Earth’s Future

Supporting Information for

Climate Forcing of Unprecedented Intense‐Hurricane Activity in the Last 2,000 Years

Jeffrey P. Donnelly1, Andrea D. Hawkes2, Philip Lane1*, Dana MacDonald3, Bryan N. Shuman4, Michael R. Toomey5, Peter J. van Hengstum6, Jonathan D. Woodruff3 1Coastal Systems Group, Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, 02543, USA, 2Department of Geography and Geology, Center for Marine Science, University of North Carolina Wilmington, Wilmington, North Carolina, 28409, USA, 3Department of Geosciences, University of Massachusetts, Amherst, Massachusetts, 01003, USA, 4Department of Geology & Geophysics, University of Wyoming, Laramie, Wyoming, 82071, USA, 5The University of Texas at Austin, Jackson School of Geosciences, Austin, Texas, 78712, USA, 6Department of Marine Sciences, Texas A&M University at Galveston, Galveston, Texas, 77554, USA

*Deceased

Contents of this file

Text S1 Figures S1 to S9 Tables S1

Introduction

This supporting information provides further details regarding attribution of event beds to historically known hurricane strikes, and a series of supplementary figures that provide further details on our data and interpretations.

1

Attribution of Historical Event Beds The instrumental record (1851 CE to present) Two hurricane landfalls and five significant extratropical storms occurred in New England in the 1982‐2005 CE interval of event bed 1. The extratropical storms (January 1987, October 1991, December 1992, January 1997, December 2005) produced storm tides between 0.87 and 1.0 m above mean high water (MHW) at the Woods Hole tide gauge (4 km southwest of Salt Pond) and thus were unlikely to breach the barrier at Salt Pond. Aerial photography from 1995, 2001, 2003, 2004 and 2007 (available on Google Earth) show no evidence of barrier overtopping. Hurricane Gloria on September 27, 1985 made landfall in western Long Island, New York as a category 2 storm (Fig. S4) and had minimal impact in southeastern Massachusetts. The Gloria storm tide at the Woods Hole gauge was 0.8 m above MHW. No evidence of barrier overtopping is obvious in aerial photographs taken 6 months following Gloria (Fig. S7). The second hurricane was Bob, which made landfall at category 2 intensity near Newport, RI on August 19, 1991 (Fig. S3). The storm tide at the Woods Hole gauge for Bob was 1.5 m above MHW. Surge and wave runup associated with Bob overtopped the western portion of the barrier fronting Salt Pond resulting in washover fans (Fig. S7). Thus, Hurricane Bob is the most plausible candidate for depositing event bed 1.

Like the severe extratropical storms between 1982 and 2005 CE, a series of three storms with storm tides of close to 1 m have struck in the last three years. Tropical Storm Irene made landfall near New York City on August 28, 2011. The maximum storm tide levels at the Woods Hole gauge for Irene was 0.76 m above MHW. Hurricane Sandy in October, 2012 resulted in a storm tide of 1.13 m above MHW and finally an extratropical storm in December 2012 caused a storm tide of 1.02 m above MHW. Localized overwash of the barrier was observed during Sandy and Irene, with overwash sediments penetrating at some locations to the north side of Surf Drive. However, none of the overwash sediment during these recent events was transported into Salt Pond. CTD water column profiles taken before and after the passage of Irene in 2011 show that Irene mixed the stratified water column in the deep basin of Salt Pond (Fig. S2).

Six additional extratropical storms resulted in storm tides between 0.9 and 1.05 m above MHW at the Woods Hole gauge between 1932 and 1982 CE and like the extratropical storms of the last few decades, none of these storms resulted in event beds in Salt Pond. A series of hurricanes made landfall on Long Island, NY in the middle part of the 20th century (e.g., 1938, 1944, 1954 CE), and while Woods Hole experienced significant storm tides related to these events [Boldt et al., 2010], none of these events appear to have left a distinguishable coarse event bed in Salt Pond. A broad peak in coarse fraction occurs at a depth of 50‐60 cm dates to the first half of the 20th century, but it does not exceed the D99 threshold above the 11 point running mean. This broad increase in coarse fraction may be the result of road construction adjacent to Salt Pond in the late 19th and early 20th century (Fig. 1c). The lack of distinct event beds related to these more distal hurricane strikes suggests that local wind speeds in Falmouth were insufficient to generate waves large enough to transport sediment the more than 400 m to the deep basin in the northeast corner of Salt Pond. For example, the storm tide of record at the Woods Hole gauge, 2.7 m above MHW, was caused by the category 3 September 21, 1938

2

hurricane that struck central Long Island. Yet sustained wind speeds at the time of landfall at Edgartown, 20 km southeast of Salt Pond, were only 74 mph [Brown, 1939](using a conversion factor of 1.07 from 5‐minute to 1‐minute sustained winds). The closest maximum sustained wind observation for the 1944 hurricane of 61 mph comes from Nantucket (~57 km to the southeast of Salt Pond) [Sumner, 1944]. In contrast maximum sustained winds in Woods Hole during Hurricane Bob were measured at 85 mph, with gusts reaching 125 mph [Mayfield, 1991].

