Paleotempestology
Paleotempestology is the study of past tropical cyclone activity by means of geological proxies as well as historical documentary records. The term was coined by American meteorologist Kerry Emanuel.
The usual approach in paleotempestology is the identification of deposits left by storms. Most commonly, these are overwash deposits in waterbodies close to the coast; other means are oxygen isotope ratio variations caused by tropical cyclone rainfall in trees or speleothems, and identifying beach ridges kicked up by storm waves. The occurrence rate of tropical cyclones can then be inferred from these deposits and sometimes also their intensity – typically the stronger events are the most easily recognizable ones –, by comparing them to deposits left by historical events.
Paleotempestological research has shown that in the Coast of the Gulf of Mexico and in Australia, the occurrence rate of intense tropical cyclones is about once every few centuries, and there are long-term variations in occurrence which are caused, for example, by shifts in their paths. Common problems in paleotempestology are confounding factors such as tsunami-generated deposits, and the fact that only some parts of the world have been investigated.
Definition and rationale
Paleotempestology is the estimation of tropical cyclone activity with the help of proxy data. The name was coined by Kerry Emanuel of the Massachusetts Institute of Technology; the field has seen increased activity since the 1990s and studies were first carried out in the United States of America on the East Coast.The realisation that one cannot rely solely on historical records to infer past storm activity was a major driving force for the development of paleotempestology. The historical record in many places is too short to properly determine the hazard produced by tropical cyclones, especially the rare very intense ones which at times are undersampled by historical records; in the United States, for example, only about 150 years of record are available, and only a small number of hurricanes classified as category 4 or 5 – the most destructive ones on the Saffir-Simpson scale – have come ashore, making it difficult to estimate the hazard level. Such records may also not be representative for future weather patterns.
Information about past tropical cyclone occurrences can be used to constrain how their occurrences may change in the future, or about how they respond to large-scale climate modes, such as sea surface temperature changes. In general, the origin and behaviour of tropical cyclone systems is poorly understood, and there is concern that human-caused global warming will increase the intensity of tropical cyclones and the frequency of strong events by increasing sea surface temperatures.
Techniques
In general, paleotempestology is a complex field of science that overlaps with other disciplines like climatology and coastal geomorphology. A number of techniques have been used to estimate the past hazards from tropical cyclones. Many of these techniques have also been applied to studying extratropical storms, although research on this field is less advanced than on tropical cyclones.Overwash deposits
deposits in atolls, coastal lakes, marshes or reef flats are the most important paleoclimatological evidence of tropical cyclone strikes. When storms hit these areas, currents and waves can overtop barriers, erode these and other beach structures, and lay down deposits in the water bodies behind barriers. Isolated breaches and especially widespread overtopping of coastal barriers during storms can generate fan-like, layered deposits behind the barrier. Individual layers can be correlated to particular storms in favourable circumstances; in addition they are often separated by a clear boundary from earlier sediments. Such deposits have been observed in North Carolina after Hurricane Isabel in 2003, for example. The intensity and impacts of the tropical cyclone can also be inferred from overwash deposits by comparing the deposits to these formed by known storms and analyzing their lithology. Additionally, thicker sediment layers usually correspond to stronger storm systems. This procedure is not always clear-cut however.Several techniques have been applied to separate out storm overwash deposits from other sediments:
- Compared to the normal sedimentation processes in such places, tropical cyclone deposits are rougher and can be detected with sieving, laser-dependent technologies or x-ray fluorescence techniques.
- In sediment cores, deposits formed by tropical cyclones may be denser due to a larger proportion of mineral content associated with overwashes, which can be detected with x-ray fluorescence techniques.
- They may contain less organic matter than deposits formed through steady sedimentation, which can be detected by combusting the deposits and measuring the resulting mass loss. This and sediment grain sizes are the most common research tools for sediment cores.
- A little used technique is the analysis of organic material in sediment cores; there are characteristic changes in carbon and nitrogen isotope ratios after flooding and the entering of seawater, including a general increase in biological productivity.
- Overwash deposits can contain elements that do not normally occur at the site, such as strontium; this can be detected with x-ray fluorescence techniques.
- Overwash deposits have usually brighter colors than those generated during steady sedimentation.
- Storm surges can transport living structures into such deposits that do not normally occur in these settings. Droughts or the entry of water unrelated to a storm can confound such records, however. Thus, this method is often supplemented with other proxies. The most common living structure employed here are foraminifera, although bivalves, diatoms, dinoflagellates, ostracods and pollen have also been used. Marine foraminifera, however, are not always present in deposits formed by historical storms.
- The absence of tsunamis in the region, as their deposits can usually not be easily distinguished from storm deposits.
- The investigation area should have low biological activity, as bioturbation can otherwise erase evidence of storm deposits. Low biological activity can be found in sites with high salt or low oxygen concentrations.
- A high geomorphic stability of the site.
- High sedimentation rates can facilitate the preservation of storm deposits.
- Tides can destroy layered storm deposits; thus non-tidal waterbodies are ideally used. In tidally active waterbodies, correlations involving various sediment cores can be applied.
