Boulder transport by waterspouts: An example from Aorangi Island, New Zealand

(2006) 115–125www.elsevier.com/locate/margeo

Marine Geology 230

Boulder transport by waterspouts: An example fromAorangi Island, New Zealand

Willem P. de Lange a,⁎, Peter J. de Lange b, Vicki G. Moon a

a Department of Earth Sciences, University of Waikato, Private Bag 3105, Hamilton, New Zealandb Terrestrial Conservation Unit, Department of Conservation, Private Bag 68908, Auckland, New Zealand

Received 20 November 2005; received in revised form 10 March 2006; accepted 9 April 2006

Abstract

In 1996 the vegetation on Aorangi Island within the Poor Knights Island Nature Reserve, New Zealand, was found to beseverely damaged. The damaged zone consisted of a distinct trail of stripped soil and broken vegetation starting at sea level andtravelling in an arc across the northeastern flank of the island. Cobbles encrusted with marine life, sea urchins and seaweeds, weredeposited at elevations up to 45 m above sea level along and beside the damaged zone. Boulders up to 2 m in size were transportedhorizontally at elevations of 5–15 m above sea level, and larger boulders were destabilised at 25 m above sea level causing them toslide and roll downslope.

Application of threshold entrainment relationships for tsunamis, storm waves and waterspouts indicates that a waterspout is themost probable mechanism for the observed damage and transport of large clasts. The addition of 2.5–5% of atomised water to thewaterspout vortex significantly increases the flow competence. Hence, intensity T3–T5 (TORRO Tornado Intensity Scale) or F2(Fujita Tornado Intensity Scale) waterspouts commonly observed around the New Zealand coast are capable of moving boulders upto 2 m in size.© 2006 Elsevier B.V. All rights reserved.

Keywords: sediment transport; tornado; waterspout; boulder deposits

1. Introduction

Aorangi Island is part of the Poor Knights IslandNatureReserve located some 24 kmoffshore from the eastcoast of Northland, New Zealand (Fig. 1). The PoorKnights Islands consist of two main islands, Tawhiti Rahi(163 ha) and Aorangi (110 ha), and a scattering of smallerislands and islets (de Lange and Cameron, 1999). Bothmain islands rise steeply from the sea to ∼200 m abovesea level, with Aorangi being the highest (216 m) and

⁎ Corresponding author.E-mail address: [email protected] (W.P. de Lange).

0025-3227/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.margeo.2006.04.006

more accessible via sloping shore platforms in Crater Bayand at Fraser's Landing.

The Poor Knights Islands represent the erodedremnants of a late Miocene rhyolite dome volcano, andconsist mostly of rhyolitic breccia and tuff that haveundergone intense hydrothermal alteration (Hayward,1991). Slow tectonic uplift over the last million years andchanges in sea level have resulted in the development ofeight marine terraces preserved between 30 and 185 mabove sea level on Aorangi Island.

The lower terraces contain rounded rhyolite pebblesthat may represent beach deposits. However, archaeolo-gists suggest that these may have been introduced by

116 W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

Maori to improve the soils for the cultivation of kumara(Hayward, 1993). Further, all the terraces have beenextensively modified by the Maori and it is probable thatmost, if not all, are artificial constructs and not marine inorigin. One feature of relevance to this paper is the relativeabsence of boulders on the terraces due to the develop-ment of gardens.

Despite evidence of repeated and prolonged humanhabitation, the Poor Knights Islands have remained rodentfree resulting in a remarkable diversity of endemic andnationally scarce invertebrates and reptiles (de Lange andCameron, 1999). This resulted in the creation of the PoorKnights Islands Nature Reserve, with access strictly con-trolled by the Department of Conservation (DOC).

2. Aorangi Island

To facilitate monitoring of flora and fauna within theNature Reserve, the Department of Conservation estab-lished a campsite in Puweto Valley above Crater Bay(Figs. 1 and 2). An intertidal rocky shore platform islocated on the eastern side of Arid Point, rising almostvertically some ∼40 m to a coastal platform. This plat-form is a slab-like unit of rhyolitic breccia that slopesdownwards northwest to The Landing, where it is ∼1 mabove high tide level (Fig. 2A). The surface of the slab iscovered in numerous small ponds fed by seepages andephemeral streams draining Puweto Valley, and bouldersup to 2 m in diameter (Fig. 2B). The slab forms a steepcliff that drops into the waters of Crater Bay.

