- Email: [email protected]
Mayon volcano, Philippines: change of monitoring strategy after microgravity and GPS measurements from 1992 to 1996
Journal of Volcanology and Geothermal Research 109 (2001) 219±234 www.elsevier.com/locate/jvolgeores
Mayon volcano, Philippines: change of monitoring strategy after microgravity and GPS measurements from 1992 to 1996 Gerhard Jentzsch a,*, Raymondo S. Punongbayan d, Ulrich Schreiber c, GuÈnter Seeber b, Christoph VoÈlksen b, Adelheid Weise a a Institute for Geosciences, FSU Jena, Burgweg 11, D-07749 Jena, Germany Geodetic Institute, UniversitaÈt Hannover, Schneiderberg 50, D-30167 Hannover, Germany c Department of Geology, UniversitaÈt Essen, UniversitaÈtsstr. 5, D-48117 Essen, Germany d Philippine Institute of Volcanology and Seismology, C.P. Garcia Avenue, University of the Philippines Campus, Diliman, Quezon City, Philippines b
Received 21 January 1999; revised 5 December 1999; accepted 3 January 2000
Abstract Mayon volcano is part of the Bicol volcanic chain on the island Luzon, Philippines. During this century there were ten activity periods distributed almost regularly. Because of the density of population (about one million people living in the vicinity of the volcano) three seismological observatories are in operation. Measurements of gravity changes started in 1992, just before the eruption of February/March 1993. Two pro®les at the slope were established, connected to a regional network around the volcano. In order to enable the determination of mass changes between the campaigns the height control was provided by parallel GPS measurements. In all, the network consists of 26 points which were remeasured with three gravimetres at least three times within one campaign. During ®ve campaigns within 4 years the differential GPS gives no signi®cant changes of the elevation (within ^4 cm). Nevertheless, the gravity increased signi®cantly by up to 1500 nm/s 2 (equivalent to 150 mGal). As no signi®cant change of elevation is observed (GPS), no extended shallow magma chamber system below the volcano can be proved. This is in accordance with geochemical results indicating a rather undifferentiated magma. The youngest lava which is of interest for the eruption dynamics belongs to the medium-K basaltic andesite ®eld of the K2O vs SiO2 diagram. A rather qualitative check of groundwater level changes reveals that these cannot be the sources for the observed gravity changes. Thus, the increase of gravity after the eruption of February 1993 can be explained by a mass redistribution in the volcanic vent from above the level of the gravity points to below. Practical conclusions of these results lead to changes in the monitoring strategy: Deformation measurements did not reveal any volcanic activities; at least for the eruption of 1993 no signi®cant deformation was observed. Gravity could be an indicator for long-term changes. Thus, repeated gravity measurements/GPS at selected points could be used in parallel to seismic monitoring to detect slow mass movements prior to changes in seismicity. q 2001 Elsevier Science B.V. All rights reserved. Keywords: physical volcanology; microgravity; GPS; petrology; volcanic modeling
1. Introduction * Corresponding author. Tel.: 149-3641-948660; fax: 149-3641948662. E-mail address: [email protected] (G. Jentzsch).
The Philippines, situated in the western Paci®c Ocean, are part of an extensive island arc system.
0377-0273/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S 0377-027 3(00)00313-9
220
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 1. Location of Mayon volcano, Luzon island, Philippines, after Ramos-Villarta et al. (1985).
The geological and geophysical situation is mainly characterized by two subduction zones. These collisions yield a complex tectonic situation with main faults and lineaments extending through the whole crust down to the upper mantle (Bischke et al., 1990; Aurelio, 1992). The active tectonics cause earthquakes and volcanic activities. Mayon is the most famous volcano of the Bicol volcanic chain in the south-east of Luzon (138 15.4 0
N, 1238 41.1 0 E, Fig. 1). Its shape is a nearly perfect cone with an altitude of 2462 m. south-west of Mayon, rather close, the Legaspi Lineament runs NW±SE across Legaspi City. According to Aurelio (1992) probably it could partly control the activity of Mayon volcano. There were ten activity periods spread almost regularly over the past century. The last important eruption was in 1984. Compared to this, some smaller eruptions took place in February/March 1993,
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
just between the ®rst and second campaigns of our gravity measurements. They resulted in ash fall and a lava ¯ow at the south-east ¯ank with a volume of about 10 Mio m 3. The decrease of activity after the eruptions of Mayon usually is rather slow; after 1 year crater glow still may be observable. More than one million people live in the vicinity of Mayon volcano. Legaspi City is situated only 12 km away from the crater. Thus, the improvement of hazard assessment requires a better understanding of the physical properties of the volcano. For the monitoring of Mayon three seismological observatories were installed (Punongbayan et al., 1990), and in previous scienti®c investigations the volumes of ejecta were evaluated and the emission cloud analysed (Ramoz-Villarta et al., 1985). Correlations with earth tides were examined (Ramoz et al., 1985). As the ocean loading tide around the Philippines varies strongly, we proposed that the suspected increase of seismicity is connected with the strain induced (Jentzsch, 1995; Jentzsch et al., 1997, 2001). Until the eruption of 1993 the seismic monitoring at the three observatories lacked optimum conditions: The equipment was not homogeneous and the noise levels were very high (except for Mayon Resthouse observatory). Therefore, it was dif®cult to detect precursors. In the present paper we discuss our results from microgravity and GPS measurements in relation to geochemical analyses of samples of the volcanics carried out by co-author Schreiber. An additional contribution came from C. Newhall (1998, personal communication), who proposes groundwater changes as the reason for the observed gravity variations. The aim of our work is to interpret all the results regarding the development of a model of the volcano to improve the understanding of the physics of further eruptions. First results were already presented by Jahr et al. (1998). 2. Microgravimetry in volcanic areas Some examples of sub-surface magma movements are documented using high precise gravimetry and concurrent elevation control, for instance at Kilauea/ Hawaii (Dzurisin et al., 1980), Pacaya/Guatemala (Brown & Rymer, 1991), and at Usu/Japan (Jousset
221
& Okada, 1999). Rymer (1996) reported gravity changes at Po'as volcano/Costa Rica, that uniquely preceded an eruption. The connection between the variation of gravity and elevation due to mass and/or density changes is given by the two well known gradients: If gravity is following the free air gradient (FAG), no sub-surface mass changed, while data following the gradient related to the standard Bouguer correction (BCFAG) imply mass changes. Departures from these gradients due to mass and/or density variations enable modelling of the volcanic process (Brown & Rymer, 1991; Brown at al., 1989; Brown et al., 1991; Jahr et al., 1995). The desired accuracy of the gravity measurements is in the order of ^100 nm/s 2 (equal to ^ 10 mGal).
3. Microgravity at Mayon volcano The measurements of repeated gravity and GPS started in November 1992. Two pro®les at the slope, towards the summit (up to 850 m) were established. They are connected to a local and a regional network (Fig. 2) around the volcano with the extension of 40 km £ 50 km. The maximum distance from the crater is 47 km. Two points at the opposite side of the Legaspi Lineament (in the south-west) are connected to the reference network. The gravity points are installed in areas of sand (lahars), pasture ground as well as basaltic rock. The total gravity range is 1800 mm/s 2. 1 They consist of iron marks cemented into drillholes in rocks or blocks of concrete. During 4 years, until December 1996, ®ve campaigns of repeated gravity measurements were carried out with three LaCoste & Romberg gravimetres of the types G and D. The instruments are equipped with electrostatic feed back systems and electronic levels. They were calibrated several times at the gravimetre calibration lines at Hannover Harz/ Germany. Even if the observed calibration factors were not valid in the used gravity range of Mayon, at least the stability of the scale factors could be checked up. Additionally, the ratio between the 1 The usual units in gravimetry are milligal and microgal; the relations to SI units are: 10 mm/s 2 1 mGal, 10 nm/s 2 1 mGal.
222
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 2. Gravity and GPS stations around Mayon volcano, the Legaspi Lineament according to Aurelio (1992): Note the circle descibing the base of the volcano and the two pro®les (Mayon Resthouse: MRH1±MRH0; Tumpa Lahar Channel: QSD1/2±TLC2/1).
scale factors of the instruments was checked to be similar at Mayon and in Germany within 1 £ 10 24. During each campaign the gravity differences were generally observed at least three times with each of the three gravimetres (minimum nine differences). The program system GRAV from G. Wenzel is used for the adjustment where a standard earth tide correction is applied with a precision better than it is required. A linear drift is adjusted (up to ^40 nm/s 2/ h) which turned out to be rather similar for each instrument in all the campaigns. The standard deviation of one observed gravity difference m0 is in the order of ^140¼150 nm/s 2. The results of each campaign are the gravity differences between the stations with an achieved accuracy
of about ^120 nm/s 2, as it was expected (Jahr et al., 1995). For the differences between the campaigns the assumption is made that gravity is constant at the most distant stations from the volcano. Between December 1992 and May 1993 no signi®cant gravity changes could be observed although Mayon was active in February/March 1993. Between May 1993 and December 1996 the increase of gravity is signi®cant, rising up to 1500 nm/s 2 (^150 nm/s 2) along the pro®les at the slope, decreasing with the distance from the crater (Figs. 3 and 4). Taking into account the dates of the individual campaigns (a) Dec. 1992, (b) May 1993, (c) May 1994, (d) Dec. 1994, (e) Dec. 1996) the observations along Tumpa Lahar Channel show a more or less steady increase in
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
223
Fig. 3. Gravity changes along the Tumpa Lahar Channel pro®le (south ¯ank) over the distance from the crater, relative to the ®rst measuring campaign, with error bars. The points from left to right are TLC1, TLC2, QSD1/2, TES, CAG, LHO (comp. Fig. 2). The campaigns are: (a) Dec. 1992, (b) May 1993, (c) May 1994, (d) Dec. 1994 and (e) Dec. 1996.
gravity of about 300 nm/s 2 per year in a rather continuous process (involving an overall error of about ^150 nm/s 2). In the network surrounding Mayon volcano (.10 km from the summit; Fig. 2) there are no signi®cant gravity variations over the whole period of the surveys; only at a few individual points at the foot of Mayon we might see gravity changes of up to 300 nm/s 2 (increase or decrease), but with regard to the errors involved these changes are not considered as signi®cant. Compared to Tumpa the picture is not quite that clear for the Mayon Resthouse pro®le: At ®rst we see that the last curve (5-1) consists of less points than the others; due to road construction works the missing points were lost. All curves show changes up to a distance from the summit of 5±6 kmÐnot signi®cant for (2-1); closer to the summit the changes are not signi®cant any more. This ®ts to the Tumpa pro®le, which ends at about 4.5 km from the crater.
