional rains were recorded again during autumn-winter 2009-10 (cf. Table 1). Until the end of January 2010, field inspections did not evidence clear signs of landslide re-activation; only small displacements were detected by the monitoring system, generally after rain events. Between Jan. 31 and Feb. 1,2010, following ca. 330 mm of cumulated rains in six days, a new phase of mobilization started, again in the central sector of the slope (cf. area dashed in red in Figure 2). On Feb. 11, a new paroxysmal phase was triggered after further 295 mm of rains cumulated in 11 days, causing serious damage to the main road infrastructures (on this occasion, part of the instruments of the monitoring system had to be re-located). On Feb. 12, 35 mm of rain coupled with a small snowfall induced a new acceleration of the landslide (Iovine et al., 2010; Capparelli et al., 2010).
As a whole, the 2009-10 landslide mobilization caused by far a greater damage to the middle sector of the slope with respect to the previous one. Starting from the end of March 2010, the landslide activity gradually reduced. In the late spring, activity persisted with displacements of few mm/month. In summer 2010, the emergency phase could be declared closed.
3. Slope Stability Analysis and Mitigation Procedures
During both the considered landslide activations, the groundwater level approached the slope surface in the middle sector of the slope (i.e. the mostly affected by the phenomenon): New springs originated, while other pre-existing ones showed a notable increase of discharge. Aiming at understanding the role of groundwater level changes on slope stability, limit equilibrium analyses coupled with seepage modelling were performed (see details in Iovine et al., 2010). In particular, the results of a first seepage model, representative of an initial steady-state summer regime, indicated groundwater levels in agreement with the observed measurements at S1, an open-pipe piezometer located in the middle portion of the displaced mass (cf. Figure 2).
Starting from these steady-state hydraulic conditions, parametric slope stability analyses were performed with respect to the groundwater level variations, by considering six landslide sub-bodies, which were defined by taking into consideration both field evidence of failure surfaces and inclinometer data (cf. Figure 2 and 3). As summarized in Figure 4, for a groundwater depth of ca. 9 m at S1, the first sub-body that becomes unstable is #4 (i.e. the sector mostly displaced during both the recent activations). After a further reduction of groundwater depth to ca. 7 and 6 m, the sub-bodies #5 and #6 (both characterized by larger extension and volume) progressively become unstable, respectively. The results of the limit-equilibrium analysis also indicated that the remaining sub-bodies (in the order: #1, #2, and #3) would mobilize only in case of shallower groundwater conditions; for a depth of about 4 m, all the sub-bodies resulted to be unstable.
The above results fairly correspond with in situ evidence collected during 2008-09 and 2009-10 activations: Observed groundwater depths and landslide evolution are in quite good agreement with the modelled activation of landslide sub-bodies #4, #5 and #6. On the contrary, no clear evidence was found in the field concerning the mobilization of the lower-most portion of the phenomenon: The observed fractures and scarps developed by the Marri stream (rather than indicating the mobilization of the sub-bodies #1, #2 and #3) may in fact be ascribed to the activation of secondary landslides.
In absence of a detailed knowledge of the phenomenon, basic procedures for risk mitigation were empirically devised beginning in early February 2009 (Iovine et al., 2009), at the beginning of the first emergency phase, with reference to thresholds of maximum daily velocities, as observed at either datum points or extensometers. In the following months, also based on the availability of further geological, piezometric and inclinometric information, parametric limit equilibrium analyses were performed to improve the emergency procedures, with reference to the Factor of Safety (Fi) of the landslide sub-bodies. The main criteria, the procedures and the emergency measures suggested to the Authority of Civil Protection are summarized in Table 2 (Iovine et al., 2010): They were defined to be performed by the technicians and volunteers of the municipality, after a short training period.
In operational terms, in case a given criterion is satisfied, the monitoring system delivers an advice (via GSM) to the Authority of Civil Protection (and to the team of consultants, i.e. the authors), to allow for an immediate and proper activation of surveillance actions (e.g. the control of specific indicators). Based on results of surveillance and on the monitoring system data processing, the level of alarm can be updated.