Dating and intensity determination
For dating purposes radiometric dating procedures involving carbon-14, cesium-137, and lead-210 are most commonly used, often in combination. Uranium series dating, optically stimulated luminescence, and correlations to human land use can also be used in some places.
Beach ridges
Beach ridges and cheniers form when storm surges, storm waves or tides deposit debris in ridges, with one ridge typically corresponding to one storm. Ridges can be formed by coral rubble where coral reefs lie at the coast, and can contain complicated layer structures, shells, pumice, and gravel. A known example is the ridge that Cyclone Bebe generated on Funafuti atoll in 1971.Beach ridges are common on the deltaic shores of China, and are indicative of increased typhoon activity. They have also been found on the Australian coast facing the Great Barrier Reef and are formed from reworked corals. The height of each ridge appears to correlate with the intensity of the storm that produced it, and thus the intensity of the forming storm can be inferred by numerical modelling and comparison to known storms and known storm surges. Ridges tend to be older the farther inland they are; they can also be dated through optically stimulated luminescence and radiocarbon dating. In addition, no tsunami-generated beach ridges have been observed, and tsunamis are important confounding factors in paleotempestology.
Wind-driven erosion or accumulation can alter the elevation of such ridges, and, in addition, the same ridge can be formed by more than one storm event as has been observed in Australia. Beach ridges can also shift around through non-storm processes after their formation and can form through non-tropical cyclone processes. Sedimentary texture can be used to infer the origin of a ridge from storm surges.
Isotope ratios
Precipitation in tropical cyclones has a characteristic isotope composition with a depletion of heavy oxygen isotopes; carbon and nitrogen isotope data have also been used to infer tropical cyclone activity. Corals can store oxygen isotope ratios which in turn reflect water temperatures, precipitation and evaporation; these in turn can be related to tropical cyclone activity. Fish otoliths and bivalves can also store such records, as can trees where the oxygen isotope ratios of precipitation are reflected in the cellulose of trees, and can be inferred with the help of tree rings. However, confounding factors like natural variation and soil properties also influence oxygen isotope ratios of tree cellulose. For these reasons, only the frequency of storms can be reliably estimated from tree ring isotopic records, not their intensity.Speleothems, deposits formed in caves through the dissolution and redeposition of dolomite and limestone, can store isotope signatures associated with tropical cyclones, especially in fast growing speleothems, areas with thin soils and speleothems which have undergone little alteration. Such deposits have a high temporal resolution, and are also protected from many confounding factors although the extraction of annual layers has become possible only recently, with a two-week resolution achieved in one case. However, the suitability of speleothems depends on the characteristics of the cave they are found in; caves that flood frequently may have their speleothems eroded or otherwise damaged, for example, making them less suitable for paleotempestology research. Caves where speleothems form mainly during the offseason are also likely to miss tropical cyclones. Very old records can be obtained from oxygen isotope ratios in rocks.
Other techniques
Historical documents such as county gazettes in China, diaries, logbooks of travellers, official histories and old newspapers can contain information on tropical cyclones. In China such records go back over a millennium, while elsewhere it is usually confined to the last 130 years. Such historical records however are often ambiguous or unclear. The frequency of shipwrecks has been used to infer past tropical cyclone occurrence, such as has been done with a database of shipwrecks that the Spaniards suffered in the Caribbean.Aside from oxygen isotope ratios, tree rings can also record information on storm-caused plant damage or vegetation changes, such as thin tree rings due to storm-induced damage to a tree canopy, and saltwater intrusion and the resulting slowdown in tree growth. The term "dendrotempestology" is used in this context. Speleothems can also store trace elements which can signal tropical cyclone activity and mud layers formed by storm-induced cave flooding. Droughts on the other hand can cause groundwater levels to drop enough that subsequent storms cannot induce flooding and thus fail to leave a record, as has been noted in Yucatan.
Other techniques:
- Rhythmites in river mouths. These are formed when storms resuspend sediments; the sediments when the storm wanes fall out and form the deposits, especially in places with high sediment supplies. Carbon isotope and chemical data can be used to distinguish them from non-storm sedimentation.
- Sand dunes at coastlines is influenced by storm surge height, and sand splays can be formed when sand is swept off these dunes by storm surges and waves; such deposits however are better studied in the context of tsunamis and there is no clear way to distinguish between tsunami- and storm-formed splays.
- Hummocky deposits in shallow seas, known as tempestites. The mechanics of their formation are still controversial, and such deposits are prone to reworking which wipes out the traces of a storm.
- Boulders and coral blocks can be moved by storms and such moved blocks can potentially be dated to obtain the age of the storm, if certain conditions are met. They can be correlated to storms with the help of oxygen isotope excursions for example. This technique has also been applied to islands formed by storm-moved blocks.
- Wave-driven erosion during storms can create scarps which can be dated with the assistance of optically-stimulated luminescence. Such scarps however tend to be altered over time – later storms can erode away older scarps, for example – and their preservation and formation is often strongly dependent on the local geology.
- Other techniques involve the identification of freshwater flood deposits by storms such as humic acid and other evidence in corals, and lack of bromine – which is common in marine sediments – in flood-related deposits, and oyster bed kills caused by sediments suspended by storms.