The results of the third campaign are already in the order of magnitude of the ®fth, whereas the forth gives smaller changes compared to Tumpa. From the data of the third campaign of the Mayon Resthouse pro®le there is the suspicion of an additional effect which shifts the rest of the curve upwards. Due to logistics it was not possible to connect the uppermost points of both pro®les to different points of the local network: The driving or walking times would have been so long or even not manageable during daylight so that the bene®t of such a connection could not been used. Therefore, these points are only connected to the base of the pro®les. Thus, if there are any changes at the ®rst points the whole pro®le is affected. Since all the data is consistent we simply must accept this result; this points to an additional signal maybe due to a recharge of the local water table which is not proved by observational data from wells.
224
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 4. Gravity changes along the Mayon Rest House pro®le (north ¯ank) over the distance from the crater, relative to the ®rst measuring campaign, with error bars. The points from left to right are MRH0, MRH9,¼,MRH1 (comp. Fig. 2). The campaigns are indicated in Fig. 3. Note that for the last campaign only points MRH1, MRH5, MRH9, MRH0 were available.
Inspite of the problems discussed, from these results we ®nd that decreasing volcanic activities are accompanied by increasing gravity. 4. The regional GPS network at Mayon In order to interpret gravity variations in terms of mass changes between the campaigns we need the height control provided by parallel GPS measurements. In each campaign three receivers of the type Ashtech (1/2. camp.) or Trimble 4000 SSE were used. The reference network was observed in triangles, resulting in 15 baselines between the stations SC1, RNSD, TAP2, TIWI, MRH1, SFES, LHO, PES2 (Fig. 2). For the data processing the program system GEONAP (University, Hannover) is used, which is based on undifferenced observations. The lay-out of the network and ®rst results were already discussed in detail by VoÈlksen & Seeber (1995).
The disturbances due to the in¯uence of the Ionosphere are major in the equatorial regions. The ®rst two or three campaigns took place during maximum solar activity. Even with dual-frequency data serious problems can occur in the data processing (Wanninger, 1993), which is re¯ected in the higher standard deviations for the ®rst two campaigns. Concerning the meteorological correction a model with a standard atmosphere is used (Hop®eld, 1979). The residual tropospheric refraction error (Residual Zenit Delay) is compensated by a scale factor, which is estimated for each station. A test was made during the third campaign introducing observed meteorological surface data for tropospheric modeling. But this did not lead to lower standard deviations, even under the tropic conditions around Mayon. This is in agreement with Brunner & Welsch (1993). We can state after comparing these results (applying a scale factor for compensating the residual tropospheric refraction errors) with those applying a standard atmosphere
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
225
Fig. 5. Gravity changes versus height differences at Mayon volcano from one campaign to the next. The crosses are indicating the mean error bars.
only, that systematic deviations in the vertical component of up to 8 cm appear. Therefore, the same meteorological model was used in all campaigns for the data processing. The reference point for the datum is the closest IGS 2 site Taipei in Taiwan with 1300 km baseline length. The result of the GPS-measurements is: No signi®cant deformations are stated, neither in the vertical nor in the horizontal components. The standard deviations (posteriory) for one campaign vary in the range of ^ 0.5¼1.5 cm in the horizontal and ^1.8¼4 cm in the vertical component. In the horizontal component isolated changes of up to 3 cm are not signi®cant. No signi®cant changes of the elevation (Fig. 5) are stated within the error bar of around ^4 cm between two campaigns. Therefore, the gravity changes drawn versus the height differences from one campaign to the next do not show any gradients (Fig. 5, crosses denote error bars for gravity and elevation). 2
International GPS and Geodynamics Service.
5. Geochemistry and petrology In 1992 some rock samples were taken of different lava ¯ows in the area of Mayon Resthouse. After the 1993 eruption samples of different layers of the new lava ¯ow were analysed (see Table 1). The rock classi®cation used in this paper is based on Le Maitre's (1989) K2O vs SiO2 diagram for orogenic igneous rocks (Fig. 6). The volcanic products of Mayon volcano are basalts, basaltic andesites and andesites of the medium-K calc-alkaline series. The lavas show cyclical variation with up to three successive basaltic ¯ows followed by six to ten andesitic ¯ows (Newhall, 1977, 1979). The youngest lava which is of interest for the eruption dynamics plots into the medium-K basaltic and andesite ®eld. MgO does not exceed 4.7% (4.12± 4.66% MgO from base to top of the ¯ow) and Fe2O3 decreases concomitantly with MgO. The samples exhibit typical arc-related trace element characters, i.e. enrichment in large ion lithophile elements
226
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 6. K2O vs SiO2 diagram of Mayon samples. Filled symbols are samples from the 1993 lava, others from Newhall (1977).
(LILEÐRb, Ba, K, Sr) and light rare earth elements (LREE) relative to heavy REE and high ®eld strength elements (HFSEÐNi, Zr, Ti). The homogeneous trace element patterns (Fig. 7), e.g. (Ba/La) n 1.9±2.3, (La/Yb)n 4.8±5.3 (normalisation constants from Sun, 1980) have typical subductionrelated negative anomalies in Nb and Ta. Sr and Y values as well as Ti and Zr values are in the range of common arc lavas. The missing Eu-anomaly indicates that plagioclase fractionation has not taken place. The youngest product of Mayon is a strongly porphyritic lava with 30±50% of phenocrysts of mainly plagioclase (An 76±An 42), augite, hypersthene, olivine and titaniferous magnetite. The same minerals and rare sanidine microliths are part of the groundmass together with glass and accessory minerals. The low degree of fractional crystallization argues against a magma source in the upper crust. The determination of the viscosity is dif®cult, depending on the crystallization and the gas content. The youngest lava contains 30 to 50% crystals of 0.3 to 2 mm in size. Here, the phenocrysts provide a
higher percentage in the lave erupted ®rst. Applying the relation for viscosity and temperature given by Richet & Bottinga (1995) for a temperature range from 1300 to 10008C relevant for the intrusion and rise of the magma we ®nd a viscosity intervall of about 2.65±5.5 Log Pas at the pressure of 1 bar. According to Scarfe et al. (1987) and Lange (1994) the viscosity decrease with pressure is less than one order of magnitude for about 20 kbar. Another factor for controlling the viscosity is the contents of water in the melt (Stolper, 1982). According to Jaupart & Tait (1990) 2% of water are enough to reduce viscosity by one order of magnitude. This effect is greater at lower temperatures (Lange, 1994). Since vesiculation depends on the pressure, the ascent of the magma is accompanied by an increase of viscosity. Already Sibree (1934) found that more than 50% gas bubbles in the melt increase the viscosity by about one order of magnitude. To estimate the viscosity of the magma of the last eruption we assume the above mentioned temperature intervall. The content of water is unknown. From the
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
227
Fig. 7. Trace element plots of the 1993 Mayon lava normalized to primitive mantle (Thompson et al., 1984). To identify the sample numbers please compare the values given in Table 1.
character and relation of the phenocrysts we assume a pressure of less than 10 kbar and a water content of ,2 wt-%. On the basis of these conditions the viscosity should be ®rst below 500 Pas; later, during the ascent in the vent this value increases due to cooling and vesiculation to 50,000±100,000 Pas. 6. Groundwater level changes Considerable drops of the water table (in the order of 2 m) occurred around Mayon before the 1993 eruption (Newhall, 1998, pers. comm.). The supposed subsequent slow rise (recharge) should have occurred over the time span of the gravity measurements and could contribute to the gravity increase. In this case we should see variations in gravity at points of the network around Mayon, especially at the foot of the volcano, with respect to the distant reference points. C. Newhall collected the observations of local residents around Mayon, which means that no really measured data exist yet. Nevertheless, since drops of
the water table can be linked to stress and, thus, strain changes he tried to use this information. Providing an internal report (Newhall, 1993) and several personal communications C. Newhall informed us about the water level changes in 30 observation wells around Mayon (even at greater distances of 10±15 km, mostly situated in the settled areas around the volcano), and about a simple experiment to prove the hypothesis that intrusion of magma is causing a drop in the water table relative to land surface. He can show by using clay as intrusion material, that in this experiment the water table drops. The model volcano is built up from unconsolidated ®ne sand; due to the intrusion dilatation occurs and water seeps deeper into the ground. It would be interesting to see this effect for a model comparable to the real volcanic edi®ce with different more or less consolidated layers and varying properties of the uppermost 10±20 m. With this model the comparison to the real situation at Mayon is still very tentative. On the other hand, the reports provided by local residents about their observations of the water
228
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Table 1 Geochemical analyses of the 1993 lava from Mayon (basaltic andesite). Main elements (XRF-analyses) in weight%, others in ppm. XRFanalyses of Mo, Ni, Pb, V, Y. Analyses of Ba, Br, Co, Cr, Cs, Cu Ga, Hf, Rb, Sc, Ta, Th, U, Zn, Zr, and REE by NAA (Inst. f. Strahlen- und Kernphysik, Bonn; Mommsen et al., 1987)
SiO2 TiO2 Al2O3 Mn0 Mg0 Ca0 Na2O K2O P2O5 L0I Cr Ni Sc V Cu Pb Zn Mo Rb Cs Ba Sr Ga Ta Nb Hf Zr Y Th U La Ce Nd Sm Eu Tb Yb Lu Br
Mayo 95-1
Mayo 95-3
Mayo 95-5
Mayo 95-7
Mayo 95-11
54.48 0.71 18.81 0.16 4.12 8.54 3.33 1.15 0.31 0.07 5 8 22 188 813 7 87 2.02 24 0.50 457 714 19 0.74 3.9 2.97 131 22 2.46 0.80 16.58 36.37 18.98 4.32 1.33 0.63 2.15 0.44 10.33
54.42 0.72 18.48 0.17 4.30 8.47 3.35 1.16 0.30 0.00 4 7 22 189 660 5 84 2.33 25 0.36 374 702 19 0.48 4.3 2.84 136 23 2.44 0.80 15.84 34.63 22.11 4.18 1.24 0.55 2.15 0.43 4.08
54.46 0.72 18.67 0.17 4.20 8.45 3.32 1.15 0.30 0.00 10 8 22 175 671 7 88 2.43 21 0.38 431 689 18 0.52 3.8 2.71 151 24 2.43 0.70 16.13 35.33 18.05 4.20 1.27 0.62 2.15 0.56 3.76
54.48 0.71 18.66 0.17 4.21 8.47 3.31 1.17 0.30 0.07 8 8 22 190 760 8 90 2.13 24 0.48 413 710 19 0.57 3.9 2.81 127 24 2.46 0.72 16.39 35.60 20.46 4.21 1.32 0.59 2.18 0.51 3.35
53.32 0.76 18.75 0.17 4.66 8.98 3.16 1.07 0.26 0.26 9 8 27 198 624 8 99 2.43 21 0.58 362 660 17 0.33 4.2 2.75 254 22 2.34 0.62 15.55 34.96 18.02 4.26 1.30 0.59 2.26 0.47 2.39
supply certainly contain interesting information. Naturally, these reports are very inhomogeneous and many of them even vague; nevertheless, the classi®cation between `strong evidences' and `no correlation' 3 is given in Table 2 (only 28 are used 3 `correlation' denotes the coincidence of volcanic activities and drop of water level.
