Elisa Kagan
Hebrew University of Jerusalem
& the Geological Survey of Israel
July 23, 2008
Hebrew
University
of
Jerusalem
(PhD advisors: Amotz Agnon, Moti Stein, Mira Bar-Matthews)
Paleoseismology is the study of the timing,
location, and size of ancient earthquakes.
San Andeas Fault, California
Interested in knowing:
•Recurrence of earthquakes
•Location
•Magnitude
•Local Intensities, site effects
•Mechanisms
•Segmentation
•Fault interactions
•Directivity
•Etc…………..
TOOLS:
•Instrumental Record
•Historical Record
•Paleoseismic Record
(Faults, deformed sediments
e.g. lake sediments, speleothems)
Instrumental is so precise!
BUT… way too short
Historical…. Quite detailed….
BUT, not totally reliable and also TOO SHORT (up to 2000 years)
Long, detailed, and well-dated paleoseismic record needed
10’s-100’s of thousands of years
 largest quakes may not be included in
historical records
 more seismic cycles
 insight into long-term recurrence times and
patterns (G&R, clustering…)
Surface rupture is recorded in the
landscape and the sediments
Modeling
$$$
Paleoseismic Data
pre-instrumental
caveats
• Site specific
• Data sets CAN be small, sparse, analog
(changing in a continuous manner relative to another quantity )
• Quantification of uncertainty - major challenge
We need:
•Earthquake-induced geological evidence (on-fault or off-fault)
•Preserved evidence
•Accessibility
•Dateable material
•Preferably continuous record
•Preferably multi-site, multi-archive
Different paleoseismic techniques
Fault scarp created by the 1959
Hebgen Lake, Montana,
earthquake
ON-FAULT STUDIES
Trenching across faults
Across Seattle Fault
Bet Zayda (near the Kinneret)
San Andreas, 1600 earthquake
ON-Fault:
•Fault-specific
• Can measure rupture
• Can measure recurrence
•Can differentiate different segments
•Can interpret magnitude
Example Fault Database from California (CDMG)
Need to “trench” each and every one!
Slip Rates (mm/yr)
By Segment
Very detailed information!
Average Recurrence Interval
(years)
At Measurement
Sites
On-Fault not always available
May be covered by soil, alluvium, lake, ocean
This includes basically all subduction
zone quakes (e.g. majority of
devastating tsunami-triggering
earthquakes)
Japan: Fault scarp, hidden deep
within a black spruce thicket...
PRO & CON: Can include evidence of earthquakes from various faults
TECTONIC
SETTING
Off-fault evidence can record earthquakes from various locations and distances
Paleo-tsunami deposits
Chile
Jody Bourgeois
Fallen Boulders
Kiryat Shemona
MSc Mor Kanari
A stream channel offset by the San Andreas fault, Carrizo Plain,
central California (photo by Robert E. Wallace)
Geomorphology
deformed landforms
Dendroseismology – tree-ring analysis, earthquake-damaged trees
New Madrid Seismic Zone -Intraplate
Clastic Dikes
in Lisan Fm., PhD - Zafrir Levy
Nahal Mishmar, Deformed Lake Sediment
Speleoseismology
Soreq Cave (Bet Shemesh), Fallen Stalagmite
Nimrod crusader fortress
offset
Archaeoseismology
Susita
Ateret - Vadum Iacob - N. Wall
Cross correlation of data types:
•Paleoseismicity
•Plate tectonics
•GPS
•Instrumental
Modified Mercalli Intensity Scale
‫עוצמה‬
The real measure of the "badness" of the earthquake
Based on human observations of damage and effects of earthquakes,
not any measurement by a machine.
• Gives a local characteristic of the earthquake at
a site.
• Based on response of people and structures.
• MMI is generally larger near the epicenter of an
earthquake, and decreases with distance.
• However, site effects can cause anomalies in
this trend.
examples:
• IV. Felt indoors by many, outdoors by few.
•
Awakened few, especially light sleepers.
Frightened no one, unless apprehensive from previous experience.
Vibration like that due to the passing of heavy or heavily loaded trucks.
Sensation like heavy body striking building or falling of heavy objects
inside.
Rattling of dishes, windows, doors; glassware and crockery clink
and clash.
Creaking of walls, frame, especially in the upper range of this grade.
Hanging objects swung, in numerous instances.
Slightly disturbed liquids in open vessels. Rocked standing motor cars
noticeably.
• VIII. Fright general -- alarm approaches panic.
•
•
Disturbed persons driving motor cars.
