The characteristics of seismic signals produced by lahars and pyroclastic flows: Volcán de Colima, México

Journal of Volcanology and Geothermal Research 179 (2009) 157–167

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research

j ourna l homepage: www.e lsev ie r.com/ locate / jvo lgeores

The characteristics of seismic signals produced by lahars and pyroclastic flows:Volcán de Colima, México

Vyacheslav M. Zobin ⁎, Imelda Plascencia, Gabriel Reyes, Carlos NavarroObservatorio Vulcanológico, Universidad de Colima, Colima, 28045, Mexico

⁎ Corresponding author.E-mail address: [email protected] (V.M. Zobin).

0377-0273/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.jvolgeores.2008.11.001

a b s t r a c t

Article history:

Keywords:seismic recordlaharpyroclastic flowVolcán de Colima

f seismic signals produced by pyroclastic flows (generated by either the collapseof a lava dome or an eruptive column) and lahars at Volcán de Colima, México are discussed. The paperconcentrates on the 2004–2006 activity associated with and following the extrusion of andesitic block-lavain October–November 2004. It is shown that the duration of the broad-band seismic records of pyroclasticflows lasts a few minutes while the duration of seismic records of lahars continues for tens of minutes orhours. The spectra of seismic records produced by pyroclastic flows are characterized by lower peakfrequencies (around 3–4 Hz) than for lahars (around 6–8 Hz). This difference in the frequency contenttogether with the difference in the duration of seismic signals allows early diagnostic of the events in realtime.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

The hazard of volcanoes is significantly related to volcanic flows,such as pyroclastic flows and lahars, which represent the naturalsediment motions generated along subaerial volcanic slopes andpropagating with a high velocity, sometimes destroying the near-bysettlements. These subaerial massive sediment motions are dividedinto two groups: the phenomena in which the particles are dispersedin the flowing body and the phenomena in which the moving bodiesare mostly the agglomerates of soil and rocks (Takahashi, 2007).Referring to volcanic flows, pyroclastic flows are included in the firstgroup; lahars belong to the second group. Both types of volcanic flowshave produced destructive effects during human history. In the list ofthe 27 largest volcanic catastrophes since 1700 to 1986 that wereresponsible for at least 1000 deaths (Bardintzeff and McBirney, 2000),11 were associated with pyroclastic flows and 11 with lahars.Therefore, the recognition of these events in real time is importantfor the mitigation of volcanic risk by decreasing their vulnerability.Seismic observations near a volcano may be basic for this type ofmonitoring. The study of seismic and acoustic signals produced byvolcanic flows for the mitigation of volcanic risk has been carried outat some active volcanoes of the world (Yamasato, 1997; Lavigne et al.,2000b; Calder et al., 2002; Zobin et al., 2005; Davila et al., 2007;among others).

Pyroclastic flows and rockfalls are common for both explosive andeffusive activity at andesitic and dacitic volcanoes. The fall-and-flow of

l rights reserved.

the pieces of lava dome or flow fronts as well as the numerousrockfalls generate seismic signals (Zobin, 2003). These signals may bethe main component of seismicity during block lava emission andsubsequent lava dome destruction. Rockfalls and pyroclastic flows areassociated with similar emergent, cigar-shaped waxing and waningseismic signals that had frequencies of 1–10 Hz and differ only in theirdurations and/or amplitudes (Calder et al., 2002). Ratdomopurbo andPoupinet (2000) note that the duration of these seismic signals(“guguran” in Indonesian) at Merapi volcano, Java usually lasts severalminutes.

A dacitic lava dome at Unzen volcano, Japan, emerged in May 1991,and its subsequent collapse resulted in the successive generation ofpyroclastic flows. Yamasato (1997) studied the seismic signalsproduced by the collapses of lava blocks from the dome, their falldown the slope and the migration of associated pyroclastic flows. Thecomparison of the seismic signals with video records allowed him toidentify seismic signals that were produced during the different stagesof the formation of a pyroclastic flow. Small amplitude seismic waveswith a predominant frequency of 2–3 Hz were excited almostsimultaneously with the collapse of the lava dome. In a few seconds,when lava blocks fell onto the slope, the amplitude of seismic signalbecome large with the appearance of a low-frequency (about 0.5 Hz)component. Then the fragmented pyroclastics started to flowgenerating high-frequency (more than 2 Hz) seismic waves.

Uhira et al. (1994) showed a good correlation between the max-imum signal amplitude recorded by long-period seismometer and thevolume of rocks within pyroclastic flows at Unzen volcano, Japan. Asnoted by Calder et al. (2002), the relationship between the volume of apyroclastic flow and the signal amplitude is not simple. More suitablecharacteristics may be the seismic signal envelope (Brodscholl et al.,

158 V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

159V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

2000) or the duration of seismic signals (Zobin et al., 2005). Thedurations of short-period signals of ground motions produced bypyroclastic flows and rockfalls during the 1998–1999 block-lavaextrusion at Volcán de Colima, México were categorized into: shortevents with durations less than 100 s; intermediate events withdurations between 100 and 250 s; and long events with durationslonger than 250 s. The longer seismic events were produced bypyroclasic flows with larger volume of material (Zobin et al., 2005).

