Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii

SEDGEO-05234; No of Pages 15 Sedimentary Geology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Sedimentary Geology journal homepage: www.elsevier.com/locate/sedgeo

Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii Catherine Chagué a,⁎, Daisuke Sugawara b, Kazuhisa Goto c, James Goff a, Walter Dudley d, Patricia Gadd e a

School of Biological, Earth and Environmental Sciences, UNSW, Sydney 2052, NSW, Australia Museum of Natural and Environmental History, Shizuoka, 5762 Ohya, Suruga-ku, Shizuoka City 422-8017, Shizuoka, Japan International Research Institute of Disaster Science, Tohoku University, Aoba 468-1, Aramaki, Aoba-ku, Sendai 980-0845, Japan d University of Hawaii - Hilo, Hilo, HI 96720, United States e Australian Nuclear Science and Technology Organisation, Locked Bag 2001, Kirrawee DC, NSW 2232, Australia b c

a r t i c l e

i n f o

Article history: Received 15 June 2017 Received in revised form 12 September 2017 Accepted 13 September 2017 Available online xxxx Editor: Dr. J. Knight Keywords: Tsunami deposit River flood deposit Geochemistry Arsenic Sediment transport modelling Hawaii

a b s t r a c t The Japanese community of Shinmachi, established on low-lying land between downtown Hilo and Waiakea, Hawaii, was obliterated by the 1946 Aleutian tsunami but was rebuilt, only to be destroyed again by the 1960 Chilean tsunami. The aim of this study was to find out if any geological evidence of these well documented events had been preserved in the sedimentary record in Wailoa River State Park, which replaced Shinmachi after the 1960 tsunami. This was achieved by collecting cores in the park and performing sedimentological, chronological and geochemical analyses, the latter also processed by principal component analysis. Sediment transport modelling was carried out for both tsunamis, to infer the source of the sediment and areas of deposition on land. The field survey revealed two distinct units within peat and soil, a thin lower unit composed of weathered basalt fragments within mud (Unit 1) and an upper unit dominated by fine volcanic sand within fine silt exhibiting subtle upward fining and coarsening (Unit 2, consisting of Unit 2A and Unit 2B), although these two anomalous units only occur on the western shore of Waiakea Mill Pond. Analysis with an ITRAX core scanner shows that Unit 1 is characterised by high Mn, Fe, Rb, La and Ce counts, combined with elevated magnetic susceptibility. Based on its chemical and sedimentological characteristics, Unit 1 is attributed to a flood event in Wailoa River that occurred around 1520–1660 CE, most probably as a result of a tropical storm. The sharp lower contact of Unit 2 coincides with the appearance of arsenic, contemporaneous with an increase in Ca, Sr, Si, Ti, K, Zr, Mn, Fe, La and Ce. In this study, As is used as a chronological and source material marker, as it is known to have been released into Wailoa River Estuary and Waiakea Mill Pond by the Canec factory between 1932 and 1963. Thus, not only the chemical and sedimentological evidence but also sediment transport modelling, corroborating the historical record, suggest that Unit 2A was deposited by the 1946 tsunami, and the sediment most likely originated from Wailoa River Estuary, beach and nearshore seafloor. The upper part of this unit, Unit 2B, is believed to have been deposited by the 1960 tsunami, as suggested by sediment transport modelling, although limited accommodation space is likely to have resulted in the thin deposit (3 cm thickness) present at that site. Limited accommodation space on the island of Hawaii has led to only rare locations where tsunami deposits are preserved, despite the repeated occurrence of tsunamis affecting the island. © 2017 Elsevier B.V. All rights reserved.

1. Introduction and background Hilo, on the east side of the island of Hawaii (Big Island), Hawaiian Islands (Fig. 1), is known to have been affected by numerous tsunamis in historical times, even before the devastating 1946 event (e.g., Macdonald et al., 1947). Indeed, because of their location in the middle of the Pacific, the Hawaiian Islands have often been impacted by tsunamis generated around the Pacific Ring of Fire (e.g., Lander and Lockridge, 1989). They also tend to trap tsunami energy, as their steep volcanic slopes lead to large-scale resonance oscillations (Munger and ⁎ Corresponding author. E-mail address: [email protected] (C. Chagué).