It can be noticed that the adopted criteria refer to a combination of values of Fi (and, thus, indirectly, to groundwater level depths) and of punctual maximum velocities measured at the slope surface. Among the different values of Fi computed for the considered landslide sub-bodies with respect to different groundwater depths, the smallest one must be considered to enter Table 2. The apparent incongruity between Fi > 1 and v > 0 can be explained by considering that the observed velocities may refer to displacements not truly representative of the dynamics of the considered landslide sub-bodies (e.g. they may be due to secondary movements of superficial portions of the slope).
As a rule, to persist in a given level of alarm (cf. Table 2), no measure or evidence must exceed the criteria for that specific level, and no other (e.g. meteoric, seismic) unusual conditions must be observed. In case of single, anomalous evidence, an immediate patrol at prefixed critical points has to verify the actual conditions of the phenomenon in situ; if direct observations confirm the suspicion of anomalous/dangerous conditions, an immediate advise is sent to the Authority of Civil Protection and to the experts. When several values of superficial displacement reach worrying amounts ÛÒ better if concerning distinct sectors of the phenomenon ÛÒ or for a combination of worrying conditions for distinct parameters (e.g. groundwater level, daily and/or cumulated rains, deep displacements), the transition to the upper level is recommended.
4. Concluding Remarks
The adopted approach to landslide risk mitigation is based on: i) detailed geomorphologic field surveys (by experts), ii) recurrent hand-made measures of superficial displacement, and field inspections (by a team of volunteers), and iii) real-time monitoring. The results of such activities were cross-analysed with respect to hydrologic data, and combined with the results of a preliminary landslide stability model.
The automated monitoring system was implemented so that the frequency of data transfer to the pre-processing unit depended on actually observed values. Accordingly, distinct automated protocols were defined to timely activate field inspections and to inform the team of experts and the Authority of Civil Protection.
The choice of an empirical approach to define the emergency procedures was imposed by the need to promptly support the mayor during the first phase of landslide activation. Based on the results of field investigations and boreholes data, a preliminary geological scheme of the slope could be defined in the following months, and limit equilibrium analyses for the interpretation of the slope stability conditions could be performed in parametric terms. According to the modelling results (fairly corroborated by field observations), the original set of emergency procedures could be refined by linking groundwater depths and punctual velocities to distinct levels of alarm. The surveillance system will be further upgraded, thanks to new field data which will be available in the near future.
Thanks to the adopted approach of risk mitigation, both the considered phases of landslide mobilization could be monitored since the very beginning, thus allowing the local Authority of Civil Protection to promptly take suitable actions.
Amodio-Morelli L., Bonardi G., Colonna V., Dietrich D., Giunta G., Ippolito F., Liguori V., Lorenzoni S., Paglionico A., Perrone V., Piccarreta G., Russo M., Scandone P., Zanettin Lorenzoni E. & Zuppetta A. (1976) – LÛªarco calabro-peloritano nellÛªorogene appenninico-maghrebide, Mem. Soc. Geol. Ital., 17, 1ÛÒ60 (in Italian).
CASMEZ (1967) – Carta Geologica della Calabria, F.229IIINE ÛÏLattaricoÛ (in scale 1/25000). Poligrafica & CarteValori, Ercolano (Napoli).
Guzzetti F. (2000) – Landslides fatalities and the evaluation of landslide risk in Italy. Engineering Geol., 58, 89-107.
Iovine G.G.R., Iaquinta P. & Terranova O.G. (2009) – Emergency management of landslide risk during Autumn-Winter 2008/2009 in Calabria (Italy). The example of San Benedetto Ullano. In Anderssen R.S., Braddock R.D. & Newham L.T.H. (Eds.), Proceedings of the 18th World IMACS Congress and MODSIM09 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand and International Association for Mathematics and Computers in Simulation, July 2009, pp. 2686-2693.
