Unzen Volcano (Mt Fugen / Fugendake)

The 1990-1995 eruption of Unzen volcano is infamous for the 43 fatalities caused by the unexpectedly large pyroclastic flow on June 3, 1991. The famous volcano filmers and photographers Maurice and Katja Krafft, and geologist Harry Glicken were among those lost during this tragic event, together with multiple journalists who had joined them in a zone into which entry was not prohibited but was recommended against by the authorities. The eruption nevertheless took a small human toll in comparison to the failure of the east flank of the Mayuyama dome following an earthquake in 1792, possibly associated with the minor eruption that had occurred in the previous months. The resulting landslide reached as far as the coast, burying half of Shimabara and further causing a devastating tsunami which ran across Ariake Bay and hit coastal communities in the Kumamoto area. In total, about 15000 fatalities were recorded.

Mount Unzen, Japan Heisei Shinzan Lava Dome, Mt Fugen, Unzen

Unzen viewed from Kita-Kamikoba

Closer view, revealing dome structure

Unzen volcano is a terminology used to refer to what is effectively a complex of several volcanic dome structures with a total volume of about 35 cubic km. These form the major geological features of the Shimabara peninsula in southern Japan. The volcanoes development can be traced back to about 500,000 years ago with four main dome structures having arisen in the last 150,000 years; Nodake, Myokendake, Fugendake (i.e. Mt Fugen) and Mayuyama. Whilst Mayuyama is the youngest of these structures (4000 years old), Fugendake has been the focus of most activity in the last 20,000 years, including the minor largely effusive eruptions of 1663 and 1792, but also the 1990-1995 eruption on which this text will focus. Activity at the site of Fugendake started 25,000 years ago, rapidly building a large dome structure. A major collapse event occurred approx. 20,000 years ago forming the Myoken caldera which still encloses the present structure on its northern and western sides. The southern and eastern flanks are basically open towards the sea, yet the remains of the Mayuyama dome shield the center of the large coastal city of Shimabara. The geology of the complex suggests that activity has largely been effusive as their are no major pumiceous deposits which would be indicative of large explosive eruptions. The complex largely consists of lava domes and their degradation products.

Sketch map of Fugendake part of Unzen volcanic complex. DOME indicates location of Heisei Shinzan dome (1483m) emplaced during 1990-95 eruption. (1) Former summit of Fugendake (1359m), (2) Myoken Caldera rim, (3) Kita-Kamikoba district, (4) Ohnokoba district, (5) Oshigadani valley, (6) Mizunashi river, (7) Akamatsudani valley, (8) Taruki height, (9) Senbongi district, (10) Amphitheatre of 1792 collapse. Note: Shimabara city was protected during the 1990-95 eruption by Mt. Mayuyama, source of the 1792 collapse.

The 1990-1995 eruption has been summarized in detail by Nakada et al. 1999 (J. Volc. Geotherm. Res. 89, p.1-22), and the following text relies strongly on this reference. In November 1989, a swarm of volcano-tectonic earthquakes was registered under the center of Tachibana Bay north of the Unzen dome complex. The swarm was short-lived and followed by several months without any notable tremors. However, renewed increased seismicity could be measured as from April 1990 with the source now located under the western flank of the volcano. A sharp increase in seismic activity was detected in July. This was centered between 500 and 1500m below the summit of Mt. Fugen and included an M4.8 event. Phreatic eruptions started from the Jigokuato and Kujukushima craters on Nov. 17 1990 yet were relatively minor. On 12. Feb. 1991, the Byobuiwa crater complex formed and activity was recorded regularly from all three crater sites. By April, a clear switch from phreatic to more violent phreatomagmatic activity was observed as juvenile material started to be emitted during eruptions. This was attributed to rising magma coming into contact with pockets of water in the volcanic edifice. So-called Cock's Tail Jets (formed by ballistics trailing ash in their wake) could be observed up to a height of about 500m above the craters and the Jigokuato crater was enlarged somewhat by this activity. This activity ceased abruptly on 12 May and in the following days shallow high-frequency tremor was recorded with the hypocenters moving upwards from 10-50m / day, indicating that magma was rapidly approaching the surface. On May 20, a first dacitic extrusion was observed in Jigojuato crater, marking the onset of the effusive phase of the eruption.

