Past variability of tropical cyclones:

1
Past variability of tropical cyclones What can
we learn from paleotempestology? Kam-biu
Liu Department of Oceanography Coastal
Sciences Louisiana State University
NCAR-ECSA-JFF-08, Boulder, Colorado
2

• What is Paleotempestology ?
• Paleotempestology is a young field of science
  that studies past hurricane activities by means
  of geological and archival techniques

• Principal approach
• Detection of storm signal in geological proxy

• Requirements for useful proxy
• instrumentally measurable and individually
  resolvable
• preservable (i.e., not eroded by subsequent
  events)
• datable
• verifiable by modern analog
• attributable solely (or at least principally)
  to storms (i.e., preclusion of other causes)

3
Overwash sand layers in coastal lake- and
marsh-sediments
Detection of overwash events caused by intense
hurricanes
Rodanthe, North Carolina after Hurricane
Isabelle, Sept. 2003
4
Overwash sand layers in coastal lake- and
marsh-sediments
5
Geomorphic proxy Australian beach ridges
(Nott and Hayne, 2001)
6
Other possibilities?

• Non-coastal lake (e.g., Besonen et al., in press)
• Need better understanding of depositional
  mechanism
• Repeatable?
• Hummocky cross-stratification
• Preservability how complete is the record?
• Visibility cannot be easily observed in cores
• Countability can individual storms be
  identified?

7

• Stable isotopic proxies from
• tree rings
• speleothems
• corals ???

Frappier et al., 2007
Miller et al., 2006
Miller et al., 2006
8
Microfossil proxies Foraminifera

• Prices Inlet, north of Charleston, SC
• Non-tidal swamp / saltmarsh
• Hurricane Hugo (9/1989)-
• Offshore/nearshore (displaced) forams

Collins et al., 1999
9
Microfossil Proxies
Phytoliths
Flat- towers
Diatom
Pollen
Uniola paniculata (Sea oats)
10
Despite the absence of clear sedimentary
evidence, microfossils (especially diatoms) can
help reconstruct the occurrence of a storm surge
event by Ivan.
DIATOMS
Thalassiosira excentrica (Marine)
Navicula delawarensis (Brackish)
Pinnularia viridis
(Freshwater)
Coscinodiscus adiatus (Marine)
11
Pollen data place the Ambrosia rise (ca. AD 1800)
at 30 cm, indicating a sedimentation rate of 15
cm per century. This suggests that the second
diatom peak at 6 cm represents Hurricane Frederic
(1979) and the two earlier peaks probably
represent hurricanes of late 1800s and late
1700s.
Marine diatoms
Ambrosia
Foram
WB06-33 core
Depth(cm)
1800 AD
Ambrosia rise
(103/cc)
12
Historical Documentary Data

• Data sources
• Ludlam, D.M. 1963. Early American hurricanes,
  1492-1870. American Meteorological Society.
• Millas, J.C. 1968. Hurricanes of the Caribbean
  and Adjacent Regions, 1492-1800. Academy of the
  Arts and Sciences of the Americas.
• Newspapers
• Plantation diaries
• Instrumental weather records
• Mariners logbooks Royal Navy ships (Public
  Records Office, National Maritime Museum, London)
• Archivo General de Indias (A.G.I., Seville)

Spanish and English historical records
Garcia et al., 2004, 2005
Mock, 2004
13
Historical Documentary Data
Chinese historical records
Jufeng
County gazette
A 1000-yr time-series of Typhoon Landfalls in
Guangdong, China
(Liu et al., 2001)
14
The expanding frontiers of paleotempestology
Liu, 2007
15
What have we learned from paleotempestology about
tropical cyclones and climate change?

• Tropical cyclone activity varies in millennial
  and centennial timescales, in addition to the
  well-documented multidecadal and interannual
  timescales.