- Luminescence of coral deposits has been used to infer tropical cyclone activity.
Timespans
Results
Paleotempestological information has been used by the insurance industry in risk analysis in order to set insurance rates. The industry has also funded paleotempestological research. Paleotempestology information is further of interest to archeologists, ecologists, and forest and water resource managers.Recurrence rates
The recurrence rate, the time gap between storms, is an important metric used to estimate tropical cyclone risk, and it can be determined by paleotempestological research. In the Gulf of Mexico, catastrophic hurricane strikes at given locations occur once about every 350 years in the last 3,800 years or about 0.48%–0.39% annual frequency at any given site, with a recurrence rate of 300 years or 0.33% annual probability at sites in the Caribbean and Gulf of Mexico; category 3 or more storms occur at a rate of 3.9–0.1 category 3 or more storms per century in the northern Gulf of Mexico. Elsewhere, tropical cyclones with intensities of category 4 or more occur about every 350 years in the Pearl River Delta, one storm every 100–150 years at Funafuti and a similar rate in French Polynesia, one category 3 or stronger every 471 years in St. Catherines Island, 0.3% each year for an intense storm in eastern Hainan, one storm every 140–180 years in Nicaragua, one intense storm every 200–300 years in the Great Barrier Reef – formerly their recurrence rate was estimated to be one strong event every few millennia – and one storm of category 2–4 intensity every 190–270 years at Shark Bay in Western Australia. Steady rates have been found for the Gulf of Mexico and the Coral Sea for timespans of several millennia.However, it has also been found that the occurrence rates of tropical cyclone measured with instrumental data over historical time can be significantly different from the actual occurrent rate. In the past, tropical cyclones were far more frequent in the Great Barrier Reef and the northern Gulf of Mexico than today; in Apalachee Bay, strong storms occur every 40 years, not every 400 years as documented historically. Serious storms in New York occurred twice in 300 years not once every millennium or less. In general, the area of Australia appears to be unusually inactive in recent times by the standards of the past 550–1500 years, and the historical record underestimates the incidence of strong storms in Northeastern Australia.
Long term fluctuations
Long-term variations of tropical cyclone activity have also been found. The Gulf of Mexico saw increased activity between 3,800–1,000 years ago with a fivefold increase of category 4–5 hurricane activity, and activity at St. Catherines Island and Wassaw Island was also higher between 2,000 and 1,100 years ago. This appears to be a stage of increased tropical cyclone activity spanning the region from New York to Puerto Rico, while the last 1,000 years have been inactive both there and in the Gulf Coast. Before 1400 AD, the Caribbean and the Gulf of Mexico were active while the East Coast of the United States was inactive, followed by a reversal that lasted until 1675 AD; in an alternative interpretation, the US Atlantic coast and the Caribbean saw low activity between 950 AD and 1700 with a sudden increase around 1700. Such fluctuations appear to mainly concern strong tropical cyclone systems, at least in the Atlantic; weaker systems have a more steady pattern of activity. Rapid fluctuations over short timespans have also been observed.In the Atlantic Ocean, the so-called "Bermuda High" hypothesis stipulates that changes in the position of this anticyclone can cause storm paths to alternate between landfalls on the East Coast and the Gulf Coast but also Nicaragua. Paleotempestological data support this theory although additional findings on Long Island and Puerto Rico have demonstrated that storm frequency is more complex as active periods appear to correlate between the three sites. A southward shift of the High has been inferred to have occurred 3,000–1,000 years ago, and has been linked with the "hurricane hyperactivity" period in the Gulf of Mexico between 3,400–1,000 years ago. Furthermore, a tendency to a more northerly storm track may be associated with a strong North Atlantic Oscillation while the Neoglacial cooling is associated with a southward shift. Atmospheric conditions favourable for tropical cyclone activity in the "main development region" of the Atlantic are correlated to unfavourable conditions along the East Coast. In West Asia, high activity in the South China Sea coincides with low activity in Japan and vice versa.
Role of climate modes
The influence of natural trends on tropical cyclone activity has been recognized in paleotempestology records, such as a correlation between Atlantic hurricane tracks and activity with the status of the ITCZ; position of the Loop Current ; North Atlantic Oscillation; sea surface temperatures and the strength of the West African Monsoon; and Australian cyclone activity and the Pacific Decadal Oscillation. Increased insolation – either from solar activity or from orbital variations – have been found to be detrimental to tropical cyclone activity in some regions. In the first millennium AD, warmer sea surface temperatures in the Atlantic as well as more restricted anomalies may be responsible for stronger regional hurricane activity.Among the known climate modes that influence tropical cyclone activity in paleotempestological records are ENSO phase variations, which influence tropical cyclone activity in Australia and the Atlantic, but also their path as has been noted for typhoons. More general global correlations have been found, such as a negative correlation between tropical cyclone activity in Japan and the North Atlantic and correlation between the Atlantic and Australia on the one hand and between Australia and French Polynesia on the other hand.