A small diameter intense tropical cyclone made landfall near the Connecticut/Rhode Island border on September 8, 1869 [Ludlum, 1963]. The small size and fast speed of the storm limited coastal flooding impacts in southeastern New England. No deposit attributable to this storm was identified at Salt Pond or Mattapoisett Marsh [Boldt et al., 2010]. Hindcast modeling indicates a storm tide at Woods Hole for the 1869 hurricane of 1.1 m above MHW [Boldt et al., 2010].

Historical hurricanes making landfall to the east of Salt Pond have failed to generate large storm tides and event beds in Salt Pond. For example, Hurricane Edna made landfall about 10 km to the east of Salt Pond on September 11, 1954 as a category 1 storm and resulted in a storm tide of only 0.78 m above MHW at Woods Hole. Similarly, the October 4, 1869 category 2 hurricane, occurring less than a month after the September 8 hurricane mentioned above, made landfall about 20 km east of Salt Pond also failed to produce an event bed.

Historically extratropical storms have not generated sufficient storm tides and wave energy to mobilize and transport coarse sediment to the depo‐center at Salt Pond. Further, despite generating enough surge to overtop the barrier fronting Salt Pond, more distal landfalling severe historical hurricanes (i.e. 1938, 1944, 1954) with relatively weak local winds (perhaps tropical storm force or marginal category one strength) failed to generate event beds in Salt Pond.

The last 162 years of sediment at Salt Pond indicates that only relatively intense storms making a close landfall (~100 km) to the west of the site have left event beds. Simulating surge and waves for a large suite of hurricane scenarios in Apalachee Bay, FL demonstrated that hurricanes with a wide variety of maximum sustained winds, sizes, and trajectories can produce complete inundation of coastal barriers there, however the vast majority of these modeled storms were category 3 or greater in intensity (83%). Further, most of these storms made a close landfall to the west, therefore causing the more damaging front‐right quadrant of the storm to pass over the site [Lin et al., 2014].

1700‐1850 CE While many weak hurricanes made landfall in southeastern New England between 1700 and 1850 CE, only a few severe hurricanes struck. Most notably, The Great September Gale of 1815 struck Long Island and southern New England on the morning of September 23, 1815. Historians have frequently highlighted the similarity of this storm to the 1938 Hurricane (e.g., [Ludlum, 1963]). Moving at close to 22 m s‐1 (or 50 mph) it made landfall on Long Island, NY, near Center Moriches, less than 16 km to the east of the landfall location of the 1938 Hurricane,

3

and resulted in a similar damage pattern [Boose et al., 2001]. Near Mattapoisett, MA a storm tide of about 3.5 m was noted and a deposit attributable to this storm was recovered at Mattapoisett Marsh was recovered [Boldt et al., 2010]. Hind‐cast surge simulations indicate the storm tide at Salt Pond was close to 2.5 meters above MHW, yet like the 1938 event the more distal landfall of the 1815 hurricane appears to have inhibited event bed deposition from this storm at Salt Pond.

Two other notable hurricanes made landfall in this interval. A hurricane made landfall in southeastern New England on October 23 and 24th, 1761 [Ludlum, 1963]. No event beds attributable to this event are recorded at Mattapoisett Marsh [Boldt et al., 2010]. Severe northeast winds in the Boston area indicate the center of this storm likely passed to the southeast of Boston, but unfortunately no accounts have been discovered from Cape Cod, so the storms track is somewhat uncertain. While no event beds at Salt Pond exceed our significance threshold at this time, a coarse fraction anomaly of about 1.3% (just below our 1.34% threshold) at ~166 cm dates to between 1742 and 1795 CE and could be related to this event. Another hurricane made landfall in southern New England on September 27, 1727, but its exact landfall position is difficult to discern from historical accounts [Ludlum, 1963]. A modest event bed was attributed to this storm at Mattapoisett [Boldt et al., 2010], but no evidence of its passage is evident at Salt Pond. This may indicate the storm passed to the west of Mattapoisett Marsh, but was far enough away that it lacked the local intensity at Salt Pond to transport sediment the distance required to preserve an event bed (perhaps similar to some of the mid‐20th century storms).

