V23D-01 13:40h
The Geochemical and Petrological History of Arenal Volcano, Costa Rica
The eruptive history of Arenal spans approximately 7,000 years, consisting of both explosive (tephra) eruptions and more effusive (lava) eruptions. Over the past 4,000 years Arenal has produced tephras layers roughly every 300 years, although there are two 500-600 year gaps in volcanic activity between 3000 and 1800 years ago. This study collected samples from 20 tephra and 10 lava units from the past 4,000 years and one lava and tephra unit dated at 7,000 years. The purpose of this study is to examine the geochemical evolution of the magma erupting at Arenal. Tephras from Arenal divide into two groups on the basis of SiO2 content. Mafic tephras have up to 53 wt.% SiO2, whereas felsic tephras have at least 56 wt.% SiO2. The gap in the SiO2 contents of the tephras is filled by lavas, which have SiO2 contents of 53-56 wt.%. The tephras that erupted prior to a hiatus at 3000 years ago consist mostly of mafic tephras. After the gap, tephras alternate between mafic and felsic. At approximately the time of this gap, the degree of melting, estimated from Zr/Hf and Nb/Ta, and the slab signal, estimated from Ba/La and Pb/Ce, also changed at Arenal. Prior to the gap the degree of melting is increasing along with the slab signal, both then begin decreasing after the gap. Over the past 4,000 years the mafic explosive tephras, which all have similar SiO2 contents, show an increase in compatible elements and a decrease in incompatible elements. This can also been seen in the felsic tephras, though not as well, since these tephras are not present throughout the whole eruptive history. This change in compatible and incompatible elements over time demonstrates fractionation occurring in the magma chamber prior to the entire eruptive sequence. The lavas, however, do not show this change in compatible and incompatible elements with time.
V23D-02 INVITED 13:55h
Closed to Open-System Crustal-level Differentiation During Eruption of Andesite: Arenal, Costa Rica, 1968-2003
Arenal volcano has erupted $\sim$ 0.5 km3 of medium-K tholeiitic basaltic andesite continuously for $>$35 years. We report new high precision whole rock major and trace element and Sr-Nd-Hf-Pb isotope data for $>$50 samples from throughout the eruption. Lava compositions during the first half of the eruption by volume can be related by $\sim$ 20% closed system fractional crystallization of the phenocryst minerals (Opx$>$Plag$>$Cpx$>$Mt) with $\Sigma$r2 = 0.01 for major elements, and acceptable D's for all trace elements. Ratios of isotopes and incompatible trace elements are constant. MELTS models best approximate this at water-undersaturated conditions: P=4 kb, T=1190oC, H2O = 2.5 wt%, and fO2 = QFM+2. The match of melt and mineral compositions is imperfect, but results can be reconciled qualitatively by polybaric crystallization extending to water-saturated conditions at $<$3 km. The differentiates erupted first from the shallowest part of the storage system and progressed to less differentiated compositions. Differentiation processes subsequently changed. Although some differentiation indices have returned to or exceeded initial values, compatible elements have remained constant or declined only slightly, and Pb isotope and initially-constant trace element ratios have changed. The change has been continuous in some cases but lasted for only several years in others (Pb isotopes, Th/U, Pb/Ce). In all cases, the more recent magmas have less of a ``subduction signature'' (lower Ba/La, Pb/Ce, excess 238U). We attribute these trends to an evolving balance between recharge, crystallization, and eruption. The recharging magma reflects less flux melting of an isotopically similar source that is transitional between that of central Costa Rica and Nicaragua.