Trees shaken strongly -- branches, trunks, broken off, especially palm
trees.
Ejected sand and mud in small amounts.
Changes: temporary, permanent; in flow of springs and wells; dry wells
renewed flow; in temperature of spring and well waters.
Damage slight in structures (brick) built especially to withstand
earthquakes.
Considerable in ordinary substantial buildings, partial collapse: racked,
tumbled down, wooden houses in some cases; threw out panel walls in
frame structures, broke off decayed piling.
Fall of walls.
Cracked, broke, solid stone walls seriously.
Wet ground to some extent, also ground on steep slopes.
Twisting, fall, of chimneys, columns, monuments, also factory
stacks, towers.
Moved conspicuously, overturned, very heavy furniture.
Modified
Mercalli
Intensity Map
Borah Peak
Earthquake
Oct 28, 1983
Ms=7.3
INQUA SCALE
“A global catalogue and mapping of earthquake environmental effects"
Using the present to interpret the past
Calibrate the scale: modern, measured earthquakes & geological effects
Then: use paleoseismic evidence and calibrate to magnitude etc…
Request: report to them ALL geological effects after an earthquake
Damaged cave
deposits as
paleoseismological
markers
Forti & Postpischl, 1984. Marine Geology
Postpischl et al, 1991. Tectonophysics
Lacave et al., 2004 J. Earthquake Engineering
Kagan et al., 2005. Geology
Gilli, 2005. Comptes Rendus Geoscience
 Seismological studies show enhancement of amplitudes (x6 and
more) may occur at depths (but also at times reduction)
 due to interference of upcoming and downgoing waves (e.g.
Bard and Tucker, 1985)
Site effect is yet unknown
Soreq Caves
Location Map
Eshtaol
N
Soreq Cave
Har-Tuv Cave
Beit-Shemesh
To
Jerusalem
‫* עבודות קודמות בנושא פלאואקלים‪ ,‬קארסט‪ ,‬והידרולוגיה‬
‫במערת שורק‪:‬‬
‫‪Asaf, 1975; Even, 1983; Frumkin et al., 1994; Kaufman et al., 1998; Ayalon et‬‬
‫‪al., 1998,1999,2002; Bar-Matthews et al., 1991,1996,1997,1999,2000,2001, 2002.‬‬
‫•השקעה רציפה ‪<185 kyr‬‬
‫• דמיון רב בין שתי המערות‬
‫• התמוטטויות ותופעות נזק רבות‬
‫• מיקום מאפשר רישום רעידות אדמה מהעתק ים‬
‫ואולי מהעתקים נוספים‬
‫המלח‬
‫• סינון של רעידות קטנות‬
‫איך יודעים שרעידות אדמה‬
‫גרמו לנזקים במערות אלו?‬
‫מה לא גרם לנזקים?‬
‫נזק אנתרופוגני? נזק מבעלי חיים?‬
‫לא!! אין כניסות טבעיות!!‬
‫חציבה רק במאה האחרונה‬
‫תיארוך התופעות פותר את בעיית החציבה‬
‫פרמה‪-‬פרוסט?‬
‫תנועת קרח?‬
‫לא במרכז ישראל!‬
‫תקופות קרח לא היו קרות מספיק‬
‫ולא היה כיסוי קרח‬
‫נהרות תת‪-‬קרקעים?‬
‫השתפלות?‬
‫לא היו במערות המחקר‬