Cheung, 2008). Furthermore, tsunamis are amplified in Hilo Bay, due to the geometry of the Bay and insular shelf, resulting in a main oscillation period between 15 and 30 min, which is similar to the period of many tsunami waves, so that resonance is almost always triggered, independently of the direction of the wave approach (e.g., Cheung et al., 2013). The first historical tsunami reported to have hit Hilo was generated by a severe earthquake around Copiapo, Chile at 11.00 pm on 11 April 1819. It struck the Hawaiian Islands around 4.00 am on 12 April and its effects were reported from both the west and east coasts of Hawaii (Big Island) and also Oahu (Lander and Lockridge, 1989; Fletcher et al., 2002). Hilo may have been affected by an earlier event on 12 December 1812, believed to have been generated off the

https://doi.org/10.1016/j.sedgeo.2017.09.010 0037-0738/© 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

2

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

Fig. 1. (a) Hawaiian Islands in the Pacific Ocean. The epicentres of the earthquakes that generated the 1946 Aleutian tsunami (+) and the 1960 Chile (Valdivia) tsunami (x) are also shown; (b) Hawaiian Islands; (c) Hawaii (Big Island); (d) Hilo and surroundings. The breakwater in Hilo Bay (bw) and Waiakea are shown, as well as the extent of the area in Fig. 7.

California coast, although there are only records of one wave on the west coast of Hawaii, not Hilo (Lander and Lockridge, 1989). More detailed historical records for tsunamis in Hilo commence with the event generated by a Ms. 8.5 earthquake in Valdivia, Chile on 7 November 1837, which caused significant damage to infrastructure and resulted in 14 fatalities (Shepard et al., 1950). Since then there have been at least 130 tsunamis reported in Hilo from both historical and instrumental records (e.g., Anon., 1902; Cox and Morgan, 1977; Soloviev and Go, 1984; Lander and Lockridge, 1989; Fletcher et al., 2002; Lander et al., 2003; NGDC/WDS, 2017; J. Goff unpublished data). Of particular note is Waiakea, an area east of downtown Hilo (Fig. 1d), which had often been damaged by tsunamis prior to 1946. Not only was damage reported as a result of the 1837 Chilean tsunami mentioned above, but also by the 1868 Arica (Chile), 1877 Arequipa (Chile) (5 people killed and 163 made homeless), 1922 Atacama (Chile) and 1923 Kamchatka tsunamis (e.g., Lander and Lockridge, 1989). There were also further tsunamis after 1946, including on 5 November 1952 when a 3.7 m (12 ft) tsunami generated by a Mw 9.0 earthquake in Kamchatka surged over Coconut Island (see Fig. 2 for location) (Dudley and Lee, 1998). Again, on 9 March 1957, a Mw 8.6 central Aleutian earthquake generated a tsunami that once more surged over Coconut Island (NGDC/WDS, 2017). Following the 1960 Chilean event, there have been further inundations, including the 1964 tsunami generated by a Mw 9.2 earthquake in Alaska (NGDC/WDS, 2017).