because two wells are too far away and do not show any changes). But, even if we do not reject those reports which are obviously exaggerated the result is not at all signi®cant, neither for `pro' nor `con'. If we analyse the reports in detail we see that some `pros' are given for wells farer away from the summit and more downslope, or people report that this drop of ground water level was never observed
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
229
Table 2 Classi®cation of 28 reports about observations regarding water level changes in wells during the 1993 eruption of Mayon (Newhall, 1998, pers. comm.) Strong evidences as seen also before previous eruptions
Correllation of drop of water level and volcanic activity
No correlation; some wells show changes, some not
No correlation, normal situation
2
12
6
8
beforeÐalthough the 1993 eruption was by far not as strong as previous ones. Inspite of these `soft' facts we followed Newhall's suggestion and we compared the data and the locations of the observation wells with our data. To put it on a more stable ground we ®rst estimate the gravity effect to be expected: If we roughly calculate the possible maximum Bouguer effect of such a water level change we arrive at 130 nm/s 2 for a change of 1 m at a porosity of 30% (for unconsolidated sand in the lahar area). Thus, if we even assume a change of 3 m, the gravity effect is in the order of ,500 nm/s 2. This is the maximum effect we could expect, and it should be observable at some of the points of the network at the foot of Mayon except at the distant reference points. Unfortunately, our gravity points and the wells do only coincide in the village Tumpa (point TES) where the ground consists of sand (lahar). At the ¯anks of the volcano there is few information about ground water variations. Upslope, the points TLC1/TLC2 (Fig. 3) are situated on consolidated rock where the observed strong gravity increase could hardly be related to ground water. Further, the steep slope of the Mayon Resthouse pro®le does not seem to allow the discussion of a prevailing water table. At least in these points we cannot assume an in®nite extension of the water table, which again reduces the estimated maximum effect. Although there are points in the lahars the data do not show the expected groundwater signal. According to our data and the localities we can only concede the small, not really signi®cant gravity variations at the foot of the Mayon Resthouse pro®le of up to 300 mm/s 2 as maybe being caused by changes of the local water level. 7. Modelling of the observed gravity changes and interpretation The observed gravity variations show a very signif-
icant distribution: They are con®ned to a radius less than 8±10 km from the summit, and they increase systematically with decreasing distances. Further, the observed maximum gravity changes are in the order of up to 1500 nm/s 2, much bigger than the estimated maximum effect of ,500 nm/s 2 due to water level changes. These observed gravity changes are not accompanied by signi®cant elevation changes as derived from GPS measurements. Usually, we expect that due to an intrusion of magma gravity is increased, and with decreasing activity gravity decreases as well. In the case of Mayon we have to explain the opposite relation, a negative correlation between gravity and volcanic activity. 4 To explain the gravity increase by applying the classical Mogi-model of a magma source below Mayon we would expect subsidence in the order of 0.5±1 m, but this is not observed. Since there are no signi®cant elevation changes, we cannot apply this model: There is obviously no extended shallow magma source which could be responsible for an uplift. This, and the shape of the gravity changes, especially the concentration of the changes around the summit points to a process taking place in the very center of the volcano. The gravity changes are restricted to the area of the volcano, increasing strongly towards the center. All this puts heavy constrains on the modelling of the physical process of this volcano. The relatively small gravity changes, especially the increase of gravity accompanying the decrease of volcanic activity, must therefore be explained by a process within the volcanic vent (and not by a shallow magma system below). This is in accordance with the geochemical analyses of rock samples indicating a rather undifferenciated magma. 4
Since our ®rst measurements were in December 1992 just before the eruption we do not have any information about development of the gravity before.
230
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Table 3 Modeled maximum gravity changes at Tumpa Lahar Channel and at Mayon Resthouse in [nm/s 2] for different model parameters of the vent: radius and density difference Model version
1
2
3
4
5
6
Max. observed
Radius [m] density difference [kg/m 3] Tumpa Lahar Channel (TLC1) Mayon Resthouse (MRH9)
200 400 155 297
200 600 233 445
300 400 349 668
300 600 524 1002
400 400 621 1188
400 600 931 1782
1579 1339
Similar results are reported by Berrino & Corrado (1991) for Vesuvius and Aeolian Islands/Italy, where the gravity changes, not accompanied by the vertical movements, are explained by mass redistribution, as well as by Eggers (1987) for andesitic volcanoes. Our observations can be explained by a process of density changes or magma movement within the
volcanic vent: Taking into account a mass redistribution in the magma column of the vent there is obviously a mass change above the gravity points (highest elevations of the gravity points are about 900 and 560 m at Mayon Resthouse and Tumpa Lahar Channel, respectively): Thus, expecting a magma drainage or density decrease with decreasing activities our concept is based on the redistribution of
Fig. 8. Modelled gravity changes along Tumpa Lahar Channel pro®le vs observation (maximum gravity changes; models 1±6 comp. Table 3).
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
231
Fig. 9. Modelled gravity changes at Mayon Rest House pro®le vs observation (maximum gravity changes; models 1±6 comp. Table 3).
mass from above the points to below causing an increase in gravity. The model is quite basic: It consists of a vertical cylinder in the place of the volcanic vent. The changes in gravity attraction are computed separately for the parts above and below point elevation. Further, the mass removed from the vent is distributed in a layer below the volcano (radius: 2.5 km). Some details are given by Jahr et al. (1998). Besides the density change the radius of the vent is the controlling variable, because it enters the formula squared (comp. Telford et al., 1976). In accordance with Brown & Rymer (1989), who assumed at Po'as volcano a diameter of several hundred metres for the vent area we tested radii between 200 and 400 m. For the density changes we took 400 kg/m 3 and 600 kg/m 3, respectively. Table 3 gives the observations and the results for the maximum gravity changes for different combinations of the para-
meters for the Tumpa Lahar Channel pro®le and the Mayon Resthouse pro®le, in both cases for the last point closest to the summit. Figs. 8 and 9 show the computed curves for this pro®le and for these parameter combinations. With this basic model we can explain both the observed increase in gravity (included in Figs. 8 and 9 for comparison) as well as the restriction of the changes to the area of the volcano. The Mayon Resthouse pro®le ends closer to the crater. Whereas at Tumpa the model only explains about two thirds of the observed effect, at Mayon Resthouse the same model gives quite a good ®t (Table 3, Fig. 9). The differences are due to the geometry: Mayon is not a `perfect' coneÐat the same distances the elevations are slightly different at both pro®les. Regarding internal unsymmetries we do not have any information. Approaching the summit the model curves also show
232
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
a turning-point to smaller changes at a distance of about 2000 m. Closer to the summit the values become negative (not shown in the ®gures). For the sake of clarity we did not merge the model curves with all the observations: Only the maximum gravity changes are given. On the other hand, because of lacking information we did not aim at a perfect ®t: The model curves should only prove that the gravity changes are con®ned to the same radius as the observations as well as the relative changes towards the crater. 8. Conclusions The observed gravity changes at Mayon volcano are not accompanied by signi®cant elevation changes (derived from GPS measurements). Applying the classical Mogi-model of a magma source below Mayon to explain the increase in gravity we would expect subsidence in the order of 0.5 to 1 m. Since there are no signi®cant elevation changes, we cannot prove an extended shallow magma source according to this model. This is supported by our geochemical analyses of latest rock samples indicating a rather undifferentiated magma. C. Newhall (pers. comm.) expects at least a small magma reservoir. He also mentions changes in the composition of the lava with time (Newhall, 1979). Generally, there are several models possible explaining the observed gravity increase. Ground water in¯ux is favoured by C. Newhall (pers. comm.), but this cannot be proved by our data in that extend. On the other hand, the respective reports are not at all signi®cant. But, since we never have enough information about the processes taking place a scienti®c program regarding the monitoring of several wells could provide additional constraints. At present, we prefer a process of mass redistribution in the volcanic vent: The relatively small gravity changes, especially the increase of gravity accompanying the decrease of volcanic activity, can be explained by density changes within a vent system. Here, we follow Eggers (1987), who explained similar observations at andesitic volcanoes by mass redistribution due to visiculation and degassing. Due to the complexity of the system and lacking information the model is quite straight forward. Therefore, the ®t of our model to the data is naturally not perfect. But a
better ®t would not improve the result at all: It is obvious that this class of models explains the magnitude and the shape of the gravity changes, and, thus, points to the physical process controlling the system. Thus, although the gravity changes are close to the threshold of being traceable, we can state that we got one step more to a better understanding of the physics of Mayon volcano. This results in a further development of the monitoring strategy: At Mayon wide-area long-term height and deformation measurements turned out to be not suitableÐat least in the time span 1992±1996. Possibly, short-term deformation measurements near the vent could be informative. The seismic monitoring as the main tool for shortterm prediction must be improved considerably. Since the gravity variations are restricted to a limited area of the volcano it seems to be suf®cient to measure only differences between some points of the regional network and the reference points to detect any future changes, preferably along both pro®les. A similar proposal was also made by Dzurisin et al. (1980) for Kilauea volcano, Hawaii. Provided that some leveling or GPS work is frequently done a special height control is not necessary in this particular case during a quiet period (of course, during a period of beginning activities the measurements should be done at the same time and at the same places like in the present study). And deformation measurements alone are not suitable, at least on the accuracy level achieved here. This results in a change of the monitoring strategy of Mayon: ² For long-term prediction microgravity can be of use; in this special case the monitoring of the gravity differences along the pro®les and connected to the reference points, maybe twice a year, is suf®cient and neither time consuming nor costly. Thus, microgravity measurements are a tool to improve hazard assessment at Mayon. ² For short-term prediction seismic monitoring is the only tool and must be improved at Mayon volcano. If there are evidences for new activities complete microgravity campaign might give insights into the processes going on. In addition, it might be useful to drill some just for the purpose to monitor groundwater
some more wells level
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
changes with data logging systems. Only in this way reliable data can be made available and, thus, the hypothesis of Newhall regarding the relation of magma intrusion and dilatation be proved.
