Journal of Volcanology and Geothermal Research 208 (2011) 5165

Contents lists available at SciVerse ScienceDirect

Journal of Volcanology and Geothermal Research

journal homepage: www.elsevier.com/locate/jvolgeores

Flank collapse scenarios at Volcn de Colima, Mexico: A relative instability analysis

Lorenzo Borselli a,, Lucia Capra b, Damiano Sarocchi a, Servando De la Cruz-Reyna c
a b c

Instituto de Geologia/ Fac. de Ingeniera - Universidad Autnoma de San Luis PotosUASLP, Av. Dr. Manuel Nava 5, C.P. 78240 San Luis Potos, Mexico Centro de Geociencias, UNAM, Campus Juriquilla, 76230 Queretaro, Mexico Departamento de Vulcanologa, Instituto de Geofsica, Universidad Nacional Autnoma de Mxico, Coyoacn 04510, D.F., Mexico

a r t i c l e

i n f o

a b s t r a c t
Previous studies on debris avalanche deposits of Volcn de Colima suggest a cyclic process of repetitive ank collapses triggered by major eruptions (VEI N 4). The recurrence interval of major collapse events during the last 10,000 years is calculated here using a stochastic approach, yielding a mean recurrence interval of 2698 yr, with an uncertainty range of 180 yr. The analysis yields an increased probability of ank collapse in the interval between 110 yr and + 345 yr from the present. This generates a series of scenarios ranging from optimistic, considering a collapse within the next 345 years, to pessimistic, derived from the 110 year delay. The analysis of relative mass/volume decit in the volcano structure, made using the new VOLCANOFIT 2.0 software, and a limit equilibrium analysis on the volcano anks point to the SW quadrant as potentially the most unstable sector of the edice under a wide range of scenarios. The TITAN2D numerical model is also used to simulate the extent of debris avalanches caused by failure of the SW ank. This approach may be applied to any volcano with a potential for ank collapse. 2011 Elsevier B.V. All rights reserved.

Article history: Received 31 December 2010 Accepted 23 August 2011 Available online 3 September 2011 Keywords: volcano ank collapse volcanic hazard slope stability analysis debris avalanche modeling stochastic numbers

1. Introduction The 1980 sector collapse and debris avalanche at Mount St. Helens triggered the recognition of characteristic hummocky deposits in many similar debris avalanche deposits worldwide (Siebert, 1984; Ui and Glicken, 1986; Siebert et al., 1987; Francis and Wells, 1988; Vallance et al., 1995; Capra et al., 2002). Since then, several studies have revealed that many volcanoes are susceptible to failure caused by exogenous or endogenous processes (McGuire, 1996), and that the associated deposits can completely change the topography around the volcano with important secondary effects, particularly on the hydrographic network (Swanson et al., 1986; Capra and Macas, 2002; Capra, 2007). Instability of a volcanic edice may be caused by many factors, either directly related to volcanic activity or to exogenous processes such as weathering. These factors include direct magmatic intrusion into the edice (Bezymianny-type activity, Gorshkov, 1962) or into the subvolcanic crust (Day, 1996; Elsworth and Voight, 1996), deposition of voluminous pyroclastic deposits on steep slopes (McGuire, 1996), hydromagmatic processes (Dzurisin, 1998), and phreatomagmatic activity (Bandai-type activity, Moriya, 1980). In some cases faulting may trigger collapse (McGuire, 1996), and the tectonic setting of the volcano may also inuence the direction of the failure (Siebert, 1984). In addition, the mass of the volcano can induce isostatic exure,

Corresponding author. Tel.: + 52 4448171039. E-mail address: lborselli@gmail.com (L. Borselli). 0377-0273/$ see front matter 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2011.08.004

compaction, and deformation that can lead directly to collapse (Borgia et al., 1992; van Wyk de Vries and Borgia, 1996). Although simple gravitational failure may occur in response to progressive weakening of an edice, discrete triggering mechanisms are commonly independent of the processes producing edice instability. Keefer (1984) established that numerous large landslides during historic time were triggered by earthquakes. Other triggering mechanisms include phreatic explosions and precipitation. Hurricane-induced rainfall triggered a ank collapse at the Casita volcano in Nicaragua in 1998, killing 2,500 people (Sheridan et al., 1999; Scott et al., 2005). Two different approaches have been used to model volcano instability; scaled analog experiments, and numerical simulation. i) Analog models have been widely used to simulate sector collapses of volcanoes, mostly focused on reproducing the direction of the collapse with respect to the stress eld affecting the volcano. Experiments of volcanic spreading have been performed to predict deformation, taking a volcano as a function of its height and the brittle-ductile ratio of the substratum, in extension and strike-slip settings (e.g. Merle and Borgia, 1996; van Wyk de Vries and Merle, 1996; van Wyk de Vries et al., 2003; Acocella, 2005; Norini and Lagmay, 2005). In addition, cryptodome intrusion has also been modeled attempting to reproduce the volcano deformation prior to the 1980 Mt. St. Helens collapse (Donnadieu and Merle, 1998; Donnadieu et al., 2001). ii) Numerical simulations have been used to understand how the stability of a volcano is affected by the increase of internal magmatic pressure (Dietrich, 1988; Russo et al., 1997), excess pore pressures due to intrusion (Voight and Elsworth, 1997; Elsworth and Day, 1999; Elsworth and Voight, 2001), hydrothermal alteration (Zimbelman et al., 2004) and even in magmatically

52

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

inactive volcanoes with signicant mass resting over a weak substratum (Borgia, 1994; van Wyk de Vries and Matela, 1998). The Colima volcano (3860 masl), also known as Volcn de Fuego, is an active composite cone with a maximum age of about 50,000 yr (Robin and Boudal, 1987) and is the youngest edice of the Colima Volcanic Complex (CVC), located in the western limit of the Trans-Mexican Volcanic Belt (Fig. 1). The older part of the edice, Paleofuego, (the ancestral Colima volcano), consists of a south-facing horseshoe-shaped crater surrounding the present active cone. Luhr and Prestegaard (1988) describe a main debris avalanche deposit exposed south of the edice, up to 70 km from the source, with an age of 4280 110 yr BP, contrasting with the age reported by Robin and Boudal (1987) of 9370 400 yr BP for the same deposit. Despite the difference in age determinations, both groups of authors agree that the deposit corresponds to a single event. In contrast, Komorowski et al. (1997) suggest that collapses have occurred at least 12 times in the last 45,000 years and perhaps as much as 9 times at the younger edice. Table 1 presents the radiocarbon ages related to these collapse events. Recently, Cortes et al. (2010) have described in great detail some of the more recent collapse events, as well as a 3600 yr BP collapse on the western ank of the edice emplacing a 1 km 3 debris avalanche deposit. Similar to older deposits (Capra and Macas, 2002; Capra, 2007), this debris avalanche deposit obstructed the Armera river, forming a temporary dam that then failed, producing a voluminous debris ow. Such secondary effects are caused by the walls of the N-S tectonic graben in which Colima is settled, acting as topographic barriers where the voluminous debris avalanches stop (Fig. 1). Modern activity of the volcano has been characterized by explosive phases, including two major Plianian eruptions such that occurred in 1818 and 1913 (Saucedo et al., 2010). Since the 1913 Plinian event, the volcano has had several eruptive phases. Its activity has been more persistent since 1998, with explosions and lava and dome extrusions (Saucedo et al., 2005). The collapse of summit domes and lava ow fronts has produced several block and ash ow deposits. Such deposits are up to several meters thick in the proximal area with lled proximal drainages up to 6 km from the vent. The block-and-ash ows at the Colima volcano consist of unwelded deposit with clasts embedded