severe damage and the largest number of deaths in Hawaiian history. The earthquake occurred at approximately 2:00 am Hawaii local time. It was registered and recorded by the US Coast and Geodetic Survey at the University of Hawaii at Manoa, Honolulu, O'ahu and the Hawaiian Volcano Observatory at Kilauea, Hawaii shortly after (Dudley and Lee, 1998). The first wave from the tsunami reached the island of Kaua'i at around 6:00 am, Honolulu at 6:33 am and Hilo at 7:06 am (Shepard et al., 1950). Luckily the first wave was small, and as people had already risen at that time, many fled as it arrived, most probably saving many lives (e.g., Muffler and the Pacific Tsunami Museum, 2015). A total of nine waves inundated Hilo (Figs. 2, 3), with the 3rd wave described as the largest (Shepard et al., 1950; Dudley and Stone, 2000). It is worth noting that the measured tsunami heights were spatially variable, ranging from 2.7 to 9.8 m (Macdonald et al., 1947; Eaton et al., 1961) (Fig. 2). As the tsunami approached Hilo, it inundated Laupahoehoe Peninsula (Fig. 1c) killing 16 school children and eight adults. A total of 159 people lost their lives on the Hawaiian Islands, 96 of those in Hilo (Shepard et al., 1950). Extensive damage was reported in Hilo, including houses and business in particular on Hilo Bay front, rail tracks, shipping port, but also part of the breakwater, with property damage reaching $26 million in 1946 dollars (e.g., Muffler and the Pacific Tsunami Museum, 2015). This tsunami led to the creation of the first tsunami warning system that would eventually become the Pacific Tsunami Warning Center (PTWC) (Muffler and the Pacific Tsunami Museum, 2015).

1.1. The 1946 tsunami

1.2. The 1960 tsunami

On 1 April 1946 at 4:28 am local time (12:28 UTC), a Mw 8.6 earthquake in the Aleutian Islands generated a tsunami that caused the most

The Mw 9.4–9.6 Valdivia (Chile) earthquake happened around 19:11 UTC on 22 May 1960 after several precursor events, starting

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

3

Fig. 2. Inundation map of the 1946 and 1960 tsunamis. The area inundated by the 1946 tsunami is shown, while only the limit of the inundation by the 1960 tsunami is marked. Inundation heights are indicated in metres above mean sea level (amsl) for the 1946 and 1960 (in italics) tsunamis, respectively. Shinmachi, Coconut Island, Waiakea, the breakwater and the location of the Coca-Cola bottling plant (C-C) are shown, as well as the extent of the area of Fig. 7 (modified after Macdonald et al., 1947; Eaton et al., 1961).

with a Mw 7.5 near Concepcion at 6:02 UTC. The first earthquake in this series occurred just after midnight, Hawaii local time, on 21 May 1960 (Eaton et al., 1961) and was registered at the US Coast and Geodetic Survey Observatory (PTWC) in Honolulu, with a tsunami watch issued at 0:45 am. This alert was cancelled at 8:49 pm when only a small wave was recorded in Hilo Bay. However, by 11:30 am Hawaii local time on 22 May 1960 it was known that a second, large earthquake had occurred some 15 min after the Valdivia event (Kanamori, 1977).

This was the largest instrumentally recorded earthquake of the twentieth century and once again, a tsunami watch was issued and the expected time of arrival was predicted to be about 15 h later, at around midnight local time (Johnston et al., 2008). The first evacuation siren sounded at 8:35 pm with the first small wave arriving at 0:07 am on 23 May 1960, which however only rose to about 1.2 m (4 ft). The second wave was larger at around 2.7 m (9 ft) flooding Kamehameha Avenue and the business district of Hilo (Dudley and Lee, 1998). At about 1:00 am, the

Fig. 3. Incoming wave of the 1946 tsunami inundating Shinmachi. As seen on the photo (taken looking NE), the waterfront of downtown Hilo and Shinmachi had already sustained heavy damage from earlier waves (James Kerschner Collection).