Acknowledgements Our measurements at Mayon were carried out with the help of E. Diao III., G. Gabriel, O. Haase, T. Jahr, J. Pantig, I. Proeck (microgravity), and E. Banganan, P. Corstens, R.B. Quiambo, R. Hejen, F. Menge, M. Rennen, K. Schmidt, A. Teuber (GPS). The adjustments of the observed gravity differences were carried out with the program system GRAV by H.G. Wenzel. C. Newhall made some of his ideas about the relation to water table changes available. This research project was started with the support of the `Deutsche Forschungsgemeinschaft' (DFG, JE 107-17) and the `Gesellschaft fuÈr Technische Zusammenarbeit' (GTZ). Under the research contract of the European Commission (ENV4-CT96-0259) additional detailed studies were possible. G. Berrino and F. Beauducel reviewed the manuscript and provided many comments and suggestions to improve the text. All this is gratefully acknowledged.
References Aurelio, M.A., 1992. Tectonique du segment central de la faille Philippine. Academie de Paris, Ph.D. thesis, Universite Pierre et Marie Curie. Berrino, G., Corrado, G., 1991. Gravity changes and volcanic dynamics. Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, 4, Proceedings of the Workshop: Geodynamical Instrumentation applied to Volcanic Areas. Oct. 1±3 1990, Walferdange, Luxembourg, 305±323. Bischke, R.E., Suppe, J., Pilar, R.D., 1990. A new branch of the Philippine fault system as observed from aeromagnetic and seismic data. Tectonophysics 183, 243±264. Brown, G.C., Rymer, H., Dowden, J., Kapadia, P., Stevenson, D., Barquero, J., Morales, L.D., 1989. Energy budget analysis for Poas crater lakeÐimplications for predicting volcanic activity. Nature 339, 370±373. Brown, G.C., Rymer, H., 1991. Microgravity monitoring at active volcanoes: A review of theory and practice. Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, 4, Proc. of the Workshop: Geodynamical Instrumentation applied to Volcanic Areas. Oct. 1±3 1990, Walferdange, Luxembourg, 279±304. Brown, G.C., Rymer, H., Stevenson, D., 1991. Volcano monitoring
233
by microgravity and energy budget analysis. J. Geol. Soc. 148, 585±593. Brunner, F.K., Welsch, H.-G., 1993. Effect of the Troposphere on GPS Measurements. GPS-World, January, 1993. Dzurisin, D., Anderson, L.A., Eaton, G.P., Koyanagi, R.Y., Lipman, P.W., Lockwood, J.P., Okamura, R.T., Puniwai, G.S., Sako, M.K., Yamashita, K.M., 1980. Geophysical observations of Kilauea volcano, Hawaii, 2. Constraints on the magma supply during November 1975±September 1977. J. Volcanol. Geotherm. Res. 7, 241±269. Eggers, A.A., 1987. Residual gravity changes and eruption magnitudes. J. Volcanol. Geotherm. Res. 33, 201±216. Hop®eld, H.S., 1979. Improvements in the tropospheric refraction correction for range measurements. Philosophical Transactions of the Royal Society of London, 294. Jahr, T., Jentzsch, G., Diao, E., 1995. Microgravity measurements at Mayon Volcano, Luzon, Philippines. Proceedings of the Workshop: New Challenges for Geodesy in Volcanoes Monitoring. June 14-16,, Walferdange, Luxembourg, Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, Vol. 8, 307±317. Jahr, T., Jentzsch, G., Punongbayan, R.S., Schreiber, U., Seeber, G., VoÈlksen, C., Weise, A., 1998. Mayon volcano, Philippines: improvement of hazard assessment by microgravity and GPSÐmeasurements? Proceedings Int. Symp. on Current Crystal Movement and Hazard Assessment (IUGG, IAG), Wuhan, Nov. 1997. Seismological Press, Beijing, pp. 599±608. Jaupart, C., Tait, S., 1990. Dynamics of eruptive phenomena. Rev. in Mineralogy 24, 213±238. Jentzsch, G., 1995, Mayon Volcano: Ocean tidal loading triggering volcanic activities?. Proceedings 12th Int. Symp. on Earth Tides, Beijing, August. 4±7 1993, 487±499. Jentzsch, G., Haase, O., Kroner, C., Winter, U., Punongbayan, R.S., 1997. Tidal Triggering at Mayon Volcano? Proceedings of the Workshop: Short Term Thermal and Hydrological Signatures related to Tectonic Activities. Nov. 15±17, 1995, Walferdange, Luxembourg, Cahier du Centre EuropeÂeen de Geodynamique et de Seismologie, Vol. 14, 95±104. Jentzsch, G., Haase, O., Kroner, C., Winter, U., 2001. Mayon volcano, Philippines: Some insights into stress balance. J. Volcanol. Geotherm. Res.109, 205±217. Jousset, P., Okada, H., 1999. Post-eruptive volcanic dome evolution as revealed by deformation and microgravity observations at Mount Usu (Hokkaido, Japan). J. Volcanol. Geotherm. Res. 89, 255±273. Lange, R.A., 1994. The effect of H2O CO2 and F on the density and viscosity of silicate melts. Rev. in Mineralogy 30, 331±369. Le Maitre, R.W., 1989. A Classi®cation of Igneus Rocks and Glossary of Terms. Blackwell, Oxford, 191pp. Mommsen, H., Kreuser, A., Weber, J., BuÈsch, H., 1987. Neutron activation analysis of ceramics in the X-ray energy region. Nuclear Instruments and methods in physics research, A257. Newhall, C.G., 1977. Geology and Petrology of Mayon Volcano, southeastern Luzon, Philippines. Master Thesis in Geology, University of California (Davis), 292 pp. Newhall, C.G., 1979. Temporal variation in the lavas of Mayon volcano, Philippines. J. Volcanol. Geotherm. Res. 6, 61±83. Newhall, C.G., 1993. Mayon volcano, February 2, 1993 and
234
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
continuing-preliminary report of collaboration with the Philippine Institute of Volcanology and Seismology, 21 pp. Punongbayan, R.S., Pena, O.D., Ramos, E.G., Umbal, J.V., Ruelo, H.B., Bautista, L.P. and Tayag, J.C., 1990. Operation Mayon. Philippine Institute of Volcanology and Seismology (unpublished). Ramos-Villarta, S., Corpuz, E.G., Newhall, C.G., 1985. Eruptive history of Mayon volcano, Philippines. Philippine J. Volcanol. 2 (1&2), 1±35. Ramoz, E.G., de Torres, C.B., Calderon, A.C., 1985. Earthtide in¯uence on the recent activities of Mayon volcano. Philippine J. Volcanol. 2 (1&2), 172±190. Richet, P., Bottinga, Y., 1995. Rheology and con®gurational entropy of silicate melts. Mineral. Soc. Am., Rev. Mineral. 32, 76±93. Rymer, H., 1996. Microgravity monitoring. In: Scarpa, R., Tilling, R.I. (Eds.), Monitoring and Mitigation of Volcano Hazards. Springer, Berlin, pp. 169±197. Scarfe, C.M., Mysen, B.O., Virgo, D., 1987. Pressure dependence of the viscosity of silicate melts. In: Mysen, B.O. (Ed.), Magmatic Processes: Physicochemical Principles, 1. The Geochemical Society, pp. 59±67 Special Publication.
Sibree, J.O., 1934. The viscosity of froth. Trans. Faraday Soc. 30, 325±331. Stolper, E.M., 1982. The speciation of water in silicate melts. Geochim. Cosmochim. Acta 46, 2609±2620. Sun, S.-S., 1980. Lead isotopic study of young volcanic ricks from mid-ocean ridges, ocean islands and island arcs. Phil. Trans. R. Soc. Lond. A297, 409±445. Telford, W.M., Geldart, L.P., Sheriff, R.E., Keys, D.A., 1976. Applied Geophysics. Cambridge University Press, Cambridge, p. 860. Thomson, R.N., Morrison, M.A., Hendry, G.L., Parry, S.J., 1984. An assessment of the relative roles of a crust and mantle in magma genesis: An elemental approach. Phil. Trans. R. Soc. Lond. A310, 549±590. VoÈlksen, C., Seeber, G., 1995. Establishment of a GPS based control network at Mayon volcano. Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, 8, Proceedings of the Workshop: New Challenges for Geodesy in Volcanoes Monitoring. June 14±16, 1993, Walferdange, Luxembourg, 99±113. Wanninger, L., 1993. Effects of the Equatorial Ionosphere on GPS. GPS-World, July, 1993.
