To Whom It May Concern:

Transcription

1 To Whom It May Concern: We submit this manuscript entitled Understanding long-term shoreline change through integration of historical aerial photographic, sedimentological, and computer modeling techniques: Lanikai and Bellows Beaches, Oahu, Hawaii for publication in your journal, Marine Geology. We feel this work is an original and relevant contribution to our knowledge of coastal sediment transport and a case study in integrating multiple diverse analysis techniques. The results of this study have great applicability to beach renourishment projects, shoreline sediment management, and to our understanding of the strengths and limitations inherent in common coastal analysis techniques. Thank you for your time. Sincerely, Christopher Bochicchio Charles Fletcher

2 1 Draft 04/06/ Title: Understanding long-term shoreline change through integration of historical aerial photographic, sedimentological, and computer modeling techniques: Lanikai and Bellows Beaches, Oahu, Hawaii Author names and affiliations: 1) Christopher Bochicchio Dept. of Geology and Geophysics, University of Hawaii at Manoa 2) Charles Fletcher Dept. of Geology and Geophysics, University of Hawaii at Manoa 3) Sean Vitousek Dept. of Geology and Geophysics, University of Hawaii at Manoa 4) Bradley Romine Dept. of Geology and Geophysics, University of Hawaii at Manoa 5) Tomas Smith United States Army Corp of Engineers, Honolulu Engineering District Corresponding author: Charles Fletcher: fletcher@soest.hawaii.edu Present/permanent address: Department of Geology and Geophysics 1

3 University of Hawaii at Manoa 1680 East-West Rd., POST 721 Honolulu, HI 96822, USA Abstract: Beach loss is significant to the economy, environment, and quality of life of many coastal regions, particularly in Hawaii. Understanding multi-decadal trends in shoreline change is a difficult and complex problem that benefits from the integration of multiple analysis techniques. Lanikai and Bellows beaches are located on the windward coast of Oahu Island, Hawaii, and separated by Alala Point. This shoreline has experienced long-term poorly understood changes in shoreline position ranging 70 meters of width over the last 80 years. In an effort to better understand long-term coastal sediment transport at this site and provide information for successful management, we integrate three commonly used techniques for studying shoreline change: 1) grain-size sediment trend analysis (GSTA), 2) hydrodynamic computer modeling (Delph 3D), and 3) historical aerial photoanalysis. Results of GSTA and hydrodynamic modeling are generally consistent with shoreline change observed in historical aerial photographs. It is shown that sediment has been shared between Lanikai and Bellows and that decadal-scale changes in the dominant trade wind direction are closely tied to shoreline change Keywords: coastal erosion, Hawaii, grain size trend analysis, historical shoreline analysis, Delph 3D model, trade wind direction 2

4 INTRODUCTION Beach loss poses a serious threat to the economy, ecology, and safety of many coastal regions. Over the latter part of the 20 th century nearly 70% of the world s beaches have experienced net erosion (Bird, 1985). Much of this is attributed to the combined influence on coastal sediment budgets of rising sea level and increasing shoreline development (NRC, 1995). On the island of Oahu, Hawaii, historical analysis of beach length shows 24% of all beaches have either narrowed or disappeared over a ~60 year interval (Fletcher et al, 1997; Coyne et al., 1999). The impact of this beach loss is particularly profound in Hawaii as sandy beaches drive a multi-billion dollar tourism industry that accounts for 60% the jobs in the state and represents an important element of cultural identity. Beach volume and shoreline position are largely governed by locally unique trends in longshore and cross-shore sediment transport. These are difficult to observe and predict over the long time scales needed to develop sustainable coastal management plans. Hence, where historical observations are available, it is important to investigate the processes driving shoreline change on poorly understood beaches. Lanikai Beach on windward (east-facing) Oahu is a developed shoreline threatened by long-term erosion that is poorly understood. Lanikai has experienced a series of decadal-scale erosion and accretion events producing >50 m changes in beach width over a 60-year period. The net trend has been erosional and consequently the total 3

5 beach length has decreased from 2.3 km to 800 m over the period from 1950 to Discussion regarding appropriate management of Lanikai Beach has continued for over 30-years without resolution. Central to this debate is the source and fate of beach sand, and specifically whether sand is exchanged around a rocky headland marking the southern littoral cell boundary of Lanikai Beach with Bellows Beach (Figure 1). In this study, we test the hypothesis that littoral sediment transport occurring between Bellows and Lanikai beaches controls historical shoreline change at Lanikai. We examine the direction of this exchange and assess factors that have potentially altered sand transport over time. We integrate grain size trend analysis and hydrodynamic modeling (Delft 3D) and compare the results with a detailed review of historical shoreline change (derived from aerial photographs) to evaluate littoral sediment transport across the Lanikai-Bellows boundary and greater shoreline. We expand our analysis to include historical shifts in wind direction as a driving factor in shoreline change. Results indicate that sediment transport does occur around Wailea Point linking Bellows and Lanikai Beach and hardening of the Bellows shoreline has starved Lanikai Beach by impounding an apparently important sediment supply. We also find evidence that historical changes in wind direction have had a significant influence on sediment transport. These results also indicate that future integration of sediment grain-trend analysis into shoreline change studies could be beneficial to coastal authorities tasked with managing poorly understood shorelines. This study is the first major reconstruction of shoreline dynamics along the Lanikai-Bellows Beach and represents the first application of sediment grain size trend analysis (GSTA) to studying shoreline change. 92 4

