Lava lake oscillations and the magma reservoir beneath a volcano

Stagnation of magma beneath a volcano very likely produces a considerable body of magma, the so called magma reservoir. Assuming an active lava lake being connected with an underneath magma reservoir through a vertical conduit, the height of the surface of the lava lake may be expected to show tidal fluctuations which are caused by squeezing out and draining back of magma from a magma reservoir due to earth tides. Examples are shown in the case of Halemaumau lava lake, Kilauea, in 1919. A similar behaviour also appeared in 1968 which showed semidiurnal tilt of the summit area. It is interesting to notice that the semidiurnal oscillation of the surface of the lava lake appeared only at the heighest level of the lava lake activity. This evidence implies that during the early stage of the activity, a part of the lava filled feeding dikes and open cracks and consequently tidal oscillations of the lava lake were masked and could not be observed.     

A Nd,Sr and O isotopic investigation into the causes of chemical and isotopic zonation in the Bishop Tuff,California

The Bishop Tuff represents a single eruption of chemically zoned rhyolitic magma. Six whole rock samples spanning the compositional and temperature range yield initial⁸⁷Sr/⁸⁶Sr of 0.7060–0.7092 andδ¹⁸O of 5.9–10.3‰. Six constituent sanidines yield smaller ranges of initial⁸⁷Sr/⁸⁶Sr of 0.7061–0.7069 andδ¹⁸O of 6.7–7.9. In contrast¹⁴³Nd/¹⁴⁴Nd ratios for the six whole rocks and two constituent magnetites exhibit negligible variation with a mean of0.51258 ± 1. These data are used to show that the phenocrysts were precipitated from an already chemically zoned liquid, that the zoning process involved negligible assimilation of, or exchange with, country rocks and that the extreme Sr and O isotopic disequilibria are probably the result of post-eruptive interaction with meteoric water. The parent magma had?Nd = ?0.9, ?Sr = +23 andδ¹⁸O = 7‰ and was formed from mantle-derived magmas and/or melts of lower crustal rocks isotopically similar to parts of the Sierra Nevada Batholith.     

On magma fragmentation by conduit shear stress: Evidence from the Kos Plateau Tuff,Aegean Volcanic Arc

Large silicic explosive eruptions are the most catastrophic volcanic events. Yet, the intratelluric mechanisms underlying are not fully understood. Here we report a field and laboratory study of the Kos Plateau Tuff (KPT, 161 ka, Aegean Volcanic Arc), which provides an excellent geological example of conduit processes that control magma vesiculation and fragmentation during intermediate- to large-scale caldera-forming eruptions. A prominent feature of the KPT is the occurrence of quite unusual platy-shaped tube pumice clasts in pyroclastic fall and current deposits from the early eruption phases preceding caldera collapse. On macroscopic and SEM observations, flat clast faces are elongated parallel to tube vesicles, while transverse surfaces often occur at ~ 45° to vesicle elongation. This peculiar pumice texture provides evidence of high shear stresses related to strong velocity gradients normal to conduit walls, which induced vesiculation and fragmentation of the ascending magma. Either an increasing mass discharge rate without adequate enlargement of a narrow central feeder conduit or a developing fissure-like feeder system related to incipient caldera collapse provided suitable conditions for the generation of plate tube pumice within magma volumes under high shear during the pre-climactic KPT eruption phases. This mechanism implies that the closer to the conduit walls (where the stronger are the velocity gradients) the larger was the proportion of plate vs. conventional (lensoid) juvenile fragments in the ascending gas–pyroclast mixture. Consequently, plate pumice clasts were mainly entrained in the outer portions of the jet and convecting regions of a sustained, Plinian-type, eruption column, as well as in occasional lateral blast currents generated at the vent. As a whole, plate pumice clasts in the peripheral portions of the column were transported at lower altitudes and deposited by fallout or partial collapse closer to the vent relative to lensoid ones that dominated in the inner column portions. The plate tube pumice proportion decreased abruptly up to disappearance during the emplacement of the main pyroclastic currents and lithic-rich breccias related to extensive caldera collapse at the eruption climax, as a consequence of an overall widening of the magma feeder system through the opening of multiple conduits and eruptive vents, along with fissure erosion, concomitant to the disruption of the collapsing block.     