On the 5th August 1996, a small party from DOC andAuckland Museum (including one the authors, P deLange) arrived at Aorangi Island to undertake a weeklongbotanical survey. They found that a stretch of the shorelineseveral hundred metres long at Arid Point had beenstripped of all encrustingmarine life; the coast appeared asif it had been steam or water-blasted. The stripped zonecontinued inland across the coastal platform and up a clifftowards the Puweto Valley (Fig. 2C).

On the coastal platform at an elevation of 5–15 mabove sea level boulders up to 2 m in diameter weredisplaced inland up to 10 m from their original positionsas determined from depressions in peat and/or absence ofground cover in rocky areas. In some areas the originalcover of peat and vegetationwas removed, but the originallocations of boulders were evident as differences in thestaining of the underlying rock. There were no evidenttracks between the inferred original locations of the boul-ders and their final resting places. The deposited bouldersoverlay broken vegetation, remnant peat, or sat directly onthe underlying rock. The deposit was chaotic and therewas no obvious imbrication, orientation or sorting. How-

ever, the survey party comprised only botanists and asedimentological survey was not undertaken.

At the top of the coastal platform a formerly dense beltof flax (Phormium tenax) and oioi (Apodasmia similis)was completely removed. The underlying oioi peat, whichhad been∼30 cm thick, was completely stripped down tothe underlying rhyolite rock (Fig. 3A). The damaged zonenarrowed away from the coast, becoming a swath∼20 mwide and extending to ∼50 m above sea level in thePuweto Valley near the DOC campsite (Fig. 3B and C),before running northwards to The Landing where itterminated at the coast (Fig. 2). Various species of sea-weed (Carpophyllum sp., Ecklonia sp. and Cystophorasp.) were found in treetops, and both seaweed and seaurchins (Evechinus chloroticus) weremixed inwith debristhroughout the damaged zone. Shears and Babcock(2004) surveyed sub-tidal flora and fauna near TheLanding, Aorangi Island. From their data, Carpophyllumare strong strap-like seaweeds that are found in the lowerlittoral zone at depths b2 m. Ecklonia are large fragilebrown algae that are easily damaged by waves and occurin dense beds in the sub-littoral zone to depths of 20 m,with the densest beds at depths N6m.Evechinus normallyfeed on Ecklonia sp. with the highest concentrations oc-curring between 4–10 m depth on the eastern coast ofAorangi Island, but they can be found from the littoralzone to depths of 50 m (Doak, 1979).

The DOC campsite at ∼25 m above sea level wasfound to be strewn with boulders with long axes up to∼5 m. This site had previously been free of large boul-ders. A boulder bank above the campsite had collapsed,and this may have contributed some of the bouldersfound within the damaged zone. However, the smallerboulders (cobble-sized) were still encrusted with marineorganisms, including barnacles and coralline algae,typical of the shallow waters and shore platforms of thePoor Knights Islands (Shears and Babcock, 2004), in-cluding the shore platform at Arid Point. Much of thebroken off foliage in the damaged zone was still green,while young seedlings were found on freshly exposedsoil. The seaweed had shriveled and blackened however,and the urchins were largely decayed. It was estimatedtherefore, that the damage occurred some 2–6 weeksprior to the arrival of the survey team.

3. Mechanisms

In the literature, there has been considerable debateabout the formation of coastal boulder deposits byextreme events. Proposed mechanisms include tsunamisand storm waves (Nott, 2004; Stone and Orford, 2004;Williams and Hall, 2004). There are no known tsunamis

117W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

in this case. The distribution of the damage is also in-consistent with wave action because it is only in localisedpatches at the coast, and has a narrow curving trackthrough Puweto Valley. Therefore, alternative mechan-isms to wave action were also considered.

Examination of synoptic weather charts published inthe New Zealand Herald newspaper during the period1st June to 20th August 1996 show that these monthswere associated with passage of intense low pressuresystems and cold fronts. These conditions often involve

Fig. 1. Location map for Aorangi Island, Poor Knights Islands, New ZealaAorangi Island are at 20 m intervals.

intense meso-scale storms and squall lines that spawnthunderstorms, hailstorms and tornadoes (Sturman andTapper, 1996). Flash flooding and wind damage wasreported on the mainland near the Poor Knights Islandsfor the periods 22–23 June, 11–12 July, and 15–18 July1996. A T3 tornado caused moderate damage to coastalcommunities near Whitianga on the Coromandel Penin-sula on the afternoon of 15th August 1996.