6 REGIONAL SETTING The Lanikai-Bellows region encompasses roughly 4.3 km of coastline along a broad, embayed headland marking the boundary between Kailua and Waimanalo Bays. The study area extends to rocky Alala Point in the north, and the mouth of Waimanalo Stream in the south. Basaltic Wailea Point, in the center of the study area, marks the boundary between northeast facing Lanikai Beach and southeast facing Bellows Beach. Northeast trade winds are dominant with an average speed of kts over 90% of the summer season (April-September) and 50-80% of the winter season (October-March) (Harney et al., 2000). Trade wind waves dominate during summer months, with average deepwater significant wave heights of 1-3 m and periods of 6-9 s. During the winter, refracted swell from the North Pacific reach significant wave heights of 4 m with periods of s. Breaking face heights at the beach are substantially lower (<0.5 m) as a shallow reef crest and the twin Mokulua Islands dissipate most incoming energy. Typical tidal range in Hawaii is <1 m. Landward of the shoreline at both Lanikai and Bellows are unconsolidated carbonate marine and dune sands (Grossman and Fletcher, 1998; Harney and Fletcher, 2003). A 0.5 to 1.0 km wide reef flat fronts the majority of the site in water generally 1.5 to 3.5 m deep. Three large sand fields extend from the beach to the reef crest, containing a total of 130 x 10 3 m 3 of sediment with average thicknesses of m (Bochicchio et al., 2009). A thin veneer of sediment, occasionally observed with ripple marks, is found over parts of the reef flat. Seaward of the Mokulua Islands the fore-reef slopes to >20 m depth. 5

7 Beaches at Lanikai and Bellows are narrow with gentle slopes and made up of poorly sorted medium to fine-grained calcareous sand (Noda, 1989). Changes in beach volume tend to be related to chronic fluctuations in alongshore sand transport and sediment deficiencies, rather than event-based erosion because the offshore reef platform diminishes incoming swell (Fletcher et al., 1997). Currently, the northern and southern regions of Lanikai as well as northern Bellows Beach are without a beach and protected by seawalls. Noda (1989) investigated transport processes at Lanikai and stated that longshore transport is responsible for substantial historical shoreline change at Lanikai Beach despite a relatively mild wave climate. Noda found no evidence of sediment transport occurring around Alala Pt. to the north, indicating that the Kailua-Lanikai cell boundary is closed. Two sandbars on the southern Lanikai shoreline, located 15 and 30 m offshore of the sea walled coast, corresponds with the node and anti-node of the mean incoming wave (Lipp, 1995). Lipp concluded that strong wave reflection off the Lanikai seawalls is, to some degree, preventing the accretion of a beach MATERIALS AND METHODS To test the hypothesis that sand transport between Bellows and Lanikai cells controls shoreline change, we use grain size trend analysis (GSTA), hydrodynamic modeling, and historical shoreline change analysis Sediment grain size trend analysis Sample collection and analysis 6

8 A total of 214 sediment samples were collected on a grid surrounding Wailea Point (Figure 2). Spacing between sample sites varied from 37.5 m near Wailea Point, to 75 m within sand fields, and 150 m between sand fields. Samples were recovered from the ocean bottom using a sediment dredge, which removed between 10 and 30 cm of the surface sediment. Between 1000 and 2000 g of sediment were recovered in each sample. The upper 5 cm layer of sample within the dredge was discarded to reduce error caused by fine sediment potentially billowing out of the dredge mouth as it was pulled to the surface. Samples were GPS positioned within 4 m. Grain size distributions of each sample were based on the weight percent of each size fraction from standard sieve analysis method ASTM C 136 (ASTM, 2006). Sieve openings ranging between -2 and 5 Ø at 0.5 Ø intervals. Mean size, sorting, and skewness were calculated for use as parameters in the trend analysis General background on method Spatial trends in the grain size of surficial sediments are a direct result of natural sediment transport processes (Russell, 1939; McCave, 1978; Swift et al., 1972; Harris et al., 1990). These trends are primarily the effect of transport processes selectively sorting and abrading sediment in the direction of transport (McLaren and Bowles, 1985; Gao and Collins, 1992; Le Roux and Rojas, 2007). Using mean size, sorting, and skewness, four trends have been found to be reliable indicators of transport direction (McLaren and Bowles, 1985; Gao and Collins, 1992; Gao et al., 1994; Le Roux 1994b) Trend 1: finer, better sorted, and more negatively skewed 7

9 Trend 2: coarser, better sorted, and more positively skewed Trend 3: coarser, better sorted, and more negatively skewed Trend 4: finer, better sorted, and more positively skewed Accordingly, transport pathways can be identified if a series of sediment samples follows one of these. Type 2 and 3 trends show distinctive coarsening along the direction of transport that at first appear counterintuitive. Type 2 and 3 trends are interpreted as indicators of more energetic transport processes in which a majority of fine-grained sediment is removed, thus creating a thin and coarse lag deposit that shields underlying deposits that have not been winnowed. This coarse upper layer may be mixed with underlying fine sediments during sampling, which results in an overall finer-grained texture upstream of the transport direction (McLaren and Bowles, 1985). GSTA encompasses a range of techniques for recovering net transport direction from naturally sorted seafloor sediments by identifying grain size trends in sediment samples collected around an area of interest. McLaren and Bowles (1985) first proposed a one-dimensional methodology to accomplish this task. This was followed by a number of two-dimensional approaches (e.g. Gao and Collins, 1992; Le Roux, 1994b,c; Asselman, 1999; Rojas et al., 2000; Rojas, 2003). These methods have been used to characterize sediment transport for a variety of engineering, environmental, and sedimentological investigations. In this study we apply two separate methods put forth by Gao and Collins (1992) and Le Roux (1992) to a dataset collected offshore of Lanikai and Bellows beaches. These two methodologies use significantly different mathematical approaches for 8

10 locating trends in sediment size data, yet are shown to detect sediment transport at similar spatial scales (Rios et al., 2002). An overview of each method is provided below to highlight the differences, provide an instructive reference, and aid in discussion of the results. Likewise, the respective authors of each method provide full descriptions in Gao and Collins (1992) and Le Roux (1994b). The Gao-Collins method is described in more detail using practical examples (Appendix A), because current publications on this method are limited to theoretical application Gao-Collins and Le Roux methodologies The method put forth in Gao and Collins (1992) determines sediment transport direction by comparing grain size parameters among a group of sampling sites. Parameters at each site are compared with those of neighboring sites within a predefined characteristic distance. The characteristic distance is defined as the spatial scale over which transport is expected to occur in the study area, generally given as the maximum interval between any two adjacent sampling sites. This study uses a characteristic distance of 200 m, which reflects the spatial scale of transport processes anticipated for this region and maximum distance between potentially related sites. In every case where either Trend 1 or Trend 2 is identified, component vectors with the unit length (i.e. equal to 1) are drawn in the direction of the neighboring site (Figure 3A). Summing all component vectors at each site produces a single vector referred to as a transport vector (large arrow in Figure 3A and 3B). Component vectors are relevant only in terms of direction. Their lengths do not reflect differences in grain size parameters or distance between points. As all component vector lengths are equal, the number and direction of neighboring sites 9