‫אירועים אקלימיים?‬
‫•לא נמצאה קורלציה‬
‫עומס סטטי ?‬
‫• רעידת אדמה תהיה‬
‫ה"טריגר"‬
‫• זקיפים גם נשברו‬
‫מה עשינו?‬
‫* פיתוח השיטה‬
‫* קביעת גילי התמוטטויות‪ ,‬והארכת הרקורד‬
‫הקיים מ‪( ky 70-‬מהליסן) ל‪ky 185-‬‬
‫* קורלציה עם הרקורד הפלאוסייסמי הקיים‬
‫•תרומה לרקורד הפלאוסייסמי של אזור המרוחק מההעתק‬
‫הפעיל‪ ,‬רקורד של הרעידות הגדולות‬
‫• מיפוי‬
‫• דיגום (בעיקר ע"י קידוח גלעינים) של כ‪70-‬‬
‫התמוטטויות‪ ,‬וזיהוי המגעים הפלאוסייסמיים‬
‫• פיענוח‬
‫• תיארוך בשיטת ‪230Th/234U‬‬
‫• השוואה עם מחקרים פלאוסייסמים נוספים‬
‫מיפוי‬
‫כיוונים מעודפים‬
‫של ההתמוטטויות‬
‫‪N‬‬
‫לפני התמוטטות‬
‫אחרי התמוטטות‬
regrowth
regrowth
‫נפול‬
‫נפול‬
‫תקרות ממוטטות‬
‫שכבות של‬
‫התמוטטויות‬
‫במשקע‬
‫זרימה‬
‫(שמהווה את‬
‫רצפת המערה)‬
‫‪After Gilli, 1999.‬‬
Collapse layers in
flowstone
‫שכבות של התמוטטויות‬
‫במשקע זרימה‬
Core in
flowstone
~10cm
Pre-collapse
Post-collapse
‫ לכוד במשקע זרימה‬,‫ נטיף נפול‬:‫חתך‬
‫• תיארוך הלמינות מעל ומתחת למגע הפלאוסייסמי בעזרת איזוטופים‬
‫רדיאואקטיביים ‪ -‬אורניום ותוריום )‪(U/Th‬‬
‫(מדידת האיזוטופים השונים בעזרת מס‪-‬ספקטרומטר)‬
‫דיוק בתיארוך רעידות אדמה‬
‫(או כל אירוע גיאולוגי אחר)‬
‫שגיאה אנליטית (שיטת התיארוך האבסולוטי)‬
‫(שגיאה של‬
‫‪)234U/230Th [MCICPMS]: 1-2 %‬‬
‫שגיאה גיאולוגית‪:‬‬
‫‪ ‬קירבת הדוגמא למגע הפלאוסייסמי‬
‫‪ ‬מספר השנים שהדוגמא מייצגת (תלוי בגודל וקצב השקעה)‬
‫‪ ‬האם הגילים הם "מינימום" או "מקסימום" או שניהם (רווח?)‬
Fallen macaroni stalactites and fallen
ceiling pieces embedded in floor
flowstone lamina
U/Th (Multi Collector) and
d18O dating,
PRE
Z PRE
MC= 53.5 ± 1.1 ky
Y
POST
X
Flowstone
has slow
growth rate
usually
W
V
MC= 82.2 ± 1.6 ky
MC= 108.1 ± 1.7 ky
MC= 129 ± 2.8 ky
Sample SO-57
U2
U1
T
S
POST
POST
PRE
PRE
Fast growth rate
POST
B=40.1 ± 0.2 ka
PRE
C=40.9 ± 1.4 ka
BC
Sample
SO-1-6
‫• דגימה‪ 70 :‬התמוטטויות‪,‬‬
‫• יותר מ‪ 70-‬גילי ‪MCICPMS‬‬
‫• זמן חזרה של בערך ‪ 10,000‬שנה‬
‫שאלות ‪ /‬בעיות פתוחות‬
‫‪.1‬א‪ .‬מהי עוצמת הסף המקומית לגרימת הנזק המתועד ומתוארך במערות?‬
‫‪ -‬פתרון ע"י ניסויים הנדסיים ותצפיות "‪ "LIVE‬של השפעת‬
‫רעידות אדמה עכשוויות‬
‫‪.1‬ב‪ .‬ומכאן מהי המגניטודה המינימאלית הצפויה לגרום אותם נזקים?‬
‫ פתרון ע"י ניסויים‬‫אילו תגובות אתר ישפיעו על העוצמות המקומיות? (איזו מגניטודה תביא לאיזו עוצמה‬
‫מקומית?)‪ ....‬ומכאן מה המגניטודה הנדרשת?‬
‫מהי עוצמת הסף ?‬
‫‪Threshold Intensity‬‬
‫דוג' מצרפת ‪1996‬‬
‫‪M 5.2‬‬
‫נזק במערה ‪ 10‬ק"מ מהמוקד‪ ,‬באזור עוצמה ‪(MSK) VI‬‬
‫בעיקר נטיפי קש שבורים‬
‫כנראה היתה תגובת אתר בעקבות טופוגרפיה‬
‫‪Gilli et al., 1999‬‬
Faulting & Paleoliquefaction in the Lisan Fm.
