Note: For a blog post on Mount Agung, see http://big-volcanic.com/agung-volcano-nearing-an-eruption-here-is-what-to-expect-from-the-bali-volcano/
This is a repost, originally written in 2015. Small edits have been made, and some new (old) images added.
It is one of the most famous volcanic eruptions in history. The loudest explosions ever recorded, dust spreading worldwide affecting sunsets for years, and it occured in a densely populated region, with many passing ships with captains adept at making scientific observations. It wreaked devastation. In spite of this, it was not an exceptionally large eruption. As a VEI6, there were larger ones, even in the 19th century, and in the 20th century Pinatubo came close. But Krakatau is more famous than any of these eruptions.
The uninhabited island of Krakatau is (or rather, was) located in the Sunda Straits, between the large islands of Java and Sumatra. Even in 1883 this was a busy shipping lane, and the onset of the eruption was well observed. But the island was not seen as a threat. Apart from a fairly minor volcanic eruption in the 17th century, there was little sign of active volcanism. The location is interesting, between the major island of Java and Sumatra. The main volcanoes in Indonesia lie along a line traced by these islands. Krakatau is at the place where this line makes a sharp bend.
Krakatau island was roughly 15 km2 in size, and had three, comparatively small, volcanoes. The tallest one was Rataka, on the southern end. The two lower cones were Danan and Perwoebatan. The word ‘Rakata’ means ‘crab’ in the old Javanese language, and this became ‘Krakatau’ in Dutch or Krakatao’ in Portugese (in English ‘Krakatoa is often used). This name became attached to the entire island. Two other islands, Verlaten island (‘deserted island’) and Long Island formed an arc, with Danan at the centre. Rataka also fell on this arc.
The arc around Danan was the remnant of a crater or caldera, 14km across, with Danan the central volcano. A large eruption would have been needed to form such a caldera. There are ancient stories about a massive eruption around AD 416 causing the opening of the Sunda straits but these seem mythological. A large eruption has been suggested to have occured in AD 535, to explain the decadal cooling at this time, but again there is a complete lack of evidence. Ash deposits in the Sunda straits indicate that no eruption larger than the 1883 one has taken place for many thousands of years. The caldera shows that a large event happened at some time, but it has not been in historical times.
Smaller eruptions did happen on Krakatau. An eruption took place from May 1680 to November 1681, probably from Perwoebatan, although one report states that activity occured at four different locations on the island. The eruption destroyed the forest on Krakatau. However, the eruption was not a large one. Afterwards the island remained quiet for 200 years. The forest regrew and became almost impenetrable. One landing party reported a hot spring, a reminder of the fire below. A lack of fresh water meant that, although people visited and occasionally settled, it did not become permanently inhabited. By 1880, no-one lived there.
In 1880, Krakatau came back to life. Frequent and strong earthquakes began to occur, one of which severely damaged a light house on Java. The first eruption occured on Sunday, May 20, 1883. Explosions were heard in Jakarta (named Batavia at the time), 150 kilometers away. Ash began to fall the next day, as far away as 500 kilometers. Over the next few days, ships in the area reported a column of ash, 11km high. The eruption continued, but not as violent as the initial eruption. On May 27th a landing noted that the explosive activity came from a 50-meter wide cavity within Perwoebatan, located within a 1km crater. Older lava flows suggested this was the same location as the 1680-81 eruption. A photograph of the eruption was taken, coloured in by an eye witness. This is the most famous picture of Krakatau.
The eruption weakened over the next month, but on June 24th a second column was seen to rise from the centre of the island. By this time the summit of Perwoebatan had disappeared, presumed destroyed in the explosions. The two ash columns existed throughout July but the activity was not seen as exceptional — for Indonesia! A group from Batavia (Jakarta) even organised a picnic trip to Krakatoa to watch the eruption!
A landing party on Aug 11 found that large columns were now rising from three separate locations: Perwoebatan, Danan, and a site north of Danan, near the base of the cone. Lesser activity was seen at no fewer than 11 locations, all from the northern half of the original island, some only meters above sea level. This is the last detailed report prior to the great explosions. The island was now off-limits and not even the local geologist wanted to attempt landings. Over the next days ash fall was seen up to 80 kilometers away, with reports of loud explosions: the eruption probably intensified during this time.