Mayon volcano, Philippines: change of monitoring strategy after microgravity and GPS measurements from 1992 to 1996 Gerhard Jentzsch a,*, Raymondo S. Punongbayan d, Ulrich Schreiber c, GuÈnter Seeber b, Christoph VoÈlksen b, Adelheid Weise a a Institute for Geosciences, FSU Jena, Burgweg 11, D-07749 Jena, Germany Geodetic Institute, UniversitaÈt Hannover, Schneiderberg 50, D-30167 Hannover, Germany c Department of Geology, UniversitaÈt Essen, UniversitaÈtsstr. 5, D-48117 Essen, Germany d Philippine Institute of Volcanology and Seismology, C.P. Garcia Avenue, University of the Philippines Campus, Diliman, Quezon City, Philippines b
Received 21 January 1999; revised 5 December 1999; accepted 3 January 2000
Abstract Mayon volcano is part of the Bicol volcanic chain on the island Luzon, Philippines. During this century there were ten activity periods distributed almost regularly. Because of the density of population (about one million people living in the vicinity of the volcano) three seismological observatories are in operation. Measurements of gravity changes started in 1992, just before the eruption of February/March 1993. Two pro®les at the slope were established, connected to a regional network around the volcano. In order to enable the determination of mass changes between the campaigns the height control was provided by parallel GPS measurements. In all, the network consists of 26 points which were remeasured with three gravimetres at least three times within one campaign. During ®ve campaigns within 4 years the differential GPS gives no signi®cant changes of the elevation (within ^4 cm). Nevertheless, the gravity increased signi®cantly by up to 1500 nm/s 2 (equivalent to 150 mGal). As no signi®cant change of elevation is observed (GPS), no extended shallow magma chamber system below the volcano can be proved. This is in accordance with geochemical results indicating a rather undifferentiated magma. The youngest lava which is of interest for the eruption dynamics belongs to the medium-K basaltic andesite ®eld of the K2O vs SiO2 diagram. A rather qualitative check of groundwater level changes reveals that these cannot be the sources for the observed gravity changes. Thus, the increase of gravity after the eruption of February 1993 can be explained by a mass redistribution in the volcanic vent from above the level of the gravity points to below. Practical conclusions of these results lead to changes in the monitoring strategy: Deformation measurements did not reveal any volcanic activities; at least for the eruption of 1993 no signi®cant deformation was observed. Gravity could be an indicator for long-term changes. Thus, repeated gravity measurements/GPS at selected points could be used in parallel to seismic monitoring to detect slow mass movements prior to changes in seismicity. q 2001 Elsevier Science B.V. All rights reserved. Keywords: physical volcanology; microgravity; GPS; petrology; volcanic modeling
1. Introduction * Corresponding author. Tel.: 149-3641-948660; fax: 149-3641948662. E-mail address: [email protected] (G. Jentzsch).
The Philippines, situated in the western Paci®c Ocean, are part of an extensive island arc system.
0377-0273/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S 0377-027 3(00)00313-9
220
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 1. Location of Mayon volcano, Luzon island, Philippines, after Ramos-Villarta et al. (1985).
The geological and geophysical situation is mainly characterized by two subduction zones. These collisions yield a complex tectonic situation with main faults and lineaments extending through the whole crust down to the upper mantle (Bischke et al., 1990; Aurelio, 1992). The active tectonics cause earthquakes and volcanic activities. Mayon is the most famous volcano of the Bicol volcanic chain in the south-east of Luzon (138 15.4 0
N, 1238 41.1 0 E, Fig. 1). Its shape is a nearly perfect cone with an altitude of 2462 m. south-west of Mayon, rather close, the Legaspi Lineament runs NW±SE across Legaspi City. According to Aurelio (1992) probably it could partly control the activity of Mayon volcano. There were ten activity periods spread almost regularly over the past century. The last important eruption was in 1984. Compared to this, some smaller eruptions took place in February/March 1993,
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
just between the ®rst and second campaigns of our gravity measurements. They resulted in ash fall and a lava ¯ow at the south-east ¯ank with a volume of about 10 Mio m 3. The decrease of activity after the eruptions of Mayon usually is rather slow; after 1 year crater glow still may be observable. More than one million people live in the vicinity of Mayon volcano. Legaspi City is situated only 12 km away from the crater. Thus, the improvement of hazard assessment requires a better understanding of the physical properties of the volcano. For the monitoring of Mayon three seismological observatories were installed (Punongbayan et al., 1990), and in previous scienti®c investigations the volumes of ejecta were evaluated and the emission cloud analysed (Ramoz-Villarta et al., 1985). Correlations with earth tides were examined (Ramoz et al., 1985). As the ocean loading tide around the Philippines varies strongly, we proposed that the suspected increase of seismicity is connected with the strain induced (Jentzsch, 1995; Jentzsch et al., 1997, 2001). Until the eruption of 1993 the seismic monitoring at the three observatories lacked optimum conditions: The equipment was not homogeneous and the noise levels were very high (except for Mayon Resthouse observatory). Therefore, it was dif®cult to detect precursors. In the present paper we discuss our results from microgravity and GPS measurements in relation to geochemical analyses of samples of the volcanics carried out by co-author Schreiber. An additional contribution came from C. Newhall (1998, personal communication), who proposes groundwater changes as the reason for the observed gravity variations. The aim of our work is to interpret all the results regarding the development of a model of the volcano to improve the understanding of the physics of further eruptions. First results were already presented by Jahr et al. (1998). 2. Microgravimetry in volcanic areas Some examples of sub-surface magma movements are documented using high precise gravimetry and concurrent elevation control, for instance at Kilauea/ Hawaii (Dzurisin et al., 1980), Pacaya/Guatemala (Brown & Rymer, 1991), and at Usu/Japan (Jousset
221
& Okada, 1999). Rymer (1996) reported gravity changes at Po'as volcano/Costa Rica, that uniquely preceded an eruption. The connection between the variation of gravity and elevation due to mass and/or density changes is given by the two well known gradients: If gravity is following the free air gradient (FAG), no sub-surface mass changed, while data following the gradient related to the standard Bouguer correction (BCFAG) imply mass changes. Departures from these gradients due to mass and/or density variations enable modelling of the volcanic process (Brown & Rymer, 1991; Brown at al., 1989; Brown et al., 1991; Jahr et al., 1995). The desired accuracy of the gravity measurements is in the order of ^100 nm/s 2 (equal to ^ 10 mGal).
3. Microgravity at Mayon volcano The measurements of repeated gravity and GPS started in November 1992. Two pro®les at the slope, towards the summit (up to 850 m) were established. They are connected to a local and a regional network (Fig. 2) around the volcano with the extension of 40 km £ 50 km. The maximum distance from the crater is 47 km. Two points at the opposite side of the Legaspi Lineament (in the south-west) are connected to the reference network. The gravity points are installed in areas of sand (lahars), pasture ground as well as basaltic rock. The total gravity range is 1800 mm/s 2. 1 They consist of iron marks cemented into drillholes in rocks or blocks of concrete. During 4 years, until December 1996, ®ve campaigns of repeated gravity measurements were carried out with three LaCoste & Romberg gravimetres of the types G and D. The instruments are equipped with electrostatic feed back systems and electronic levels. They were calibrated several times at the gravimetre calibration lines at Hannover Harz/ Germany. Even if the observed calibration factors were not valid in the used gravity range of Mayon, at least the stability of the scale factors could be checked up. Additionally, the ratio between the 1 The usual units in gravimetry are milligal and microgal; the relations to SI units are: 10 mm/s 2 1 mGal, 10 nm/s 2 1 mGal.
222
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 2. Gravity and GPS stations around Mayon volcano, the Legaspi Lineament according to Aurelio (1992): Note the circle descibing the base of the volcano and the two pro®les (Mayon Resthouse: MRH1±MRH0; Tumpa Lahar Channel: QSD1/2±TLC2/1).
scale factors of the instruments was checked to be similar at Mayon and in Germany within 1 £ 10 24. During each campaign the gravity differences were generally observed at least three times with each of the three gravimetres (minimum nine differences). The program system GRAV from G. Wenzel is used for the adjustment where a standard earth tide correction is applied with a precision better than it is required. A linear drift is adjusted (up to ^40 nm/s 2/ h) which turned out to be rather similar for each instrument in all the campaigns. The standard deviation of one observed gravity difference m0 is in the order of ^140¼150 nm/s 2. The results of each campaign are the gravity differences between the stations with an achieved accuracy
of about ^120 nm/s 2, as it was expected (Jahr et al., 1995). For the differences between the campaigns the assumption is made that gravity is constant at the most distant stations from the volcano. Between December 1992 and May 1993 no signi®cant gravity changes could be observed although Mayon was active in February/March 1993. Between May 1993 and December 1996 the increase of gravity is signi®cant, rising up to 1500 nm/s 2 (^150 nm/s 2) along the pro®les at the slope, decreasing with the distance from the crater (Figs. 3 and 4). Taking into account the dates of the individual campaigns (a) Dec. 1992, (b) May 1993, (c) May 1994, (d) Dec. 1994, (e) Dec. 1996) the observations along Tumpa Lahar Channel show a more or less steady increase in
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
223
Fig. 3. Gravity changes along the Tumpa Lahar Channel pro®le (south ¯ank) over the distance from the crater, relative to the ®rst measuring campaign, with error bars. The points from left to right are TLC1, TLC2, QSD1/2, TES, CAG, LHO (comp. Fig. 2). The campaigns are: (a) Dec. 1992, (b) May 1993, (c) May 1994, (d) Dec. 1994 and (e) Dec. 1996.
gravity of about 300 nm/s 2 per year in a rather continuous process (involving an overall error of about ^150 nm/s 2). In the network surrounding Mayon volcano (.10 km from the summit; Fig. 2) there are no signi®cant gravity variations over the whole period of the surveys; only at a few individual points at the foot of Mayon we might see gravity changes of up to 300 nm/s 2 (increase or decrease), but with regard to the errors involved these changes are not considered as signi®cant. Compared to Tumpa the picture is not quite that clear for the Mayon Resthouse pro®le: At ®rst we see that the last curve (5-1) consists of less points than the others; due to road construction works the missing points were lost. All curves show changes up to a distance from the summit of 5±6 kmÐnot signi®cant for (2-1); closer to the summit the changes are not signi®cant any more. This ®ts to the Tumpa pro®le, which ends at about 4.5 km from the crater.