Effusive activity was recorded almost continuously from May 20 1991 to mid-February 1995. Effusion rates were highest in 1991 and essentially gradually declined thereafter, although the trend was interupted by a second pulse of slightly increased effusion in 1993. Effusion was often accompanied by significant endogenous dome growth, which was proportionally more significant when extrusion rates were low. This growth at times led to significant deformations of the dome, sometimes raising fears that it could destabilize the entire summit structure. Dome growth was correlated to continued deflation of the western flank. This is consistent with seismic data in localizing the source of the magma to a region at a depth of around 13km below the western shore of the Shimabara Peninsula. From here, the magma rose at an angle of about 45' towards the surface.

During the eruption, a series of 13 distinct extrusion lobes was formed. These typically were about 300-400m long, 200-300m wide and 50-100m thick. Once lobes reached near-maximal size, it appears that they could no longer by mobilized by the pressure from lava effusing at the crater and this would then be forced to take a different path. This often simply resulted in a new lobe riding over the preceding one, but also several different sites of effusion over and to the east of Jigokuato crater were formed during the course of the eruption. Lobes tended to advance at up to 10m/day, yet their upper surfaces could be detected moving at up to 50m/day. The tendency for the upper surfaces to advance faster often resulted in overhangs which were extremely prone to failure, triggering numerous pyroclastic flows. These flows, and larger ones resulting from failure of significant parts of the dome complex were the most notable features of the eruption and will be focussed on seperately below.

By the end of the eruption, a dome with dimensions of 1.2x0.8km and a thickness of 230-540m had been emplaced. Its volume approaches 0.1 cubic km and a slightly greater amount of deposits attributable to the eruption (pyroclastic and debris flow deposits) was spread around its flanks. In total, about 0.21 cubic km of magma was erupted. The lava dome consists of plagioclase-phyric dacite with silicate content of between 64.5 and 66%, which was erupted at temperatures of 780-870 'C. Effusion rates of about 7 cubic meters per second were achieved in 1991, with a second minor peak of about 2.5 in 1993. Sulphur dioxide emissions also peaked near the beginning of the eruption at a level of about 250 tonnes / day. The summit of the lava dome resulting from the eruption is 1483 meters high and referred to as Heisei Shinzan. It rests to the east of the former summit of Mt Fugen (1359m).

Nearly 10000 Pyroclastic Flow events were detected during the course of the eruption. All but one of these were block and ash flows resulting from collapse events at the lava dome. Only the 8 June vulcanian explosion resulted in a pumiceous pyroclastic flow which was particularly mobile, surpassing all previous flows in length and having a more extensive surge component as evidenced by the greater zone of singed vegetation around the main flow body. This flow was triggered by preceding major collapse events which removed significant amounts of the "cap" over the pressurized conduit, allowing a sudded release of pressure. Several other vulcanian explosions occurred in summer 1991, but none were accompanied by pyroclastic flows. The 11 June explosion was the most powerful and resulted in pumice up to nearly half a meter in diameter falling up to 5km to the NE of the crater in the populated Senbongi district, resulting in breakage of numerous car windscreens. About 20000 tonnes of volcanic bombs are thought to have been ejected by this explosion alone.

Pyroclastic flows initially proceeded to the east along the Mizunashi River. Run-out distances gradually increased, reaching 3km on 26 May 1991, 3.2km on 3 June (accompanied by a fatal 4km long surge - see details below), and 5.5km on 8 June (the longest runout of the whole eruption). Dome growth temporarily shifted more to the NE flank in the following months. Consequently, PFs started to flow down towards Taruki Height before turning right into the narrow Oshigadani River gorge which feeds into the Mizunashi River about 3km from the dome. On September 15, 1991, a large collapse event led to an extensive flow along this drainage, the surge cloud from which crossed the path of Mizunashi River and impacted infrastructure including the Onokoba Elementary School. This was the most voluminous flow of the whole eruption. A further notable flow occurred on June 23, 1993. This was one of several flows that entered the Senbongi district after crossing Taruki Height NE of the volcano, and was responsible for one fatality (a local viewing his house that had been burnt during a previous flow). Senbongi had become exposed to PFs since previous activity had flattened the topography somewhat by filling Oshigadani valley with deposits.