V23D-03 14:10h
Vent , Voluminous Lava Emissions, Steep Slopes and Pyroclastic Flows at Arenal Volcano, Costa Rica
The initial explosion of Arenal Volcano, occurred July 29-31, 1968, opened three new craters on the west flank of the volcano. The first lava flow was issued from crater A (1050 m) on September 20, 1968. This crater continued the emission of lava flows until November, 1973. In April, 1974 crater C (1460m) erupted its first lava flow. Between 1974 and 2004 crater C has been constructing a cone which is off set 500 m from the prehistoric summit Arenal cone (1633 m) and has reached an elevation of 1670 m. The slopes of the present day cone developed on crater C are very steep, specially north, northeast and northwest of the cone. Crater C lavas were issued from two vents: south and north vents. The location of the vents inside crater C controlled the direction of flow of the lavas issued from each vent. During the last 30 years both vents of crater C have been active at different times, and it was only in the late eighties during a period of intense strombolian activity, that both vents were active at the same time. During the highest intensity of strombolian activity (1987-1989) fall back ejecta from strombolian explosions produced pyroclastic flows. The shift of activity from the south vent to the north vent coincided with the generation of the two largest pyroclastic flows issued by Arenal volcano since 1968. During July, 1975 a very voluminous lava flow erupted from the north vent of crater C was descending the NW flank of Arenal when gravitational collapse of the flow produced the 1975 Tabacon River valley pyroclastic flow. This was the first time that a lava was flowing on the very steep, partially altered unconsolidated slopes of the NW flank of Arenal. This pyroclastic flow reached Arenal River located 4.3 km from the summit, this is the largest pyroclastic flow produced by gravitational collapse at Arenal volcano. A similar event occurred in August, 1993 when the migration of activity from the south vent to the north vent of crater C and the emission of a voluminous lava flow, preceded the collapse of its NW rim. This rim consisted mainly of loose bombs and blocks accumulated after the period of strombolian eruptions during the late eighties and early nineties. After 1999, the south vent of crater C deflated showing a crateric depression. In contrast, the north vent of crater C became inflated and filled with lava. New vents opened on the north vent rim of crater C. They formed cones located NE and SW of the vent. Both cones have been active, nevertheless the most active vent has been the NE vent after 1999. The nature of these cones have been discussed extensively as there are evidences that small lobes have been extruded from their flanks suggesting a dome-like behavior. In parallel, after 1999, lava flows were shorter, developing blocks that cascaded the slopes of the volcano at very short distances from the vent. Several small volume, gravitational collapse pyroclastic flow events originated from the NE cone of the north vent of crater C since 1999. They represent the most important volcanic hazard for visitors of the volcano that get too close to see the volcano. Plume heights, erupted volumes and the number of daily explosions have been decreasing after 1999, suggesting an important decrement in the activity of Arenal volcano.
V23D-04 INVITED 14:25h
Support from Mineral and Melt-Inclusion Data for a Flux-controlled Ascent Model to Explain Longevity, Magnitude, and Composition of the Current Arenal Eruption
Arenal volcano is famous for its ongoing, small-scale, continuous activity ever since it reawakened in July of 1968. Although activity levels have generally declined from early in the eruption to now, multiple daily explosions and associated lava production have persisted and continue to be characteristic for Arenal's current eruption. The combination of small eruption volumes, decades-long continuous activity, and eruption products that are remarkable compositionally similar, phenocryst-rich basaltic andesites provoke pressing questions about eruption driving force and petrogenetic history. Our key results indicate all eruption products exhibit significant but comparable complexities in mineral compositions, zoning and distributions requiring multi-stage mixing. Bulk compositions are mostly too mafic to have crystallized the majority of ferromagnesian minerals yet appropriate liquids are preserved in melt inclusions from tephras. Furthermore, no progressive changes in zoning near rims of phenocrysts over the course of the current eruption are apparent which complicates scenarios in which one, evolving magma batch has been tapped for the last 36 years. Evidence for replenishment events with more primitive magma is often found in cpx phenocrysts in form of discrete growth bands of high Mg\#, Cr-rich cpx bound by low Mg\# cpx. Modeling the diffusive equilibration of Fe-Mg gradients across bands yielded times of the latest recharge within a crystal prior its eruption of $<$ 1 to $\sim$200 years suggesting variable residence times of phenocrysts and that mafic recharge can closely predate eruption. Our results fit best a model in which similar basaltic andesites are repeatedly generated from mantle magma batches during their ascent as they mix with resident magmas, fractionate, and assimilate crystals. We infer that new increments of basaltic andesite are continuously blended into the eruption-feeding reservoir concurrently to the current activity. We propose that observed longevity, magnitude, and monotonous composition of the current Arenal eruption mirror the combined result of duration, mass, and frequency of mantle inputs moving through the crust.
