Influence of town on PDC temperature - Earth-printsInfluence of town on PDC temperature 1 1 Influences of urban fabric on pyroclastic density currents at Pompeii (Italy), part II:

Influence of town on PDC temperature

1

Influences of urban fabric on pyroclastic density currents at Pompeii (Italy), part II: 1

temperature of the deposits and hazard implications 2

E. Zanella 3

Dipartimento di Scienze della Terra, Università di Torino, Via Valperga Caluso 35, 10125, Torino, Italy 4

ALP – Alpine Laboratory of Paleomagnetism, Peveragno, Italy 5

L. Gurioli 6

Istituto Nazionale di Geofisica e Vulcanologia, Via della Faggiola 32, 56126, Pisa, Italy 7

Present address: Geology & Geophysics, University of Hawaii, 1680 East-west Rd, Honolulu HI 96822, USA 8

M.T. Pareschi 9

Istituto Nazionale di Geofisica e Vulcanologia, Via della Faggiola 32, 56126, Pisa, Italy 10

R. Lanza 11

Dipartimento di Scienze della Terra, Università di Torino, Via Valperga Caluso 35, 10125, Torino, Italy 12

ALP – Alpine Laboratory of Paleomagnetism, Peveragno, Italy 13

14

During the AD 79 eruption of Vesuvius, Italy, the Roman town of Pompeii, was covered by 15

2.5 m fallout pumice and then partially destroyed by pyroclastic density currents (PDCs). Thermal 16

remanent magnetization (TRM) measurements performed on the lithic and roof tile fragments 17

embedded in the PDC deposits allow us to quantify the variations in the temperature (Tdep) of the 18

deposits within and around Pompeii. These results reveal that the presence of buildings strongly 19

influenced the deposition temperature of the erupted products. The first two currents which entered 20

Pompeii at a temperature around 300-360 ºC, show drastic decreases in the Tdep, with minima of 21

100-140 ºC found in the deposits within the town. We interpret these decreases in temperature as 22

being the result of localized interactions between the PDCs and the city structures, which were only 23

able to affect the lower part of the currents. Down flow of Pompeii, the lowermost portion of the 24

PDCs regained its original physical characteristics, emplacing hot deposits once more. The final, 25

dilute PDCs entered a town that was already partially destroyed by the previous currents. These 26

PDCs left thin ash deposits which mantled the previous ones. The lack of interaction with the urban 27


2

fabric is indicated by their uniform temperature everywhere. However, the relatively high 28

temperature of the deposits, between 140 and 300 ºC, indicates that even these distal, thin ash 29

layers, capped by their accretionary lapilli bed, were associated with PDCs that were still hot 30

enough to cause problems for unsheltered people. 31

KEYWORDS: Pompeii, temperature, magnetic fabric, pyroclastic density currents. 32

33

1. Introduction 34

Pyroclastic density currents (PDCs) are the primary cause of death during explosive eruptions 35

[Tanguy et al., 1998]. These currents are hot mixtures of gas, pumice and lithic fragments, ranging 36

in size from fine ash up to metric blocks and bombs [e.g. Freundt and Bursik, 1998; Druitt, 1998]. 37

In the proximal locations their destructive effect is mainly due to their high momentum and 38

temperature [e.g. Baxter et al., 2005]. In distal locations, where the currents have already lost a 39

large portion of their solid load, their hazard is associated with their high concentration in fine ash 40

[e.g. Horwell and Baxter, 2006], coupled with their high velocity and temperature [e.g. Todesco et 41

al., 2002]. Thus, in order to assess the hazard posed by PDCs to human populations, it is extremely 42

important to understand PDC dynamics and their physical characteristics. In addition, educational 43

efforts, based on scientific analyses of PDC emplacement and their effects, will likely improve the 44

chances of survival of, or correct response to, future eruptions. 45

Observations made in villages and areas devastated by PDCs report evidence of sudden death 46

and survival among groups of people sheltering in the same house and even in the same room, as 47

during the eruption of Mt. Pelée, Martinique, in 1902 [Anderson and Flett, 1903,]. There is also 48

evidence for survival of certain individuals, when the majority of people in area impacted by PDC 49

were killed, as for example during the 1980 eruption of Mt St Helens (USA) [Bernstein et al., 1986] 50

or during the dome-collapse-fed pyroclastic flows of 1991 at Mount Unzen [Baxter et al., 1998]. In 51

addition, during the 1902 eruption of Mt. Pelée un-burnt material was found in close proximity to 52

incinerated objects [Lacroix, 1904]. The variable impact PDCs on human populations has been 53


3

explained at Montserrat in terms of survivors being located in areas marginal to the PDCs [Baxter et 54

al., 2005]. However, this explanation cannot be applied for unconfined, diluted PDCs, as shown by 55

Gurioli et al. [2005]. Further investigations are thus required if we are to understand these local 56

variations in PDC dynamics, and how a PDC interacts with a town and its human population. 57

We attempt to resolve this issue through analysis of the PDC deposits of the AD 79 eruption 58

of Vesuvius (Italy) which crop out within and around the Roman town of Pompeii. All PDCs that 59

entered the town, even the most dilute ones, were density stratified currents whose lower part 60

interacted with the urban fabric [Gurioli et al., submitted]. We use a combination of thermal 61

remanent magnetization (TRM) and anisotropy of the magnetic susceptibility (AMS) analyses to 62

obtain information regarding flow direction of these parent currents and deposit temperatures. 63

These data, when integrated with volcanological field investigations, reveal that the presence of 64

buildings strongly affected the distribution and accumulation of the erupted products; where the 65

results of our analyses from the single, most destructive unit in the sequence have been presented in 66

Gurioli et al. [2005]. Here, we build on these previous findings by now considered the TRM results 67

from the entire eruptive sequence cropping out within and around Pompeii to infer the temperature 68

of all of the deposits. This allows us to assess the cooling effect of the urban fabric on these currents 69

and the effect of these currents on the inhabitants. 70

71

2. Volcanological setting and sampling 72

Pompeii is located 9 km southeast of Vesuvius (Figure 1) and the following 73

reconstruction of events during the AD 79 eruption at that location is based on the analysis of 74

Sigurdsson et al. [1985], Carey and Sigurdsson, [1997] and Cioni et al. [2000]. The town first 75

experienced 7 hours of air fall comprising white pumice, lapilli and bombs (up to 3 cm in 76

diameter), with scarce lithic blocks of up to 3 cm in diameter. This emplaced unit EU2 (Figure 77

2). The town then underwent 18 hours of grey pumice, lapilli and bombs (up to 10 cm in 78

diameter) fallout to emplace EU3 (Figure 2); a deposit that is rich in lithic blocks of up to 7 cm 79