The results of the third campaign are already in the order of magnitude of the ®fth, whereas the forth gives smaller changes compared to Tumpa. From the data of the third campaign of the Mayon Resthouse pro®le there is the suspicion of an additional effect which shifts the rest of the curve upwards. Due to logistics it was not possible to connect the uppermost points of both pro®les to different points of the local network: The driving or walking times would have been so long or even not manageable during daylight so that the bene®t of such a connection could not been used. Therefore, these points are only connected to the base of the pro®les. Thus, if there are any changes at the ®rst points the whole pro®le is affected. Since all the data is consistent we simply must accept this result; this points to an additional signal maybe due to a recharge of the local water table which is not proved by observational data from wells.
224
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 4. Gravity changes along the Mayon Rest House pro®le (north ¯ank) over the distance from the crater, relative to the ®rst measuring campaign, with error bars. The points from left to right are MRH0, MRH9,¼,MRH1 (comp. Fig. 2). The campaigns are indicated in Fig. 3. Note that for the last campaign only points MRH1, MRH5, MRH9, MRH0 were available.
Inspite of the problems discussed, from these results we ®nd that decreasing volcanic activities are accompanied by increasing gravity. 4. The regional GPS network at Mayon In order to interpret gravity variations in terms of mass changes between the campaigns we need the height control provided by parallel GPS measurements. In each campaign three receivers of the type Ashtech (1/2. camp.) or Trimble 4000 SSE were used. The reference network was observed in triangles, resulting in 15 baselines between the stations SC1, RNSD, TAP2, TIWI, MRH1, SFES, LHO, PES2 (Fig. 2). For the data processing the program system GEONAP (University, Hannover) is used, which is based on undifferenced observations. The lay-out of the network and ®rst results were already discussed in detail by VoÈlksen & Seeber (1995).
The disturbances due to the in¯uence of the Ionosphere are major in the equatorial regions. The ®rst two or three campaigns took place during maximum solar activity. Even with dual-frequency data serious problems can occur in the data processing (Wanninger, 1993), which is re¯ected in the higher standard deviations for the ®rst two campaigns. Concerning the meteorological correction a model with a standard atmosphere is used (Hop®eld, 1979). The residual tropospheric refraction error (Residual Zenit Delay) is compensated by a scale factor, which is estimated for each station. A test was made during the third campaign introducing observed meteorological surface data for tropospheric modeling. But this did not lead to lower standard deviations, even under the tropic conditions around Mayon. This is in agreement with Brunner & Welsch (1993). We can state after comparing these results (applying a scale factor for compensating the residual tropospheric refraction errors) with those applying a standard atmosphere
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
225
Fig. 5. Gravity changes versus height differences at Mayon volcano from one campaign to the next. The crosses are indicating the mean error bars.
only, that systematic deviations in the vertical component of up to 8 cm appear. Therefore, the same meteorological model was used in all campaigns for the data processing. The reference point for the datum is the closest IGS 2 site Taipei in Taiwan with 1300 km baseline length. The result of the GPS-measurements is: No signi®cant deformations are stated, neither in the vertical nor in the horizontal components. The standard deviations (posteriory) for one campaign vary in the range of ^ 0.5¼1.5 cm in the horizontal and ^1.8¼4 cm in the vertical component. In the horizontal component isolated changes of up to 3 cm are not signi®cant. No signi®cant changes of the elevation (Fig. 5) are stated within the error bar of around ^4 cm between two campaigns. Therefore, the gravity changes drawn versus the height differences from one campaign to the next do not show any gradients (Fig. 5, crosses denote error bars for gravity and elevation). 2
International GPS and Geodynamics Service.
5. Geochemistry and petrology In 1992 some rock samples were taken of different lava ¯ows in the area of Mayon Resthouse. After the 1993 eruption samples of different layers of the new lava ¯ow were analysed (see Table 1). The rock classi®cation used in this paper is based on Le Maitre's (1989) K2O vs SiO2 diagram for orogenic igneous rocks (Fig. 6). The volcanic products of Mayon volcano are basalts, basaltic andesites and andesites of the medium-K calc-alkaline series. The lavas show cyclical variation with up to three successive basaltic ¯ows followed by six to ten andesitic ¯ows (Newhall, 1977, 1979). The youngest lava which is of interest for the eruption dynamics plots into the medium-K basaltic and andesite ®eld. MgO does not exceed 4.7% (4.12± 4.66% MgO from base to top of the ¯ow) and Fe2O3 decreases concomitantly with MgO. The samples exhibit typical arc-related trace element characters, i.e. enrichment in large ion lithophile elements
226
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Fig. 6. K2O vs SiO2 diagram of Mayon samples. Filled symbols are samples from the 1993 lava, others from Newhall (1977).
(LILEÐRb, Ba, K, Sr) and light rare earth elements (LREE) relative to heavy REE and high ®eld strength elements (HFSEÐNi, Zr, Ti). The homogeneous trace element patterns (Fig. 7), e.g. (Ba/La) n 1.9±2.3, (La/Yb)n 4.8±5.3 (normalisation constants from Sun, 1980) have typical subductionrelated negative anomalies in Nb and Ta. Sr and Y values as well as Ti and Zr values are in the range of common arc lavas. The missing Eu-anomaly indicates that plagioclase fractionation has not taken place. The youngest product of Mayon is a strongly porphyritic lava with 30±50% of phenocrysts of mainly plagioclase (An 76±An 42), augite, hypersthene, olivine and titaniferous magnetite. The same minerals and rare sanidine microliths are part of the groundmass together with glass and accessory minerals. The low degree of fractional crystallization argues against a magma source in the upper crust. The determination of the viscosity is dif®cult, depending on the crystallization and the gas content. The youngest lava contains 30 to 50% crystals of 0.3 to 2 mm in size. Here, the phenocrysts provide a
higher percentage in the lave erupted ®rst. Applying the relation for viscosity and temperature given by Richet & Bottinga (1995) for a temperature range from 1300 to 10008C relevant for the intrusion and rise of the magma we ®nd a viscosity intervall of about 2.65±5.5 Log Pas at the pressure of 1 bar. According to Scarfe et al. (1987) and Lange (1994) the viscosity decrease with pressure is less than one order of magnitude for about 20 kbar. Another factor for controlling the viscosity is the contents of water in the melt (Stolper, 1982). According to Jaupart & Tait (1990) 2% of water are enough to reduce viscosity by one order of magnitude. This effect is greater at lower temperatures (Lange, 1994). Since vesiculation depends on the pressure, the ascent of the magma is accompanied by an increase of viscosity. Already Sibree (1934) found that more than 50% gas bubbles in the melt increase the viscosity by about one order of magnitude. To estimate the viscosity of the magma of the last eruption we assume the above mentioned temperature intervall. The content of water is unknown. From the
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
227
Fig. 7. Trace element plots of the 1993 Mayon lava normalized to primitive mantle (Thompson et al., 1984). To identify the sample numbers please compare the values given in Table 1.
character and relation of the phenocrysts we assume a pressure of less than 10 kbar and a water content of ,2 wt-%. On the basis of these conditions the viscosity should be ®rst below 500 Pas; later, during the ascent in the vent this value increases due to cooling and vesiculation to 50,000±100,000 Pas. 6. Groundwater level changes Considerable drops of the water table (in the order of 2 m) occurred around Mayon before the 1993 eruption (Newhall, 1998, pers. comm.). The supposed subsequent slow rise (recharge) should have occurred over the time span of the gravity measurements and could contribute to the gravity increase. In this case we should see variations in gravity at points of the network around Mayon, especially at the foot of the volcano, with respect to the distant reference points. C. Newhall collected the observations of local residents around Mayon, which means that no really measured data exist yet. Nevertheless, since drops of
the water table can be linked to stress and, thus, strain changes he tried to use this information. Providing an internal report (Newhall, 1993) and several personal communications C. Newhall informed us about the water level changes in 30 observation wells around Mayon (even at greater distances of 10±15 km, mostly situated in the settled areas around the volcano), and about a simple experiment to prove the hypothesis that intrusion of magma is causing a drop in the water table relative to land surface. He can show by using clay as intrusion material, that in this experiment the water table drops. The model volcano is built up from unconsolidated ®ne sand; due to the intrusion dilatation occurs and water seeps deeper into the ground. It would be interesting to see this effect for a model comparable to the real volcanic edi®ce with different more or less consolidated layers and varying properties of the uppermost 10±20 m. With this model the comparison to the real situation at Mayon is still very tentative. On the other hand, the reports provided by local residents about their observations of the water
228
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Table 1 Geochemical analyses of the 1993 lava from Mayon (basaltic andesite). Main elements (XRF-analyses) in weight%, others in ppm. XRFanalyses of Mo, Ni, Pb, V, Y. Analyses of Ba, Br, Co, Cr, Cs, Cu Ga, Hf, Rb, Sc, Ta, Th, U, Zn, Zr, and REE by NAA (Inst. f. Strahlen- und Kernphysik, Bonn; Mommsen et al., 1987)
SiO2 TiO2 Al2O3 Mn0 Mg0 Ca0 Na2O K2O P2O5 L0I Cr Ni Sc V Cu Pb Zn Mo Rb Cs Ba Sr Ga Ta Nb Hf Zr Y Th U La Ce Nd Sm Eu Tb Yb Lu Br
Mayo 95-1
Mayo 95-3
Mayo 95-5
Mayo 95-7
Mayo 95-11
54.48 0.71 18.81 0.16 4.12 8.54 3.33 1.15 0.31 0.07 5 8 22 188 813 7 87 2.02 24 0.50 457 714 19 0.74 3.9 2.97 131 22 2.46 0.80 16.58 36.37 18.98 4.32 1.33 0.63 2.15 0.44 10.33
54.42 0.72 18.48 0.17 4.30 8.47 3.35 1.16 0.30 0.00 4 7 22 189 660 5 84 2.33 25 0.36 374 702 19 0.48 4.3 2.84 136 23 2.44 0.80 15.84 34.63 22.11 4.18 1.24 0.55 2.15 0.43 4.08
54.46 0.72 18.67 0.17 4.20 8.45 3.32 1.15 0.30 0.00 10 8 22 175 671 7 88 2.43 21 0.38 431 689 18 0.52 3.8 2.71 151 24 2.43 0.70 16.13 35.33 18.05 4.20 1.27 0.62 2.15 0.56 3.76
54.48 0.71 18.66 0.17 4.21 8.47 3.31 1.17 0.30 0.07 8 8 22 190 760 8 90 2.13 24 0.48 413 710 19 0.57 3.9 2.81 127 24 2.46 0.72 16.39 35.60 20.46 4.21 1.32 0.59 2.18 0.51 3.35
53.32 0.76 18.75 0.17 4.66 8.98 3.16 1.07 0.26 0.26 9 8 27 198 624 8 99 2.43 21 0.58 362 660 17 0.33 4.2 2.75 254 22 2.34 0.62 15.55 34.96 18.02 4.26 1.30 0.59 2.26 0.47 2.39
supply certainly contain interesting information. Naturally, these reports are very inhomogeneous and many of them even vague; nevertheless, the classi®cation between `strong evidences' and `no correlation' 3 is given in Table 2 (only 28 are used 3 `correlation' denotes the coincidence of volcanic activities and drop of water level.