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

4

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

ocean receded 2.1 m (7 ft) below normal, and the largest 3rd wave arrived at 1:04 am as a ‘20-foot [6.1 m]-high nearly vertical front’ that rose as high as 10.7 m (35 ft) (Eaton et al., 1961) (Fig. 2). Spatial variation of the measured tsunami heights was again evident, ranging from 3.0 to 10.7 m (Fig. 2). Boulders of up to 22 tons from the offshore breakwater were deposited in downtown Hilo, and the area around Waiakea was hit particularly hard; at 1:05 am, the power plant at the southern end of Hilo Bay was hit, plunging Hilo into darkness. By 2:15 am, wave heights had reduced sufficiently for people to re-enter Hilo (Eaton et al., 1961). Despite the warning, 61 people lost their lives in the tsunami; property damage was estimated at $23.5 million dollars (Lander and Lockridge, 1989). 1.3. The history of Shinmachi Early 1885 saw the first sponsored Japanese workers arrive in the Hawaiian Islands to work in sugar cane and pineapple plantations as part of the ‘Kanyaku Imin’ immigration system, an agreement between the Japanese government and the Kingdom of Hawaii (Iida, 1994; Hosok, 2010). In Hawaii (Big Island) the first Japanese labourers arrived in Waiakea in February 1885 to work in sugar cane plantations, followed by many more who settled in other areas, mostly around Hilo (Iida, 1994). Around 1900 they established the Shinmachi (‘New Town’) community on low-lying land on Hilo's waterfront (Fig. 4). Like the Bayfront business district in downtown Hilo, Shinmachi was obliterated by the 1946 Aleutian tsunami (Figs. 2, 3), with the Coca-Cola bottling plant, located between the shore and Wailoa River Estuary (Fig. 2), one of the few buildings left standing (Fig. 5). Almost immediately after the event, Hilo businesses and residents, particularly those in the Waiakea area, rebuilt in the vulnerable, low-lying lands of Shinmachi (Dudley and Stone, 2000). Shinmachi was however destroyed again by the 1960 Chile tsunami (Fig. 6). It was never rebuilt and the area was turned into a greenbelt, the Wailoa River State Park, a complex of sport fields, picnic areas and parking lots (e.g., Dudley and Lee, 1998). 1.4. The Canec plant and arsenic contamination One of the by-products of sugar cane is bagasse, the fibrous matter left after crushing of sugar cane to extract the juice, with 1.9 t of bagasse produced for 1 t of raw sugar extracted. It was initially burned to heat

the boilers to operate the sugar cane mills, although this was phased out as fuel oil was used instead. Excess bagasse was then used as fertiliser, pulp or just dumped into the sea, until a new use for it was sought. Between 1932 and 1963, wallboard was manufactured from bagasse in the Hawaiian Cane Products Company Ltd. (commonly referred to as the Canec plant or factory). However, calcium arsenate and arsenic trioxide were added to the bagasse to make it termite-proof, with wastewater containing high amounts of arsenic (As) released into the Waiakea Mill Pond and Wailoa River (Kelly et al., 1981). This led to contamination of sediments in Wailoa River Estuary and Hilo Bay, with the highest concentrations measured in Wailoa River Estuary (Hallacher et al., 1985; Glendon-Baclig, 2007). Hallacher et al. (1985) showed that the highest As concentrations occurred in the anaerobic sediments below 50 cm depth, both near the location of the former Canec factory and also downstream of the exit point of the effluent pipe of the factory, near the confluence of Waiakea Mill Pond and Wailoa River Estuary. Glendon-Baclig (2007) also reported higher As concentrations at these locations, although generally at shallower depth (between 15 and 55 cm depth). Near the confluence of the Waiakea Mill Pond and Wailoa River Estuary, maxima in As concentrations occurred below 30 cm depth in all cores, and were associated with remnants of bagasse in at least one of the cores taken at that site. This spatial distribution most likely reflects the effects of the historical release of As in the waterbodies in the area, which mostly ceased after 1963, when the Canec factory closed. 2. Study aims Although the inundation and destruction of Shinmachi, Hilo, Hawaii, by the 1946 and 1960 tsunamis are well documented, it was not known whether any geological evidence of these events had been preserved in the sedimentary record. The aim of this study was to search for the sedimentary evidence of these two tsunamis that may lay buried in Wailoa River State Park, by collecting cores in the park and carrying out sedimentological, geochemical and chronological analysis to decipher the record of environmental changes that might have been preserved there. Sediment transport modelling was also performed to estimate the origin of the deposits, and results compared with the field data and the findings of our multi-disciplinary study. We also briefly report on issues of accommodation space on Hawaii (Big Island).

Fig. 4. Aerial view of Shinmachi prior to the 1946 tsunami (photo taken NNE) (Futoshi Okumara Collection).

