How pyroclastic surges sometimes “bounce” landwards when they reach the sea.



M Edmonds and R A Herd, Inland-directed base surge generated by the explosive interaction of pyroclastic flows and seawater at Soufrière Hills volcano, Montserrat. Geology (2005) 33: 245-248.




This story began in 1982 when Fisher and Heiken published a paper in the Journal of Volcanology and Geothermal Research about the 1902 eruption of Mount Pelée in Martinique that wiped out the town of St Pierre and its inhabitants [ée]. A memorable misprint on page 341 of this paper called it Mt. Melée! Very few eyewitnesses survived the eruption and so the details of how it caused the disaster had largely to be reconstructed by studying the deposits laid down during the event. The climactic eruptions produced both pyroclastic flows (PFs) and a “halo” of pyroclastic surges around them, as happens on Montserrat. It was the surges that destroyed St Pierre although volcanologists have bickered continuously since about the precise details of their deduced behaviour.


There is a large published literature, based on observed modern examples, describing the diagnostic features of the deposits laid down by various types of pyroclastic flows and surges. The pages called "Introduction to pyroclastic flows and surges" elsewhere in this website explained these phenomena to the extent essential for a Montserrat resident. Of course the Montserrat eruption has provided many such examples for scientists to study. Parts of surge deposits are layered and the shapes of the layers, and relationships between individual layers (Wikipedia, “cross bedding”), show which way the surge was moving when it deposited the sediment.


For the 1902 Mt Pelée eruption climax, Fisher and Heiken deduced from such measurements that two big PFs flowed from the summit area towards the coast. Surges spread out around these PFs and one destroyed St Pierre. The weird thing about the Mt Pelée surges was the map of their flow directions. Figure 1 shows what Fisher and Heiken deduced to have happened:


The main PFs ran out to sea down a steep valley (the Rivière Blanche) that closely resembled the Tar River Valley on Montserrat. The flow directions of surge deposits south of this valley (the ones that destroyed St Pierre) show that these simply burst over the bank of the Rivière Blanche and expanded downhill and along the coast, as might be expected. But something different happened on the other side of the valley, which had a high ridge running along it – almost but not quite down to the coast. Here the surge deposit flow directions suggest that the surge was trapped by the ridge but then expanded violently as it reached the seaward end of the ridge and rushed back uphill up a small adjacent valley, reaching about 100 metres above sea level.



Fig. 1  Map of main pyroclastic flows and surges during the 1902 Mt Pelée eruption climax, plus flow directions in the latter.



Fisher & Heiken suggested that the surge refracted around the seaward end of the ridge, like a wind rushing around the corner of a building, but there are other possibilities. The amazing videos taken from helicopters offshore from Tar River have shown us far more than any previous eruption about how PFs and their associated surges behave when they reach the sea. The dense main PF sinks and runs along the seafloor whilst the surge spreads out across the surface of the water (Fig. 2). All Tar River surges cause the sea surface to boil and splutter, and this activity cools and slows the front of the surge (it takes heat to boil water). This creates a situation familiar to many drivers, where the front cars in a group slow down and the ones behind pile into the back of them. This gassy pileup over the sea may help to force the following surge to “bounce” back towards the land, whereas intuition would expect that it could escape harmlessly out to sea.



Fig. 2  A relatively small pyroclastic surge reaching the sea below Tar River Valley (Copyright NERC).

Fig. 3  Pyroclastic flow with almost no accompanying surge clouds reaching the sea near Bramble Airport on 21 Sept 1997. Because the surge clouds do not obscure everything, you can see how most of the PF material just generates steam (white) when it hits the water but at one spot in the middle there is a small hydromagmatic explosion , causing a grey billowing ash cloud (Copyright NERC).




Now fast forward to the account of the eruption in July 2003 by Edmonds and Herd (Geology 2005, v. 33, p. 245), when the largest dome collapse during the present Montserrat eruption took place over a few hours. Fortunately this material passed down the Tar River Valley to the sea and thus caused the rest of the island little pain, except for a huge ash fall. Because the collapse occurred during the evening and night, only part of the process could actually be observed. These observations showed that the biggest PFs and surges did more than boil surface water when they reached the sea. As surges travelled 2-3 km out across the sea, distinctive jets of black ash rose here and there, marking where the surge material had actually exploded as it mixed with water. These are called phreatic or hydromagmatic explosions (Wikipedia) and they happen as the result of what is called a “fuel-coolant interaction” (Wikipedia) between the hot surge ash and the water (Fig. 3). Google produces about 9000 links about fuel-coolant interaction (FCI) because it is thought to occur if a nuclear reactor, like Chernobyl, goes out of control.


This type of explosion provides a convincing new reason why a surge reaching the sea might suddenly appear to “bounce back” landwards. Edmonds and Herd found one other crucial piece of evidence for violent explosions just offshore at the mouth of Tar River Valley. Here and there, on top of the distinctive fine-grained deposits out of the “bouncing” surge cloud, there are scattered boulders (up to 1 metre in size) that must have been thrown inland by the explosions. Here are the summary diagrams from their paper:





In the 2008 "first edition" of this website, I suggested that the results of this research would have little practical relevance to Montserrat residents, so long as the dome stayed relatively small and sent all its future major collapses down Tar River Valley. But of course everything has changed by April 2010, as follows:


1.  For some inexplicable reason the wholesale dome collapses down Tar River Valley whenever the dome grows very large have ceased, allowing the dome to build ever higher and more massive.


2.  The next really big dome collapse, on 11 February 2010, went NE towards the sea in the Trants and Spanish Point area.


3.  Although there were substantial explosions when these hot PFs hit the sea, there were none on the scale of the 2003 hydromagmatic events studied by Edmonds and Herd, or the similar blasts in 2006, studied by Sue Loughlin (BGS) and her colleagues.


The next two photos were taken by Bennette Roach, Editor of the Montserrat Reporter, at approximately 14.30 on 11 February 2010. This was near the end of the main phase of the dome collapse. Between them, these remarkable images show evidence of very large hydromagmatic (or hydrovolcanic or phreatomagmatic) explosions, as the pulses of pyroclastic flow material hit the sea. The swirling mushroom-shaped dark rising clouds mark the hot, ash-rich products of individual explosions. It is clear in the right-hand photo that falling ash is pouring out of them over the sea, but there are no visible (from here) splashes made by large flying boulders





What is conspicuously absent from these photos is a shower of boulders, up to a metre in diameter, being hurled by the explosions towards the photographer. Likewise a massive lethal surge cloud is not sweeping towards Bennette and about to destroy him! Therefore, despite the spectacle, this dome collapse is not producing a sufficiently intense flux of hot material into the sea to trigger explosions on the scale of those that happened during the dome collapses in 2003 and 2006. This February 2010 partial dome collapse was much smaller than the two previous ones, and Spanish Point is much further from the volcano than the sea below Tar River Valley. This gives us a hint that, even if a massive future dome collapse was to occur directly above the Belham Valley, it might be impossible for a volcano of the height and dome-size of SHV to deliver enough PF material fast enough to the mouth of the valley to trigger damaging hydromagmatic explosions, which is good news.