The control of lava flow during the 1991-1992 eruption of Mt. Etna
F. BarberP, M.L. Carapezza ~', M. Valenza" a n d L. V i l l a f f ~ ~Dipartimento di Scienze della Terra. Universiti~ di t%a, 17a ,S'..llaria 53, 5(i 120 l%a. l tah
blstitulo di Mineralogia. Pelrogratia e GeoclHmica, Unlvo:sil& di Pah'rm< t'alerm~>, l lah ~CNR, lslilUtO per la Geochimica det t.Tutdi. Palermo, Italy
aCNR, Islituto lnlernazionale di l'ulcano/n~,ia. ('alu~Hu, lla/I
( Received November 11, 1992; revised version accepted December 3. [ 992 )
ABSTRACT
All the actions carried out in 1992 to protect the village of Zafferana Etnca from being in\ adcd by lava arc described. An earthen barrier 234 m long and 21 m high was firstly built in January 1992 by accumulating with mechanical cscavators 370,000 m 3 of earth, scoriae and stones. This embankment contained the lava for about one month and was ovcrilowcd by April 9, 1992. Three additional smaller earthen barriers (lenght: 90-160 m; height: 6-12 m ) were built in ~pril to gain time while the lava front was descending towards Zafferana from the overflowed first embankment.
The major effort of the 1992 operation consisted of several attempts at stopping the lava fiont advance b x diverting the flow out from the natural and extensively tunnelled channel through a skylight near the vent. The main intervention point was located in Valle del Bore at an elevation of 2000 m, at 8 km from Zafferana, in a zone almost unaccessiblc from land: helicopters were hence extensively used during the whole operation. Initial interventions called for attempts at plugging a tunnel by dumping into it linked concrete blocks, hedgehogs and blasted portions of the solid levee. Each intervention caused the partial obstruction of the tunnelled channel, which determined major increases of lava overflow in Valle del Bove and the consequent halt of the most advanced fronts. However, benefits were of brief duration, at the most two weeks of respite, before new lava fronts approached again and again the outskirts of Zafferana. The final successful intervention was carried out on May 27-29. An artificial channel was dug departing from the natural one. The solid separation lcwe was thinned to 3 m and blasted by 7000 kg of explosives. After the explosion, 2/3 of the lava flowed spontaneously in the artificial channel and then the total diversion was obtained, the tunnel being plugged by dumping into the natural tlow 230 m 3 of lava boulders. As a consequence of the intervention the active natural lava front, that on May 27 was only 850 m flom Zafl'erana, came to an halt, as did the entire flow downhill t¥om the diversion point, bringing back the situation as ~t was five months earlier, a few days alter the beginning of the eruption, with the new front of the dixertcd flow at 6-7 km fi-om Zafferana. In June 1992, the effusion rate halved from 30 to 15 m3/s and with this reduced thrust the lava was no longer capable of covering long distances. Five months after the conclusive intervention, the diverted lava continues to flow over its initial natural field but remaining confined in the upper Vallc del Bore. without an~ nc,a threat to Zaffcram~.
Introduction
Lava flows, because of their usually low ad- vance velocity, only rarely threaten human lives. A remarkable exception is the very fluid lava flow produced in 1977 by Niragongo (Zaire) that descended at an average speed of 30 km/h and caused 72 casualties in the vil-
('orrespondence to. F. Barberi.
lage of Goma at 10 km from the volcano (Ta- zieff, 1977 ). Lava flows destroy everything that cannot be removed from their paths and have frequently caused great economic losses. For a long time, volcanologists have conceived plans and actions to control lava flows and protect cities or settlements. However, only very few field interventions have been carried out so far: the walls trown up in 1669 to protect Catania (Lyell, 1875), the containment dikes built in
1960 near the village of Kapoho (Hawaii) (Macdonald, 1962); the earthen barriers and the lava diversion at tempt of 1983 at Etna (Barberi and Villari, 1984); the attempts at cooling advancing lava with jets of water (Ha- waii 1960, Heimay 1973, Motomaki 1986: Macdonald, 1975 and Shimozuru, 1988); the attempts at Hawaii to block a tunnel (1935 ) and to break the levee of a channeled flow (1940) by bombing and the bombing experi- ments carried out on a prehistoric tunnel and a spatter cone of Mauna Loa ( 1975 and 1976: Macdonald, 1975; Lockwood and Torgerson, 1980).
During the 1991-1992 eruption ofMt. Etna, which began on 14 December 1991 and is pres- ently ( 31 October, 1992 ) still going on, several interventions have been carried out in order to protect the town of Zafferana Etnea from being invaded by lava. The 1992 Etna operations in- cluded the building of four lava-containment earth barriers, several attempts at plugging a tunnel by concrete blocks, steel hedgehogs and large fragments of solid lava obtained by blast- ing the channel levee, and finally, the total di- version of the lava flow from a skylight near the vent into an artificial channel by blasting the separation wall between the two channels
and obstructing the lava tunnel by dumping into it big boulders by mechanical means.
As a whole, the 1992 Etna interventions are the most complete series of actions carried out so far on a volcano to control a lava flow. The aim of this paper is to provide a detailed de- scription of all the interventions carried out. and to illustrate the reasons for the technical solutions adopted, the difficulties encountered and the results achieved, so to share the expe- rience with the volcanological communityo
Chronology of the eruption
After 23 months of rest, and preceded by an uplift of the volcanic edifice and a short se- ismic crisis, a new eruption of Mt. Etna began on 14 December 1991 (GNV, 1992). Two fractures, with N and SSE trends, originated from the base of the SE crater at about 3000 m elevation (Fig. 1). For a few hours, ash and bombs were emitted from the N fracture in the proximity of the SE crater. At the same time, the SE fracture propagated for about 1300 m down to an elevation of 2700 m, along the frac- ture system opened in the same zone in 1989 (Barberi et al., 1990; Fig. 1 ). Lava fountains were produced from the upper part of the frac-
TABLE 1
Characteristics of the 1991-1992 Mt. Etna eruption
Lava composition Porphyricity Index Normative Color Index Mg number Temperature
Covered surface Travelled distance Viscosity (*) Density Effusion rate Total volume of emitted lava
bawaiitc 30-33% 33 53.2 1080 'C (January 3, Valle del Bove: lava channel 600 m from the vent i 1060C (April 18, Portella Calanna: ephemeral vent at 7 km from the vent I 7 km 2 8.5 km 447-624 Pa.s 2.65 g/cm 3 30 m3/s ( 15 m3/s since June 1, 1992 ) 500X 106 m 3 (October 31, 1992 )
(*) Viscosity was calculated at 1080°C using the method of Shaw (1972), the mean composition of 22 analyses and correcting for the porhyricity index assuming R = 1.67 (Marsh, 1981 ). Chemical and petrographic data are from Armienti et al. (1992). The relatively high viscosity, however similar to that of 1989 lava (GNV, 1990), depends on the low water content, assumed equal to kO.l . (0.46 wt.%).
THE CONTROL OF LAVA FLOW, 1991-1992 ERUPTION OF MI. ETNA 3
Pizzt Denerl
M~. ETNA
Torre del
Mt. S lmone
z '9~ 9
ML Centenad $ k
/ d i LE oJsl 8OrE /"~ Oi3, rnazz?,
xX,X, ~ La Monlagnola i
2 2 0 0 ~ _ ~ . . ~ . . _ ~ , .
Rif. Sapienza . - - ~ ~ ,,. / ' ~ - - S P 92_, / /~--~,'~"f \\\,
y h
, . , I
Mt. Fontane
'Portoila
5
ZTZ77t ,
C%% /
~ . i ° 1
,.-
Zaf[e?ana ' \ xac~C~ ; ~ - . etnea
/ 1 l, : ,.>., . . . . . " "~.~
Fig. 1. Schematic map of the area affected by the 1991-1992 Mt. Etna eruption, l = eruptive fracture opened in December 14-15. 1991; 2 = t h e 1989 fracture system: 3= lava field from the vent to the first barrier: 4= la~a overflowed dox~nhill from the first barrier: 5=d ive r t ed lava flow (on June 15, 1992).
ture together with two small lava flows that de- scended eastward for 600-800 m (Fig. 1), while the fracture continued to propagate downslope. In the night between 14 and 15 De- cember, lava began to pour out from the lowest extremity of the fracture on the western wall of Valle del Bove, at an elevation of about 2200 m, 3 km from the SE crater (Fig. 1 ). From this date, the eruptive vent did not move from that location and the fracture did not further prog- ress. For nearly two months the vent was char- acterized by strombolian and mild phreato- magmatic activity and then only by quiet degassing. The effusion rate was initially eval- uated to be 18-25 m3/s and from 12 January
15-10 m3/s. These estimates were largely in defect, as a precise measurement made on the diverted flow on 27 May 1992 gave a value of 30 m3/s, decreased to 15 m3/s by 1 June 1992.
During ten and a half months of eruption, over 500 × 106 m 3 of lava have been emitted. This eruption, whose main characteristics are summarized in Table 1, is still in progress at the moment of writing (31 October 1992 ). It is already the biggest Etna eruption in over 300 years in terms of total volume of emit ted lava, mean effusion rate and distance reached by the flow. It is second only to the terrible 1669 eruption, that emitted over 900 × 10 6 m 3 of lava that travelled 17 km down to the sea shore, de-
4 F BARBERI El ,\l
stroying dozen of villages and the western part of Catania town. In Table 2 the 1991-1992 eruption is compared with those which oc- curred on Etna during the last 20 years.
Five days after the beginning of the erup- tion, because of the persistence of the inflation of the volcano, the Civil Defense Minister was advised that most probably the eruption would have a long duration. A computer simulation of the most probable paths for the lava was carried out by G. Macedonio (Volcanic Simu- lation Group, Pisa) (Fig. 2). This simulation was based on the identification of the steepest slopes ( 500,000 runs) on a digitised 1 : 10,000 topographic map, down to an arbitrary dis- tance of 10 km from the eruptive vent. It was
TABLE 2
Mt. Etna effusive activity between 1971 and 1992
clear from the morphology and the simulation that the village of Zafferana Etnea (7000 in- habitants ) was on the lava path. It was not easy, however, to establish whether the lava flow could travel the 9 km separating the outskirts ofZafferana Etnea from the vent. The distance that can be reached by a flow depends on many factors, such as effusion rate, lava viscosity, vent location, slope steepness and formation of thermally insulating tunnels during lava ad- vance (Chester et al., 1985 ). Observations on the historical Etna eruptions (Walker. 1978) indicate that a correlation exists, although with a large dispersion of values, between the mean lava output rate (m.o.r) and the distance trav- elled by the lava. Using the initial estimate of
Year Onset Duration Vent Vent altitude A.A. L, ...... (days ) location f m.a.s.l. ) ( m.a.s.l. )
1971 Apr. 5 69 SE and N E flank 31;100-1800 600 5 1974 Jan. 30 17 W flank 1670 1500 1.5 1974 Mar. 11 18 W flank 1650 1400 1.3 1974-78 (*) 824 NE crater and NW flank 3200-2600 1900 2 t 978 Apr. 29 37 SE crater and E flank 3000-2500 1600 4 1978 Aug. 24 6 E flank 3000-2300 1650 3.5 1978 Nov. 18 12 SE flank 3000-1700 1100 5 1979 Aug. 3 6 E and NE flank 3000-1800 870 6 1980 Sept. 1 2 NE crater 3200 2200 4 1980 Sept. 6 2 NE crate, 3200 2600 2 1980 Sept. 26 1 NE crater 3200 2700 2 1981 Feb. 5 ~'~ NE crater 3200 2600 .~ 198 l Mar. 17 6 N W flank 2600 -1150 600 8 1983 Mar. 28 131 S flank 3000-2200 1100 7 1984 Apr. 27 172 SE crater 3000 1900 3 1985 Mar. 8 127 SE crater and S flank 3000-2650 1800 3 1985 Dec. 25 4 E flank 2900-2500 1750 3.5 1986 Sept. 14 10 NE crater 3200 2900 1.2 1986 Oct. 30 120 E, NE flank 2900 -2200 1300 5 t 989 Sept. 11 17 SE crater 3000 1900 2 1989 Sept. 28 12 E, NE flank 2640-2550 1100 7.6 1991 (**) Dec. 14 321 SE flank 2200 740 8.5
v5
2.4 2~
~5 23
3 5 75 ! l
0.6 1.7
18 100
10 30
1 I
60 12.2 26.2
5O0
2 5
i,4
].]2
4 4
9 8 34 7
5,.9
7 L~
24.3 23.7
A.A.: Altitude attained by lava flows; Lma x = ma x i mum length of lava flows (km); V= total volume of lavas (m3X 106 ). Accuracy of eruption volumes ranges between +_ 10% when activity was continuously recorded and _+ 30%; m.o.r. = mean output rate (m ~ S - I ) .
* An estimate of the cumulative duration and volumes has been given tbr the long series of short eruptions from the nor theas tern crater and on the northwestern upper flank of the volcano from 1974 to 1978. ** Until October 31, 1992. 1971 - 1989 data from Armienti et al. ( 1989a,b ).
F H E C O N T R O L O F LAVA F L O W , 1 9 9 1 - 1 9 9 2 E R U P T I O N O F Mr. E ~ N A
Fig. 2. Computer simulation of the most probable lava paths made on 20 December 199 l,
I O0
. . . . - r - w r ~ " V ~ . . - .~" 7"~T~: -~ - --?*, ..'~
L (km) ; ~ ~ i r__. ~ : I
.1
.1 1 0 1 O0 1 0 0 0 Mean effusion rate (mS/s)
Fig. 3. Lenght of lava flows (L) against mean effusion rate (m.e.r.) as observed in historical erupHons of Etna. Circle = m.e.r of 15 m3/s, triangle= m.e.r of 30 m3/s.
the m.o.r. (15-25 m3/s) and the empirical graph of Figure 3, a max imum travel distance of 20 km (mean of 9 km) was estimated. These figures rise to 28 and 10 km respectively, using the actual m.o.r, of 30 m3/s. Considering the vent elevation (2200 m) and using the data of Lopes and Guest (1982) , a max imum dis- tance of 11 km could be estimated. Although inaccurate, these data indicated that the vil- lage of Zafferana was exposed to the risk of being invaded by the lava.
Actually, on May 12 the flow front reached a distance of 8.5 km from the vent, only 700 m
from the village. Without the actions carried out to prevent the further advancement of the flow, the lava would certainly have severely damaged Zafferana.
Progressive advance of the lava flow
During the first days of the eruption the lava descended rapidly from the vent. Until De- cember 24 the flow remained confined within the Valle del Bore (Figs. 1 and 4); then it de- scended the steep slope of Salto della Giu- menta ( 1300-1400 m elevation) and accu- mulated on the floor of Val Calanna (Fig. 1 ).
6 F. BARBER1 U I ..'~I
Fig. 4. The eruptive vents and the lava field in Valle del Bove on December 20, 1991,
By the end of December, the flow had de- scended the entire Val Calanna, covering a dis- tance of 6.5 km from the vent, and its front was near to the narrow eastern termination of the valley (Portella Calanna) at an elevation of 990 m. The construction of an earthen barrier at Portella Calanna began on January 1, 1992. On January 9, the front stopped and there was no further advance of the lava until March 10. During this period, lava accumulated by sev- eral overflows, in Valle del Bove and Val Cal- anna, with progressive development of tunnels and ephemeral vents. By early March, the flows fed by ephemeral vents in Val Calanna over- passed the old front. On March 14, a thin lava tongue first reached the barrier on its central- northern side. Other small tongues hit against the dam in the following days: the lava was di- verted laterally and constrained to expand on the Val Calanna basin. The entire basin was progressively filled and on March 28 the lava level, raised by accumulation of many thin ov-
erflows (max. thickness 2-3 m), reached 2/3 of the barrier height (21 m ) (Fig. 5 ). On April 7, a small lava tongue began to overflow the dam. By April 10, several lava tongues over- flowed the dam joining downslope into a sin- gle flow that began to descend toward Zaffer- ana (Fig. 6 ). In order to slow the advancement of the front, three new barriers were built at 830, 810 and 770 m elevation. By mid-April, the lava overflowed the last barrier and then the front stopped at about 800 m from Zaffer- ana, 1.5 km downslope from the barrier of Portella Calanna. Several times during the fol- lowing weeks, and until the diversion of the flow near the vent was carried out on May 27, the front again descended the narrow valley down from Portella Calanna. On May 11, an overflow reached the previous front and over- passed it by 120 m. A few days after the inter- vention of May 27, the lava fronts retreated, the ephemeral vents of Val Calanna closed one after the other, and the lava cooled down within
THE O )NTROL OF LAVA FLOW. 1991-1992 ERUPTION OF MI. ETNA 7
@)ii)i: =ii~ ii!i ̧̧ ~
Fig. 5. F r o m M a r c h 14 to Apr i l 9 the Por le l la C a l a n n a e a r t h e n bar r ie r c o n s t r a i n s the lava to e x p a n d lateral ly a n d accu- m u l a t e in the ba s in uphi l l .
the tunnels. There remained only the new ac- tive front produced by the diversion of lava, in the upper Valle del Bove, as in the initial days of the eruption, at over 6 km from Zafferana (Fig. 1 ).
Actions carried out to slow or prevent the lava advance
Lava containment interventions
The first intervention to slow the advance of the lava flow consisted in building the barrier
of Portella Calanna. The decision to construct this embankment was made on January 1, 1992, and was ratified by the Commissione Grandi Rischi (the scientific advisory board to the Civil Defense minister) the following day.
The lava front had advanced swiftly during the last days of 1991, arriving about 2 km from Zafferana. The embankment, whose essential characteristics are indicated in Table 3, was constructed at Portella Calanna, at the east end ofVal Calanna (Fig. 7), where the valley nar- rows (Fig. 5 ).
8 F. BARBERI ET AL
Fig. 6. The Portella Calanna barrier being overflowed by the lava.
The purpose of the barrier, oriented orthog- onally to the direction of the lava flow, was to prevent or slow down the advance of the flow (not to divert it ), by creating a morphological obstacle which was to favour the superposi- tioning of the lava flows and lateral expansion of the lava in the large basin of Val Calanna. The barrier was constructed by specialised units of the Army and the Fire Brigade, with backup from private earth-moving equipment.
TABLE 3
Characteristics of the barriers built to contain the lava in Val Calanna
It was erected by digging into the bottom of the valley on the front of the advancing lava and accumulating loose material (earth, scoriae and fragments of lavic stone) on a small natu- ral escarpment. The original plan called for building a dike about 20 m high, with a suffi- ciently wide plane at the top to allow mobility of mechanical equipment for possibly increas- ing its height in the future. In reality, the heavy rains of the following weeks and the incoher- ent nature of most of the filling caused the top of the embankment to erode and made re- sumption of the work impossible.
On January 9, the lava front drew to within a few tens of metres of the embankment under construction, then stopped due to a sudden de- crease in the thrust caused by extensive over- flowing of the lava several kilometres farther uphill. Only on March 14, two months later, did the first tongue of lava manage to touch the base of the barrier. In the following days the flows, originating from the numerous ephem-
T H E CI ) N T R O L O F L A V A F L O W , 1 9 9 1 - 1 9 9 2 E R U P T I O N O F MI. ETNA_ 9
",4~t~ • ~#-~. , - ~ / , V / / / _ ~ . ~ r ~ M I . , ~ ~ " t
• ' . . ' , . . ' " " ' . ## . ~ , / / / / / i oD
1 0 F. BARBERI ET A L
eral vents present in Val Calanna, slowly reached the bottom of the barrier and dis- persed laterally, gradually spreading out along its whole length, with a maximum thickness of a few metres (Fig. 8). The lava continued to accumulate, through overlapping of small flows, along the whole width of the embank- ment until April 10, when the barrier was ov- erflowed (Fig. 6). Thus, the Portella Calanna enbankment contained the lava for about one month, impeding its advance towards Zaffer- ana and forcing it to spread out in the basin uphill.
Some Italian volcanologists, according to what was reported in the press, expressed doubts that the Val Calanna embankment might have a negative effect on the course of the lava flow because it could have: - caused the formation, just uphill from it, of
a lava lake, that is, an accumulation of liquid lava beneath a thin solid crust which would
later have fed the ephemeral vents of Por- tella Calanna
- facilitated tunnelling of the lava, delaying its cooling and favouring downhill migration of the ephemeral vents. The continuous monitoring of the evolution
of the lava field in Valle del Bove and in Val Calanna, permits to rule these possibilities out completely. No accumulation of lava, no "lava lake" ever existed behind the Portella Calanna embankment. As already described, the lava accumulated against it very slowly, with over- lappings of very thin flows that expanded lat- erally, displaying thin pahoehoe tongues which rapidly cooled and forming successive layers of consolidated lava. The embankment was never broken through; it was only gradually covered and then buried by a successive series of lava flows (Fig. 9).
Furthermore, no particular acceleration in the tunnelling of the lava and in the migration
Fig. 8. The lava accumulation against the earthern barrier (right) occurred by several overlappings of thin flows.
THE CONTROL OF LAVA FLOW, 1991-1992 ERUPTION O F M I ETNA I l
Fig. 9. Lava begins Io overflow the first earthen barrier (April 7/.
of the ephemeral vents was observed before or after March 14, the date the lava began to build up against the embankmen t .
The operations to slow down the lava led to the building of three other barriers between Portella Calanna and Piano dell 'Acqua be- tween April 10 and 14 (Fig. 7, Table 3). These were minor constructions, erected in a couple of working days, which could only stop the lava for a short t ime on account of the narrow mor- phology of the valley, without any natural basin where the lava could spread out.
All these embankments were oriented per- pendicularly to the valley and were con- structed so as to present a morphological ob- stacle of equal height, from one side to the other, to the advancing lava. This was done so as to absolutely avoid the possibility, and even the doubt, that they might cause a diversion of the lava outside its natural path, with the re- sulting legal consequences. Only for the last embankment , the one situated near Piano
dell 'Acqua, had the Commiss ione Grandi Ris- chi suggested the possibility that it be oriented differently in order to facilitate the flow to- wards the middle reaches of Valle San Gia- como (Fig. 7). This path in fact presented lower risks for the town of Zafferana since it was longer and, therefore, farther from the in- habited zone. It would also have concerned a much larger basin, the only one where concrete possibilities of creating new dams existed. And in fact the Army officers had identified a point in Valle San Giacomo that was morphologi- cally suitable for a new, large retaining em- bankment and had also pertbrmed a detailed feasibility study. The people of Zafferana re- fused this possibility, preferring that the lava be allowed to follow its natural course. As of mid-April, therefore, any possibility of opera- tions downhill was excluded because they could have caused a diversion of the lava from its path. Hence, all interventions had to be con- centrated uphill and aimed at trying to slow or
1 2 F. BARBERI kl" ~l
prevent the advance of the lava fronts by de- creasing the thrust behind them.
Before describing the group of uphill opera- tions, other activity performed in the course of the emergency in the zone near Portella Cal- anna must be mentioned.
Between April 13 and 16, specialized Army units conducted tests, using explosives, to blow up the sides and roofs of recently formed lava tunnels on the Portella Calanna embankment . A first test, with a 20-kg shaped charge, con- firmed that the scoriaceous lavic material ab- sorbed the energy of the explosion to a consid- erable extent, reducing its effects. In order to increase the thrust of the explosion, the second test used a steel plate interposed between the explosives (eight 20-kg charges) and the wall of solid lava to be destroyed. The explosion opened a breach in the flank of the tunnel, with a small outpouring of lava. Other tests by the Army regarded experimentation with insu- lated metal containers for the explosives to solve the problem of the high temperatures (400-600°C) on the thin walls of newly formed tunnels (S. Di Palma and V. Pennisi, pers. commun. , 1992).
Another operation was carried out between April 19 and 21 by units of the Army and the Fire Brigade and regarded an at tempt to stop the supply to the lava fronts which in those days were swiftly descending downhill from Por- tella Calanna, creating at the elevation of 970 m a lateral by-pass for the lava flow through a specially dug artificial channel . The operation was not completed because on April 21 the lava overflowed spontaneously into the artificial channel.
Interventions to divert the lava flow in upper Valle del Bove
The project
At the meeting of April 10, 1992, the Com- missione Grandi Rischi approved, in addition to the construction of the containment dikes
for the lava flow downhill from Portella Cal- anna, the experimentation of techniques of various kinds to try to achieve the diversion of the lava flow from the upper part of the supply channel. The basic idea was the same one that guided the lava flow diversion operation car- ried out on Etna in 1983: stopping or slowing down the downhill advance of the lava fronts by cutting off or greatly reducing the supply near the eruptive vents. The rheologic proper- ties of lava are such (Hulme, 1974; Chester et al., 1985 ) that, if the thrust exerted by the lava issuing from the eruptive vents is eliminated or significantly reduced, the most advanced fronts come to a halt. In practice, interrupting the lava flow near the vent by forcing the lava out of its flow channel means making the erup- tion start all over again, with a precious gain of t ime in terms of civil defence. An intervention of this type must be carried out: - w h e n the lava flow has already covered a
considerable distance from the eruptive vents and there is a high probability that it will strike inhabited or particularly valuable areas
- in proximity to the eruptive vents, where the entire lava flow is still confined within a sin- gle channel and where the distance that will have to be covered by the artificially created flow is greatest
- in such a way as to reduce to the min imum the invasion of new terrain, that is, diverting the lava on a path mostly overlapping the al- readly formed lava field. An unforeseen but resolutive phenomenon
which occurred during the 1983 Etna opera- tion guided the initial strategy of the 1992 in- tervention, The 1983 plan called for making, by means of explosives, an opening in the solid levee of the lava channel (Fig. 10) so as to get the lava to flow down into an artificial chan- nel, specially prepared and parallel to the nat- ural one. The plan also called for building earthen barriers to guide the path of the di- verted lava by impeding lateral expansions in built-up or farmed areas (Barberi and Villari, t 984 ). In reality, a number of difficulties which
Ttt E CONTROL OF LAVA FLOW, 1991 - 1992 ERUPTION OF MI. ETNA 13
Fig. 10. The 1983 intervention point. The solid levee has been thinned from 6 to 3 m and drilled to place the explosive charges. Resulting cooling caused the lava to ovcrflov~ in the .~ard.
cropped up during the preparatory phase for the intervention with the explosives (Aber- sten, 1984) prevented placing charges in the deepest part of the lava channel. The opera- tion therefore yielded partial results: a rather modest diversion of the lava flow into the ar- tificial channel which only lasted a couple of days. The most important phenomenon was the plugging of a tunnel located just downhill from the point of intervention, caused by dumping a large amount of big solid fragments pro- duced by the explosion into the lava channel. Nearly all the lava overflowed out of the plugged tunnel, thus unexpectedly producing the desired result.
The aim of the initial operations in upper Valle del Bove was therefore to repeat what had happened in 1983: a substantial outpouring of lava from the natural channel by the plugging of a tunnel by introducing a large quantity of solid material in the lava channel.
Right after the April 10 meeting of the Com- missione Grandi Rischi, the Italian govern- ment declared a state of emergency. The oper- ations downhill from Portella Calanna were entrusted to the Army (Colonels Di Palma and Pennisi) and to the Fire Brigade (Commander Murgia). The uphill operations were entrusted to Colonel Roberto Vassale of the "Arditi In- cursori" of the Navy. He was to be aided by the Fire Brigade, the Etna Guides, the Forestry Service and units of the Mountain Rescue Service. It was also decided to request the co- operation of the United States' CH53 "Black Stallion" helicopters from the NATO air base at Sigonella for transporting the heavy mate- rial up to 2000 m. For the transport of men and machinery the participation of the helicopters o f MARISTAELI o f Catania, commanded by Colonel Stefano Leuzzi, was requested and proved to be absolutely fundamental.
] 4 F BARBERI E] A [
Fig. 11. The intervention point at the elevation of 2,000 m was located just uphill from the channel [ork (Januar3 1'992 picture ).
The intervention point
In early April, the lava flow in Valle del Bove was almost entirely tunnelled, with some sky- lights here and there, particularly in the upper part of the lava field. The point of intervention necessarily had to correspond to one of these skylights, since destroying the thick roofs of the tunnels (over 5 m) proved to be practically impossible. After a test on the effects of the ex- plosion, performed on April 13 in a skylight situated at an elevation of approximately 1800 m, all the subsequent operations were concen- trated on a skylight located at about 2000 m, on the west face of Valle del Bove about 750 m away from the eruptive vents, where the whole lava was flowing in a single channel. The point was located just uphill from a morphologic rupture, with a sharp increase in slope, which should have favoured drainage of the lava pouring out from the natural channel. How-
ever, it should be noted that in the initial plan, overflow of the lava was not so much expected from the intervention point, where the tunnel was of considerable size (section approxi- mately 40 m 2, about two-third filled with lava; velocity 1 m / s ) , as from some point farther downstream, where the lava flow divided into at least two distinct channels (Fig. 11 ),
Two operations were performed on a sky- light at about 2100 m, 200 m uphill from the main intervention point. All these points were accessible from the ground only on foot, and with difficulty. In all phases of the operations, therefore, the use of helicopters was necessary.
The initial plan
In order to plug the tunnel in which the lava flowed, the simultaneous introduction of many large solid blocks was necessary. Tests with ex- plosives applied directly on the walls of the lava
THE I "ONTROL OF LAVA FLOW, 1991 1992 ERUPTION OF Mt. E1NA 15
channel had demonstrated the difficulty of ob- taining such large solid fragments. For exam- ple, the test using 480 kg of explosives (Bofors NSP72; demolition charges D3 made of com- pound B; cylindrical charges: Cyclotol) blew up about 30 m 3 of solid lava, but fragmenting it into small blocks which were swept away with- out any trouble by the stream of lava. It was therefore decided to proceed in the following two ways:
(a) Utilize the explosives on the lava chan- nel levees only where there already existed rock septa limited by natural fractures within which it was possible to place the charges in order to blow up a large spur of consolidated lava. Two points with these characteristics were identi- fied, one for each edge of the skylight at 2100 m .
(b) Transport large solid blocks to the site by helicopter. Owing to their immediate avail- ability, concrete blocks with a truncated pyra- mid shape were chosen (lower rectangular base 2.00×0.98 m; upper base 2.00X0.52 m; height 1 m: weight 3.6 t). The density of these blocks (2.4 g/cm 3 as opposed to the 2.65 g/cm 3 of the lava ) was not ideal and a denser material would have been preferable. However, it was thought it was not essential for them to sink immedi- ately into the lava flow, as they could effi- ciently contribute to plugging the lava tunnel as long as they remained wedged in. On the other hand, the low thermal conductivity (0.20-0.30 W / m ° K ) of the concrete offered sufficient guarantees as to the resistance of the mechanical properties of these blocks after their immersion in the lava at a temperature of over 1000°C. In the final phase of the opera- tions the concrete blocks were replaced by large blocks of basaltic lava (density 2.78 g/cm~). At the suggestion of prof. G. Oliveto and col- laborators of the University of Catania, it was also decided to experiment steel hedgehogs (2 x 2 X 2 m). Because of their high density (7.8 g/cm3), the hedgehogs should easily sink into the lava flow and jam in the tunnel, form- ing an obstacle for the concrete blocks. Their
breaking strength is almost totally annulled at 800 ° C, a temperature which, considering the high thermal conductivity of the steel (54 W/ re°C), should be reached in a little over 2 hours. This time was deemed sufficient for the experimentation, although it would have been preferable to insulate the hedgehogs with a layer of refractory cement: however, such an operation would have required too much time.
The first intervention planned provided for (Fig. 12):
- Creating, by means of small explosions, of a slide plane (surface area about 75 m-~; 5 m wide by 15 m long) on the roof of the tunnel immediately uphill from the skylight
- Placing a metal plate on the surface of the slide to facilitate the sliding of the blocks
- Helicopter transport and emplacement of the cement blocks
- Tying the blocks with steel chains and cov- ering the chains with a heat insulating mate- rial ( rock wool )
- Dragging the tied blocks into the lava chan- nel by means of jacks and explosives. It was also planned to use cables to keep the blocks under tension and restrain them; these hold- ing cables were to be cut with explosives when all the blocks were dumped onto the lava The plan also provided for introducing a
large volume of solid lava into the lava chan- nel, right after the blocks, to be accomplished by blowing up a septum on the right levee of the lava channel at 2100 m.
Description qf the interventions
First intervention On April 19 the operation had to be modi-
fied. It proved to be impossible to transport the steel plate up to the elevation of 2000 m (it was released in flight for safety reasons) and it was thought that without the plate it would have been impossible to make the blocks slide into the lava channel. It was then decided to set the blocks in a crown pattern around the skylight (Figs. 13 and 14 ), to place a metal grid over the lava channel fFig. 15), and bind it all
16 F. BARBERI E~I AL
i / g A /
-' > o
/"
Fig. 12. Initial intervention project at the 2,000 m skylight. (A) Preparation of a slide plane on the roof of the tunnel uphill, emplacement o f a sliding metal plate to be charged with concrete blocks tied with insulated metal chains. (B~ Dragging the tied blocks into the lava channel by means of explosive to block the tunnel downhill•
THE C O N T R O L OF LAVA. FLOW, 1991 1992 E R U P T I O N O [ M t . E / N A 17
, ~ . "5 f..<-:#/, ",~'#'X//~ : / / / / y . d ' : / 0
/ i 2 ~ ! ,/, ,' / • [ ' [ i , ' /
/
J I
f /,¢(q i ~' : ,f
Fig. 13. First intervention. Concrete blocks are arranged around the skylight and linked together with thermally insulated chains. Some are anchored to a rock spur ( 1 ). A grid (2) is placed on the lava channel. Large concrete platforms ( 3 ) arc dropped on the grid.
together with the insulated chains. The inten- tion was to make the grid fall into the lava channel by dropping large concrete platforms (3 .40×2.40×0.50 m) onto it; the grid should drag the blocks along with it, dumping a large volume of solid, bound material in the lava flow all at one time. Immediately thereafter, an explosion at 2100 m should make another large volume of rock fall into the channel. The inter- vention took place on April 21. The metal grid was installed and bound to 28 blocks (another 4 blocks were arranged farther from the chan- nel and served for anchoring); the solid sep- turn ( 1 0 × 2 X 4 m ) at 2 1 0 0 m w a s prepared and the explosives placed (ten 15-kg T2 shear charges of compound B; time 0; one hundred 3-kg charges of TNT, short delayed). Six plat- forms were dropped: 2 were swept away by the lava flow and the other 4 piled up on the grid,
resting on the slide plane without managing to drag the blocks into the channel. At this point the U.S. helicopters were asked to drop some other blocks on the piled up platforms. The helicopters made several interventions with attempts, three of them successful, to make some blocks fall into the channel. At the end of the helicopter operation a total of 24 blocks and 4 platforms remained around the skylight, with the 4 platforms and 6 blocks piled up on the slide plane. During this whole first phase a to- tal of 11 blocks and 2 platforms, in pairs or in- dividually, went into the lava channel. The spur at 2100 m was then blown up. The explosion caused the detachment of the solid lava levee. In addition, the propagation of the shock wave caused all the material piled up on the slide plane to slide into the lava channel. As a result of the explosion 4 platforms, 12 blocks and the
18 F. BARBERI ~!q A[ .
Fig. 14. The concrete blocks are crowned on the 2,000 m skylight and tied together with insulated metal chains (April 21).
crest of rock detached by the explosion spilled into the channel simultaneously, for a total volume of approximately 100 cubic metres.
The effects Immediately after the explosion a rapid
swelling of the lava was observed at the inter- vention point and lava overflowed from the skylight situated about 40 m uphill, and de- scended several dozen metres.
The next day, the skylight at the interven- tion point was tunnelled. An increase in the
number of ephemeral vents and of the lava outflow from a lava tumulus situated in upper Valle del Bove was observed. This could be the result of the partial obstruction of the left branch of the channel, which forks just down- stream from the point of intervention (Fig. 11 ). As a result of this obstruction, a greater flow of lava was channelled irrto the right branch, causing the observed phenomena.
On April 25, a large ephemeral vent opened on the left branch of the tunneled flow, in the middle part of Valle del Bove at an elevation
THE ( ONTROL OF LAVA FLOW, 1991-1992 ERUPTION OF M~. ETNA 19
Fig. 15. The metal grid on the skylight.
of about 1600 m. Since the rate of effusion from the eruptive vents had remained constant, the appearance of this ephemeral vent attested to a significant modif icat ion in the flow regime downhill from the point of intervention at 2000 m. This caused a substantial halt of the most advanced fronts and a general regression of the activity of the ephemeral vents in Val Calanna, which displayed a tendency to close, the low- ermost ones first.
These positive effects lasted for about two weeks; the ephemeral vent was gradually ex- hausted and on May 4 a new lava front began
to descend downhill from Portella Calanna again.
Second intervention In the light of the results of the first interven-
tion it was decided to make a few changes (Fig. 16). The metal grid would no longer be used; all the blocks would be placed on the thin roof of the newly formed tunnel and on the wall of the channel and made to fall in by means of explosives. It was also decided to dig, again us- ing explosives, an artificial channel to facili- tate the drainage of the lava overflowing from
2 0 F. BARBERI Eq[ AL.
..... . W - J
, t . . i - i " - 7 ~ ~ , ~ . ~ ~ ~ ., ' ~ / i , ,' st I . - , "
. . ~ ....... s'i '~ 'I \ \ ,
/ # / l " ' I ; ' / / , , . r ; ~
Fig. 16. Second intervention. The skylight at 2,000 m has been reopened. Blocks and hedgehogs are placed on the hedge and on the thin roof of the newly formed tunnel. A channel (large arrow) is being excavated by exptosivc and removing the rock fragments by hand.
the natural channel. The artificial channel had to be dug on the right side of the skylight in order to prevent the diverted flow from de- scending toward the town of Milo.
First of all, it was necessary to reopen a sky- light on the tunnelled lava channel. The heli- copters dropped 3 concrete blocks which broke through the thin roof of the tunnel creating a 3 X 3.5 m opening. On April 25, 20 blocks and 6 hedgehogs were transported up to 2000 m and they were then tied together with chains. The digging of the artificial channel began. A total of 10 explosions was made using kits ofcrater- ing charges composed of an 11-kg charge (compound B) and a 30-kg thrust charge (granular explosive); the charges, the line and the wires of the detonators were protected with asbestos sheets. The rock fragments were re- moved by hand. These operations proved to be long, dangerous and tiring. The need rose of using earth-moving equipment for the excava- tion, which had thus far been avoided due to
the difficulty of access and the opposition of environmentalist groups.
On April 29, an unforeseen event required that the intervention be moved forward. The explosion of a charge to excavate the artificial channel made part of the roof of the tunnel cave in, and 13 blocks and 2 hedgehogs fell into the lava channel. Eight blocks remained around the skylight over the channel. The level of the lava quickly rose 1.5 m, but there was no overflow into the artificial channel due to the presence of a solid separating septum about 2 m higher than the level of the lava. In view of the contin- uing swelling of the lava, which indicated the existence of an obstruction downstream, it was decided to dump all the available material into the lava channel. Two charges were set off and made 7 blocks and 1 hedgehog fall in. The Army helicopters dropped into the skylight 5 blocks and 3 metal containers half full of ce- ment (1 .30× 1.50X 1.80 m; volume 3.5 m3;
weight 3.5t). Altogether, aside from the solid
THE CONTROL OF LAVA FLOW, 1991-1992 ERUPTION OF MI. ETNA 21
rock fragments of the roof of the tunnel de- stroyed by the two explosions, 25 concrete blocks and 3 containers (total volume 35 m 3) were dumped into the lava channel.
The effects As after the intervention of April 21, the level
of the lava at the intervention point immedi- ately rose (this time without overflow from the channel), and the flow from the ephemerous vents situated in upper Valle del Bove increased.
Third intervention On May 3, lava started overflowing from the
skylight at 2100 m. In this case, too, it was dif- ficult to establish whether this overflow was caused by the April 29 attempt to block the lava channel, as small lava overflows had already been observed earlier at this point. On May 4, since the level of the lava at 2100 m was still very high, it was decided to try to facilitate ad- ditional overflow by blowing up the lava chan- nel levee at that point. Holes were made with five l l-kg charges of explosives each; in the blastholes (diameter 0.24 m; depth 1.5 m) 600 kg of granular explosive was placed, primed with instantaneous detonators. The explosion blew up the solid spur (19 X 5 X 6 m), which collapsed into the lava channel, obstructing the tunnel immediately downstream and increas- ing the lava overflow. The overflow was addi- tionally facilitated by the dropping into the skylight of blocks of solid lava (weight 3-3.5 tons each) by Army helicopters.
The overflow lasted three days and fed small branches of lava, that descended for about 700 m. In addition to the overflow at 2100 m, the May 4 intervention caused an obstruction far- ther downhill, as indicated by the considerable rise of the level of the lava that was observed in the skylight at 2000 m.
Fourth intervention On May 7, work was resumed on digging the
artificial channel at 2000 m (Fig. 17). Seven
blasts were performed using cratering kits (but with a mixture of granular and plastic explo- sives in the thrust charges) and removing the solid fragments by hand. It became absolutely clear that this method of proceeding was not producing results and it was decided to request the intervention of mechanical excavators. On May 13, a Fiat Allies FL14 excavator reached the point of intervention at 2000 m, but owing to mechanical failures it went into operation only on May 16. From May 20, a heavier ma- chine, a Caterpillar CAT 235 excavator, also went into action.
This new intervention provided for the Fol- lowing operations: - D i g g i n g an artificial channel with a depth
greater than those of the lava flow and of the adequate section, leaving between it and the natural channel a solid septum approxi- mately 3 m thick; transport of hedgehogs and lava boulders obtained on site, piling them on the roof of the lava tunnel immediately downstream from the skylight, where the roof was thinnest; chaining the transported ma- terial together.
-Blasting with explosives and short-delay charges of the solid septum and the roof of the tunnel, in order to achieve a partial di- version of the flow into the escavated channel.
- Dumping into the lava channel, by mechan- ical means, the hedgehogs and boulders in order to achieve obstruction of the natural channel, swelling of the lava and its overflow into the artificial channel. Once again, an unforeseen circumstance
made it necessary to move the intervention forward (Fig. 18 ).
The artificial channel had been deepened about 4 m below the level of the lava in the natural channel and work was proceeding on thinning the solid dividing septum between the two channels. The top part of the solid parti- tion was made of about 3 m of massive lava layers and the bottom of a bank of loose scor- iaceous pyroclastic material. In the meantime
22 F. BARBERI E l AL.
Fig. 17. Artificial channel being excavated by explosive on the right side of the 2,000 m skylight (May 7 ).
the skylight on the lava tunnel had been en- larged with mechanical equipment to 4 m in width and 8 in length. A few large boulders of lava had been piled on the left edge of the sky- light. On May 22, the pressure of the lava on the thinned solid septum caused it to give way and the lava poured into the artificial channel (Figs. 18 and 19). A lava flow, estimated at about 1/4 of the total, initially overflowed from a breach about 2.5 m wide. With the mechani- cal equipment the breach was enlarged to 5 m and the diverted flow increased to some 35- 40% of the total. The plugging of the natural channel was then at tempted by dumping 5 concrete blocks, 2 hedgehogs and numerous large lava boulders into it.
The effects Another modest overflow of lava occurred
from the skylight at 2100 m and on May 24 the opening of ephemeral vents in the middle reaches of Valle del Bove was again observed.
The flow from the artificial channel descended for about 1 km, over the central part of the lava field. The diverted flow remained around 30- 40% of the total until May 26, when it began to diminish and a tunnel was formed just down- stream from the breach. This occurred because the artificial channel had not yet been com- pleted when the lava poured into it sponta- neously and therefore the morphology was not yet favourable for substantial drainage. It was therefore decided to complete the intervention by digging a new channel, contiguous and par- allel to the first one.
Cbnclusive intervention On May 27, without any other unforeseen
events, everything was ready for the final in- tervention, On the left side of the natural lava channel there were transported 17 hedgehogs, chained together and anchored with steel ca- bles to a concrete block and from the block to a large spur downhill. The earth-moving vehi-
TH E CONTROL OF L.A.V.% FLOW. 1991-1992 ERU PTION OF Mr. ETNA 2 3
/"
++ __ ___,+..+_,.,,..,
;~-+~+~~>~,~, ~ ~+,W";";! !`A. ,~,,; ,
'-) )
Fig. 18. F o u r t h i n t e r v e n t i o n . An ar t i f ic ia l c h a n n e l ( l ) is be ing excava ted b) m e c h a n i c a l m e a n s when tile pressurc oi the lava opens a b r e a c h at the base o f the t h i n n e d solid s e p a r a t i o n scp tum, caus ing the laxu to I1o~ into the channe l < I ) . P lugging of the t u n n e l is a t t e m p t e d by d u m p i n g sol id b locks in to the na tura l lava channe l ( 2 ).
Fig. 19. The lava pressure opens a breach through the solid separating wall and the lava begins to flow into the artilicial channel (May 22 ).
24 F. BARBERI ET A L
Fig. 20. Conclusive intervention of May 27. An artificial channel 10 m wide and 15 m deep has been excavated ( 1 ), parallel to the previous artificial channel (partly tunnelled) where the lava poured out spontaneously on May 22 (5). Metal containers with 7,000 kg of explosive have been placed into a trench against the thinned solid separation wall ( 2 ). Steel plates (4) have been interposed between the explosive and the trench filling. Shaped charges (3) have been placed against a massive lava bank. Lava boulders (6) have been piled on the left side of the skylight near hedgehogs ( 7! tied together and anchored.
cles had accumulated at the same point large boulders of lava for an overall volume of about 60 m 3. The excavation of the artificial channel had been completed; it had a section of 20 m 2, was 10 m wide and reached down 7 m below the level o f the lava in the natural channel (Fig. 20). The solid septum that divided the two channels had a thickness of some 2-3 m and a height o f !1 m. The top (1.5 m) was com- posed of minutely fractured lava that rested on a layer of massive lava about 2 m thick. At the bo t tom there were banks of scoriae and loose pyroclastic material. Before the explosives were placed against the septum, the solid wall was insulated by covering it with various layers o f sheets and panels o f rock wool (Fig. 21 ). The heat insulation worked extremely well, as the temperature stayed below 36°C. The explo- sives were placed at two points. Most of it
(6000 kg of plastic explosive PE4; 1000 kg of granulated Tri tonal) , arranged inside 6 sheet metal containers (4 with PE4 only and two with a 50% mixture of PE4 and Tri tonal) was placed at the bo t tom of the thinned solid levee in a trench a couple of metres wide (Fig. 20). The bo t tom of the trench was heat-insulated with a 50-cm-thick layer of rock wool panels. The opposite wall of the trench was made of filling for a height o f 5 m and a thickness of 3 m. In order to prevent excessive absorption of the thrust o f the explosion, two steel plates, 2 m X 4 m X 2 cm, were placed between the cases with the explosives and this wall. The empty spaces between the charges were filled with small bags of sand and loose material. The whole thing was wetted down with jets of water to make it homogeneous.
In the upper part of the solid septum, against
THE ( ONTROL OF LAVA FLOW, 1991-1992 ERUPTION OF MI. ETNA 25
Fig. 21. Final preparation of the May 27 intervention. The solid separation wall has been thermally insulated by a rock- wool cover.
Fig. 22. The May 27 explosion.
26 V BARBER[ ET A I
Fig. 23. The separation wall is destroyed by the explosion, the natural tunnel is partially obstructed, Most of the Iav~ flow~ into the artificial channel (May 27 )
the massive lava bank, 6 shaped charges were placed, each with 11 kg of explosives, short-de- layed (30/1000 s) with respect to the bulk of the charges. The purpose of the shaped charges, which were oriented with a crossed direction, was to minutely shatter the 2-m-thick massive lava bank to prevent the explosion from form- ing large blocks of stone which might have blocked the artificial channel. The detonator wires were wrapped with glass wool covered by a luminium foil; the cable was raised and pro- tected by panels of rock wool in the points where it crossed the still hot lava that over- flowed from 2100 m during the preceding days.
The explosion occurred at 4:36 p.m. (Fig. 22 ). A seismograph placed about 200 m from the point of the blast recorded a shock of mag- nitude 2.3. The explosion almost entirely de- stroyed the solid septum, opening a large gap on the side of the natural channel and causing a notable flow of lava into the artificial chan-
nel. Immediately thereafter, the mechanical equipment was used to dump the chained hedgehogs and the boulders of lava into the natural channel. The chains resisted for sev- eral minutes, forming a barrier that holded the blocks; this created an obstruction of the nat- ural channel and increased the flow of lava into the artificial one. The obstruction was en- hanced by dumping another hundred cubic metres of rock boulders into the lava channel. In this way an outflow of lava from the natural channel equal to at least 2 /3 of the total flow was obtained (Fig. 23).
On May 28, the mechanical equipment was used to enlarge the breach produced by the ex- plosion and it was at tempted to maintain the partial blockage of the natural channel by dumping rock fragments into it. On May 29. the total plugging of the natural tunnel was achieved with the mechanical equipment. Twenty blocks of lava, each about 4 m 3, in ad- dition to about 150 m 3 of smaller sized frag-
THE C O N T R O L OF LAVA FLOW, 1991 1992 E R U P T I O N O F M I . EI'N~. 2-]
\ I;/, ~']~" ~,~ ~.~ / / ~ , ~
Fig. 24. On May 28 and 29, a large quantity of lava blocks of various size. obtained ill site (,X). is accumulalcd ~md dumped into the skylight up to totally block the natural tunnel and gcl the total diversion ot the (1o~ I B
ments, were piled up, again on the left edge of the natural lava channel. The blocks were pushed into the channel, followed by the dumping of the smaller material and the tun- nel was completely blocked (Fig. 24).
One hundred percent of the lava flow ran into the artificial channel: for the first time in history man had succeeded in totally diverting a lava flow.
As the dimensions of the artificial channel
2~ F. BARBERI E] AL.
Fig. 25. The diverted lava flow produced by the May 27-29 intervention.
were known, by measuring the flow rate of the lava ( 1.5 m/s ) , the total output of the lava was estimated at about 30 m3/s, a value nearly double the estimates made earlier when the real dimensions of the lava channel were not known.
The artificial flow descended along the south wall ofValle del Bove (Fig. 25 ). After 24 hours it had covered around 1300 m; by June 4 it had covered a little over 2000 m and the fronts came to a halt. Unfortunately, the lava also in- vaded a part ofValle del Bove which until then had escaped damage from the eruption and also submerged the ruins of Rifugio Menza. This caused the polemical reaction of some envi- ronmentalist groups who complained about the destruction of "mountain meadows" and "for- ests of centuries-old beech trees". In reality the few beeches on the south face of Valte del Bove are still there and the benefit-cost analysis of the operation is also positive from a purely en-
vironmental standpoint. Indeed, it should be considered that without the interventions car- ried out the lava would have destroyed the middle reaches of Val San Giacomo, a natural environment of particular beauty and value.
Meanwhile, downhill, the positive effects of the intervention began to be observed. The most advanced fronts, which on May 27 had arrived at the elevation of 840 m, 7.5 km away from the eruptive vents, came to a halt. One after another, all the ephemeral vents of Val Calanna closed, all the lava flows in the tun- nels downhill from the point of intervention blackened and stopped. Months would have to pass before the lava, which should again flow through all Valle del Bove and Val Calanna, would be able to threaten Zafferana again. But this eventuality would no longer be possible in the course of this eruption: on June 1. for the first time after nearly 6 months of eruption, the rate of effusion from the eruptive vents sud-
THE ~ 'ONTROLOF I.AVA FLOW, 1991 1992 ERUPTION OF MI. EFNA 29
denly dropped by half. On October 31, all the lava still continues to flow in the artificial channel, which has become the new, definitive route of the lava flow. But with the thrust ca- pacity halved, the fronts do not advance, the lava flows cover itself and creates new branches, but remains confined in the upper part of Valle del Bove (Fig. 1 ).
Reflections on the techniques of controlling lava flows
Diversion barriers" and retaining embankments
The slowing and diversion of lava flows caused by the lava striking against barriers built for that or entirely different purposes (the city walls and the walls thrown up by the inhabit- ants of Catania in 1669; a railroad supereleva- tion only 2-3 m high in 1955 near the village of Kapoho on the island of Hawaii) and the observation of the behaviour of lava flows in correspondence to natural morphologic obsta- cles have long inspired the idea of using artifi- cial barriers to guide the course of lava flows. In 1955, small dikes were hurriedly erected to try to divert a flow of Kilauea: the success was " l imited" (Macdonald, 1975). In 1960, near the village of Kapoho, a natural rise, running almost orthogonally to the direction of the lava flow, was increased in height to protect a light- house of great importance. Six 150 to 2295 m long dikes were built, between 1.5 and 7.6 m in height and with a max imum width of the base of 22-27 m (Macdonald, 1962). The lava passed over the obstacle but an important part of the flow was diverted laterally and the light- house was saved. Also in Japan, in the 1970s, a preventive intervention was carried out by closing a breach on the wall o fOsh ima caldera, but the lava did not arrive there. In 1937 and 1950, it was proposed to build barriers in ad- vance to protect the city of Hilo, the capital of the island of Hawaii; so far, however, these barriers have not been built. In commenting on
the encouraging results achieved in 1955 and 1960 in Hawaii, Macdonald ( 1975 ) expressed the doubt that the diversion barriers could work only for very fluid and thin lavas such as the hawaiian "pahoehoe" ones, whereas they could prove to be ineffective for the more vis- cous lavas of the "aa" or "block" type of many continental volcanoes, among which Etna. Moreover, he expressed the concern that if the barrier does not have a very wide base and is not composed of material denser than the liq- uid lava that it must contain or divert, it could be broken through and carried away by the thrust of the lava flow.
In reality, already at Catania in 1669 the walls, 18 m high, withstood the thrust of the lava very well: they were overflowed but not carried away ( Lyell, 1875 ). In any case, the ex- perience at Etna in 1983 dispelled all these doubts and preoccupations. Four earthen bar- riers, 580 to 300 m long and 8 to 20 m high (Colombrita, 1984), worked extremely well and kept the lava (one of the usual "'aa" flows of Etna) from spreading out laterally and de- stroying important installations ( Rifugio Sap- ienza; Osservatorio Astroflsico and other buildings nearby). The barriers were con- structed with mostly incoherent, low-density material (scoriae, lapilli, ash t but stood up to the lateral thrust of the flow without any prob- lems. One of the barriers continued to function even after having been totally submerged by the lava, as had already been observed at Kapoho.
Aside from the cited intervention in Hawaii in 1960, up to 1992 only barriers built on the flanks of the lava flow and oriented diagonally (less than 45 ~ ) with respect to its direction had been experimented with ( and very few times ). Their aim was not to stop the lava, but to di- rect it towards zones where the damage would be less. The embankment built in January 1992 at Portella Calanna is the first true dam, ori- ented orthogonally to the direction of the lava flow, built with the aim of slowing its advance. It is the highest (21 m ) construction ever built to control lava. This structure contained the
30 F. BARBERI ET AL
lava for about one month, forcing it to spread out in the large basin ofVal Calanna. Built with low-density filling (earth, scoriae with small blocks of lava stone) it withstood without dif- ficulty the thrust exerted by the flow, which came up against it and accumulated with re- peated thin flows of fluid lava emitted from the ephemeral vents that opened on the fronts of the more powerful "aa" flows. The 1992 expe- rience at Etna teaches, therefore, that it is pos- sible to contain the advance of every type of "basaltic" lava with earthen retaining em- bankments of adequate height. The utility of an intervention of this type depends on the morphological conditions. A large valley with a narrow outlet is required so that the embank- ment does not require excessive dimensions and costs and the basin uphill can contain large volumes of lava, preventing the advance of the flow for a long period of time. Less important is the type of material used in constructing the dam and its shape, although a gentle slope is preferable because it favours lateral expansion of the lava and hinders erosion of the dam dur- ing overflow. Both parameters must obey only the laws of engineering, economy and respect for the environment. It is obvious, however, that the success of a retaining dam depends on the duration and the characteristics of the eruption. Like the other types of intervention, the construction of a retaining embankment is aimed only at gaining time: any embankment is destined to be overflowed by the lava if the rate of effusion remains high for a sufficiently long time.
Cooling lava with jets of water
During the 1960 eruption of Kilauea, while a flow was destroying the houses of the village of Kapoho one after the other, the Fire De- partment had the idea of trying to slow the lava by cooling it with jets of water. In this way they managed to control the expansion of the lat- eral margin of the flow for a few hours, buying the time needed to remove the objects from a
few houses and even to move one (Macdon- ald, 1975). A more massive use of water was made in 1973 on the island of Heimay (Ice- land), when seawater was pumped at a rate of 900 litres per second on the flank and front of a lava flow advancing towards the town of Westmannaeyjar, with some success. In 1986, some success was also achieved using jets of water to slow the advance of the lava whose front had arrived just 200 m away from the town of Motomaki on the island of Izu-Osh- ima in Japan (Shimozuru, 1988).
Also on Etna in 1983, for a few hours the lava was prevented from overflowing laterally from the lava channel into the yard by sprinkling the approaching lava with jets of water pumped from tank trucks of the Fire Brigade. It was ob- served that a modest sprinkling of water on the incandescent lava forms a sufficiently solid crust to hold back the lava that swelled in the channel. In this way a wall of solid lava was gradually made to grow above the previous dike. The lava overflowed into the yard only when the water in the tank truck finished.
From all these experiences one can conclude that spraying water can be useful in some cases if it is directed at the front and a flank of a flow to try to make it easier for the lava to advance in the opposite direction. Acting only on the front is useless and even dangerous because it may favour tunneling of the lava or uncon- trollable overlappings. Also in this case it is necessary to intervene in morphologically fa- vourable zones for guiding the lava towards the desired path, which in any case cannot be very different from the natural one. It is also ob- vious that easy access to large amounts of water is required. At any rate these interventions, as in 1960 in Hawaii, can only help to buy a little time and can therefore only be useful in some critical situations (emptying a building, etc. ).
Interruption of the lava flow
Along with containment or diversion bar- riers, interventions aimed at bringing the fronts
r i t e CONTROL OF LAVA FLOW, 1991-1992 ERUPTION OF MI. ETNA ~ I
of a lava flow to a halt by cutting off or reduc- ing its supply upstream are the ones that offer the best prospects of success. Among other things, like retaining embankments, they offer the advantage of limiting to the maximum the invasion of new terrains by the lava, thus avoiding delicate legal problems, compensa- tion for damage, etc. So far there have been very few interventions of this type. There is the famous historical precedent of the inhabitants of Catania who, in 1669, while using rudimen- tary means to open a breach in the flank of the flow advancing towards their town, were forced to flee by the enraged inhabitants of nearby villages who feared that the diverted lava might threaten their homes (Gemmellaro, 1858: Lyell, 1875 ).
Since then, until the interventions on Etna in 1983 and 1992, there have been only two attempts in Hawaii to block a tunnel by bomb- ing its roof (1935) and to break, again by bombing, the levee of a channelled flow ( 1942 ). According to Macdonald ( 1975 ), the results were "encouraging but inconclusive"'. According to Lockwood and Torgerson (1980) the bombing did not produce any "significant effect". Again in Hawaii, in 1975 and 1976, a few experiments were conducted by bombing a scoria cone, the flank of a lava channel and the roof of a tunnel produced by a prehistoric eruption of Mauna Loa. Bombs weighing up to 900 kg were used. The roof of the tunnel was barely affected; the flank of the channel was destroyed to an appreciable extent only where it was thin; in the scoria cone, instead, the bombs opened up wide craters (Lockwood and Torgerson, 1980).
The use of bombs or missiles, or artillery pieces, is reproposed at Etna in a more or less confused fashion each time that the necessity of intervening to interrupt the flow of a stream of lava comes up. However, the officers and pilots of the Air Force and the artillery experts of the Army, expressly consulted, declared that they cannot ensure the precision required in the case of Etna. Both in 1983 and in 1992, it was
not a matter only of interrupting the lava flow, but the lava ought to be made to overflow from just one side of the natural channel (in both cases, the right), avoiding flow from the op- posite side, where serious problems would have arised (possible destruction of the zone of Ri- fugio Sapienza in 1983; possible trajectory to- wards the village of Milo in 1992). In the course of the 1992 emergency, bombing the lava channel in the upper part of Valle del Bove was therefore left, with the maximum discre- tion, as a last, desperate intervention, to be at- tempted only in the event that the lava had reached the outskirts of Zafferana, other means of stopping it having failed.
Various lessons are learned in comparing the Etna experiences of 1983 and 1992. First of all, the technique of placing and using explosives was much improved in 1992. By putting the charges directly against the solid septum to be knocked down and not, as in 1983. in holes made in the levee, the following advantages are obtained: the time necessary to drill the holes is saved; the problem of avoiding heating of the charges is solved much more easily: the flow regime is not perturbed at the point of inter- vention (narrowing of the channel due to the cooling of the thinned, drilled wall was lhe cause of the biggest problems in 1983): a greater amount of explosives can be used: the working conditions are safer.
On the other hand, the 1992 experience clearly indicates that trying to block a lava tun- nel by introducing solid material into it pro- duces only short-term benefits. In each of the interventions carried out before the last two, in which an artificial channel for the lava ex- isted, the tunnel was always partially blocked and some results were obtained, such as o~ er- flow of the lava near the intervention points, the opening of ephemeral vents downhill or in- crease of the flow from already existing vents, with the consequent halt or slowing of the fronts. However, they were always benefits of brief duration, at the most two weeks of respite.
Hence, as in the conclusive interventiop of
32 r:, B A R B E R I ET Al
May 27, it is also necessary to dig an artificial channel with an adequate section, deeper than the natural bed of the lava flow, if possible with a diversion of only a few degrees from the nat- ural direction of the flow, and with a steep slope that favours efficient drainage of the diverted lava. It follows that the use of suitable power- ful mechanical equipment is absolutely indis- pensable for the success of operations of this kind. The existence of a morphology favour- able for the runoff of the lava that one tries to make flow out of the natural channel repre- sents an indispensable condition for the suc- cess of the operations. The intervention point must therefore be chosen with great care. One very delicate problem is that of the prepara- tion of the solid-separating septum between the natural channel and the artificial channel, that is, the solid wall that must be blown up with the explosive charges. In 1983 this septum turned out to be too thick (from 3.5 to 6 m) , partly because the abovesaid cooling phenom- ena produced by the work had increased its thickness. On the other hand, the preoccupa- tion of thinning the wall to the maximum caused, on May 22, 1992, it to be broken through by the lava before completion of the artificial channel. These problems arise from the difficulty of estimating with precision the dimensions and the shape of the tunnelled lava channel when it is only seen from a small skylight.
Finally, it must be underscored that, al- though the mechanical equipment was funda- mental for the success of the 1992 operation, the use of explosives is indispensable in this type of intervention, since only with an explo- sion is it possible to suddenly open a breach of large dimensions on the edge of the lava chan- nel and make a large volume of lava run off into the artificial channel, thus avoiding the lava cooling immediately and plugging up the gap.
One last reflection regards the time needed to carry out an intervention of this type. On Etna in 1992, between the decision to attempt
to interrupt the lava flow (April 10) and the last, decisive intervention (May 27), 47 days passed. Of these days 22 were lost due to me- chanical breakdowns (3), bad weather (15), and interruptions of the work for reasons of safety or appropriateness on a couple of holi- days (4). But only 7 days passed between the decision to bring the second, decisive earth- mover into action and the final result. It can therefore be concluded that an intervention such as that of May 27 can be accomplished on Etna in about ten days, including supply of the material and unforeseen events, the duration depending to a considerable extent on the con- ditions of access to the point of intervention.
Final remarks
On March 14, the lava front arrived at Por- tella Calanna, 2 km from Zafferana Etnea. From then to May 27, the eruptive vents emit- ted 240 million m 3 of lava. In the course of the eruption the lava flow always displayed great ease in forming tunnels, up to over 8 km away from the vents, even in the narrow valley that separates Zafferana from Portella Calanna, maintaining high temperature and fluidity and a resultingly high capacity of advance. Despite the retaining earthworks and the interventions uphill carried out up to May 22, on two occa- sions the lava front arrived just 700-800 m away from the outskirts of Zafferana. It is therefore evident that without these interven- tions the lava would have reached Zafferana and would have caused extremely heavy ma- terial damage. The damage produced by the eruption consisted in the destruction of the water collection system of the Commune of Zafferana, and of a few isolated houses, mostly rural; in the devastation of the orchards of Val Calanna and of others in the zone of Piano dell'Acqua, and in the destruction of the access road to Val Calanna. Damage of some conse- quences, estimated at some 3 million USD, but hardly anything compared to the damage the lava would have been caused if it had sub-
THE (- ONTROL OF I.AVA FLOW, 1991 - 1992 ERU P I I O N OF M1. ETN A 3 3
merged the town. Also in terms of civil def- ence, therefore, the 1992 operat ion represents a big success.
The involvement of the Military Forces in the operation proved an excellent decision. They proved to possess an excellent technical expertise and an outstanding operation capa- bility. They were gratified by being involved in a succesfull civil defense peace mission that drew a lot at attention within and outside Italy. Under these conditions, it is difficult to esti- mate the cost of the interventions considering that military personnel and helicopters would have been employed in any case for less useful routine military operations.
Despite the success, it would be wrong if this experience gave rise to the conviction that it is always possible to intervene to protect towns and villages threatened by lava. The possibili- ties of intervening depend on the location of the eruptive vents, the distance between them and the inhabited areas, the morphology of the zone where the lava flows and above all on its speed of advance, which in turn depends on the slope, the viscosity and the rate of effusion. It would have been difficult, for example, to de- fend Randazzo in 1981 if the first lava flow had taken that direction, considering that the town was only 7 km from the eruptive vents and that the lava initially travelled 5 km in just 4 hours (Chester et al., 1985 ).
Attention must also be called to the fact that an intervention like the effective one of May 27-29 cannot be carried out immediately. In addit ion to favourable morphology, it is nec- essary that the eruptive situation has stabilized and that the lava flow is confined inside a well- developed channel to avoid the intervention being thwarted by a downhill migration of the eruptive vents or by a sudden lateral shift of the flow. Reaching these condit ions may re- quire several days, during which the lava fronts can advance swiftly downhill, causing enor- mous damage if the eruptive vents open at low elevation.
In the most terrible of the historical erup-
tions of Etna, that of 1669, the vents opened at an elevation of about 800 m, near the town of Nicolosi. In just 20 hours the lava reached and destroyed Malpasso (present-day Belpasso) 3.5 km away. Various other inhabited areas were destroyed on the following days, but it took 33 days for the flow to reach the city walls of Catania (Gemmellaro, 1858 ). It is obvious that if a similar eruption were to occur again, there would be no possibility of intervening for the inhabited zone situated in close proximity to the eruptive vents, which would already be devastated anyway, as Nicolosi was in 1669, by the pre-eruptive seismic activity. It would also be difficult to intervene to protect the built-up areas located a short distance down- hill: 20 hours is obviously too short a time to set the complex apparatus required for an in- tervention of this type in motion. However, an eccentric eruption, such as the one that began on 12 March 1669, is preceded by several days of anomalous activity (seismicity, ground de- formations, geochemical and geophysical anomalies) which allows the monitoring sys- tem operating on the volcano to identify' the zone where the eruptive vents will soon open and therefore to get an emergency plan under- way before the start of the eruption.
In the first hours of an immediately danger- ous eruption, the only interventions possible are the construction of containment or diver- sion barriers, which also have the aim of buy- ing the time necessary for setting up the inter- ventions to interrupt the lava flow near the vents. However, the construction of barriers that guide the lava toward a less damaging course poses, in densely inhabited zones, grave problems of responsibility and may not meet a favourable consensus among the populations concerned, as was also seen at Zafferana in 1992. Again with reference to the Etna erup- tion of 1669, there is no doubt that it would be possible to intervene to cut off the flow near the eruptive vents in the month of time that the lava took to reach the town of Catania. In this case too, since it would be necessary to op-
34 I:. B A R B E R I E ' I . g L
erate in inhabited areas, there would be the risk that the flow produced by the diversion could destroy new zones. It follows that interven- tions on Etna in the case of eruptions with vents at low elevation inevitably pose serious legal problems and can only be carried out where there is strong political determination which, although seeking popular consensus, enforces the right to choose the lesser damage.
In any case it is important to stress that, if all these difficulties are overcome and the con- ditions for an intervention are created, the ex- perience of 1992 makes it possible to perform this intervention in a short period and with the maximum probability of success through im- mediate application of the technical solutions that proved to be effective after having been refined in one and a half months of tentatives and experimentation.
References
Abersten, L., 1984. Diversion of a lava flow from its nat- ural bed to an artificial channel with the aid of explo- sives: Etna, 1983. Bull. Volcanol., 47-4 (2): 1165- 1177.
Armienti, P., Innocenti, F., Petrini, R., Pompilio, M. and Villari, L., 1989a. Petrology and Sr-Nd isotope geo- chemistry of recent laves from Mt. Etna: bearing on the volcano feeding system. J. Volcanol. Geother. Res.. 39: 315-327.
Armienti, P., Calvari, S. Innocenti, F., Pompilio, M. and Villari U, 1989b. Petrografia e chimismo delle lave dell'Etna del settembre-ottobre 1989. Boll. GNV, 5 (2): 521-532.
Armienti, P., Calvari, S., Innocenti, F., Petrini, R., Pom- pilio, M. and Villari, L., 1990. Petrography and chem- ical composition. In: Mt. Etna: the 1989 Eruption. GNV, Giardini, Pisa, 75 pp..
Armienti, P., D'orazio, M., lnnocenti, F., Petrini, R., Pompilio, M., Tonarini, S. and Villari, L., 1992. Pe- trologia e geochimica delle lave dell'eruzione del 1991- 1992. GNV (unpubl. rep.).
Barberi, F., 1984. Etna: anatomia di un vutcano..\nonc. 38: 69-95.
Barberi, F. and Villari, L. (Editors). 1984. Mr. Etna and its 1983 Eruption. Bull. Volcanol., 4%4 (2): 877. 1177.
Barberi, F., Bertagnini, A. and Landi. P. (Editors ~. 1991i. Mr. Etna: The 1989 Eruption. Giardini, Pisa, 7~ pp,
Chester, D.K., Duncan, A.M., Guest, J.E. and Kilburn. C.R.J., 1985. Mount Etna: The Anatomy ofa V()lcano. Chapman and Hall, London, 404 pp.
Colombrita, R., 1984. Methodology for the construction of earth barriers to divert lava flows: the Mt. Etna t983 eruption. Bull. Volcanol., 47-4 ( 2 ): 1009-1038
Gemmellaro, C., 1858. La vulcanologia dell'Etna, fipo- grafia dell'Accademia Gioenia, Catania, 266 pp..
GNV, 1992. gtna. Smithsonian Institution, Global Vol- canism Network, pp. 12-16.
Hulme, G., 1974. The interpretation of lava flow mor- phology. Geophys. J. R. Astron. Soc., 39: 36t-383.
Lockwood, J.P. and Torgerson, F.A.. 1980. Diversion of lava flows by aerial bombing. Lessons from Mauna Loa volcano, Hawaii. Bull. Volcanol., 43-4: 727- 74 !.
Lopes, R. and Guest, J.E., 1982. Lava flows oll Etna, a morphomelric study. In: The Comparative Stud? of the Planets. D. Reidel, Dordrecht.
kyell, C., 1875. Principles of geology, XII ed. John Mur- ray, London.
Macdonald, G.A., 1962. The 1959 and 1960 eruption of Kilauea volcano, Hawaii and the construction of walls to restrict the spread of the lava flow. Bull. Volcanol.. 2: 24,249-294.
Macdonald, G.A, 1975. Volcanic hazards. In: Geological Hazards. Springer, New YorL NY, 328 pp.
Marsh, B.D., 198t. On the cristallinity, probabilit.~ of oc- currence and theology of lava and magma. Contrib. Mineral. Petrol., 78: 85-98.
Shaw, H.R., 1972. Viscosities of magmatic silicate liq- uids: an empirical method of prediction. Am. J. Sci. 272: 870-893.
Shimozuru, D., 1988. Role of the coordinating commit- tee, hazard mitigation and evacuation. In: The 1986- 1987 Eruption of Izu-Oshima Volcano. Earthquake Res. Inst., Univ. Tokyo, pp. 54-59.
Tazieff, H., 1977. An exceptional eruption: M1. Nira- gongo, January 10th, 1977. Bull. Volcanol., 40(3): 189-200.
Walker, G.P.L., 1978. Lenghts of lava flows. Philos. Trans. R. Soc. London, Set. A, 274:107--118.