in a silty to sandy matrix. During the last 15 yr the volcano had several eruptive episodes; in 1991, 1994, 19981999, 20012003, 2004 and 2005. Despite this persistent eruptive activity, the emitted products have not signicantly affected the surrounding inhabited area. During heavy rains, which usually occur from June through October at this latitude, these deposits are often remobilized, producing lahars (Capra et al., 2010). Although numerous studies on the textural characteristics of the avalanche deposits have been published, we are not aware of any results concerning the edice conditions prior to the failure or the possible triggering mechanism. Considering the present condition of the active cone, it is extremely important to understand its stability and recognize which sector could be destabilized by any endogenous or exogenous triggering process. The aim of the present work is to evaluate the relative ank instability of the Colima volcano using a set of new tools; recurrence intervals of cyclic debris avalanche events, the analysis of mass/volume decit with respect to a homogeneous stable reference shape, and a limit equilibrium method (LEM); and to evaluate the possible debris avalanche scenario after estimation of potential volume of the Debris Avalanche Events (DAE). 2. Materials and methods 2.1. Recurrence time of Debris Avalanche Events (DAE) The published average ages (BP) and associated uncertainties for each DAE in the Colima Volcanic Complex (Komorowski et al., 1997; Cortes et al., 2005, 2010) are listed in Table 1. The number of DAE is indeed much lower than the number of explosive events. De la CruzReyna (1993) established a Poissonian model for the recurrence intervals and occurrence frequency of explosive eruptions, and MendozaRosas and De la Cruz-Reyna (2008) analyzed the distribution of events with VEI N 4, which may be related to large DAE, nding an 85% probability of a VEI N 4 event within the next 500 yr, and an average recurrence time for VEI N 5 over 2500 yr. The fundamental problem for VEI N 4 events derives from the reduced number and reliability of event dating.

Fig. 1. Colima Volcan de Fuego. DEM.

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165 Table 1 Available ages of debris avalanche in the last 10,000 years BP, VEI and calculated intervals between the successive collapses and their corresponding band of uncertainty. Data source 1,2,3 2,3 2,3 2,3 1,2 Event ID Number () 4 3 2 1 0 VEI* () 5 5 6 56 56 Tei Debris avalanche events (DAE) (years BP) 2580 3600 7040 9671 13370 Tei Uncertainty on DAE (years) 140 120 160 88 120 Tei Interval from previous DAE (years) 1020 3440 2631 3699 n.a

53

Tei Uncertainty on interval from previous DAE (years) 184 200 183 149 n.a Te Standard deviation associated to mean DAE interval (years) 180

Mean interval of last four DAE (expressed as stochastic number)

Te Mean interval of last four DAE (years) 2698

1 Komorowski et al. (1997); 2 Cortes et al. (2005); 3 Cortes et al., 2010; *from Mendoza-Rosas and De La CruzReyna (2008).

The approach used in this paper is based on a simplied statistical approach to assessing the recurrence time of DAE with the limited number of events available. We use stochastic arithmetic (Vignes, 1993; Markov and Alt, 2004) adapted to the mean age of DAE and its band of uncertainty. This technique accounts for the error propagation and uncertainty associated with the computation of successive intervals between collapses. The proposed methodology resembles that proposed by Akiz et al. (2010) for the assessment of large earthquake recurrence times at the San Andreas Fault (California). Denoting the age in years before present (BP) of a DAE by Tei, and the uncertainty band around the average age by Tei (years), we perform an analysis of the mean recurrence time between DAEs, dening for each DAE a stochastic number characterized by the mean and standard deviation and then use stochastic number methodology (Vignes, 1993; Alt and Markov, 2001; Markov and Alt, 2004) for the computation procedure. Following the denition and procedure in Markov and Alt (2004), a stochastic number X is a Gaussian random variable with a known mean m and a known standard deviation s and is denoted X = (m; s). The set of stochastic numbers is denoted by S fm; sjmR; sRg: Under these denitions, arithmetic operations between stochastic numbers are identical to the operations between independent Gaussian distributions (Vignes, 1993). The application of the simple stochastic arithmetic allows the interval between two consecutive DAEs to be calculated as the difference Tei (years) between of the means of any given Tei and its preceding event Tei 1 as: Tei Tei1 Tei 1

v u n u u Tei 2 t Te i1 ; n and their values are included in Table 1.

3b

2.2. Digital elevation model analysis and treatment (by VOLCANOFIT 2.0) In order to estimate possible anomalies in the Colima volcano edice structure, we analyzed the current volcano's DEM by means of a special program developed by us, VOLCANOFIT 2.0 (www.volcanot.org). A theoretical description of the analytical procedure is given in Appendix A. The available DEM of the edice was obtained from the original LIDAR (provided by INEGI, the Mexican National Institute of Geography and Statistics) data. It has a spatial resolution of 5 5 m (Fig. 1). In our analysis we resample the original DEM at a grid resolution of 1010 m. The part of the edice that could be affected by an instability (which we call the Upper Edice) is shown in Fig. 2. This part of the DEM has been used for the analysis carried out with VOLCANOFIT 2.0. The aims of this analysis are to show a possible decit or surplus of edice volume with respect to an ideal 3D surface and a stable isotropic shape without preferential direction of collapse due to intrinsic geometric nonhomogeneities. The portion of edice suitable for this type of analysis is the new cone built on the Paleofuego collapse caldera. 2.3. Edice parameterization and slope stability computation In the past, the methodologies most used to analyze volcano ank collapses were based on slope stability computation, namely the limit equilibrium method (LEM) and the nite element method (FEM) (Duncan, 1996). Although several authors used both methodologies, the use of LEM still prevails for its relative simplicity and the limited number of parameters required. (Voight, 2000; Donnadieu et al., 2001; Zimbelman et al. 2004; Apuani et al., 2005a; Simmons et al., 2005; Reid et al., 2006; Hewitt, 2007; Reid et al., 2010). The basic steps of the analysis are assumption of a basic slope geomechanical model (including 2D section with strata, aquifer or piezometric lines, and geomechanical parameters assigned to each independent stratum). External forces and seismic effects can be simulated, as well as magma intrusions or uid pressure and overpressure (Iverson, 1995; Apuani et al., 2005a; Apuani and Corazzato, 2009). In the case of active volcanoes there are several problems, because sampling of the internal rock bodies cannot be done with standard direct geotechnical methods, and thus one is limited to surface outcrop sampling, or to assuming a range of possible values (Apuani et al., 2005a, 2005b). The Colima stratovolcano structure is assumed to be composed of three main bodies: 1) a body composed of andesitic lava ows and pyroclastic deposits, 2) the conduit and the active andesitic dome, and 3) a hydrothermally altered body surrounding the conduit and

We can also calculate the new width of the band of uncertainty (standard deviation) Tei (years) associated to each Tei as: Tei q Tei1 2 Tei 2 2

Table 1 shows the available ages of Volcn de Colima DAEs in the last 10,000 yr BP, the VEIs, and the calculated intervals between successive collapses and their corresponding bands of uncertainty. We now use the data of Table 1, considering each DAE as a stochastic number, to derive the mean recurrence interval in the last 10,000 yr BP and its band of uncertainty. The mean DAE interval of recurrence in the last 10,000 yr BP, Te and its band of uncertainty Te are thus: Tei
i1 n

Te

3a

54

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

Fig. 2. Portion of actual DEM used in VOLCANOFIT 2.0 analysis.

the dome (Fig. 3). This reconstruction is based on previous knowledge of similar volcanoes and of the Colima volcano itself, and is useful for carrying out the relative degree of stability analysis. For the present case, we used a LEM slope stability analysis based on Janbu's rigorous method (Janbu, 1973) and generalized computational and searching strategies to nd the sliding surface with minimum stability factor (Fs) (Siegel et al., 1981; Chen, 1992; Zhu et al., 2003, 2005). We used a random search strategy for all possible surfaces compatible with a potential sliding not constrained to a specic shape (e.g. circular only), which is implemented in the SSAP 4.0.6 software (only available in full freeware and non-commercial, license) (Borselli, 2011). As input data we used the reconstructed prole of the slope and a geotechnical parametrization using the Geological Strength Index (GSI) (Hoek et al., 2002; Marinos et al., 2005; Hoek and Diederichs,

Fig. 3. Colima volcano structure.