The characteristics of magma reservoir failure beneath a volcanic edifice

Eruptions fed from subsurface reservoirs commonly construct volcanic edifices at the surface, and the growth of an edifice will in turn modify the subsurface stress state that dictates the conditions under which subsequent rupture of the inflating reservoir can occur. We re-examine this problem using axisymmetric finite element models of ellipsoidal reservoirs beneath conical edifices, explicitly incorporating factors (e.g., full gravitational loading conditions, an elastic edifice instead of a surface load, reservoir pressures sufficient to induce tensile rupture) that compromise previous solutions to illustrate why variations in rupture behavior can occur. Relative to half-space model results, the presence of an edifice generally rotates rupture toward the crest of a spherical reservoir, with increasing flank slope (for an edifice of constant volume) and larger edifices (or greater reservoir scaled depths) normally serving to enhance this trend. When non-spherical reservoirs are considered, the presence of an edifice amplifies previously identified half-space failure characteristics, shifting rupture to the crest more rapidly for prolate reservoirs while forcing rupture closer to the midpoint of oblate reservoirs. Rupture is always observed to occur in the σ_t orientation, and depending on where initial failure occurs rupture favors the initial emplacement of either lateral sills, circumferential intrusions or vertically ascending dikes. Ultimately, integration of our numerical model results with other information, for instance the sequence of intrusion/eruption events observed at a given volcano, can provide useful new insight into how a volcano's subsurface magma plumbing system evolved. We demonstrate this process through application of our model to Summer Coon, a well-studied stratocone on Earth, and Ilithyia Mons, a large conical shield volcano on Venus.     

Mechanical behavior of magma reservoir envelopes: Elasticity of the olivine tholeiite solidus

A study of the elasticity of the solidus of igneous rocks provides a means of evaluating several high temperature mechanical characteristics of the zone that forms the transition region between solid rock and rock with a small fraction of partial melt — that is, the magma reservoirenvelope. This type of study has been performed for Hawaiian olivine tholeiite, by employing the theoretical treatment ofWalsh (1969). The effective aggregate bulk modulus (K*), aggregate rigidity (μ*), aggregate Young’s modulus (E*) and the aggregate Poisson’s ratio (n*) have been used to characterize the range in elastic behavior expected for this partially molten envelope. The details of mineralogy, mineral crystal chemistry and microstructure for a tholeiitic basalt from a deep-ponded flow unit (the Boiling Pots), as well as the geometry of melt pocket aspect ratios have been used to provide a basis for computing the elasticity for a shallow reservoir solidus. For a broad range in modal mineralogy, the internal 0–1% melt represents a transition region from an aggregate elasticity controlled by mineral composition, to an elastic behavior dominated by the melt phase (evaluated at a melt pocket aspect ratio ofε=0.001). For vesicular rocks in a low pressure environment, the internal 0–1% melt represents a transition from porosity-controlled elasticity to melt-controlled elasticity, evaluated at a prevalent aspect ratio ofε=0.001, and over the porosity range Φ=0.0?0.20 (20%). Thin melt films found to separate plagioclase microlites (0.0001=ε) are capable of effecting a 94.7% reduction in Young’s modulus, E*, over the 0–1% melt range, and suggest that the reservoir envelope may be critically weakened by rather small volume fractions of liquid, if dispersed in such narrow packets.     

Controls on magma permeability in the volcanic conduit during the climactic phase of the Kos Plateau Tuff eruption (Aegean Arc)

X-ray computed microtomography (μCT) was applied to pumices from the largest Quaternary explosive eruption of the active South Aegean Arc (the Kos Plateau Tuff; KPT) in order to better understand magma permeability within volcanic conduits. Two different types of pumices (one with highly elongated bubbles, tube pumice; and the other with near spherical bubbles, frothy pumice) produced synchronously and with identical chemical composition were selected for μCT imaging to obtain porosity, tortuosity, bubble size and throat size distributions. Tortuosity drops on average from 2.2 in frothy pumice to 1.5 in tube pumice. Bubble size and throat size distributions provide estimates for mean bubble size (~93–98 μm) and mean throat size (~23–29 μm). Using a modified Kozeny-Carman equation, variations in porosity, tortuosity, and throat size observed in KPT pumices explain the spread found in laboratory measurements of the Darcian permeability. Measured difference in inertial permeability between tube and frothy pumices can also be partly explained by the same variables but require an additional parameter related to the internal roughness of the porous medium (friction factor f ₀ ). Constitutive equations for both types of permeability allow the quantification of laminar and turbulent gas escape during ascent of rhyolitic magma in volcanic conduits.     

