Ash cloud surges have been fatal during numerous volcanic events in history. Despite the numerous episodes, ash cloud surge hazard is still underestimated and not fully contemplated in hazard maps. According to 2 the vulnerable zone should be double extended respect to the associated basal high concentration pyroclastic currents.
The under-consideration of the hazard of such turbulent, dilute, and high temperature ash cloud surges lies in the little preservation in the stratigraphic record of their thin and easy to wash-away deposits. Few centimeters of very fine ash deposit are what remains after the passage of these flows despite their high thermal impact on people and objects.
The recent volcanic episodes occurred in 1991 at Unzen 7,8,26,59, in 2010 at Merapi Volcano 9,12, in 2018 in Guatemala 11,60,61, and in 2019 in New Zealand 62 highlight the need to deepen our understanding of ash surge hazard in terms of their thermal impact.
In the case of Fuego de Guatemala 2018 eruption when more than 300 people died (although independent evaluations sadly suggest up to 2,900 deaths) 11,61, ash cloud surges detached from valley confined high concentration flows killed many for suffocation and severe burns 60. Some victims’ bodies were found on topographic highs, far from the valley ponds where thick and high concentration pyroclastic currents accumulated, and they were only partly covered by thin ash layers displaying pugilistic attitude as a result of exposure to high temperatures 60. They were therefore certainly killed by high temperature of ash cloud surges at the periphery of high concentration flows, rather than by the impact of dynamic pressure. Similarly, the base surge generated during the Whakaari Volcano phreatic eruption in White Island, New Zealand, killed 22 tourists and injured other 25. Some victims died in hospitals after being rescued, for the consequences of severe burnings due to a combination of heat and of acidic gasses 63. These tragic volcanic events display remarkable similarities with the most iconic eruption of the 79CE Vesuvius.
The heat-induced effects suffered by the victims, notably the explosion and charring of skulls, vaporization of brains, cracked and charred bones, cracked teeth, contraction of limbs and thermal degradation of blood haemoproteins 14–16 indicate the occurrence of an early extremely high thermal event higher than the previously estimated temperature of about 500°C. Unlike Pompeii, where many bodies show the typical post-mortem stance known as pugilistic attitude, the lack of such corpse attitude at Herculaneum testifies to the rapid disappearance of soft tissue, as the pugilistic stance is due to dehydration and shortening of muscles induced by intense heat 14,16. However, until now, no direct measures of such high temperature early PC event were made at Herculaneum.
Our study on charcoal reflectance records for the first time the occurrence of subsequent thermal events at decreasing temperature which affected Herculaneum. By comparing the pyroclastic stratigraphy (Fig. 1e) and the thermal stratigraphy recorded by the polymodal distribution of charcoal reflectance we can reconstruct the succession of PC events that impacted the city.
The first ash cloud surge (S1 or EU2/3pf according to authors) entered in Herculaneum with a temperature exceeding 550°C recorded by samples collected at the Collegium Augustalium and the Decumanus Maximus (Table 1). This is the minimum temperature of the ash cloud as the polymodal charcoal reflectance records an early uncomplete carbonification which testifies a short-lived event, unable to reach full equilibrium, which in experiments is usually attained after at least 24hr, at constant temperature, for pieces of wood measuring 2 x 5 cm 38.
This early > 550°C event was later followed by the succession of PCs which finally buried the town under 20 meters thick volcanic deposits (Fig. 1e). These later flows were characterized by lower temperatures as testified by the presence of multiple modes within the same charcoal sample from which we inferred at least two carbonization events at temperatures ranging from 390 to 465°C and from 315 to 350°C, respectively. The lower temperatures of these later events can be explained by the progressive involvement of ground water during the course of the eruption (see phreatomagmatic phase 31 ; Fig. 1e).
The occurrence of an early > 550°C short lived ash cloud event leaving only a thin ash layer on the ground, and later followed by the deposition of lower temperature but thicker pyroclastic deposits, allows to understand the conditions for the formation and preservation of a vitrified brain recently discovered within a victim’s skull in the Collegium Augustalium 17. The transformation into glass of fresh cerebral tissue in a hot environment is only possible if two conditions are met: i) the heating event is short-lived, so that the tissue is not fully vaporized 15, and ii) once the ash cloud has vanished, the body is not fully entombed in a hot deposit, a necessary condition to allow the very rapid cooling required to attain vitrification 17,66. This allows to recognize that S1 was an ephemeral, dilute event, and that a sufficient time interval occurred for the fast cooling of the body still partly exposed to air before the following PCs progressively entered and covered the town. The lower temperature of these later PC deposits explains the preservation of the vitrified brain, as well as of the high reflectance values within polymodal distributions. If subsequent PC were at higher temperatures, the vitrified brain would have been reheated above the glass transition temperature and gone lost in its neuronal ultrastructure, which is instead integrally preserved 32, as well as the charcoal fragments would had been totally reset at higher, unimodal Ro% values.
The temperature of the first S1 ash cloud surge, previously only generically inferred by heat effects on both the victims’ skeletons 13–16 and the vitrified brain 17,64, is now recorded at minimum temperature of > 550°C by high reflectance values in polymodal charcoal datasets, whereas all other paleo-thermal data from the rest of the pyroclastic sequence indicate lower temperatures of diachronic processes related to the later burial of the town 34,36,37,65,66.