2006). This type of geomechanical parametrization provides a system for estimating the reduction in rock mass strength for different geological conditions as identied by eld observations, and derives the local equivalent shear strength MohrCoulomb parameters for fractured rock masses. This methodology to characterize different parts of a stratovolcano has been previously implemented in the analysis of other volcanic edice stabilities (Voight, 2000; Watters et al., 2000; Zimbelman et al., 2004; Apuani et al., 2005a,b; Hewitt, 2007; Gonzlez-Mellado and De la Cruz-Reyna, 2008). Table 2 shows the parameters that we assumed for each stratum in the volcanic structure. The values in Table 2 are based on average values from previous studies of stratovolcano slope stability, considering also a possible increase of 50% in the GSI values as the potential band of uncertainty. Seismic effects are also considered in the analysis, adopting a pseudostatic method in the computation (Ashford and Sitar, 2002). To apply this method, the horizontal and vertical seismic coefcients kh and kv are considered to be a function of the maximum peak ground acceleration (PGA) (Makdisi and Seed, 1978), and account for the increase in the vertical and horizontal inertial force components in the mass due to seismic action. The seismic horizontal and vertical coefcients kh and kv used are respectively 0.20.25 and 0.10.125, corresponding to an equivalent PGA in the range 0.40.5 (g). Accelerations exceeding 0.4 g (40% of the acceleration of gravity) in the region of the Colima volcano would require the occurrence of an earthquake with magnitude 7.5 or greater at a hypocentral distance of 50 km or less. The mean recurrence time estimated for this condition is about 1200 years (Mario Ordaz, 2011; Instituto de Ingeniera UNAM Mexico, personal comm.). The seismic effect was considered in separate scenarios (Table 3) associated with the analysis using higher GSI values (Table 2). The internal uid pressure in a volcanic edice has been recognized as one of the main triggering factors for potential instability (Day, 1996; Voight and Elsworth, 2000; Reid, 2004; Thomas et al., 2004). The uid

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165 Table 2 Shear strength parameterization of main bodies of the stratovolcano following the Hoek and Brown strength criterion (Hoek et al., 2002). unsaturated unit weight (kN/m3) Strato volcano main body Hydrothermal altered body Dome and conduct 24.5 24.0 24.0 s saturated unit weight (kN/m3) 25.0 24.5 24.5 I uniaxial compressive strength of intact rock element (MPa) 50 40 25 GSI geological strength index (adimensional) 40, (60)* 30, (45)* 20, (30)* mi lithological index (adimensional) 22 22 22

55

D disturbance factor (adimensional) 1.0 1.0 1.0

*In parentheses the GSI value for scenario analysis Nos. 2, 3 and 4 (50% increase assumed with respect to GSI of scenario no. 1).

pressure produces a local reduction of shear strength along a portion or all of the potential sliding surface, producing a global reduction of the stability factor. For the present analysis we adopt a new type of uid pressure function that acts like neutral pressure reducing the effective shear strength of rock masses. The new type of function, implemented in the SSAP 4.0.6 software, is described in Appendix B. The function is used because of the need to equilibrate the uid pressure at a given depth of volcanic edice with the lithostatic pressure, the presence of two uid pressure phases (liquid vapor), and the possibility of progressive pressure dissipation or overpressurization in the proximity of the suface. The SSAP software offers the possibility of excluding some portions of the edice from the calculation of uid neutral pressure. We used also this option considering the uid pressure computation only for the hydrothermalized and dome conduit systems. The main body of the stratovolcano was then excluded from the computation of uid neutral pressure. Fig. 4 shows the pressure eld inside the inner portion of the volcanic edice. The pressure increases up to 20 MPa at a depth of about 900 m from the volcano summit, and shows a non-linear decrease due to depressurization at the dome surface close to the surface. The parameters and uid pressure function assumed in the analysis are described in Appendix B and Fig. B.1. The analysis divides the volcanic edice into 12 sections (Fig. 5). These sections are drawn at angular steps of 30, in clockwise order, and centered around the summit of volcano structure. For each section with the basic internal structure assumed in Fig. 3, the SSAP software searched for the weakest sliding surface, characterized by the minimum stability factor Fs, in a set of randomly generated samples of 10,000 of potentially sliding surfaces. The main objective of the relative stability analysis is to establish which part of the present edice is the weakest and prone to an unstable condition. This type of analysis cannot be performed on a single userspecied surface, as many authors have done after a ank collapse (Apuani et al., 2005a) and the search for a minimum Fs must be applied to a signicant volume of the edice structure based on its morphological features that can be directly quantied. Other possible structural instability factors, such as faulting or criptodome intrusion should not be considered without clear evidence of their presence. In addition, the parameterization of geotechnical and uid parameters within the volcanic structure can be only relative, and not absolute, because of limited geotechnical knowledge of the internal volcanic structure.

2.4. Debris avalanche simulation Debris avalanches are dry, gravity-driven granular ows that follow topographic features during their emplacement. Various models have been proposed to explain the mobility of such ows and how the energy generated dissipates during owage (e.g. Hayashi and Self, 1992). A better way to simulate the inundation limits and thickness of this type of ows is by computation routines that simulate the ow over real topography. The TITAN2D code was designed to simulate a dry granular ow from an initial point of collapse over a natural terrain (Pitman et al., 2003; Patra et al., 2005) highly suited to simulating block-and-ash ows and debris avalanches. The code is based on a model for an incompressible Coulomb ow adapted from the work of Savage and Hutter (1989) using a shallow-water, depth-averaged approximation (Iverson and Denlinger, 2001). Mass and momentum conservation equations are solved with a Coulomb friction term for the interface between the granular material and basal surface and for the internal friction of the owing media (Pitman et al., 2003). The resulting hyperbolic system of equations is solved using a parallel adaptive mesh (Patra et al., 2005). We used the 2.0.1 version to perform the simulations. Terrain data are entered into the algorithm via GRASS GIS (Geographic Resources Analysis Support System). The main input parameters for running simulations are: i) the volume of the collapsed mass; ii) the basal friction angle (b), and iii) the internal frictional angle i. Initial conditions such as the coordinates of the starting point, elongation and orientation of the collapsed mass, initial velocity

Table 3 Characteristics of scenario analysis adopted for limit equilibrium analysis. Scenario Description no. 1 1 2 3 4 Geomechanical parameters as in Table 2 Geomechanical parameters as in Table 2 with GSI increase of 50% The same as scenario 2, but seismic coefcients Kh = 0.2; Kv = 0.1 The same as scenario 2, but seismic coefcient Kh = 0.25; Kv = 0.125 Notes No seismic effect No seismic effect Seismic effect by LEM pseudostatic analysis Seismic effect by LEM pseudostatic analysis

Fig. 4. Fluid pressure eld assumed in the relative stability analysis.

