Catastrophic Hurricane - Louisiana State University

14 downloads 203 Views 3MB Size Report
Louisiana State University, Baton Rouge, LA. Miriam L. Fearn. Department of Geology and Geography,. University of South Alabama, Mobile, AL. Gulf Coast.
~

Holocene History of Catastrophic Hurricane Landfalls along the Gulf of Mexico Coast Reconstructed from Coastal Lake and Marsh Sediments Kam-biu Liu Department of Geography and Anthropology Louisiana State University, Baton Rouge, LA Miriam L. Fearn Department of Geology and Geography, University of South Alabama, Mobile, AL

Introduction The Gulf of Mexico coast is highly vulnerable to catastrophic strikes by landfalling hurricanes. During the pastcentury, over 100hurricanesha\'e made landfalls along the northern Gulf of Mexico coast from the Florida Panhandle to Texas (Neumannet al., 1987). About 44% of these were major hurricanesof category3 or aboveaccording to the Saffir-Simpsonintensity scale. However, only 10were catastrophichurricanesof categof)'4 and 5 intensities, including hurricane Camille of 1969,the only category5 hurricaneto have struck the U.S. mainland in the past century of instrumental record. Thus, landfalls by catastrophic hurricanes, here defined as those of category4 and 5 intensities,are very rare occurrences.In the light of this, severalquestionsare most relevantto scientific issuesof Gulf Coastclimate changes:(1) What are the landfall probabilities (or alternatively, "return periods")* of catastrophichurricanes along the Gulf Coast?(2) Did these landfall probabilities change temporally, and at what time scale?(3) How are these variations in hurricane activity and landfall probabilitiesrelatedto global and regionalclimate changes,suchas those associated with the Little Ice Age and the anticipated global warming? The answersto these questions are important not only becausethey are vital to a better scientific understandingof the patternsand processesof global and regional climate change, but alsobecauseof their practicalvalue in the prediction and assessmentof hurricane risks for the

Gulf Coast. For example, what is the probability for a major metropolitanarea like New Orleansto be directly hit by a category 5 hurricane? The answerto this questionwould be of great importance to socioeconomicplanners, civil engineers, and insurancecompanies. Obviously,these questions cannot be adequately answered by the actuarial approach, which is based on probability estimatesderived from the hurricane landfall statisticsof the last 100years a period too short for estimating the return periods of theserare but most destructivehurricanes. Therefore,geologic data provide the only means for extendingthe observationaldatabaseof hurricane landfalls from the past century to the last severalcenturiesor millennia. Among the geological proxies,those that offer some, though as yet unproven, potential for the reconstruction of paleo-hurricaneactivities, include tempestites in marine and coastallagoonalsediments (Keenand Slingerland, 1993a,1993b; Duke, 1985; Davis et al., 1989),oxygenisotopesin rain water and cave deposits (Lawrence and Gedzelman, 1996; Malmquist, 1997),and pollen in lake and marsh sediments(Gathen,1994;Zhou et al., 1998;Bravo et al., 1997).Recently,Uu and Fearn(1993a)have demonstrated that a chronologically and stratigraphically distinct record of past catastrophic hurricane strikes can be obtained by the detailed stratigraphic study and radiocarbon-dating of overwashdepositsidentified in coastallake sediments.Since 1990,we have used this approachof

38 Section 3.2: Holocene History of Catastrophic Hurricane Landfalls

"'\;r

\ :;7'

\

GA

.;

/..,

1X

~

\.

~

"-..

Fig. 1. Location of lakes and marshes along the northern Gulf of Mexico coast cored for the hurricane paleoclimate study, in relation to the storm tracks of major landfalling hurricanes (categories 3, 4, 5) recorded in this century.

using storm depositsin coastallakes and marshes as a proxy for paleo-hurricanelandfalls in a comprehensivestudy to reconstructthe Holocenehistory of catastrophichurricane activities along the northern Gulf of Mexico coast.So far, more than a dozencoastallakesand marshesfrom Texasto the Florida Panhandlehave beencored for this study (Liu and Fearn, 1993a,1993b,1997,1998;Gathen, 1994; Li, 1994; Zhou et al., 1998) (Fig. 1). In this paper, we will present an overview of our meth~ ods and a summaryof our key findings in the light of the above questions relating to Holocene climate change and long-term variations in hurricane activity and landfall probabilities along the Gulf of Mexico coast.

hurricane eye. Generallyfor the northern Gulf of Mexico coast, the coastal segmentsituated in the northeastern quadrant (i.e., forward right) of a landfalling hurricane usually experiences the greatestdestructiondue to the strongestonshore winds and the highest storm surge potential, whereas the northwestern quadrant tends to receivelessdamagedue to the predominantlyoffshore wind field. Thus, the sediment-stratigraphic proxies recovered from coastal lakes and marshes along the northern Gulf Coast (a predominantly east-westoriented coastline) are primarily sensitive to hurricanes that made landfall within < 100km (the typical radius of hurricaneforce wind) to the west of the site.