There were no weather, wave or water level recorderson the Poor Knights Islands during the period 1st June to

nd. Also shown are localities mentioned in the text. The contours for

118 W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

20th August 1996. A tidal model for New Zealand isavailable online from the National Institute of Water andAtmosphere (NIWA) at http://www.niwascience.co.nz/services/tides/index.html and this was used to hindcast thetidal conditions at Crater Bay for the period of interest(Fig. 4A). At Crater Bay the maximum tide during theperiod of interest was a perigean spring tide on 1st August1996 (largest during 1996) with a maximum elevation of1.22 m above mean sea level. The spring tide a monthearlier (3–4 July 1996) was slightly smaller (∼1 cm).These data do not include storm surges. For this section ofNew Zealand coast the storm surge elevation is approx-imately twice the inverse barometric response, althoughas the islands are near the edge of the continental shelf themaximum elevation is likely to be less than this (Barnett,2002; de Lange, 2003).

Weather data were obtained from automatic weatherstations (AWS) on the mainland (Purerua Peninsula,

Fig. 2. Views of the Poor Knights: (A) Aorangi Island as seen from Tawhiti RaPoints; (B) Crater Bay, Aorangi Island, looking westwards towards Castle Cr(C) Zone of damage on Aorangi Island.

Bay of Islands, and Whangarei Airfield) and the Mo-kohinau Islands in the outer Hauraki Gulf (Fig. 1, Table 1)for the period 1st June to 11th August 1996 (Juliandays 153 to 224). These show that there were severalevents involving rapid pressure drops, storm surges(Fig. 4A), strong winds and heavy rain during the 2–6 weeks prior to 5th August 1996 consistent with thenewspaper reports. The Whangarei AWS was equippedwith a lightning sensor during the period of interest.This recorded the number of lightning flashes per hour(Fig. 4C), which is taken as a proxy for the intensity ofthunderstorm activity. The Mokohinau Islands AWSlocation is a good indicator of the wind and hence waveconditions for the outer Hauraki Gulf. This site is nowequipped with a wave recorder, but this was not presentduring the period of interest.

From the AWS data, two storm events are consideredas being capable of causing damage at Crater Bay,

hi Island showing the sloping coastal platform between Arid and Urupaag showing the small ponds and boulders on the coastal platform; and

119W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

Aorangi Island. The first was a north-easterly stormduring 22–23 June (Fig. 4B) associated with averagewind speeds up to 35 m.s−1 and intense rainfalls on the23rd of June 1996. This storm was capable of generatinglarge waves within Crater Bay. The second storm eventinvolved intense thunderstorm activity around Whan-garei on the 18th of July (Fig. 4C). In this case, it issuggested that the presence of the Poor Knights Islandsland masses and disturbed westerly conditions triggeredat least one waterspout in the lee of Aorangi Island.Meaden et al. (2005) document the generation of water-spouts and tornadoes by the Isle of Wight along theEnglish South Coast under similar conditions.

A review of waterspout and tornadoes in New Zealandbetween 1961 and 1975 by Tomlinson and Nicol (1976)

Fig. 3. Damage observed on Aorangi Island on 5th August 1996: (A) region of(B) flattened flax (Phormium tenax) marking the damage swath; broken branencrusted cobbles (left foreground) deposited near the DOC campsite.

notes that they tend to be associated with thunderstorms.Further, their data indicate a higher frequency of water-spouts around offshore islands, although tornadoes andwaterspouts are thought to be under-reported and the datamay be biased by the location of manned lighthouses andshipping routes. Typical New Zealand tornadoes havedamage paths 10–30 m wide and 1–5 km long, and mostreported tornadoes correspond to TORRO Tornado In-tensity Scale (Elsom et al., 2001) values of T3–T4 withmaximumwind speeds of 40–60m.s−1 (Fig. 5), althoughsome have been T5 (Intense Tornado) with maximumwind speeds of 72 m.s−1 (Tomlinson and Nicol, 1976). AT4 (Severe) Tornado is capable of uprooting trees andliftingmotor vehicles andmobile homes, while a T5 eventcan lift heavy motor vehicles (Elsom et al., 2001).

stripped peat and vegetation exposing the underlying rhyolitic breccia;ches, (C) damaged cabbage tree (Cordyline obtecta) and seaweed and