56

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

Fig. 5. Position of sections of the upper edice, where the slope stability computation has been performed. A constant angular step of 30 has been assumed.

and direction, duration and mass ux rate of the initial pile of material are also required (Sheridan et al., 2005; Charbonnier and Gertisser, 2009; Procter et al., 2010). After the ow starts, a criterion for determining its stopping is crucial for an accurate assessment of the runout, since in numerical simulations a ow never stops if its basal friction angle is higher than the slope angle. This is an unrealistic condition for real ows, where other physical parameters are involved in ow stopping (e.g. aspect ratio of the sliding pile, kinetic and potential energy of the moving ow, shape of the valley or channel for conned ows). Here we applied the global stopping criterion proposed by Yu et al. (2009) in which a dimensionless average velocity is used to determine the minimum acceptable average ow velocity below which the ow stops, as also described in Sulpizio et al. (2010). To perform simulations, we used a DEM of 50 m in resolution (vectorial data from the INEGI). The input parameters, such as the internal friction angle and basal friction angles are set as those calibrated to reproduce the 3600 yr BP collapse event (b = 8 and i = 30) (Cortes et al., 2010), which represent the younger and better-studied ank collapse episode at the Colima volcano.

A forecast for the time of a future collapse Tnext (year) counting from the present time can be calculated and expressed in terms of the stochastic number Tnext Tnext ; Tnext : q Tnext Tnext ; Tnext Te4 Te ; Te4 2 Te2 118; F228

where Te4 is the time of the most recent event and Te4 is its respective band of uncertainty (see Table 1).

3. Results 3.1. Scenario analysis from interval analysis of Debris Avalanche Events (DAE) Two possible scenarios of future debris avalanche occurrences due to partial edice collapse can be estimated using stochastic arithmetic. Using Eq. (3a) and (3b) we obtain Te 2698 and Te=180 as the mean DAE interval of recurrence in the last 10,000 yr BP, and its band of uncertainty (see Table 1). The result is itself a new stochastic number.

Fig. 6. DAE events vs. time interval from previous debris avalanche event. The projection of a possible scenario for the next DA event is included in the horizontal axis.

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

57

Finally, considering the signs of the stochastic component in Eq. (4), we have two possible scenarios. The more optimistic scenario Top may be calculated by means of the following equation: Top Tnext Tnext 345years

Table 4 VOLCANOFIT 2.0 tting parameters and statistics applied to Colima Volcan de Fuego upper edice 10 10 Dem. VOLCANOFIT 2.0 Input grid Data source Adopted Z limit (masl) Percentage random selected data used for tting Volcanoid function Fitting parameters Z1 (masl) X0 (m local UTM grid) Y0 (m local UTM grid) a b c Fitting statistics No. data Average absolute deviation (m) R2 EF

The negative sign in Eq. (5) indicates a projection into the future of 345 years from the present. The more pessimistic scenario Tpe may be calculated from: Tpe Tnext Tnext 110years

COLIMA_UPPEREDIFICE.DAT (INEGI) 2000 50% Negative exponential type 3860.8 645101.0 2158131.0 2494.9 3334.5 1429.0 80146 69.5 0.956 0.912

The positive sign of the result of Eq. (6) indicates a projection into the past with respect to the present time, or in other words, that the next DAE is 110 years overdue. Fig. 6 gives a graphical and simplied portrayal of the data of Table 1 and the results of the stochastic analysis. 3.2. DEM analysis and treatment (Volcanot 2.0) The application of VOLCANOFIT 2.0 to the upper edice of the Colima volcano (Fig. 7a) yields a good statistical t as indicated in

Fig. 7. a) Upper edice of Colima volcano DEM (2005) b) tted volcanoid 3D surface Eq. (A.5); c) Upper edice Colima Volcan de Fuego DEM with overlaid volcanoid Eq. (A.5); d) plot of local decit (negative values) or surplus (positive values) calculated with Eq. (A.6).

58

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

Fig. 8. a) Position of most critical surface in the section, azimuth 210; b) position of most critical surface in the section azimuth 240; c) position of most critical surface in the section azimuth 60; d) position of most critical surface in the section azimuth 30.

Table 4. The volcanoid of Eq. (A.5) (Fig. 7b) ts the actual DEM well, with a high coefcient of determination (r 2 = 0.956) and efciency modeling (Nash and Sutcliffe, 1970) of 0.912 over a large number of data points N80,000 (random selection of 50% of the total points of the DEM).

Fig. 7c shows the actual DEM and the overlay of the tted volcanoid. The f decit-surplus analysis procedure described in Appendix A produced the results shown in Fig. 7d, with a clear diffuse volume decit in the SW ank and a signicant surplus in the NE ank. The total decit volume in the SW ank is around 0.4 km3.

Fig. 9. Diagrams with the corresponding safety factor and index of relative instability vs. azimuth of volcano section. a) safety factor (Fs); b) running average of Fs; c) index of relative instability; d) running average of index of relative instability. The arrows indicate the azimuth with most unstable conditions.

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

59

3.3. Edice slope stability The LEM slope stability analysis performed for all sections at 30 azimuthal angular steps is shown in Figs. 8 and 9. Fig. 8a, b, c and d show the Fs obtained in four sections of the volcano edice at azimuthal angles of 210, 240, 60 and 30 for Scenario 1. The minimum stability factor is found at azimuth 240 (Fs = 0.997) and the maximum Fs at azimuth 60 (Fs = 1.700). It is worth recalling that Fs b1 indicates unstable conditions and Fs.N1.0 stable conditions. Fig. 9a shows the Fs values by azimuth. Fig. 9b shows the relative stability factors Rfsi for each section analyzed as dened by the following equation:

3.4. Debris avalanche simulation To simulate the extent of the area that could be affected by the collapse of the SW sector we modied the DEM of the volcano to nd the sliding surface over which the collapsing pile is built (Fig. 10). The sliding surface is set based on the result of the slope stability analysis described above. The total mass removed from the edice is 0.9 km 3. Considering that the mass dilates during emplacement by at least 30% in volume, the nal deposit would be close to 1 km 3. As previously stated, the basal friction angle, one of the main parameters affecting the simulation, was set at 8, as calibrated by Cortes et al. (2010) for the Los Ganchos debris avalanche originating from the 3600 yr BP partial sector collapse of the SW ank. Fig. 11 shows the distribution of the simulated ow and its thickness. The main path would be southwards, lling the Montegrande ravine, with the greater thickness (up to 40 m) accumulating in the La Yerbabuena plain, a morphological feature bounded by hummocks of older debris avalanche deposits. On the SW sector, the village of La Becerrera would be completely buried by the deposit. Based on the simulation, part of the mass would be able to emplace towards the SE, reaching the towns of Queseria and Tonila and affecting the main interstate road to Manzanillo, one of the main ports on the Pacic coast. 4. Discussion and conclusions The proposed new methodology for analysis of major DAE recurrences gives a cyclic recurrence time with a relative narrow band of uncertainty as well as the possibility of a consistent delay in generation of the next DAE. This delay, of approximately 110 years, is only

Rfs i

Fsi Fs max

where Fsmax is the maximum Fs value, and Fsi is the stability factor corresponding to each section of the entire set of sections analyzed. Eq. (7) enables the relative degree of stability of each section to be expressed with respect to the most stable part of volcanic edice. In other words, Rfsi is a normalized stability factor Fs that helps to efciently display the more problematic sections in terms of global instability. Fig. 9c and d are similar to Fig. 9a and b but they enable a comparison between scenarios 1, 2, 3 and 4, showing as well the inuence of varying mechanical parameters and seismic effects. Fig. 9a, b, c, and d indicate that the most potential unstable area of the volcano's edice is the SW quadrant, particularly the direction centered between azimuths 210 and 270.