11 showing a positive transport trend determine both the direction and length of the resulting transport vector. Determining transport vectors for every point produces a field of transport vectors (Figure 3B), which can be filtered to reduce noise and reveal the dominant trends, by averaging the vector at each site with surrounding transport vectors (Figure 3C). Details of the steps and calculations used in Figure 3 are included in Appendix A. The method of Le Roux (1994b) functions by comparing grain size parameters of a central site with the closest four neighboring sites in all cardinal directions (i.e. one site is selected from the North, East, South, and West quadrants) (Figure 4A). The Le Roux method searches for all four trend types individually, producing a vector field of transport for each trend. Trend determination begins with the normalization of all three grain size parameters between all five sites. These values are combined into a single value (E) representing the strength of transport along that axis. The process of normalizing and combining the parameters is modified in a manner depending on the trend type. For example, Equation 1 is used in the case of Trend 3, where all parameters are expected to decrease along the direction of transport min ( mn mn ) + ( var var ) E = min min mn max mn min varmax var Equation (1) min + sk max sk ( sk sk ) min In this process, sites with the smallest values receive the highest score (E) indicating stronger transport potential in the direction of that site. Conversely, to achieve 10

12 the same effect with Trend 1, Equation 1 must be modified so that increasing mean grain size results in a lower value of E. This is done simply by subtracting from the normalized mean size parameter (Equation 2) E = mn sk sk max min mn max ( sk sk ) min min ( mn mn ) + ( var var ) min var var max min min Equation (2) Similar adjustments are made to the normalized skewness parameter for Trend 2 and the normalized variance parameter for Trend 4 to so that increasing values on these parameters result in higher values of E. Values (E) are defined for every site (Figure 4B) then the value of the central site is subtracted from each adjacent site and the relative difference between sites is used to define the length of component vectors, which are summed to produce a final transport vector (Figure 4C). This process is repeated at every site to produce a field of transport vectors for each trend type. Trend 1 results are shown in Figure 4D. Commonly, the strongest vectors from each trend type are incorporated into a final vector field. The Watson (1966) non-parametric test is used to ensure that the final transport vectors are sufficiently non-random before smoothing the data to reduce noise (Le Roux et al., 2002). The Gao-Collins and Le Roux methods both determine transport direction by searching for predefined trends between a single site and adjacent sites, but Gao-Collins uses only Trends 1 and 2, while Le Roux checks for all four trends. The Gao-Collins method checks a variable number of sites (all those that fall within the characteristic distance), while Le Roux only uses a central site and four adjacent sites. With Gao- 11

13 Collins, direction of transport is determined by relative position of all neighboring sites showing a trend to the central site, with transport occurring in the direction of the most trend positive sites. In contrast, using Le Roux, transport direction and strength is determined from the calculated difference between the actual grain size parameters of all five sites. Both methods have been shown to give comparable and informative results (Rios et al, 2002) Computer hydrodynamic model (DELFT 3D) The Delft3D-FLOW module (v used here) solves the unsteady shallowwater equations with the hydrostatic and Boussinesq assumptions. In 2D mode the model solves two horizontal momentum equations (see Eq. 3-4), a continuity equation (Eq. 5) and a transport (advection-diffusion) equation (Eq. 6) shown below: τ F ( + ) = 0 (3) u u u η bx x u u u v g fv ν e 2 2 t x y x ρw( h η) ρw( h η) x y 263 τ F ( + ) = v v v η by y v v u v g fu ν e 2 2 t x y y ρw( h η) ρw( h η) x y (4) 264 η [( h+ ) u] [( h ) v] u η v η = 0 (5) t x y [ hc] [ huc] [ hvc] c c + + = h DH + DH t x y x x y y (6) where u and v = the horizontal velocities in the x and y directions respectively; t = time; g = gravity; η = free surface height; h = water depth; f = coriolis force; ρ w = density of 12

14 water; τ b = bed friction; F = external forces due to wind and waves, ν e = horizontal eddy viscosity; D H = horizontal eddy diffusivity; and c = concentration of suspended sedimen t. The equations are solved on a staggered finite difference grid using the Alternating Direction Implicit (ADI) method after Stelling (1984). In this study the Delft 3D model is employed to examine the potential for different transport regimes developing under changing forcing conditions. This element of the study focuses on trade winds, as it is the most persistent type of forcing on Oahu s windward shore and most likely to determine equilibrium shoreline conditions. Figure 5 shows a 58-year time series of trade wind direction recorded at Kaneohe Marine Corps, located on the coastline approximately 9 kilometers north of the study area. These data show periodic shifts in trade wind direction that persist over decadal-scale time periods and are in some cases rapid (e.g. 1964, 1974, and 1987). Changes in trade wind direction were first documented by Wentworth (1949) and implicated as a possible factor in shoreline change in Lanikai in a report by Noda (1988). This study is the first to extend the directional dataset presented by Wentworth (1949). The exact cause of these directional shifts is not currently understood. Using the range of observed wind directions, this study uses Delft 3D to model the potential influence that changing wind direction could have on sediment transport in the Lanikai region. The model was calibrated using current and sea-level data collected by two acoustic doppler velocimeters deployed from August 10 th to September, 12 th 2005 on the southern and northern bounds of the study area (Figure 6). The model parameters included wind driven currents, tidal forcing, open ocean waves, and wave-driven currents. Ocean swell direction and height was simulated using a representative dataset 13

15 from a deepwater directional wave buoy located 2 km north-east of the study area in Kailua Bay (National Data Buoy Center number 51001). Tidal forces were modeled using standard harmonic components. Separate model runs used directional extremes from the historical dataset to simulate time periods when north-east (51 ), east-north-east (71 ), and east (85 ) wind conditions dominated Shoreline change analysis Analysis of shoreline change draws from data collected by Romine et al. (in press). Historical shoreline positions were hand digitized from survey quality aerial photos and T-sheets acquired in: 1911, 1928, 1949, 1951, 1959, 1963, 1967, 1971, 1975, 1982, 1988, 1989, 1996, and Distortion errors from scanning the photos were corrected (Thieler and Danforth, 1994), and following the methodology of Fletcher et al. (2003), all photos were orthorectified and mosaicked using software from PCI Geomatics, Inc. Seaward and landward boundaries of the sub-aerial beach were defined as the position of mean lower low water (MLLW, using the toe of the beach, or base of the foreshore as a proxy) (Bauer and Allen, 1995) and the vegetation line. Horizontal error in shoreline position was ± m RESULTS 4.1. Textual and transport trend analyses Sediment texture over much of the study area is characterized by distinct, isolated zones of varying size. As a whole, sediments offshore of Lanikai tend to be coarser (Figure 7), more poorly sorted (Figure 8), and positively skewed (Figure 9) than those at Bellows. 14

16 While sediment textures directly adjacent to shore tend to be finer and more negatively skewed along the entire sample area. Offshore of Wailea Point, in the central portion of the study area, sediments are generally finer, better sorted, and more negatively skewed towards the tip of the point. However, closer examination of the entire study area shows a close juxtaposition of alternating sediment textures indicative of lag and lead deposits. Results of the Gao-Collins (Figure 10) and Le Roux (Figure 11) methods indicate the direction and relative probability of sediment transport. A fundamental difference between the two methodologies is well illustrated by the smooth appearance of the Gao- Collins results and the noisier appearance of the Le Roux results. As described in the methodology, the Le Roux method is more sensitive to small differences in grain size and to small-scale isolated trends than the Gao-Collins method. Results of the two methodologies generally agree, with the only major exception being the box labeled A in the southern part of the study area, where results differ considerably. In A, the Gao- Collins results show primarily north-to-northeast trends, while the Le Roux results show an opposing southeast trend converging with a north-to-northwest trend Within 100 m of the Bellows shoreline, sediment textures alternate between coarse-positively skewed and fine-negatively skewed with all sediments becoming better sorted to the north (box B). Results from the Le Roux method show a majority of transport to the north and Gao-Collins also shows a consistent northern trend adjacent to Bellows Beach. Offshore sediment immediately south of Wailea Point (box C) becomes finer, better-sorted, and more negatively skewed toward the north, which is the signature of 15

17 Type 1 transport. Gao-Collins results indicate uniform northern transport of sediment from Bellows toward Wailea Point. Similarly, Le Roux results show northwesterly transport toward Wailea Point where it meets an opposing transport trend. This is mirrored to the north in box D, where sediment becomes finer, better-sorted, and more negatively skewed toward the south. Resulting transport vectors in D from both Le Roux and Gao-Collins methods are southeasterly and directly oppose transport in B. Near the northern slightly embayed portion of Wailea Point (box E), sediment within 250 m of southern Lanikai Beach shows two distinct textures. Nearshore sediments are finer, better-sorted, and more negatively skewed than sediment farther offshore. This contrast in sediment texture produces onshore and southeasterly transport vectors in the both Gao-Collins and Le Roux methodologies. To the north, both sets of results show an opposite northwesterly trend in box F along Lanikai Beach. Sediment at F tends to be coarse, poorly sorted, and positively skewed relative to areas to the south. In general, transport trends in F and E show divergence between northerly and southerly transport. Similar divergence occurs between A and B, while both results show convergence near Wailea Point between C and D. Le Roux transport vectors seem to indicate a gyre-like circulation pattern across C and D Shoreline change analysis Changes in shoreline position are visible in historical aerial photography (e.g. Fletcher et al., 1997, Romine et al., in press). Historical shorelines representing major shifts in position are overlain on a modern (2005) aerial photograph in Figures 12 and 13. Plots 16

18 show relative shoreline position over time along transects A through G centered on sites with the greatest shoreline movement. Lanikai Beach (Figure 12) shows multi-decadal trends of either accretion or erosion, indicated on transects B, C, and D. Shoreline position appears relatively stable between 1912 and 1949, although this could be due to under sampling (only 3 shorelines). In southern Lanikai (transect D) from 1949 to 1967 the shoreline accreted significantly, adding approximately 60 m of new land. Beach to the north (transects C, B, and A) was either stable or lightly accreting over the same period. After 1967, accretion turned to erosion until halted by seawalls in To the north, accretion intensified and central Lanikai (transect C) added approximately 30 m of new land. This ended by 1987 and the beach has chronically eroded since. At transect B accretion has persisted and accelerated. At transect A the shoreline was relatively stable throughout much of the 20 th Century until 1975 and has subsequently eroded, culminating with the construction of seawalls in the early 1980 s. Figure 13 summarizes chronic erosion at Bellows south of Wailea Point (transect E) persisting since the beginning of the dataset in To the south, transect F reveals an accretionary trend between 1916 and 1928, followed by erosion until 1961 resulting in the loss of approximately 40 m of land. At transect G, shorelines remain stable until a general erosional trend developed during the 1961 to 1962 time period. This continued until seawall construction in Erosional trends at transects E and F were also halted by seawalls Hydrodynamic modeling results 17

19 A hydrodynamic model is useful for envisioning nearshore currents that develop under various forcing conditions. In Figures we use a scale along the shoreline to define locations for discussion. Results show substantially different nearshore current configurations in the region when north-east (51 ), east-north-east (71 ), and east (90 ) winds a re used to force the model. The most noticeable effect of changing wind direction is the shifting location of longshore current convergence and divergence. Eastern winds (Figure 14) induce northward transport along the entire study area. East-North-East winds (Figure 15) create a divergence point in longshore flow near location 40 along the Bellows shoreline and induce a distinct gyre to the north of Wailea Point. Under northeast winds (Figure 16) the divergence point shifts to the north along the Bellows coast to location 10 and divergence develops in southern Lanikai between locations 90 and 100. In general, southern transport becomes more common along both shorelines as the northern component of wind direction becomes more prominent, which is reasonable considering the geometry of the shoreline DISCUSSION 5.1. Comparing Historical Shoreline Data with Results A reconstruction of historical shoreline change in the study area is presented in Figure 17. The color grid is interpolated from historical shoreline position data with color indicating the rate of shoreline change (m/yr) during a time period (horizontal axis) over a length of beach (vertical axis). A number of decadal-scale erosion and accretion events are visible in this data. The most prominent pattern is the aforementioned expansion and erosion of south Lanikai, followed by central Lanikai. From these 18

20 changes in shoreline position we can derive an empirical analog for historical littoral sediment transport direction. Arrows on Figure 17 show the direction of longshore sediment transport inferred from changes in shoreline position. It is apparent that there are coinciding system-wide shifts in sediment transport at certain times throughout the dataset. Figure 18 shows major shifts in wind direction delineated into eight periods (periods II through VIII). These same periods are overlain on Figure 17. The good agreement between changes in wind direction and shoreline change suggests that wind direction plays a significant role in driving sediment transport along the Lanikai-Bellows shoreline. The relationship between shoreline position and wind direction supports the results of the hydrodynamic modeling experiments, in which different wind directions showed the formation of different nearshore currents (Figures 14, 15, and 16). Presumably, these changes in current structure would have a notable effect on shoreline position, however in some cases the currents predicted in the model are not expressed in shoreline positions or are at odds with the observed shoreline history. This pairing of datasets allows evaluation of the model performance and, where observations match, provides insight to processes driving shoreline change. Gao (1992) suggests that the time period represented by GSTA is related to the deposition rate of sediment but the issue is still poorly understood. In coastal settings, transport patterns can shift on time scales ranging from days to decades, during which grain size distributions representing previous transport patterns are overwritten or mixed with new transport. Figure 19 summarizes both Le Roux and Gao-Collins results. The nearshore vectors show divergence near Lanikai transect 90 and Bellows transect

21 Convergence occurs near Lanikai transect 70 and in the vicinity of Wailea Point, as well as an overall northerly transport trend. This pattern is typical of trends observed since Period IV (Figure 17), suggesting the GSTA results are valid back to the mid-1960 s, coinciding with the onset of the modern erosive trend on south Lanikai Analysis of Historical Transport Trends This section takes the form of a timeline detailing observed shoreline changes in the context of wind direction, model predictions, and GSTA results. A generalized summary of historical sediment transport trends (Figure 17) and dominant wind direction (Figure 18) is presented for periods II through VIII in Figure 20. The goal for these data is to create a predictive, empirical model relating shoreline change to wind direction Period I Southern Lanikai eroded over the first half of this period (1911 to 1928), while the remainder of Lanikai and Bellows both accreted. Source material for accretion may have come from erosion in south Lanikai Period II Wind records during period II show that eastern winds dominated (Figures 18 and 20). Modeling indicates that sediment eroding from Bellows was likely transported around Wailea Point into south Lanikai, which shows accretion during this time. Northward transport along Lanikai Beach supported the supported the stable or accreting shoreline during this period. 20

22 Modeling under east winds (Figure 14) indicate northward longshore currents developed along Bellows and Lanikai Beaches and around Wailea Point. Current intensities are greatest in North Bellows corresponding to with the highest observed erosion rates during this period. Modeling also shows a slackened current in south Lanikai thus accounting for sand deposition. During this period Lanikai Beach volume was likely dependant on the delivery of Bellows sediment to balance the sediment budget in south Lanikai Period III Wind direction during this period shifted to the northeast, which resulted in a more complex system of littoral transport. The previously uniform transport from Bellows to Lanikai was interrupted as erosion slowed previously uniform in central Bellows and turned to accretion in south Bellows. However, erosion accelerated in north Bellows indicating the development of a divergent current. Current divergence also developed in central Lanikai, which eroded substantially during this period. South Lanikai experienced accretion rates in excess of 3 m/yr over this entire period as large volumes of sediment arrived from central Lanikai and north Bellows. Modeling using 71-degree winds fails to capture converging currents in south Lanikai or diverging currents in central Lanikai. It does however capture divergence centered on Bellows transect 35 near the observed erosion hotspot. Bellows Air Force Base underwent substantial development during and after World War II ( ) as the shoreline eroded. A series of coastal revetments appear in aerial photos where erosion rates were highest, and expand in later decades. These revetments 21

23 effectively impounded landward sediment reducing sources to down-drift littoral cells. South Lanikai especially reflects this event as some part of the source material for the large accretion event was likely provided by erosion in north Bellows IV Wind during this period shifts further to the northeast causing a relocation of convergence and divergence zones. Accreting and eroding areas essentially reversed during this period marking the start of long erosive trend in south Lanikai. Central Lanikai experiences accretion during this time,, suggesting the convergence zone has shifted north. Slight accretion on both sides of Wailea Point suggests the divergence zone in north Bellows has shifted to the south. A convergence zone seems to have also fo rmed in south Bellows where the beach undergoes accretion. The model, when forced with 51 degree winds (Figure 16), shows divergence near Lanikai transect 80, which is in agreement with historical observations. The model also indicates a southerly current moving around Wailea Point that could cause the accretion observed in north Bellows. The model fails to capture convergence in central Lanikai as there is no opposing current moving south from north Lanikai. Similarly, this model does not show convergence at the southern edge of the study area where accretion is observed. This period is the earliest in which GSTA results match observed shoreline changes. The dominant littoral transport trends in Figure 19 indicate convergence near Lanikai transect 70 and on the flanks of Wailea Point. GSTA results also show divergence near Bellows transect 30. These patterns closely match the observed 22

24 shoreline trends, suggesting that the GSTA results are incorporating grain size distributions that were established by transport patterns during this period V Wind direction during this period shifts back to the east (75 degrees), but shoreline changes show a unique pattern that was not previously seem. Erosion rates in south and north Lanikai increased during this period while central Lanikai accreted, likely gaining source material from the north and south. Mild accretion developed in north Bellows while sediment in south Bellows was seemingly dispersed to the north and south. Sediment buildup against the jetties stabilizing Waimanalo Stream suggest southerly transport along the Bellows coast. We might expect transport under 75 degree winds to be similar to the 71-degree winds during Period III, but there is no sign of shoreline accretion in south Lanikai. Possibly: 1) current patterns between 75 and 71 degree winds are sufficiently different to shift the convergence zone to central Lanikai and/or 2) hardening of both north Bellows and south Lanikai prevented sand accumulation along the shoreline. With a nearly perpendicular onshore wind, it is reasonable to expect small changes in wind direction to have a disproportional effect on current patterns. The reduced sediment availability from north Bellows and wave reflection off the hand shoreline may have prevented accretion at this time. Model results for 71 degrees (Figure 15) are the closest approximation to wind direction during this time period and match the observations poorly. GSTA results do 23

25 show sediment transport from south to central Lanikai, but transport patterns in Bellows are not well represented VI Wind direction makes a relatively brief northeastern excursion during this period causing subtle changes in the shoreline. As in period IV, in which wind was also more northeasterly, there is slight accretion in southern Lanikai adjacent to Wailea Point as sediment moved to the south. This excursion also marks the start of wide-spread erosion on Bellows beach, a northward expansion of the eroding area in south Lanikai, and a northward shift in the accreting area in central Lanikai VII Wind direction shifts back to an easterly inducing northerly transport along the entire study area. The slight accretion on the southern edge of Lanikai disappears sharply and sediment accumulated along the Bellows shoreline erodes. Farther evidence of increased northerly currents is the northern shift and general decrease in accretion rate of the accreting zone in central Lanikai. As the accreting area moves north, central Lanikai develops a strong erosive tendency and portions of south Lanikai experience complete beach loss. At this point the north Bellows, south Lanikai, and portions of central Lanikai shorelines are completely hardened by seawalls and revetments. This has reduced the volume of sediment available to the overall system and increased the reflectivity of the 24

26 shoreline. Sediment eroding from central and south Bellows is likely bypassing south and central Lanikai to accrete in north Lanikai VIII Similar to Period VII, north Bellows and south Lanikai continue to erode to the point of complete beach loss. The accreting zone in north Lanikai narrows by shifting its border further to the north. A slight accretion at the southern tip of Bellows reveals the possibility of southward transport, though the sediment volumes involved are relatively small. GSTA results indicate southerly transport in south Bellows and diverging transport on either side of Wailea Point. This transport might still be occurring, but with sediment volumes too small to cause notable beach accretion. In general, the modern trend of shoreline change appears compatible with the results of the GSTA procedure Future coastal change Since 1987 wind direction has steadily shifted to the east (Figure 18). If this trend continues, littoral transport along the Lanikai-Bellows shoreline will become increasingly northern. As this happens there is the chance that more sediment will be transported from the Bellows shoreline into the Lanikai system. However the presence of revetments in north Bellows will severely limit the amount of sediment available for transport. Similarly, seawalls along most of south and central Lanikai might make beach accretion problematic. A 1995 study of wave energy and shore-perpendicular bottom profiles at south Lanikai revealed a sand bar offshore containing approximately m 3 of 25

27 sediment (Lipp, 1995). The study also showed that height and period of incoming and outgoing waves was nearly identical due the highly reflective seawalls. The sand bar is located at a distance from shore near the exact anti-node between the incoming and outgoing waves (1/2 x the mean wave length) Directional changes in the wind record The cause of the decadal directional shift in trade winds is not fully understood. An explanation might lie in small shifts in the North Pacific High pressure system, north of the Hawaii islands. This system is already known to control the occurrence other aspects of Hawaiian weather (e.g. Kona storms). The influence of a decadal scale cycle, such as the Pacific Decadal Oscillation, could be affecting the North Pacific High and resulting in the directional shifts seen in the data. It is likely that these long-term changes in trade wind direction have had a similar impact on sediment availability on other windward shorelines. The primary obstacle to understanding is the relative scarcity of long-term continuous directional wind records for islands other than Oahu CONCLUSIONS This study integrates sedimentological data, hydrodynamic modeling, and historical shoreline analysis to investigate large-scale enigmatic changes in the Lanikai-Bellows shoreline, windward Oahu. The results show wind direction to be a major controlling factor on littoral processes. Most major accretion and erosion events can be linked to periodic shifts in the dominant trade wind direction. Hydrodynamic modeling was 26

28 employed to examine the exact mechanism linking wind direction and littoral transport. Modeled nearshore currents under easterly wind conditions matched observed historical shoreline change trends, however only portions of the model results matched historical observations under more northeasterly winds. Two different methods of grain size sediment trend analysis were applied to samples taken over a portion of the study area (le Roux, 1996 and Gao-Collins 1992). This study represents the first application of GSTA methods to specifically target coastal change. The results of both methods were in general agreement. By comparing GSTA results to the observed historical record we determined that grain size trends reflect the last major shift in sediment transport in the mid-1960 s. Revealing the relationship between wind direction and coastal change in the Lanikai-Bellows beach system represents a major step in the creation of a regional sediment budget and demonstrates the great utility of integrating multiple analysis techniques. There is strong evidence to support sediment transfer across Wailea Point, indicating Bellows Beach and Lanikai Beach are dynamically linked. It is also very likely that coastal hardening in response to erosion in both Lanikai and Bellows has worsened the over all erosion and further complicated the littoral sediment transport. Given more recent trends in both wind direction and shoreline position, northern sediment transport will likely dominate in the immediate future. However, the coastal armoring in place along both Lanikai and Bellows shorelines will likely negatively influence sediment availability and the possibility of accretion on the Lanikai shoreline

29 ACKNOWLEDGEMENTS Support for this project was provided by the United States Army Corps of Engineers: regional sediment management program, Hawaii Department of Land and Natural Resources, United States Geological Survey, University of Hawaii Sea Grant, and the Harold K.L. Castle Foundation

30 APPENDICES Figure 3 illustrates the Gao-Collins method. This appendix details the application of the Gao-Collins method to a synthetic dataset. Calculations associated with site 9 are included Table 1 Coordinates and grain size data for the calculations used in Figure * Trend type and component vectors calculated in table for site 9 only Step 1. Determine which sites are within the characteristic distance from the site of consideration (site 9). In the example, the characteristic distance is equal to two, which encompasses ten sites: 2, 5, 6, 8, 10, 11, 12, 13, 14, and Step 2. Check for the existence of trends 1 or 2 between the central site ( site 9) and the proximal sites listed above. Trend 1: sites 2 and 5. Trend 2: sites 6 and Step 3. Define component vectors r(x,y) i between the central site and those showing a transport trend. All component vector magnitudes are assumed to be equal (i.e. value = 654 1). When a trend is found the vectors are assigned to the site with the highest sorting 29

31 coefficient. As an example, calculations to determine the component vector from site 9 to site 5 are below: r ( x) 5 ( X = 5 X d 9 ) (1.5 2) = = r ( y) ( Y = Y ) (3 2) = = d where d is the distance between site 2 and the central site 0, given as: d = ( X 5 X 9 ) + ( Y5 Y9 ) = = Step 4. Sum all component vectors r(x,y) i to make a sum vector R(x,y): R( x, y) 17 = r( x, y 9 ) i i= 1 = [ ] Step 5. Repeat steps 1 4 on every site in the data set to define sum vectors at every site. Results of this step are presented in Table 2. 30

32 Table * Result of the average vector Step 6. Remove noise by averaging each sum vector with the neighboring sum vectors determined to be within the characteristic distance (i.e. sites identified in Step 1). This effectively serves as a low-pass filter with a search radius of 2. For site 9 this process is expressed as: R = av 1 1 ( x, y) 9 = R( x, y) 9 + R( x, y) q = ( k + 1) (10 + 1) [ ] ([ ] + [ ] ) where q is a list of all sites within the characteristic distance of site 0: q = [ ] 692 and k is the total number of such sites: k =

33 Thus, the final averaged transport vector at site 0 has an x-component of 0.09 and a y- component of Step 7. Convert average vector into azimuth direction Θ (exact formula will vary) and vector length VL: arctan Θ = 7 degrees VL ( R av ( x) ) + ( R ( y) ) = ( 0.77) + ( 0.09) =. 78 = i av i

34 REFERENCES Asselman, N.E.M Grain-size trends used to assess the effective discharge for floodplain sedimentation, river Waal, The Netherlands. Journal of Sedimentary Research. v. 69 no ASTM, 2006 American Society for Testing and Materials Annual Book of ASTM Standards, 2006: ASTM C Bauer and Allen, 1995 Bauer, B. O. and J. R. Allen Beach steps: an evolutionary perspective. Marine Geology 123: Bird, 1985 Bird E.C.F 1985 Coastline changes. Wiley Interscience, Chichester, 219pp Bochicchio et al., 2009 Christopher Bochicchio, Charles Fletcher III, Matthew Dyer, and Thomas Smith Reef-Top Sediment Bodies: Windward O ahu, Hawai i Pacific Science 63(1): Coyne et al.,

35 Coyne, M.A., C.H. Fletcher, and B.M. Richmond Mapping erosion hazard areas in Hawaii: Oberservations and errors. Journal of Coastal Research Spec Iss (28): Fletcher et al, 1997 Fletcher, C.H., R.A. Mulllane, and B.M. Richmond Beach loss Along Armored Shorelines on Oahu. Journal of Coastal Research 13 (1): Folk and Ward, 1957 Folk R L & Ward W C. Brazos River bar: a study in the significance of grainsize parameters. J. Sediment. Petrol. 27:3-26, Gao and Collins, 1992 Gao, Shu, and M. Colliins Net sediment transport patterns inferred from grain-size trends, based upon definatioin of "transport vectors". Sedimentary Geology 80 (1/2): Gao et al., 1994 Gao et al., S. Gao, M.B. Collins, J. Lanckneus, G. De Moor and V. Van Lancker, Grain size trends associated with net sediment transport patterns: an example from the Belgian continental shelf. Marine Geology v. 121 no. 3/4 (1994), pp Gerritsen,

36 Gerritsen, F. Gerritsen, Beach and Surf Parameters in Hawaii. In: Sea Grant Technical Report, UNIHI-SEAGRANT-TR-78-2, University of Hawaii (1978) Grossman and Fletcher, 1998 Grossman, E, and C Fletcher Sea level higher than present 3500 years ago on the northern main Hawaiian Islands. Geology 26 ( ) Harney and Fletcher, 2003 Harney, J.N., and C.H. Fletcher A budget of Caronate Framework and Sediment Production, Kailua Bay, Oahu, Hawaii. Journal of Sedimentary Research 73 (6): Harris et al., 1990 Harris, P.T., Patiaratchi, C.B., Keene, J.B., Cole, A., Modelling the evolution of a linear sandbank field, Moreton Bay, Queensland. Report of results obtained during the cruise of A.M. Brolga in July, 1989, Ocean Sciences Institute, University of Sydney, Vol. 41, 172pp Le Roux and Rojas, 2007 J. Le Roux and E. Rojas, Sediment transport patterns determined from grain size parameters: overview and state of the art, Sedimentary Geology 202 (2007), pp Le Roux, 1994a 35

37 J. Le Roux, An alternative approach to the identification of net sediment transport paths based on grain-size trends, Sedimentary Geology 94 (1994), pp Le Roux, 1994b J. Le Roux, Net sediment transport patterns inferred from grain-size trends, based upon definition of transport vectors comment, Sedimentary Geology 90 (1994), pp Le Roux, 1994c J. Le Roux, A spreadsheet template for determining sediment transport vectors from grain-size parameters, Computers & Geosciences 20 (3) (1994), pp Lipp, 1995 Lipp, David, Changes in Beach Profiles Due to Wave Reflections off Sea Walls at Lanikai, Hawaii MS Plan A Thesis Aug McCave, 1978 McCave, I.N. McCave, Grain-size trends and transport along beaches: an example from eastern England. Marine Geology v. 28 no. 1/2 (1978), pp. M43 M McLaren and Bowles, 1985 McLaren, P., and D. Bowles The Effects of Sediment transport and Grain-Size Distributions. Journal of Sedimentary Petrology 55 (4):

38 McLaren, 1985 P. McLaren, An interpretation of trends in grain-size measurements. Journal of Sedimentary Petrology 51 (1981), pp Noda, 1989 Noda, E.K. and Associates, Inc., Hawaii shoreline erosion management study, overview and case studies^makaha, Oahu; Kailua-Lanikai, Oahu; Kukuiula-Poipu, Kauai; Report for the Hawaii Coastal Zone Management Program NRC, 1995 National Research Council (NRC) Beach nourishment and protection, National Research Counsel. Marine Board, Commission on Engineering and Technical Systems Washington, DC: National Academy Press. pp Rios et al, 2002 J. P. le Roux, R. D. O Brien, F. Rios and M. Cisternas Volume 28, Issue 5, June 2002, Pages Romine et al., in press Romine, B.M., Fletcher, C.H., Frazer, L.N., Genz, A.S., Barbee, M.M., and Lim, S.C. (in press) Historical shoreline change, southeast Oahu, Hawaii: Applying polynomial models to calculate shoreline change rates. Journal of Coastal Research. 37

39 Russell, 1939 Russell, R.D. Russell, Effects of transportation of sedimentary particles. In: P.D. Trask, Editor, Recent Marine Sediments, Society of Economic Paleontologists and Mineralogists, Tulsa, Oklahoma (1939), pp Stelling (1984) Stelling, G.S., On the construction of computational methods for shallow water equations. Rijkswaterstaat communication No. 35/ Swift et al., 1972 Swift et al., D.J.P. Swift, J.C. Ludwick and W.R. Boehmer, Shelf sediment transport: a probability model. In: D.J.P. Swift, D.B. Duane and O.H. Pilkey, Editors, Shelf Sediment Transport Process and Pattern, Dowden, Hutchinson and Ross, Stroudsburg, Pennsylvania (1972), pp Thieler and Danforth, 1994 Thieler, E.R., and Danforth, W.W., Historical shoreline mapping (I): improving techniques and reducing positioning errors. Journal of Coastal Research, v. 10, no. 3, p Watson (1966) Watson, G.S., The statistics of orientation data. Journal 38

40 856 of Geology 74 (2), Wentworth (1949) Wentworth, C.K Directional shift of tradewinds at Honolulu. Pacific Science 3(1):

41 FIGURES LEGENDS Figure Study area. Bathymetric contours are in meters Figure Location of surficial sediment sampling sites for sediment grain size trend analysis. Inset: samples in the vicinity of Wailea Pt Figure Gao-Collins method for determining sediment transport. See Appendix for data and calculations used in figure. A) Illustration of transport determination at site 9 with characteristic distance equal to 2 (dashed circle). Circles represent sampling sites; those containing x show either a trend 1 or trend 2 relationship in grain size parameter with site 9. Dashed arrows indicate component unit vectors (length = 1) drawn in the direction of each trend positive site, while the bold arrow is the summation of the component vectors. B) The process is repeated at each site producing a transport vector field, which is filtered (C) by averaging adjacent vectors. Figure 4 40

42 Le Roux method for determining sediment transport. Grain size parameters are identical to those of Figure 3. This method considers each trend type separately, only trend 1 is considered in this example. A) The closest site in the northern, eastern, southern, and western quadrants is selected for used; dotted lines illustrate quadrants and x on a site indicates selection. B) All sites are transformed to lie at an equal distance of the central site on the cardinal radials; site 5 is at the position of site 5A, 10 is moved to 10A, etc. Grain size parameters are modified to reflect the new positions and summed using the appropriate form of equation (1) for the trend type being investigated. C) The value of the central site is subtracted from all sites. The resulting values indicate transport magnitude in each direction, with negative values indicating transport away from the central site and positive values towards the central point. Summation of component vectors determines the final transport vector. D) The process is repeated at every site with available adjacent sites to produce a vector field for that trend type Figure 5 Directional wind data from Kaneohe Marine Corps Air Base. Values range between degrees Figure 6 Sea level and wave energy calibration for ADVs Figure 7 Mean size, sorting, and skewness interpolated from seafloor sediment samples. 41

43 Figure 8 Results of Gao-Collins method for sediment grain trend analysis Figure 9 Results of Roux method for sediment grain size trend analysis. Figure 10 Gao-Collins sediment grain-size analysis results Figure 11 Le Roux sediment grain-size analysis results Figure 12 Summary of historical shoreline position at Lanikai Beach. Graphs A, B, C, and D show representative datasets for each corresponding transect location. Positions are given as meters from an offshore baseline, thus positive shifts indicate accretion and negative shifts erosion. Gray boxes track the development of a sudden accretion trend. Left map shows a period of accretion in southern Lanikai ( ). Right map shows erosion trend in the south Lanikai and subsequent accretion at central Lanikai Figure 13 42

44 Summary of historical shoreline position at north Bellows Beach. Graphs A, B, and C show representative datasets for each corresponding transect location. Positions are given as meters from an offshore baseline, thus positive shifts indicate accretion and negative shifts erosion. Map shows persistent erosion across region. Arrows mark beginning of seawall construction in response to erosion Figure 14 Hydrodynamic model result for 51 degree winds Figure 15 Hydrodynamic model result for 71 degree winds Figure 16 Hydrodynamic model result for 90 degree winds Figure 17 Historical Shoreline record for Lanikai-Bellows beach. Red indicates erosion rate, blue indicates accretion rate Figure 18 The wind record showing divisions used to separate periods of sediment transport Figure 19 43

45 972 Combined interpretation of results from Le Roux and Gao-Collins methods Figure 20 Each panel shows the generalized sediment transport pattern inferred from changes in beach width shown on Figure 17 and discussed in section 5.2. The dotted line represents 977 an exaggerated view of accreting and eroding areas during each time period. This dotted line is designed to help the reader track the movement of littoral sand as it is redistributed from period-to-period. Wind direction from Figure 18 is also displayed on each panel. 44

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65