Influence of long-term temperature variations
The effect of general climate variations have also been found. Hurricane and typhoon tracks tend to shift north during warm periods and south during cold periods, patterns that might be mediated by shifts in the subtropical anticyclones. These patterns has been observed as a consequence of man-made global warming and the end of the Little Ice Age but also after volcanic eruptions ; some volcanic eruptions have been linked to decreased hurricane activity, although this observation is not universal.During the 1350 to present interval in the Little Ice Age, there were more but weaker storms in the Gulf of Mexico while hurricane activity did not decrease in western Long Island. Increased hurricane activity during the last 300 years in the Caribbean may also correlate to the Little Ice Age.
The response of tropical cyclones to future global warming is of great interest. The Holocene Climatic Optimum did not induce increased tropical cyclone strikes in Queensland and phases of higher hurricane activity on the Gulf Coast are not associated with global warming; however warming has been correlated with typhoon activity in the Gulf of Thailand and marine warming with typhoon activity in the South China Sea, increased hurricane activity in Belize and during the Mesozoic when carbon dioxide caused warming episodes such as the Toarcian anoxic event.
After-effects of tropical cyclones
A correlation between hurricane strikes and subsequent wildfire activity and vegetation changes has been noted in the Alabaman and Cuban paleotempestological record. In St. Catherines Island, cultural activity ceased at the time of increased storm activity, and both Taino settlement of the Bahamas and Polynesian expansion across the Pacific may have been correlated to decreased tropical cyclone activity. Tropical cyclone induced alteration in oxygen isotope ratios may mask isotope ratio variations caused by other climate phenomena, which may thus be misinterpreted.On the other hand, the Classic Maya collapse may coincide with, and have been caused by, a decrease in tropical cyclone activity, because tropical cyclones are more important for preventing droughts in the southeastern US.
Problems
Paleotempestological reconstructions are subject to a number of limitations, including the presence of sites suited for the obtainment of paleotempestological records, changes in the hydrological properties of the site due to e.g. sea level rise which increases the sensitivity to weaker storms and "false positives" caused by for example non-tropical cyclone-related floods, sediment winnowing, wind-driven transport, tides, tsunamis, bioturbation and non-tropical storms such as nor'easters or winter storm, the latter of which however usually result in lower surges. In particular, tsunamis are a problem for paleotempestological studies in the Indian and Pacific Ocean; one technique that has been used to differentiate the two is the identification of traces of runoff which occurs during storms but not during tsunamis.Not all of the world has been investigated with paleotempestological methods; among the places thus researched are Belize, the Carolinas of North America, northern coasts of the Gulf of Mexico, the northeastern United States, the South Pacific islands and tropical Australia. Conversely China, Cuba, Florida, Hispaniola, Honduras, the Lesser Antilles and North America north of Canada are poorly researched. The presence of research institutions active in paleotempestology and suitable sites for paleotempestological research and tropical cyclone landfalls may influence whether a given location is researched or not. In the Atlantic Ocean, research has been concentrated on regions where hurricanes are common rather than more marginal areas.
Paleotempestology records mostly record activity during the Holocene and tend to record mainly catastrophic storms as these are the ones most likely to leave evidence. In addition, as of 2017 there has been little effort in making comprehensive databases of paleotempestological data or in attempting regional reconstructions from local results.
Also, paleotempestological records, especially overwash records in marshes, are often grossly incomplete with questionable geochronology. Deposition mechanism are poorly documented, and it's often not clear how to identify storm deposits. The magnitude of overwash deposits is fundamentally a function of storm surge height, which, however, is not a function of storm intensity. Overwash deposits are regulated by the height of the overwashed barrier and there is no expectation that it will remain stable over time; tropical cyclones themselves have been observed eroding such barriers and such barrier height decreases may induce a spurious increase of tropical cyclone deposits over time. Successive overwash deposits can be difficult to distinguish, and they are easily eroded by subsequent storms. Storm deposits can vary strongly even a short distance from the landfall point, even over few tens of metres, and changes in tropical cyclone activity recorded at one site might simply reflect the stochastic nature of tropical cyclone landfalls.
Application to non-tropical storms
Paleotempestological research has been mostly carried out in low-latitude regions but research in past storm activity has been conducted in the British Isles, France and the Mediterranean. Increases in storm activity on the European Atlantic coast have been noted AD 1350–1650, AD 250–850, AD 950–550, 1550–1350 BC, 3550–3150 BC, and 5750–5150 BC. In southern France, a recurrence rate of 0.2% per year of catastrophic storms has been inferred for the last 2,000 years.Storm records indicate increased storm activity during colder periods such as the Little Ice Age, Medieval Dark Age and Iron Age Cold Epoch. During cold periods, increased temperature gradients between the polar and low-latitude regions increase baroclinic storm activity. Changes in the North Atlantic Oscillation may also play a role.