4

in diameter. During this Plinian phase, some PDCs were generated by the discontinuous collapse 80

of marginal portions of the convective column. Only the last of these events (EU3pf1, Figure 2) 81

was able to reach the north-western edge of Pompeii, but did not enter the town itself. This PDC 82

left only a 2-3 cm thick ash layer interbedded with the fallout deposit. The town was then 83

completely covered by a 5-30 cm thick ash layer (EU3pf, Figure 2) emplaced by very dilute 84

PDCs, derived from the total collapse of the Plinian column. Locally EU3pf was able to interact 85

with obstacles of a few decimetres in height, suggesting that its denser, thin lower part was 86

capable of only filling minor negative depressions [Gurioli et al., submitted]. 87

EU3pf was next mantled by a 3-6 cm thick, lithic rich (blocks up to 3 cm in diameter), 88

grey pumice, lapilli and bomb (up to 3 cm in diameter) fall deposit (EU4, Figure 2), emplaced 89

from a second, short-lived, lithic-rich column. The collapse of this second column generated the 90

most powerful, turbulent, PDC of the eruption. This was able to partially destroy the town and 91

left relatively coarse-grained, cross stratified, meter-thick deposits (EU4pf, Figure 2). EU4pf was 92

able to interact with obstacles of a few meters in height, showing significant interaction with the 93

town [Gurioli et al., 2005]. 94

Finally, the settlement was buried by a 1 m thickness of very dilute PDC and air-fall 95

deposits (EU7 and EU8, Figure 2) consisting of ash and accretionary lapilli emplaced during the 96

last, phreatomagmatic phase of the eruption. The EU7 sequence (Figure 2) comprises two 97

centimetre-thick grain supported, lithic-rich lapilli beds, separated by a 1-to-3-cm thick cohesive 98

ash layer, and capped by a coarse ash layer, which is in turn covered by a massive pisolite 99

bearing fine ash bed. EU8 then comprises an alternation of normally graded, ashy bedsets, each 100

up to 10 cm thick (Figure 2). Each bedset is characterised at its base by a massive-to-crudely 101

stratified facies followed by accretionary lapilli facies. In general, the EU7 and EU8 deposits 102

mantled a town that had already been severely damaged by the preceding currents (mainly 103

EU4pf) [Gurioli et al., submitted]. 104


5

Within this sequence we sampled undisturbed sites in open country around the city, as 105

well as sites along the city walls and within the town, both along the roads and inside the rooms 106

(Figure 1). As described in Gurioli et al. [2005] we collected lithics and stripped roof tiles on 107

which we could perform TRM analyses. At least 5 lithic or tile fragments were collected from 108

each unit at each site giving a total of more than 200 samples. The tile fragments and the larger 109

lithic clasts (Figure 3) were first oriented using clinometer, magnetic compass and, whenever 110

possible, sun compass, before being removed from the outcrop. Usually two specimens were cut 111

from individual fragments to improve accuracy in the estimate of the deposition temperature 112

interval. The great majority of lithic clasts in the AD 79 deposits, and thus around 70% of our 113

samples, are fragments of a few mm in dimension (Figure 3). They were considered as un-114

oriented samples, because their small size prevents accurate orientation. The presence of 115

fragments of plaster and roof tiles picked up and heated by the PDCs makes the Pompeii deposits 116

particularly suitable for such a study. As already discussed in Evans and Mareschal [1986], 117

Marton et al. [1993], Zanella et al. [2000] and Cioni et al. [2004], building fragments within the 118

deposits are reliable magnetic thermometers, because they were cold when they were picked up 119

by the PDC. 120

121

3. Measurement of deposition temperature using TRM 122

The thermal remanent magnetization (TRM) acquired during the cooling of a magmatic rock 123

records the polarity, direction and intensity of the Earth’s magnetic field at the time of cooling. 124

Such paleomagnetic information is widely used in geodynamics and stratigraphy. However, 125

thorough investigation of the TRM features may yield information on the physical processes 126

which led the remanence acquisition and help in understanding the emplacement mechanisms of 127

magmatic rocks. This applies to pyroclastic deposits, whose formation results from combination 128

of thermal and sedimentological processes. Paleotemperature investigation of pyroclastic rocks, 129

emplaced by PDCs, was pioneered by Aramaki and Akimoto [1957] and Chadwick [1971]. Their 130


6

assumption was straightforward. If a deposit was emplaced hot (where hot means at higher 131

temperature than the Curie point of magnetite), then the primary remanence of the embedded 132

lithic fragments would have been erased. A secondary remanence was then re-acquired during 133

the subsequent cooling. All fragments were magnetized at the same time and their TRM 134

directions are therefore well clustered. In contrast, if the emplacement temperature was cold each 135

fragment would have retained its own primary TRM acquired when the parent rocks formed. In 136

this case, the TRM directions are randomly distributed because of the chaotic movements during 137

emplacement. 138

Following Chadwick [1971], Hoblitt and Kellog [1979] applied a more quantitative 139

procedure, based on thermal demagnetization of the fragments. This technique allows derivation 140

of the TRM blocking temperature (Tb) spectrum and thus identification of the distinct TRM 141

components acquired in different temperature ranges, usually referred to as high-Tb and low-Tb 142

components. Such an analysis yields an estimate of the actual temperature reached by the 143

fragment upon re-heating within the pyroclastic material, this being the maximum unblocking 144

temperature (Tbmax) of the low-Tb component. 145

Further work has improved this methodology [e.g. McClelland and Druitt, 1989; Bardot, 146

2000] and have revealed some complications to the initial assumptions. According to the 147

simplest model, paleomagnetic estimates of the deposition temperature relies on two basic 148

assumptions: 149

1) During PDC transport and deposition heat is transferred from the hot gas and fine-grained 150

material to the cold clasts. For clasts larger than 2-5 cm, thermal equilibrium is mainly reached 151

during residence in the deposit rather than the emplacing current. In fact, the time required to 152

reach thermal equilibrium within the current is longer than the time of residence in the flow 153

[Cioni et al., 2004]. For this reason, rock magnetic investigations give an estimate of the deposit 154

(Tdep) rather than the emplacement (Temp) temperature. 155

2) The clasts’ remanence is a pure thermal remanent magnetization (TRM). 156


7

In cases where these assumptions are fulfilled, the natural remanent magnetization (NRM) of a 157

clast consists of two TRM components characterized by different blocking temperatures (Tb). 158

The high-Tb component pre-dates the PDC emplacement. This was acquired when the lithic clast 159

was originally formed. A part of this remanence is erased when the clast is re-heated within the 160

PDC and then acquired as low-Tb component when the PDC comes to rest and cools down. 161

Considering the full Tb spectrum of the clast, Tbmax is the temperature value which separates the 162

two components. It can be identified by stepwise thermal demagnetization because the directions 163

of the two components are different (Figure 4). The low-Tb direction is close to that of the 164

geomagnetic field at the time of the PDC emplacement, and it is the same in all clasts. The high-165