120 W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

4. Transport by waves and waterspouts

With respect to the boulders affected by the da-maging event at Aorangi Island, we consider threephases of transport:

1. The vertical and horizontal displacement of cobblesand small boulders (b0.4 m diameter) from the shoreplatform at Arid Point to elevations of 20–45 m inPuweto Valley;

2. The horizontal displacement of boulders 5–15 mabove sea level on the coastal platform with long axes

Fig. 4. Tide and weather data for the period 1st June to 11th August 1996 (Juliastorm surge elevation at Aorangi Island (solid line), taken as twice the inverseWhangarei Airfield (effectively a mirror image of the pressure record); (B) wperiod 20th to 25th June 1996 (Julian days 172–177); and (C) rainfall and lighJuly 1996 (Julian days 197–202). The periods enlarged in (B) and (C) are m

of ∼2.0 m, intermediate axes of ∼1.5 m, and shortaxes of 0.5–1.5 m; and

3. The downslope movement of tabular boulders withlong axes of 3–5 m, intermediate axes of 1–2 m, andshort axes of 0.2–1 m at the DOC campsite (∼25 mabove sea level).

4.1. Cobble transport

Boulders that were transported on the shore platformcould have been moved by waves or a waterspout. Nott(1997) proposed a relationship for predicting theminimum

n days 153 to 224): (A) height of low and high tide (dots) and estimatedbarometric response calculated from sea level atmospheric pressure atind velocity and direction data from the Mokohinau Islands during thetning count data fromWhangarei Airfield during the period 15th to 20tharked on the tide height and storm surge plot (A).

Table 1Automatic weather stations used to obtain data on storms for the period 1 June to 15 August 1996

Location (Fig. 1) Parameters available

Distance from Crater Bay (km) Pressure Wind velocity Wind direction Rainfall Lightning

Purewa Peninsula 76.6 x x x xWhangarei Airfield 25.4 x x x x xMohohinau Islands 58.2 x x x x

121W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

wave height required for a high-energywave to initiate themovement of boulders, and applied it to deposits inside theGreat Barrier Reef, Australia. This relationship was prog-ressively improved and applied to boulder deposits aroundthe Australian coast (Nott, 2000, 2003a,b; Nott andBryant, 2003). It may be expressed as

Hmin ¼1d

qsq −1

� �2a−4Cm

ab

� �u Vg

� �j kCd

acb2� �þ Cl

ð1Þ

where Hmin=minimum wave height required to initiatetransport of the boulder; a, b and c are the long-intermediate- and short-axes of the boulder respectively;ρs=density of the boulder; ρ=fluid density; u′=peak flowacceleration; g=gravitational acceleration; Cm=masscoefficient; Cd=drag coefficient; Cl= lift coefficient; andδ=wave type coefficient (1 for storm wave and 4 fortsunami). The full equation applies to a subaerial boulder(Nott, 2003b). Setting Cm=0 produces the originalequation (Nott, 1997), which corresponds to a fullysubmerged boulder (Nott, 2003b). Finally, setting Cm=0,Cd=0 and a=half long-axis length applies to a joint-bounded block (Nott, 2003b). The range of values for thecoefficients reported in the literature have beenCd=1.2–2,Cm=1–3, Cl=0.178 and u′=1 m s−2 (Nott, 1997, 2003a,b; Nott and Bryant, 2003).

The coastal platform boulders in Crater Bay weresubaerial, so the full version of Eq. (1) applies. The largerboulders were tabular to blocky with long axes (a) of∼2m, intermediate axes (b) of∼1.5 m, and short axes (c)of 0.5 to 1.5 m. Sampling of the boulders from AorangiIsland was not possible, but similar rhyolite materialswere analysed by Stevenson (1989). He determined bulkdensities of 1695–2390 kg m−3, with a mean of 2143 kgm−3. Taking the mean rhyolite density and a fluid densityof 1025 kg m−3, the boulders on the coastal platformcould be moved by wave heights of 1.4–3.7 m.

The Mokohinau wind data indicate that wavegeneration conditions for the June storm were dura-tion-limited, with a mean wind speed of 32 m s−1 lastingfor 1–2 h. Using the parametric JONSWAP relation-ships of Carter (1982), these conditions correspond tosignificant wave heights of 1.3–2.1 m, which would be

sufficient to move most of the boulders if they werelocated on the shore platform. However, they werelocated on the coastal platform at elevations of 5–20 mabove mean sea level, and the combination of tides andstorm surge for the June Storm (Fig. 4A) is insufficientto allow the waves to traverse the coastal platform. Thelarge tabular boulders transported into DOC campsitewere originally located at higher elevations, and are alsovery unlikely to have been moved by wave action.