Examples
| Place | Country/state | Data sources | Duration of the record, in years | Conclusions | Sources | Approximate coordinates |
| Actun Tunichil Muknal | Belize | Oxygen and carbon isotopes in a quickly growing stalagmite | AD 1977 – 2000 | Strong correlation of hits by named tropical cyclones with isotope ratio variations | ||
| Amur Bay | Russia | Sediments from floods | 1,800 | Low storm activity in the last 500 years, probably correlated to the Little Ice Age but continuing into the 19th and 20th century | ||
| Ara River | Japan | River terraces formed by typhoon flooding | 11,600 | Intense flooding during the late glacial to 5,000 – 4,500 years ago indicate increased typhoon activity, followed by a period of less intense activity until about 2,350 years ago | ||
| Barbuda | Antigua and Barbuda | Sediments in a coastal lagoon | 5,000 | Inactive period between 2,500 – 1,500 years, preceded and followed by more active periods | ||
| Belize, central | Belize | Overwash deposits | 500 | 1.2-1 catastrophic storms per century including one very strong storm before 1500AD | ||
| Belize, south-central | Belize | Sediments | 7,000 | Several active periods, between 6,900 – 6,700, 6,050 – 5,750, 5,450 – 4,750, 4,200 – 3,200, 2,600 – 1,450 and 600 – 200 years ago | ||
| Big Pine Key | Florida | Tree ring evidence of storm damage | AD 1700–present | Decreased activity correlated to decreased shipwreck rates in the Maunder Minimum | ||
| Blackwood Sinkhole | Bahamas | Sand deposits in sinkhole | 3,000 | A stage without intense storms between 2,900 – 2,500 years ago, followed by an active period that lasted until 1,000 years ago. Two intense events about 500 years ago and an increase between 300 – 100 years ago | ||
| Brigantine, New Jersey | New Jersey | Sediments | 1,500 | Two strong storms between 600–700 and 700–1,400 AD; nor'easters are also recorded here | ||
| Cenote Chaltun Ha | Yucatan | Mud layers in speleothems | AD 365 – 2007 | Frequent flooding during the 7th, 9th and 19th century with less common flooding during the 13th and 15–17th centuries. Also, evidence of strong tropical cyclone strikes during the Terminal Classic Maya | ||
| Commerce Bight Lagoon | Belize | Sediment cores | 7,000 | Active periods between 600 and 200, 1,450 – 2,600, 3,200 – 4,200, 4,750 – 5,450, 5,750 – 6,050 years ago | ||
| Charlotte Harbor | Florida | Sediments | 8,000 | Increased activity between 3,000 – 2,000 years ago and also during El Nino-leaning periods | ||
| Chenier Plain | Louisiana | Sediments in coastal plain | 600 | 7 hurricanes with category 3 or more intensity are known in the last 600 years, giving a frequency of 1.2 storms per century. Among the storms are Hurricane Audrey and Hurricane Rita | ||
| Chezzetcook Inlet | Nova Scotia | Sediment analysis | 1,000 | Potential storm deposits at 1200 AD, AD 1831 and AD 1848, the middle of which is correlated to a major storm; also an inactive phase in the 1950s and 1970s | ||
| Cowley Beach | Queensland | Beach ridges | 5,740 | Low activity between 1,820 – 850 and 2,580 – 3,230 years ago | ||
| Croatan National Forest | North Carolina | Tree rings | AD 1771 – 2014 | Low activity in 1815–1875 | ||
| Culebrita | Puerto Rico | Sediment deposits | 2,200 | Several sand layers may correlate to hurricanes, including one perhaps linked to the 1867 San Narciso hurricane | ||
| Curacoa Island | Queensland | Beach ridges | 6,000 | 22 hits by intense storms in 6,000 years, implying return periods of 280 years | ||
| Duri Island | South Korea | Shell-gravel deposits | 1,300 | Storms in 720 ± 60, 880 ± 110, 950 ± 70, 995 ± 120 and 1535 ± 40, the latter occurring during the Little Ice Age and the others during the Medieval Climate Anomaly | ||
| Eshaness | British Isles | Boulders perched on cliffs | 1,400 | Probably not tropical cyclones, but intense storm activity occurred since AD 1950, between 1,300–1,900 AD, 700–1,050 AD and 400 – 550 AD | ||
| Exmouth Gulf | Australia, northwestern | Washover fans | 3,000 | Tropical cyclone strikes took place 170 – 180 ± 16, 360 ± 30, 850 – 870 ± 60, 1,290 – 1,300 ± 90, 1,950 – 1,960 ± 90, 2,260 – 2,300 ± 120 and 2,830 – 2,850 ± 120 years ago, consistent with expectations based on sea surface temperature variations | ||
| Falso Bluff Marsh | Nicaragua | Sediment deposits | 5,400 | Last 800 years have an active climate with a return period of about 140–180 years, while between 800–2,800 the return period was only once between 600–2,100 years and another quiet period between 4,900 – 5,400 years ago; between 2,800–4,900 no records | ||
| Folly Island | South Carolina | Back-barrier marshes | 4,600 | The last 4,600 years may have seen 27 storms, as well as 11 major storms in the last 3,300 years | ||
| Frankland Islands | Queensland | Coastal ridges and coral mortality | 510 | Active periods are known from 1980–2000, 1940–1960, 1860–1880, 1800–1830, 1760–1780, 1700–1720, 1630–1650, 1570–1590 | ||
