One of the most dramatic events following the devastating eruptions of Mount Pelee in Martinique was the growth of a towering spine on the summit lava dome. It began to rise above the dome on November 3, By May 31, , the spine reached a height of m above the dome, temporarily creating a m-high peak at the summit.
It then slowly disintegrated and was gone by the end of the eruption. This March 11, , photo shows the spine near its peak height, with a smoothly extruded eastern side. Photo by A. Lacroix, from the collection of Maurice and Katia Krafft. Slow, piston-like extrusion of a solidified portion of a lava dome sometimes produces vertical lava spines that rise above the surface of the dome.
The world's largest known spine rose to a maximum height of m, more than twice that of the Washington Monument, at Mount Pelee on Martinique in Growth of the spine began in November and reached its maximum on May 31, It slowly disintegrated and was gone by the end of the eruption two years later. This photo was taken on March 15, A pyroclastic flow, similiar to the one that destroyed the city of St. Pierre on Martinique on May 8, , sweeps down the flanks of Mount Pelee volcano on December 16, A towering column of ash and steam rises above the advancing pyroclastic flow, which was formed by collapse of gas-rich rocks on a growing lava dome in the summit crater.
The May 8 pyroclastic flow was substantially larger than this one, and would have covered an area wider than the entire shoreline of this photo. The devastated city of St. Pierre lies in ruins after a catastrophic eruption on 8 May , in which pyroclastic flows and surges swept over the city, killing 28, people. The high-temperature pyroclastic surges devastated a 58 sq km area SW of the volcano and swept out to sea, capsizing all but two ships in the harbor.
This March photo from the south shows Mount Pelee towering over the remnants of the city, capped by a dramatic lava spine that grew above the summit lava dome.
A pyroclastic flow produced by the collapse of a growing lava dome in the summit crater, sweeps down the SW flank of Mount Pelee on 1 January Small pyroclastic flows continued for 16 months following the catastrophic eruption of 8 May , that destroyed St. A second large pyroclastic flow, comparable in size to that of 8 May, devastated the SE flank on 30 August, killing an additional people.
The frequency of pyroclastic flows diminished after September , but they continued at longer intervals until October Photo by Hayot, from the collection of Maurice and Katia Krafft.
Pierre on the coast at the left , which the volcano destroyed during a catastrophic eruption in The modern volcano was constructed on the rim of a large SW-facing horseshoe-shaped caldera whose northern wall is the ridge in the shadow on the left horizon.
Photo by Richard Fiske, Smithsonian Institution. This procedure, sometimes referred to as "dry tilt," detects deformation of the volcano that often precedes an eruption by measuring the precise differences in elevation between two stadia rods placed on fixed points.
This technique is part of monitoring efforts by the observatory to help detect future eruptions of this scenic, but deadly volcano.
Photo by Lee Siebert, Smithsonian Institution. This photo portrays an unusual combination of geology and history. The light-colored deposits in this outcrop south of St.
Pierre are pyroclastic-flow deposits similar to those of eruptions that destroyed the city in The abundant large holes in the outcrop are not a volcanological phenomenon, but were produced by cannon balls blasted into the unconsolidated deposit during British-French wars for control of the island of Martinique. The lava dome at the left, seen from the east rim of the crater, was formed during an eruption that began in The vegetated knob halfway down the right skyline is a lava dome from the eruption.
The eruption was similiar to that of , but smaller in scale. After explosive removal of part of the dome, growth of a new dome began in January Pyroclastic flows accompanied dome growth until the end of Photo by William Melson, Smithsonian Institution.
Pierre in northern Martinique a century after the catastrophic eruption that destroyed the city in Lava domes formed during the eruption and one in form the present summit, which was constructed within a large scarp visible on the lower left horizon that formed when the volcano collapsed about years ago.
This prison cell in the city of St. Devastating pyroclastic flows and surges swept down the SW flank of the volcano early in the morning and destroyed the city, killing 28, people in the world's deadliest eruption of the 20th century.
Photo by Paul Kimberly, Smithsonian Institution. The area beyond the grassy knoll was part of the first portion of the ancestral volcano that underwent massive edifice collapse more than , years ago. This massive collapse produced a 25 cu km debris avalanche that swept into the Caribbean Sea up to 70 km from the coastline.
The steep-sided grassy knob in the right foreground is part of the Aileron lava dome, which formed during an eruption about years ago. This view looks to the SE towards the town of Morne Rouge left-center , which was devastated by pyroclastic flows during the eruption. The Pleistocene Piton du Carbet volcano lies in the clouds on the right-center horizon.
