The sedimentology of storm-emplaced coastal boulder deposits in the northeastern Atlantic region by Kalle L. Jahn R6nadh Cox, Advisor A thesis submitted in partial fulfillment of the requirements for the Degree of Bachelor of Arts with Honors in Geosciences WILLIAMS COLLEGE Williamstown, Massachusetts May 2014 Acknowledgements I would like to express my deepest gratitude to R6nadh Cox for all of her wisdom and guidance. This thesis would not have come to fruition without her. Special thanks to Bud Wobus for his insight and constructive comments as my second reader. Thanks to Fabio Sacchetti for his amazing work in the field and his GIS help in post-processing. Many thanks to Ward Lopes for setting up Funwave, and assisting me with the modeling. Thanks to Will Wicherski for his help in the field, and for joining me on some great morning workouts. Finally, I would like to thank my loving family for their constant support and interesting in this project. Abstract Northeastern Atlantic coasts exposed to the open ocean have coastal boulder ridges formed and modified by storm waves, providing an opportunity to examine the sedimentology of these understudied deposits. This study documents the morphology and sedimentology of boulder ridges through surveying and systematic grain size measurements. A number of interesting conclusions are reached: some intuitively obvious, others surprising. It makes sense that ridge size and largest clast sizes tend to decrease as ridge height above high water increases. It is surprising then, that there is a significant trend showing both larger ridges and larger clast sizes correlated with distances further inland. Noise in the data inhibits prediction of ridge morphology and sedimentology, but a low coastal slope appears to be a criterion for the transport of large boulders far inland. A mass-based analysis was performed in addition to direct measurements of transect population statistics, because most grain size analyses are performed with mass-based sieving. When a large proportion of the population's total mass is located in a few large clasts, mass-based population statistics deviate strongly from statistics calculated directly from clast measurements. Mass-based statistics are therefore not a good proxy for direct measurements of boulder ridge clast populations, which tend to contain several outsized clasts. The literature Jacks other systematized studies of ridge sedimentology, so the ridges of this study are compared to the most analogous deposits: boulder beaches. The rarity of fine-tailed grain size distributions among boulder ridges distinguishes them from boulder beaches, which tend to be fine­tailed. The morphology and sedimentology of boulder ridges are inherently variable, but this study represents a first step in expanding our understanding of these deposits. Table of Contents ACKNOWLEDGEMENTS ....................................................................................................................1 ABSTRACT ............................................................................................................................................. 11 LIST OF FIGURES ................................................................................................................................. V LIST OF TABLES ................................................................................................................................ VII LIST OF EQUATIONS ..................................................................................................................... VIII INTRODUCTION ................................................................................................................................... 1 WHY IS THIS WORK IMPORTANT? ...................................................................................................................... 2 BOULDER RIDGES ................................................................................................................................................... 3 DIFFERENTIATING STORM AND TSUNAMI DEPOSITS ............. .................................................. ... ........... .... ... 3 EVIDENCE OF STORM ACTIVITY IN THE NORTHEASTERN ATLANTIC REGION .....................................6 OUR APPROACH ....... ................................................................................................................................................6 STUDY SITES ..........................................................................................................................................8 THE ARAN ISLANDS ...... ....... .... ............................................................ ........................................... ... .... ................ 9 THE MUL LET PENINSULA .. ........... ............................ ......................................................................................... 11 Annagh Head ........................................................................................................................................................ 12 Erris Head .......................................................... ....................... ..................... ................. ............ .......................... 13 SHETLAND .............................................................................................................................................................. 14 Eshaness.......................................................................... .. ..................................................................................... 15 Grind of the Navir ................ ............. .................................................................. ... .......................... ....... ..... ....... 16 Head ofStanshi.. ..................................................................................... ............................................................. 18 Unnamed Bay ..................................................................................................................... ....................... . .......... 19 Villians of Hamnavoe ...... .......................................................................................................... ........................ 20 The Burr .......................................................................... ........... ............ ................................................................ 21 Ness o.f Burgi ................................................................. .............................................. ......................................... 22 METHODS .................................................................................... ......................................................... 23 FIELD METHODS.................................................................................................................................................... 23 CAL CULAT ION OF ROCK DENSITIES ............................................................................................................... 26 GRAIN SIZEANALYSIS ........................................................................................................................................ 27 GRADISTAT (mass-based) analysis ............. ............................................................................................... 27 Line count (clast-based) analysis ................ ..................... ..................... ...................................... ................. 29 SYNTHESIS AND STANDARDIZATION OF FIEL D TRANSECT DATA ......................................................... 30 IMBRICATION . ........ ............. ..... .... ............................................................................................ .............................. 31 BATHYMETRIC DATA COLL ECTION ..... ..................................................................... ...................................... 32 GIS............................................................................................................................................................................ 33 WAVE MODELING USING FUNWAVE .......................................................................................................... 34 RESULTS ............................................................................................................................................... 34 FIEL D SEDIMENTOL OGY AND GEOMORPHOLOGY .. .............................................. ...................................... 34 GRAIN SIZEANALYSIS ........................................................................................................................................ 36 FUNWAVE ...............................................................................................................................................................37 DISCUSSION ......................................................................................................................................... 38 RELATIONSHIPS BETWEEN RIDGES AND THEIR LOCATION .... ................................................................. 38 GRAIN SIZEANALYSIS ........ ............................... .................................................................................................41 Discrepancies in population statistics ........................................................................................................ 