Fig. 10. Titan 2D debris avalanche simulation. a) Direction of debris avalanche, b) assumed 3D sliding surface of debris avalanche, c) section of present topography and main sliding surface.

60

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

Fig. 11. Titan 2D nal stopping phase with accumulated debris depth above previous topography.

partially counter-balanced by the more optimistic scenario of the next DAE occurring within in the next 345 years. The scarce number of extreme DAE events (VEI N 4) in the last 10,000 yr BP does not allow a reliable Poisson analysis of recurrence times. The proposed new method, partially based on a recently applied principle to assess the long earthquake recurrence time at the San Andreas fault in California, offers a new perspective that can be applied not only to the Colima volcano but also to other similar volcanoes in Mexico and elsewhere. The observed SW ank decit seems to be correlated to the major relative instability in the same quadrant, in which steeper slopes prevail. It is well known that a non-homogeneous distribution of mass volume in a slope can produce either local or global instabilities. The general volcanoid surface obtained by the VOLCANOFIT program indicates the ideal surface and shape characterized by a non-prevalent unstable portion in terms of the mass volume distribution. The mass/volume decit analysis indicates that the SW quadrant ank has a larger difference with respect to the reference volcanoid. This difference may be explained in terms of greater lahar and erosion process activity in that quadrant (Capra et al., 2010). In fact, the SW quadrant is intensively incised and presents the most active lahar sources and gullies. The local erosion rate may be larger than in the other quadrants where the morphological surface is dominated by deposition. The LEM relative stability analysis and the mass decit analysis using VOLCANOFIT and SSAP are here proposed as an integrated procedure to better support the hypothesis of differential stability in a stratovolcano. The proposed analysis does not consider the possible effects of cryptodome intrusions, faulting, or specic seismic extreme events. But these instability factors may be included if a better knowledge of the internal volcano structure becomes available. Previous debris avalanche simulations made on the Colima volcano (Cortes et al., 2010) were based on reproduction of possible volumes calculated from eld evidence of previous DAE. In the present case the simulation is based on the most probable volume estimated by LEM analysis. However, it must be kept in mind that the volume assumption may be inuenced by many non-predictable trigger

factors such as cryptodome intrusion, or high uid circulation associated with heavy rains that can modify the ow behavior (Roverato et al., 2011). Our results conrm and support the hypotheses from previous studies and eld evidence of cyclic DAE at the Colima volcano and provide the possibility of forecasting a future event. The multi-sector LEM relative instability analysis results, strongly supports previous hypotheses of a dominant SW ank instability (Cortes et al., 2010; Norini et al., 2010). Acknowledgments We wish to thank Roberto Bartali for his help in the revision of the manuscript. We are grateful to the Instituto de Geologia, UASLP, for the logistic facilities. This work was partially supported by CONACyT and PROMEP (Project UASLP-PTC-241) funds. Roberto Sulpizio and an anonymous reviewer provided useful suggestions that greatly improved the manuscript. Appendix A. VOLCANOFIT 2.0: Concept description and use VOLCANOFIT 2.0 is a freeware software available for the scientic community. The aim of this software is to identify departures from the actual DEM of a volcano in shape or volume with respect to an ideal 3D surface. The authors believe that identication of these anomalies is important to understand the evolution of the edice; an important tool for volcanic risk scenario analysis, particularly for sector collapse events. VOLCANOFIT 2.0 performs a non-linear tting of the actual volcano DEM to an ideal 3D surface, with some constraints determined by the selected volcanoid surface. The non-linear tting is obtained by minimization of the average absolute deviation between the actual DEM and the 3D volcanoid surface. VOLCANOFIT uses the Differential Evolution Genetic Algorithm proposed by Storn and Price (1997) as an optimization engine. This algorithm performs the optimal solution search with intrinsic properties of global

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

61

optimization (Corne et al., 1999). (the details of the software are available at the ofcial site http://www.volcanot.org). In this appendix the theoretical basis for the development of a generalized 3D ideal surface that we call volcanoid and the procedure for computing the local volume decit/surplus of the volcanoid surface with respect to the actual volcano edice DEM are summarized. A.1. The volcanoid equation A surface of revolution with constant negative curvature may obtained assuming the following generatrix function Z(r): Zae
r b

A:1

where r is the radius or distance with respect to the axis of rotation, and a, b, c are coefcients. The fundamental property for a surface of revolution around a Z axis in the Cartesian system of coordinates X, Y, Z is: X Y r Z
2 2 2

Fig. A.2. Example of volcanoid with constant negative curvature (Eq. (A.5)).

A:2 Z Z1 if Z Z1 A:5b

Combining Eqs. (A.1) and (A.2) we obtain a function explicitly describing the surface of revolution Z(X, Y):
p 2 2 X bY

Zae c A:3 The function A.3 is the prototype of the basic shape of the symmetric and rotational source of revolution that appears in many stratovolcanoes, as well as in monogenetic cones. For simplicity we call this function a volcanoid. Using Cartesian coordinates, we need to specify and normalize the surface of revolution with respect to a center positioned at X0, Y0 inside the assumed Cartesian grid (e.g. a UTM grid) (Fig. A.1). In this way we can represent Eq. (A.3) using two additional variables: p 2 2
XX0
YY0 b

subject to the constraints z1 N c and z1, a, b, c N 0 . Fig. A.2 displays an example of a volcanoid obtained with Eq. (A.5a) and (A.5b). A.2. The Mt. St. Helens 1980 pre-eruption volcanoid An example of the use of VOLCANOFIT 2.0 is the application to the pre-eruption DEM (30 m grid) of Mt. St. Helens freely available from University of Washington, Earth and Space Science (2010). Fig. A.3a shows the original DEM with the elevations above 1200 m asl. The VOLCANOFIT non-linear tting output is listed in Table A.1. Fig. A.3.b displays the volcanoid tted surface. Fig. 4a shows the volcanoid surface overlapped on the original DEM. A.3. Local decit and surplus calculation

Zae

A:4

An additional element should be inserted allowing a truncation of the vertices to allow for the presence of a crater at a given specic elevation Z1. The nal shape of the volcanoid becomes: Zae

For relative stability scenario analysis it is useful to calculate the local relative difference between actual DEM and the tted volcanoid surface. The computation of local decit/surplus is obtained from: Zi Zi Zvi A:6

p 2 2
XX0
YY0 b

c if Z Z1

A:5a

where Zi is the local volume decit or surplus expressed as difference in elevation between the local actual elevation Zi and the local elevation of tted volcanoid Zvi. A map showing the local decit or surplus can be produced easily. Fig. A.4b shows the decit/surplus map of the 1980 pre-eruption Mt. St. Helens volcano. A.4. Other volcanoid genratrix functions Other generatix functions can be implemented. Two examples are given: Z a cosh r c b A:7

for r b c and a, b, c N 0. Z
Fig. A.1. Local reference system of coordinates used for the tting.

z1 a rc 1e b

A:8

with z1 N a and z1, a, b, c N 0.

62

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

Fig. A.3. a) Mt. St. Helens 1980 pre-eruption DEM, b) tted volcanod 3D surface Eq. (A.5).