because two wells are too far away and do not show any changes). But, even if we do not reject those reports which are obviously exaggerated the result is not at all signi®cant, neither for `pro' nor `con'. If we analyse the reports in detail we see that some `pros' are given for wells farer away from the summit and more downslope, or people report that this drop of ground water level was never observed
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
229
Table 2 Classi®cation of 28 reports about observations regarding water level changes in wells during the 1993 eruption of Mayon (Newhall, 1998, pers. comm.) Strong evidences as seen also before previous eruptions
Correllation of drop of water level and volcanic activity
No correlation; some wells show changes, some not
No correlation, normal situation
2
12
6
8
beforeÐalthough the 1993 eruption was by far not as strong as previous ones. Inspite of these `soft' facts we followed Newhall's suggestion and we compared the data and the locations of the observation wells with our data. To put it on a more stable ground we ®rst estimate the gravity effect to be expected: If we roughly calculate the possible maximum Bouguer effect of such a water level change we arrive at 130 nm/s 2 for a change of 1 m at a porosity of 30% (for unconsolidated sand in the lahar area). Thus, if we even assume a change of 3 m, the gravity effect is in the order of ,500 nm/s 2. This is the maximum effect we could expect, and it should be observable at some of the points of the network at the foot of Mayon except at the distant reference points. Unfortunately, our gravity points and the wells do only coincide in the village Tumpa (point TES) where the ground consists of sand (lahar). At the ¯anks of the volcano there is few information about ground water variations. Upslope, the points TLC1/TLC2 (Fig. 3) are situated on consolidated rock where the observed strong gravity increase could hardly be related to ground water. Further, the steep slope of the Mayon Resthouse pro®le does not seem to allow the discussion of a prevailing water table. At least in these points we cannot assume an in®nite extension of the water table, which again reduces the estimated maximum effect. Although there are points in the lahars the data do not show the expected groundwater signal. According to our data and the localities we can only concede the small, not really signi®cant gravity variations at the foot of the Mayon Resthouse pro®le of up to 300 mm/s 2 as maybe being caused by changes of the local water level. 7. Modelling of the observed gravity changes and interpretation The observed gravity variations show a very signif-
icant distribution: They are con®ned to a radius less than 8±10 km from the summit, and they increase systematically with decreasing distances. Further, the observed maximum gravity changes are in the order of up to 1500 nm/s 2, much bigger than the estimated maximum effect of ,500 nm/s 2 due to water level changes. These observed gravity changes are not accompanied by signi®cant elevation changes as derived from GPS measurements. Usually, we expect that due to an intrusion of magma gravity is increased, and with decreasing activity gravity decreases as well. In the case of Mayon we have to explain the opposite relation, a negative correlation between gravity and volcanic activity. 4 To explain the gravity increase by applying the classical Mogi-model of a magma source below Mayon we would expect subsidence in the order of 0.5±1 m, but this is not observed. Since there are no signi®cant elevation changes, we cannot apply this model: There is obviously no extended shallow magma source which could be responsible for an uplift. This, and the shape of the gravity changes, especially the concentration of the changes around the summit points to a process taking place in the very center of the volcano. The gravity changes are restricted to the area of the volcano, increasing strongly towards the center. All this puts heavy constrains on the modelling of the physical process of this volcano. The relatively small gravity changes, especially the increase of gravity accompanying the decrease of volcanic activity, must therefore be explained by a process within the volcanic vent (and not by a shallow magma system below). This is in accordance with the geochemical analyses of rock samples indicating a rather undifferenciated magma. 4
Since our ®rst measurements were in December 1992 just before the eruption we do not have any information about development of the gravity before.
230
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
Table 3 Modeled maximum gravity changes at Tumpa Lahar Channel and at Mayon Resthouse in [nm/s 2] for different model parameters of the vent: radius and density difference Model version
1
2
3
4
5
6
Max. observed
Radius [m] density difference [kg/m 3] Tumpa Lahar Channel (TLC1) Mayon Resthouse (MRH9)
200 400 155 297
200 600 233 445
300 400 349 668
300 600 524 1002
400 400 621 1188
400 600 931 1782
1579 1339
Similar results are reported by Berrino & Corrado (1991) for Vesuvius and Aeolian Islands/Italy, where the gravity changes, not accompanied by the vertical movements, are explained by mass redistribution, as well as by Eggers (1987) for andesitic volcanoes. Our observations can be explained by a process of density changes or magma movement within the
volcanic vent: Taking into account a mass redistribution in the magma column of the vent there is obviously a mass change above the gravity points (highest elevations of the gravity points are about 900 and 560 m at Mayon Resthouse and Tumpa Lahar Channel, respectively): Thus, expecting a magma drainage or density decrease with decreasing activities our concept is based on the redistribution of
Fig. 8. Modelled gravity changes along Tumpa Lahar Channel pro®le vs observation (maximum gravity changes; models 1±6 comp. Table 3).
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
231
Fig. 9. Modelled gravity changes at Mayon Rest House pro®le vs observation (maximum gravity changes; models 1±6 comp. Table 3).
mass from above the points to below causing an increase in gravity. The model is quite basic: It consists of a vertical cylinder in the place of the volcanic vent. The changes in gravity attraction are computed separately for the parts above and below point elevation. Further, the mass removed from the vent is distributed in a layer below the volcano (radius: 2.5 km). Some details are given by Jahr et al. (1998). Besides the density change the radius of the vent is the controlling variable, because it enters the formula squared (comp. Telford et al., 1976). In accordance with Brown & Rymer (1989), who assumed at Po'as volcano a diameter of several hundred metres for the vent area we tested radii between 200 and 400 m. For the density changes we took 400 kg/m 3 and 600 kg/m 3, respectively. Table 3 gives the observations and the results for the maximum gravity changes for different combinations of the para-
meters for the Tumpa Lahar Channel pro®le and the Mayon Resthouse pro®le, in both cases for the last point closest to the summit. Figs. 8 and 9 show the computed curves for this pro®le and for these parameter combinations. With this basic model we can explain both the observed increase in gravity (included in Figs. 8 and 9 for comparison) as well as the restriction of the changes to the area of the volcano. The Mayon Resthouse pro®le ends closer to the crater. Whereas at Tumpa the model only explains about two thirds of the observed effect, at Mayon Resthouse the same model gives quite a good ®t (Table 3, Fig. 9). The differences are due to the geometry: Mayon is not a `perfect' coneÐat the same distances the elevations are slightly different at both pro®les. Regarding internal unsymmetries we do not have any information. Approaching the summit the model curves also show
232
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
a turning-point to smaller changes at a distance of about 2000 m. Closer to the summit the values become negative (not shown in the ®gures). For the sake of clarity we did not merge the model curves with all the observations: Only the maximum gravity changes are given. On the other hand, because of lacking information we did not aim at a perfect ®t: The model curves should only prove that the gravity changes are con®ned to the same radius as the observations as well as the relative changes towards the crater. 8. Conclusions The observed gravity changes at Mayon volcano are not accompanied by signi®cant elevation changes (derived from GPS measurements). Applying the classical Mogi-model of a magma source below Mayon to explain the increase in gravity we would expect subsidence in the order of 0.5 to 1 m. Since there are no signi®cant elevation changes, we cannot prove an extended shallow magma source according to this model. This is supported by our geochemical analyses of latest rock samples indicating a rather undifferentiated magma. C. Newhall (pers. comm.) expects at least a small magma reservoir. He also mentions changes in the composition of the lava with time (Newhall, 1979). Generally, there are several models possible explaining the observed gravity increase. Ground water in¯ux is favoured by C. Newhall (pers. comm.), but this cannot be proved by our data in that extend. On the other hand, the respective reports are not at all signi®cant. But, since we never have enough information about the processes taking place a scienti®c program regarding the monitoring of several wells could provide additional constraints. At present, we prefer a process of mass redistribution in the volcanic vent: The relatively small gravity changes, especially the increase of gravity accompanying the decrease of volcanic activity, can be explained by density changes within a vent system. Here, we follow Eggers (1987), who explained similar observations at andesitic volcanoes by mass redistribution due to visiculation and degassing. Due to the complexity of the system and lacking information the model is quite straight forward. Therefore, the ®t of our model to the data is naturally not perfect. But a
better ®t would not improve the result at all: It is obvious that this class of models explains the magnitude and the shape of the gravity changes, and, thus, points to the physical process controlling the system. Thus, although the gravity changes are close to the threshold of being traceable, we can state that we got one step more to a better understanding of the physics of Mayon volcano. This results in a further development of the monitoring strategy: At Mayon wide-area long-term height and deformation measurements turned out to be not suitableÐat least in the time span 1992±1996. Possibly, short-term deformation measurements near the vent could be informative. The seismic monitoring as the main tool for shortterm prediction must be improved considerably. Since the gravity variations are restricted to a limited area of the volcano it seems to be suf®cient to measure only differences between some points of the regional network and the reference points to detect any future changes, preferably along both pro®les. A similar proposal was also made by Dzurisin et al. (1980) for Kilauea volcano, Hawaii. Provided that some leveling or GPS work is frequently done a special height control is not necessary in this particular case during a quiet period (of course, during a period of beginning activities the measurements should be done at the same time and at the same places like in the present study). And deformation measurements alone are not suitable, at least on the accuracy level achieved here. This results in a change of the monitoring strategy of Mayon: ² For long-term prediction microgravity can be of use; in this special case the monitoring of the gravity differences along the pro®les and connected to the reference points, maybe twice a year, is suf®cient and neither time consuming nor costly. Thus, microgravity measurements are a tool to improve hazard assessment at Mayon. ² For short-term prediction seismic monitoring is the only tool and must be improved at Mayon volcano. If there are evidences for new activities complete microgravity campaign might give insights into the processes going on. In addition, it might be useful to drill some just for the purpose to monitor groundwater
some more wells level
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
changes with data logging systems. Only in this way reliable data can be made available and, thus, the hypothesis of Newhall regarding the relation of magma intrusion and dilatation be proved.