| France | France | Tempestites | Kimmeridgian | Intense tropical cyclone activity from storms coming off the Tethys | Inapplicable | |
| Gales Point | Belize | Sediment cores | 5,500 | In the last 5,500 years 16 major hurricanes | ||
| Grand Case | St. Martin | Sediments | 4,280 | Active period between 3,700 – 1,800 years ago, while 1,800 –800 years ago was inactive | ||
| Great Bahama Bank | Bahamas | Coarse sediment deposits | 7,000 | Active periods occurred within the last 50 years, between 1,200 and 500 years ago, 2,400 – 1,800 years ago and 4,600 – 3,800 years ago, with low activity before 4,400 years | ||
| Great Blue Hole | Belize | Overwash deposits | 1,200 | Active periods between 800 and 500, 1,300 – 900 or 650 – 1,200 years ago and coinciding with the Medieval Warm Period | ||
| Gulf of Carpentaria | Australia | Beach ridges | 7,500 | Low activity/intensity between 5,500–3,500, 2,700–1,800 and 1,000–500 years ago, the former coinciding with the Neoglacial | ||
| Gulf of Thailand | Thailand | Beach ridges and a coastal marsh | 8,000 | 18 typhoon strikes in the last 8,000 years, with increased activity in the mid-Holocene until 3,900 years ago either due to a warmer climate or higher sea level induced better sensitivity to storms | ||
| Hainan Island | China | Deposits in lakes | 350 | 1–2 typhoons per decade, with higher solar activity, positive Pacific Decadal Oscillation, La Nina and positive North Atlantic Oscillation correlating with decreases | ||
| Hainan Island | China | Coastal dunes | 3,400 | 8 storms in 1095 ± 90 BC, 900–1000 BC, 975 ± 50 AD, 1720 ± 20 AD, 1740 ± 35 AD, 1790 ± 25 AD, 1850 ± 15 AD, and 1895 ± 10 AD | ||
| High Atlas | Morocco | Tempestite | Toarcian | Increased tropical cyclone activity during the hot Toarcian Oceanic Anoxic Event | Inapplicable | |
| Ilan Plain | Taiwan | River erosion sediments in a lake | 2,000 | Between 500 – 700 and after AD 1400 intense typhoon rainfall | ||
| Israel | Israel | Oxygen isotope ratios in rocks | Cretaceous-Miocene | Intense tropical cyclone activity in the Tethys until its closure 20 million years ago | Inapplicable | |
| Kamikoshiki-jima | Japan | Sediments in coastal lagoons | 6,400 | Higher typhoon activity at the time of the Kamikaze typhoons, with high activity between 3,600 – 2,500 and between 1,000 – 300 years ago | ||
| Island Bay | Florida | Overwash deposits | 1,000 | 3–4 storms in the last 500 years, 1–2 in 150 – 500 years before present and 11 storms between 1,000 – 500 years ago, all probably major hurricanes; one of the storms in the last 50 years is Hurricane Donna while the other might either be 1926 Miami hurricane, 1910 Cuba hurricane or the 1873 Central Florida Hurricane | ||
| Kimberley | Australia | Flood deposits in stalagmites | 2,200 | Moderate activity between 1,450 – 850 AD and low activity between 500 – 850 and 1,450 – 1,650 AD | ||
| Lady Elliot Island | Queensland | Beach ridges | 3,200 | Strong storms occur every 253 years | ||
| Laguna Alejandro | Dominican Republic | Sediment analysis | 910 | Strikes 910, 800, 730, 530, 500, 330, 260, 210, 200 and 170 years ago | ||
| Laguna Negra | Nicaragua | Deposits in a coastal lake | 8,000 | One very strong storm 3,340 ± 50 years ago, at the same time as increased storm activity in Alabama and Florida | ||
| Laguna Madre | Texas | Storm deposits | 3350 BC–AD 1050 | 0.46% probability of landfall any given year | ||
| Laguna Playa Grande | Puerto Rico | Overwash sediments | 5,000 | 0.48% probability of landfall any given year, but an active period in the last 250 years and previous active periods between 2,500 – 1,000 and 3,600 – 5,400 years ago. El Nino is linked with lower activity, a strong West African Monsoon with higher activity | ||
| Lake Daija | Japan | Sediments in a coastal lagoon | 2,000 | Beginning at 250 AD increased activity, while a quiet period has lasted from 1600 AD to today. Typhoon Jean, Typhoon Grace and others have been identified, including two deposits that may correlate to the Kamikaze typhoons which also coincide within an active period. Recorded storms appear to be of category 3 or higher strength | ||
| Lake Shelby | Alabama | Storm deposits | 4,800 | 11 intense storms between 3,500 and 700 years ago, a quiet period before 3,200 radiocarbon years ago may be either a stage of inactivity or a change in the lake environment. Comparisons to Hurricane Frederic and Hurricane Ivan imply that the intense storms reached category 4 or 5 intensity | ||
| Lake Tiriara | Cook Islands | Minerals from simultaneous seawater intrusion and island erosion | 3,500 | Two storms between 3,200 – 2,800 and 200 years ago | ||
| Lingyang Reef | South China Sea | Storm deposits | 3,500 | Between 3,100 – 1,800 years ago only weak activity, followed and preceded by strong activity; intense storms about once every ten years in the last 3,500 years and the storm activity correlates to sea surface temperatures | ||