V23D-05 14:40h
High water contents in basaltic melt inclusions from Arenal volcano, Costa Rica
Despite the importance of water to arc magma genesis, fractionation and eruption, few quantitative constraints exist on the water content of Arenal magmas. Early estimates, by electron microprobe sum deficit, suggested up to 4 wt% H$_{2}$O in olivine-hosted basaltic andesite melt inclusions (MI) from pre-historic ET-6 tephra (Melson, 1982), and up to 7 wt% H$_{2}$O in plagioclase and orthopyroxene-hosted dacitic MI from 1968 lapilli (Anderson, 1979). These high water contents are consistent with abundant hornblende phenocrysts in Arenal volcanics, but inconsistent with geochemical tracers such as $^{10}$Be and Ba/La that suggest a low flux of recycled material (and presumably water) from the subduction zone. In order to test these ideas, and provide the first direct measurements of water in mafic Arenal magmas, we have studied olivine-hosted MI from the prehistoric (900 yBP; Soto et al., 1998) ET3 tephra layer. MI range from andesitic ($>$ 58% SiO$_{2}$) to basaltic compositions ($<$ 50% SiO$_{2}$), the latter of which are similar to the most primitive whole rocks analyzed from Arenal. SIMS analyses yield up to 4 wt% H$_{2}$O in the basaltic MI, and water declines systematically (to 1-2 wt%) with increasing silica content. Water also correlates strongly with sulfur (up to 2500 ppm S) and CO$_{2}$ (up to 300 ppm). H$_{2}$O and CO$_{2}$ in the MI define a closed-system degassing path that begins at 2 kb. Chlorine ($\sim$ 2000 ppm) and fluorine ($\sim$ 400 ppm) show less variation, as expected from their higher solubilities in these melts. The high sulfur contents ($\sim$ 2000 ppm on average for basaltic MI) would provide more than enough "petrologic" sulfur to balance recent (1982, 1995, and 1996; Williams-Jones et al., 2001) COSPEC measurements. Although host olivines are quite evolved ($<$ Fo$_{76}$), the high CO$_{2}$ and sulfur contents indicate that their inclusions are not highly degassed. The high water contents ($>$ 4 wt%) found here for Arenal basaltic MI support the semi-quantitative data from earlier studies, but are somewhat unexpected given predictions from slab tracers. Arenal water contents (4%) approach those of the 1995 eruption of Cerro Negro in Nicaragua (4-5 wt% in basaltic MI; Roggensack et al., 1997), despite the fact that the latter has Ba/La of $>$ 100, while Arenal has Ba/La $<$ 50. Thus, Ba/La, which has commonly been used to infer the proportion of slab-fluids and water-fluxed melting in Central American magmas, seems to be a poor water proxy. Even more surprising are the very high Cl contents of Arenal MI, which exceed those in Cerro Negro MI by more than a factor of two. The high Cl and high Cl/K$_{2}$O (0.3-0.5) of Arenal, are also inconsistent with an enriched, Galapagos-type mantle (Cl/K$_{2}$O $<$ 0.1), as has been proposed for this part of Costa Rica. Thus, contrary to previous inferences, Arenal receives a major flux of H$_{2}$O and Cl from the subduction zone. References: Anderson, A.T. (1979) Journal of Geology; Melson, William G. (1982) Boletin de Volcanologia; Roggensack et al. (1997) Science; Soto et al. (1998) OSIVAM; Williams-Jones et al. (2001) Journal of Volc. and Geoth. Res.