Tb direction is fully random. According to the first assumption above, Tdep may thus be derived 166

from the mean Tbmax value derived from a number of clasts. 167

Estimation of Tdep, however, is not always as straightforward as in the above case 168

[Grubensky et al., 1998]. The first problem is that we do not know the thermal history of the 169

clast. It might have been either picked up cold along the slopes of the volcano, or ejected hot 170

from the conduct, possibly being even hotter than the PDC within which it became entrained. In 171

the second case, when the current stops the lithic fragment is still hotter than the surrounding 172

fine-grained matrix and the Tbmax value found from rock-magnetism will be higher than Tdep. 173

The second problem concerns NRM which, in addition to the thermal (TRM) components, 174

may also comprise chemical (CRM) and viscous (VRM) components. It has been shown by 175

McClelland [1996] that a chemical remanence (CRM) may develop due to mineralogical 176

changes during reheating. This will hinder the identification of the low-Tb and high-Tb 177

components (Figure 4b). VRM is typical of ferromagnetic grains with low relaxation time, and 178

thus low blocking temperature. According to theory [Pullaiah et al., 1975, Bardot and 179

McClelland, 2000] a VRM component acquired at 20 °C in the course of the ~1930 years 180

elapsed since the AD 79 eruption, is erased by heating at 125 °C in a time of 25 to 30 minutes, 181

typical of a thermal demagnetization step in the laboratory. It may thus be indistinguishable from 182


8

very low Tb secondary components. Moreover, a small VRM overprint often occurs in most 183

rocks. The presence of a VRM overprint could partially affect identification of the true, low-Tb 184

TRM component and hamper the determination of its direction by principal component analysis 185

[Kirschvink, 1980]. 186

The problems outlined above do not always occur and when they do they can often be 187

overcome. In the case of the Vesuvius AD 79 eruption, archaeological remains are of great help. 188

Bricks and tiles, which have their own TRM typical of baked-clay artefacts, were picked up by 189

PDCs at the ambient temperature and could not be heated at values higher than Tdep. The 190

problems with CRM are also reduced sampling clasts of as different lithologies as possible, 191

because different lithology means different magnetic properties. 192

In conclusion, the various clasts collected at an individual site may have had different 193

thermal histories and have recorded more or less faithfully the Tdep. Following the approach of 194

Cioni et al. [2004], the thermal overprint of a group of clasts by a PDC is better represented by 195

what they have in common. The Tbmax values derived from each individual clast may differ from 196

each other because of the reasons summarized above, whereas the re-heating was a single event. 197

The traces it left, Tdep, must be consistent at the site scale. 198

199

4. Standard measurements and peculiar cases 200

A total of 379 specimens were measured at the ALP laboratory (Peveragno, Italy) using a 201

JR-5 spinner magnetometer, and Schonstedt and ASC TD-48 thermal demagnetizers. Small bits 202

were measured using the plastic box + Plasticine technique of Cioni et al. [2004]. Thermal 203

demagnetization was carried out in steps of 40 °C between a starting temperature of 100 °C and 204

a maximum of ~520 °C. Whenever sister specimens from individual clasts were available, a 205

second demagnetization was carried out using the same 40 °C steps but starting at 80 °C. The 206

data were then interpreted using the principal component analysis available as part of the 207

Paleomac program [Cogné, 2003]. 208


9

On the basis of the demagnetization patterns, Cioni et al. [2004] distinguished four kinds of 209

thermal behaviour in clasts embedded within the AD 79 PDC deposits. Type A is characterized 210

by blocking temperatures higher than Tdep and its primary TRM is therefore not affected by the 211

re-heating. Type B has blocking temperatures lower than Tdep and its primary TRM is completely 212

erased during the re-heating. Type C is the archetype of lithic clasts with a TRM comprising two 213

components with distinct Tb spectra, which are well evident in the Zijderveld diagrams (Figure 214

4a). In Type D the two components are not very clear because the spectra more or less overlap, 215

show a zigzag pattern, or have points that are too close to each other to be well distinguished 216

(Figure 4b). Identification of Tbmax is thus straightforward in Type C, more complicated in Type 217

D. A similar classification was adopted by McClelland et al. [2004] whose Types 1a, 1b, 2 and 3 218

respectively correspond to the Types C, D, A, B types of Cioni et al. [2004]. In the present 219

paper, we add Type E, which comprises a few tile fragments which show no evidence of re-220

heating. The normalized intensity decay curve (Figure 4c) shows that a fraction of their 221

ferromagnetic grains have low blocking temperatures. The clasts NRM directions, however, do 222

not change throughout demagnetization up to the highest values close to the Curie point and are 223

different from that of the ambient field in AD 79. In the example, the direction of the 224

characteristic remanent magnetization (ChRM), calculated using all steps and with maximum 225

angular deviation MAD = 1° (Figure 4c), is D = 356.5°, I = 2.3°, where D is declination and I 226

inclination. We have no explanation for the occurrence of these clasts and can only consider 227

them as outliers, but which bear witness to the fact that the temperature distribution within thin 228

pyroclastic deposits is far from uniform and can in fact be highly variable. 229

The pie diagram in Figure 5 summarizes the per cent occurrence of the five types in all 230

clasts investigated in the present paper. Type C accounts for about 50 % of the clasts, Type D for 231

about 45%, with the remainder are either being Type A or E. No Type B clasts were found in the 232

present investigation. 233


10

An uncommon case is given by the plaster fragment sampled at site 18 (Figure 1). Hueda-234

Tanabe et al. [2004] have shown that plaster may be used as archaeomagnetic material. Small 235

ferromagnetic grains are free to move when plaster is applied to a wall or floor, orienting their 236

magnetic moment parallel to the Earth’s magnetic field and then being blocked when the plaster 237

dries. The vector sum of the NRM of a plaster specimen is thus given by the NRM of its 238

individual grains. The process is similar to the orientation of the grains of ferromagnetic 239

pigments in red coloured murals, which also have been shown to record the ambient field 240

direction at the time of painting [Zanella et al., 2000]. The plaster sampled at Pompeii is a kind 241

of pozzolana, one of the most outstanding results of Roman civil engineering. It was made from 242

lime and grains of volcanic rocks from the sandy deposits of neighbouring rivers. A large 243

fraction of grains are 1-2 mm in size (Figure 6a), too large to be effectively oriented by the 244

Earth’s field. Thus, it can reasonably be assumed that the remanence direction varies from grain 245

to grain. At a first glance the thermal demagnetization diagram looks quite odd (Figure 6b). 246

Because each grain has its own Tb spectrum, thermal demagnetization erases different fractions 247

of remanence in grains with different TRM directions, so that the direction of the resultant vector 248

varies randomly from one step to another, without any coherence. However, the low-temperature 249

demagnetization steps show a linear trend in the initial part of the Zijderveld diagrams (up to 250