| Little Lake | Alabama | Overwash deposits | 1,200 | Seven strikes in 1,200 years, including Hurricane Ivan | ||
| Little Sippewissett Marsh | Massachusetts | Overwash deposits | 400 | Annual landfall probability is about 2.3%, 4% in the last 50 years | ||
| Long Island | New York | Overwash deposits | 3,500 | Increased activity during the Little Ice Age and an inactive period between 900–250 years ago | ||
| Lower Mystic Lake | Massachusetts | Varves formed by post-storm sedimentation | 1000 | Up to eight Category 2–3 hurricanes occurred per century in the 12th to 16th century, while the preceding and the two subsequent ones only saw 2–3 such storms per century | ||
| Mattapoisett Marsh | Massachusetts | Storm inundation deposits | 2,200 | Inactive period between 2,200–1,000 followed by an active period in the last 800 years | ||
| Miaodao | China | Storm deposits | 80,000 | Marine isotope stage 5e storm frequency comparable to that of Holocene low-latitude China | ||
| Mullet Pond | Florida | Sediments in a sinkhole | 4,500 | Active periods with intense storms 650 – 750 years ago, 925 – 875 years ago, 1,250 – 1,150 years ago, 2,800 – 2,300 years ago, 3,350 – 3,250 years ago, 3,600 – 3,500 years ago and 3,950 – 3,650 years ago; the maximum occurrence rate between 2,300 and 2,800 years ago saw six storms per century while the last 150 years have been fairly inactive. Mullet Pond records also somewhat weaker storms and shows a recurrence rate of 3.9 events per century. | ||
| Onslow Bay | North Carolina | Backbarrier deposits | 1,500 | Poor preservation; only 5–8 deposits in 1,500 years | ||
| Oyster Pond | Massachusetts | Sand layers in organic deposits | 1,250 | One of the earliest paleotempestological records; nine sand layers were interpreted as evidence for hurricanes | ||
| Pascagoula Marsh | Louisiana | Sediments | 4,500 | Storms occur about all 300 years; hyperactive period between 3,800 and 1,000 years ago | ||
| Pearl River Marsh | Louisiana | Sediments | 4,500 | Storms occur about all 300 years; hyperactive period between 3,800 and 1,000 years ago | ||
| Princess Charlotte Bay | Queensland, Australia | Beach ridges | 3,000 | 12 hits by intense storms in 6,000 years, implying return periods of 180 years | ||
| Chillagoe | Queensland | Stalagmites | 800 | 2 strong storms between AD 1400 – 1600 after two centuries without one, seven strong storms between AD 1600 and AD 1800 and only one strong storm after that | ||
| Robinson Lake | Nova Scotia | Sediments in lake | 800 | Storms at 1475, 1530, 1575, 1670 and Hurricane Juan. The record probably reflects storms of at least category 2 | ||
| Rockingham Bay | Queensland | Sand ridges | 5,000 | Intense storms occurred between 130 and 1,550 years ago as well as between 3,380 – 5,010 years ago, while the time between 1,550 – 2,280 years ago had very weak storms | ||
| Salt Pond | Massachusetts | Sediments in a lake | 2,000 | 35 hurricanes with active periods between 150 -1,150 AD and 1,400 – 1,675 AD; one historical hurricane recorded; some storms are stronger than the most intense hurricane there, the Great Colonial Hurricane of 1635 | ||
| San Salvador Island | Bahamas | Lake sediments | 4,000 | Increased storm activity between 3,400 and 1,000 years ago. Recurrence rate of strong hurricanes appears to be much less than the historical rate, which may be due to measurement issues | ||
| Santiago de Cuba | Cuba | Deposits in a coastal lagoon | 4,000 | Active periods occurred between 2,600 – 1,800 years ago and between 500–250 years ago | ||
| Sea Breeze | New Jersey | Sediments | AD 214 – present | Storm deposits were emplaced between AD1875-1925, before AD1827, before AD1665-1696, in the 14th–15th century, before AD950-1040, AD429-966 and before AD260-520 | ||
| Seguine Pond | New York | Overwash deposits | 300 | Severe storm surges associated with the 1821 Norfolk and Long Island hurricane and Hurricane Sandy | ||
| Shark Bay | Western Australia | Shell beach ridge | 6,000 | An inactive period between about 5,400 and 3,700 years ago accompanied by drought. Storm intensity indicated by the ridges is about category 2–4 on the Saffir-Simpson scale, while no case of category 5 is inferred | ||
| Shark River Slough | Florida | Sediment cores | 4,600 | Decrease of storm activity after 2,800 years ago | ||
| Shinnecock Bay | New York | Sediments | Older than 1938AD | Several historical deposits by the 1938 New England hurricane, Hurricane Carol, either Hurricane Donna or Hurricane Esther and the Ash Wednesday Storm of 1962 | ||
| Singleton Swash | South Carolina | Sediments in tidal deposits | 3,500 | Historical storms like Hurricane Hazel and Hurricane Hugo are recorded, with more storms until 1050 BC. Between 3050–1050 BC there are no storm deposits, but one deposit dating to 3750 BC appears to relate to a very intense event, perhaps due to a warmer climate at that time | ||