Eq. (A.7) is characterized by a constant negative curvature as Eq. (A.1) and the nal volcanoid equation requires only 5 parameters to be dened. Eq. (A.8) is a portion of a sigmoid function and it is suitable for a prole with variable curvature (e.g. convexconcave prole) with a exus point indicated by the parameter c. An example of prole given by Eqs. (A.7) and (A.8) is shown in Fig. A.5.

edice is an important element for magma and dome buoyancy, setting a particular uid pressure eld inside. We assume a simple hydrostatic eld that coincides with the land surface, and then modify the basic eld w (Eq. (B.1)) with a dissipation or overpressure modier factor function FD (Eq. (B.2)): w w z B:1

Appendix B. Fluid pore pressure function In stability computation the neutral pressure is usually obtained by the hydrostatic pressure or seepage pressure obtained by groundwater level or ltration nets. In our case, uid pressure beneath the volcanic
Table A.1 Volcanot 2.0 tting parameters and statistics for Mt. St. Helens 1979. VOLCANOFIT 2.0 Input grid Data source Adopted Z limit (masl) Percentage random selected data used for tting Volcanoid Function Fitting parameters Z1 (masl) X0 (m local UTM grid) Y0 (m local UTM grid) a b c Fitting statistics No. data Average absolute deviation (m) R2 EF

where

w Z w and

is the hydrostatic uid pore pressure at a given dept (kPa) is the vertical distance from the surface (m) is the uid density (kN m -3)

OLD_ST_HELENS. DAT (University of Washington, Earth and Space Science, 2010) 1200 50% Negative exponential type 2890.1 562678.0 5116128.4 2505.3 2076.7 943.9 37808 45.6 0.987 0.974

FD 1Ae where: FD

kD

B:2

is the dissipation or overpressure function modier factor (adimensional) A is the dissipation type coefcient (adimensional) which can assume one of the three values: A=0 no dissipation or overpressure A = 1 to produce an increase of dissipation at decreasing distance from the surface A = +1 to produce increasing overpressure at decreasing distance from the surface k is the constant of dissipation (or overpressure) with values in the range 0.000001 to 0.1 (m 1)

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165

63

Fig. A.4. a) Pre-eruption 1980 DEM with overlaid volcanoid Eq. (A.5). b) Plot of local decit (negative values) or surplus (positive values) calculated with Eq. (A.6).

is the minimum distance from a point on the slope and the surface (m).

U0MIN

is the minimum pressure at Z = 0 and D = 0 to account for the exit pressure of a fumarolic and hydrothermal system (kPa).

The nal pore uid pressure at a given point beneath the surface is given by Eq. (B.3): f w zFD U0MIN where f is the uid pore pressure at a given point (kPa) B:3

Fig. B.1 shows the dissipation function for A = 1, w = 25 (kN m 3), U0MIN = 65 (kpa) and k = 0.005 (m 1), as used during the simulations. Notice that at large depths the dissipation function in Fig. B.1 coincides asymptotically with the lithostatic equivalent pressure (equilibrium with lithostatic pressure). Near the surface this pressure dissipates rapidly and asymptotically to U0MIN.

Fig. A.5. Alternative generatrix function of 3D volcanoid.

Fig. B.1. Fluid pore pressure assumed in the slope stability calculation.

64

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165 Gonzlez-Mellado, A.O., De la Cruz-Reyna, S., 2008. A simplied equation of state for the density of silicate hydrous magmas: an application to the Popocatpetl buoyancy-driven dome growth process. Journal of Volcanology and Geothermal Research 171, 287300. Gorshkov, G.S., 1962. On the classication and terminology of Pele and Katmai type eruption. Bulletin of Volcanology 24, 155165. Hayashi, J.N., Self, S., 1992. A comparison of pyroclastic ow and debris avalanche mobility. Journal of Geophysical Research 97 (B6), 90639071. Hewitt D., 2007. Risk analysis associated with ank failure from Putauaki, Bay of Plenty, New Zealand. Master Sc. Thesis. University of Waikato. http://researchcommons. waikato.ac.nz/handle/10289/2337 (last accessed 21-12-2010). Hoek, E., Diederichs, M.S., 2006. Empirical estimation of rock mass modulus. International Journal of Rock Mechanics and Mining Sciences 43, 203215. Hoek, E., Carranza-Torres, C., Corkum, B., 2002. HoekBrown criterion 2002 edition. Proc. NARMS-TAC Conference, Toronto 1, pp. 267273. Iverson, R.M., 1995. Can magma injection and groundwater forces cause massive landslides on Hawaiian volcanoes? Journal of Volcanology and Geothermal Research 66, 295308. Iverson, R.M., Denlinger, R.P., 2001. Flow of variably uidized granular material across three-dimensional terrain: 1. Coulomb mixture theory. Journal of Geophysical Research 106, 537552. Janbu, N., 1973. Slope stability computations. The Embankment Dam Engineering Casagrande Volume. John Willey & Sons. 4786 pp. Keefer, D.K., 1984. Landslides caused by earthquakes. Geological Society of America Bulletin 95, 406421. Komorowski, J.C., Navarro, C., Cortes, A., Saucedo, R., Gavilanes, J.C., Siebe, C., Espndola, J.M., Rodriguez-Elizarrars, S.R., 1997. The Colima Volcanic Complex. Field Guide #3, IAVCEI, General Assembly, Puerto Vallarta, Mexico. Luhr, J.F., Prestegaard, K.L., 1988. Caldera formation at Volcn de Colima, Mexico, by large Holocene volcanic debris avalanche. Journal of Volcanology and Geothermal Research 35, 335348. Makdisi, F.I., Seed, H.B., 1978. Simplied procedure for estimating dam and embankment earthquake-induced deformations. Journal of Geotechnical Engineering, American Society of Civil Engineers 104, 849867. Marinos, V., Marinos, P., Hoek, E., 2005. The Geological Strength Index: applications and limitations. Bulletin of Engineering Geology and the Environment 64, 5565. Markov, S., Alt, R., 2004. Stochastic arithmetic: addition and multiplication by scalars. Applied Numerical Mathematics 50, 475488. McGuire, W.J., 1996. Volcano instability: a review of contemporary themes. In: McGuire, W.J., Jones, A.P., Neuberg, J. (Eds.), Volcano Instability on the Earth and Other Planets: Geological Society Special Publication, 110, pp. 123. Mendoza-Rosas, A.T., De la Cruz-Reyna, S., 2008. A statistical method linking geological and historical eruption time series for volcanic hazard estimations: applications to active polygenetic volcanoes. Journal of Volcanology and Geothermal Research 176, 277290. Merle, O., Borgia, A., 1996. Scaled experiments of volcanic spreading. Journal of Geophysical Research 101 (B6), 1380513817. Moriya, I., 1980. Bandaian eruption and landforms associated with it. Collection of Articles in Memory of Retirement of Prof K Nishimura from Tohoku University. . 214 219 pp. Nash, J.E., Sutcliffe, J.V., 1970. River ow forecasting through conceptual models part I a discussion of principles. Journal of Hydrology 10, 282290. Norini, G., Lagmay, A.M.F., 2005. Deformed symmetrical volcanoes. Geology 33, 605608. Norini, G., Capra, L., Groppelli, G., Agliardi, F., Pola, A., Cortes, A., 2010. Structural architecture of the Colima Volcanic Complex. Journal of Geophysical Research 115, B12209. doi:10.1029/2010JB007649. Ordaz, M., 2011. (Mario Ordaz, Instituto de Ingeniera UNAM., Mexico City, Mexico, personal comm.). Patra, A., Bauer, A., Nichita, C.C., Pitman, E.B., Sheridan, M.F., Bursik, M.I., Rupp, B., Webber, A., Stinton, A.J., Namikawa, L., Renschler, C., 2005. Parallel adaptive numerical simulation of dry avalanches over natural terrain. Journal of Volcanology and Geothermal Research 139, 121. Pitman, E.B., Patra, A., Bauer, A., Sheridan, M.F., Bursik, M.I., 2003. Computing debris ow and landslides. Physics of Fluids 15, 36383646. Procter, J.N., Cronin, S.J., Platz, S.J., Patra, A., Dalbey, K., Sheridan, M.F., Neall, V.E., 2010. Mapping block-and-ash ow hazards based on Titan2D simulations a case study from Mt Taranaki NZ. Natural Hazards 53 (3), 483501 doi:10.1007/s11069-0099440-x. Reid, M.E., 2004. Massive collapse of volcano edice triggered by hydrothermal pressurization. Geology 32, 373376. Reid, M.E., Brien, D.L., Waythomas, C.F., 2006. Preliminary slope-stability analysis of Augustine Volcano. In: Power, J.A., Coombs, M.L., Freymueller, J.T. (Eds.), The 2006 Eruption of Augustine Volcano, Alaska: U.S. Geological Survey Professional Paper, 1769. Reid, M.E., Keith, T.E.C., Kayen, R.E., Iverson, N.R., Iverson, R.M., Brien, D.L., 2010. Volcano collapse promoted by progressive strength reduction: new data from Mount St. Helens. Bulletin of Volcanology 72, 761766. Robin, C., Boudal, C., 1987. A gigantic Bezymianni-type event at the beginning of modern volcano Popocatpetl. Journal of Volcanology and Geothermal Research 31, 115130. Roverato, M., Capra, L., Sulpizio, R., Norini, G., 2011. Stratigraphic reconstruction of two debris avalanche deposits at Colima Volcano (Mexico): insights into pre-failure conditions and climate inuence. Journal of Volcanology and Geothermal Research 207 (12), 3346.