16
Decadal-scale variability
North Atlantic SST total power dissipation
index (PDI) (Emanuel, 2005)
Cold AMO vs warm AMO (Goldenberg et al., 2001)
17
16thC
18thC (1766-1780)
Centennial-scale variability
Caribbean basin (Garcia-Herrera et al., 2005)
quiet
Guangdong Provinces, China (Liu et al., 2001)
1880-1910
1830s
Charleston, South Carolina (Mock, 2004)
1660-1680
18
Millennial-scale variability from the Gulf Coast

• Hyperactive period 3800-1000 cal yr BP
• Relatively quiet period 5000-3800 ca. yr BP and
  recent millennium

19
Laguna Playa Grande, Vieques, Puerto Rico
Identification of overwash deposits based on
grain-size analysis
5400 3600 yr BP Active 3600 2500 yr
BP Quiet 2500 1000 yr BP Active 1000 250 yr
BP Quiet 250 yr BP present Active
Donnelly Woodruff, 2007
20
2. Paleotempestology helps to identify climate
mechanisms affecting hurricane activity

• For example
• Bermuda High NAO
• ITCZ
• SST - AMO
• ENSO

Bermuda High Hypothesis (Liu Fearn, 2000)
21
Proxy record from Puerto Rico suggests that
hurricane activity was positively linked to the
strength of the West African monsoon (and
strength of the African easterly jet) and
negatively with the frequency of strong El Nino
events. However, the Puerto Rico record is out
of phase with the Gulf coast record.
Donnelly Woodruff, 2007
22
Climate forcing mechanism for 50-year cycles?
50-yr periodicity for Guangdong typhoon activity
(Liu et al., 2001)
50-yr periodicity for severe droughts in Lake
Huguang Maar, southern China, and in Cariaco
Basin AD 810, 860, 910 (Yancheva et al., 2007)
23
3. Hurricane activity varies spatially as well
as temporally, often in a see-saw pattern between
regions, on decadal to millennial timescales
1660-1680
Hypothesis of southward track shift (Liu et al.,
2001)
Steering tracks are important
24
4. Paleotempestology helps to reveal return
periods of extreme events

• For Gulf Coast sites, return period for cat 4-5
  hurricanes 300 years (Liu, 2004)
• About 9-12 events in 3500 years

Waveland, Mississippi was hit by Camille in 1969
and again by Katrina in 2005
25
5. Paleotempestology helps to reveal worst
(worse)-case scenarios not seen in historical
record
Example
Honduras was not directly hit by any cat 4-5
hurricanes in last 100 years
Proxy record from Mosquito Coast suggests it was
directly hit by catastrophic hurricanes at least
three times during last 700 years, around AD
1290, 1580, 1780
26
Some Research Needs and Priorities

• Expansion of geographical coverage and spatial
  data network

• Basically, we need more proxy records from more
  study sites (paleo-weather stations)
• So far data mainly come from U.S. Gulf Atlantic
  coasts and pan-Caribbean region, and Australia
• Need data from W. and E. Pacific basin Indian
  Ocean

27
2. Use multiple proxies in reconstruction
The expanding frontiers of paleotempestology
(Liu, 2007)

• Different proxies have different temporal
  resolutions and spatial sensitivities
• Coastal sediments a record of landfalls
  intense hurricanes only greater uncertainties in
  age estimates
• Isotopic record (tree rings) coastal
  hurricanes (up to 400 km) annual resolution
• Isotopic luminescence records (corals)
  indirect proxies for AMO, SST, vertical wind
  shear, etc

28
3. Improved calibration of proxy records with
modern analog studies
Example Hurricane Ivan
Landfall (Sept 16, 2004) Gulf Shores,
Alabama Category 3 121 mph wind (54 m/s) Storm
surge flooding 10-16 ft
129 sediment cores 52 from Shelby Lakes 77
from Weeks Bay
Weeks Bay
The Shelby Lakes
29
LS-37
LS-36
LS-G
Lake Shelby No Ivan sand layer at top
0
cm
Loss-on-ignition curves
130
80

0
30
Middle Lake Prominent sand layer attributable to
Ivan
ML-TV2
ML-06
ML-01
31
Little Lake Sand Layer at top attributable to
Ivan
32
What do we learn from modern analog study?

• Absence of Ivan sand layer in Lake Shelby
  supports previous interpretation that prehistoric
  overwash sand layers represent cat 4 cat 5
  hurricanes.
• Presence of Ivan sand layer in Middle Lake and
  Little Lake suggests that these lakes are
  sensitive to upper-cat-3 hurricanes as well.
• Can we independently verify these conclusions?