V23D-06 14:55h
Degassing Mechanisms and Timescales of Implied by ($^{210}$Pb) Values for Andesites Erupted from Arenal Volcano
The ongoing eruption of Arenal, which began in 1968, is an ideal laboratory for investigating magmatic processes that occur over short time periods during eruptions. To identify and place time constraints on these processes, lavas from throughout this eruption have been analyzed for ($^{210}$Pb) (t$_{1/2}$ = 22.6 y). Because Pb is both incompatible and only weakly volatile, variations in ($^{210}$Pb)/($^{226}$Ra) largely monitor decade-scale fluxes of $^{222}$Rn through magmas. At present, only one lava has been analyzed for $^{226}$Ra, and the following discussion assumes that Ra varies in concert with other highly incompatible elements in Arenal lavas. By meeting time, additional whole rock $^{226}$Ra values will be available to further constrain this discussion. The eruption has been divided into two principal stages based on variations in bulk composition (Ryder, C., 2004, MS Thesis, UCSC). The first stage lasted from the beginning of the eruption until the early 1970s when Pb isotopes shifted. This shift marked the end of the eruption from one reservoir and the appearance of a new magma presumably from a deeper chamber. Lavas and tephras erupted in 1968 have small excesses in $^{210}$Pb over calculated $^{226}$Ra values, whereas those erupted in 1969 have $^{210}$Pb deficits. These data are consistent with decade-scale transfer of $^{222}$Rn from the less-differentiated lower portion of the original magma reservoir to the more differentiated and more phenocryst poor upper reservoir. This could occur either by diffusion of $^{222}$Rn through the melt within the chamber or by transfer in a separate gas phase in vapor saturated magma. In 1971, just before the shift in Pb isotopes, lavas erupted with an approximately 2-fold $^{210}$Pb excess over $^{226}$Ra, which suggests that the deeper magma that eventually erupted and shifted Pb-isotope values contributed volatiles to the lower portion of the original reservoir. If the excess in 1971 was due to $^{222}$Rn fluxing since 1968, then the average ($^{222}$Rn)/( $^{226}$Ra) in the lava-gas mixture must have been about 10-fold. Such large magnitude excesses can only occur if the Rn in streaming gasses was generated by ingrowth and extracted from a larger volume of magma than for other volatiles. By 1973, just before a several month pause in the eruption, ($^{210}$Pb)/( $^{226}$Ra) approached equilibrium values showing that these lavas did not received the sustained flux of $^{222}$Rn. Upon resumption of eruption, $^{210}$Pb excesses remained small but systematically climbed until 1996. By 1999, these excesses decreased again. These excesses are explainable by a constant source of excess $^{222}$Rn through the rising magmas from 1974 until about 1996 when the source of the Rn was terminated by a slackening recharge (see Ryder, 2004). The recent decrease in excess $^{210}$Pb could be a precursor for a different mode of activity for Arenal. Continued crystallization combined with lessened recharge could increase viscosity, which could terminate the eruption or cause it to become more explosive.
V23D-07 15:10h
Phase Equilibrium Experiments at 0.5 GPa and 1100-1300 deg. C on a Basaltic Andesite From Arenal Volcano, Costa Rica.