Sketch map showing cumulative extent of pyroclastic flow (PF) and Lahar (Debris Flow) deposits from 1990-95 eruption.

The impact of the eruption included the burning of 800 buildings by PFs and associated surges, and the destruction of 1700 houses by debris flows / lahars. A total of 44 fatalities were recorded and over 11,000 people were displaced at the height of the eruption when the restricted areas were widened as an immediate response to the fatal 3rd June 1991 event.

Fatal Pyroclastic Flow of June 03, 1991

Sketch map showing approximate extent of fatal pyroclastic flow and surge on June 03, 1991. Stars indicate victim locations. Star labelled (1) is site at which bodies of Maurice & Katia Krafft were found.

Information board apparently showing ash clouds from fatal PF rising behind school. The school survived this flow.

Pyroclastic flows are amongst the most impressive and dangerous volcanological phenomema. The famous french volcano filmers, Maurice and Katia Krafft were increasingly drawn to their documentation, following previous experiences viewing small flows at Alaskan volcano Mt. Augustine. The Unzen eruption appeared to provide a perfect opportunity for viewing pyroclastic flows. On June 3rd, 1991, the Kraffts, accompanied by US Geologist and 40 Japanese, largely members of the press, entered the evacuated zone around the volcano to take up position in a slightly elevated area (about 40m above the river bed) in Teiten village, Kita-Kamikoba district. The position basically was in the direct line of pyroclastic flows proceding down the Mizunashi River, which however took a slight turn to the right before Kita-Kamikobe. Previous flows had reached a maximum run-out distance of about 3km a week earlier, but had yet to reach as far as the Kraffts position.

In the afternoon (at 16:08 local time), the eastern half of the lava dome suddenly collapsed, together with some older underlying material. About 0.5 million cubic meters of material was mobilized and rushed down Mizunashi drainage in the form a massive PF, far surpassing all prior events during the eruption. The PF body reached as far as 3.2 km, but the highly energetic ash surges accompanying it swept up into Kita-Kamikobe, reaching a maximum run-out of 4km by about 16:10. The flows were so powerful that cars of some of the journalists were swept away, one being moved as far as 80m. Trees were flattened in the area as were most of the houses. All people in the area were killed by the impact of the surge and/or by the intense heat. Many of the bodies, including those of Katia and Maurice (which were brought to the Anyoji Shrine in Shimabara, dedicated to the victims of the 1792 catastrophe) were recovered 2 days later by military personnel. The carbonized bodies of several victims were found in a taxi in which they had evidently sought refuge. Post-mortem examinations of the victims showed that clothes and hair had been burned off, hands were often near the face as if trying to protect from the hot ashes. Mucous membranes and gums were severely decomposed due to the heat and mouths were full of ash and other small particles. Asphyxiation and thermal distress could thus be determined as the cause of death in most of the bodies actually recovered. The ash surges may have been particularly intense as a result of the flow cascading over several waterfalls as it proceeded downwards, this increasing fragmentation and production of the fluidization zone overlying the main flow body. A fatal event at Merapi in 1994 has similarly been attributed to such a phenomena.