180-200 °C in Figure 6b). This suggests that, in each individual grain, the fraction of remanence 251

with Tb < Tdep was erased when the plaster was re-heated to the Tdep of the pyroclastic material 252

filling the room and burying the walls. During cooling, all grains re-acquired a low-Tb 253

component showing the same direction, i.e. that of the Earth’s field, which is therefore coherent 254

throughout the specimen. 255

Another peculiar case is that of the lithics embedded in the fall deposits. The 16 lithics we 256

sampled were characterized by three TRM components: these being the expected high- and low-257

Tb components, as well as an intermediate-Tb component with a distinct direction around 258


11

temperatures of 340 to 480 °C (Figure 7). No evidence for an intermediate-Tb component was 259

found in the tile fragments embedded in these deposits. 260

As discussed in the previous section, most of our samples were fragments too small to be 261

oriented in the outcrop. Out of a total of 379 specimens, 145 were oriented and 75 gave two 262

clearly isolated components whose directions could be transformed to the geographical reference 263

system. The directions of the high-Tb component were widely scattered, as expected (Figure 8); 264

those of the low-Tb were more clustered. The site mean ChRM direction (D = 350.4°, I = 61.8°, 265

Fisher’s semi-angle of confidence α95 = 7.4°) is close to the AD 79 Earth’s magnetic field 266

direction as given by archaeomagnetism [Tema et al., in press], even if its statistical definition is 267

lower than usual in paleomagnetic investigations. This is often the case in paleotemperature 268

investigations [McClelland et al., 2004; Tanaka et al., 2004]. In our case, this is mainly due to 269

the fact that the analysis of the low-Tb component relies on few clustered points due to the low 270

rate of decrease in the intensity of magnetization. Furthermore it may also be biased by a VRM. 271

Even if the low-Tb directions of individual clasts show some dispersion, their overall mean (D = 272

352.0°, I = 53.7°, number of specimens N = 75, length of the resultant vector R = 69; α95 = 4.9°) 273

passes the Watson [1956] randomness test and is consistent with the archaeomagnetic direction 274

(D = 354.3°; I = 58.0°; α95 = 1.7°) derived from various materials studied at Pompeii 275

(archaeological remains, fine-grained pyroclastics, lithic clasts) [Evans and Hoye, 2005; Tema et 276

al., in press and references therein]. 277

Final estimation of paleotemperature was completed following the technique of Cioni et al. 278

[2004]. For each site and each eruptive unit, first the temperature interval which contains the 279

maximum unblocking temperature (Tbmax) of the low-Tb component of each individual clast was 280

derived from the demagnetization path. This was done by analysis of the normalized intensity 281

decay curve, the Zijderveld diagrams and the equal-area plot of directions. As a conservative 282

approach, directions separated by less than 15° were not considered as significantly different 283

[Porreca, 2004], because a small angular deviation might be due to post-depositional settling of 284


12

the fragment within the still unconsolidated rock. Tdep of each site and eruptive unit was then 285

estimated from the overlap range of the temperature intervals of the individual fragments 286

(Figures 9a, 9b, 9c1, 9c2, 9c3 and 9d). 287

288

5. The temperature of the AD 79 deposits around and within Pompeii 289

In the following discussion we present our data for the whole eruptive sequence present 290

in Pompeii, unit by unit (Figures 9a, 9b, 9c1, 9c2, 9c3 and 9d). Where ever possible, we compare 291

these results with the data obtained for the same units cropping out around Vesuvius by Cioni et 292

al. [2004]. 293

5.1 The fallout sequence (EU2-EU3 and EU4) 294

We collected just a few samples from the three main fallout deposits (EU2-EU3 and 295

EU4), mostly to check whether they were emplaced hot and whether they were able to maintain a 296

temperature high enough to heat artefacts. As already shown by several authors [e.g. Thomas and 297

Sparks, 1992; Tait et al., 1998; Hort and Gardner, 2000] pumice clasts larger than 6 cm in 298

diameter suffer little heat loss during their fall and can be emplaced at temperatures within 10-20 299

% of their magmatic temperature [Thomas and Sparks, 1992]. Thus Plinian deposits, depending 300

on their grain size, thickness and distance from the vent, can remain sufficiently hot to pose 301

hazards to life and property [Thomas and Sparks, 1992]. For the fallout deposits in Pompeii, we 302

found that the white fallout deposit had a temperature high enough to warm the tiles up to 120-303

140 °C (Figure 9a). Unfortunately, we could not collect any artefacts in the grey fallout, EU3, 304

but we can assume that this deposit was even hotter than the white one, because of its coarser 305

grained texture. 306

Within both the EU3 and EU4 fall deposits we also sampled lithic blocks. As previously 307

discussed, all these lithics are characterized by three components (Figure 7). We assume that, 308

while the high-Tb component was acquired during clast formation, the low-Tb component 309

represents the temperature of the lithic at the point at which it fell to the ground. If this 310


13

hypothesis is correct the lithic fragments reached Pompeii at a minimum temperature of 180-220 311

°C during the EU3 fallout and 220-260 °C during the EU4 fallout (Figure 9a). We do not fully 312

understand the meaning of the intermediate temperature component. It may represent heating 313

during passage as part of the gas-thrust section of the plume and/or cooling experienced in the 314

umbrella portion of the plume. 315

5.2 The first PDC entering Pompeii (EU3pf) 316

EU3pf deposits were sampled around and within Pompeii (Figure 10a). They show a 317

large variation in their Tdep, ranging from 140 to 300 °C (Figures 9b and 10a). Nowhere else 318

around Vesuvius do we find such high variability in EU3pf deposit temperatures [Cioni et al., 319

2004] (Figure 10b). Even the coldest outcrops located north of Vesuvius are relatively warm in 320

comparison to those within Pompeii (Figure 10). 321

EU3pf in Pompeii shows the highest Tdep (240-300 °C) in the northern sector (Figures 9b 322

and 10a), outside the town. However, these values are lower than the 300 to 360 °C values 323

obtained for the EU3pf deposits emplaced upstream of Pompeii, in the proximal sector (Figure 324

10b). We explain these results as the consequence of uniform cooling experienced by the current 325

at this distance from the vent, due to the reduction of its total load which will decrease its 326

thermal energy. 327

Lowest temperatures were recorded within the town and in the western sector, where Tdep 328

drops to 140-220 °C. Slightly higher values have been found in rural areas south of Pompeii, 329

where the Tdep is 220-260 °C [Cioni et al., 2004] (Figure 9b and 10a). These temperatures show 330

that the EU3pf current, even if diluted and capable of only emplacing thin deposits, entered 331

Pompeii with a minimum temperature of 260-300 °C. The decrease in temperature within 332