1. Coastal Lakes Coastallakes situatedbehind sandybarriers (e.g., barrier beaches,sand dunes, beach ridges) are subject to overwash if the storm surge height caused by a landfalling hurricane exceeds the

--0..,

~

-0..,..

LAKE core

---

-,,-

Rationale and Methods For a given coastalsite during a hurricane strike, the geomorphic impact (e.g., storm surge height) depends on the interplay of a number of factors (Simpson and Riehl, 1981). Among the more important factors are: (1) the intensity, size, and structure of the hurricane; (2) the duration and timing of the hurricane landfall, especiallyin relation to tidal parameters;(3) the angle of hurricane approach in relation to the configuration of the coastline; and (4) the proximity and direction of the site in relation to the location of landfall or the

Fig. 2. A model of sediment stratigraphy in a coastal lake containing multiple sand layers representing overwash events caused by past catastrophic hurricane strikes. These sand layers can be identified in sediment cores taken from different parts of the lake. Cores taken from nearshore sites should contain more and thicker sand layers than those taken farther away from shore. Only sand layers that are the most widespread, probably reflecting the severest overwash caused by the strongest hurricanes, are likely to be recorded in cores taken from the lake center. However, sand layers may coalesce to form a massive sand unit at sites too close to shore, so that individual overwash events may not be distinguishable in the core.

Section 3.2: HoloceneHistory of Catastrophic Hurricane Landfalls 39

-~~~

height of the coastalbarrier. During an overwash event, much sand is transportedinto the lake by waves, currents,and wind. The modelwe use as a working hypothesis(Fig. 2) is that a sand layer will be deposited in the lake bottom when an overwashoccurs due to a hurricane strike. This sand layer, an overwash deposit, can be distinguished from the more organic and finer sediments that accumulate continuously throughout the year. Generally,this sand layer is thickestnear the edgeof the lake and thins out towardsthe center. If other factorsare moreor less equal,impacts by strongeror major hurricanesare more likely to produce a thicker and more extensivesand layer than minor hurricanes and winter storms. Thus, corestakenfrom the centerof the lake shouldcontain fewer, and generally thinner, sand layers, which record impacts only by the most intense hurricanes. On the other hand, cores taken from closerto shore should contain more, and thicker, sand layers, which may provide a more sensitive record of smaller hurricane impacts. However, sedimentsfrom the littoral zone too closeto shore would be more susceptibleto reworking and disturbance by subsequenthurricanes so that the stratigraphy may not be preserved,or the thick sand layers may coalesceto form one massive sand unit so that individual eventsmay not be distinguishable. Moreover, cores taken from the nearshorezone would tend to be less organicand the sedimentaryrecordthereis likely to be shorter or incomplete due to wave disturbance and sediment focusing (Davis and Ford,1982). Therefore,the frequency,extent,and thickness of these sand layers, which can be traced and cross-correlatedin a suite of cores taken in different parts of the lake, can be used to reconstruct a history of paleo-hurricaneactivities. The organicsediments adjacentto thesesandlayers canbe radiocarbon-datedeither by the conventional radiometric method or by the accelerator mass-spectrometricCAMS)method to provide a chronology of these past hurricane landfalls.

2. Coastal Marshes Like their lacustrine counterparts,coastalor estuarine marshesare also highly susceptibleto hurricaneimpacts. Stormsurgescausedby landfalling hurricanestypically result in increasedsedimentation in coastalmarshes,althoughhere the pattern of sedimentation could be more complex and more variable both spatially and temporally than in lacustrine environments (Cahoonet al., 1995; Gunterspergenet al., 1995). In the Mississippi River deltaic plain of Louisiana, where normal riverine flooding contributes very little to marsh accretion, hurricanes may result in significant depositionof clastic sedimentson coastalmarshesdue to the resuspensionof sedimentsfrom shallow bays and estuaries (Rejmaneket al., 1988; Baumannet al., 1984). Remarkably,sand layers have beenreportedto occur in coastalmarsh sediments along the tectonically active Pacific coast of North America where hurricanes are absent; thesesandlayerswere attributed to tsunamistriggered by submarine or nearshore earthquakes (Bensonet al., 1997).It is reasonableto infer that tsunamis and hurricane-driven storm surges would probably have similar sedimentological impacts in coastal marshes, even though these two types of extremeeventsare somewhatdiffer-