Many will know the famous video sequence reportedly showing the fatal pyroclastic flow. However, whilst the sequence beautifully illustrates the phenomena of pyroclastic flows, showing a break-off from an extrusion lobe, progression down a valley and the huge flow front coming into view and finally to a halt as a fire-engine drives away, it is probably only in part a true documentation of the main 3. June flow. This point has been raised before (see e.g. youtube comments), so i thought it should be addressed. Firstly, the dome was not visible from Kita-Kamikobe at the time due to poor weather conditions (however, being over 1300m high it may arguably have been above the clouds - clouds are visible at the bottom of the frame). Secondly, more crucially, the break-off shown is minor and clearly does not correspond to a failure of the whole eastern half of the dome complex, neither does it leave a significant collapse scar as was visible after the 3. June event. The second scene showing the flow progressing down the valley is presumably also from a smaller flow (possibly same as first scene) since with the low cloud at least covering the middle of the flank, and extent of overlying ash cloud over the June 3 PF, it is inconceivable to fly over it in the manner shown (and no helicopter is visible above flow in 3rd scene where the PF emerges from cloud cover). The 3rd scene, which is the most dramatic as it shows the huge extent of the flow, is however possibly correctly attributed. The flow is clearly highly energetic, fast-flowing and cascading over previously unaffected landscape (green vegetation), further its orientation appears to be correct compared to scientific documentation of the location of deposits. The argument that the village is not reached is incorrect since according to the detailed map showing houses and positions of victims bodies in the Encyclopedia of Volcanoes (Chapter "Hazards from Pyroclastic Flows and Surges" (Nakada), Fig.6) there were numerous houses both in the area affected and just beyond the affected zone (as can be seen in the picture). The camera that took the footage in the 3rd scene is not that of the Kraffts. I am unable to exactly correlate the topography seen to map of the area due to the cloud cover and changes in the landscape occurring as a result of the eruption as a whole. Neither can i deduce the exact camera position, although a fair guess is that it was from a road running parallel to the Mizunashi River about 50m to its north at about 150m asl.

A further issue that may be raised is whether the risk taken by the Kraffts was irresponsible. According to a report that was presented to the French National Assembly (Rapport No.1540-08; Available online), the area entered by the Kraffts was in the "Recommended" (meaning not recommended) area, but not in the "Restricted" area. Nevertheless, the Kraffts and Harry Glicken would have known that the extent of pyroclastic flows was increasing almost weekly and thus rapidly approaching the area they chose to visit. Further, they would have known that an elevation of 40m would provide little protection from an energetic surge. Indeed, Maurice Krafft is reported to have considered Unzen the most dangerous volcano he had ever visited. Nevertheless, the immense contribution the Kraffts made, especially in the area of hazard management, would probably not have been possible without taking such risks to obtain to date unsurpassed eruption footage. Using previously obtained footage, the Kraffts were apparently instrumental in persuading the Philippine authorities to evacuate the area around Pinatubo before its massive eruption and thereby may have saved thousands of lives. Hence, in the case of the Kraffts the risk can probably be justified by the potential benefits.

Major Pyroclastic Flow of September 15, 1991

The pyroclastic flows of Sept. 15, 1991, included the longest flow passing through the Oshigadani drainage during the course of the eruption and involved mobilization of about 2.4 million cubic meters of dome material. The events on this day have been analysed in detail (Fujii and Nakada 1999. J. Volc. Geotherm. Res. 89, p.159-172). The main flow initially descended the NE flank and was then deflected by Taruki height, directing it in a SE-orientation through a narrow gorge, after which it passed the already ravaged Kita-Kamikoba area and flowed into the lower reaches of Mizunashi river which channelled the flow eastwards towards the coast. As the flow turned at the Mizunashi river, a large surge detached and continued in a SE direction for another about 600m. This surge set light to about 150 houses in the Ohnokoba district as well as the local elementary school which remains as an outdoor exhibit today. The destructive nature of the flow and the large volume of the associated surge, especially as it exited the Oshigadani gorge can possibly be attributed to increased collision of materials in the flow body as the flow passed through the gorge, resulting in generation of more fine materials.

Sketch map showing extent of 15 September 1991 Pyroclastic flow and surge. Location of burnt Elementary School is indicated.

Progressive failure of the NE flank of the lava dome started at 16:44 local, with further major events following at 17:59 and 18:45, before the climactic flow began at 18:54. The first flow stopped just north of Kita-Kamikoba, whilst the next two preceding the main flow were slightly shorter. The 18:54 event was far more energetic and surge clouds flanking the main flow body flattened vegetation either sides of the flow and even remobilized trees already toppled by the fatal June 3 event in the Kita-Kamikoba area, also transporting the remains of a car about 120m. As the surge subsequently detached from the flow it rapidly lost energy but remained dense and extremely hot. At the school, vinyl pipes of the north side (facing the surge) were melted, but most windows withstood the de-energized surge. Interestingly, about 20cm of ash was deposited in the playground and on this side more windows were broken. Hot ash appears to have entered the building and set the classrooms alight on the playground side. The deposits of the flow have been studies in detail and generally contain 3 units: a 20cm layer of well-sorted ash, an up to 2m thick block and ash layer rich in fresh, slightly vesiculated dacite lava, and a few cm thick layer of ash which would have settled on the flow after its passage. The bottom layer of well-sorted ash is attributed to the surge running ahead of the main flow body and being overrun by the flow immediately after deposition. This correlates to video-footage showing highly mobile lobes (of the surge) "jetting ahead" at the base of the main flow body.