Acknowledgements Our measurements at Mayon were carried out with the help of E. Diao III., G. Gabriel, O. Haase, T. Jahr, J. Pantig, I. Proeck (microgravity), and E. Banganan, P. Corstens, R.B. Quiambo, R. Hejen, F. Menge, M. Rennen, K. Schmidt, A. Teuber (GPS). The adjustments of the observed gravity differences were carried out with the program system GRAV by H.G. Wenzel. C. Newhall made some of his ideas about the relation to water table changes available. This research project was started with the support of the `Deutsche Forschungsgemeinschaft' (DFG, JE 107-17) and the `Gesellschaft fuÈr Technische Zusammenarbeit' (GTZ). Under the research contract of the European Commission (ENV4-CT96-0259) additional detailed studies were possible. G. Berrino and F. Beauducel reviewed the manuscript and provided many comments and suggestions to improve the text. All this is gratefully acknowledged.
References Aurelio, M.A., 1992. Tectonique du segment central de la faille Philippine. Academie de Paris, Ph.D. thesis, Universite Pierre et Marie Curie. Berrino, G., Corrado, G., 1991. Gravity changes and volcanic dynamics. Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, 4, Proceedings of the Workshop: Geodynamical Instrumentation applied to Volcanic Areas. Oct. 1±3 1990, Walferdange, Luxembourg, 305±323. Bischke, R.E., Suppe, J., Pilar, R.D., 1990. A new branch of the Philippine fault system as observed from aeromagnetic and seismic data. Tectonophysics 183, 243±264. Brown, G.C., Rymer, H., Dowden, J., Kapadia, P., Stevenson, D., Barquero, J., Morales, L.D., 1989. Energy budget analysis for Poas crater lakeÐimplications for predicting volcanic activity. Nature 339, 370±373. Brown, G.C., Rymer, H., 1991. Microgravity monitoring at active volcanoes: A review of theory and practice. Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, 4, Proc. of the Workshop: Geodynamical Instrumentation applied to Volcanic Areas. Oct. 1±3 1990, Walferdange, Luxembourg, 279±304. Brown, G.C., Rymer, H., Stevenson, D., 1991. Volcano monitoring
233
by microgravity and energy budget analysis. J. Geol. Soc. 148, 585±593. Brunner, F.K., Welsch, H.-G., 1993. Effect of the Troposphere on GPS Measurements. GPS-World, January, 1993. Dzurisin, D., Anderson, L.A., Eaton, G.P., Koyanagi, R.Y., Lipman, P.W., Lockwood, J.P., Okamura, R.T., Puniwai, G.S., Sako, M.K., Yamashita, K.M., 1980. Geophysical observations of Kilauea volcano, Hawaii, 2. Constraints on the magma supply during November 1975±September 1977. J. Volcanol. Geotherm. Res. 7, 241±269. Eggers, A.A., 1987. Residual gravity changes and eruption magnitudes. J. Volcanol. Geotherm. Res. 33, 201±216. Hop®eld, H.S., 1979. Improvements in the tropospheric refraction correction for range measurements. Philosophical Transactions of the Royal Society of London, 294. Jahr, T., Jentzsch, G., Diao, E., 1995. Microgravity measurements at Mayon Volcano, Luzon, Philippines. Proceedings of the Workshop: New Challenges for Geodesy in Volcanoes Monitoring. June 14-16,, Walferdange, Luxembourg, Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, Vol. 8, 307±317. Jahr, T., Jentzsch, G., Punongbayan, R.S., Schreiber, U., Seeber, G., VoÈlksen, C., Weise, A., 1998. Mayon volcano, Philippines: improvement of hazard assessment by microgravity and GPSÐmeasurements? Proceedings Int. Symp. on Current Crystal Movement and Hazard Assessment (IUGG, IAG), Wuhan, Nov. 1997. Seismological Press, Beijing, pp. 599±608. Jaupart, C., Tait, S., 1990. Dynamics of eruptive phenomena. Rev. in Mineralogy 24, 213±238. Jentzsch, G., 1995, Mayon Volcano: Ocean tidal loading triggering volcanic activities?. Proceedings 12th Int. Symp. on Earth Tides, Beijing, August. 4±7 1993, 487±499. Jentzsch, G., Haase, O., Kroner, C., Winter, U., Punongbayan, R.S., 1997. Tidal Triggering at Mayon Volcano? Proceedings of the Workshop: Short Term Thermal and Hydrological Signatures related to Tectonic Activities. Nov. 15±17, 1995, Walferdange, Luxembourg, Cahier du Centre EuropeÂeen de Geodynamique et de Seismologie, Vol. 14, 95±104. Jentzsch, G., Haase, O., Kroner, C., Winter, U., 2001. Mayon volcano, Philippines: Some insights into stress balance. J. Volcanol. Geotherm. Res.109, 205±217. Jousset, P., Okada, H., 1999. Post-eruptive volcanic dome evolution as revealed by deformation and microgravity observations at Mount Usu (Hokkaido, Japan). J. Volcanol. Geotherm. Res. 89, 255±273. Lange, R.A., 1994. The effect of H2O CO2 and F on the density and viscosity of silicate melts. Rev. in Mineralogy 30, 331±369. Le Maitre, R.W., 1989. A Classi®cation of Igneus Rocks and Glossary of Terms. Blackwell, Oxford, 191pp. Mommsen, H., Kreuser, A., Weber, J., BuÈsch, H., 1987. Neutron activation analysis of ceramics in the X-ray energy region. Nuclear Instruments and methods in physics research, A257. Newhall, C.G., 1977. Geology and Petrology of Mayon Volcano, southeastern Luzon, Philippines. Master Thesis in Geology, University of California (Davis), 292 pp. Newhall, C.G., 1979. Temporal variation in the lavas of Mayon volcano, Philippines. J. Volcanol. Geotherm. Res. 6, 61±83. Newhall, C.G., 1993. Mayon volcano, February 2, 1993 and
234
G. Jentzsch et al. / Journal of Volcanology and Geothermal Research 109 (2001) 219±234
continuing-preliminary report of collaboration with the Philippine Institute of Volcanology and Seismology, 21 pp. Punongbayan, R.S., Pena, O.D., Ramos, E.G., Umbal, J.V., Ruelo, H.B., Bautista, L.P. and Tayag, J.C., 1990. Operation Mayon. Philippine Institute of Volcanology and Seismology (unpublished). Ramos-Villarta, S., Corpuz, E.G., Newhall, C.G., 1985. Eruptive history of Mayon volcano, Philippines. Philippine J. Volcanol. 2 (1&2), 1±35. Ramoz, E.G., de Torres, C.B., Calderon, A.C., 1985. Earthtide in¯uence on the recent activities of Mayon volcano. Philippine J. Volcanol. 2 (1&2), 172±190. Richet, P., Bottinga, Y., 1995. Rheology and con®gurational entropy of silicate melts. Mineral. Soc. Am., Rev. Mineral. 32, 76±93. Rymer, H., 1996. Microgravity monitoring. In: Scarpa, R., Tilling, R.I. (Eds.), Monitoring and Mitigation of Volcano Hazards. Springer, Berlin, pp. 169±197. Scarfe, C.M., Mysen, B.O., Virgo, D., 1987. Pressure dependence of the viscosity of silicate melts. In: Mysen, B.O. (Ed.), Magmatic Processes: Physicochemical Principles, 1. The Geochemical Society, pp. 59±67 Special Publication.
Sibree, J.O., 1934. The viscosity of froth. Trans. Faraday Soc. 30, 325±331. Stolper, E.M., 1982. The speciation of water in silicate melts. Geochim. Cosmochim. Acta 46, 2609±2620. Sun, S.-S., 1980. Lead isotopic study of young volcanic ricks from mid-ocean ridges, ocean islands and island arcs. Phil. Trans. R. Soc. Lond. A297, 409±445. Telford, W.M., Geldart, L.P., Sheriff, R.E., Keys, D.A., 1976. Applied Geophysics. Cambridge University Press, Cambridge, p. 860. Thomson, R.N., Morrison, M.A., Hendry, G.L., Parry, S.J., 1984. An assessment of the relative roles of a crust and mantle in magma genesis: An elemental approach. Phil. Trans. R. Soc. Lond. A310, 549±590. VoÈlksen, C., Seeber, G., 1995. Establishment of a GPS based control network at Mayon volcano. Cahier du Centre EuropeÂen de Geodynamique et de Seismologie, 8, Proceedings of the Workshop: New Challenges for Geodesy in Volcanoes Monitoring. June 14±16, 1993, Walferdange, Luxembourg, 99±113. Wanninger, L., 1993. Effects of the Equatorial Ionosphere on GPS. GPS-World, July, 1993.