Fig. 3. A model of sediment stratigraphy in a coastal marsh containing multiple storm deposits. The storm deposits, in the form of clastic layers embedded in marsh peat or organic-rich sediments, can be identified in sediment cores. Each storm deposit may vary spatially in thickness and in texture and mineral composition due to variations in the direction and source of sediment supply, and in the hydrology and topography of the marsh surface. The environmental impacts of past hurricanes can also be detected by means of biostratigraphic (e.g., pollen, diatom, foraminifera) studies of cores taken along salinity and ecological gradients from fresh marsh to saltmarsh.

40 Section 3.2: Holocene History of Catastrophic Hurricane Landfalls

ent in duration and intensity. The occurrence of tsunamis is highly unlikely, however, along the tectonicallyand seismicallystable Gulf of Mexico coast,so that a non-stormorigin for thesesand or silt layers can be discounted. Along the Gulf Coast, winter storms may also contribute significantly to marsh accretion and clastic deposition (Reed,1989). However,except in some extraordinary cases such as the 1993 "Storm of the Century" (Goodbred and Hine, 1995), winter storms' surgeheights are typically lower and their geomorphic and sedimentological impacts less pronouncedthan those of catastrophichurricanes. Storm depositsattributable to past winter storms have not beenidentified from sub-recentor older marsh sediments(Goodbredand Hine, 1995). Basedon these studies, we may hypothesize that the sedimentologicaland stratigraphic signatures of paleo-hurricanescan be identified from coastal marshesin the form of distinct clastic (clay,silt, or sand, dependingon sediment source) or inorganic layers embeddedin peat deposits. Becauseof the topographic and hydrographic complexity of the marsh $urface in relation to the source and direction of sediment supply during and after a s(orm surge event, this storm deposit may be uneven in thickness or spatially discontinuous (Fig. 3). Hurricaneimpacts may also result in hiatus in the marsh stratigraphy because certain areasof the marsh may be subjectedto significant sediment compaction and lateral compression, wrack deposition, reworking, and erosion (Cahoonet al., 1995; Gunterspergenet al., 1995). Therefore,multiple cores from different parts of the marsh and adequateradiocarbondating control are necessaryfor reconstructinga more complete history of prehistoric hurricane activities. Finally, since storm surgeevents often causedrastic physical, geochemical(e.g., salinity), and ecological changes in the coastal marsh ecosystems (Gunterspergenet al., 1995; Jackson et al., 1995; Conneret al., 1989;Valentine,1977;Chabreckand Palmisano,1973),suchchangesshould be readily detectableby means of pollen and diatom analysesof marsh sedimentstratigraphies (Zhou et al., 1998; Li, 1994; Hemphill-Haley,1996).

Results During the past eight yearswe have retrievedmultiple cores from over a dozen coastal lakes and marshesalong the northern Gulf of Mexico coast from Texasto northwesternFlorida (Fig. 1). The sediment- and bio-stratigraphies of these cores were studied by means of loss-on-ignitionanalysis, pollen analysis, and diatom analysis to help identify storm depositsand overwashsand layers, and to detect any vegetationaland environmental changesassociatedwith such hurricane impacts. Most of these cores were supported by radiocarbon dates, including many AMS dates. The following is a summaryand highlight of the key findings from selectedstudy sites.

1. Lake Shelby, Alabama Lake Shelby,Alabama,was the site of our pioneer study to use overwashdepositsin lake sediments asa proxy for paleo-hurricanereconstruction (Liu and Fearn, 1993a),This lake is ideally situated for testing our model because it is a closed-basin freshwater lake isolated from the Gulf; it is surrounded by systems of beach ridges and sand dunesand separatedfrom the Gulf by a relatively narrow barrier beach;and it was directly struck by a category 3 hurricane (Hurricane Frederic) in 1979 which can provide a "modern analog" for reconstructing the intensity of prehistoric hurricanes. Sedimentcoresretrieved from the center of Lake Shelby contained multiple sand layers that were attribut:ableto catastrophicimpacts by prehistoric hurricanes of category 4 or 5 intensity (Liu and Fearn,1993a).Validationthat thesesandlayersdo indeed representcategory4 or 5 hurricanes came from a study of the geomorphicimpacts and stratigraphic signaturesof Hurricane Frederic (a strong category3 hurricane),whose overwashsand layer was confined to the nearshoresedimentsof Lake Shelby. Based on the radiocarbon chronology of theseolder sandlayers, it was establishedthat catastrophichurricanesdirectly struck Lake Shelbyat ca. 3200yr BP (radiocarbonyearsbeforepresent),