The steep-sided lava dome at the left is Aileron, which was formed about years ago. The eastern rim of l'Etang Sec, the current summit crater, cuts horizontally across the photo at the upper right in front of the dome on the right horizon. The and lava domes fill much of this crater.
The town with reddish roofs at the far right is Morne Rouge, affected by pyroclastic flows from the eruption. The modern volcano was constructed within a scarp produced by collapse of the volcano about years ago. The irregularity on the right-hand flank is part of the eastern summit crater rim and the Aileron lava dome, which erupted about years ago. The lava dome fills much of the l'Etang Sec summit crater, as seen here from Morne Macouba, north of the summit.
What exactly happened on May 8 is unclear, but one thing is certain: an immense blast of hot ash and gas exploded from the volcano obliterating the city of St. Pierre burned for days, and when rescuers were finally able to enter the city, they found little left.
Pierre jail ruins. Mount Pelee continued to erupt throughout the summer of , including an eruption on August 30 that destroyed the village of Morne Rouge. Ominously, a black obelisk-shaped lava dome began to rise rapidly above the caldera.
During one 8-day period, it grew thirty feet and during another 4-day period, it grew another 20 feet. The tower eventually rose over 1, feet above the crater and glowing lava could sometimes be seen through cracks in the rock.
The tower finally crumbled in but not before it was thoroughly documented by Angelo Heilprin, a Hugarian-born American geologist. The May 8, eruption of Mount Pelee is believed to be the deadliest in the 20th century and the 3rd deadliest in recorded history, behind the eruption of Tambora and the explosion of Krakatoa.
The choice to focus primarily on the ash-cloud surge was motivated by the fact that: 1 its extent and limits, as extracted from the field, are based on robust evidence and therefore only contain small uncertainties, and 2 it covered a much larger area than the block-and-ash flow, restricted to valleys.
However, the maximum runout in the south of St Pierre is underestimated where the simulated ash-cloud surge traveled m less that the real flow. Also, a large part of the area inundated by the ash-cloud surge around the northern part of the crater is not reproduced by the simulations.
Since the location of the initial mass flux in the simulations was set to be in the southern crater outlet, the simulated ash-cloud surge derived from the block-and-ash flow in the proximal area was unable to spread northward and inundate that part of the crater. Figure 7. Results of the best fit simulation obtained using the input parameters presented in Table 2.
A Final distribution of the simulated deposits from the best-fit simulation, which include the extent of the simulated ash-cloud surge green color scale for the thickness and the simulated block-and-ash flow pink color scale. For ease of comparison, outlines of the observed ash-cloud surge and block-and-ash flow, as extracted from the field, have been added with a white and black outline, respectively.
B—E Sequence of four snap shots of the best-fit simulation at 30, , , and s after the mass starts to overflow through the crater outlet, showing the propagation of the flows overlain on the DEM. Comparisons between the simulated surge deposit thicknesses with those measured at 20 locations in the field Bourdier et al.
Surge velocities are either in good agreement or underestimated by a factor of 3, depending on the reference value taken Table 3 but are within the typical range of 40—90 m s —1 for ash-cloud surge estimations elsewhere Calder et al. Indeed, the simulated flow inundated an area similar to the real flow with a model precision coefficient of The simulated block-and-ash flow travels at an average velocity of 19 m s —1 , relatively common for this type of flow, as described elsewhere Calder et al.
Figure 8 presents the maximum dynamic pressure and the mean direction of the ash-cloud surge extracted from the best-fit simulation. In VolcFlow, the dynamic pressure P dyn is calculated following Valentine :.
Figure 8. Map of the maximum dynamic pressure and a mean direction of the ash-cloud surge from the best fit simulation. The distribution of the dynamic pressure is shown as a color scale, and isobar lines indicate the 1,2 and 3 kPa blue, green and yellow lines, respectively pressure fields.
The average direction of the current is represented by white arrows, whose lengths correspond to the velocity of the ash-cloud surge, calculated from the center of the arrow.
Values gradually decrease from more than 5 kPa toward the block-and-ash flow to a few Pa only toward the edges. This pattern was also observed at Merapi volcano by Jenkins et al. The mean direction of the simulated surge is radially dispersed around the block-and-ash flow and perfectly matches the direction measured in the field by Fisher et al.
However, the model does not match the backward direction measured by Fisher et al. Toward the east, especially in the St Pierre area, flow directions slowly change from south to southeast as the simulated surge expanded eastward.
The same observation can be made on the western side of the area inundated by the surge with a flow direction that changes from southwest to west.