Iovine G.G.R., Lollino P., Gariano S.L. & Terranova O.G. (2010a) – Coupling limit equilibrium analyses and real-time monitoring to refine a landslide surveillance system in Calabria (southern Italy), Nat. Hazards Earth Syst. Sci., 10, 2341ÛÒ2354.
Capparelli G., Iaquinta P., Iovine G.G.R., Terranova O.G. & Versace P. (2010) ÛÒ Modelling the rainfall-induced mobilization of a large slope movement in northern Calabria. Natural Hazards, DOI: 10.1007/s11069-010-9651-1
Monaco C. & Tortorici L. (2000) – Active faulting in the Calabrian arc and eastern Sicily. Journal of Geodynamics, 29, 407-424.
PAI (2001) – Piano Stralcio di Bacino per lÛªAssetto Idrogeologico (PAI). AutoritÌÊ di Bacino Regionale. Regione Calabria, 59 pp. (in Italian).
Terranova O. (2004) – Caratteristiche degli eventi pluviometrici a scala giornaliera in Calabria. Proceedings of XXIX Convegno di Idraulica e Costruzioni Idrauliche, 7-10 settembre 2004, Trento, Italy, Editoriale Bios, Cosenza, Italy, 343-350 (in Italian).
Giulio G.R. Iovine, is a scientist of the CNR-IRPI (National Research Council – Research Institute for Geo-Hydrological Protection) since 1994. He began its activities in applied geomorphology and, specifically, in landslide recognition and mapping, and has been responsible of research projects since 1999. His scientific interests are mainly directed towards the understanding and modeling of hazardous phenomena (mainly, slope instability processes) for hazard mapping and risk mitigation purposes. Recently, he has focused on modeling of flow-type landslides and on innovative techniques of evaluation of landslide susceptibility and hazard, on identification of rainfall thresholds of landslides, on sinkhole inventory and related hazard evaluations, on relationships among Radon anomalies, recent/active faulting and large scale slope movements, and on landslide risk mitigation approaches through remote surveillance networks.
Stefano L. Gariano, is research assistant at CNR-IRPI (National Research Council – Research Institute for Geo-Hydrological Protection) since July 2008. His main expertise is in GIS, in evaluation of environmental vulnerability, in risk assessment and in hydrological modeling.
Pasquale Iaquinta, is research assistant at CNR-IRPI (National Research Council – Research Institute for Geo-Hydrological Protection) since July 2002. His main expertise lies in understanding and modeling of hydraulic and hydrological processes, and of interaction between rainfall and slope stability. More specifically, he has worked in data base management and processing, and GIS implementation. Recently, he has been involved in evaluation of soil erosion hazard, in analyses on rainfall triggering of landslides, in landslide susceptibility and hazard assessment, and in landslide risk mitigation through remote monitoring systems.
Piernicola Lollino, is a scientist of the CNR-IRPI (National Research Council – Research Institute for Geo-Hydrological Protection) since 2001. His main expertise lies in numerical analyses of deformation and stability in soil slopes (Finite Element Method, Finite Differences Method) and of the interaction between soil and structures, in interpreting slope behavior through data gathered through monitoring systems, and in numerical analyses of slope stability in discontinuous rocks (Distinct Elements Method, Finite Elements Method).
Oreste G. Terranova, is a scientist of the CNR-IRPI (National Research Council – Research Institute for Geo-Hydrological Protection) since 1984. He began its activities in hydraulics and hydrology, and has been responsible of research projects since 1989. His scientific interests are mainly directed towards the understanding and modeling of hydrological processes and water interaction between precipitation and slope. In particular, he conducted studies on rainfall-runoff and sediment transport modeling, on soil water erosion, on the influence of precipitation over the triggering of landslides, on the evaluation of susceptibility to landslide at regional scale and on the risk management during emergencies through remote surveillance networks.