Section3.2: HoloceneHistory of CatastrophicHurricaneLandfalls41

Co.. ...th

"C

"C

DATE E

(em)

OATEA

(YrB.P.1

(yr",)

~s

20. 40-

80SO100120-

140-

2650tSOl

1S01S0. 200-

220240280-

(3730 %100)1

280- 3110%001 300320-

3241)tSOll

340380-

\ 35S0fl001

3S0

860. 880-

\

\

D

900-' 4760t601

Fig. 4. Stratigraphic columns of Lake Shelby cores E, 8, A, S, and L showing the occurrence of overwash sand layers. Core L is a short core taken from a nearshore site where the Hurricane Frederic sand layer was deposited. Core E is a 9 m long core taken from the center of the lake. The most prominent sand layers in these cores were radiocarbondated. The date in brackets was rejected as being too old due to contamination by shell fragments. All dates shown are uncalibrated "C ages (After Liu and Fearn, 1993a). Inset photograph shows an overwash sand layer in core E.

2600 yr BP, 2200 yr BP, 1400 yr BP,and 800 yr BP, or 5 times during the past 3200 years (Uu and Fearn, 1993a) (Fig. 4). This implies a "return period" of approximately 600 years for hurricanes of category 4 or 5 intensity, or a landfall probability of about 0.16% per year for the Alabama coast near Lake Shelby.

(Liu and Fearn, 1993a). Severalless prominent sand layers were not radiocarbon-dated,and several distinct sand layers dated to about 3200-3000 yr BP were conservativelyregardedas one event (Fig. 4). If all the sand layers had been radiocarbon-dated and counted, the frequency of catastrophic hurricane strikes would have been 9 in the last 3200 years; hence a "return period" of approximately350 years or a landfall probability of 0.28% per year.This estimateis not unreasonable, consideringthe fact that no hurricane of category 4 or 5 intensity has made landfall on the Alabama coastduring the past 120yearsof documentary record, eventhough this coastwas directly struck by category3 hurricanes four times-in 1906, 1916, 1926, and 1979 (Neumann et al., 1987). The absence of sand layers in the Lake Shelby cores prior to 3200 yr BP implies a significant environmentalchangeoccurring around that time. The discovery of corn (Zea mays) pollen in the Lake Shelbysedimentsdating to 3500 yr BP suggests that the Lake Shelbyarea had been settled before this environmental change took place (Fearnand Uu, 1995, 1997). This environmental changewas at least regionalin scaleand probably affected the whole Gulf Coast, because similar stratigraphicchangesalso occur in the sediments of WesternLake, Florida,dating to aboutthe same time (Uu and Fearn, 1993a,1997). It is likely that a significant shift in global climate and oceanic circulation occurred during 3500-3200yr BP, as evidenced from broadly synchronous changes documentedfrom various paleoclimaticproxies at different sites around the world (Uu and Fearn, 1993a;Uu et al., 1994).

2. Western Lake, Florida This "return period" derivedfrom the Lake Shelby sedimentaryrecord is the first empirically based WesternLake in the Florida Panhandlehas yieldestimate of landfall probability for catastrophic ed another detailed record of catastrophichurrihurricanes of category4 or 5 intensity. This is cane strikes during the Holocene.Since 1992,we probably a minimum estimate,becauseonly the have raised 16 sedimentcores from this lake for most prominent sand layers in the Lake Shelby our hurricane paleoclimate study. The Western sediment cores were used in the reconstruction Lake area was severely impacted by Hurricane