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

5

Fig. 5. After the 1946 tsunami with the Coca-Cola bottling plant one of the few buildings left standing in the foreground (C-C) (James Kerschner Collection). See Fig. 2 for the location of the plant.

3. Study site The study site is in Wailoa River State Park, in Hilo District, which includes the area previously occupied by Shinmachi that was abandoned

after the 1960 tsunami. The Waiola River State Park borders Waiakea Mill Pond and Wailoa River Estuary, which drain two ephemeral rivers or streams, namely Alenaio Stream and Waiakea River, and an underground river, Palai Stream (Fig. 7). The area is underlain by a lava flow

Fig. 6. After the 1960 tsunami, with Waiakea and Coconut Island (CI) in the foreground and Shinmachi and the Wailoa River in the background (Hawaii Tribune – Herald Collection). Photo taken looking SW.

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

6

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

Fig. 7. Hilo Bay/Shinmachi map with locations of sites sampled and stratigraphy recorded at these sites. The dark grey shaded areas are built areas, buildings or roads, while light grey shaded areas are parks. M = memorials (one to honour the tsunami victims, one to honour the Vietnam veterans), C-C marks the original location of the Coca-Cola bottling plant (See Fig. 5). A photo of cores at site SH-11 (cores WP494a and WP494c) is also shown, with the white dashed lines marking the upper and lower contacts of Unit 1 (U1) and Unit 2 (U2).

that originated from the Mauna Loa volcano and slopes gently downward along the northeast rift zone of the volcano toward Hilo Bay (Buchanan-Banks, 1993). While there are some uncertainties about the age of this lava flow, it is believed to be either 3500 years old (Buchanan-Banks et al., 1989) or even N14,000 years old (Lipman and Moore, 1996). 4. Methods 4.1. Sampling A field survey was carried out in December 2014 and focussed on the area previously occupied by Shinmachi, and which is now Wailoa River State Park, although a reconnaissance survey also took place around the island of Hawaii. Twelve short cores, between 10 and 150 cm length, were collected in the park along Alenaio Stream and on the shores of Waiakea Mill Pond and Wailoa River Estuary (Table S1) with a geoslicer in an attempt to recover the geological evidence of the 1946 and 1960 tsunamis. 4.2. Laboratory analyses One core, in which two anomalous units were visible within peat and soil, was analysed using the ITRAX core scanner equipped with a magnetic susceptibility meter at the Australian Nuclear Science and Technology Organisation. An optical image and a X-radiograph were also produced. The setting used for X-Ray Fluorescence (XRF) scanning was 55 mA, 30 kV and exposure time of 10 s at 1 mm resolution, thus producing a high resolution continuous multi-elemental dataset. For the X-radiograph, the setting was 30 mA, 50 kV and exposure time of 500 ms at 500 μm resolution.

A molybdenum (Mo) tube was used for the XRF scanning, which allowed the analysis of elements from silicon (Si) to lead (Pb), although only elements of interest are reported here (Table S2). As the incoherent (Mo Inc) and coherent (Mo Coh) scattering are also measured during the analysis, we can use the Mo Inc/Mo Coh ratio as a measure of organic matter content, as shown in previous studies (e.g., Guyard et al., 2007; Chagué-Goff et al., 2016). All data obtained with the ITRAX with total of counts per second (cps) below 34,000 were removed from the dataset, as variations in elemental counts at these intervals are more likely to be due to changes in surface geometry and matrix effects. Data were then normalised over cps, to correct for grain size and moisture effect, as well as difference in surface geometry (e.g., ChaguéGoff et al., 2016), and then plotted using the R Studio v0.99.467 package. Normalised data and Mo Inc/Mo Coh (Mo Ratio) were also processed using principal component analysis (PCA). Grain size analysis of each cm-interval in the coarser deposits and a few samples taken at variable intervals in the remaining of the sedimentary sequence, as well as replicate samples from the beach on Hilo waterfront (Table S1), was carried out. The few samples containing a coarse fraction N1.4 mm were dried and sieved, with the weight of the coarse particles determined. The remaining fraction was analysed with a Malvern Master 2000 Multisizer, after removal of organic matter with hydrogen peroxide and addition of sodium hexametaphosphate, following methods outlined in Sperazza et al. (2004). Results were processed using GRADISTAT v.8 (Blott and Pye, 2001), and results reported in metric units following Folk and Ward (1957), including mean grain size and sorting. Results of samples containing larger particles were however first processed following methods outlined in Dinis and Castilho (2012). The geochronological control was obtained using radiocarbon and geochemistry. One peat sample taken immediately below the lower coarse layer at 42–43 cm depth was submitted to the Waikato