41 Line count distributions and comparisons to other deposits .............................................................47 CONCLUSIONS ................................................................................................................................... 50 APPENDIX A. TRANSECT POPULATION STATISTICS ...................................................... 52 REFERENCES ...................................................................................................................................... 55 List of Figures Figure I. Generalized ridge profile identifying common ridge elements .......................... 3 Figure 2. Map of study location in Ireland and the UK. (1.) Aran Islands, (2.) Mullet peninsula, (3.) Shetland... .. .................... ............ . .... .. ...... . ... ........... . .. 15 Figure 3. The Aran Islands, where data was collected from 2008 to 2010. Figure from Cox et al. (2012) . .................................................................................................... 9 Figure 4. View of a ridge front on the Aran Islands. Large clast pictured has a mass of -70 tonnes. Notice the scoured bedrock platform in front of the ridge ............... 10 Figure 5. Study sites on the Mullet peninsula. Data from these locations were collected in the summer of 2012. Figure by Leaf Elliott.. ..................................... ...... ........ 12 Figure 6. View of the Annagh Head ridge facing north. People in center right are standing on the ridge crest. The ridge back and scattered boulder field lie to the East (not pictured), while the ridge front, oceanward boulder field, and scoured platform are seen to the west. ....................... ........................................................ 13 Figure 7. View of Erris head deposits facing south. 40 m cliffs located to the west, and long sloping turf fields (not pictured) to the east. . ................................................ 14 Figure 8. Study sites on Shetland. Inset: Northmavine peninsula, where most of the sites are located. The one exception is the Ness of Burgi, located on the southern tip of Shetland. Data from these locations were collected in the summer of 2013 ....... 15 Figure 9. View of the accumulations at Eshaness facing inland to the west. Notice that bedrock is almost completely covered by gravel, with larger clasts scattered throughout. ...... ............ .... ..................... ................................................................. 16 Figure 10. View ofthe Grind ofthe Navir, facing the ocean. Image taken from the front of ridges. Gap is 12 m wide and 8 m tall, seperated from ridges by scoured platform with many bedrock steps ...................... .......... ......... .. ......... ... ... ... ... ... ... .. 17 Figure 11 . View of the ridges at the Grind of the Navir, facing east. Picture shows the front of a ridge complex made up of several ridges. The jointed bedrock, broken into series of steps, shows some fresh surfaces where new clasts have recently been quarried (e.g. light bedrock in the bottom right of image.) .......................... 18 Figure 12. Image of the deposit at the Head of Stanshi, facing south. Scoured, jointed bedrock to the west, gravel and boulders to the east. Some turf is growing in behind the de post. ................................................................................................. 19 Figure 13. Image of the Unnamed Bay ridge taken from the crest, facing west. Shoreline lies to the north, and the beginning of the turf covered scattered boulder field is noticible to the south. Boulders are covered in lichen, indicating lack of recent movement................................... ............ ......... ... ... ... ............................................. 20 Figure 14. View of the Villians of Hamnavoe deposits, facing north. Clusters of large lichen covered boulders punctuate the continous cover of gravel. In the distance, scoured platform can be seen to the west, and the scattered boulder field can be seen to the east. .. ..... ....... ... ... ...... ....... ........ ... ... ...... .. ...... ... ... .. ...... ..... ............ ......... 21 Figure 15. Image of the boulder beach located at the Burr, facing east. Deposit extends below low tide, and the green algae mark the furthest extent of normal high tide . ............................. .................................................................................................. 22 Figure 16. Image of a ridge at the Ness of Burgi, taken on the ridge crest facing south. Flaggy sandstone clasts make up the ridge, many of them are covered in lichen. Cliffs drop into the ocean to the east and west of the ridge .................................. 23 Figure 17. Data collection techniques: (a.) surveying performed using tape measure, Brunton compass, and laser rangefinder; (b.) clasts counted using the ribbon count method; (c.) boulder roundedness measured using device designed by Kirkbride (2005); (d.) clast imbrication measured with Brunton compass ........................... 24 Figure 18. Details of the horizontal correction calculations used for sloping cliffs. The stick figure is standing at the cliff edge, inland is to the right of the figure ......... 31 Figure 19 -Location of bathymetric data collection around the Aran Islands. Lines represent path taken by ship while collecting data. Image by Fabio Sacchetti and INFOMAR............................................................................................................ 32 Figure 20. Relationships between topographical variables (Height AHWM, Horizontal distane inland, Coastal slope) and measures of ridge size (Main ridge width, Main ridge height, Avg. mass of 5 largest clasts). Negative trends are associated with Height AHWM and Coastal slope, while a positve trend is associated with 2 Horizontal distance inland. Correlations have the following p-values and R 2 values: (a.) p = 0.0327, R2 = 0.03 ; (b.) p = 0.0012, R2 = 0.07; (c.) p = 0.0553, R 2 2 = 0.03; (d.) p = <0.0001, R= 0.11; (e.) p = 0.0002, R= 0.09; (f.) p = <0.0001, 2 2 R= 0.13; (g.) p = <0.0001, R= 0.14; (h.) p = <0.0001, R2 = 0.17; (i.) p = 2 0.0001, R= 0.16 . ................................................................................................. 35 Figure 21. Grain size distributions for transects with more than 200 measured clasts. The majority are unimodal, but AH1, BR1, BR2, GN4, HS2, and Z8 all appear to be at least slightly bimodal. The population statistics for these and and other transects can be found in Table 4 ......................................................................................... 37 Figure 22. Numerically modeled distribution of bore velocity after cliff overtopping. Time (t) refers to time after overtopping. The Ryu et al. (2007) study is based on a scaled model, but the patterns found can be related to real waves through Froude scaling. Adapted from Ryu et al. (2007) .............................................................. 39 Figure 23. Graphs depicting relationship between line count population statistics and mass-based populations statistics. No strong trends are evident, and correlations 2 are not significant. P-values and Rvalues for the correlations: (a.) p = 0.90, R2 = 2 0.0003; (b.) p = 0.60, R2 = 0.0055; (c.) p = 0.50, R= 0.0092 . ............................ 43 Figure 24. Relationships between population statistics on the x axes and L\s on the y axes. L\aeoM and L\Med are strongly correlated with the mass-based statistics, while L\so is strongly correlated with the line count SD. Correlations have the following 2 2 p-values and Rvalues: (a.) p = 0.<0.0001, R2 = 0.89; (b.) p = <0.0001, R= 0.92; (c.) p = 0.0074, R2 = 0.13; (d.) p = <0.0232, R2 = 0.10; (e.) p = 0.0115, R2 = 0.12; (f.) p = <0.0001, R2 = 0.87 ........................................................................... 45 List of Tables Table 1. Rock densities calculated from samples collected in the field. Average volume is derived from 4 water displacement tests performed on each sample. Mass measured using a small scale that is precise to the nearest gram .......................... 27 Table 2. Comparison of GRADIST AT population statistics calculated for transect AH l using .25 and 0.5 intervals, and rounded and un-rounded 0.5 intervals. The difference in results is at least an order of magnitude smaller than the differences found between transects . .................................................................... 29 Table 3. Modified ranges for characterizing sorting and skewness calculated using the Folk and Ward (1953) graphical measures as modified for metric units by Blott and Pye (2001) ............................................................................................. ......... 30 Introduction Many areas of the world are subject to intense wave energy produced by large storms, tsunami, or both. As coastal populations grow and sea levels rise, coastal risk assessment studies require a better understanding of the affects of these high-energy, destructive processes. Coastal boulder deposits are a characteristic feature of the highest energy wave events, and understanding how these deposits form is therefore key to risk assessment. To date, however, there has been no consensus about whether these deposits are produced by storms, tsunami, or both. It is best to study places where the depositional mechanism is well understood and then compare those places to locations where the mechanism is unknown. Therefore, the storm-rich, recently tsunami-free northeastern Atlantic region (NEA) is an ideal location for studying coastal boulder deposits. The spectacular NEA gravel and boulder deposits generally take the form of well­developed ridges containing clasts up to 78 t (Cox et al., 2012). These ridges are often located at considerable elevation (up to 40 meters above sea level), as well as at substantial horizontal distances from the high water mark (up to 120 m inland). NEA examples occur in Ireland (Williams and Hall, 2004; Hall et al., 2006; Scheffers, Kelletat, Haslett, et al., 201 0; Cox et al., 20 12), Shetland (Hall et al., 2006, 2008), France (Suanez et al., 2009), and Iceland (Etienne and Paris, 201 0). The NEA is not the only place where coastal boulder deposits form; examples are also found in Japan (Go to et al., 2007, 2009; Go to, Miyagi, et al., 201 0), southeast Asia (Go to et at., 201 0), Hawaii (Richmond et al., 201 1 ), the Caribbean (Scheffers, 2004; Morton et al., 2008), and Australia (Nott, 1997; Saintilan and Rogers, 2005). The main difference between these locations and the NEA is that they can be products of tsunami as well as large storms. Although it has been argued that the NEA deposits are relict deposits recording ancient tsunami activity (Scheffers, Kelletat, Haslett, et a!., 201 0; Scheffers, Kelletat, and Scheffers, 20 1 0), it is now clear that the deposits are currently being shaped as a result of storm wave action (Hallet a!., 2006; Cox eta!., 20 12). The absence of recent tsunami is unique to the NEA, so unlike other locations, any recent depositional characteristics identified in the NEA can be confidently categorized as features of storm deposits. Why is this work important? Documentation of the specific effects of storm and tsunami events is important for quantitatively assessing future risk in coastal communities. Coastal boulder deposits are vital to this assessment, because they provide an extensive record of powerful depositional events. Preparation for tsunami disasters is very different from preparation for extreme storms, as tsunamis occur quite suddenly, while storms can be tracked before they hit. Therefore, if boulder deposits are to be useful for risk assessment and preparation, it is necessary to first differentiate between tsunami and storm-emplaced deposits. Despite abundant work (discussed below), no one has devised a diagnostic tool that both accurately models wave complexity and functions irrespective of location. If the quantification of deposit sedimentology on Ireland and Shetland reveals commonalities among storm deposits in different locations, it might serve as a diagnostic tool that could be applicable worldwide. Boulder Ridges Generally, boulder ridges are gravel and megagravel accumulations found at elevations ranging from sea level to 50 m above the high water mark (AHWM). Ridges can be found as close as 4 m from the shoreline and as far as 250 m inland. Most ridges are separated from the ocean by cliffs and scoured bedrock platforms swept clean of clasts. The archetypal boulder ridge has a steep, ocean-facing front that ends in a distinct crest. Behind the crest the deposit slopes gently inland (Figure 1 ). These structures can reach up to 7 m tall from platform to crest, and over 250 m wide from front to back. Most locations have only a single ridge, but there are also ridge complexes made up of a series of 2 to 4 ridges that become smaller inland. A scattered boulder field of solitary and clustered clasts often characterizes the area most distal part of the deposit. Turf usually grows behind ridges in the NEA region, and many clasts become partially buried by turf growth after being deposited in the scattered boulder field. Ridge crest Subsidiary Ridge ridge j Ridge front Scattered back boulder field Scoured platform Cliff Bedrock Ocean Figure I. Generalized ridge profile identifying common ridge elements. Differentiating storm and tsunami deposits Our study is not the first to try and differentiate between storm and tsunami deposits. A common approach has been to use wave transport equations, such as those reworking. Consequently, the mere presence of organized ridges at a location could be a good first-order indication of a powerful storm regime at work. The difference in deposition extent between storm waves and tsunami waves may also be an important discriminatory feature (Go to et al., 20 I0; Goto et al., 20 I 0). In many cases, tsunami deposits are characterized by extensive fields of sand, gravel, and boulders spread over thousands of square meters (Goff et al., 2004; Kortekaas and Dawson, 2007; Richmond et al., 20 1 1 ). In contrast, it is generally thought that the effects of storms are limited to the shoreline area, a notion often based on differences in the degree of weathering of the deposits as one moves inland. For example, in their study of the 1975 Kalapana tsunami, Goff et al. (2006) distinguished between storm and tsunami deposits using an arbitrary point that was selected based on the distinction of yellow (fresh) sand and grey (weathered) sand around I00 m inland. They supposed that anything deposited beyond the separation point had probably not been affected by recent storm waves and must therefore be associated with the 1975 tsunami. This approach may work in locations where the tsunami and storm history is well documented, but in places where the history is not well known, there would be no direct evidence again massive storms being responsible for deposits beyond I OOm inland. Without site history, distance alone cannot be used to identify the depositional events responsible for deposits within only a few hundred meters of the shoreline. Despite the numerous attempts to identify suitable methods for differentiating storm and tsunami deposits, no one has been able to develop an all-encompassing diagnostic tool. The basic morphology and sedimentology have been described in many studies (Hall et al., 2006; Morton et al., 2008; Etienne and Paris, 201 0; Richmond et al., 2011 ), but a true quantitative understanding of the mechanisms for ridge formation remains elusive. Evidence of storm activity in the Northeastern Atlantic region Recent studies present strong evidence that ridges on the Aran Islands and Shetland have moved over the last 1.5 centuries. Comparison of 19th century maps to present-day orthophotos (Cox et al., 20 12), inspection of early 20th century film footage (Cox, 2013), existence of clasts with unweathered surfaces that correspond to unweathered bedrock sockets (Williams and Hall, 2004), photographic evidence from 1900 to present (Hall et al., 2006), and modern trash pinned beneath boulders (Williams and Hall, 2004; Hallet al., 2006; Cox et al., 2012) all definitively show that movement of individual large boulders and of entire ridge sections has taken place within the last century. The most recent tsunami to affect the NEA region was caused by the 1755 Lisbon earthquake, and had a minimal impact on the locations of our study sites (Cox et al., 20 12). Unlike most ridge locations worldwide, which are subject to both storms and tsunami, our study sites cannot have experienced recent tsunami activity. Consequently, our sites are in a rare position to provide an in-depth examination of the sedimentology of known storm deposits. Our approach We had two major goals going in to this study: to quantify the sedimentologic parameters of the ridges, and to examine the relationship between bathymetry and wave amplification. First, we investigated correlations between grain size parameters and topography in an attempt to characterize storm deposits and better understand the depositional mechanisms creating them. Because of the focus on wave transport (discussed above), coastal boulder studies have often focused on the largest clasts (Williams and Hall, 2004; Scheffers et a!., 2008; Etienne and Paris, 201 0; Richmond et al., 201 1), and the sedimentology of these deposits has not yet been documented in toto. Our study is the first to document full grain size distributions and ridge morphologies, allowing us to quantify sedimentologic parameters. This is an important first step in understanding boulder ridges and how they are deposited. Second, we attempted to determine the effects of bathymetry on the generation of extremely large coastal waves. One of the most vexing problems associated with these deposits is how storm waves become large enough to transport clasts to such great heights and so far inland. Wave heights in the Northern Atlantic are reported to reach a maximum of 29 m (Hall et a!., 2006), but storm-emplaced boulders are found up to 40 m above sea level in both Ireland and Shetland. This discrepancy was given as one of the strongest arguments against a storm origin for the ridges (Scheffers, Kelletat, Haslett, et a!., 201 0; Scheffers, Kelletat, and Scheffers, 201 0). However, we are now confident in the role of storms in generating these deposits, so there must be some explanation for the wave-cliff height inconsistency. Reflection off cliffs and amplification by bathymetry are two important factors to investigate. Simple wave reflection can cause constructive interference, increasing wave height twofold. Additionally, preliminary models (Leaf Elliot' 13 and Ward Lopes, personal communication) indicate that a stepped bathymetric profile close to shore can increase wave height by more than two times the initial height. These two effects could be additive, potentially increasing wave height more than four-fold. We used bathymetric data from the Atlantic-facing coast of the Aran Islands in conjunction with wave models to shed some light on storm wave amplification by bathymetry. Study sites NEA boulder deposit locations (Figure 2.) share a number of common characteristics. Most important is their exposure to extreme weather. All locations directly face the open Atlantic and its storms. They are all treeless: areas closest to the shore have been stripped of vegetation by wave scour, and inland areas sustain only turf and (in some cases) low shrubs. Due to the storminess and exposed topography, deposit locations are uninhabited. They are uniformly barren, windswept, and rocky. Despite the broad similarities among the study sites, differences in site lithology, topography, and bathymetry cause profound differences among deposits at different locations. The density and orientation of joint systems has a major effect on the size and shape of clasts found in boulder accumulations. Topography and bathymetry have a significant impact on the type of deposit that can develop at a location, because they appear to affect both potential incoming wave strength and the maximum distance inland a wave can reach. The Aran Islands The Aran Islands (Inishmore, Inishmaan, and Inisheer) (Figure 3) span the mouth of Galway Bay, and their southwestern coasts face the Atlantic Ocean. The Aran coastlines are fairly linear and lack large bays and inlets. The islands are composed of Carboniferous limestone dipping about 3° to the south-southwest (Cox et al., 2012). Steep vertical cliffs characterize the western coasts, reaching heights of 80 min places (Williams and Hall, 2004). The limestone is interbedded with shale units that are preferentially eroded by wave action, leading to horizontal undercutting of cliff faces (Williams and Hall, 2004). These shale beds, along with vertical and sub-horizontal joint systems, allow wave action to form stepped cliff profiles on the western coasts. The joint systems also increase the ability of waves to quarry tabular blocks from the cliff edge (Williams and Hall, 2004). Figure 3. The Aran Islands, where data was collectedfrom 2008 to 2010. Figure from Cox et at. (2012). Many of these quarried blocks are incorporated into boulder ridges separated from the cliff edges by scoured platforms of varying width (Figure 4). Fresh sockets marking sites of recent block removal are often visible on the platforms, and occasionally a block quarried from a specific socket can be identified in the nearby ridge. Well-developed ridges are the most common type of deposit found on the Aran Islands. The seaward facing side of a ridge (which we refer to as the front) is steeper than the back, probably as a result of wave bores plowing clasts into the ridge accumulation (Cox et a!., 20 12). Both the ridges and platforms have a high coverage of the black lichen Verrucaria maura. Figure 4. View of a ridge front on the Aran Islands. Large clast pictured has a mass of -70 tonnes. Notice the scoured bedrock platform in fi·ont of the ridge. The islands are regularly subjected to intense winds and large waves. The Atlantic facing coasts have an onshore mean wind speed of about 9 meters per second (SEAl, 2003). The steep cliffs ofthe Aran Islands drop into very deep waters, limiting wave attenuation close to shore. The largest significant wave recorded by the M6 buoy, located 400 km west of the Islands, is 17.2 m on December 9, 2007 (Cox et al., 20 12). The Mullet peninsula The Mullet peninsula (Figure 5) extends about 23 km north-south along the west coast of Ireland, with its longest coastline facing the Atlantic. The peninsula is predominantly composed of the granitic Annagh Gneiss (Daly, 1996), with a well developed system of joints running approximately east-west. The ocean coast of the Mullet is topographically diverse, ranging from low-lying and gently sloping to tall vertical cliffs. Unlike the Aran Islands, large bays and inlets break up the coastline. Although there is very little detailed bathymetry for the area, the water depth must increase fairly rapidly offshore in order to limit wave attenuation that would prevent boulder emplacement inland. Similar to the Aran Islands, the Mullet Peninsula is extremely windy. The mean onshore wind speed (measured 50 m above sea level) ranges from 10 to 11 m/s (SEAl, 2003), which is higher than the mean speed is on the Aran Islands. I have been unable to find data on mean wave heights off the Mullet. Figure 6. View of the Annagh Head ridge facing north. People in center right are standing on the ridge crest. The ridge back and scattered boulder field lie to the East (not pictured), while the ridge front, oceanward boulder field, and scoured platform are seen to the west. Erris Head Nine km north of Annagh, Erris Head (EH) (Figure 5) has a very different set of features. The deposits are located on steep, craggy bedrock as far as 50 m from the edge of40 m high cliffs. Erris deposits do not have the well-developed ridge structure of either the Annagh or Aran deposits. They form discontinuous clusters only a few boulders thick and lack a distinct ridge face or crest (Figure 7). Behind these diffuse deposits, a turf covered slope rises gently but steadily. As far back as 60 m from the cliff edge, the turf is strewn with small (1-5 em long axis) clasts that must have been flung inland from the rocky cliff edge, because there are no inland outcrops to serve as pebble sources. The turf beneath many ofthese far-flung pebbles is yellowed but living, signifying recent clast emplacement and indicating that intense storminess is a regular occurrence. Figure 7. View of Err is head deposits facing south. 40 m cliffs located to the west, and long sloping turf fields (not pictured) to the east. Shetland Shetland is an elongate archipelago located 180 km north of mainland Scotland (Figure 2). Due to its small landmass and isolated location in the path of the northern Westerlies, Shetland's climate is characterized by frequent strong winds. Storms making landfall on Shetland in 1992 and 1993 created some of the most severe sea states of the 201h century (Hansom et al., 2008). Modeling of deep water waves and data from nearby buoys indicates that 160 km west of Eshaness, wave heights above 20 m occur about 1 00 times per year (Hansom et al., 2008). The steep coasts give way to deep waters, which means that large storm waves undergo very little attenuation as they approach the coast. Three types of boulder deposits can be found in Shetland: boulder beaches, diffuse boulder fields, and well-developed boulder ridges. The deposits on Shetland are sea. Behind the boulder accumulations, the turf is peppered with small clasts similar in size to those found thrown inland on Erris Head. As at EITis Head, there is no inland source for the scattered pebbles, and the living (albeit yellowed) grass beneath some pebbles indicates recent emplacement. Figure 9. View of the accumulations at Eshaness facing inland to the west. Notice that bedrock is almost completely covered by gravel, with larger clasts scattered throughout. Grind of the Navir The Grind of the Navir (GN) (Figure 8) is a 12m wide, 8 m tall gap between two rock towers that rise above a 10 m stepped cliff that leads out to the ocean (Figure I 0). The Grind allows waves to enter into a natural bedrock amphitheater. The jointed and stepped bedrock within the amphitheater is free of clasts, and while much of the outcrop is covered by the black lichen Verrucaria maura, there are many fresh sockets from recent quarrying of blocks. On the inJand end of the amphitheater, about 60 m from the Grind, lies a massive well-developed ridge complex (Figure 1 1) consisting of highly angular clasts, a product of orthogonal joint systems in the rhyolitic ignimbrite bedrock (Hall et al., 2008). The ridges are characterized by very steep fronts and impressive crest heights reaching more than 3m. Most of the clasts within the ridge complex are very fresh and lichen-free, with lichen cover appearing only on some of the back-ridge clasts. Vil/ians of Hamnavoe The deposits at the Villians (VH) (Figure 8) are a fusion of boulder ridge and diffuse field: a series of boulder clusters with a discontinuous front, separated by gravel accumulations (Figure 14). While the clusters are not visually striking, they are located about 20m above the high water mark (AHWM), and as much as 130m inland. Between the deposits and the sea is a series of scoured bedrock steps and platforms. The bedrock is covered by black lichen, and many of the deposit clasts have lichen cover as well. Directly behind the deposits, a boulder-strewn turf slope rises gently inland. The Burr The deposit at the Burr (BR) (Figure 8) differs from most other deposits in that its front descends into the ocean below the low tide mark such that it is always in contact with the sea (Figure 15). No more than 10m behind the beach lies a smalllochan ringed by boulders covered in the black lichen V maura, which is not found on the beach boulders. Between the lochan and the beach is a scattered boulder field of partially buried and very lichen covered boulders. Although the boulders on the crest have some lichen cover, the clasts on the beach front are clean of lichen. The beach clasts are also notably more rounded than clasts at all of the other Shetland locations. lichen cover, regardless of their location on the ridge, indicating that this deposit does not experience regular wave reworking. Methods Field methods During the 20 13 summer field season in Shetland, we conducted boulder ridge and cliff-top surveys, as well as measurements of clast size, imbrication, roundness, and lichen cover. We recorded a total of 16transects in Northmavine: three at the Burr, three at Eshaness, four at the Grind of the Navir, two at the Head of Stanshi, two at the Villians ofHamnavoe, and two at Unnamed Bay between the Burr and Head ofStanshi. We also recorded three transects at the Ness of Burgi on the southern tip of Shetland near Surnburgh (Figure 8). For each survey, we set up a transect line parallel to the average imbrication direction of clasts in the ridge front (Cox et al., 20 12). On multi-transect ridges, transect locations were evenly spaced along the length of the ridge. Transect line location and bearing were defined by GPS points at the cliff edge, at the fronts, crests, and backs of the main ridge and any secondary ridges, and at the end of the landward scattered boulder field (where present) (Figure 1) We surveyed the ridge and adjacent topography using a tape measure or laser rangefinder to determine distances, and a Brunton compass to measure associated angles (Figure 17a). We then used trigonometry to calculate ridge dimensions, horizontal distance from the ridge front to the sea's edge, and vertical distance from the ridge front to sea level. Finally, we systematically documented each site by taking photograph sets from the ridge front and from the crest, documenting the seaward, landward, and ridge­parallel views. At each survey site we collected clast size data using a line-count or ribbon-count method (Cox et al., 2012) (Figure 17b). We laid a tape along the transect line, and measured the X, Y and Z axes of every clast (with Y axis greater than 1 em) touched by the tape. On every clast with a Y axis greater than 40 em, we measured the imbrication direction as the trend and plunge of the maximum dip direction of the upper surface (Figure l7d). Additionally, when clast size and exposure permitted, we measured edge roundness using the device and protocol of Martin Kirkbride (2005), as amended by Ward Lopes (Figure 17c). In some cases, clasts were partially buried by other clasts in the ridge or, in the case of the scattered boulder field, by turf. When this occurred, we attempted to exhume the clast to measure all three dimensions. If this was not possible, we measured the two longest visible axes, and designated them X and Y. These clasts were omitted from later volumetric calculations. Many clasts had lichen on their exposed surfaces. On the basis that lichen could be a proxy for time spent in the ridge (Williams and Hall, 2004), we assigned them to one of three qualitative classes based on the amount of surface coverage: "Some" (<50% coverage), "Half' (-50% coverage), or "Most" (>50% coverage). We also made note of clasts with significant (>50%) cover of black lichen. However, given the subjectivity of the observers and the wide number of variables influencing lichen growth, we believe no reliable and broadly applicable conclusions can be drawn from lichen cover. Calculation of rock densities To calculate the masses of the clasts measured in the field, we needed to know the densities of the constituent lithologies. I measured the masses of hand samples collected on the Aran Islands and Shetland using a digital scale that was precise to the nearest gram. Due to the irregular shape of the samples, I measured their volume by water displacement. I placed a 5L bucket on the center of a large aluminum-foil cooking pan, and filled it with water such that the surface tension just barely prevented the water from spilling. I then slowly submerged a sample in the bucket, and caught the displaced water in the foil pan. I removed the bucket from the pan, moving very carefully so no excess water was spilled. I poured the captured water from the pan into a graduated cylinder and recorded the volume. This procedure was performed four times for every rock sample. The average of those four measurements was used to calculate the sample densities (Table 1). location Average Mass Density Vol l Vol 2 Vol 3 Vol 4 Rock Type (distinguishing Volume (g/cm"'3) (cm"'3) (cm"'3) (cm"'3) (cm"'3) feature) (cm"'3) Limestone Aran (bivalve) 1252 475 3 470 480 460 490 Limestone Aran (larger) 1764 691 3 675 695 690 705 Limestone Aran (smaller) 510 194 3 190 180 200 205 Rhyolite Srind of the Navi1 1044 423 2 430 420 400 Rhyolite The Burr (cobble) 1232 511 2 525 500 525 Rhyolite Geo of Ure 810 328 2 330 325 320 Sandstone Ness of Burgi 853 315 3 310 325 325 300 Table I. Rock densities calculated from samples collected in the field. Average volume is derived from 4 water displacement tests performed on each sample. Mass measured using a small scale that is precise to the nearest gram. Grain size analysis GRADISTAT (mass-based) analysis Part of the grain size analysis was performed using GRADISTAT v8.0, a macro- enabled Excel spreadsheet written for Excel2010 (Blott and Pye, 2001). GRADISTAT expects sieved data as input, so I had to prepare artificial sieve datasets by grouping the line count data into size classes. Using the mass of sediment in each size class, GRADIST AT calculates sample mean, sorting, skewness, and kurtosis using both the method of moments (Krumbein and Pettijohn, 1938; Friedman and Johnson, 1982) and the Folk and Ward graphical method (Folk and Ward, 1957). To prepare the clast size data for GRADISTAT, I first calculated the nominal diameter of each clast, which is defined as the diameter of a sphere with volume equal to that of the clast (Wadell, 1932): d II =2· Sorting aG Skewness SkG Very well (VW} <1.27 Very fine (VF) -1.0 to -0.3 Weii (W) 1.27 -1.41 Fine (F) -0.3 to -0.1 Moderately well (MW) 1.41 -1.62 Symmetrical (S) -0.1 to +0.1 Moderately {M) 1.62 -2.00 Coarse (C) -+0.1 to -+0.3 Poorly {P) 2.00 -4.00 Very course (VC} +0.3 to -+1.0 Very poorly {VP} 4.00 -16.00 Extremly poorly {EP) >16.00 Table 3. Modified ranges for characterizing sorting and skewness calculated using the Folk and Ward (1953) graphical measures as modified for metric units by Blatt and Pye (2001). Synthesis and standardization of field transect data To date, students working on this project have accumulated data during five field seasons on the Aran Islands, the Mullet peninsula in western Ireland, and Shetland. Each year, different tield teams collected the data, and several different students worked on the data post-fieldwork. Our methods evolved over time, leading to discrepancies in both data collection and analysis. As one of the aims of my thesis is to conduct a comparative analysis of all the sites studied so far, the initial challenge was to collate all existing transect data into one master file, and to standardize all calculations. For example, previous workers used different formulae to apply the tide corrections necessary for calculating the heights of ridge fronts above the high water mark (HWM) and the ridge fronts' horizontal distance inland. To directly compare the characteristics of different transects, corrections of those characteristics must all be calculated the same way. Thus, it was necessary to create and apply standard formulae for correcting both the horizontal and vertical distances. For both cotTections, it was necessary to calculate the tide height at the time of field measurement. Using tide tables, we found the high tide and low tide that bracketed the time of measurement and calculated the tidal rate of change (meters per hour) for It is important to note that although the correlations with height AHWM, horizontal distance, and coastal slope are statistically significant, the data are very scattered and show lots of variability (Figure 20). So while we certainly see descriptive trends in the data, prediction should be done cautiously, if at all. Grain size analysis Estimated grain size distributions differ substantially depending on the calculation method used. For each transect, the mean and median calculated by GRADISTA T (i.e. grain sizes estimated from distribution of masses in different size classes) are larger than the mean and median of the grain population as represented by the raw line count data (i.e. grain sizes measured directly) (Appendix A). Furthermore, the standard deviations from the mass-based analysis are consistently smaller (i.e. the populations appear more sorted) than standard deviations derived line count. The line count grain-size distributions vary from poorly to moderately well sorted, whereas those calculated using GRADIST AT range up to very well sorted. Most transects are unimodal, but there are a few cases with indications of bimodality, including: AG2, BR l, BR2, AH2, Z8, GN4 (Figure 21 and Appendix A). A few other transects also have slight bimodality (NB 1, Z7, Zll, Zl5, Zl9), but I cannot confidently label them as such because they have such small N. Within one ridge, both unimodal and bimodal transects can be present. Sorting values from mass based analyses range from poorly to very well sorted, while sorting from line count analyses ranges from poorly to moderately sorted (Appendix A). Fun wave was unable to model any sort of realistic wave-cliff interaction using the shorter periods. In all tests, the wave trains would quickly stabilize to a steady state where the amplitude progressively decreases away from the origin, eventually reaching zero at the cliff face. There was no wave reflection and no cliff overtopping, which contradicts what we see at the cliffs in the Northeastern Atlantic region. It is not entirely clear whether we were using the wrong settings in Funwave, but after reviewing the results of our tests, Jim Kirby believed that we were not seeing a model error. Instead, the progressive loss of wave height was due to "dissipation dominated conditions" (Jim Kirby, personal communication). In the field, however, conditions at the cliffs of the Aran Islands do not appear to be "dissipation dominated." In fact, wave amplification was visible at the cliffs every day during the relatively calm summer season. Therefore, it is more likely that the boussinesq Funwave model simply is not the appropriate model for the conditions we were attempting to replicate. Unfortunately, due to time constraints, we did not complete the work with Funwave. Discussion Relationships between ridges and their location Some of the relationships we find in the data are intuitively obvious. As one expects, ridge height, ridge width, and the average mass of the five largest clasts (AM5) decrease with increasing height AHWM. The negative correlation between these parameters and coastal slope also falls in line with our expectations. It is surprising however, that the height, width, and the AM5 increase with distance inland (Figure 20). Explaining this trend is the hardest puzzle in this study. Inland transport of large clasts is not surprising in itself, as bores generated by green water overtopping cliffs maintain high horizontal velocity far inland from the cliff face (Ryu et al., 2007) (Figure 22). According to Ryu et al., (2007) "this dominant horizontal motion of water could generate a large horizontal load on any objects located on the deck. " (p. 559). Although this can explain the existence large clasts far inland, it does not explain the inland coarsening seen in Figure 20. Figure 2 2. Numerically modeled distribution of bore velocity after cliff overtopping. Time (t) refers to time after overtopping. The Ryu et a/. (2007) study is based on a scaled model, but the patterns found can be related to real waves through Froude scaling. Adapted from Ryu et a/. (2007). There is significant variability for any given distance inland, height AHWM, or coastal slope. So, as we evaluate the implications of the positive correlation between the horizontal distance and the AM5, it is important to also consider the noise in the data. Not all of the largest AM5s occur far inland. In fact, transects with an AM5 greater than 5 t have horizontal distances AHW ranging from 18 m to 210 m. A similar spread is evident in transects with AM5s less than 1 t, which are found as close as 2 m to the ocean at HWM, and as far inland as 260 m. A mixture of both small and large AM5s is found across the entire spectrum of horizontal distances. It appears therefore that the movement and final depositional location of similarly sized clasts will vary case by case. Sometimes large blocks will make it far inland and sometimes not, probably determined by complex interactions between incoming bores and backwash, and variability in location topography. Clast movement will largely be controlled by the relative strengths of these mechanisms, which will undoubtedly fluctuate at different times and locations. Yet the complexity and variability surrounding our data need not prevent us from identifying geomorphologic settings that might promote the inland transport of larger clasts. The relationship between coastal slope and AM5 provides the best starting point for this exploration, because coastal slope factors in both the vertical and horizontal components of location. Most transects with AM5s > 5 t have low coastal slopes: they are never located more than 20 m AHWM, but are found as far inland as 210 m. Once slopes exceed 1, the AM5s drop to mostly < 2 t (Note that there are 3 transects with AM5s above 2 t. The variability is still present.). Transects with the largest vertical components (slopes -7), have AM5s < 0.5 t. Evidently, large heights AHWM present a difficult obstacle to the generation of bores that can quarry and transport large clasts inland. Conversely, a bore that does not have to overcome a significant vertical component is likely to maintain strength over large horizontal distances, pushing large clasts far inland. Bores of such strength will not occur regularly, so once large clasts have been pushed far inland, they will be relatively safe from further movement. Unfortunately, there are no studies in the literature that directly investigate the impact of location geomorphology on the distribution and accumulation of many clasts over time. The work by Perez-Alberti eta!. (20 12) is a rare example of an investigation into relationships between geomorphology and boulder deposits, but their study is not particularly relevant to ours. They studied deposits located on bedrock platforms that slope gently into the ocean, and are primarily concerned with how deposits armor platforms against further weathering. The most relevant work is wave tank experiments performed by Hansom et al. (2008), but they offer explanations only for transport mechanisms acting on individual clasts, and do not consider transportation distance or deposition location. There are no existing studies of how clasts accumulate to form ridges. Future work, therefore, should include experiments that test the transportation and deposition of many clasts over time, which could shed light on the processes and interactions that result in different ridge morphologies and average clast sizes. Grain size analysis Discrepancies in population statistics Grain size distributions provide information about provenance, transport history, and deposition (Folk and Ward, 1957). Different methods, including sieving, point counting, and line counting, are used to estimate the size distribution of sedimentary populations, and it is known that these methods do not always yield directly comparable results (Rosenfeld et al., 1953; Friedman, 1958). Due to the importance of grain size distributions in sedimentologic analysis, several studies have investigated the relationships between parameters derived from sieving (i.e. mass-based determination of size distribution) and those derived from direct grain counting. Rosenfeld et al. (1953) showed that that "uncorrected thin-section analysis yields coarser sizes than sieving," but were unable to find a constant relationship between the means of the two analytical techniques. Friedman (1958) also determined that thin section counting yielded coarser median grain sizes, but in addition found a positive correlation between the medians derived from the two analyses. These studies, and many others like them (e.g. Friedman, 1962; Friedman and Johnson, 1982), found that the discrepancy between mass-based and point count analysis was not egregiously large, and that mass-based analysis could function as a suitable proxy for direct measurements of grain sizes. Although they made use of point count data, their results are broadly applicable to analysis of grain-size data because both line count and point count methods involve direct measurement of randomly selected individual grains. Consequently, most studies of unconsolidated sediment have performed grain-size analysis by sieving, because it is faster and easier. We therefore decided to run our clast-count data through GRADISTA T, calculating the population statistics based on distribution of masses in grain-size classes, because we wanted to directly compare our data to published mass-based distributions from other sedimentary deposits. count. Differences between standard deviations (6so) are as large as 2 em, a difference that could change a sorting determination of "very well sorted" to "poorly sorted." Although we see no correlation between the statistics from the mass-based analysis and the statistics from the line count analysis (Figure 23), there is a strong positive correlation between mass-based means and 6GeoM (Figure 24). The same holds for mass-based medians. We see a negative correlation between 6Med and median clast sizes, as well as between 6GeoM and mean clast sizes. Interestingly, when we compare standard deviations and 6so we find a more significant positive correlation with the clast size standard deviations and a less significant negative correlation with the mass-based standard deviations (Figure 24). than our mass-based analysis. Additionally, it is strange that we see such a large difference between the results of our two analyses. We would expect thin section count analysis to deviate more from sieve analysis than the line count. Thin section analysis involves extrapolating a three-dimensional object from a two-dimensional one. The long axis may not be correctly identified, which leads to an incorrect estimation of grain volume. Additionally, the comparison is not direct: the grains measured in thin section are not the same grains run through the sieve. In contrast, the clast data we entered into GRADIST AT came directly from the line counts, so there cannot be disparities due to different sample populations. Therefore, our analyses should theoretically be more similar than the analyses compared in the Friedman (1958) and Rosenfeld eta!. (1953) studies. Although the correlations between the D.s and the line count population statistics possess p values indicating significance at the 99% level, we find that the correlations become less significant when we separate the data by location. For example, the correlation between DoMed and median clast sizes has a p-value of 0.6449 for the Inishmore transects. Separation by location does not have the same effect on correlations between D.oeoM and DoMed and the mass-based population statistics. For example, the correlation between DoMed and the mass-based medians from Inishmore has a p-value of <0.000 1. We interpret this to mean that D.GcoM and DoMed depend mainly on the distribution of mass in the grain size population. If a substantial proportion of the mass is located in a few large clasts, the mass-based analysis fails to function as a suitable proxy for direct measurement of grain-size population statistics. As the proportion of the mass in those large clasts increases, so do the D.s in the population statistics. This conclusion is nicely illustrated by the transect EN 1. Five clasts (of the 43 1 counted in this transect) contain more than 70% of the mass, and EN 1 has the largest GeoM (91 em) of any unimodal transect (certain bimodal transects, such as AG2 and AH2, have much larger GeoM, because they contain a second population of larger grain size). For comparison, NB 1 has 17% of total transect mass in the five largest clasts, and has a GeoM that is about 1;2 that ofENl (Appendix A). EN1 is a model illustration of the mass distribution influence on population statistics. Clearly, we cannot use the mass-based grain-size analysis as a proxy for clast population statistics. Therefore, we should use only the line count distributions when we compare the ridge distributions to other sedimentary deposits, even if those analyses were performed using sieves. Line count distributions and comparisons to other deposits The majority of transect distributions are unimodal, which corroborates findings that storm deposits are unimodal (Goff et a!., 2004; Switzer and Jones, 2008). There are however, a few bimodal exceptions (see Figure 21 and Appendix A). The morphologies ofVH2 and AH2 provide the most convincing answers to the issue of bimodality. VH2 does not have a standard ridge shape, but is instead made up of two separate clast populations: a series of boulder clusters resting on a bed of gravel. Annagh Head has large boulders located in front of the ridge, some of which may have been pushed into the ridge front area of AH2. Some of the other transects share morphological characteristics, but their morphologies do not offer as clear-cut an answer as VH2 or AH2. Both GN4 and Z8 have a double crest, while GN4 and AG2 both have subsidiary ridges. Double crests and extra ridges might be indicative of changes in wave strength over time, which might also change the clast size favored for deposition. Ridges are long-lived formations, so it is very possible that changes in local climate, bathymetry, and topography could change the size of the clasts added to a ridge. BR1 and BR2 (the Burr) have no obvious morphological features that might account for their bimodality, but a change in storm regime could be a possible factor. The occurrence of both unimodal and bimodal transects within individual ridge systems suggests inherent variability in ridge sedimentology (Figure 21 and Appendix A). Although ridges at some locations show little variability in grain size distributions (e.g. Esha Ness and Head of Stanshi), transects at other locations reveal very dissimilar distributions along the length of the deposit (e.g. Grind of the Navir, the Burr, and Annagh Head). For example, AHl and AH3 are both right skewed unimodal distributions, while AH2 has a bimodal shape. In some cases, transects from ridges in similar geomorphic settings show similar distributions, most notably Eshaness and Erris Head. Both of those deposits are located on the highest cliffs of any of the study locations, and both are strongly right skewed with means close to 5 em. Unfortunately, only four transects exist for Eshaness and Erris Head combined, so we cannot confidently conclude that such distributions are characteristic of high-cliff ridges until more data are gathered. Before this study, detailed grain size analyses of other boulder ridges had not yet been performed. Other studies have measured clast size, but documented only the largest clasts (Go to et a!., 201 0; Etienne and Paris, 201 0). No equivalent systematic grain size collection exists, but we can compare our data to analogous deposits that have been more closely studied. The closest analogue to boulder ridges are boulder beaches, and although boulder beaches are not as well studied as finer-grained beaches, some literature is available (Emery, 1955; Bluck, 1967; Oak, 1984). Oak's (1 984) work is particularly useful, because she compared the population statistics of boulder beaches in Australia to the known characteristics of pebble and gravel beaches. She (1 984) found that the sedimentary characteristics of boulder beaches differ from those of finer-grained beaches. One of the most important differences was skewness. Oak found that boulder beaches have fine-tailed grain size distributions. This distinguishes them from finer-grained beaches, which tend to have coarse skewed distributions (Leeder, 1982). Although it appears no one has replicated her work, her conclusions are accepted in the literature (McKenna, 2005; Etienne and Paris, 201 0). Our distributions have a wide range of skewness, from "very fine" to "very coarse" (Appendix A). There are fewer fine-tailed distributions than coarse-tailed, which in turn are fewer than symmetric distributions. This might suggest that predominantly symmetrical distributions -with the potential for both fine-and coarse-tailed exceptions -distinguish boulder ridges from other coastal sediment deposits. There is, however, some division in the skewness based on location. Shetland has more variability than both Inishmore (mostly symmetric) and Mullet (even split, symmetric and coarse), which could indicate that the skewness of boulder ridge populations is largely dependent upon location-based factors. Despite this difference, boulder beaches and ridges are not completely sedimentologically different. Both Oak (1984) and Etienne & Paris (20 1 0) noted that boulder beaches show consistent inland fining of clast size. Our data shows that inland clast size fining is also present in boulder ridges in Ireland and Shetland. In her thesis on Aran Islands boulder ridges, Danielle Zentner (2009), found that the fining trend on Inishmore ridges is significant at the 98% level. In contrast, pebble and cobble beaches are reported to have inland coarsening (Biuck, 1967; Oak, 1984). Conclusions The negative coiTelations between height AHWM/coastal slope and ridge width, height and AM5 are intuitively understandable (Figure 20). The positive correlations between these variables and the ridge distance inland are unexpected. While the correlations are all statistically significant, the data is noisy. The prediction of ridge characteristics based on location geomorphology is limited, as clast movement seems to vary on a case-to-case basis. Nevertheless, it appears as though low coastal slopes are a prerequisite for the inland transport of large clasts. Substantial discrepancies exist between the population statistics determined from mass-based analysis and the population statistics calculated actual clast measurements. GeoM and Med are strongly positively correlated with the means and medians derived from the mass-based analysis (Figure 24). The larger the proportion of total transect mass located in a few large clasts, the more the mass-based statistics deviate from the line count statistics. Therefore, when comparing our grain size distributions with other sedimentary deposits, we choose to base our comparison on the line count statistics. The majority of our studied boulder ridges are unimodal, which supports previous studies that characterized storm deposits as unimodal (Goff et al., 2004; Switzer and Jones, 2008). The bimodality of a few transects could be due to changes in wave strength brought on by local changes in climate, bathymetry, and topography. We find that the skewness of boulder ridge distributions seems to be largely location dependent. The whole range of possible skewness is represented in our data (Appendix A), but very few boulder ridges are fine-tailed. This distinguishes ridges from boulder beaches, which tend to be fine-tailed (Oak, 1984). Boulder ridges do however, exhibit a landward grain size fining trend akin to boulder beaches (Oak, 1984; Etienne and Paris, 201 0). GRAOISTAT Folk and Ward (1957) Line count graphical (em) graphical (em) Geometric Skewness Geometric Modality Median so Sorting Skewness Median so Sorting Transect N mean value Mean Difference between methods (em) Geometric Median so mean B 22 19 2.0 M -0.21 F so z 23 60 1.5 MW 28 26 0.5 z 24 40 B? 20 19 1.9 M -0.09 s 40 1.4 w 20 19 0.5 Note: See Table 3 for sorting and skewness shorthand definitions. Transect abbreviations: BR -The Burr, EN ­Eshaness, GN -Grind of the Navir, HS -Head of Stanshi, NB -Ness of Burgi, VH -Villains of Hamnavoe, XX­Unnamed Bay, AG-An Gleib, AH-Annagh Head, EH -Err is Head, Z-Inishmore. References Blott, S.J., and Pye, K., 2001, GRADIST AT: a grain size distribution and statistics package for the analysis of unconsolidated sediments: Earth Surface Processes and Landforms, v. 26, p. 1237-1248, doi: 10.1002/esp.261. Bluck, B., 1967, Sedimentation of beach gravels: examples from South Wales: Journal of Sedimentary Research, v. 37, p. 128-156. Cardozo, N., and Allmendinger, R.W., 2013, Spherical projections with OSXStereonet: Computers & Geosciences, v. 51, p. 193-205. Cox, R., 2013, 1930s movie "Man of Aran" as a documentary source for studying boulder transport by storm waves:. Cox, R., Zentner, D.B., Kirchner, B.J., and Cook, M.S., 2012, Boulder Ridges on the Aran Islands (Ireland): Recent Movements Caused by Storm Waves, Not Tsunamis: The Journal of Geology, v. 120, p. 249-272, doi: 10.1086/664787. Daly, J.S., 1996, Pre-Caledonian History ofthe Annagh Gneiss Complex North-Western Ireland, and Correlation with Laurentia-Baltica: Irish Journal of Earth Sciences, v. 15, p. 5-18. Emery, K.O., 1955, Grain size of Marine beach gravels: Journal of Geology, v. 63, p. 39­ 49. Etienne, S., and Paris, R., 2010, Boulder accumulations related to storms on the south coast of the Reykjanes Peninsula (Iceland): Geomorphology, v. 114, p. 55-70, doi: 10.1016/j.geomorph.2009.02.008. Folk, R., and Ward, W., 1957, Brazos River bar: a study in the significance of grain size parameters: Journal of Sedimentary Research, v. 27, p. 3-26. Friedman, G.M., 1962, Comparison of Moment Measures for Sieving and Thin-Section Data in Sedimentary Petrological Studies: SEPM Journal of Sedimentary Research, v. Vol. 32, doi: 1 0.1306/74070C36-2B21-11D7-86480001 02Cl865D. Friedman, G., 1958, Determination of sieve-size distribution from thin-section data for sedimentary petrological studies: The Journal of Geology, v. 66, p. 394-416. Friedman, G.M., and Johnson, K.G., 1982, Exercises in Sedimentology: New York, Wiley. Goff, J., Dudley, W.C., deMaintenon, M.J., Cain, G., and Coney, J.P., 2006, The largest local tsunami in 20th century Hawaii: Marine Geology, v. 226, p. 65-79, doi: 10.1 01 6/j.margeo.2005.09.0 17. Goff, J., McFadgen, B.G., and Chague-Goff, C., 2004, Sedimentary differences between the 2002 Easter storm and the 15th-century Okoropunga tsunami, southeastern North Island, New Zealand: Marine Geology, v. 204, p. 235-250. Goto, K., Chavanich, S.A., Imamura, F., Kunthasap, P., Matsui, T., Minoura, K., Sugawara, D., and Yanagisawa, H., 2007, Distribution, origin and transport process of boulders deposited by the 2004 Indian Ocean tsunami at Pakarang Cape, Thailand: Sedimentary Geology, v. 202, p. 821-837, doi: 10.101 6/j .sedgeo.2007 .09 .004. Goto, K., Miyagi, K., Kawamata, H., and Imamura, F., 2010, Discrimination of boulders deposited by tsunamis and storm waves at Ishigaki Island, Japan: Marine Geology, v. 269, p. 34-45, doi: 10.1016/j.margeo.2009.12.004. Goto, K., Okada, K., and Imamura, F., 2009, Characteristics and hydrodynamics of boulders transported by storm waves at Kudaka Island, Japan: Marine Geology, v. 262, p. 14-24, doi: 10.1016/j.margeo.2009.03.001. Goto, K., Okada, K., and Imamura, F., 2010, Numerical analysis of boulder transport by the 2004 Indian Ocean tsunami at Pakarang Cape, Thailand: Marine Geology, v. 268, p. 97-105, doi: 10.1 016/j.margeo.2009.1 0.023. Hall, A.M., Hansom, J.D., and Jarvis, J., 2008, Patterns and rates of erosion produced by high energy wave processes on hard rock headlands: The Grind of the Navir, Shetland, Scotland: Marine Geology, v. 248, p. 28-46, doi: 10.1 016/j.margeo.2007.1 0.007. Hall, a. M., Hansom, J.D., Williams, D.M., and Jarvis, J., 2006, Distribution, geomorphology and lithofacies of cliff-top storm deposits: Examples from the high­energy coasts of Scotland and Ireland: Marine Geology, v. 232, p. 131-155, doi: 10.101 6/j .margeo.2006.06.008. Hansom, J.D., Barltrop, N.D.P., and Hall, A.M., 2008, Modelling the processes of cliff­top erosion and deposition under extreme storm waves: Marine Geology, v. 253, p. 36-50, doi: 10.1016/j.margeo.2008.02.015. Kirkbride, M.P., 2005, Boulder edge-roundness as an indicator of relative age: A lochnagar case study: Scottish Geographical Journal, v. 121, p. 21 9-236, doi: 10.1080/00369220518737232. Kortekaas, S., and Dawson, a. G., 2007, Distinguishing tsunami and storm deposits: An example from Martinhal, SW Portugal: Sedimentary Geology, v. 200, p. 208-221, doi: 1 0.1016/j.sedgeo.2007.01.004. Krumbein, W.C., 1934, Size frequency distributions of sediments: Journal of Sedimentary Research, v. Vol. 4, p. 65-77, doi: 10.1306/D4268EB9-2B26-l lD7­8648000102Cl865D. Krumbein, W.C., and Pettijohn, F.J., 1938, Manual of Sedimentary Petrography: New York, Appleton-Century-Crofts. Leeder, M.R., 1982, Sedimentology: Process and Product:. Mastronuzzi, G., and Sanso, P., 2004, Large boulder accumulations by extreme waves along the Adriatic coast of southern Apulia (Italy): Quaternary International, v. 120, p. 173-184, doi: 10.10 16/j .quaint.2004.0 1.01 6. McKenna, J., 2005, Boulder beaches: Encyclopedia of Coastal Science, v. 61, p. 206­208, doi: 10.2307/1937439. Morton, R.A., Richmond, B.M., Jaffe, B.E., and Gelfenbaum, G., 2008, Coarse-Clast Ridge Complexes of the Caribbean: A Preliminary Basis for Distinguishing Tsunami and Storm-Wave Origins: Journal of Sedimentary Research, v. 78, p. 624-637, doi: 10.211 O/jsr.2008.068. Nott, J., 1997, Extremely high-energy wave deposits inside the Great Barrier Reef, Australia: determining the cause-tsunami or tropical cyclone: Marine Geology, v. 141, p. 193-207. Nott, J., 2003, Waves, coastal boulder deposits and the importance of the pre-transport setting: Earth and Planetary Science Letters, v. 210, p. 269-276, doi: 10.101 6/SOO 1 2-821 X(03)001 04-3. Oak, H., 1984, The boulder beach: a fundamentally distinct sedimentary assemblage: Annals of the Association of American Geographers, v. 74, p. 71-82. Perez-Alberti, A., Trenhai1e, A.S., Pires, A., L6pez-Bedoya, J., Chamine, H.I., and Gomes, A., 2012, The effect of boulders on shore platform development and morphology in Galicia, north west Spain: Continental Shelf Research, v. 48, p. 122­137, doi: 10.1 016/j.csr.201 2.07.014. Richmond, B.M., Watt, S., Buckley, M., Jaffe, B.E., Gelfenbaum, G., and Morton, R.A., 20 11 , Recent storm and tsunami coarse-clast deposit characteristics, southeast Hawai'i: Marine Geology, v. 283, p. 79-89, doi: 10.1016/j.margeo.2010.08.001. Rosenfeld, M.A., Jacobsen, L., and Ferm, J.C., 1953, A Comparison of Sieve and Thin­Section Technique for Size Analysis: The Journal of Geology, v. 61, p. 114-132. Ryu, Y., Chang, K.-A., and Mercier, R., 2007, Runup and green water velocities due to breaking wave impinging and overtopping: Experiments in Fluids, v. 43, p. 555­567, doi: 10.1 007/s00348-007-0332-0. Saintilan, N., and Rogers, K., 2005, Recent Storm Boulder Deposits on the Beecroft Peninsula, New South Wales, Australia: Geographical Research, v. 43, p. 429-432, doi: 10.1111/j.1745-5871.2005.00344.x. Scheffers, A., 2004, Tsunami imprints on the Leeward Netherlands Antilles (Aruba, Cura9ao, Bonaire) and their relation to other coastal problems: Quaternary International, v. 120, p. 163-172, doi: 10.1016/j.quaint.2004.01.015. Scheffers, A., Kelletat, D., Haslett, S., Scheffers, S., and Browne, T., 20 l 0, Coastal boulder deposits in Galway Bay and the Aran Islands, western Ireland: Zeitschrift fur Geomorphologie, Supplementary Issues, v. 54, p. 247-279, doi: 10.1 127/0372­8854/201 0/0054S3-0027. Scheffers, A., Kelletat, D., and Scheffers, S., 2010, Wave-Emplaced Coarse Debris and Megaclasts in Ireland and Scotland: Boulder Transport in a High-Energy Littoral Environment: A Reply: The Journal of Geology, v. 11 8, p. 705-709, doi: l 0.1086/656356. Scheffers, A., Kelletat, D., Vott, A., May, S.M., and Scheffers, S., 2008, Late Holocene tsunami traces on the western and southern coastlines of the Peloponnesus (Greece):, 271-279 p. Scicchitano, G., Monaco, C., and Tortorici, L., 2007, Large boulder deposits by tsunami waves along the Ionian coast of south-eastern Sicily (Italy): Marine Geology, v. 238, p. 75-91, doi: 10.1016/j.margeo.2006.12.005. SEAl, 2003, Wind Atlas:. Shi, F., Kirby, J.T., Harris, J.C., Geiman, J.D., and Grilli, S.T., 2012, A high-order adaptive time-stepping TVD solver for Boussinesq modeling of breaking waves and coastal inundation: Ocean Modelling, v. 43-44, p. 36-5 1, doi: 10.1016/j.ocemod.20ll.12.004. Suanez, S., Fichaut, B., and Magne, R., 2009, Cliff-top storm deposits on Banneg Island, Brittany, France: Effects of giant waves in the Eastern Atlantic Ocean: Sedimentary Geology, v. 220, p. 12-28. Switzer, A.D., and Jones, B.G., 2008, Large-scale washover sedimentation in a freshwater lagoon from the southeast Australian coast: sea-level change, tsunami or exceptionally large storm?: The Holocene, v. 18, p. 787-803, doi: 10.1177/0959683608089214. Wadell, H., 1932, Volume, shape, and roundness of rock particles: The Journal of Geology, v. 40, p. 443-45 1. Wentworth, C., 1922, A scale of grade and class terms for clastic sediments: The Journal of Geology, v. 30, p. 377-392. Williams, D.M., and Hall, A.M., 2004, Cliff-top megaclast deposits of Ireland, a record of extreme waves in the North Atlantic-storms or tsunamis?: Mar. Geol., v. 206, p.101-1 17. Zentner, D., 2009, Geospatial, hydrodynamic, and field evidence for the storm-wave emplacement of boulder ridges on the A ran Islands, Ireland: Williams College.