42 Section3.2: HoloceneHistory of Catastrophic HurricaneLandfalls

~

Opal in October1995 when this category3 hurricane made landfall about 75 km to the west. The Opal sand layer was found only in a core taken closeto the southshoreof the lake (Liu and Fearn, 1997). Using that as a modern analog, we interpret that the prominent sand layers occurring in cores taken farther away from the south shore must have beencausedby prehistoric hurricanes of category 4 or 5 intensity. If only the most prominent sand layers are counted, at least eight such eventsoccurred during the last 3400 years, suggestinga "return period" of approximately400 yearsfor thesecatastrophichurricane strikes.This translatesto a landfall probability of about 0.25% per year (Liu and Fearn,1997). Notably, eventhis is a minimum estimate.If four other distinct but less prominent sand layers in the core are counted, the "return period" would be about 280 years (12 strikes in 3400 years) (Fig. 5). The radiocarbon chronology of the sand layers from WesternLakerevealssignificantvariations in catastrophic hurricane frequencies at millennial timescales.Of the 12 paleo-hurricanestrikes discussedabove,11 occurredduring the two millennia between3400and 1400 yr BP,while only one struck during the past 1400 years (Fig. 5). Sedimentary records from two adjacent lakes (Campbell Lake, Camp Creek Lake), plus new data from other cores taken from Western Lake, seem to suggest that the first millennium AD (2000-1000yr BP) may be one of the "stormiest" periods during the late-Holocene,with multiple strikes on the Florida Panhandle coast by catastrophic hurricanes at an average frequency of once every 200years, or an annual landfall probability of 0.5% (Liu and Fearn, 1998). The notion that the current millennium (1000 yr BP to present)is a relatively quiescentepisodein catastrophic hurricane activities has significant scientific as well as practical implications. What caused such millennial-scale variations in catastrophic hurricane activities? What mechanisms triggeredthe switch from a hyperactiveto a quiescenthurricane regime,and vice-versa?How are these apparent"mega-cycles"relatedto globalcli-

matic changes?Answers to these questions are of utmost importance to the scientific community, risk managers,insurance companies, and socioeconomicplanners.

3. Horn Island, Mississippi Horn Island ,is one of the barrier islands situated offshore from the coast of Mississippi. The lakes on Horn Island are very sensitive to catastrophic hurricane strikes becausethe barrier island forms the "frontline" to the Mississippi coast against landfalling hurricanes. The beaches and sand dunes on the island were completely overwashed by the storm surgeassociatedwith an intense hurricane strike, such as the one by Hurricane Camille (the only category5 hurricane instrumentally recorded to have directly struck the u.s. coast) in 1969. We cored nine lakes on Horn Island and found a sand layer near the top of each 0, ,

500,

1000-

c:

m '- 1500>.

}>

I

~

Q) 2000-

01

«

2500-

3000-

3500-1

Fig. 5. Abbreviated chronostratigraphic column of Western Lake core I for the past 3500 years showing the radiocarbon-dated ages of the major sand layers (solid lines) and minor sand layers (stippled lines). The vertical axis is time measured in uncalibrated 14Cyears before present. At least 12 sand layers occur within the past 3400 years.

Section3.2: HoloceneHistory of CatastrophicHurricaneLandfalls43

core that was datedto the Camillestrike by means of the Cesium-137OOCs) dating method (Gathen, 1994). Radiocarbondating of the sedimentcores shows that the lakeswere formed about 600years ago. The sedimentrecord from FearnLake seems to suggestthat few intensehurricanesstruck Horn Island between240and 600 yr BP-a time coincident with part of the Little Ice Age, a climatic episode of global cooling (Grove, 1988). Pollen analysis of a core from Momma's Diner Lake revealsthat the lake environmentchangesepisodically from fresh to brackish in a way consistent with repeatedsaltwaterintrusionscausedby overwash events associatedwith catastrophichurricane strikes (Gathen,1994).

4. Pearl River Marsh, Louisiana/Mississippi

approximatelyonce evef)' 450 years.The landfall probability, about 0.22% per year, is consistent with the estimatesderived from the Lake Shelby and WesternLake records.

S. Atchafalaya Marsh, Louisiana The AtchafalayaMarsh in southernLouisianawas severelyimpactedby Hurricane Andrew in August

Pearl River Marsh Depth

(cm) 14CDates

50

1240:1:60-

The PearlRiverMarsh is an estuarinemarshsituated along the border between Louisiana and Mississippi. In August 1969 when Hurricane Camille (a category 5 hurricane) made landfall near PassChristianon the Mississippicoastabout 16 km to the east, the Pearl River Marsh was affected by the 6 m storm surgerecordednearthe hurricane eye, althoughthe hurricane-forcewind was predominantlyoffshore. We have identified the Camille-induced overwash sand layer in a transectof corestakenfrom the BuccaneerMarsh, Mississippi, situated directly under the path of Hurricane Camille at landfall (Uu and Fearn, 1993b). In the Pearl River Marsh, Camille left a clay-rich layer in a core taken at a site about 4 miles inland from the coast (U, 1994). The 8.5m core, however,contains other distinct clay or silt layers that were probablycausedby storm surges associatedwith prehistoric hurricane strikes. At least 9 suchlayersare presentin the upperhalf of the sedimentcore spanning the last 4000 years (Fig. 6). If these clay or silt layerswere all formed by catastrophichurricanescomparableto Camille in terms of storm surgeheight and sedimentological impacts, as it seems likely, then the record implies that catastrophichurricanesof category4 or 5 intensity directly struckthe PearlRiver Marsh

OJ !