References
Acocella, V., 2005. Modes of sector collapse of volcanic cones: insights from analogue experiments. Journal of Geophysical Research 110, B02205. doi:10.1029/ 2004JB003166. Akiz, S.O., Grant Ludwig, L., Arrowsmith, J.R., Zielke, O., 2010. Century-long average time intervals between earthquake ruptures of the San Andreas fault in the Carrizo Plain, California. Geology 38 (9), 787790. doi:10.1130/G30995.1. Alt, R., Markov, S., 2001. On the algebraic properties of stochastic arithmetic. Comparison to interval arithmetic. In: Kraemer, W., von Gudenberg, J.W. (Eds.), Scientic Computing, Validated Numerics, Interval Methods. Kluwer Academic, Dordrecht, pp. 331341. Apuani, T., Corazzato, C., 2009. Numerical model of the Stromboli volcano (Italy) including the effect of magma pressure in the dyke system. Rock Mechanics and Rock Engineering 42, 5372. Apuani, T., Corazzato, C., Cancelli, A., Tibaldi, A., 2005a. Stability of a collapsing volcano (Stromboli, Italy): limit equilibrium analysis and numerical modeling. Journal of Volcanology and Geothermal Research 144, 191210. Apuani, T., Corazzato, C., Cancelli, A., Tibaldi, A., 2005b. Physical and mechanical properties of rock masses at Stromboli: a dataset for volcano instability evaluation. Bulletin of Engineering Geology and the Environment 64, 419431. Ashford, S.A., Sitar, N., 2002. Simplied method for evaluating seismic stability of steep slopes. Journal of Geotechnical and Geoenvironmental Engineering ASCE 128, 119128. Borgia, A., 1994. Dynamic basis of volcanic spreading. Journal of Geophysical Research 99 (B9), 1779117804. Borgia, A., Ferrari, L., Pasquar, G., 1992. Importance of gravitational spreading in the tectonic and volcanic evolution of Mount Etna. Nature 357, 231235. Borselli, L., 2011. Slope Stability Analysis Program (SSAP). http://www.ssap2005.it (last accessed 13-07-2011). Capra, L., Macas, J.L., 2002. The cohesive Naranjo debris ow deposit (10 km3): a dam breakout ow derived from the pleistocene debris-avalanche deposit of Nevado de Colima volcano (Mexico). Journal of Volcanology and Geothermal Research 117, 213235. Capra, L., Macias, J.L., Scott, K.M., Abrams, M., Garduo-Monroy, V.H., 2002. Debris avalanche and debris ow transformed from collapses in the Trans-Mexican Volcanic Belt, Mexico behavior, and implication for hazard assessment. Journal of Volcanology and Geothermal Research 113, 81110. Capra, L., 2007. Volcanic natural dams: identication, stability and secondary effects. Natural Hazards 43, 4561. Capra, L., Borselli, L., Varley, N., Norini, G., Gavilanes, J.C., Sarocchi, D., Caballero, L., 2010. Rainfall-triggered lahars at Volcn de Colima, Mexico: surface hydro-repellency as initiation process. Journal of Volcanology and Geothermal Research 189, 105117. Charbonnier, S.J., Gertisser, R., 2009. Numerical simulations of block-and-ash ows using the Titan2D ow model: examples from the 2006 eruption of Merapi Volcano, Java, Indonesia. Bulletin of Volcanology 71, 953959. Chen, Z.Y., 1992. Random trials used in determining global minimum factors of safety of slopes. Canadian Geotechnical Journal 29, 225233. Corne, D., Dorigo, M., Glover, F., 1999. New Ideas in Optimization. McGrowHill0077095065. 450 pp. Cortes, A., Garduno, V.H., Navarro, C., Komorowski, J.C., Saucedo, R., Macias, J.L., Gavilanes, J.C., 2005. Carta Geolgica del Complejo Volcnico de Colima, Con Geologa del Complejo Volcnico de Colima. CARTAS GEOLGICAS Y MINERAS 10, 01854798. Cortes, A., Macias, J.L., Capra, L., Garduo, V.H., 2010. Sector collapse of the SW ank of Volcn de Colima, Mxico. The 3600 yr BP La Lumbre-Los Ganchos debris avalanche and associated debris ows. Journal of Volcanology and Geothermal Research 197, 5266. Day, S.J., 1996. Hydrothermal pore uid pressure and stability of porous, permeable volcanoes. In: McGuire, W.J., Jones, A.p., Neuberg, J. (Eds.), Volcano Instability on Earth and Other Planets: Geological Society Special Publication, 110, pp. 7793. De la Cruz-Reyna, S., 1993. Random patterns of occurrence of explosive eruptions at Colima Volcano, Mexico. Journal of Volcanology and Geothermal Research 55, 5168. Dietrich, J.H., 1988. Growth and persistence of Hawaiian volcanic rift zones. Journal of Geophysical Research 93, 42584270. Donnadieu, F., Merle, O., 1998. Experiments on the indentation process during cryptodome intrusions : new insights into Mount St. Helens deformation. Geology 26, 7982. Donnadieu, F., Merle, O., Besson, J.C., 2001. Volcanic edice stability during cryptodome intrusion. Bulletin of Volcanology 63, 6173. Duncan, J.M., 1996. State of the art: limit equilibrium and nite element analysis of slopes. Journal of Geotechnical Engineering, American Society of Civil Engineers 122, 577597. Dzurisin, D., 1998. Geodetic detection of inating stratovolcanoes: a potential breakthrough for mitigating volcanic hazards in the 21st century. Eos 79 973 pp. Elsworth, D., Day, S.J., 1999. Flank collapse triggered by intrusion: the Canarian and Cape Verde Archipelagoes. Journal of Volcanology and Geothermal Research 94, 323340. Elsworth, D., Voight, B., 1996. Dike intrusion as a trigger for large earthquakes and the failure of volcano anks. Journal of Geophysical Research 100, 60056024. Elsworth, D., Voight, B., 2001. The mechanics of harmonic gas pressurization and failure of lava domes. Geophysical Journal International, Royal Astronomical Society 145, 187198. Francis, P.W., Wells, G.L., 1988. Landsat thematic mapper observation of debris avalanche deposits in Central Andes. Bulletin of Volcanology 50, 258278.