The difference in chemical composition of the 1962 lava from miyakezima magma reservoir in Japan

Many examples of mixed magmas in banded lavas have been studied. Another type of mixed magmas or inhomogeneity of magma reservoir found in the 1962 lava flows of Miyake-zima Volcano erupted from fissures is reported.     

A thermal model of the magma reservoir feeding Plinian eruptions at Mt. Vesuvius (Italy)

A theoretical thermal model has been worked out for the magma reservoir that would have fed the two last Plinian eruptions of Mt. Vesuvius (Barberi et al., 1981). The effect of convective motions is discussed, and it is shown that the size of convective cells and the efficiency of the process in smoothing out temperature gradients evolves in time due to the progressive viscosity increase produced by the heat lost by conductive heat transfer through the host rock. Although convection will be important throughout the history of the reservoir, until very high viscosities are reached, the pure conductive model seems to account satisfactorily for the cumulative heat loss by the reservoir. Gravitative crystal settling can occur, even in presence of convective motions, mostly during several hundred years after the magma emplacement when viscosity is not yet increased to high values.     

Pre-1991 sulfur transfer between mafic injections and dacite magma in the Mt. Pinatubo reservoir

Before the 1991–1992 activity, a large andesite lava dome belonging to the penultimate Pinatubo eruptive period (Buag ∼ 500 BP) formed the volcano summit. Buag porphyritic andesite contains abundant amphibole-bearing microgranular enclaves of basaltic–andesite composition. Buag enclaves have lower K₂O and incompatible trace element (LREE, U, Th) contents than mafic pulses injected in the Pinatubo reservoir during the 1991–1992 eruptive cycle. This study shows that Buag andesite formed by mingling of a hot, water-poor and reduced mafic magma with cold, hydrous and oxidized dacite. Depending on their size, enclaves experienced variable re-equilibration during mixing/mingling. Re-equilibration resulted in hydration, oxidation and transfer of mobile elements (LILE, Cu) from the dacite to the mafic melts and prompted massive amphibole crystallization. In Buag enclaves, S-bearing phases (sulfides, apatite) and melt inclusions in amphibole and plagioclase record the evolution of sulfur partition among melt, crystal and fluid phases during magma cooling and oxidation. At high temperature, sulfur is partitioned between andesitic melt and sulfides (Ni-pyrrhotite). Magma cooling, oxidation and hydration resulted in exsolution of a S–Cl–H₂O vapor phase at the S-solubility minimum near the sulfide–sulfate redox boundary. Primary magmatic sulfide (pyrrhotite) and xenocrystic sulfide grains (pyrite), recycled together with olivines and pyroxenes from old mafic intrusives, were replaced by Cu-rich phases (chalcopyrite, cubanite) and, partially, by Ba–Sr sulfate. Sulfides degassed and transformed into residual spongy magnetite in response to fS₂ drop during final magma ascent and decompression. Our research suggests that a complete evaluation of the sulfur budget at Pinatubo must take into account the en route S assimilation from the country rocks. Moreover, this study shows that the efficiency of sulfur transfer between mafic recharges and injected magmas is controlled by the extent and rate of mingling, hydrous flushing and melt oxidation. Vigorous mixing/mingling and transformation of the magmatic recharge into a spray of small enclaves is required in order to efficiently strip their primary S-content that otherwise remains locked in the sulfides. Hydrous flushing increases the magma oxidation state of the recharges and modifies their primary volatile concentrations that cannot be recovered by the study of late-formed mineral phases and melt inclusions. Conversely, S stored in both late-formed Cu-rich sulfides and interstitial rhyolitic melt represents the pre-eruptive sulfur budget immediately available for release from mafic enclaves during their decompression.     

Formation of the Green Tuff,Pantelleria

The strongly peralkaline Green Tuff, Pantelleria, is an example of a thin, densely welded air-fall tuff which mantles an area of at least 85 km². Offshore the tuff is correlated with the Y-6 ash layer in the central Mediterranean Sea, and the total volume of the eruption is estimated at 7 km³ D.R.E. New petrological data suggests that the tuff was erupted from a zoned magma chamber containing a cooler, more fractionated upper zone relative to be bulk of the magma. Analysis of the distribution of accessory lithic fragments in terms of existing models of eruption dynamics indicates emplacement by a plinian-type eruption. It is shown that, due to the low viscosity of pantelleritic ejecta, dense welding can occur at moderate tephra accumulation rates and a rate of the order of 1 cm/minute is suggested for the Green Tuff; this yields an estimate for the eruption duration of rather less than one day. It is predicted that welded tuff should be formed during large plinian eruptions of pantelleritic magma, and therefore that welded airfall tuffs should be common in areas of peralkaline volcanism.     