1001502002503003504004490=60-

450500550600650700-

5830:1:90.

750800~ 850J

Fig. 6. Loss-on-ignition curves for a core from the Pear! River Marsh, Louisiana, showing the organic matter contents (dark) and water contents (stippled) of the sediments. The occurrence of clastic layers (storm deposits), denoted by arrows, is reflected by an abrupt drop in both water contents and organic matter contents. At least 9 such clastic layers occur within the last 4000 years.

44 Section3.2: HoloceneHistory ofCatastrophicHurricaneLandfalls

1992 when this category 3 hurricane passed immediately to the west of this coastal marsh

before making landfall near Morgan City, Louisiana. In a series of shortcorestakenalong a transectfrom the fresh marshto the saltmarsh,we found a 5-16 cm thick sand or silt layer at or near the top that is attributable to the impact of Hurricane Andrew and the 1-2 in stormsurgethat it caused,thus confirming earlier findings reported by Cahoonet al. (1995)and Gunterspergenet al. (1995) on Andrew's sedimentologicaland ecological impacts. Data from pollen and diatom analysessuggestthat this storm depositwas probably derived from the estuarineand bay-bottom sedimentsin the AtchafalayaBay that were resuspended and redepositedby waves and currents during the storm surge. Long cores taken from two saltmarshsites in the southernmostpart of the AtchafalayaMarsh permit reconstructionof a millennial history of catastrophic hurricane strikes in southernLouisiana. While this work is still in progress (Zhou et al., 1998),preliminary resultsfrom a 2.85m corenear Goreausuggestthat intensehurricanesequalto or exceedingAndrew in intensity (i.e., categories3, 4, 5) directly struck the AtchafalayaMarshat least 9 times during the past 1200years.This implies a "return period" of 133 years, or a landfall probability of about 0.75 % per year.Stratigraphicwork is being conductedon another7.9 m long core to confirm this preliminary reconstruction.

Conclusions By using overwashsand layers and storm surge deposits in coastallake and marsh sedimentsas proxies of paleo-hurricanestrikes,we were ableto extend the record of catastrophichurricane landfalls from the past century to the last severalmillennia. Thus the geologicalrecord offers the only means by which empirically based,realisticlandfall probabilities can be estimatedfor these rare but most destructive hurricanes. Our data from Lake Shelbyand WesternLake suggestthat these catastrophic hurricanes typically have "return periods" of 600 to 300 years, with the former

undoubtedly being a minimum estimate. If these estimatesare representative of places along the northern Gulf of Mexico coast-an assumptionto be tested with proxy records from more sitesthen the annual.landfall probabilities for the catastrophic hurricanes are in the order of 0.16 0.33%. One of the most important findings of our studies is that there are long-term variations in catastrophic hurricane activities at millennial timescales. The inter-millennial variability is superimposed on the inter-decadal and interannual variability that has been documented basedon the instrumental hurricane record of this century (Gray, 1990; Landsea et al., 1996). Data from WesternLake and adjacentlakes suggestthat catastrophic hurricanes struck the Florida Panhandlemuch more frequently during the first millennium AD than the last 1000 years. The notion that the current millennium is relatively quiescentin catastrophichurricane activities has important scientific and practical implications. If future global climate changeresults in a return to the hyperactive hurricane regime typical of the first millennium AD, the Gulf of Mexico coast would experiencea dramatic increase in the risk of catastrophichurricane landfalls. Understanding the climatological mechanisms controlling such long-termvariations in hurricane activities is one of the most challengingtasks for climate scientists and global changeresearchers. *Landfall probability is defined as the reciprocal of the "return period" of a specified intensity category of hurricanes. Note that the term "return period" is used here loosely and in quotes to denote the long-term averagein the frequency of hurricane landfalls; it does not imply cyclicity or periodicity in the occurrenceof these events.