Pompeii, indicates that the local interaction with the city structures had a cooling effect on the 333

lower part of the current. South of Pompeii, the EU3pf current was unable to restore the same 334

temperature it had before entering Pompeii, but it was still able to emplace hotter deposits than 335

found within Pompeii with temperatures of up to 220-260 °C. 336


14

5.3 The most powerful PDC (EU4pf) 337

Temperature data obtained from the thick deposits of EU4pf display the largest 338

variability in temperature at the scale of individual sampling sites, ranging from 100 to 320 °C 339

(Figures 9c1, 9c2, 9c3 and 11a). These values differ from those obtained from the same deposits 340

in non-urban areas around the volcano, (Figure 11b). In such non-urban locations, all sites yield 341

temperatures of around 300°C (260-340 °C), irrespective of their distance from the vent [Cioni et 342

al., 2004]. This relatively uniform temperature suggests a substantial homogeneity of the 343

transport system of the EU4pf deposits in the Vesuvius area [Cioni et al., 2004]. 344

In contrast, a large decrease in temperature occurs within Pompeii. In sites examined 345

orientated along the axis of the flow direction (from the northwest edge of the city towards its 346

southern edge), temperatures generally decline. As found for EU3pf, EU4pf also has highest Tdep 347

values in the northern sector of Pompeii, where the Tdep range from 240 to 320 °C, with an 348

average value of around 280 °C. A low value of 200-240 °C was found up flow the Villa dei 349

Misteri, at site 2a (Figure 11a). We speculate that this anomalous value may be due to the 350

presence of some structure up flow of this area, or some morphological high, not now visible 351

because is covered by the deposits and the modern soil and vegetation. 352

Inside the town Tdep ranges from 100 to 220 °C, with an average value of 160 °C. Here 353

the lowest values are found in three rooms with collapsed roofs, aligned parallel with the main 354

flow direction. The lowest value that we found is located in the third of these three rooms, i.e. 355

the furthest down flow [Gurioli et al., 2005]. These low values are consistent with cooling due to 356

strong disturbances caused by the town and morphological features, such as the 3-meter-deep 357

cavities presented by rooms with collapsed roofs (Figure 11a, sites 10, 12a and 12b), the 10-m-358

high cliff on the southern edge of the town (Figure 11a, sites 22c and 25), or collapsed walls 359

(Figure 11a, sites 11 and 16). Down flow of the city walls, where there are no morphological or 360

urban disturbances, the deposit temperatures are high once more (Figure 11a, sites 22a, 22b, 19a 361

and 19b). 362


15

These results show that the presence of the settlement resulted in substantial cooling of 363

the current over short distances. Roughness of the topographic/depositional surface increases the 364

ability of the basal portion of the flow to decouple from the main flow and to form local 365

vortexes, ingesting ambient air. Increases in turbulence, due to the surface irregularities caused 366

by the presence of the town, are evident from upstream particle orientations which develop 367

down-flow of obstacles or inside cavities [Gurioli et al., 2005] and from characteristic 368

sedimentary structures such as fines-poor, undulatory, lenticular bedded facies on the lee side of 369

the obstacles [Gurioli et al., submitted]. This would also cause air ingestion. Air ingestion into 370

the lower system of the EU4pf current during passage over the urban canopy is the most 371

reasonable cause of the observed strong temperature decreases. As shown in Cioni et al. [2004], 372

the very high thermal diffusivity of air with respect to magma, and the intimate mixing between 373

the air and gas-ash mixture, results in instantaneous thermal equilibrium during this process. 374

Furthermore, the EU4pf current lasted for 8-10 minutes [Gurioli, 1999]; because we witness 375

cooling this interval of time must be sufficient for the lower part of the current to entrain air and 376

undergo cooling. 377

The amount of building material entrained by the current seems not to play an important 378

role in cooling of the deposits, where we found no correlation between low temperatures and 379

amount of building material. Furthermore, tile-fragment-rich zones within EU4pf deposits are 380

present as small lenses (around 1-2 m long, 1 m high and less than 1 m wide) which probably did 381

not have sufficient volume or extent to cool the deposits. 382

The roof tiles have very high thermal conductivity, as a result they will heat up very 383

quickly. We estimate the characteristic time it takes to heat a cool object (roof tile) buried in a 384

hot medium (the deposit) from τ = D2/α, in which D is the object dimension and α is the thermal 385

diffusivity of the object. Thermal diffusivity is calculated using the density and specific heat 386

capacity for common brick [obtained from Holman, 1992], as well as the temperature dependent 387

thermal conductivity which we calculate for clay following Vosteen and Schellschmidt [2003]. 388


16

This gives τ of ~1 minute for our smallest (1 cm) objects, increasing to ~1.5 hours for our largest 389

(10 cm) objects for heating from ambient to 200-400 °C. This means that all of our objects 390

should have equilibrated with the temperature of the deposit within 1.5 hours, with the smallest 391

objects reaching equilibrium in just a few minutes. Thus, although individual tile fragments 392

entrained within the deposit were heated quickly, they robbed insufficient thermal energy from 393

the surrounding hot body to cause significant cooling. 394

As shown in Gurioli et al [submitted], the urban canopy encouraged deposition and the 395

upper portion of the current was not able to fully restore the sediment supply to the lower 396

current. Our temperature results also show that the increase in surface roughness caused by the 397

presence of the town caused strong variations in temperature. However, these variations where 398

short-lived and confined to the lower part of the current. South and south east of Pompeii, at 399

distances of up to 8 km from the town, EU4pf was able to emplace hot deposits once more 400

[Cioni et al., 2004] (Figure 11b). 401

5.4 The final PDCs (EU7 and EU8) 402

Temperature data obtained from the thin deposits of EU7 and EU8 display the lowest 403

variability in temperature at the scale of individual sampling sites. In EU7 we sampled the ash 404

interlaid between the two lapilli beds (Figure 2d). This gives a temperature of between 210 and 405

260 °C, in agreement with the temperature range (180-240 °C) found by Cioni et al. [2004] 406

between the vent area and Pompeii for this deposit. The second ash layer shows a highest 407

temperature of 260-300 °C at Site 1 (Figure 12) but around and within Pompeii Tdep ranges from 408

180-260 °C, without showing strong variations within the city. This behaviour agrees with a 409

scenario within which there was interaction between this current and a town that was already 410

partially covered by the previous deposits [Gurioli et al., submitted]. 411

EU8 was sampled at just one site because we had difficulties in finding lithic fragments 412

large enough to be suitable for measurement. It displays a Tdep of 180-220 °C. A fragment of tile 413

was also collected at site 16 (Figures 9d and 12) within the coarse grained ash layer of this 414


17

deposit. Even if the top of the deposit shows evidence of water condensation, the deposit at the 415

base had a temperature able to heat the tile up to 130-180 °C. 416

417

6. Implications for volcanic hazard 418

In the early afternoon of August 24 AD 79 a pumice fall began which lasted until early in 419

the morning of the following day. During this period Pompeii was covered by 3 meters of pumice. 420