The on-going eruption of Arenal volcano has undergone complex changes in eruptive behavior. To investigate pre-eruptive crustal processes, we conducted phase equilibrium experiments on a basaltic andesite. The anhydrous synthetic starting mix has a Mg\# of 52 and concentrations of SiO$_{2}$, Na$_{2}$O and K$_{2}$O of 55, 3.0 and 0.6 wt.%, respectively. Al(OH)$_{3}$ was used to obtain bulk water concentrations of 2 and 4 wt.%. Starting mixes were loaded into Pt/C capsules and each run contained three capsules with nominal water contents of 0, 2, and 4 wt.%. Experiments were run in a piston-cylinder apparatus with NaCl/pyrex/MgO assemblies and S-type thermocouples. Water loss is a potential problem in wet experiments: the actual water contents in quenched glasses remain to be quantified and results presented here are only from runs in which melt fractions systematically increased with initial water content at a given constant temperature. At $1300\deg$C the dry composition yields glass coexisting with traces of spinel and plag, and both hydrous compositions are above the liquidus. Glass, plag (An$_{75-82}$) and traces of spinel are present for all three compositions at $1250\deg$C. At $1200\deg$C, cpx is present in the dry composition with plag and glass; increasing water content enhances the plag stability (An$_{77-69}$) at the expense of cpx, which is replaced by low-Ca pyroxene. Traces of spinel are present and are found in all experiments at lower temperatures, independent of water content. Cpx, low-Ca pyroxene, plag and andesitic glass are present at $1150\deg$C in the dry and 2% water compositions, but the 4% water composition yields only cpx and no low-Ca pyroxene. Due to the overall small grain size (Íš10 $\mu$m) in our experiments, low-Ca pyroxene may have been overlooked, yet the lower plag mode and its decreasingly anorthitic character (An$_{64-59}$) may account for the absence of low-Ca pyroxene. At $1100\deg$C, glasses become dacitic to rhyolitic and coexist with low-An plag and cpx, and in the case of both hydrous compositions, also ilmenite. Olivine was not identified in any experiment. The plag range seen at and above $1150\deg$C covers the lower end of the plag range in erupted lavas but fails to account for An$_{85-95}$ phenocrysts also present in Arenal lavas.
V23D-08 15:25h
Diffusion-Reaction Between Basaltic Andesite and Gabbro at 0.5 GPa: an Explanation for Anorthitic Plagioclase?
Despite the remarkably smooth variation in bulk composition of erupted lavas at Arenal volcano (1968-2003), mineral compositions vary widely. Plagioclase ranges from An$_{52}$ to An$_{95}$ while Cr2O3 in CPX varies from 0.7 to 0.05 wt % (Streck et al., 2003). To address the question "how do bulk compositions remain near-steady-state while crystal compositions vary widely," we have performed 2 diffusion-reaction experiments in the piston cylinder at 0.5 GPa. These juxtaposed Arenal basaltic andesite AR-8 at $1200\deg$C with a Stillwater Complex gabbro, lying in a thermal gradient toward the piston. In one experiment, we synthesized a glass-plagioclase (An$_{67-75}$) aggregate of AR-8 in a graphite-Pt-Ti capsule at P-T, polished one end, dried tracer solutions of 45Ca, 6Li, 84Sr and 136Ba on its surface, and juxtaposed it with gabbro for 13 days. Profiles of bulk composition as a function of distance from the interface show that AR-8 gains Al2O3, MgO and CaO from the gabbro and loses Na2O, K2O, SiO2 and FeO to it. Notably, a plagioclase rich (65%) layer develops at the interface between the two materials as CPX disappears. This layer and the compositional profiles are reproduced by diffusion-reaction models using IRIDIUM (Boudreau, 2003). Plagioclase at the interface develops a texture of homogeneous anorthitic cores (An$_{90}$) that abruptly shift to 10$\mu$m rims having compositions (An$_{67}$) in Na-Ca exchange equilibrium with the co-existing melt. A beta track map shows that 45Ca is incorporated into the plagioclase cores while SIMS analyses indicate isotopic equilibration between core and melt. Thus, these anorthitic plagioclase result from diffusion-reaction with efficient chemical communication between the melt and the plagioclase core. Microchannels cutting through the rim, rather than solid-state diffusion, appear to control re-equilibration. Other observations from the experiment parallel Arenal lavas: Mg# variation in OPX is small in both experiments and lavas while profiles of Cr show that significant amounts of Cr move from gabbro to AR-8, possibly explaining Cr2O3 variations in Arenal CPX. The near-steady-state behavior at Arenal could reflect a flux balance between ascending magmas and melt from the surrounding crust reflecting diffusion-reaction.