The critical phase began on the afternoon of Sunday, Aug 26, leading to a terrible night and morning.
The big explosions
Up to Aug 25, the eruption of Krakatau was not a major one. The largest event had been the initial explosion, May 20th 1883. The long continuation, and the presence of multiple eruption sites, was similar to the 1680 event. Magma had traveled from a deep chamber (estimated at 30 kilometers deep) to a shallow reservoir around 9 km below the surface. This magma migration could have caused the earthquakes which began three years earlier. The pathway from the reservoir to the surface used a rift or dyke, with multiple exit sites. But now the eruption changed, and became unique.
Reports on the events of Aug 26 and 27 are patchy, and the full story was only pieced together over the next few years. People on the coast were too busy surviving, and ships in the Sunda Strait were in complete darkness. The main events are clear, though. This post is based heavily on the description in a report from the Royal Society, dated 1888 (see reading list below).
The eruption rapidly intensified during August 25 and 26. By 1pm on Sunday, Aug 26th, the explosions were heard in Batavia, 150km away. By 2pm, a ship 100km away reported a black mass rising to to a height of 25 kilometer. By this time explosions occured every 10 minutes. By 3pm the explosions were heard 250 km away and by 5pm, all over Java. The explosions were not accompanied by earthquakes. The nearest ship, 10 miles from Krakatau, reported that the explosions came from the northeast of the island. Between 5pm and 6pm, a rain of large pieces of warm pumice began to fall on the ship. On the shore, waves from the eruption began to arrive from 5:30pm. They were at first just feet high, but continued through the night.
Sleep became impossible that night in the main cities of Batavia and Buitenzorg. The explosions were loud and frequent, rattled windows and cracked buildings, in spite of the great distance from Krakatau. During the night, ships in the Sunda Strait battled a hot stormy wind, carrying a sulphurous smell. Cinders fell, and the sea was warm to the bottom, 50 meters down. From midnight to 4am, explosions were continuous and one second, the sky was intensely black, the next a blaze of light. The roar grew less only towards morning. By that time many places on the Sunda Strait coast had already been devastated, but even worse was to come. Around 5am, the first of the large explosions must have happened. But in the blackness and deafening roar, this did not seem to have been noticed.
In the morning, the ash column was seen to have reached a height of 30 km. As it spread, blackness followed and ships had to take anchor. The clouds reached Java and Sumatra around 10:30am. Batavia was in complete darkness from 11am to 3pm, while a rain of dust fell. The dust contained some 10% water. In the Strait, falling pumice the size of pumpkins was reported. This continued during the morning; later the dust became falling mud.
The darkness precluded observations of Krakatau during the 27th and the exact sequence of events there is not known. The explosions reached their peak during the morning, ending after 11am. After a quieter afternoon, there was another set of smaller explosions during the following night. Activity finally ceased at 2:30am on the 28th. The next day, the Sunda Strait was almost impassable due to the mass of floating pumice and other debris. And although Krakatau’s peak was still there, the rest of the island had gone.
Together, the confused reports indicate four distinct major eruptions. Air pressure measurements of the shock waves at Batavia provide their times. Large explosions occured at 5:07pm on the 26th, 5:30am, 6:44am, and 10:02am on the 27th, and a smaller one at 10:52 (all local time). The fourth was by far the largest and loudest. There may have further strong explosions at 4:30am and 9am. The 10:02 explosion has an estimated energy equivalent to 150 megatons TNT – comparable to a M8.6 earthquake, and three times larger than all the other explosions combined.
The dust and ash reached very high altitudes. The ash columns were reported at 30 km high, but that was before the 10am explosion. Estimates for that explosion give values up to 50 km.
The sound of the 10:02 explosion was reported as far away as Diego Garcia, 3600 km distant, and possibly Rodriguez at 5000 km(!), four hours later. They sounded like distant cannons and in many places were thought to come from ships in distress. The map below shows where the sound was heard.