Incremental growth of a large volume,chemically zoned magma body: a study of the tephra sequence beneath the Rainier Mesa ash flow sheet of the Timber Mountain Tuff

The Rainier Mesa ash-flow is a large (1200 km³), 11.6 My old, chemically zoned unit that ranges in composition from 55 to 76% SiO₂ — one of the largest chemical ranges ever observed in a large volume ash-flow sheet. Two chemical trends occur in this sheet, a low silica (55–66% SiO₂) and a high silica (>66% SiO₂) trend. Ninety per cent of the Rainier Mesa sheet occurs in the high silica trend. Immediately beneath the Rainier Mesa sheet is a thick tephra sequence. The chemical variation of this sequence is nearly equivalent to the high silica portion of the Rainier Mesa ash-flow sheet (about 66–78% SiO₂). Throughout the tephra sequence numerous small ash-flow layers occur, and each ash-flow layer is chemically zoned from more evolved at the base to less evolved at the top. This is consistent with having been erupted from a zoned magma body. The lowest silica tephra units are at the base of the sequence and the highest silica units are at the top — that is, the large-scale chemical trend of the entire sequence is opposite to that of the individual ash-flow layers. These ash-flow layers are of very small volume. The tephra sequence provides a unique record of the incremental development of the zoned, high silica portion of the Rainier Mesa magma body.     

Metals contamination along the watershed and estuarine areas of southern Bohai Sea,China

Distributions and magnitude of metals in water, sediment and soil collected from the watershed and estuarine areas of southern Bohai Sea, were investigated. The largest dissolved concentrations of As, Cu and Zn in water were 347.70, 2755.00, 2076.00 μg/L, respectively, much higher than corresponding drinking water guidelines. The greatest concentrations of Cu, Zn, Cr, Ni, Pb, As and Cd in sediments were 1462.2, 1602.17, 196.43, 67.15, 63.54, 73.86 and 1.41 mg/kg, dw, respectively. The mean concentrations of Cu, Ni, Cd, Zn, Cr, Pb and As in soils were 24.67, 24.73, 0.14, 64.75, 56.52, 25.12 and 9.34 mg/kg, dw, respectively. Land use was confirmed to be an important factor of influence on soil metal concentrations. Metal contents along the watershed of Jie River were significantly greater than in other locations. The detection of metals in relatively high concentrations from different environmental matrices in this region indicates the necessity of further studies.     

First results on the study of metal contamination along the Corsican coastline using Posidonia oceanica

This study aims at determining the metal concentrations in blades and sheaths of Posidonia oceanica adult leaves, in 16 stations of the Corsican coastline. It shows that except for Cr, all the metals are preferentially accumulated in the blades. This result is particularly interesting as it means that trace metals analyses may be carried out only on the blades avoiding thus the removal of the shoots. Moreover, this study shows that metal concentrations generally fall within the range of the lowest values available in literature and may reflect the "background noise" of the Mediterranean. Station 15 (Canari) can however be distinguished from the others due to its high Co, Cr and Ni concentrations. This result may be related to the presence of a previous asbestos mine, located near this station. Therefore, this study reinforces the relevance of the use of P. oceanica as a tracer of metal contamination.     

Evidence for magma in the Katmai volcanic range

Several independent observations during the summer of 1965 suggest the presence of magma in the volcanic range of Katmai. A high value of 0.3 for Poisson’s ratio and the screening of predominantly the vertical component of the elastic shear waves have been observed. Narrow negative Bouguer anomalies possibly indicate the presence of low density material at shallow depth. The location of magma reservoirs has been attempted, using the calculated wave path and the screening of the mainly vertically polarized shear wave. Of the possible ten chambers thus located, the ones of shallow depth (to 20 km) correspond to the location of active volcanoes. The ones between the 20 km level to the upper mantle seem to spread over a rather wide area and are not clearly related to the geographical position of a particular volcano. Theoretical considerations on the propagation of elastic waves substantiate the observed absence of vertically polarized shear waves.     