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

Radiocarbon Laboratory, New Zealand, for accelerator mass spectrometry (AMS) 14C analysis. We however used geochemistry, namely the temporal distribution of arsenic in the sedimentary sequence, to estimate the timing of deposition of the upper unit. As mentioned above, arsenic (As) had been released in the Waiakea Mill Pond and Wailoa River Estuary from the Canec factory between 1932 and 1963 (Hallacher et al., 1985). While As is often mobile in sediment and soil (e.g., Bhumbla and Keefer, 1994; Bissen and Frimmel, 2003), its distribution in the sedimentary sequence, in combination with the distribution of other elements, such as Ca and Sr, can help pinpoint the source of specific units. No attempt was made to use 210Pb or 137Cs for dating control, as the upper sequence consisting of soil, sand and gravel is disturbed, thus preventing any meaningful interpretation of 210Pb and 137Cs activities that would have been measured in the samples. 4.3. Modelling Source and transport processes of sediments due to the 1946 Aleutian and 1960 Chilean tsunamis were investigated using numerical modelling. A nesting grid system, which consists of five layers of different areas and resolution, was used to simulate tsunami generation

7

around the Aleutian and Chilean coasts and propagation across the Pacific Ocean. The General Bathymetric Chart of the Oceans (GEBCO; http://www.gebco.net) was used for the bathymetry data of the first to fourth layers (Fig. S1a-d). In the third and fourth layers, the GEBCO data were partly replaced by higher-resolution data from the Main Hawaiian Islands Multibeam Bathymetry and Backscatter Synthesis (http://www.soest.hawaii.edu/HMRG/multibeam/bathymetry.php) to include the accurate shape of the coastline and detailed bathymetric features (Fig. S1c–d). The first, second, third and fourth layers cover the entire Pacific basin, Hawaiian Islands, Molokai to Hawaii (Big Island) and the northeast of Hawaii, respectively. These layers were constructed on the geographical coordinate system with decreasing spatial resolution from 5 arc-min to 2.4 arc-sec by a constant ratio of 1/5. Tsunami inundation and sediment transport in Hilo Bay were modelled using the 1/3 arc-sec bathymetry and topography data, which were provided by the National Tsunami Hazard Mitigation Program (NTHMP) Mapping and Modelling Benchmarking Workshop (http://coastal.usc.edu/ currents_workshop/). The NTHMP data resolve the shape and elevation of the breakwater, Wailoa River Estuary and Waiakea Mill Pond (Fig. 8). The bathymetry data were transformed into a Cartesian coordinate system using the UTM projection (Zone 5) and resampled with a spatial resolution of 5 m.

Fig. 8. Bathymetry data used for numerical modelling. Hilo Bay (dx = 5 m). The red solid circle indicates the virtual tide station TS3. The island of Hawaii is shown in the inset, indicating the location of the area presented in the main figure. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