Dilute ash clouds are characterized by high air-entrainment coefficients27,67, therefore the recorded high temperature of S1 cannot be explained by a dilute current generated at vent and propagating as a surge for 7 km along the Vesuvius slope. Instead, we propose a new interpretation for the first Herculaneum ash cloud surge as detached from nearby high-concentration pyroclastic currents, as occurred at Unzen in 1991 7,68 and Volcan de Fuego in 2018 11. High-concentration pyroclastic currents, especially where valley confined, can maintain very high temperatures for kilometers from the vent as air entrainment is very limited (e.g., 20–22), whereas the overriding ash clouds may also maintain a similar high temperature for as long as they are supplied with mass and heat from below 23. However, as soon as the ash cloud detaches from the basal high-concentration current and becomes an independent dilute surge, the temperature rapidly drops due to fast air entrainment and heat exchange promoted by the fine grain size of the pyroclasts.
Herculaneum was built front facing the seashore on a relief some 10–15 m higher than sea level (Fig. 1), therefore on a topographic high likely sided north and south by valleys along which the denser parts of the pyroclastic currents would have been confined (Fig. 3a,c).
The most likely scenario for S1 at Herculaneum is that of an ash cloud decoupled 2 from its parent high concentration valley confined pyroclastic current just in the proximity of the city, so that the ash cloud could form at high temperature (Fig. 3a,b,c). A similar scenario was reconstructed for the early phases of the Fogo A eruption (Azores) 69.
Once crossed the city, 200 m downstream respect to the Decumanus Maximus, the S1 ash cloud surge jumped onto the beach and into the waterfront chambers (Fornici, Fig. 1), where it instantly killed the people who had taken refuge there 13. The thermal effects detected on the victims’ bones found in the Fornici 13–16, well match with the > 550°C ash-cloud temperature measured upstream at the Collegium Augustalium and the Decumanus Maximus. The scattered preservation of bone collagen does not appear to be evidence of a low temperature of the ash cloud surge as claimed by some author 33, but it seems to be related to the amount of heat transfer the victims' bodies and bones were exposed to during the short-lived ash cloud event. Actually, the greater or lesser extent of heat effects on the skeleton, or even on a single bone element, has been shown to be closely related to the lesser or greater crowding of victims inside the Fornici, and also amount of fleshy mass present in different anatomical districts, even at the level of a single bone 15,16. However, the persistence of proteins such as collagen and other organic components of bones at Herculaneum is most likely independent by exposure to more or less intense heat but can be rather correlated to the burial environment in which the victims’ skeletons were embedded until their discovery after about 2000 years. At Herculaneum, after sudden death and rapid thermally induced soft tissue vanishing, the skeletons were buried in an alkaline, anoxic soil permanently waterlogged 70, environment able to inhibit chemical changes from microbial attack 71,72 thus allowing long-term survival of organic matter in the bone.
The average lower temperature between 325 and 350°C (Table 1) detected by charcoal at Fornici may be also explained by the fast heat exchange due to the interaction of the ash cloud with both human bodies and the nearby seawater. Direct observations of pyroclastic currents entering the sea after travelling along the slopes of stratovolcanoes (e.g., 73,74) show that they rapidly inflate and cool due to entrainment of seawater. This effect has been documented at the nearby Villa dei Papiri 37, and it is in agreement with well documented examples, including the 1902 ash surge that destroyed Saint Pierre, Martinique 4,5 and the ash surges of Secche di Lazzaro, Stromboli 75, where the combination of topographic effects and seawater engulfment at the coastal surge-water interaction, promoted the sudden surge expansion and even the backflow uphill of the cooled and dilute surge.
We therefore interpret the first ash cloud surge S1 to have been very short-lived, reaching the coast and the Fornici still at > 500°C, leaving almost no deposit but killing the people there13,14. The interaction of the ash cloud surge and seawater caused the surge inflation and the settling of cooled ash immediately after, which then embedded the skeletons of the people already killed instantly by the extreme heat (Fig. 3b A-A’ profile). This interpretation explains the apparent disagreement of recorded temperatures at the seashore and also the difference in thickness of S1, which is maximum 20 cm thick in town, whereas it reaches 50 cm along the pre-eruption coast (up to 150 cm in the Fornici according to 30), where deposition was controlled by the slowed and cooled water-mixed ash cloud.
The results of this study bear unprecedented implications for the mitigation of volcanic risk at Vesuvius and possibly elsewhere. The red zone at Vesuvius, where full evacuation of ca 700.000 people is planned in case of a future eruption 76 was designed based on the probability of PC invasion derived from the geological record 77. While this is certainly the goal to be achieved, it remains uncertain whether the progression of the volcanic unrest will allow enough time to reach the expected full evacuation prior to eruption 78. In addition, Plinian ignimbrites 76 from directional and partial collapses of the eruptive column flowing confined along valleys and prone to ash cloud detachments, are more likely respect to axisymmetric caldera-forming ignimbrite 79 from PCs covering all at once the entire red zone 77. Given these premises, we suggest that the edifices within the red zone irrespective of the need to evacuate all people before the eruption, should be reinforced to be able to shelter people from the thermal impact of ash cloud surges in case full evacuation is not achieved on time. In facts, while zones exposed to high dynamic pressure of high concentration and high velocity PCs will inevitably see the collapse of edifices and structures with very little chances of survival, other zones may be impacted by short-lived detached ash clouds where potential for survival critically depends on the ability of shelters to prevent infiltration of the hot dusty gas. This could allow people who may not had the chance to evacuate earlier to survive and wait for rescue or be able to leave before other PCs may impact the area.