Geochemical modeling of magma mixing and magma reservoir volumes during early episodes of Kīlauea Volcano’s Pu‘u ‘Ō‘ō eruption

Geochemical modeling of magma mixing allows for evaluation of volumes of magma storage reservoirs and magma plumbing configurations. A new analytical expression is derived for a simple two-component box-mixing model describing the proportions of mixing components in erupted lavas as a function of time. Four versions of this model are applied to a mixing trend spanning episodes 3–31 of Kilauea Volcano’s Puu Oo eruption, each testing different constraints on magma reservoir input and output fluxes. Unknown parameters (e.g., magma reservoir influx rate, initial reservoir volume) are optimized for each model using a non-linear least squares technique to fit model trends to geochemical time-series data. The modeled mixing trend closely reproduces the observed compositional trend. The two models that match measured lava effusion rates have constant magma input and output fluxes and suggest a large pre-mixing magma reservoir (46±2 and 49±1 million m³), with little or no volume change over time. This volume is much larger than a previous estimate for the shallow, dike-shaped magma reservoir under the Puu Oo vent, which grew from ∼3 to ∼10–12 million m³. These volumetric differences are interpreted as indicating that mixing occurred first in a larger, deeper reservoir before the magma was injected into the overlying smaller reservoir. Electronic Supplementary Material Supplementary material is available at and is accessible for authorized users.     

Age and isotope geochemistry of the Archaean Pongola and Usushwana suites in Swaziland, southern Africa: a case for crustal contamination of mantle-derived magma

The igneous rocks of the Pongola Supergroup (PS) and Usushwana Intrusive Suite (UIS) represent a case of late Archaean continental magmatism in the southeastern part of the Kaapvaal craton of South Africa and Swaziland.

U-Pb dating on zircons from felsic volcanic rocks of the PS yields a concordia intercept age of 2940 ± 22Ma that is consistent with a Sm-Nd whole rock age of 2934 ± 114Ma determined on the PS basalt-rhyolite suite. The initial ε_Nd of−2.6 ± 0.9 is the lowest value so far reported for Archaean mantle-derived rocks. Rb-Sr whole rock dating of the PS yields a younger isochron age of 2883 ± 69Ma, which is not significantly different form the accepted U-Pb zircon age.

An internal (cpx-opx-plag-whole rock) isochron for a pyroxenite from the younger UIS yields an age of 2871 ± 30 Ma and initial ¹⁴³Nd/¹⁴⁴Nd that lies off the CHUR growth curve by ε_Nd −2.9 ± 0.2. However, Sm-Nd whole-rock data for the UIS yield an excessively high age of 3.1 Ga that conflicts with firm geological evidence showing the UIS to be intrusive into the PS.

The negative deviations of initialε_Nd from the chondritic Nd evolution curve suggest significant contamination of the PS and UIS melts by older continental crust. A mixing process with continental crust after magma segregation is supported by a high initial ⁸⁷Sr/⁸⁶Sr ratio of0.703024 ± 24 for a clinopyroxene sample from a UIS pyroxenite, compared with an expected value of 0.701 for the 2.9 Ga mantle. We therefore interpret the linear array of data points for the UIS gabbros as a mixing line between 2.87 Ga old magma and older continental crust.

Parallel LREE-enriched REE patterns, negative Nb-Ti anomalies, a distinctive and uniform ratio of Ti/Zr 46 and a narrow span of initial Nd indicate a common source for both the PS and UIS suites which is different from primitive mantle.     

Gravitational spreading causes en-echelon diking along a rift zone of Madeira Archipelago: an experimental approach and implications for magma transport

Many volcanic rift zones show dikes that are oriented oblique rather than parallel to the morphological ridge axis. We have evidence that gravitational spreading of volcanoes may adjust the orientation of ascending dikes within the crust and segment them into en-echelon arrays. This is exemplified by the Desertas Islands which are the surface expression of a 60 km long submarine ridge in southeastern Madeira Archipelago. The azimuth of the main dike swarm (average = 145°) deviates significantly from that of the morphological ridge (163°) defining an en-echelon type arrangement. We propose that this deviation results from the gravitational stress field of the overlapping volcanic edifices, reinforced by volcano spreading on weak substratum. We tested our thesis experimentally by mounting analogue sand piles onto a sand and viscous PDMS substratum. Gravitational spreading of this setup produced en-echelon fractures that clearly mimic the dike orientations observed, with a deviation of 10°–32° between the model’s ridge axis and that of the main fracture swarm. Using simple numerical models of segmented dike intrusion we found systematic changes of displacement vectors with depth and also with distance to the rift zone resulting in a complex displacement field. We propose that at depth beneath the Desertas Islands, magmas ascended along the ridge to produce the overall present-day morphology. Above the oceanic basement, gravitational stress and volcano spreading adjusted the principal stress axes’ orientations causing counterclockwise dike rotation of up to 40°. This effect limits the possible extent of lateral dike propagation at shallow levels and may have strong control on rift evolution and flank stability. The results highlight the importance of gravitational stress as a major, if not dominant factor in the evolution of volcanic rift zones.Editorial responsibility: M Carroll     