Within 6 hours the roofs and parts of the walls of the buildings had collapsed under the pumice load 421

[Sigurdsson et al.,1985; Luongo et al., 2003a]. Luongo et al. [2003a, b] identify a significant 422

number of victims within this deposit (38 % of the total number of victims, estimated to be about 423

1150). They probably died as a consequence of building collapse. Our new data reinforce this 424

reconstruction, where we find that these deposits were emplaced with sufficient heat to be able to 425

heat cold materials to 140 °C. We can therefore speculate that in some sites they could have been 426

capable of causing damage by carbonisation of wood, such as roof beams, and skin burn upon direct 427

contact. This hazard scenario was made even more dangerous by the scattered rain of large, very 428

hot, lithic blocks. 429

In the early morning of August 25, the PDCs started to enter and devastate Pompeii 430

[Cioni et al., 2000]. In Pompeii, features of both EU3pf and EU4pf show that the two currents 431

were able to interact with the urban structures even in the first, very dilute case, suggesting that 432

the currents were stratified, and capable of interacting with objects of the same height as the 433

thickness of their depositional systems [Gurioli et al., submitted]. EU3pf interacted little with the 434

fabric of the town, due to its very thin depositional system. However, its content of fine ash and 435

relatively high temperatures would have made it hazardous to the human population, causing 436

asphyxia and lung damage. Recent studies [Luongo et al., 2003a, b] suggest that the first diluted 437

PDC (EU3pf) caused minimal damage in Pompeii. This is true for the building structures, but 438

our evidence indicates that even this current was extremely dangerous for the inhabitants [Cioni 439

et al., 2000]. Here we have been able to quantify this hazard, where the hazard results from the 440


18

temperature of this current and its composition of fine ash. Although the current left only a thin 441

deposit, the current itself had a thickness of more than 10 m (the height of the ridge on which 442

Pompeii was built). Furthermore, the current was hot, with a temperature of 140-300 °C. This 443

temperature would have been very close to the average temperature within the current at 444

Pompeii, an assumption made plausible by the abundance of fine-grained fragments that 445

comprised the PDC at this distance from the vent [Cioni et al., 2004]. Finally, the grain-size of 446

the EU3pf deposits indicates that 40-50 % of its mass was accounted for by particles of less than 447

0.1 mm in diameter [Gurioli et al., submitted]. Such a size represents dust that can be inhaled by 448

humans. Assuming a minimum fractional particle volume concentration of about 1-5 x10-4 for 449

this current and a minimum bulk density of 1000 kg/m3 for the particles [Freundt and Bursik; 450

1998], the concentration of fine materials is between 0.03 and 0.15 kg/m3. These values fall 451

within the concentration range for inhalable dust capable of causing asphyxia [0.1 kg/m3, Baxter 452

et al., 1998] at ordinary temperatures. A temperature of 200 °C and ash concentration of 0.1 453

kg/m3 are considered threshold values above which human survival is likely to be impossible 454

[Baxter et al., 1998]. Furthermore this atmosphere would have persisted for several minutes, as 455

suggested by the low terminal velocity of the fine particles and the aggradational model 456

proposed for this current [Gurioli, 2000; Gurioli et al., submitted]. Thus, this PDC was 457

extremely hazardous and, following Baxter et al. [1998], would have led to asphyxia and severe 458

lung damage. 459

Invasion of Pompeii by EU4pf was almost instantaneous. If we assume flow velocities of 460

50-60 m/s [Esposti Ongaro et al., 2002], in agreement with the average mean grain size of the 461

deposits [Gurioli et al., submitted], the EU4pf front took less than 10 s to cross the town. This 462

second, more concentrated current had a more profound influence on the town. Its dense 463

depositional system, coupled with its high velocity, was able to tear down some walls orientated 464

at right angles to the main flow direction (i.e. east-west trending structures). This is in agreement 465

with masonry vulnerability of the old Pompeii buildings that fall in the range of 1-5 kPa 466


19

[Nunziante et al., 2003] and simulated dynamic pressures of 1 kPa calculated for a PDC with a 467

mass effusion rate of 5 x 107 kg/s and at a distance of 7.5 km from the vent [Esposti Ongaro et 468

al., 2002]. The case of EU4pf is more clear-cut, in that it would have been completely lethal for 469

any inhabitants surviving the previous PDC, a consequence of its high velocity, density, mass 470

and temperature. 471

The final, dilute PDCs entered a town that was already partially destroyed by the previous 472

currents. At this stage of the eruption, all the remaining population were dead [Sigurdsson et al., 473

1985; Cioni et al., 2000; Luongo et al., 2003a]. These currents just mantled the ruins left by 474

EU4pf without inflicting further damage upon the buildings. However, even though these 475

currents caused no further damage to Pompeii and its population, our data are significant in that 476

they show that even distal, thin ash layers can be emplaced by currents that are still hot enough 477

to cause problems for unsheltered people. 478

479

Acknowledgement. 480

We are indebted to P.G. Guzzo, G. Stefani , A. d’Ambrosio, A. Varone and C. Cicirelli for 481

archaeological assistance; G. Di Martino, B. Di Martino, A. Cataldo for their logistic help in the 482

archaeological sites; A.J.L. Harris for informal review. We thank C. Frola and E. Deluca for 483

some TRM measurements and their help in the field. M. Bisson for the GIS database of Pompeii. 484

A.J.L. Harris, E. Tema, S. Ranieri and M. Lanfranco contributed to field work. This work was 485

partially supported by the European Commission, Project Exploris EVR1-CT-2002-40026 and 486

the financial support of INGV and University of Torino. 487

488

References 489

Anderson T., and J.S. Flett, Report on the eruption of The Soufriere in St Vincent 1902 and on a 490

visit to Montagne Pelee in Martinique, Phil Trans R Soc London, 100, 353–553, 1903. 491


20

Aramaki, S., and S. Akimoto, Temperature estimation of pyroclastic deposits by natural 492

remanent magnetism, Am. J. Sci., 255, 619-627,1957. 493

Bardot, L., Emplacement temperature determinations of proximal pyroclastic deposits on 494

Santorini, Greece, and their implications, Bull. Volcanol., 61, 450-467, 2000. 495

Bardot, L., and E. McClelland, The reliability of emplacement temperature estimates using 496

palaeomagnetic methods: a case study from Santorini, Greece, Geophys. J. Int., 143, 1, 39-497

51, 2000. 498

Baxter, P.T., A. Neri, and M. Todesco, Physical modeling and human survival in pyroclastic 499

flows, Natural Hazard, 17, 163–176, 1998. 500

Baxter, P.J., R. Boyle, C. Paul, A. Neri, R. Spence, and G. Zuccaro, The impacts of pyroclastic 501

surges on buildings at the eruption of the Soufriere Hills volcano, Montserrat, Bull. 502