| Silver Slipper West | Mississippi | Overwash deposits and microfossils | 2,500 | Deposits from Hurricane Katrina and Hurricane Camille are present and serve as modern analogues to reconstruct storm surge height for stormy intervals between 350 BC–AD 50 and AD 1050–1350. The decline in activity after AD 1350 coincides with a southward shift in the mean position of the Loop Current | ||
| South Andros Island | Bahamas | Deposits in blue holes | 1,500 | Mainly intense tropical cyclones recorded, including unnamed 1919 and 1945 Category 3 hurricanes although a weaker storm in 1945 might have also contributed. In general there are phases of high and low activity associated with phase changes of the ITCZ volcanic activity and the Little Ice Age | ||
| St. Catherines Island | Georgia | Sediment cores | +3,000 | 7 storms in 3,300 years, equating a recurrence rate of 1 every 471 years. An active period ended 1,100 years before present | ||
| Spring Creek Pond | Florida | Storm layers | 4,500 | An active period between about 600 and 1,700 years ago, but fewer major hurricanes in the last 600 years | ||
| Succotash Marsh | Rhode Island | Sediment overwash | 700 years | Over 6 intense storms in the last 700 years | ||
| Tahaa | French Polynesia | Overwash deposits | 5,000 | Increased activity between 5,000 – 3,800 and 2,900 – 500 years ago with relative inactivity since | ||
| Thatchpoint Bluehole | Bahamas | Sediments | AD 1010–present | Recorded storms include Hurricane Jeanne in 2004; active periods between 1050–1150 AD, a very active period between 1350-1650AD, a reincrease in the late 18th century | ||
| Tutaga | Tuvalu | Coral blocks moved by storms | 1,100 | Increased storminess 1,100, 750, 600 and 350 years ago; correlated with storminess in French Polynesia and a recurrence rate of about 100–150 years | ||
| Tzabnah Cave | Yucatan | Oxygen isotope ratios in stalagmites | AD 750 and earlier | Low tropical cyclone activity at the time of the Classical Maya collapse, and more generally coinciding with drought | ||
| Valdosta State University | Georgia | Oxygen isotope ratios in tree rings | AD 1770 – 1990 | Historical storms have been recorded, as well as a trio in 1911–1913 and a strong event in 1780 | ||
| Wallaby Island | Australia | Beach ridges | 4,100 | Strong storms occur every 180 years | ||
| Walsingham Cavern | Bermuda | Sediments in submarine cave | 3,100 | Increased storm activity between 3,000 – 1,700 and 600 – 150 years ago; however this record might include extratropical storms | ||
| Wassaw Island | Georgia | Overwash | 1,900 | At least eight deposits from strong hurricanes between 1,000 – 2,000 years ago, with a quiet period between 1,100 and 250 years ago | ||
| Western Lake | Florida, northwestern | Overwash deposits | 7,000 | Between 3,800 – 1,000 years ago strike probability was about 0.5% per year, followed and preceded by relative inactivity | ||
| Whale Beach | New Jersey | Sand sheets in marshes | AD 1300–present | Two major hurricanes in 700 years, one between 1278–1438 and the other is the 1821 Norfolk and Long Island hurricane | ||
| Wonga Beach | Queensland, northern | Beach ridges | 4,500 | An inactive period between about 3,800 and 2,100 years ago was followed by an active on between 2,100 and 900 years ago | ||
| Xincun Bay | China, southern | Lagoonal sediments | 7,500 | Seven storm periods in the last 7,500 years, including active periods between 5,500 and 3,500 and from 1,700 years ago onwards, with inactive period in between; there are also active periods embedded within these active ones and there is more generally a correlation to storm activity elsewhere in southern China and to ENSO variations | ||
| Yok Balum Cave | Belize | Oxygen isotope ratios in speleothems | AD 1550 – 1983 | After an inactive phase in the middle 16th century, an increase to ~8 storms/year in the 17th century associated with the Little Ice Age. Then a steady decrease until 1870, when occurrence halved and dropped to ~2 storms/year | ||
| Yongshu Reef | South China Sea | Coral blocks relocated by storms | 4,000 | Six strikes in 1,000 years, with two during the Little Ice Age and four during the Medieval Climate Anomaly. Also high storm activity around 1200 AD, 400 BC and 1200 BC |
Non-tropical examples
| Place | Country/state | Data sources | Record duration in years before present | Conclusions | Sources | Approximate coordinates |
| Île d'Yeu | France | High-energy sedimentation | 8,000 | Between around 5720–5520 BC and 5050 BC–AD 360, storm activity was less meaningful. Increased storminess occurred AD 1350–1450, 150 BC–year 0, 900–400 BC, 1550–1320 BC, 3450–3420 BC, and 4700–4560 BC. | ||
| Pierre Blanche and Prevost lagoons | France | Overwash deposits | 1,500 | Four intense storms in the last 1,500 years |