1621‐1700 CE (Early Colonial period) Historical records document two hurricanes between 1668 and 1695 CE (event bed 2; mean age is 1679 CE; Supplemental Fig. 5e), with one on September 7, 1675 and the second on August 25, 1683 [Ludlum, 1963]. The historical accounts suggest that the impacts of the 1675 event were similar to the intense Great Colonial Hurricane of 1635 in southeastern New England (see below). Conversely, the 1683 event appears to largely have consisted of a precipitation driven flood in the Connecticut River. Thus, we attribute the event bed at 235 cm to the 1675 event.

Two hurricane landfalls are also documented in New England between 1621 and1660 CE (Supplemental Fig. 5e; the age range for event bed 3 is 1614‐1660 CE (mean = 1639 CE), however continuous European accounts of hurricane landfalls did not begin until 1621 CE). The first of these is the Great Colonial Hurricane of August 25, 1635, which passed across southeastern New England and caused widespread damage consistent with a category 3 hurricane [Boose et al., 2001] and a storm tide of about 6 m in Buzzards Bay. Hindcast surge modeling indicates surge likely reached about 3.5 m at Salt Pond [Boldt et al., 2010]. The second hurricane in this interval is less well documented and occurred on August 13, 1638. This storm may have struck further to the west of Falmouth as wind observations indicate the center of the storm passed to the west of Boston, MA. A storm tide of over 4 m was noted for Narragansett Bay, about 60 km west of Salt Pond. Of these two events the 1635 hurricane

4

appears to be the most logical candidate for the placement of the event bed at 253 cm, given its more proximal track and well‐documented storm tide and damage in the region.

Given the sampling resolution is approximately annual, we should be able to differentiate two storms occurring only 3 or 8 years apart, but only one event bed is distinguished in the Salt Pond archive in each interval. This suggests that only one storm in each of these intervals was locally intense enough to generate an event bed.

5

south Distance (m) north Barrier 0 100 200 300 400 500 0

1 stratigraphic 2 truncations SP2 SP6 3 (m)

Salt Pond 4 gas Depth depo‐center 5

6 SP2 gas 7 SP6 Fig. S1. Representative Chirp subbottom sonar profile from Salt Pond. Track line is shown in the inset image in yellow. The cross‐sectional image shows the geometry of the Salt Pond and the 5.5 m deep basin in the northeast corner of the Pond (depo‐center). Subsurface north reflectors indicate lithological boundaries and are likely related to sand layers within the dominantly fine grained sediment. Gas charged organic‐rich sediments obscure the stratigraphy below approximately a meter in the southern portion of the basin and below about 0.5 m in the deepest portion of the basin. Stratigraphic truncations are present in the Chirp data in the southern portion of the line in water depths between 2 and 3.5 m. These truncations indicate erosional features likely related to high energy events. Conversely, in the limited window of subbottom available from the deepest portion of the pond suggests uniform deposition. One reflector is obvious is the upper ~0.5 m in the deep portion of the basin and is likely event bed 1 which we attribute to Hurricane Bob in 1991 CE. The location of cores SP2 and SP6 are shown. 8/26/2011 (pre‐Irene) 8/31/2011 (post‐Irene) 0 0

1 1

2 2

3 3 Depth (m) Depth (m)

4 4

5 5 Temp (C) Temp (C) Salinity (PSU) Salinity (PSU)

6 6 15 20 25 30 15 20 25 30 Temp (C) Temp (C)

Fig. S2. Water Column Data. Salinity (Practical Salinity Units PSU) and temperature (degrees Celsius C) profiles from the location of SP2 (see Fig. 1c) taken before and after the passage of tropical storm Irene in 2011. SP2 Age Model Thatchpoint Age Model LPG4, Vieques Age Model

abc

Fig. S3. Age models for Salt Pond, Laguna Playa Grande, and Thatchpoint were derived using Bacon software. Here we show the resulting output with 95% uncertainty for each of these reconstructions. 1938 1954 1960 1869

Fig. S4. Tracks of category 2 or greater hurricanes at landfall in southern New England over the last 162 years from the NOAA “best track” data set (1851‐2013). Dashed line is a 325 km east west line used to estimate expected probability. Thicker lines are category 3 storms. Thinner are category 2 storms. % HERBS % Secale cereale % Plantago lanceolata Opaque spherules a 01020304050bc0 0.2 0.4 0.6 0.8 1 0 0.5 1 1.5 2 2.5 3 3.5 d0 100 200 300 400 500 600 Event 1 0 0 0 0 Industrial Revolution 100 100 100 100 (ca. 1900 CE) Introduction of English plantain (ca. 1800 CE) 200 200 200 200 Event 2 European clearance/agriculture Event 3 (ca. 1670 CE) 300 300 300 300 e 400 400 400 400 800