Lahars at volcanoes are usually triggered by heavy rain. At Merapivolcano, for an example, about 20 mm/h is supposed to be critical forlahar triggering if it falls on unconsolidated materials such as ash,sand, or gravel (Lavigne et al., 2000a). At Pinatubo volcano,Philippines, hot lahars were triggered by rain with a similar thresholdvalue of about 20 mm/h (Rodolfo et al., 1996). The factors thatdetermine the lahar triggering are the rainfall distribution, intensityand duration, the morphology of the upper drainage and sedimento-logical characteristics of the source deposits. In some instances, laharsmay have been triggered by steam explosions that generated strongground vibrations, which induced failure of partially water-saturatedpyroclastic materials (Rodolfo et al., 1996). The typical 1991 post-eruption lahars at Pinatubo were estimated to be 2 to 3 m deep, 20 to50 mwide, and moving at 4 to 8 m/s (Pierson et al., 1996). The frontalmean velocity of lahars at Merapi ranges between 2.5 and 5.9 m/s; thepeak flow velocity may reach 6–15 m/s (Lavigne et al., 2000a). Theamount of sediments transported during a single lahar is extremelyvariable, depending primarily on the intensity and duration of rainfall.Small- to moderate-sized lahars at Pinatubo remobilized between2×105 to 106 m3 of sediment; larger events lasting up to 4 hours havetransported about ten times this amount (Rodolfo et al., 1996).

The duration of seismic records produced by lahars is rather longand may reach a few hours (Purbawinata et al., 1997; Lavigne et al.,2000b; Tuñgol and Regalado, 1996). Lahar discharge at Merapi is wellcorrelated with the RSAM (Real-time Seismic Amplitude Measure-ment) seismic amplitudes. Within a single lahar event, SSAM (SpectralSeismic Amplitude Measurement) signals are higher at the flow frontthan at the flow tail for a given discharge. The best frequency band formonitoring Merapi lahars ranges from 8.75 to 9.25 Hz (Lavigne et al.,2000b). The observations at Pinatubo showed that the energy fromlahars was concentrated within 10–100 Hz (Marcial et al., 1996).

Our paper discusses the general characteristics of seismic signalsproduced by these volcanic flows at Volcán de Colima, México, duringthe 2004–2006 activity associated with and following the extrusion ofandesitic block-lava during October–November 2004 (Zobin et al.,2006a,b). This volcano is the most active in México and its recentactivity produces numerous examples of the both types of volcanicflows (Davila et al., 2007; Zobin et al., 2005).

2. The nature of volcanic flows and their modeling

Pyroclastic flows and debris flows (lahars) are characterized bydifferent mechanical parameters. The equivalent friction coefficientsH/L, whereH is the difference between theheights of the initial andfinalpoints in the flow and L is the length of the run-out, in pyroclastic flowsare systematically higher than for the same volume ofmaterial in debrisflows; this indicates a larger mobility of lahars (Takahashi, 2007).

2.1. Pyroclastic flows

Pyroclastic flows (or pyroclastic density currents) are the inflatedmixtures of hot volcanic particles and gas that flow in variable con-centrations and varying velocity along the ground (Schminke, 2004).

Fig. 1. Position of Volcán de Colima (VdeC) within the TransMexican Volcanic Belt (A) and theare shown by open diamonds. In B, the red areas show the zones of propagation of pyroclasLumbre and Montegrande, respectively. Seismic station EZ5 is shown as a star. The map of vothe WEB site of Colima Volcano Observatory (http://www.ucol.mx/volcan).

They may be generated by three main processes (Francis, 1993): thegravitational collapse of a growing lava dome (Merapi type), explosivedestruction of growing lava dome (Peléean type) or the collapse froman eruption column (Soufrière type). Pyroclastic flows are distin-guished from rockfalls by their larger size, longer run-out (≥0.5 km),greater production of fine ash and appreciable buoyant hot ash clouds(Calder et al., 2002). Small-volume pyroclastic flows are named block-and-ash flows (Freundt et al., 2000).

Modeling of the seismic signals produced by pyroclastic flowsfollows the three elements noted by Takahashi (2007): fall, flow andslip. The first element is variable: it may be the fall of the massivesediments caused by the collapse of a lava dome segment or the fall ofthe rocks from a volcanic eruption column. The flow and slip elementswould be similar.

The observations of the generation of pyroclastic flows at Unzenvolcano, Japan (Yamasato, 1997; Uhira et al., 1994) demonstrated threestages in the formation of a pyroclastic flow with a volume equal to5.7×104m3, recorded by video and seismic instruments: the lava domecollapse, the downfall of lava blocks onto the slope, and the generationof the actual pyroclastic flow. This process may be described by thefunction representing a single-force system (Uhira et al.,1994). The useof the shape of this function for the construction of a seismic sourcetime function allows the calculation of synthetic seismograms similarto the observed seismic signals produced by the pyroclastic flow.