It seems strange that the later explosions were not heard closer to Krakatau (such as Batavia), but were heard at very distant locations. Many people would have been stone deaf by now. But the ash in the atmosphere also dampened the sound, in the same way that falling snow brings silence. But sound could still travel over the top of the ash layers, at high altitude. Far away these waves curved back to the ground, causing the sound to be heard there. Over short distances sound weakens with the square of the distance, quickly rendering distant noise inaudible. But over much larger distances, the sound travels horizontally (the lack of atmosphere above 10km altitude stops it carrying much energy further up), and now the sound only decays linearly with distance. It weakens much slower, and this is why sound could be carried over such large distances.
The pressure wave from the explosion traveled much further, and circled the world several times. This was detected by barometers everywhere and lasted for five days. But this wave should not confused with the sound: it is not audible. The sound ‘only’ traveled 5000 km.
Pyroclastic flows were not known at the time. But some of the descriptions show that they happened. They may have come directly from the explosions, driving material laterally outwards, or from the collapse of the columns of ash. People received burn wounds on various ships in the Sunda Strait. At Ketimbang, Sumatra, 40km away, as many as 1000 people were killed by a ‘rain of hot ash’. One family survived inside a hut, and describe how the hot ash was pushed through the cracks and floorboards, and the air was sucked out. This was clearly a pyroclastic flow coming over the sea. Ships 80 km away experienced hot winds but escaped burning. Thus, the pyroclastic flows became survivable only between 40km and 80km from their origin.
How can pyroclastic flows travel such large distance over water, and still remain so hot? Pyroclastic flows have temperatures up to 600°C, are denser than air, but have density comparable to sea water. This allows them to travel along the sea surface and reach large distances. Much of the sub-sea deposits around Krakatau have been shown to come from pyroclastic flows. The pressure of the flows can also push up the sea – causing tsunamis. In this way, Tambora, exploding well in-land, caused a tsunami.
Waves from the eruption began to reach the town of Anjer, to the east of Krakatau, from 5:30pm on the 26th, at first only a few feet high. Around 7pm, houses on the coast were damaged both on Java and Sumatra, and a camp of labourers was washed away. The waves continued irregularly through the night, at first not at damaging heights. But at 1:30pm two towns were submerged, and Anjer was destroyed at 6:30am and again submerged at 7:30am. After 10am, a much larger wave completed the destruction and affected the entire coast along the Sunda Strait. However, there are few eye witness reports on this wave. By this time the Sunda Strait was in darkness. At Anjer, noone was left on shore to see the large wave. Few had survived the two earlier waves, and those survivors had fled to the hills. This must have been the case at many locations along the coast.
The 6:30 wave at Anjer was 10 meters high. The height of the later wave is not known. Damage to buildings along the coast was seen up to 15 meters above sea level, but at one location at the head of a bay, to 38 meters. A men-of-war ship, called Berouw, was carried 2.5 km inland, left 10 meters above sea level.
The earlier waves were local, causing destruction in one place but not in another not far away. The big wave traveled in all directions and caused devastation everywhere along Sunda Strait. But little damage was done outside of the Strait. At Batavia, the earlier waves were recorded by a tide gauge at 5-10 cm, but the big wave was a wall of water, and the height was 2.5 meters. This shows how much larger this event was. The large wave started as a rise. This indicates that the tsunami was not caused by a sub-sea collapse, but by water being pushed up.
The height of the wave rapidly diminished beyond the Sunda Strait. Still, the waves reached all over the Indian ocean, with a measurable wave even at Honolulu. Reported waves at Cardiff and New Zealand, attributed to Krakatau, probably were not related and had a local origin.
The cause of the tsunamis is still debated. Four explanations have been proposed: the collapse of the caldera, atmospheric shock waves from the explosions pushing up the water, pyroclastic flows from the explosions pushing the water out, and the fall-back of a cubic kilometre of ejecta into the sea. The second and third one seem most likely. They explain why the times of the tsunamis agree with the explosions (especially for the big explosion at 10am), and why the wave started with a rise. However, for the other waves the other explanations may also have contributed.
After the end
There was no volcanic activity seen after Aug. 28th. Reports of steam can be explained by the hot deposits, and an explosion on Oct. 10 may have been due to rock fall from the unstable slopes.