The passage of the simulated ash-cloud surge over the flat sea surface promotes its lateral spreading as it covers a larger area to the west of St.
To better investigate the behavior of the simulated ash-cloud surge toward St Pierre, Figure 9A shows a snapshot of the simulated flow dynamics in this area over the DEM while Figure 9B superimposes these simulation results and field observations over the topographic map of St Pierre in from Lacroix The external, low dynamic pressure zone of the simulated surge with a maximum pressure of 1.
Thus, the dynamic pressure in St Pierre seems to be underestimated by the model compared to field observations. In this area, the surge direction changes from southeast to south as it propagates toward the southern part of the city Figures 7A,B , matching approximately the direction of the Victor Hugo street red line as observed by Lacroix However, the mean direction of the simulated surge does not match the surge direction measured in Fort Cemetery by Boudon and Lajoie In summary, after entering the sea at Fort district, the simulated surge is first deflected to the east toward St Pierre, and then further deflected to the south by the hills on the east of St Pierre south of Morne Abel, Figure 8.
The direction of the simulated surge seems to be highly variable when it passes through St Pierre due to high turbulence induced by the complex pattern of the city infrastructures. Figure 9. When the simulated block-and-ash flow entered the sea, it formed unrealistic thick and large lobes Figure 7. Focusing all the mass through the crater outlet as the primary source condition for our simulations, as previous workers have commonly hypothesized from field observations, results in good correlations with the real event.
The resulting self-regulated volume rate Figure 6 , generated by passive overflowing of the mass through the lowest elevated part of the crater rim, produces a realistic simulated pyroclastic current. The direction of the ash-cloud surge seems to corroborate quite well with the field direction measurements and the damage in St Pierre. However, simulations did not reproduce the up-valley movement of the surge at Fond Canonville, inferred by Fisher et al.
Charland and Lajoie questioned the reliability of the flow directions measured by Fisher et al. But beyond these conflicting measurements, the landward flow direction obtained by Fisher et al. Unfortunately, if such a process had occurred, our simulations did not capture it because VolcFlow does not model such flow temperature and energy variations. While the deposit extent and paleo-current directions are well reproduced by our simulations, the dynamic pressure seems to be underestimated.
Given the equation used here to calculate the dynamic pressure Eq. Simulated flow velocities seem to be accurate if we compare them with the field estimations from Fisher et al.
Underestimation of the dynamic pressure could also be explained by an underestimation of the surge density at the base of the flow. In fact, the shallow-water modeling approach used in VolcFlow implies the use of an averaged density across the entire current depth, which provides accurate reproduction of the general surge dynamics but constitutes an important simplification from natural density-stratified surges Valentine, Therefore, the actual density at the base of surges the part that interacts with buildings is much higher than a depth-averaged value.
This density difference could potentially explain the resulting underestimation of the dynamic pressures in our simulations. In order to reproduce the actual runout of the ash-cloud surge, the particle drag coefficient C d used in our simulations had to be set to an unrealistically high value see Table 2.
In fact, the chosen value of 35 does not match any previous estimation of this coefficient for volcanic particles 0. C d has been tuned in our model because it is the only parameter linked to the settling velocity Eq. With a smaller settling velocity, the simulated ash-cloud surge settles much slower, keeping particles in suspension for a longer time, and subsequently covers a larger area before becoming buoyant.
Therefore, some process seems to have hindered sedimentation in the May 8th pyroclastic current. A similar process has already been inferred for the simulation of the November 5th, pyroclastic current at Merapi by Kelfoun et al. Different hypotheses are proposed to explain the hindering of the sedimentation: i if the base of the May 8th, ash-cloud surge was relatively dense, as suggested by the high dynamic pressures obtained from field observations, particle settling in the density-stratified surge could have been reduced and particles transported further away i.
The factor of 30 obtained for the best-fit value of C d could be applied to the surge density instead, thus giving similar modeling results. Moreover, the resuspension of soft material i. Further model development is needed to include air entrainment in VolcFlow and to investigate whether this process has a significant influence on the dynamics of two-layer, depth-averaged simulated currents, as recently proposed by Shimizu et al. The model of Fisher et al. Instead, the two different layers of the simulated pyroclastic current i.
Indeed, the simulated surge spreads radially around the crater without following the southward spreading of the block-and-ash flow. Moreover, the shape of the simulated ash-cloud surge area differs from the pear-like shape characterizing our best-fit simulation as well as the May 8th, surge area Figure 8.