2.2. Lahar

A lahar, or volcanic mudflow, is a turbulent-muddy debris flow or ahyper-concentrated flood flow on volcanic slopes irrespective oftemperature, in which heavily ejected fine ash or pyroclastic materialsare the main constituent. A lahar begins its movement during the firstkilometers as a debris flow but then transforms into a hyper-con-centrated flood flow. The key factor in the generation of lahar is therainfall intensity (Takahashi, 2007). Floods of water moving across theeasily erodable, loose clastic sediments common on the flanks andaprons of volcanoes easily incorporate that debris andmay quickly bulkup to form lahars. Lahars of this type are commonly small but frequent inoccurrence during rainy periods (Valance, 2000). Lahars may be alsoformed directly during volcanic eruptions through the emptying craterlakes or on snow or glacier-clad volcanoes (Schminke, 2004).

Because lahars are unsteady and non-uniform, owing to particlesegregation processes, no simple mechanical model can be success-fully applied to these flows (Valance, 2000). Nevertheless, somemodels developed for the turbulent-muddy and stony debris flowsmay be applied to modeling of lahars (Takahashi, 2007).

3. Volcán de Colima, México and its volcanic flows

The andesitic, 3860-m high, stratovolcano Volcán de Colima is themost active volcano in Mexico. It is located in the western part of theMexican Volcanic Belt, and together with the Pleistocene volcanoNevadodeColima, forms theColimaVolcanic Complex (Fig.1). Volcán deColima displays a wide spectrum of eruption styles, including smallhreatic explosions,major block-lavaeffusions, and large explosive eventsfollowed by pyroclastic flows and lahars (Breton González et al., 2002).

3.1. Recent activity of Volcán de Colima

The recent unrest at Volcán de Colima began on November 28,1997with a sharp increase in seismic activity and a significant shortening ofgeodetic lines around the volcano. This then developed into three

extract of the map of volcanic hazards of Volcán de Colima (B). In A, the active volcanoestic flows; the blue areas are zones of lahar propagation. 1 and 2 mark the ravines of Lalcanic hazards of Volcán de Colima, prepared by C. Navarro and A. Cortés, is taken from

Table 1Seismic characteristics of pyroclastic flows and lahars

Date, yyyy_mmddhh Type of seismicrecord

Duration of thebroad band signal, s

Spectral peakfrequency, Hz

2005_06051920 Explosion+FP 362 1.32005_05300826 Explosion+FP 397 0.82005_05160201 Explosion+FP 422 12005_05101415 Explosion+FP 439 1.32005_03132128 Explosion+FP 282 1.22005_06051920 FP_ex 42005_05300826 FP_ex 4.12005_05160201 FP_ex 3.22005_05101415 FP_ex 3.62005_03132128 FP_ex 3.32004_10021715 FP_col 214 3.92004_10010100 FP_col 253 2.62004_10030816 FP_col 328 3.92004_10011405 FP_col 220 2.42004_09300610 FP_col 250 6.32005_06300500 Lahar_MG 5029 6.72005_07070014 Lahar_MG 5100 3.62005_08032330 Lahar_MG 4925 6.62006_08132025 Lahar_MG 3286 6.62006_08082000 Lahar_MG 3100 6.82006_06272030 Lahar LL 6812 7.32005_07292313 Lahar LL 5015 7.22006_06220130 Lahar LL 2953 7.12006_07150414 Lahar LL 3123 72006_06052000 Lahar LL 4347 7.6

Note: Explosion+FP means the seismic record of the explosion and the following PF;FP_ex, seismic record of the PF following the explosion; FP_col, seismic record of the PFfollowing the partial collapse of lave dome; Lahar_MG, seismic record of the lahar thatwent along the Montegrande raven; Lahar LL, seismic record of the lahar that wentalong the La Lumbre raven.

160 V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

stages of activity during the period 1998–2007. Each of these stagesconsisted of the extrusion of andesitic (SO2 between 59–61%; Savovet al., 2008) block-lavawith the formation of a lava dome andflows andthe following destruction of the lava dome by a sequence of volcanicexplosions (Zobin et al., 2002, 2006a,b).

3.2. Pyroclastic flows at Volcán de Colima

Pyroclasticflows atVolcán deColimaoccur as the result of the partialcollapse of a lava dome or the collapse of a volcanic column (Saucedoet al., 2002, 2005;Macías et al., 2005). Theseflowsproduce anunweldeddeposit with clasts embedded in a silty–sandy matrix (Davila et al.,2007). During recent activity, the pyroclastic flows entered the Colimaravines (Fig. 1B) and reached distances up to 5–6 km from the summit(Saucedo et al., 2002; Davila et al., 2007). Regardless of the mechanismby which pyroclastic flows are generated, each segregates into a basalavalanche that moves by granular flow, and an upper cloud in whichparticles are in dilute turbulent suspension (Saucedo et al., 2002). Thevelocity of their movement has reached 80–90 km/h (Navarro-Ochoaet al., 2002). During the 1998–2007 activity, the total volume of thedeposits of pyroclastic flows was estimated to be about 7.4×106 m3

(Zobin et al., 2008a).