Volcanol., 67, 292–313, 2005. 503

Bernstein, R.S., P.J. Baxter, H. Falk, R. Ing, L. Foster, F. Frost, Immediate public health 504

concerns and actions in volcanic eruptions: lessons from the Mount St. Helens eruption, 505

May 18-October 18, 1980. Am. J. Public Health, 76, 25-37, 1986. 506

Carey, S., and H. Sigurdsson, Temporal variations in column height and magma discharge rate 507

during the 79 AD eruption of Vesuvius, Bull. Geol. Soc. Am., 99, 303-314, 1987. 508

Chadwick, R.A., Paleomagnetic criteria for volcanic breccia emplacement, Geol. Soc. Am. Bull., 509

82, 2285-2294, 1971. 510

Cioni, R., L. Gurioli, A. Sbrana, and G. Vougioukalakis, Precursory phenomena and destructive 511

events related to the Late Bronze Age Minoan (Thera, Greece) and AD 79 (Vesuvius, 512

Italy) Plinian eruptions; inferences from the stratigraphy in the archaeological areas, in The 513

Archaeology of Geological Catastrophes, edited by W. G. McGuire et al., Geol. Soc. Spec. 514

Publ., 171, 123–141, 2000. 515

Cioni, R., L. Gurioli, R. Lanza, and E. Zanella, Temperatures of the AD 79 pyroclastic density 516

current deposits (Vesuvius, Italy), J. Geophys. Res., 109, 1–18, 2004. 517


21

Cogné, J.P., Paleomac: a Macintosh application for treating paleomagnetic data and make plate 518

reconstructions, Geochem. Geophys. Geosyst, 4, 10007, doi:10.1029/2001GC000227, 519

2003. 520

Druitt, T.H., Pyroclastic density current, in The physics of explosive volcanic eruptions, edited 521

by J.S. Gilbert and R.S.J.Sparks, Geol. Soc. London, Spec. Publ., 145, 145-182, 1998. 522

Esposti Ongaro, T., A. Neri, M. Todesco, and G. Macedonio, Pyroclastic flow hazard assessment at 523

Vesuvius (Italy) by using numerical modeling. II. Analysis of flow variables, Bull. Volcanol., 524

64, 178–191, 2002. 525

Evans, M.E., and G.S. Hoye, Archaeomagnetic results from southern Italy and their bearing on 526

geomagnetic secular variation. Phys. Earth Planet. Int., 151, 155-162, 2005 527

Evans, M.E., and M. Mareschal, M., An archaeomagnetic example of polyphase magnetization: 528

Journal of Geomagnetism and Geoelectricity, 38, 923–929, 1986. 529

Freundt, A., and M.I. Bursik, Pyroclastic flow transport mechanism, in From magma to tephra: 530

modelling physical processes of explosive volcanic eruptions, edited by A. Freundt and M. 531

Rosi, Elsevier, Amsterdam, 173-246, 1998. 532

Gurioli, L., Pyroclastic flow: classification, transport and emplacement mechanisms, Plinius, 23, 533

84-89, 2000. 534

Gurioli, L., M.T. Pareschi, E. Zanella, R. Lanza, E. Deluca, and M. Bisson, Interaction of 535

pyroclastic currents with human settlements: evidences from ancient Pompeii, Geology, 536

33, 6, 441–444, 2005. 537

Gurioli, L., E., Zanella, M.T., Pareschi, and R., Lanza, Influences of urban fabric on pyroclastic 538

density currents at Pompeii (Italy), part I: Flow direction and deposition, J. Geophys. Res., 539

submitted. 540

Grubensky M.J., G.A., Gary, and J.W., Geissman, Field and paleomagnetic characterization of 541

lithic and scoriaceous breccias at Pleistocene Broken Top volcano, Oregon Cascades, J. 542

Volcanol. Geotherm. Res., 83, 93-114, 1998. 543


22

Hoblitt, R.P., K.S., Kellogg, Emplacement temperatures of unsorted and unstratified deposits of 544

volcanic rock debris by paleomagnetic tecniques, Geol. Soc. Am. Bull., 90, 633-642, 1979. 545

Hort M., and J. Gardner, Constrains on cooling and degassing of pumice during Plinian volcanic 546

eruptions based on model calculations, J. Geophys. Res., 105, 25,981–26,001, 2000. 547

Horwell, C.J. and P.J. Baxter, The respiratory health hazards of volcanic ash: a review for 548

volcanic risk mitigation. Bull. Volcanol., 69, 1-24, 2006. 549

Hueda-Tanabe, Y., A.M., Soler-Arechalde, J., Urrutia-Fucugauchi, L., Barba, L., Manzanilla, 550

M., Rebolledo-Vieyra, and A., Goguitchaichvili, Archaeomagnetic studies in central 551

Mexico – dating of Mesoamerican lime-plasters, Phys. Earth Planet. Int., 147, 269-283, 552

2004. 553

Kirschvink, J.L., The least-squares line and plane and the analysis of palaeomagnetic data, 554

Geophys. J. Roy. Astron. Soc., 62, 699–718, 1980. 555

Lacroix, A. La Montagne Pelée et ses Eruption (Masson, Paris, 1904), 1904. 556

Luongo, G., A. Perrotta, and C. Scarpati, Impact of the AD 79 eruption on Pompeii, I. Relations 557

amongst the depositional mechanisms of the pyroclast products, the framework of the 558

buildings and the associated destructive events, J. Volcanol. Geotherm. Res., 126, 201-223, 559

2003a. 560

Luongo, G., A. Perrotta, C. Scarpati, E. De Carolis, G. Patricelli, and A. Ciarallo, Impact of the 561

AD 79 eruption on Pompeii, II. Causes of death of the inhabitants inferred by stratigraphic 562

analysis and areal distribution of the human causalities, J. Volcanol. Geotherm. Res., 126, 563

169–200, 2003b. 564

Marton, P., D.H. Tarling, G. Nardi, and Pierattini D.,An archaeomagnetic study of roof tiles 565

from temple E: Selinute, Sicily, Science and Technology for Cultural Heritage, 2, 131–566

136, 1993. 567


23

McClelland, E., Theory of CRM acquired by grain growth, and its implications for TRM 568

discrimination and paleointensity determination in igneous rocks, Geophys. J., Int., 126, 569

271-280, 1996. 570

McClelland E., C.J.N., Wilson, and L., Bardot, Palaeotemperature determinations for the 1.8-ka 571