33
Improved estimates of paleohurricane intensity
from proxy record

• application of statistical methods to calibrate
  proxy record with historical record
• (Elsner et al., 2008)

Lake Shelby
Example Lake Shelby, Alabama

• Lake Shelby proxy record shows 11 events in 3500
  years (Liu and Fearn, 1993)
• Thus return period 318 years
• Modern analog from Ivan suggests that these are
  events of cat 4 or 5 intensity (gt59 m/s).
• Can we independently validate these frequency and
  intensity estimates?

34

• Developing an extreme value model linking return
  period (frequency) and return level (intensity)
  (Elsner et al., 2008)
• Based on historical hurricane record
  (1851-2005)
• Parameter estimates made using maximum
  likelihood procedure

Location map of all tropical cyclones passing
within 495 km of Lake Shelby, 1851-2005 (265
storms in 155 years)
35
Model Results
Cat 1 33-42 m/s Cat 2 43-49 m/s Cat 3
50-58 m/s Cat 4 59-69 m/s Cat 5 gt69 m/s

• Model results suggest an expected return level of
  72.7 m/s (cat 5) for a return period of 318 years
  using a radial distance of 45 km from Shelby
• 95 confidence interval is between 63.8 and 77.5
  m/s (cat 4 5)
• In summary, the model results based on the
  historical record suggest that a hurricane of at
  least 64 m/s (cat 4) is needed to produce a sand
  layer in Lake Shelby (Ivan 54 m/s)
• Model results are consistent with return level
  estimates based on the analog method using Ivan.

r 318 yr
Return level curves as a function of increasing
radial distance (red arrow)
36
5. Integration of proxy data with modeling
results
Example Modeling of Hurricane Bobs storm surge
on New England lakes (Cheung et al., 2007)
Modeled wave field
Modeled maximum wave field
Modeled storm water level
37
Homer Pond, MA
(Cheung et al., 2007)
38

• 6. Improved understanding of sedimentology and
  sedimentary processes
• Distinction between storm deposits and other
  event deposits (e.g., tsunami deposits, tephra,
  floods)
• Better estimates of paleo-hurricane intensity
• e.g., Advective-settling model (Woodruff et al.,
  2008)

Example Complexity of Caribbean cores
Tracking Donna across the Caribbean
39

• Collaboration with scientists in cognate fields
• (Paleo)oceanographers
• e.g., what is the role of the Loop Current in
  affecting paleo-record of landfalling hurricanes
  along Gulf Coast?
• Hurricane dynamics modeling communities
• (Frappier et al., 2007)
• Archeologists (e.g., Gulf Coast Mayan
  societies)
• Ecologists paleoecologists
• (e.g., hurricanes fire interactions)

40
Is there a link between hurricane and fire ?
?
Liu et al., 2008
Hypothesis of hurricane-fire interactions Fire
hazard increases significantly after a major
hurricane strike due to fuel accumulation (dead
biomass).
Charcoal analysis
41

• Hurricane-Fire Interaction
• Dinoflagellate data confirm overwash origin of 7
  sand layers.
• At least 11 charcoal peaks during past 1,300
  years.
• 3 of 4 prominent charcoal peaks occur
  immediately above sand layers.
• Data support the hypothesis that fire hazard
  increases significantly after major hurricane
  strike.

42

• Enhance policy relevance and practical value to
  decision-makers and stakeholders in society
• Rebuilding New Orleans how high should the
  levee be? Cat 3 or cat 5?
• Mosquito Coast of eastern Honduras risk
  assessment and adaptation to future climatic
  changes

Laguna Ibans, Honduras
Interface with social sciences
43

• Conclusions
• Paleotempestology is an interdisciplinary
  science that integrates the geosciences and
  climate sciences
• Paleotempestology provides a long-term
  perspective in understanding the spatial and
  temporal variability of hurricane activity, which
  is vital for identifying climate mechanisms and
  predicting the future
• Paleotempestology offers many opportunities for
  collaboration between natural social sciences
• Paleotempestology is a policy-relevant science
  that has many applications in society.