3.3. Lahars at Volcán de Colima

Lahars at Volcán de Colima occur during the rainy season, betweenJune and October (Davila et al., 2007). They are formedwhen the run offrain water has saturated the fine and coarse unconsolidated deposits ofpyroclasts and ash accumulated during the eruptive activity. Recent

Fig. 2. The episode of the 2004 extrusion of andesitic block-lava at Volcán de Colimaand associated events. A, variations in the rate of lava extrusion (Zobin et al., 2006).B, seismic activity associated with the occurrence of pyroclastic flows and rockfalls(PF&RF) and with small gas-and-ash explosions (S_EX). The period of large explosions isshown. C, occurrence of lahars.

lahars at Volcán de Colima went along the ravines; more frequentlyalong the ravines of Montegrande and La Lumbre, and their continua-tions (Fig.1B). These two ravines differ in their morphology and activity.La Lumbre contains a perennial stream; whilst Montegrande is filledduring rainy seasons only. La Lumbre is characterized by amore straightdrainage system compared to the strongly curved Montegrande ravine.Both ravines are characterized by the vertical sides with widthschanging from about 10 m (gradient about 32°) at a distance of about1.6 km from the crater of Colima volcano to about 20 m (gradient about19–7°) at a distance of about 10 km. Then their channels widen from40m to 100m (Davila et al., 2007; Abel Cortes, personal communication,2008). The peak flow velocity of some lahars was estimated from videorecords as about 10 m/s. They reach about 8–12 km from the volcaniccone. A rain gauge, installed in 2006 near the Montegrande ravine at analtitude of 2,450m a.s.l., showed that the triggering of lahars occurred ifthe rain intensity reached 60–120 mm/h.

3.4. Subject of study

Here we discuss the most recent 2004–2006 activity that followedthe extrusion of andesitic block-lava in October–November 2004(Zobin et al., 2006b, 2008a). Fig. 2 demonstrates the development ofthis activity. The lava extrusion generated numerous pyroclastic flowsand rockfalls in October 2004 due to partial collapses of the lava dome.The next, and more voluminous, portion of pyroclastic flows wasproduced by the fall of volcanic columnmaterial during the large 2005explosions. The lahars were observed during the rainy seasons (fromJune to October) in 2005 and 2006. Their number was larger in 2005just after the lava extrusion and during the large explosions; in 2006,their number decreased with decreasing eruptive activity.

4. Seismic signals produced by the Colima volcanic flows

The seismic stations of the local seismological network (RESCO)systematically record the seismic signals produced by the Colima

Fig. 3. Seismic records of pyroclastic flows produced by the partial collapse of lava dome (station EZ5, vertical component). The index near seismic record here and in the followingfigures gives the date (yyyy_mmddhhmm) of the events. The rectangle on the seismogram of 2004_10011405 shows the selected sample for Fig. 4.

161V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

volcanicflows. In this study,weuse the records of thedigital broad-bandseismic station EZ5 (vertical component) situated at a distance of 4 kmfrom the summit (Fig. 1B). This digital seismic station was installed in2003 and allows the recording of the seismic signal related to volcanicflows without saturation.

For this study, we selected the seismic records of pyroclastic flowsassociated with the 2004 block-lava extrusion and the following 2005explosive activity as well as the 2005–2006 lahars. We analyze thewaveforms and spectral characteristics of the signals. We selected 20

Fig. 4. The structure of seismic record of pyroclastic flow (station EZ5, verticalcomponent). The sample was taken from the seismogram of 2004_10011405 (Fig. 4; theselected sample is shown as a rectangle). Arrows with numbers shows the arrivals ofseismic impulses (see text).

representative broadband seismic records, listed in Table 1, whichincludes 5 records of pyroclastic flows generated by the partialcollapse of a lava dome, 5 records of pyroclastic flows generated by thecollapse of volcanic material from an eruption column, 5 records oflahars passing along the La Lumbre ravine and 5 records of laharspassing along the Montegrande ravine.

4.1. Waveforms

4.1.1. Pyroclastic flows generated by the partial collapse of a lava domeThe waveforms (Fig. 3) represent rather similar records that begin

with a low amplitude signal which gradually increases in its amp-litude and then more slowly decreases. Fig. 4 shows the details of theinitial part of the record. We have an initial relatively short-periodimpulse (1) of duration of about 2 s, then a long-period impulse (2) ofduration of about 9 s, and then a long short-period impulse (3). Acomparison with the seismograms and video images observed for theUnzen volcano pyroclastic flows (Yamasato, 1997) showed that thefirst impulse may be produced by the lava dome collapse, the secondcorresponds to the downfall of the lava blocks onto the slope andthe third is associated with the generation of pyroclastic flows. Thedurations of the first two impulses are comparable with thosereported for the Unzen pyroclastic flows (Yamasato, 1997).