Acknowledgments Our hurricane researchhas been supported by the Risk Prediction Initiative (RPI) of the Bermuda Biological Station for Research(RPI-96-01S),and by the NationalScienceFoundation (SES-8922033,

Section3.2: HoloceneHistory of CatastrophicHurricaneLandfalls45

SES-91220S8). We thank Caiming Shen, Xinyu Florida. Journal of Sedimentary Petrology 59: 1052Zhou, Zuju Yao, Don Cahoon, and Stephen 1061. Fournetfor assistancein the field and in the laboratory. Over the years many graduateand under- Duke, W. L., 1985. Hummocky cross-stratification, tropical hurricanes, and intense winter storms. graduatestudentshelped in the field work at var- Sedimentology 32: 167-194. ious locations. We also thank the staff of Grayton Beach State Park, Florida, and Gulf State Park, Fearn,M.L. and Uu, K-b., 1995.Maize pollen of 3500 Alabama, for logistical support and field assis- B.P. from southern Alabama. American Antiquity 60: 109-117. tance.

Cited References Baumann, R.H., Day, l.W., and Miller, C.A., 1984. Mississippi deltaic wetland survival: sedimentation versus coastal submergence.Science224: 1093-1095. Benson, B.E., Grimm, K.A., and Clague, J.J., 1997. Tsunami deposits beneath tidal marshes on northwestern Vancouver Island, British Columbia.

QuaternaryResearch48: 192-204. Bravo, J., Donnelly, J.P., Dowling, J., and Webb III, T., 1997. Sedimentary evidence for the 1938 hurricane in southern New England. Preprint volume of the 22nd conference on hurricanes and tropical meteorology, American Meteorological Society, Boston, pp. 395-396. Cahoon, D.R., Reed, D.l., Day, l. W., Steyer, G.D., Boumans, R.M., Lynch, l.C., McNally, D., and Latif, N., 1995. The influence of Hurricane Andrew on sediment distribution in Louisiana coastal marshes. Journal of Coastal Research,Special Issue 21: 280-

294. Chabreck, R.H. and Palmisano, A. W., 1973. The effects of hurricane Camille on the marshes of the Mississippi River delta. Ecology 54: 1118-1123. Conner, W.H., Day, J.W., Baumann, R.H., and Randall, J.M., 1989. Influence of hurricanes on coastal ecosystems along the northern Gulf of Mexico. Wetland Ecology and Management 1: 45-

56. Davis, M.B. and Ford, M.S., 1982. Sedimentfocusing in Mirror Lake, New Hampshire. Limnology and Oceanography 27: 137-150.

.

Davis, R.A., Jr., Knowles, S.C., and Bland, M.J., 1989. Role of hurricanes in the Holocene stratigraphy of estuaries: Examples from the Gulf Coast of

Fearn, M. L. and Uu, K-b., 1997. Identification of maize pollen: Reply to Eubanks, American Antiquity 62: 146-148. Gathen, K., 1994.Ecological impacts of historic and prehistoric hurricanes on Horn Island. Mississippi. M.Sc. thesis, Department of Geography and Anthropology, Louisiana State University, Baton Rouge, 134 p. Goodbred, S. L., Jr. and Hine, A. C., 1995. Coastal storm deposition: Salt-marsh response to a severe extratropical storm, March 1993, west-central

Florida.Geology23: 679-682. Gray, W.M., 1990. Strong association between West African rainfall and U.S. landfall of intense hurricanes. Sdence 249: 1251-1256. Grove,J.M., 1988. The Little Ice Age. Methuen, New York, 498 p. Gunterspergen,G.R., Cahoon,D.R., Grace,J., Steyer, G.D., Fournet, S., Townson, M.A and Foote, A.L., 1995. Disturbance and recovery of the Louisiana coastal marsh landscape from the impacts of Hurricane Andrew. Journal of Coastal Research, SpecialIssue 21: 280-294. Hemphill-Haley, E., 1996. Diatoms as an aid in identifying late-Holocene tsunami deposits. The Holocene 6: 439-448.

Jackson, L.L., Foote,A.L., and Balistrieri, L.S., 1995. Hydrological, geomorphological, and chemical effects of Hurricane Andrew on coastal marshes of Louisiana. Journal of Coastal Research, Special Issue 21: 306-323. Keen,T.R. and Slingeriand, R.L., 1993a. Four stormevent beds and the tropical cyclones that produced them: A numerical hindcast. Journal of Sedimentary Petrology 63: 218-232.

46 Section3.2: HoloceneHistory of CatastrophicHurricaneLandfalls

~

Keen,T.R. and Slingeriand,R.L., 1993b. A numerical study of sediment transport and event bed genesis during 1i'opical Storm Delia. Journal of Geophysical Research98: 4775-4791. Landsea, C.W., Nicholls, N., Gray, W.M., and Avila, L.A., 1996. Downward trends in the frequency of intense Atlantic hurricanes during the past five decades. Geophysical Research Letters 23: 1697-

1700. Lawrence,l.R. and Gedzelman,S.D., 1996. Low stable isotope ratios of tropical cyclone rains. Geophysical ResearchLetters 23: 527-530.