After the eruptions had ended, only the southern third of Krakatau, including half of Rakata, remained. Rataka now showed a vertical cliff to the north, from the peak of the mountain. The peak had increased in height by 40 meters, probably due to the blanket of ejecta. The rest of Krakatau had disappeared. Verlaten Island and Long island had grown considerably. At a distance of 15 kilometers, in all directions, the sea had become less deep by 3 to 15 meters. The channel to the north had become completely blocked, and some new islands had formed (which were quickly destroyed by the sea). But around the centre of the eruption, over several kilometers, the depth of the sea had increased. This is the 1883 caldera.
The floating pumice traveled across the Indian ocean, eventuallt reaching Australia and South Africa. There is an unconfirmed report that in Zanzibar, human skeletons were found on the pumice rafts. The dust at great height in the atmosphere spread across the world. Along the equator, the dust encircled the earth twice, each circumference taking 13 days. It caused strangely coloured Sun and Moon, and suppressed the rains over India. Deeply red sunsets were seen for many months. A small drop in temperature occured. Many descriptions claim that temperatures dropped by 1.2 degrees the next year, but published temperature data do not confirm this. The global effect was minus 0.1-0.3 degrees, lasting 2 years. This was similar to, or a bit less, than Pinatubo which caused a drop of 0.5°C
Over time, the sea claimed back the new islands leaving only the Rataka remnant. But eventually, in 1927, new volcanic activity was seen, from a location midway between Danan and Perwoebatan, at the edge of the 1883 caldera. This volcano, Anak Krakatau, has been above sea level since 1930 and is still growing with frequent eruptions. Krakatau is not finished with us yet.
Cause and effect
The Krakatau eruption was not extraordinary, and behaved very much like the the 1680 eruption, up to mid-August. But the enormous explosions set the event apart. Was this caused by interaction with sea water? If so, why not in 1680? Or did something else occur? The ejecta did contain some water, but the volcanic deposits do not show alteration by magma-water interaction. Sea water rushing into the magma chamber cannot have been the direct cause of the explosions.
A strong hint comes from seismological maps of the shallow magma chamber, 9km deep. They show a double chamber, separated along an east-west line, a few kilometers apart. The two chambers are on either side of the 1883 caldera. Shouldn’t there have been a chamber underneath Danan, the centre of the 1883 caldera? Perhaps there was. The best explanation for the events of August 1883 is that the roof over the centre of the 1500-meter deep magma chamber collapsed. The collapse pushed the remaining magma into the two magma pockets on either side, and upward. The collapse caused chaos near the surface.
The different explosions can have been caused by the collapse of different parts of the island. The 5:30am explosion may have been related to Perwoebatan collapsing, the later ones to Danan. The fact that mud falls were not reported until after 10am suggests that the explosions happened above sea level: submarine explosions did not happen until the end.
Krakatau: volcanic danger
The death toll was officially 36417. It included the entire population of nearby Sebesi island where there were no survivors, hit by pyroclastic flows and tsunamis. The death toll is the second largest reported from an eruption in historical time, after Tambora. The majority of the casualties came from the tsunamis.
Still, Krakatau was a large but not exceptionally large eruption. The total ejecta are estimated at between 15 and 25 cubic kilometer. Tambora ejected in excess of 50 km3, Pinatubo around 10km3, and the Novarupta eruption (1912) and Laki (1783) both around 15 km3. The Taupo eruption in New Zealand, AD 180, erupted some 120 km3. So Krakatau was large, but there have been comparable and larger eruptions in historical times. The island was not recognized as a dangerous volcano, even when erupting in May to August 1883. If the cause of the calamity that followed is the collapse of a deep magma chamber, this could happen at any volcano, whether in the sea and on land.
The conclusion is that what makes a volcano dangerous is not primarily the eruptive history. Its location, including how many people live within 100 km, and the size and stability of the magma chamber are important. Krakatau is much more than gripping history. It is a warning from the past.
Albert, Jan 2015
The eruption of Krakatoa, and subsequent phenomena (1888): report by the Royal Society. https://archive.org/details/eruptionkrakato00whipgoog
http://blogs.scientificamerican.com/history-of-geology/2011/08/27/august-27-1883-krakatoa/ Reports from ships in the Strait.