The progressive generation of an ash-cloud surge during the southward propagation of the main block-and-ash flow seems to be the more suitable process to explain both the shape of the inundated area and maximum runout of the surge toward St Pierre, as inferred by previous field investigations Fisher et al.
Thus, this also shows that simulating only one of the two conflicting scenarios to investigate the dynamic of the May 8th, pyroclastic current was satisfactory. Indeed, regarding our results, simulating the pyroclastic current as a blast flow appears to be unnecessary since a blast is exclusively formed in the crater, as the complementary simulation, and would probably have been unable to reproduce the pear-like shape of the surge deposit.
Figure Result of a complementary simulation with the ash-cloud surge and the block-and-ash flow supplied directly into the crater, without any surge production from the block-and-ash flow during the transport. The source conditions are adapted to supply A source with an initial vertical component, like an explosion, that collapses and spreads volcanic products radially around the crater seems to be required in order to inundate this area.
If an explosion had occurred on May 8th , it was most likely from the sudden decompression of a dome, resulting in a block-and-ash flow deposit similar to those described at Merapi Charbonnier et al. Thus, we hypothesize that the source of the May 8th, pyroclastic current was most likely an explosive dome collapse, during which the shape of the crater enhanced the lateral spreading of the pyroclastic current by redirecting the products of the explosion through the crater outlet to the south.
The remaining portion of the vertical jet formed during the explosion collapsed, providing enough lateral momentum for a dilute current to form and overflow the northern rim of the crater, inundating a small area downslope before rapidly stopping and becoming buoyant. The occurrence of a blast during the May 8th, eruption has been strongly debated Fisher and Heiken, , ; Sparks, ; Boudon et al.
The idea of a blast like Mount St Helens was first rejected by Fisher and Heiken , , mainly from field evidence i. When using the more recent blast description of Belousov et al. Paradoxically, discarding the hypothesis of a violent decompressive phase as a source condition for the May 8th, pyroclastic current is limiting.
In fact, the May 8th, pyroclastic current exhibits remarkable features that can be attributed to a decompressive blast-like event as described nowadays:. The lower unit in the May 8th deposit is composed of relatively coarse particles, which require a high level of turbulence to be sustained in suspension for a long time Lajoie et al. Moreover, even if the deposit stratigraphy does not match the blast stratigraphy of Belousov et al.
In summary, a few arguments support the fact that the initiation of the May 8th, pyroclastic current could have been partially driven by a decompressive blast-like phase, however, most of the characteristics of the current and its deposits are accurately reproduced by an ash-cloud surge generated from the block-and-ash flow, for which the surge production increases gradually with the spreading of the concentrated flow.
Since our model is currently unable to simulate a laterally-directed blast and its associated initial burst phase , our results show instead that: i such a lateral explosion was not a compulsory component of the source conditions needed to correctly reproduce the characteristics of the May 8th pyroclastic current numerically, and ii the May 8th events can be modeled with relatively simple physics, commonly attributed to small-scale eruptions generating small-volume pyroclastic currents.
A sudden decompression of the lava dome with a small vertical component could have likely taken place in the first few seconds of the eruption, explaining the presence of surge deposits on the northern part of the upper flank of the volcano, as well as the coarser nature of the basal layer in the deposit sequence, likely responsible for the high dynamic pressure of the distal surge.
The May 8th, pyroclastic current may fit instead inside a spectrum between a blast i. The physical behavior of such a blast is likely not dissimilar to one of a simple ash-cloud surge. This idea is reinforced by the work of Esposti Ongaro et al. This corresponds, to our knowledge, to the first estimation of the total volume of the May 8th, pyroclastic current deposit.
Nevertheless, because of the simplified physics used in our model, this value should be considered with caution and rather corresponds only to a first-order estimation. Nevertheless, the relatively good correlations of the best-fit simulation with the observed deposit in terms of thickness distribution and repartition Figure 7 and Table 3 support the validity of the total deposit volume proposed here.
This is also supported by the componentry analysis by Bourdier et al. Again, this ratio must be taken with caution due to simplifications made in the model to simulate the sea surface. Nevertheless, measuring only the volume deposited on land to estimate the total volume of the eruption is too restrictive and underestimates the value. Pelee by Heilprin, May 26, Photograph of the remains of St. Pierre by Heilprin, Fisher, R.
Pelee, Martinique; May 8 and 20, , pyroclastic flows and surges: Journal of Volcanology and Geothermal Research, v. Pierre, Martinique by ash-cloud surges, May 8 and 20, Geology, v. Hirn, A. Perret, F.
0コメント