Evidence for magma mixing in the Mariana arc system

Volcanoes of the Mariana arc system produce magmas that belong to several liquid lines of descent and that originated from several different primary magmas. Despite differences in parental magmas, phenocryst assemblages are very similar throughout the arc. The different liquid lines of descent are attributed to differences in degree of silica saturation of the primary liquids and in the processes of magmatic evolution (fractional crystallization vs magma mixing). Pseudoternary projections of volcanic rocks from several arc volcanoes are used to show differences between different magmatic suites. In most of the arc, parental liquids were Ol- and Hy-normative basalts that crystallized olivine, augite, and plagioclase (± iron-titanium oxide) and then plagioclase and two pyroxenes, apparently at low pressure. Eruptive rocks follow subparallel liquid lines of descent on element–element diagrams and on pseudoternary projections. Magmas at North Hiyoshi are Ne-normative and have a liquid line of descent along the thermal divide due to precipitation of olivine, augite, and plagioclase. Derived liquids are large ion lithophile element (LILE)-rich. Magmas at other Hiyoshi seamounts included an alkaline component but had more complex evolution. Those at Central Hiyoshi formed by a process dominated by mixing alkaline and subalkaline magmas, whereas those at other Hiyoshi seamounts evolved by combined magma mixing and fractional crystallization. Influence of the alkaline component wanes as one goes south from North Hiyoshi. Alkaline and subalkaline magmas were also mixed to produce magmas erupted at the Kasuga seamounts that are behind the arc front. The alkaline magmas at both Hiyoshi and Kasuga seamounts had different sources from those of the subalkaline magmas at those sites as indicated by trace element ratios and by _Nd.     

Temporal change in chemistry of magma source under Central Kyushu,Southwest Japan: progressive contamination of mantle wedge

Volcanism related to subduction of the Philippine Sea (PHS) plate began in Central Kyushu at 5 Ma, after a pause of igneous activity lasting about 10 m.y. It formed a large volcano-tectonic depression, the Hohi volcanic zone (HVZ), and has continued to the present at a decreasing eruption rate. The products are largely andesite and dacite, which became enriched in K with time. The proportion of tholeiitic to calc alkalic rocks also increases with time. Calc-alkalic high-Mg basaltic andesites (YbBs) were erupted in the early stage of the HVZ activity (5–3 Ma), and high-alumina basalts (KjBs) were erupted in the later stage (2–0 Ma). In contrast to the basalts in the HVZ, Northwest Kyushu basalts (NWKBs) have been erupted on the backarc side of the HVZ since 11 Ma, and hence are not related to the PHS plate subduction. They are mainly high-alkali tholeiitic to alkali basalt that shows no notable chemical change with time. NWKB, YbB, and KjB have MORB-normalized incompatible-element spectra that differ from each other, as is well expressed in both Nb and Sr anomalies. The patterns of KjB and NWKB are typical of those for island-arc basalt (IAB) and ocean-island basalt (OIB), respectively. YbB shows a pattern intermediate between the two. We suggest that the magma source beneath the HVZ changed in composition from an OIB-type mantle to an IAB-type mantle as the subduction of PHS plate advanced. However, the magma source remained fertile under Northwest Kyushu. In order to explain the temporal change of source mantle beneath the HVZ, we propose a model for progressive contamination of the mantle wedge, in which three processes (contamination by a slab-derived component, subtraction of magma from the mantle, and mixing of the mantle residue and slab-derived component) are repeated as subduction continues. As long as the progressive contamination of mantle wedge proceeds, its trace-element composition converges at a steady-state value for a short period. This value does not depend on the initial composition of the mantle wedge but instead on the composition of the slab-derived component. The trace-element composition of the magma produced in such a mantle wedge approaches that of the slab-derived component with time, but the major-element composition is determined by the phase relations of mantle peridotite. The slab-derived component may be basaltic liquid that is partially melted from rutile-bearing eclogite.