Taupo ignimbrite, New Zealand, and implications for the emplacement history of a high-572

velocity pyroclastic flow, Bull., Volcanol., 66, 492-513, 2004. 573

McClelland, E.A., and T.H. Druitt, Palaeomagnetic estimates of emplacement temperatures of 574

pyroclastic deposits on Santorini, Greece, Bull. Volcanol., 51, 16-27, 1989. 575

Nunziante, L., M. Fraldi, L. Lirer, P. Petrosino, S. Scotellaro, and C. Cicirelli, Risk assessment 576

of the impact of pyroclastic currents on the towns located around Vesuvius: A non-linear 577

structural inverse analysis, Bull. Volcanol., 65, 547–561, 2003. 578

Porreca, M., Applicazioni di metodi paleomagnetici per lo studio della messa in posto di flussi 579

piroclastici. Il caso delle unità vulcaniche recenti del cratere di Albano (Italia Centrale), 580

unpublished PhD Thesis , University of Roma Tre, 1-118, 2004. 581

Pullaiah G.E., E., Irving, L., Buchan, and D.J., Dunlop, Magnetization changes caused by burial 582

and uplift, Earth Planet. Sci. Lett., 28, 133-143, 1975. 583

Sigurdsson, H., S. Carey, W. Cornell, and T. Pescatore, The eruption of Vesuvius in 79 AD, Nat. 584

Geogr. Res., 1, 332-387, 1985. 585

Tait S., R. Thomas, J. Gardner, C. Jaupart, Constraints on cooling rates and permeabilities of 586

pumice in an explosive eruption jet from colour and magnetic property. J. Volcanol. 587

Geotherm. Res., 86, 79–91, 1998. 588

Tanaka, H., H., Hoshizumi, Y., Iwasaki, and H., Shibuya, Applications of paleomagnetism in the 589

volcanic field: a case study of the Unzen Volcano, Japan, Earth Planets Space, 56, 635-590

647, 2004. 591

Tanguy, J.C., C. Ribiere, A. Scarth, and W.S. Tjetjep, Victims from volcanic eruptions: A 592

revised database, Bull. Volcanol., 60, 137–144, 1998. 593


24

Tema E., I. Hedley, and P. Lanos, Archaeomagnetism in Italy: a compilation of data including 594

new results and a preliminary Italian secular variation curve, Geoph. J. Int. (in press) doi: 595

10.1111/j.1365-246X.2006.03150.x 596

Thomas, R.M.E. and R.S.J. Sparks, Cooling of tephra during fallout from eruption columns, 597

Bull. Volcanol., 54, 542-553, 1992. 598

Todesco, M., A. Neri, T. Esposti Ongaro, P. Papale, G. Macedonio, R. Santacroce, A. Longo, 599

Pyroclastic flow hazard assessment at Vesuvius (Italy) by using numerical modelling. 1. 600

Large-scale dynamics, Bull. Volcanol., 64, 155–177, 2002. 601

Watson, G.S. A test for randomness of directions, Mon. Not. Roy. Astornom. Soc. Geophys. 602

Supp., 7, 160-161, 1956. 603

Zanella, E., L. Gurioli, G. Chiari, A. Ciarallo, R. Cioni, E. De Carolis, and R. Lanza, 604

Archaeomagnetic results from mural paintings and pyroclastic rocks in Pompeii and 605

Herculaneum, Phys. Earth. Planet. Inter., 118, 227-240, 2000. 606

607

Figure captions 608

Figure 1. Ancient Roman town of Pompeii. Dots = sites of studied sections; light areas = portions 609

of ruins still buried by undisturbed AD 79 deposits. Inset upper right, shaded relief map of the of 610

Vesuvius region. 611

612

Figure 2. AD 79 deposits at Pompeii. On the left, the schematic stratigraphy according to the 613

nomenclature of Cioni et al. [1992, 2004]. The numbers in the brackets are the variation 614

thickness of the studied deposits. From a) to d) particulars of the deposits within and around 615

Pompeii (see figure 1 for site location). (1) White pumice lapilli and bombs. (2) Grey pumice 616

lapilli and bombs. (3) Massive to stratified coarse-grained ash and grey pumice lapilli. (4) 617

Accretionary lapilli in in coarse and fine ash matrix. 618

619


25

Figure 3. Hand sampling of lithic fragments and roman tiles from the PDC deposit matrix. The 620

orientation of the main face of the tile is traced. Inset: standard paleomagnetic specimens and 621

little bits embedded in the plasticine. 622

623

Figure 4. Stepwise thermal demagnetization of lithic clasts from the AD 79 deposits: a) type C 624

clast; b) type D clast; c) type E clast (see text for further explanation). 625

Left: Zijderveld diagrams . Symbols: full dot = declination; open dot = apparent inclination. 626

Right: equal-area projections of the directions of magnetization. Symbols: full/open dot = 627

positive/negative inclination. Directions in the Zijderveld diagrams are represented in the sample 628

reference system; in the equi-areal projections the geographic reference system is used. 629

630

Figure 5. Pie diagram of the percentage occurrence of different types of fragments (see text for 631

further explanation). 632

633

Figure 6. Stepwise thermal demagnetization of a plaster specimen from the AD 79 deposits. 634

a) plaster bit (front and transverse section). 635

b) normalised intensity deacy curve and Zijderveld diagram of specimen T19a. Symbols as in 636

Fig. 4 637

638

Figure 7. Zijderveld diagrams of specimens of ballistic blocks from the fall-out deposit. For 639

symbols see figure 4. 640

641

Figure 8. Equal-area projection of high-Tb (HT) and low-Tb (LT) component directions of the 642

oriented specimens from EU4 pf at site 12a. Symbols: full/open dot = positive/negative 643

inclination; star = site mean direction with α95 confidence ellipse. 644

645


26

Figure 9. Evaluation of the AD 79 deposits temperature (Tdep), by overlap of individual fragments 646

reheating temperature range (see text for further explanation). Types are shown left of the bar. 647

Color: black = lithic fragment (mainly lavas); grey = tile; black dot = plaster. 648

a) EU2, EU3 and EU4 fall deposits. 649

b) EU3pf deposits 650

c) EU4pf deposits 651

d) EU7pf I ash, EU7pf II ash and EU8 deposits. 652

653

Figure 10. Tdep variation of EU3pf: 654

a) within and around Pompeii. 655

b) around the Vesuvius area [modified from Cioni et al.,2004]. 656

657

Figure 11. Tdep variation of EU4pf: 658

a) within and around Pompeii. 659

b) around the Vesuvius area [modified from Cioni et al.,2004]. 660

661

Figure 12. Tdep variation of EU7pf I ash, EU7 II ash and EU8, within and around Pompeii. 662

663

Figure 13. Destructive effects of the AD 79 deposits in Pompeii: a) collapsed roof tiles in the 664

fall-out deposit (photo courtesy Soprintendenza di Pompeii); b) skeletons in the ash deposits 665

inside a room (photo courtesy Soprintendenza di Pompeii); c) collapsed wall by EU4pf; d) 666

bodies rolled in EU4pf deposits. In c) and d) the arrow indicates the main flow direction. 667

Documents

Influence of town on PDC temperature - Earth-printsInfluence of town on PDC temperature 1 1 Influences of urban fabric on pyroclastic density currents at Pompeii (Italy), part II: