Long Term Forecasting - Cornell Geological Sciences

Volcanic Hazard
Assessment and Long-Term
Forecasting
EAS 458
Volcanology
Introduction
 Volcanic hazards generally cannot be
eliminated: eruptions cannot be prevented and it
is not practical to permanently move large
populations out of harms way.
 Volcanic hazards can be managed and the risk
mitigated.
 Land use planning can limit settlement in the highest
risk areas.
 Building codes can be modified to reduce the risk of
structural collapse from ash or volcanic earthquakes.
 Structures can constructed to divert lahars, and in
some cases, small pyroclastic and lava flows.
 Early warning systems can be implemented that allow
limited use of high-risk areas.
1
Role of the Volcanologist
 The first steps in this process are identification
and assessment of volcanic hazards, and these
tasks are primarily the responsibility of the
volcanologists.
 In the subsequent efforts to reduce hazards, for
example evacuation route planning, construction
of diversion structures, or implementation of
improved building codes, volcanologists play
only an advisory role.
Case History: Mt. Pelee, Martinique
– How not to do Hazard
Management
 Picturesque Mt. Pelée,
Pelée,
situated on the northern end of
the Caribbean island of
Martinique, killed some 28,000
people in 1902.
 Avoiding this kind of
catastrophe is one of the goals
of modern volcanology (which
really got its start from
Lacroix’
Lacroix’s study of the volcano
after the eruption).
 What happened in 1902
provides an excellent example
of what not to do.
2
Mt. Pelée Volcanic History
 1851-1852: phreatic eruptions
 No juvenile material
 No casualties, damage restricted to
uninhabited region near the vent
 1792 phreatic eruptions
 No juvenile material
 No casualties, damage restricted to
uninhabited region near the vent
 ~1600 AD
 Pyroclastic flows in the direction of
St. Pierre
 St. Pierre founded in 1635
 Summit of Pelée unvegetated
(hence the name – bald mountain)
 ~1300 AD
 Dome formation, pyroclastic flows
 23 magmatic eruptions during
preceding 5000 years (judged from
14 C dates)
Events Preceding the Eruption
 1889 - new fumaroles open in crater Grand Etang just below the
summit
 April, 1902 - fumarole activity intensified; odor noticed in St. Pierre,
6 km away
 Phreatic explosions begin on April 23
 Ash falls on inhabited areas
 Ash consisted entirely of lithic fragments, no juvenile material
 First small earthquakes
 Numerous submarine telegraph cables break, beginning on April 22
 Probably causes by submarine landslides
 May 3
 Intensity of explosions increases, 1-5 cm ash falls on St. Pierre; > 5 cm
on Le Precheur to the north; small amounts over entire island.
 Small earthquakes constant enough to occasionally produce constant
shaking.
 Wide fluctuations in flow of Riviere Blanche that are unrelated to
rainfall during April & early May.
3
First Casualties
 Morning of May 5, large lahar flows down the
Riviere Blanche, destroying the Guerin rum
distillery and killing 23.
 Most likely caused by failure of east wall of Grand
Etang,
Etang, releasing the water in the crater lake into
Riviere Blanche
 3 successive lahars in 3 minutes
 Upon reaching the sea, the first lahar produced
a tsunami of about 3 or 4m, pushing water 30 m
inland at St. Pierre
 No casualties and limited damage from the tsunami
Magmatic Phase
 Blue glow observed above the summit on May 5
 Probably combustion of magmatic gases
 Incandescent blocks of lava observed on the
summit on May 6



Probably marks the emergence of a lava dome
Vulcanian explosions on the evening of May 6
Dome becomes more apparent on the evening of May
7
 Phreatic explosions continue from southern flank
 Produced continuing ash fall at Le Precheur and
occasional ash fall at St. Pierre
4
 May 7
First Pyroclastic Flows
 Soufriere of St. Vincent (2
islands south) erupts,
pyroclastic flows kill 1600;
blasts heard in St. Pierre
 Evening of May 7
 “At 10:30…
10:30…we saw a great
cloud leave the summit and
descend towards Fonds-Coré.
Fonds-Coré.
But it stopped on midslope…
slope…then, almost at the
same time, the same
phenomenon was repeated,
always at the same place.”
place.”
 Morning of May 8
 Several large Vulcanian
explosions that blasted ash
over and beyond St. Pierre.
Societal Background
 What were people and the gov’
gov’t doing during all this?
 Martinique was and is part of metropolitan France (full citizenship,
vote in elections, representation in parliament, etc.)
 St. Pierre, though not the capital, was the largest city on the island
and the cultural and commercial center of the island.
 1901 census lists a population of 26,011
 Mt. Pelée was recognized as a volcano (certainly after 1852), but no
thorough geologic study had been undertaken and certainly no
hazard assessment.
 Events of 1851-1852 produced a false sense of security since those
explosions had done no real harm. People generally expected the
same from the 1902 eruption – at least in the beginning.
 People were concerned about floods, earthquakes, tsunamis, and
lava flows and were bothered by the ash flows and stench of sulfur,
but the threat of pyroclastic flows and surges was not known or
understood.
5
Government Response
 Initially, the government did nothing
 There were no scientists, much less geologists or volcanologists on the
island, but the government did not request any kind of assistance from
Paris.
 The Governor only notified Paris of volcanic activity on May 3.
 Once ash falls began, the governor made frequent visits to the area,




arranged for aid and assisted those wanting to leave the most
affected areas - such as Le Precheur.
Precheur.
People in mountain and coastal villages were encouraged to take
refuge in St. Pierre.
1000-2000 people from neighboring villages likely took refuge in St.
Pierre
Roughly 1000 fled St. Pierre on their own initiative, mostly to Fort du
France - sulfur stench, ash falls, and lahars finally managed to scare
some.
Mayor of St. Pierre requested 30 soldiers to keep order and help
distribute aid. The governor sent them on the morning of May 8.
They never arrived.
Investigative Commission
 The governor appointed a commission of
informed and educated men to assess the
situation in early May.
 The most informed member of this commission was
the high school science teacher in St. Pierre.
 The commission published a report in the local
newspaper on May 7 concluding that St. Pierre
was safe.
 The governor, his wife, and all but one member of the
commission perished in the eruption.
 (Another psychological factor was that Fort du
France was considered unsafe because of the
damage it suffered in an 1839 earthquake).
6
 St. Pierre was destroyed and all




but 2 inhabitants killed at 8:02 AM
(as judged from the loss of
telegraph & telephone
connection).
Witnesses reported seeing an
emerging eruption column, a flash
of light, followed by the sound of a
large explosion (the delay
representing the slower travel of
sound; blast was heard as far
away as Venezuela), and a
“glowing cloud”
cloud” moving toward St.
Pierre.
The blast was supersonic and
ultimately generated a hurricane
force return wind.
Another major eruption on May 20
completed destruction of the city
(though there was essentially no
one left to kill).
What happened?
The Cataclysm
Column Collapse or Directed Blast?
 General, but not complete,
consensus that what destroyed St.
Pierre is better called a surge and
than a pyroclastic flow.
 Deposit was thin, less than 1 m of
material in much of the city (much
less, depending on interpretation).
 Thickness decreases to south, but
varies irregularly
 Cross bedding abundant
 Variable grading
 Event was energetic enough to
destroy almost all structures,
including the cathedral.
 Victims burned; fires ignited in St.
Pierre burned throughout the day
 Nevertheless, 2 very different
interpretations of what happened
have been put forward.
Rue Victor Hugo
7
Isopach map of Fisher (1982)
Column Collapse Interpretation
 Fisher and others (1982, 1983) believe that
column collapse occurred within a few seconds
of the eruption, the material overtopped the
crater, and generated a pyroclastic flow down
the Riviere Blanche.
 A low density, turbulent ash cloud surge then
segregated from the high density, laminar
pyroclastic flow and moved around and over
topography toward St. Pierre.
 This interpretation is consistent with eyewitness
reports of an eruption column preceding the flow
and the generally agreed small size of the dome
at the time.
8
May 8, 1902 deposits as mapped by Bourdier et al. (1989)
Directed Blast
Interpretation
 Lacroix (1904), Sparks (1983) and
Bourdier et al. (1989) believe that
the event was generated by an
explosion of the admittedly small
dome (gas pressure simply
exceeded its strength); i.e. a
directed blast, Mt. St. Helens style
(though much smaller and without a
debris avalanche).
 Sparks (1983) considers it a high
velocity pyroclastic flow (i.e., high
concentration, laminar flow)
 Bourdier et al. (1989) consider it a
pyroclastic surge (low
concentration, turbulent flow).
 This interpretation depends on
eyewitness reports of a flash and
explosion, on a different
interpretation of stratigraphy, and
on the continuity of stratigraphy
from St. Pierre into Riviere Blanche.
9
Survivors
 Often said that there were only
1 or 2 (or 3) survivors of the
event.
 This is true only with respect
to the actual city of St. Pierre.
 151 people were hospitalized
in Fort du France, of whom
111 recovered.
 Some of these were from the
surrounding countryside - on
the edge of the devastated
area; most were rescued from
ships in the harbor.
 Many others survived for a
few hours; dozens others
were rescued but died en
route to Fort du France.
Louis-August Sylbaris
Aftermath
 Though many were rescued
from the harbor, there was no
organized effort to hunt for
survivors in St. Pierre and the
surrounding countryside (at
least until much later).
 With St. Pierre destroyed,
commerce on the entire island
was disrupted.
 Subsistence economy in the
surrounding countryside
effectively destroyed as ash,
sulfur, and continuing
eruptions made farming
impossible.
 25,000 refugees, with some
15,000 in Fort du France.
10
Government Response
 Paris did send a scientific
team, led by Alfred Lacroix,
Lacroix,
which arrived in late June.
 Refugees initially provided
with subsistence allowance,
but little organization of
housing.
 Unloading and distribution of
international donations of food
and supplies very badly
organized
 In at least one case, custom
officials tried to charge duty on
donated supplies.
Critical Failures
 Lacroix left in early
August, concluding
that the worst was
over.
 Gov’
Gov’t ordered
refugees back to their
homes in early
August.
Fonds Coré before and after
11
August 30 Eruption
 Little activity between mid-July
and mid-August
 Mid-August
 Nuée down the Riviere
Blanche
 New dome growth
 Large explosive eruptions
beginning August 24
 Delegation from Morne Rouge
requests evacuation on August
26; request refused by new
governor
 On August 30, another blast,
less powerful but more widely
distributed, kills >1000 in
Morne Rouge - many of whom
had recently returned.
Epilogue
 St. Pierre was eventually




resettled and partly
rebuilt.
By 1910, it had 500
residents.
By 1923, it had a
population of 5000
Residents fled during the
eruption of 1929-1932,
but St. Pierre was not
damaged.
Today, it is a sleepy town
of 5000 (Fort du France
grew from 22,000 in
1900 to >100,000 in
2000).
12
Lessons
 Full volcanic history needs to be determined - not just
one or two previous eruptions
 All threats need to be evaluated
 Must be done by competent volcanologists
 Need for continued monitoring and evaluation of the
state of the eruption
 Period of quiescence does not mean an eruption is over
 Evacuation and emergency response planning needs to
be carried out ahead of time




Including evacuation routes
Supply routes
Search and rescue plans
Sheltering and sustenance of refugees
2500 years of Mt. Pelée Eruptions
 Modern hazard
assessment studies
(e.g., Westercamp and
Traineau,
Traineau, 1983) have
delineated the typical
kinds of eruptions of Mt.
Pelée and their
frequency.
 These include plinian
eruptions, pyroclastic
flows, directed blasts,
and, in the more distant
past, sector collapse.
Pyro. flow
Pumaceous ash
Cloud deposit
Fine-grained
Surge & blast
fall deposits
(isopachs)
13
Monitoring Mt. Pelée
 Lacroix established a volcano observatory in 1902.
 This functioned until 1925 - after 20 years of quiescence, its was
decided to shut down the observatory.
 4 years later (in 1929), Mt Pelée erupted again.
 This led to the establishment of the present observatory.
 Facilities were upgraded following the eruption of Soufriere of
Guadaloupe (neighboring island, also French) in 1976.
 Today, the Mt. Pelée Volcanological Observatory
(http://volcano.ipgp.jussieu.fr:8080/martinique/stationmar
.html)
.html) is a modern facility that operates telemetered
seismometers, tiltmeters, magnetometers, and cameras
spread around the mountain.
Instrumentation on Mt. Pelée
14
Long-term Forecasting and Repose
Times
 Once the eruptive history of a volcano is known,
and provided the volcano has been reasonably
active in recent past (historic + time accessible
to 14C dating), it is possible make statistical
inferences about future activity.
 To assess eruption probability, Wickman (1966)
proposed a ‘survivor function’
function’, ƒ, defined as:
ƒ(t) = N.d(t)/No.
 Where N.d(t) is number of observed reposes that
lasted longer than time t, and No. is the total number
of observed reposes.
 ƒ(t) is then the probability that a repose has not
ended after time t; i.e., the probability of the volcano
not erupting.
Wickman Diagram
diagram, the log
of N.d(t) (number
of observed
reposes lasting
longer than t) is
plotted against
the length of
repose (t
(t).
Log N.d(t)
 In a Wickman
Length of repose, t
15
Simple Wickman Diagrams
 Popocatepetl
 Linear relationship indicates
repose time is random
 Probability of eruption is
constant (I.e.., repose times
of 1 and 100 yrs equally
likely).
 Hekla (Iceland)
 Compared to Popocatepetl,
Popocatepetl,
Hekla shows enhanced
probability of repose periods
lasting 20-40 years.
 This is interpreted as a
“loading time”
time” - time required
to refill the magma system
after an eruption.
Mt. Pelée Wickman Diagram
 Mt. Pelée exhibits more
complex patterns. This
diagram shows all
eruptions over the last
5000 yrs
 Until 1 century has past,
eruption probability is high
and constant
 After 1 century, probability
drops and remains low for
2 centuries.
 Indicates a pattern of
alternating periods of
intermittent eruption over
a century or so followed
by ~2 centuries of
repose.
16
Mt. Pelée -Pyroclastic Flows
 Adjacent figure is a Wickman
diagram for only those Mt.
Pelée eruptions over the last
5000 years that produced
pyroclastic flows
 It similar to the preceding
diagram, but emphasizes the
pattern of eruption periods
lasting ~125 years
interspersed with ~200 year
repose times.
 Implication for the future:
 Probability of renewed activity
on Mt. Pelée remains high
through ~2055
 After 2055, ~2 centuries of
repose can be expected.
Identifying High Risk Volcanoes
 Some 1500 volcanoes are known to have
erupted in past 10,000 yrs. It is too large
a task to provide thorough hazard
assessment of all. The approach,
therefore, is to identify those presenting
highest risk and focus efforts on those.
 Both Mt. Pelée in Martinique and El Chichon
in Mexico are examples of highly hazardous
volcanoes that were not recognized as such
until after deadly eruptions.
17
High Risk Volcanoes - Yokoyama
Rating System
 To provide an objective means of determining which
volcanoes represented the greatest risk, and were
therefore the most appropriate targets of detailed hazard
assessment, Yokoyama et al. (1984) developed a rating
system. The system was designed to assess both
hazard and risk factors.
 Hazard and risk factors





Type of past activity
Age of last major explosive activity
Size of area affected during past activity
Occurrence of seismic activity or deformation
Size of population at risk and past fatalities.
Yokoyama Rating System- Hazard
Factors
Value
Hazard Factor
1
High silica content of eruptive product
2
Major explosive activity within 5000 yrs
3
Major explosive activity within 500 yrs
4
Pyroclastic flows within 500 yrs
5
Mudflows within 500 years
6
Destructive tsunami within 500 years
7
Area of destruction within 5000 yrs >10km2
8
9
Area of destruction within 5000 yrs >100km2
Frequent volcano-seismic swarms
10
Significant ground deformation within 50 yrs
18
Yokoyama Rating System- Risk
Factors
Value
Risk Factor
1
Population at risk > 100
2
Population at risk > 1000
3
Population at risk > 10,000
4
Population at risk > 100,000
5
Population at risk > 1,000,000
6
Historic fatalities
7
Historic evacuations
Shortcomings
 Prehistoric activity poorly known
 Unprecedented events can occur
 Some volcanoes may not have been
identified
 Long repose time, especially for very
explosive volcanoes
 Low priority volcanoes may be ignored
 Examples: neither Nevado del Ruiz nor
Pinatubo made the list
19
USGS NVEWS Assessment
 Hazard Factors:
 Series of factors for which the value can be 0 or 1
(except recurrence, which can be 0-4)
 Idea is to have many factors, so a missing or wrong
one is not critical
 Exposure (or Risk) Factors
 Mostly also 0 or 1 values
 Relative Threat Ranking = Hazard x Risk
 Current monitoring levels: score of 0-4
 0: no ground based monitoring, 1: minimal, 2: limited,
3: basic real time, 4: well monitored
A Well Monitored Volcano
 Seismic:
Seismic: 12-20 stations within 20 km of vent. Borehole instruments
where practicable.
 Deformation:
Deformation: Routine surveys along with sufficient continuous
stations (GPS, tiltmeters, and/or borehole dilatometers).
 Gas:
Gas: Frequent airborne or campaign gas measurements. Arrays of
continuous sensors.
 Hydrologic:
Hydrologic: Comprehensive database on temperatures and
chemistry of springs and fumaroles, real-time monitoring of hill-slope
soil moisture, stream discharge, etc., as appropriate. AFM systems
for lahar detection where warranted.
 Remote sensing: Regular processing and review of satellite images,
high-resolution thermal-infrared images and frequent, high
resolution, multi-channel visible images. Where practicable,
continuous ground-based thermal imaging and Doppler radar
coverage.
20
A Minimally Monitored Volcano
 Seismic:
 Volcano lies within a regional network; no near-field
stations are in place but at least one station is within
50 km of the volcano. Or, a single near-field station is
present, but no regional network exists.
 Remote Sensing:
 Baseline inventory exists of Landsat-class satellite
images. Routine scans for eruption clouds are
conducted by meteorological agencies.
USGS Hazard Factors
Hazard Factors
Scoring Ranges
Volcano type
0 or 1
Maximum Volcanic Explosivity Index
0 to 3
Explosive activity in past 500 years?
0 or 1
Major explosive activity in past 5000 years?
0 or 1
Eruption recurrence
0 to 4
Holocene pyroclastic flows?
0 or 1
Holocene lahars?
0 or 1
Holocene lava flow?
0 or 1
Hydrothermal explosion potential?
0 or 1
Holocene tsunami?
0 or 1
Sector collapse potential?
0 or 1
Primary lahar source?
0 or 1
Observed seismic activity
0 or 1
Observed ground deformation
0 or 1
Observed fumarolic or magmatic degassing
0 or 1
21
USGS Exposure (Risk) Factors
Log10 of Volcano Population Index (VPI) at 30 km
0 to 5.4
Log10 of approximate population downstream or
downslope
0 to 5.1
Historical fatalities?
0 or 1
Historical evacuations?
Local aviation exposure
0 or 1
0 to 2
Regional aviation exposure
0 to 5.15
Power infrastructure
0 or 1
Transportation infrastructure
0 or 1
Major development or sensitive areas
0 or 1
Volcano is a significant part of a populated island
0 or 1
Assessment Results
 18 very high threat volcanoes, 3 of which
are well monitored
 37 high threat volcanoes, 5 of which have
no monitoring (4 in Marianas, 1 in Alaska),
and 4 of which (including Clear Lake and
Mono Craters in CA) have minimal
monitoring.
22
Priorities for Increased Monitoring
Aviation- Threat
Score
Volcano
Kilauea
State
Threat
Score
Required
Monitoring
Level
Current
Monitoring
Level
Monitoring
Gap
HI
48
324
4
4
Eruption
St. Helens
WA
56
267
4
4
Eruption
Rainier
WA
35
244
4
2
2
Hood
OR
28
213
4
2
2
Shasta
CA
37
210
4
2
2
South Sister
OR
28
194
4
2
2
Lassen
CA
31
186
4
2
2
Mauna Loa
HI
4
170
4
3
Unrest
Redoubt
AK
44
164
4
3
1
Crater Lake
OR
35
161
4
1
3
Baker
WA
14
156
4
2
2
Glacier Peak
WA
35
155
4
1
3
Makushin
AK
34
152
4
3
1
Akutan
AK
42
140
4
3
1
Spurr
AK
44
130
4
3
Unrest
Long Valley Caldera
CA
29
128
4
4
0
Newberry Volcano
OR
28
126
4
2
2
Augustine
AK
44
123
4
3
1
23