Marco and Agnon, 1995; Marco et al,., 1996; Agnon et al., 2006
‫בקרקעית נוצר נוף‬
‫מדורגת הרבדה‬
‫שלבים‬
‫ביצירת שכבת‬
‫רסק‬
‫‪Breccia‬‬
‫‪Layer‬‬
‫‪-b‬וניזול במים גלים יצירת‬
‫‪-c‬הרחפה‬
‫‪-c‬הרחפה‬
‫‪ -d‬התרחיף ושקיעת‬
‫‪e‬השקעה המשך ‪-‬‬
U-Th dating
70 000 year record
Longest worldwide at the time
1996, JGR
Different sites show
somewhat different records
Holocene lake sediment paleoseismology
Nahal Ze’elim
Ken-Tor et al., JGR, 2001; Ken-Tor et al., Radiocarbon, 2001
?
31 B.C.
?
? 64 B.C.
‫מחשוף‬- ‫נחל צאלים‬
outcrop
Agnon et al., 2006
Ken-Tor et al., 2001
2 sigma, until 8 meters depth
y = -0.2946x + 472.14
Age BC/AD
-500
0
500
500
600
700
33
175
400
-140
-525
300
1500
C14 calibrated ages
seismites, historical
correlation
-750
Depth, m
200
363
419
100
1000
1456
-1000
660
749
-1500
0
2
R = 0.9829
Linear (C14 calibrated
ages)
800
900
‫ אין היאטוסים‬,"‫ אבל "עמוק‬,‫צאלים גם‬
‫דוקטורט שלי‬
1997 coring campaign
Migowski et al., 2004
‫מבחן לשיטה‬
‫מופיע‬
‫בחתך‬
‫מגניטודה‬
‫מרחק‬
‫נעדר‬
‫בחתך‬
‫מרחק אפיצנטר ק”מ‬
‫‪Agnon et al., 2006‬‬
‫מגניטודה‬
‫במשך מפורט רישום‬
‫שנה אלפים עשרת‬
‫אנו חיים בתקופה פעילה‬
‫‪Migowski et al., 2004‬‬
Identifying the Largest Earthquakes in Lisan Lacustrine Breccias
by Correlation with Cave Seismites and Asphalt-bearing Breccias
‫זיהוי רעידות האדמה החזקות ביותר ברקורד הסייסמיטים בתצורת ליסן‬
‫ע"י קורלציה עם ספליאוסייסמיטים וברקציות המכילות אספלט‬
15,000-75,000 yr BP
Lake Lisan deformed varves
Soreq Cave
deformed
speleothems
Late Pleistocene earthquake history
of Dead Sea Basin and Judea Mt. area
Documented by:
Lake Lisan & stalagmite cave archives
Massada Plain (M1b)
Perazim Valley (PZ)
Nahal Tovlan (NT)
Nahal Tamar (TM)
Nahal Mishmar (MR)
Soreq Cave
Searching for
matching
events in the
different archives
compare seismites
from various types of sediments & locations
Lake
Cave
• Different
number/type/thickness of
seismites
• Different number/type of
seismites
•Location, source distance
•Water depth
•Lithology
•Sediment compaction
•Slope & basin structure
•Location, source distance
•Depth underground
•Size of cave room
•Type of speleothem
Motivation
Paleoseism records.
Normally lucky to find one suitable site
Records and recurrence rates are typically based on one site
Multi-archive study :
 different medium (dif. response to EQ)
 different location (dif. distance to EQ)
different physical conditions (e.g. water depth)
site effects (amplification)
Dead Sea basin,
central Dead Sea Transform
sites
(Modified after Garfunkel et al., 1981)
Site locations
Tovlan
Soreq
Caves
Mishmar
Massada
Perazim
Tamar
Lisan maximum extent (LGM)
Soreq caves
+400 m
60 m
Dead Sea Transform
40 km (filters out smaller events)
+200 m
Sea level
Lake Lisan levels
-200 m
-400 m
LGM ~26 ka
eg: 35 ka
Perazim & Tamar
Massada, Mishmar & Tovlan
consequences for seismite formation
eg: 46 ka
Lacustrine intraclast breccias
SEISMITES
(Marco, Agnon et al. 1995, 1996, 2005;
Ken-Tor et al., 2001; Migowski et al., 2004;
Agnon et al., 2006; Kagan et al., 2006)
Brecciated, homogenated,
folded, faulted
Association of Asphalt Inclusions and Breccias in Lisan
Association of Asphalt
Inclusions and Breccias in
Lisan
• Observed in many sites
• May represent asphalt or oil
discharge into lake before strong
earthquake
•Turbulence after quake may cause
floating asphalt/oil to be trapped in
sediment before oxidation takes
place
Historical accounts of
asphalt floating on Dead Sea
after earthquakes
(Arie Nissenbaum, 1977)
Methods
(1)
Field :
Lacustrine section- detailed description, sampling for dating
and chemical analysis
Cave- core drilling & hand samples for dating and chemical