Liu, K-b., Fearn,M. L., and Li, X., 1994. Abrupt climatic change along the northern Gulf of Mexico coast at 3200 yr B.P. Abstracts, Annual Meeting of the Association of American Geographers, San Francisco, March 28 -April 2, 1994. Malmquist, D. L., 1997. Oxygenisotopes in cave stalagmites as a proxy record of past tropical hurricane activity. Preprint volume of the 22nd conference on hurricanes and tropical meteorology, American Meteorological Society, Boston, pp. 393-394. Neumann, C.J., Jarvinen, B.R., and Pike, A.C., 1987.

1tupical Cyclonesof the North Atlantic Ocean, 1871-1986.National Climate Data Center,Ashville.

Li, X., 1994. A 6,200-year environmental history of the Pearl River Marsh, Louisiana. Ph.D. dissertation, Department of Geography and Anthropology, Louisiana State University, Baton Rouge,208 p. Liu, K-b. and Fearn, M. L., 1993a. Lake-sediment record of late Holocene hurricane activities from coastalAlabama. Geology21: 793-796. Liu, K-b. and Fearn, M.L., 1993b. Where in the mud is hurricane Camille? Abstracts, Annual Meeting of the Assodation of American Geographers,Atlanta, April 6-10, 1993. Liu, K-b. and Fearn, M. L., 1997. Lake sediment records of hurricane Opal and prehistoric hurricanes from the Florida Panhandle.Preprint volume of the 22nd conference on hurricanes and tropical meteorology, American Meteorological Society, Boston, pp. 397-398. Liu, K-b. and Fearn,M. L., 1998. Landfall probabilities of prehistoric catastrophic hurricanes estimated from Gulf Coast lake-sediment records. Abstracts, Annual Meeting of the Association of American Geographers,Boston, March 25-29, 1998.

Reed,D.l., 1989. Patterns of sediment deposition in subsiding coastal saltmarshes, Terrbonne Bay, Louisiana: The role of winter storms. Estuaries 12: 222-227. Rejmanek, M., Sasser, C.E., and Peterson, G. W., 1988. Hurricane-induced sediment deposition in a Gulf Coast marsh. Estuarine, Coastal and Shelf Science27: 217-222.

Simpson,R.H. and Riehl, H., 1981.The humcane and its impact. LouisianaState University Press, BatonRouge,398p. Valentine, J.M., 1977. Plant succession after sawgrass mortality in southeastern Louisiana. Annual Proceedings of the Southeastern Assodation Fishel}' and Wildlife Agencies 30: 634-640.

of

Zhou, X., Liu, K-b., Fearn, M.L., Cahoon, D.R., and Fournet, S.C., 1998. A 1,200-yearhistory of intense hurricane strikes in the Atchafalaya Marsh of southern Louisiana. Abstracts, Annual Meeting of the Association of American Geographers, Boston, March 25- 28, 1998.

Section3.2: HoloceneHistory of Catastrophic HurricaneLandfalls47

-930129-61-0

Editorial/production supervisionand interior design: Franklin Press,Inc Photos:EyewireStudios TechnicalEditing: Zhu Hua Ning and KamranK. Abdollahi Printed / Publishedin 2000by Franklin Press,Inc. and GCRCC

ISBN

Preassignedcard number (PCN)is available from the Library of Congress, This book contains information obtained from authentic peer reviewed and highly regardedsources. Reprinted material is quoted with permission,and sourcesare indicated. A wide variety of references are listed. Collaboratingagenciesand institutions and their works are acknowledged.Reasonableefforts have beenmade to publish reliable data and information. However, the authorsand the publisher can not assumeresponsibilityfor the validity of all materialor for the consequencesof their use.This book is designed,edited and printed by Franklin Press,Inc. This book may not be reproducedor transmitted in any form or by any means,electronic or mechanical, without permission in writing from the publisher,Gulf CoastRegionalClimate Change Council (GCRCC).In obtaining permissionrefer to this book by its ISBN1-930129-61-0. Direct all inquiries to GCRCC.U.S. FederalBuilding. BatonRouge.LA 70821-4292 c 2000 by GCRCC No claim to original U.S. Governmentworks InternationalStandardBook Number (ISBN)1-930129-61-0. Printed in the United Statesof America by Franklin Press.Inc. 10 9 8 7 6 5 4 3 2 1

ii

I; '