The cobbles and small boulders from the shoreplatform at Arid Point clearly could be transported bythe waves during the June storm. In order to reach thePuweto Valley however, the cobbles and boulders neededto overtop the ∼20 m vertical cliff backing the shoreplatform, and possibly a further 40–50 m vertically tocross the ridge at Arid Point (Fig. 1).

Williams and Hall (2004) discuss the emplacement ofmegaclasts by waves at elevations of 0–50 m abovecliffs along the Atlantic coast of Ireland. They appliedversions of Eq. (1) to determine the wave conditionsnecessary to transport the clasts and found that resultswere inconsistent with historical records of wave con-ditions (both storm waves and tsunami). Since plasticdebris embedded in the deposits indicated that theemplacement of megaclasts had occurred relatively re-cently, it appears that the ability of waves to entrainmegaclasts is underestimated by current theory. Com-pared to the material transported at Aorangi Island, themegaclasts discussed by Williams and Hall (2004) werederived from the upper parts of the cliff and not from ashore platform at the cliff base. Although approachessuch as Eq. (1) may underestimate the competence ofwaves, it seems unlikely that wave action was respon-sible for the emplacement of the cobbles and smallboulders in Puweto Valley.

The terminal velocity of cobbles and boulders in aircan be found from the following relationship (Ward-Smith, 1984; Greeley and Iversen, 1985)

WT ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi43qsqgDCD

s¼

ffiffiffiffiffiffiffiffiffi4

3CD

rU ð2Þ

whereWT=terminal velocity,D=clast diameter,CD=dragcoefficient, andΦ=threshold parameter (U ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

gDqs=qp

).

122 W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

Wind tunnel measurements of the behaviour of gianthailstones up to 0.3 m in diameter indicate that the dragcoefficient is in the range from 0.5 to 0.8 (Ward-Smith,1984). Taking a mid-point value of CD=0.65 and a fluiddensity of 1.21 kg m−3, the terminal velocities for a rangeof clast diameters were calculated (Fig. 6). The calculatedvelocities are consistent with the terminal velocity of 40 ms−1 assessed for bricks, which have a lower density thanrhyolite, in tornadoes (Dowell et al., 2005).

Assuming that the vertical (updraft) velocities near thebase of a waterspout or tornado are similar to the horizontaltangential and inflow velocities, a comparison betweenFigs. 5 and 6 demonstrates that typical NewZealand eventsare capable of suspending rhyolite cobbles (Db0.256 m).Allowing for the 20–30% reduction in fall velocity thatoccurs when objects fall within vortices compared to stillair (Dowell et al., 2005), it is possible for small boulders tobe suspended by the larger events (T5). Doppler velocitymeasurements of tornadoes in theUnited States of Americaindicate that the velocities close to ground level are inten-sified and may be 20–60% higher than the tangentialvelocities (Wurman and Alexander, 2005). However theDoppler signatures are modified by processes such asdebris centrifuging (Dowell et al., 2005) and tornado ta-pering, which makes measurement of true updraft ve-locities extremely difficult (Wurman andAlexander, 2005).

Fig. 5. The relationship between wind speed and the TORRO Tornado IntenBeaufort Scale and Fujita Tornado Intensity Scale. The TORRO scale is moreand the light to intense tornadoes typically experienced.

The threshold of sediment motion for winds may beexpressed in terms of a threshold friction speed as(Greeley and Iversen, 1985):

u⁎t ¼ A

ffiffiffiffiffiffiffiffiffiffiffiqsqgD

r¼ AU ð3Þ

Where u⁎t= threshold friction velocity (u⁎t ¼ffiffiffiffiffiffiffiffiffist=q

p),

τt= threshold surface shear stress, A=threshold speedcoefficient. For large clasts the threshold speed coefficientis constant, with a value in the range 0.10–0.12 (Greeleyand Iversen, 1985), with a value of 0.11 suggested as asuitable average based on wind tunnel measurements(Greeley et al., 2003). Since the rhyolite clasts are muchlarger than the clasts tested, a coefficient ofA=0.12,whichfits the observations for silt to fine sand clasts of lead(ρs=11,350 kgm

−3), bronze (ρs=7800 kgm−3) and steel

(ρs=7600 kg m−3), is considered more appropriate.For coarse particles, the wind velocity is related to the

threshold friction velocity by (Greeley and Iversen, 1985):

Uu⁎t

¼ 10:4

lnzzo

� �ð4Þ

Where U=wind velocity at elevation z above theground, and z0= roughness height.(vertical intercept ofthe logarithmic velocity profile). The roughness height

sity Scale as described by Elsom et al. (2001). Also included are theuseful than the Fujita scale for New Zealand, due to the finer divisions

123W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

for large clasts may be expressed as a fraction of theclast diameter. Typical values for wind flowing overrough ground range between D / 30 and D / 8, with themaximum roughness height occurring for clasts sepa-rated by 2D (Greeley and Iversen, 1985). CombiningEqs. (3) and (4), and assuming a roughness height ofD / 30 due to the wide separation of clasts on the shoreplatform at Arid Point, the wind speed at z=5 m re-quired to initiate clast movement is plotted in Fig. 6. It isevident that typical New Zealand tornadoes and wa-terspouts are capable of moving cobble-sized clasts.Therefore, we conclude that a waterspout was the likelymechanism that transported cobbles, as well as seaweed,from the shore platform and shallow depths at AridPoint into Puweto Valley.

4.2. Larger clasts

Fig. 6 indicates that the wind speeds required to initiatemovement of the boulder-sized rhyolite clasts at CraterBay are greater than those associated with typical NewZealand tornadoes and waterspouts, or even or the largertornadoes experienced in the USA (F4–F6 in Fig. 5). Thelargest boulders moved near the DOC campsite in PuwetoValley probably rolled or slid downslope from the base ofthe Western Cliffs (Fig. 2B). Therefore, we conclude thata waterspout need only destabilise the pre-existing talusmaterial to initiate themovement. This was probably not adirect consequence of the forces exerted by the water-spout, particularly since the source of the boulders lay

Fig. 6. Terminal fall velocity for giant hailstones (Ward-Smith, 1984) and rhyvelocity at an elevation of 5 m required to initiate sediment transport predicsediment clasts is included.

outside the damage path. Instead, we consider that impactby cobbles and/or water thrown out of the waterspout wassufficient to destabilise the talus and cause the downslopemovement of large boulders.

This mechanism does not account for the movementof the tabular boulders on the coastal platform in CraterBay. Within a waterspout or tornado, there are twomechanisms that can lift clasts off the ground (Greeleyet al., 2003): the upward component of force due to thefrictional drag as the wind blows over the clasts asdefined by Eq. (3); and the lift resulting from the re-duction in pressure within the vortex. Hence the forcebalance on a clast at the point of vortex threshold may bewritten as (Greeley et al., 2003):

k1Dpþ k2s ¼ k3qsgDþ k4rp ð5Þ

Where k1–4=constants defining the relative importanceof the terms, Δp=pressure drop within the vortex,τ=vortex shear stress, and σp=inter-clast cohesion.Greeley et al. (2003) demonstrate that for fine clastsentrained by small vortices (Dust devils) the pressurereduction can reduce the threshold friction velocity by80% (k1≫k2), even though cohesion increases theresistance to entrainment. Their data indicates that apressure effect occurs for large clasts, but that thereduction in threshold friction velocity decreases withincreasing clast size (k1≪k2). However, for large claststhe cohesion is negligible (σp≈0).

olite clasts in still air as predicted by Eq. (2). Also plotted is the windted by Eqs. (3) and (4). The Udden–Wentworth size classification for

124 W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

The reduction in pressure in a vortex may beestimated by (Greeley and Iversen, 1985):

Dp ¼ q 1þ 15qsgD

� �U2 ð6Þ

This equation was developed for silt and clay sizedparticles (b63 μm). Measured pressure changes forintensity F2 tornadoes comparable to New Zealandwaterspouts are in the range 24–55 hPa (Samaras andLee, 2004), which corresponds to grain sizes of0.8–1.5 mm for the assumed rhyolite density. This isconsistent with the laboratory measurements of Greeleyet al. (2003) for silica sand. Clearly it is insufficient toaccount for the movement of the large clasts in CraterBay.

Zavolzhenskii (1980) developed a theoretical modelfor the hydrodynamics of an oceanic waterspout, andgives an example for a waterspout with a column of10 m diameter at the water surface. This diameter is lessthan the widths of the observed damage trail on AorangiIsland, but is a reasonable approximation of themagnitude of the waterspout considered to have causedthe damage observed. The modelled waterspout createsa column of atomised water 143.5 m high, with amaximum velocity of 53 m s−1. A conical depressionextending to a depth of 5.3 m, which is consistent withthe depths at which the seaweed and sea urchins in-corporated in the debris are normally found, precededthe column.

Fig. 7. Threshold wind velocity at an elevation of 5 m predicted by Eqs. (classification for sediment clasts is included.

It was also noted that if the waterspout progressesonto land, the column can uplift and atomise water fromstreams, ponds and lakes (Zavolzhenskii, 1980). Thearea in Crater Bay where the larger clasts were movedby this event is characterised by numerous small ponds(Fig. 2B). These were scoured clean, appearing as if thearea had been water-blasted to reveal fresh rock.Therefore, it is probable that the vortex did not consistentirely of air, but contained an unknown amount ofatomised water that increased the fluid density. Fig. 7plots the threshold velocity at z=5 m for fluid densitiesof 1.21, 25 and 50 kg m−3 (corresponding to approx-imately 0%, 2.5% and 5% atomised water by volumewithin the vortex). It is evident that the addition of smallquantities of atomised water to the vortex significantlyincreases the flow competence. Further 2.5% to 5%added water is sufficient to entrain the larger claststransported at Crater Bay, Aorangi Island.

5. Conclusions

It ismost probable that the damage observed atAorangiIsland, Poor Knights Island Nature Reserve on the 5thAugust 1996 was caused by a waterspout associated withthunderstorms 3 weeks earlier. The waterspout upliftedclasts up to cobble size, sea urchins and seaweeds from amaximum depth of∼5 m, and transported them inland toelevations of 20–45 m above sea level. As the waterspouttraversed the coastal platform in Crater Bay, boulders at anelevation of 5–15 m above sea level were transported

3) and (4) for a range of fluid densities. The Udden–Wentworth size

125W.P. de Lange et al. / Marine Geology 230 (2006) 115–125

short distances due to the presence of seawater and waterentrained from local ponds in the vortex. At a higherelevation around 25 m above sea level, the waterspoutdestabilised a boulder talus slope causing large boulders toslide and roll downslope.

This event indicates that waterspouts are a mecha-nism for the formation of boulder deposits in coastalareas, particularly in areas where waterspouts may beconcentrated, such as around islands relatively close tothe coast.

Acknowledgements

The authors wish to thank Ngati Wai (who exerciseMana Whenua over the Poor Knights Islands) and theDepartment of Conservation, Northland Conservancy,especially Ms Lisa Forester for facilitating the necessarypermits to visit the Poor Knights Islands Nature Re-serve. Further, Mr Ewen Cameron (Auckland Museumand Institute Herbarium) and Dr Peter Heenan (Land-care Research) assisted with gathering field data duringthe 1996 visit to Aorangi Island. Finally, the authorsthank Dr James Goff and an anonymous reviewer fortheir useful comments.

References

Barnett, K.M.C., 2002. The behaviour of storm surges in WhangareiHarbour and Bream Bay, Northland, New Zealand. MSc Thesis,Waikato University, Hamilton. 202 pp.

Carter, D.J.T., 1982. Prediction of wave height and period for aconstant wind velocity using the JONSWAP results. OceanEngineering 9, 17–33.

de Lange, W.P., 2003. Tsunami and storm surge hazard in NewZealand. In: Goff, J., Nichol, S., Rouse, H. (Eds.), The NewZealand Coast: Te Tai o Aotearoa. Dunmore Press, PalmerstonNorth, pp. 79–95.

de Lange, P.J., Cameron, E.K., 1999. The vascular flora of AorangiIsland, Poor Knights Islands, northern New Zealand. New ZealandJournal of Botany 37, 433–468.

Doak, W., 1979. The cliff dwellers: An undersea community. Hodderand Stoughton, Auckland. 80 pp.

Dowell, D.C., Alexander, C.R., Wurman, J.M., Wicker, L.J., 2005.Centrifuging of hydrometeors and debris in tornadoes: Radar-reflectivity patterns and wind-measurement errors. MonthlyWeather Review 133 (6), 1501–1524.

Elsom,D.M.,Meaden, G.T., Reynolds, D.J., Rowe,M.W.,Webb, J.D.C.,2001. Advances in tornado and storm research in the UnitedKingdom and Europe: the role of the Tornado and Storm ResearchOrganisation. Atmospheric Research 56 (1–4), 19–29.

Greeley, R., Iversen, J.D., 1985. Wind as a geological process : on Earth,Mars, Venus, and Titan. Cambridge Planetary Science Series. Cam-bridge University Press, Cambridge. 333 pp.

Greeley, R., et al., 2003. Martian dust devils: Laboratory simulationsof particle threshold — art. no. 5041. Journal of GeophysicalResearch—Planets 108 (E5), 5041.

Hayward, B.W., 1991. Geology and geomorphology of the PoorKnights Islands, Northern New Zealand. Tane 33, 23–37.

Hayward, B.W., 1993. Prehistoric archaeology of the Poor KnightsIslands, northern New Zealand. Tane 34, 89–105.

Meaden, G.T., et al., 2005. Influence of an island land mass on thefrequency ofwaterspout and tornado formation in its vicinity. Exampleof the Isle ofWight with regard to waterspouts and tornadoes affectingthe English South Coast. Atmospheric Research 75 (1–2), 71–87.

Nott, J., 1997. Extremely high-energy wave deposits inside the GreatBarrier Reef, Australia: determining the cause— tsunami ortropical cyclone. Marine Geology 141, 193–207.

Nott, J., 2000. Records of prehistoric tsunamis from boulder deposits—evidence from Australia. Science of Tsunami Hazards 18 (1), 3–14.

Nott, J., 2003a. Tsunami or storm waves?—Determining the origin of aspectacular field of wave emplaced boulders using numerical stormsurge and wave models and hydrodynamic transport equations.Journal of Coastal Research 19 (2), 348–356.

Nott, J., 2003b. Waves, coastal boulder deposits and the importance ofthe pre-transport setting. Earth and Planetary Science Letters 210(1–2), 269–276.

Nott, J., 2004. The tsunami hypothesis — comparisons of the fieldevidence against the effects, on theWestern Australian coast, of someof themost powerful storms onEarth.MarineGeology 208 (1), 1–12.

Nott, J., Bryant, E., 2003. Extreme marine inundations (tsunamis?) ofcoastal Western Australia. Journal of Geology 111 (6), 691–706.

Samaras, T.M., Lee, J.J., 2004. Pressure measurements within a largetornado. Eighth Symposium on Integrated Observing andAssimilation Systems for Atmosphere, Oceans and Land Surface,84th AMS Annual Meeting. American Meteorological Society,Seattle. pp. Paper 4.9.

Shears, N.T., Babcock, R.C., 2004. Community composition andstructure of shallow subtidal reefs in northeastern New Zealand.Science for Conservation, vol. 245. Department of Conservation,Wellington. 65 pp.

Stevenson, R.J., 1989. Physical volcanology, emplacement history andinferred viscosity of two rhyolites. PhD Thesis, Waikato University,Hamilton. 182 pp.

Stone, G.W., Orford, J.D., 2004. Editorial: storms and their significancein coastal morpho-sedimentary deposits. Marine Geology 210 (1–4),1–5.

Sturman, A., Tapper, N., 1996. The weather and climate of Australiaand New Zealand. Oxford University Press, Melbourne. 476 pp.

Tomlinson, A.I., Nicol, B., 1976. Tornado reports in New Zealand1961–1975. New Zealand Meteorological Service Technical Note,vol. 229, p. 13.

Ward-Smith, A.J., 1984. Biophysical aerodynamics and the naturalenvironment. John Wiley & Sons, Chichester. 172 pp.

Williams, D.M., Hall, A.M., 2004. Cliff-top megaclast deposits ofIreland, a record of extreme waves in the North Atlantic— stormsor tsunamis? Marine Geology 206 (1–4), 101–117.

Wurman, J., Alexander, C.R., 2005. The 30 May 1998 Spencer, SouthDakota, storm. Part II: Comparison of observed damage and radar-derived winds in the tornadoes. Monthly Weather Review 133 (1),97–119.

Zavolzhenskii, M.V., 1980. Hydro- and thermodynamics of tornadoesand oceanic waterspouts. Journal of Applied Mechanics andTechnical Physics (Historical Archive) 21 (4), 538–548.