4.1.2. Pyroclastic flows generated by eruption column collapseFig. 5 shows the seismic records of this typeof pyroclasticflowswhich

are more complicated. They begin with the large-amplitude records ofthe explosions; then the collapse of volcanic material from an eruption

Fig. 5. Seismic records of explosions and the following pyroclastic flows produced by the collapse of volcanicmaterial from the volcanic column (station EZ5, vertical component). Theheavy-line rectangle on the seismogram of 2004_10011405 shows the selected sample for Fig. 6. The thin-line rectangles show the selected samples for Fig. 7.

Fig. 6. The structure of seismic record of explosion and the following pyroclastic flow(station EZ5, vertical component). Arrowswith number shows the arrivals of the seismicimpulses (see text).

162 V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

columnwhich generates an associated pyroclastic flow. The record of anexplosion has significantly larger amplitudes than the signal of apyroclastic flow. The comparative analysis of the video and seismicrecords of an explosion (Zobin et al., 2008b) showed that the initial 15–20 s of seismic record describe themovement of the fragmentedmagmain the conduit and the explosionwithin the shallowconduit. Only thendowehave the superficialmanifestation of the explosion. The fall of volcanicrocks from the eruptive column begins, therefore, 20–25 s after thebeginning of the seismic record. The detailed structure of the seismicrecord “explosion—pyroclasticflow” is shown in Fig. 6.Wecan identifyatleast 5 impulses. As it was proposed in (Zobin et al., 2006b), the first twoimpulses are generated by the fragmented magma movement from thefragmentation surface in the shallow conduit (1) and by an explosion intheconduit (2). The impulses3, 4 and5are equivalent to the impulses1–3in Fig. 4 for the pyroclastic flows originating from a lava dome collapse:impulse 3 is generated by the rockfall from the eruptive column, impulse4 corresponds to themovementof the rocks along the slope and impulse5 is generated by the pyroclastic flow. The waveform of the impulse 5 ismore heterogeneous as compared to the impulse 3 in Fig. 4. The con-tinuous fall of the rocks from the column and their followingmovementadds the low-frequency component in this impulse.

Fig. 7 shows the 100-s-length parts of seismic records of Fig. 6without their explosive parts. To remove the low-frequency noise,they were high-pass filtered at 2.5 Hz. After filtering, they are similarenough to the records of pyroclastic flows generated by the collapse ofa lava dome. At the same time, their amplitudes are one order higherthan for the seismic records of the block-and-ash flows initiated by thepartial collapse of a lava dome.

4.1.3. LaharsFig. 8 shows one-hour seismic records produced by lahars passing

along the ravines La Lumbre (Fig. 8A) andMontegrande (Fig. 8B). It is seenthat both groups of the records are similar in their waveforms. Theyrepresent a sequence of impulses that gradually increase in time, reachingtheirmaximumamplitude, and thengradually, butmore slowly, decrease.

The detailed analysis of the initial records of lahars is practicallyimpossible due to the large noise that hides the small-amplitude initialimpulses produced by the initial stage of a lahar. Nevertheless, it ispossible to separate some groups of impulses. For the case of the2006_07150414 La Lumbre lahar (Fig. 9A), the first low amplitude

Fig. 7. Seismic records (samples in rectangles selected in Fig. 5) of pyroclastic flows following explosions and produced by the collapse of volcanic material from an volcanic column(station EZ5, vertical component).

163V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

impulse has a duration of about 8 min. The second impulse sharplyarrived; it has larger initial amplitude than the first impulse. Thesecond impulse gradually decreased during about 40–50 min. Therecord of the 2005_0506300500 Montegrande lahar (Fig. 9B) consistsof at least three impulses with increasing amplitude. The final longimpulse 3 is similar to the final long impulse 2 of the La Lumbre event.The presence of the impulsive large-amplitude long-duration finalimpulses may indicate a change in the nature of lahars during theirmovement or a change in their path conditions.

4.2. Spectral characteristics

4.2.1. Pyroclastic flowsFig. 10 shows the Fourier spectra calculated for the broad-band

seismic records of two types of pyroclastic flows. There are three groupsof spectra: the spectra of the seismic records including the record ofexplosion and the following pyroclastic flow (Explosion+pyroclasticflow) and of the pyroclastic flow only (pyroclastic flow_explosion) andthe spectra of the seismic records of pyroclastic flows initiated by thepartial lava dome collapse (pyroclastic flow_collapse). It is seen that thespectra of both types of pyroclastic flows are similar with close peakfrequencies (between 2.4 and 4.4 Hz). At the same time, the spectra ofthe “total explosion+pyroclastic flow” records show the peak frequen-cies in the range of low-frequencies (around 1 Hz).

4.2.2. LaharsFig. 11 shows the Fourier spectra calculated for the broad-band

seismic records of two groups of lahars generated along two ravines,

La Lumbre andMontegrande. They are similar; no special features dueto different relief of the ravines are seen. Their peak frequencies arebetween 6.3 and 7.6 Hz.

4.3. Comparison of the seismic characteristics of the pyroclastic flowsand lahars

Fig.12 represents the seismic characteristics of pyroclastic flow andlahars in terms of the relationships “duration of seismic signal vsspectral peak frequency”. It is clearly seen that the group ofcharacteristics for each type of volcanic flow are separated. Thesecond-to-minute durations of the seismic signals of pyroclastic flowsignificantly differ from the ten-minute-to-hours durations of theseismic signals of lahars. The peak frequencies of Fourier spectra ofpyroclastic flows are significantly lower than the same of lahars.

5. Results and discussion

Our study allows us to define the main characteristics of seismicsignals produced by the movement of lahars and pyroclastic flowsoriginating at Volcán de Colima:

1. The seismic records of pyroclastic flows generated by the collapseof a lava dome and from the collapse of an eruptive column have asimilar spectral content.

The similarity in the spectral content of both types of pyroclasticflows shows that the mechanical model for the pyroclastic flowsgenerated by the rocks falling from the eruptive column principally

Fig. 8. Seismic records of lahars propagating along the La Lumbre (A) and Montegrande (B) ravines (station EZ5, vertical component).

164V.M

.Zobinet

al./Journal

ofVolcanology

andGeotherm

alResearch179

(2009)157

–167

Fig. 9. The structure of the seismic records of lahars (station EZ5, vertical component).Arrows with number shows the arrivals of the seismic impulses (see text).

Fig. 11. The spectra of the seismic records of lahars propagating along the La Lumbre(thin lines) and Montegrande (heavy lines) ravines.

165V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

has no difference with the mechanical model for the pyroclastic flowsgenerated by the lava dome collapse. The only difference is in thenature of the first impulse but then we have the same stages of agranular flow, which later changes to a fluidized flow.

2. The seismic records of pyroclastic flows and lahars are similar intheir shapes but significantly differing in their durations andspectral content. This observation allows a simple recognition ofthese events in real time.

Fig. 10. The spectra of the seismic records including the record of explosion and thefollowing pyroclastic flow (Explosion+PF; thin lines) and of a pyroclastic flow only(PF_explosion; heavy black lines) and the spectra of the seismic records of pyroclasticflows initiated by the partial lava dome collapse (PF_collapse; heavy grey lines).

The duration of the broad-band seismic records of pyroclasticflows continues for seconds or a few minutes while the duration ofseismic records of lahars continues for tens of minutes or hours.Therefore, a spindle-shape signal completely recorded on a seismo-gram during the first three–five minutes would indicate a pyroclasticflow but if a signal gradually increases in amplitude during the sametime interval, it may refer to the beginning of lahar movement. It maybe used for early diagnostics of the type of volcanic flow in real time.

3. The high-frequency content of the seismic records of lahars andlow-frequency content of the seismic records of pyroclastic flowsindicate the different mechanism of their movement.

The difference in the spectral content may be explained in terms ofthe mechanical models proposed by Takahashi and Tsujimoto (2000)

Fig. 12. Distribution of the seismic characteristics of lahars propagating along the LaLumbre (black stars) and Montegrande (grey stars) ravines; of explosions with thefollowing pyroclastic flow (grey circles) and of pyroclastic flows initiated by the partiallava dome collapse (black circles).

Fig. 13. The mechanical models of pyroclatic flows (A) and lahars (B). Details in text.

166 V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

for pyroclastic flows and by Takahashi and Satofuka (2002) for lahars.They are shown in Fig. 13. The model of Takahashi and Tsujimoto(2000) illustrates the process consisting of the following stages:collapse of lava dome (or the rocks from the eruptive column) inducesa granular flow, which later changes to a fluidized flow (Fig. 13A). As itwas shown by Yamasato (1997) and noted in our study, the main partof the seismic signal is produced by the fluidized flow stage ofpyroclastic flows.

Takahashi and Satofuka (2002) proposed a lahar as a hybrid-typedebris flow that would be between a turbulent-muddy and stonydebris flows. The hybrid flow in the inertial range consists of twolayers: the upper layer of depth h1 is the particle suspension layer, andin the lower depth h2 layer particles move with frequent collision orenduring contact with each other (Fig.13B). The dominance of stony orturbulent-muddy type of flow depends on the relative thickness ofthese two layers h1 / h2. The movement of the lower stony layer alongthe volcanic slope ravine (angle of inclination θ in Fig. 13B) producesthe main part of the seismic signal of lahar.

4. The durations and spectral content of seismic records produced bypyroclastic flows and lahars at Volcán de Colima are similar to thesame previously mentioned for other volcanoes of the world.

Acknowledgements

The information and comments provided by Abel Cortés, JuanCarlos Gavilanes and Nicholas Varley were very valuable. Thecomments of two anonymous reviewers helped to improve themanuscript. Miguel Gonzalez helped uswith the acquisition of seismicdata. The processing of the short (up to 150 s) digital signals wasrealized using the program DEGTRA provided by Mario Ordaz; for theprocessing of the longer digital signals it was used the programprovided by Philippe Lesage and modified by Miguel Gonzalez.

Nicholas Varley strongly improved our English grammar. This studywas partially supported by the European Commission, 6th FrameworkProject — ‘VOLUME’, Contract No. 08471.

References

Bardintzeff, J.-M., McBirney, A.R., 2000. Volcanology, Second ed. Jones & Bartlett Publ.,Sudbury. 268 pp.

Breton González, M., Ramírez, J.J., Navarro, C., 2002. Summary of the historical eruptiveactivity of Volcán de Colima, Mexico 1519–2000. J. Volcanol. Geotherm. Res. 117,21–46.

Brodscholl, A., Kirbani, S., Voight, B., 2000. Sequential dome-collapse nuées ardentesanalysed from broadband seismic data, Merapi Volcano, Indonesia. J. Volcanol.Geotherm. Res. 100, 363–369.

Calder, E.S., Luckett, R., Sparks, R.S.J., Voight, B., 2002. Mechanisms of lava domeinstability and generation of rockfalls and pyroclastic flows at Soufrére Hillsvolcano, Montserrat. In: Druitt, T.H., Kokelaar, B.P. (Eds.), The eruption of SoufriéreHills volcano, Montserrat, from 1995 to 1999. Geol. Soc. London, Memoirs, vol. 21,pp. 173–190.

Davila, N., Capra, L., Gavilanes-Ruiz, J.C., Varley, N., Norini, N.G., Gómez-Vazquez, A.,2007. Recent lahars at Volcán de Colima (Mexico): drainage variation and spectralclassification. J. Volcanol. Geotherm. Res. 165, 127–141.

Francis, P., 1993. Volcanoes. A Planetary Perspective. Clarendon Press, Oxford. 443 pp.Freundt, A., Wilson, C.J.N., Carey, S.N., 2000. Ignimbrites and block-and-ash flow

deposits. In: Sigurdsson, H. (Ed.), Encyclopedia of Volcanoes. Academic Press, SanDiego, pp. 581–599.

Lavigne, F., Thouret, J.C., Voight, B., Suwa, H., Sumaryono, A., 2000a. Lahars at Merapivolcano, Central Java: an overview. J. Volcanol. Geotherm. Res. 100, 423–456.

Lavigne, F., Thouret, J.C., Voight, B., Young, K., LaHasen, R., Marso, J., Suwa, H.,Sumaryono, A., Sayudi, D.S., Dejean, M., 2000b. Instrumental lahar monitoring atMerapi volcano, Central Java, Indonesia. J. Volcanol. Geotherm. Res. 100, 457–478.

Macías, J.L., Saucedo, R., Gavilanes, J.C., Varley, N., Velasco-García, S., Bursik, M., Vargas-Gutiérrez, V., Cortés, A., 2005. Flujos piroclásticos asociados a la actividad explosivadel Volcán de Colima y perspectivas futuras. GEOS 25, 340–351.

Marcial, S., Melosantos, A.A., Hadley, K.C., LaHusen, R.G., Marso, J.N., 1996. Instrumentallahar monitoring at Mount Pinatubo. In: Newhall, C.G., Punongbayan, R.S. (Eds.),Fire and Mud. Eruptions and lahars of Mount Pinatubo, Philippines. University ofWashington Press, Seattle. Pp.

Navarro-Ochoa, C., Gavilanes-Ruíz, J.C., Cortés-Cortés, A., 2002. Movement andemplacement of lava flows at Volcán de Colima, Mexico: November 1998–February1999. J. Volcanol. Geotherm. Res. 117, 155–167.

Pierson, T., Daag, A.S., Delos Reyes, P.J., Regalado, M.T., Solidum, R.U., Tubianosa, B.S.,1996. Flow and deposition of posteruption hot lahars on the east side of MountPinatubo, July–October 1991. In: Newhall, C.G., Punongbayan, R.S. (Eds.), Fire andMud. Eruptions and lahars of Mount Pinatubo, Philippines. University ofWashington Press, Seattle, pp. 921–950.

Purbawinata, M.A., Ratdomopurbo, A., Sinulingga, I.K., Sumarti, S., Suharno (Eds.), 1997.Merapi Volcano. A guide book. The Volcanological Survey of Indonesia, Bandung.64 pp.

Ratdomopurbo, A., Poupinet, G., 2000. An overview of the seismicity of Merapi volcano(Java, Indonesia). J. Volcanol. Geotherm. Res. 100, 193–214.

Rodolfo, K.S., Umbal, J.V., Alonso, R.A., Remotigue, C.T., Paladio-Melosantos, M.L.,Salvador, J.H.G., Evangelista, D., Millar, Y., 1996. Two years of lahars on the westernflank of Mount Pinatubo: initiation, flow processes, deposits, and attendantgeomorphic and hydraulic changes. In: Newhall, C.G., Punongbayan, R.S. (Eds.), Fireand Mud. Eruptions and lahars of Mount Pinatubo, Philippines. University ofWashington Press, Seattle, pp. 989–1014.

Saucedo, R., Macías, J.L., Bursik, M., Mora, M., Gavilanes, J.C., Cortés, A., 2002.Emplacement of pyroclastic flows during the 1998–1999 eruption of Volcán deColima, Mexico. J. Volcanol. Geotherm. Res. 117, 129–153.

Saucedo, R., Macías, J.L., Sheridan, M.F., Bursik, M.I., Komorowski, J.C., 2005. Modeling ofpyroclastic flows of Colima volcano, México: implications for hazard assessment.J. Volcanol. Geotherm. Res. 139, 103–115.

Savov, I.P., Luhr, J.F., Navarro-Ochoa, C., 2008. Petrology and geochemistry of lava andash erupted from Volcán de Colima, México, during 1998–2005. J. Volcanol.Geotherm. Res. 174, 241–256.

Schminke, H.-U., 2004. Volcanism. Springer-Verlag, Berlin-Heidelberg. 324 pp.Takahashi, T., 2007. Debris Flow. Taylor & Francis, London. 448 pp.Takahashi, T., Tsujimoto, H., 2000. Amechanical model forMerapi-type pyroclastic flow.

J. Volcanol. Geotherm. Res. 98, 91–115.Takahashi, T., Satofuka, Y., 2002. Generalized theory of stony and turbulent muddy

debris flow and its practical model. J. Jpn. Soc. ECE 55, 33–42.Tuñgol, N.M., Regalado, M.T.M., 1996. Rainfall, acoustic flow monitor records, and

observed lahars of the Sacobia River in 1992. In: Newhall, C.G., Punongbayan, R.S.(Eds.), Fire and Mud. Eruptions and lahars of Mount Pinatubo, Philippines.University of Washington Press, Seattle, pp. 1023–1032.

Uhira, K., Yamasato, H., Takeo, M., 1994. Source mechanism of seismic waves excited bypyroclastic flows observed at Unzen volcano, Japan. J. Geophys. Res. 99, 17757–17773.

Valance, J.W., 2000. Lahars. In: Sigurdsson, H. (Ed.), Encyclopedia of Volcanoes.Academic Press, San Diego, pp. 601–616.

Yamasato, H., 1997. Quantitative analysis of pyroclastic flows using infrasonic andseismic data at Unzen volcano. Jpn. J. Phys. Earth 45, 397–416.

Zobin, V.M., 2003. Introduction to Volcanic Seismology. Ámsterdam, Elsevier. 203 pp.Zobin, V.M., Luhr, J.F., Taran, Y.A., Bretón, M., Cortés, A., De la Cruz-Reyna, S., Domínguez,

T., Galindo, I., Gavilanes, J.C., Muñiz, J.J., Navarro, C., Ramírez, J.J., Reyes, G.A., Ursúa,

167V.M. Zobin et al. / Journal of Volcanology and Geothermal Research 179 (2009) 157–167

M., Velasco, J., Alatorre, E., Santiago, H., 2002. Overview of the 1997-2000 activity ofVolcán de Colima, Mexico. J. Volcanol. Geotherm. Res. 117, 1–19.

Zobin, V.M., Orozco-Rojas, J., Reyes-Dávila, G.A., Navarro, C., 2005. Seismicity of anandesitic volcano during block-lava efusión: Volcán de Colima, México, November1998-January 1999. Bull. Volcanol. 67 (No 7), 679–688.

Zobin, V.M., Navarro-Ochoa, C.J., Reyes-Davila, G.A., 2006a. Seismic quantification of theexplosions that destroyed the dome of Volcan de Colima, Mexico, in July-August2003. Bull. Volcanol. 69, 141–147.

Zobin, V.M., Navarro, C., Reyes-Dávila, G., Orozco, J., Bretón, M., Tellez, A., Reyes-Alfaro,G., Vázquez, H., 2006b. The methodology of quantification of volcanic explosions

from broadband seismic signals and its application to the 2004-2005 explosions atVolcán de Colima, México. Geophys. J. Int. 167 (No 1), 467–478.

Zobin, V.M., Varley, N.R., González, M., Orozco, J., Reyes, G.A., Navarro, C., Bretón, M.,2008a. Monitoring the 2004 andesitic block-lava extrusion at Volcán de Colima,México from seismic activity and SO2 emission. J. Volcanol. Geotherm. Res. 177,367–377.

Zobin, V.M., Reyes, G.A., Guevara, E., Breton, M., 2008b. Seismological constraints on theposition of the fragmentation surfaces in the volcano conduit. Earth Planet. Sci. Lett.275, 337–341.