analysis, description of seismites, spatial analysis
(2)
Chronology :
U-Th on calcite cave deposits and on Lisan aragonite
(MC-ICP-MS at Geological Survey of Israel)
MC-ICP-MS
From these different and distant paleoseismic
sites, three to four events stand out
(~ 10% of total)
RESULTS
Speleoseismite age ranges
sample
Seismite
speleoseismites
pre
post
15
20
25
30
35
40
45
50
1000's of years
70 ka-15 ka
- 27 damaged
speleothems dated
- Define minimum 6
tectonoseismic
events
55
60
65
70
75
Findings
Lisan Lake sediment field work and dating
• Massada: 21 seismites, thinner
seismites
• Perazim: 29 seismite ages
Massada
recalculated, very thick seismites
(data from Marco et al., 2006; ages
recalculated after Haase-Schramm et
al., 2004)
• Tovlan: ONE seismite
• Tamar: small part of section studied
• Mishmar: 2/3 of entire Lisan (10
seismites, in period when PZ has 19
and M1b has 9)
Tovlan
Massada - west
Massada - east
Nahal Tamar
34.8 ka
Mas-6
36.2 ka (detrital
Tas-24 contamination)
34.8 ka
Mas-3
32.6 ka
Tas-22
34.1 ka
Mas-1
RESULTS
38.4 ka (ss)
Mas-2
(preliminary)
Legend
Chronology of
Asphalt in breccias
aad layer
breccia layer
conglomerate
36.2 ka
Mas-4
asphalt
34.9 ka
Tas-20
dating sample
38.7 ka
Mas-5
50 cm
36.8 ka
Tas-21
Schematic diagram of outcrops of asphalt-bearing breccia layers at Massada and
Nahal Tamar, all yielding ages from approximately 33 to 39 ka. Ages given are
isochron ages, except for the one marked ss (single sample).
Massada Section
Additional Gypsum Unit
Top Gypsum
Unit
The White
Cliff
30
27
24
21
Broken Gypsum
Unit
18
15
Gypsum 5
Dating of Massada Lisan site
(multi-sample isochrons)
Almost complete
Torfstein, Kagan, in progress
12
9
Small
Gypsums Unit
6
Three Gypsum
Unit
3
Samra-Lisan
transition
COMPARISON
RESULTS
sample
Speleoseismite age ranges
Soreq
Cave
pre
post
15
20
25
30
35
40
45
50
55
60
65
70
75
1000's of years
Massada
±1 to 1.5
Massada
Massada
15
20
25
30
35
40
45
50
55
Perazim
60
65
70
75 ±1 to 1.5
Perazim
Perazim
HIATUS
15
20
25
30
35Tovlan
40
45
50
55
60
65
70
75
±1.5 ka
a ge k a
15
20
25
30
35
40
45
Tovlan
Tovlan
50
55
60
65
70
Tam ar
Tamar
Tam ar
25
30
35
40
ABS: asphaltbearing seismites
75
seismites
Recurrence Interval
IN LAKE & CAVE
- 3-4 earthquakes show at most sites in 55
kyr
- 14-18 kyr recurrence interval for the largest
events expected for DST
Such long recurrence intervals are rarely reported in the literature, but
probably because such long paleoseismic records have rarely been dated
and most existing ones don’t actually include full seismic cycles.
But according to Marco et al., 1996:
Mean recurrence interval for largest events: M 7.9
M
is 50,000 yrs
M 7.5 is 20,000 yrs
May accommodate slip deficit
Calculation:
Assume Guttenberg Richter
log10N = 2.66 – 0.93 M (from 1983-1993 (Shapira & Shamir, 1994)
According to the Lisan mixed layer record a M 6.3 event will occur once in 1600
yrs and from here M 7.9 is 50,000 yrs and M 7.5 is 20,000 yrs
Summary & Conclusions
1. Differences in records can shed light on how different media and
environmental conditions affect recording of earthquakes
2. Different locations and different media record earthquakes differently but the
large earthquakes show through most medium
3. Asphalt Bearing Seismites may be ancient precursors to large earthquakes
4. Distinctive large earthquakes occurred at central DST at ~ 38-40, 52, 71 ka
5. These are probably the largest earthquakes on the DST
Work in Progress
1. Similar Analyses in Holocene Records
2. Small-scale spatial analyses (on order of meters) of seismite variability
3. Lithological, grain-size analysis
4. Detailed analysis of lake levels correlation to seismite record