700 500 500 500 500 600 Depth (cm) Depth (cm) Depth (cm) Depth (cm) 600 600 600 600 Event 2 500 Event 3 700 700 700 700 400

Frequency 300 1635 800 800 800 800 1675 200 900 900 900 900 100

1000 1000 1000 1000 0 1580 1600 1620 1640 1660 1680 1700 1720 Age (years CE)

Fig. S5. Selected pollen taxa used for chrono‐stratigraphic dating. The depths of event beds 1‐3 are noted. (a) The increase in total herb pollen is indicative of European land clearance. (b) Rye (Secale cereale) was among the first crops planted by the colonists and the first appearance of Secale cereale pollen, in combination with the increase in total herb pollen, indicates European clearance and agriculture which began ca. 1670 CE. (c) The introduction of English plantain (Plantago lanceolata) marks the earliest 19th century. (d) Opaque spherules, thought to be associated with the combustion of fossil fuels, mark the onset of the Industrial Revolution. (e) Histogram of ages from Bacon age model for event bed 2 (235 cm depth) and event bed 3 (253 cm depth). 95% confidence interval is noted with black bar and mean age is noted with green circle. The age of historically documented severe hurricanes in 1635 and 1675 are shown with dashed lines. b 137Cs (Bq/g) 0 0.005 0.01 0.015 0.02 0.025

a % coarse c Pb (ints) 0246810 0 50 100 150 200 250 300 350 400 0 Pb 20 pt 5

10 1991 (Bob)

15 ~1986 20 Phase out of Pb gas Depth (cm) 25 1963 30

35

40

Fig S6. Core SP6 Data. Upper 45 cm of percent sand fraction data with 137Cs and bulk Pb pollution chrono‐horizons. The event bed attributed to Hurricane Bob is noted. Salt Pond

March 12, 1986

Salt Pond

August 22, 1991

Fig. S7. Aerial photographs of barrier fronting Salt Pond on March 12, 1986 (six months following Hurricane Gloria) and August 22, 1991 (three days after Hurricane Bob made landfall). (source: http://www.falmouthmass.us/gis/aerial/1991.php). 5 250 AMM summer (JJAS) ACE 200 ACE

150 (10 4 summer 0

kts) 100 AMM

50 r = 0.59 ‐5 0 1950 1960 1970 1980 1990 2000 2010 Year

Fig S8. Accumulated cyclone energy (ACE) and summer AMM, where ACE is calculated by taking the sum of each tropical cyclone’s maximum wind speed squared for all storms in the basin. From NCEP/NCAR reanalysis (http://www.esrl.noaa.gov). Aguada X’caamal [Hodell et al., 2005] Cariaco Basin [Haug et al., 2001]

Lake Bosumtwi [Shanahan et al., 2009]

Pumacocha [Bird et al., 2011] Quelccaya Ice Cap [Thompson et al., 2013]