Ohnokoba Elementary School burnt by Pyroclastic surge Unzen Volcano Ohnokoba Elementary School burnt by Pyroclastic flows Unzen Volcano Burnt classroom, Ohnokoba Elementary School, Unzen Volcano

Burnt shell of Ohnokoba Elementary School

School and burnt playground feature

Remains of burnt out science (?) classroom

Ohnokoba Elementary School, burnt by pyroclastic flows Ohnokoba Elementary School, melted vinyl drainpipe Building burnt by pyroclastic flows, Unzen volcano

Burnt entrance of Ohnokoba Elementary School

Vinyl drainpipe melted by surge and following fires

Frame of building burnt by surges in Kita-Kamikoba area

Oshiga Valley, Oshigadani, Fugendake, Unzen Oshiga Valley, Oshigadani, Mt Fugen, Unzen

Oshigadani valley and its gorge which connects it to Mizunashi River (bottom left)

Looking westwards over Oshigadani valley

Heisei Shinzan lava dome extrusion lobes, Mount Unzen Dome fragments in Oshigadani valley

Remains of two large extrusion lobes

Boulders on dome talus slope in Oshigadani valley

- - - - -

Heisei Shinzan Lava Dome, Mt Fugen, Unzen Heisei Shinzan Lava Dome, Mt Fugen, Unzen Heisei Shinzan Lava Dome, Mt Fugen, Unzen

Mt Fugen / Heisei-Shinzan Peak

Heisei Shinzan Lava Dome, Mt Fugen, Unzen Heisei Shinzan Lava Dome, Mt Fugen, Unzen Heisei Shinzan Lava Dome, Mt Fugen, Unzen

Mt Fugen Heizei-Shinzan dome complex

As a result of the 1990-95 eruption, huge amounts of easily mobilizable material remained and indeed still remains on the lower flanks of the volcano. To prevent lahars which can be triggered by heavy rainfall, a large number of Sabo (erosion control) structures are in lace on the flanks of the volcano. Many already existed before the eruption and extensive work on the structures was carried out afterwards. A more detailed discussion of Sabo structures can be found in the section on Sakurajima volcano. In addition to Sabo structures, a program was implemented to increase vegetation on the flanks in order to reduce mobilization of deposits. Between 1994 and 1997, helicopters were used to reseed the slopes of the volcano that had been stripped of vegetation during the course of the eruption. Further revegetation measures followed.

Sabo channels, Unzen Sabo dams, stepped check dams, Mizunashi River, Unzen

Sabo channels for erosion control

Series of Sabo dams, Mizunashi River

Sabo dams, stepped check dams, Unzen Sabo dams, stepped check dams, Mizunashi River,  Unzen

Sabo dams with school and observatory behind

Series of Sabo dams, Mizunashi River

Sabo dams, slit dam,  Mizunashi River, stepped check dams, Unzen Sabo dams, stepped check dams, Oshiga gorge, Unzen

Series of Sabo dams (slit dam at left side), Mizunashi River

Series of Sabo dams (stepped check dams) in Oshigadani River gorge

Readers interested in lava domes, pyroclastic flows, and their impact on humans and infrastructure will probably find the section on Soufriere Hills volcano interesting. Those wishing to visit Unzen may be interested to know that in a cable-car leads up to near the summit of Mt Fugen, and trails lead to the summit of Mt Fugen, although climbing on the dome itself is prohibited. During my visit in January, we encountered freezing temperatures and light snowfall on the higher slopes of the volcano and the cable-car was closed. The burnt elementary school and adjoining information center can be visited and there is the Mount Unzen Disaster Museum near town as well as a shrine dedicated to the victims of the 1792 event in Shimabara. There is also a small information center on Taruki Height which can be reached by road.

Photovolcanica Full Index