L. Borselli et al. / Journal of Volcanology and Geothermal Research 208 (2011) 5165 Russo, G., Giberti, G., Sartoris, G., 1997. Numerical modeling of surface deformation and mechanical stability of Vesuvius Volcano, Italy. Journal of Geophysical Research 102, 2478524800. Saucedo, R., Macas, J.L., Sheridan, M.F., Bursik, M.I., Komorowski, J.C., 2005. Modeling of pyroclastic ows of Colima Volcano, Mexico: implications for hazard assessment. Journal of Volcanology and Geothermal Research 139, 103115. Saucedo, R., Macias, J.L., Gavilanes, J.C., Arce, J.L., Komorowski, J.C., Gardner, J.E., ValdezMoreno, G., 2010. Eyewitness, stratigraphy, chemistry, and eruptive dynamics of the 1913 Plinian eruption of Volcn de Colima, Mxico. Journal of Volcanology and Geothermal Research 191, 149166. Savage, S.B., Hutter, K., 1989. The motion of a nite mass of granular material down a rough incline. Journal of Fluid Mechanics. 199, 177215. Scott, K.M., Vallance, J.V., Kerle, N., Macias, J.L., Strauch, W., Devoli, G., 2005. Catastrophic precipitation-triggered lahars at Casita Volcano, Nicaragua: occurrence, bulking and transformation. Earth Surface Processes and Landforms 30, 5979. Sheridan, M.F., Bonnard, C., Carreno, C., Siebe, C., Strauch, W., Navarro, M., Calero, J.C., Trujilo, N.B., 1999. Report on the 30 October 1998 rock fall/ avalanche and breakout ow of Casita Volcano, Nicaragua, triggered by Hurricane Mitch. Lanslide News 12, 24. Sheridan, M.F., Stinton, A.J., Patra, A., Pitman, E.B., Bauer, A., Nichita, C.C., 2005. Evaluating Titan2D mass-ow model using the 1963 Little Tahoma Peak avalanches, Mount Rainier, Washington. Journal of Volcanology and Geothermal Research 139, 89102. Siebert, L., 1984. Large volcanic debris avalanches: characteristics of source areas, deposits and associated eruptions. Journal of Volcanology and Geothermal Research 22, 163197. Siebert, L., Glicken, H., Ui, T., 1987. Volcanic hazards from Bezymianny- and Bandaitype eruptions. Bulletin of Volcanology 49, 435459. Siegel, R.A., Kovacs, W.D., Lovell, C.W., 1981. Random surface generation in stability analysis. Journal of Geotechnical Engineering 107 (7), 9961002. Simmons, J., Elsworth, D., Barry, Voight B., 2005. Classication and idealized limitequilibrium analyses of dome collapses at Soufrire Hills volcano, Montserrat, during growth of the rst lava dome: November 1995March 1998. Journal of Volcanology and Geothermal Research 139 (34), 241258. Storn, R., Price, K., 1997. Differential evolutiona simple and efcient heuristic for global optimization over continuous spaces. Journal of Global Optimization 11, 341359. Sulpizio, R., Capra, L., Sarocchi, D., Saucedo, R., Gavilanes, J.C., Varley, N., 2010. Predicting the block-and-ash ow inundation areas at Volcn de Colima (Colima, Mexico) based on the present day (February 2010) status. Journal of Volcanology and Geothermal Research 193, 4966. Swanson, F.J., Norio, O., Masaki, T., 1986. Landslide dams in Japan. In: Schuster, R.L. (Ed.), Landslide Dams Processes, Risk and Mitigation: American Society of Civil Engineers Geotechnical Special Publication, pp. 131145.

65

Thomas, M.E., Petford, N., Bromhead, E.N., 2004. The effect of internal gas pressurization on volcanic edice stability: evolution towards a critical state. Terra Nova 16, 312317. Ui, T., Glicken, H., 1986. Internal structural variations in a debris-avalanche deposit from ancestral Mount Shasta, California, USA. Bulletin of Volcanology 48, 189194. University of Washington, Earth and Space science, 2010. http://rocky.ess.washington.edu/ data/raster/thirtymeter/mtsthelens/OldMtStHelens.zip (last accessed : july 2010). Vallance, J.V., Siebert, L., Rose, W.I., Girn, J.R., Banks, N.G., 1995. Edice collapse and related hazard in Guatemala. Journal of Volcanology and Geothermal Research 66, 337355. van Wyk de Vries, B., Borgia, A., 1996. The role of basement in volcano deformation. In: McGuire, M.J., Jones, A.P., Neuberg, J. (Eds.), Volcano Instability on the Earth and Other Planets: Geological Society Special Publication, 110, pp. 95110. van Wyk de Vries, B., Matela, R., 1998. Styles of volcano-induced deformation: numerical models of substratum exure, spreading and extrusion. Journal of Volcanology and Geothermal Research 81, 118. van Wyk de Vries, B., Merle, O., 1996. The effect of volcanic constructs on rift fault patterns. Geology 24, 643646. van Wyk de Vries, B., Wooller, L., Cecchi, E., Murray, J.B., 2003. The importance of strikeslip faulting in spreading volcanoes. EGS-AGU-EUG Meeting, Nice, France. Vignes, J., 1993. A stochastic arithmetic for reliable scientic computation. Mathematics and Computers in Simulation 35, 233261. Voight, B., 2000. Structural stability of andesite volcanoes and lava domes. Philosophical Transactions of the Royal Society of London 358, 16631703. Voight, B., Elsworth, D., 1997. Failure of volcano slopes. Geotechnique 47, 131. Voight, B., Elsworth, D., 2000. Stability of gas-pressurised lava-domes. Geophysical Research Letters 27, 14. Watters, R.J., Zimbelman, D.R., Bowman, S.D., Crowley, J.K., 2000. Rock mass strength assessment and signicance to edice stabilityMount Rainier and Mount Hood, Cascade Range volcanoes. Pure and Applied Geophysics 157, 957976. Yu, B., Dalbey, K., Webb, A., Bursik, M.I., Patra, A., Pitman, E.B., Nichita, C., 2009. Numerical issues in computing inundation areas over natural terrains using Savage-Hutter theory. Natural Hazards 50, 249267. Zhu, D.Y., Lee, C.F., Jiang, H.D., 2003. Generalised framework of limit equilibrium methods for slope stability analysis. Geotechnique 53 (4), 377395. Zhu, D.Y., Lee, C.F., Quian, Q.H., Chen, G.R., 2005. A concise algorithm for computing the factor of safety using the MorgensternPrice method. Candadian Geotechnical Journal 42, 272278. Zimbelman, D.R., Watters, R.J., Firth, I.R., Breit, G.N., Carrasco-Nuez, G., 2004. Stratovolcano stability assessment methods and results from Citlaltpetl, Mexico. Bulletin of Volcanology 66 (1), 6679.