Fig. S9. Correlation map of precipitable water and AMM (1950‐2010 CE). Locations of proxy records of hydroclimate used to infer changes in ITCZ position (e.g., Figs. 4 and 5). Note that during a positive AMM phase tropical West Africa, Venezuela, and Yucatan, Mexico are wetter and southern Peru is drier, with the reverse for the negative AMM phase. Thus, the centennial scale fingerprint of hydroclimate changes between 1300 and 1800 CE are consistent with those observed during AMM variability over the last 60 years. From NCEP/NCAR reanalysis (http://www.esrl.noaa.gov). Supplementary Table 1 ‐ Radiocarbon results Depth 14 13 Lab # (cm) Material C age ± 2 calibrated ranges (CE) (probability) 1 calibrated ranges (CE) (probability)  C section 1523‐1572 (0.105), 1630‐1683 (0.485), 1735‐ 1639‐1672 (0.608), 1778‐1799 (0.320), 1942‐ OS‐81850 93 misc. plant fragments 240 35 ‐26.69 SP2 D4 1:1 93cm 1805 (0.339), 1931‐1950 (0.071) 1950 (0.072) 1471‐1681 (0.857), 1738‐1752 (0.012), 1762‐ 1521‐1579 (0.449), 1581‐1591 (0.056), 1620‐ OS‐81390 192 misc. plant fragments 270 50 ‐22.1 SP2 D2 1:3 108cm 1802 (0.106), 1937‐1950 (0.025) 1666 (0.415), 1783‐1796 (0.080) 1494‐1530 (0.283), 1538‐1602 (0.539), 1615‐ OS‐75691 252 misc. plant fragments 330 45 1461‐1646 (1.000) ‐25.98 SP2 D2 2:3 18cm 1635 (0.178) 1490‐1522 (0.423), 1574‐1603 (0.373), 1610‐ OS‐71460 286 sassaphras leaf 345 15 1472‐1527 (0.402), 1554‐1633 (0.598) ‐26.98 SP2 D2 2:3 52cm 1627 (0.205) OS‐75831 343 deciduous leaf 465 30 1411‐1462 (1.000) 1426‐1447 (1.000) ‐27.09 SAP2 D2 2:3 109cm OS‐87873 392 deciduous leaf 680 30 1270‐1316 (0.635), 1354‐1389 (0.365) 1278‐1301 (0.680), 1368‐1381 (0.320) ‐26.72 SAP2 D6 2:3 43cm OS‐75832 410 misc. plant fragments 805 25 1189‐1271 (1.000) 1220‐1257 (1.000) ‐24.74 SAP2 D2 3:3 17cm 1038‐1049 (0.173), 1084‐1124 (0.613), 1136‐ OS‐71394 482 deciduous leaf 940 15 1031‐1059 (0.217), 1066‐1154 (0.783) ‐27.36 SP2 D2 3:3 97‐98cm 1150 (0.213) 1019‐1048 (0.426),1087‐1123 (0.445), 1138‐ OS‐81851 460 deciduous leaf fragments 975 35 997‐1005 (0.017), 1011‐1155 (0.983) ‐28.03 SP2 D7 2:4 71cm 1149 (0.129) 778‐791 (0.031), 805‐842 (0.059), 860‐988 OS‐81852 545 leaf stems 1130 30 888‐908 (0.256), 913‐968 (0.744) ‐26.6 SP2 D7 3:4 28cm (0.910) OS‐75825 557 woody stems 1110 25 887‐989 (1.000) 896‐926 (0.492), 942‐974 (0.508) ‐26 SP2 D6 3:4 95 cm 2 plant fragments (possible leaf OS‐87423 584 1250 30 677‐779 (0.785), 790‐869 (0.215) 688‐774 (1.000) ‐28.41 SP2 D7 3:4 68 cm vascular tissue) 350‐367 (0.009), 379‐717 (0.977), 743‐766 OS‐87764 611 misc. plant fragments 1480 100 432‐490 (0.271), 531‐652 (0.729) ‐16.96 SP2 D6 4:4 21‐22cm (0.015) 663‐777 (0.968), 792‐803 (0.013), 819‐821 OS‐75681 657 deciduous leaf 1270 30 687‐726 (0.564), 738‐768 (0.436) ‐28.34 SP2 D6 4:4 73.5cm (0.002), 842‐859 (0.017) 426‐436 (0.116), 446‐472 (0.295), 486‐535 OS‐81856 698 deciduous leaf fragments 1580 30 410‐546 (1.000) ‐26.69 SP2 D7 4:4 80cm (0.590) 421‐434 (0.176), 453‐470 (0.186), 487‐534 OS‐87878 698 deciduous leaf 1590 25 412‐539 (1.000) ‐26.15 SP2 D7 4:4 59cm (0.638) 244‐439 (0.852), 442‐474 (0.044), 485‐535 OS‐79589 740 misc. plant fragments 1670 55 258‐284 (0.139), 322‐427 (0.861) ‐16.26 SP2 D11 1:2 ~45cm (0.104) OS‐81857 756 deciduous leaf fragments 1810 30 128‐258 (0.907), 283‐322 (0.093) 140‐196 (0.589), 208‐241 (0.411) ‐28.18 SP2 D8 2:2 71cm OS‐87950 772 deciduous leaf fragments 1770 45 134‐358 (0.966), 364‐380 (0.034) 214‐339 (1.000) ‐28.22 SP2 D11 1:2 69‐71cm OS‐79365 808 misc. plant fragments 1860 40 65‐243 (1.000) 88‐102 (0.124), 122‐215 (0.876) ‐27.66 SP2 D11 2:2 36cm OS‐75828 817 misc. plant fragments 1910 25 24‐ 137 (1.000) 70‐93 (0.468), 97‐125 (0.532) ‐25.89 SP2 D9 1:2 43cm