8

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

The 1946 Aleutian earthquake was modelled using the fault parameters determined by Butler et al. (2017). A uniform slip of 22.3 m was applied for two contiguous segments, which have a total fault length of 200 km and width of 50 km (Table 1). Butler et al. (2017) showed a good agreement between the observed and simulated tsunami heights in Hilo and at other locations in the Hawaiian Islands. The fault parameters of the 1960 Chilean earthquake determined by Kanamori and Cipar (1974) were used for the tsunami source model, with a uniform slip of 24 m applied to single fault plane with a total length of 800 km and width of 200 km (Table 1). The location of the fault plane was set based on Fujii and Satake (2013), while the elastic model of Mansinha and Smylie (1971) was used to compute the vertical seafloor deformation from the given fault parameters, and the initial tsunami waveform was set according to the seafloor deformation. The conventional depth-averaged tsunami hydrodynamic model (TUNAMI CODE; Goto et al., 1997) was used to simulate tsunami propagation across the Pacific. The model is based on the nonlinear shallowwater theory and constructed on the finite differential method with the leap-frog scheme. Effects from the Coriolis force and wavenumber dispersion were considered to simulate tsunami propagation from far-field. In the fifth layer, the sediment transport model (STM; Takahashi et al., 2000) was coupled with the TUNAMI CODE to simulate tsunami-induced sediment erosion, transport, deposition and resulting morphological change. The STM calculates the exchange of bed- and suspension loads and resulting morphological change. The coupled TUNAMI-STM model was validated through the case studies of the 2011 Tohoku tsunami on the Sendai Plain and Sanriku coasts (e.g., Sugawara et al., 2014; Yamashita et al., 2016). The transport formula proposed by Takahashi et al. (2011) was employed to determine the bedload and pick-up rates of suspended sediments. The STM includes a slope effect for bedload and saturation concentration for suspended sediments. The STM can handle erosion, transport and deposition of sediments of a single grain size class. In this study, the movement of sand-sized sediments was considered in the modelling. Based on the grain size analysis of the tsunami deposit, a uniform grain size of 166 μm was chosen for the target grain size of the modelling. As the Mauna Loa lava flow and an overlying thin muddy to sandy soil make up the ground surface of Hilo and the surrounding area (including Shinmachi), sources of the sand were assumed to be the beach, river bed and sea bottom. The initial sand thickness was assumed to be infinite in the source area, and zero on land. In addition, entire land areas, except for the beach and water bodies, were assumed to be non-erodible beds. A Manning's roughness coefficient of 0.025 m−1/3 s was used to calculate shear velocity for the sediment transport modelling. 5. Results 5.1. Stratigraphy Cores were retrieved at a number of sites in Wailoa State River Park along the western and eastern shores of Waiakea Mill Ponds and Wailoa River Estuary and on the southern shores of Alenaio Stream, along a transect almost parallel to the shore of Hilo Bay (Fig. 7). Our survey revealed that only a shallow sedimentary sequence (max. 150 cm thick (site SH-3), but as little as 10 cm (site SH-1)), consisting of peat, sandy soil and rare coarser layers, is present over the basalt lava flow (Fig. 7).

Two anomalous coarse layers, consisting of sand, coarse silt and gravel, intercalated within peat and soil were recorded at only a few locations (SH-3, SH-7, SH-9, SH-11) in Wailoa River State Park. At other sites, only one sandy layer was recorded (SH-2, SH-5, SH-6), while elsewhere, there was no visible coarse layer within the peat and soil, most probably reflecting anthropogenic disturbance following the abandonment of Shinmachi after the 1960 tsunami and establishment of Wailoa River State Park. Anthropogenic debris, including glass and painted wood, was however found, although rarely (SH-6, SH-9, SH-10) (Fig. 7). One of the cores collected at site SH-11 was retained for further analysis, as it presented a stratigraphy with two anomalous layers that also occurred around the site (Fig. 7). These layers consisted of a lower brown-grey gravelly sandy mud (Unit 1) exhibiting sharp lower and upper contacts, and an upper fining upward brown fine sand to silt (Unit 2) with a sharp lower contact but a somewhat diffuse upper contact, although a sharp upper contact was visible in another core taken a few metres away (core WP494c) (Fig. 7). A faint change in colour was observed at 20 cm depth, and was also marked by rare sand grains. 5.2. Grain size distribution As illustrated in Figs. 9 and S2, the peat and soil units are generally characterised by a unimodal distribution, while the anomalous layers have either a polymodal (Unit 1) or bimodal (Unit 2) grain size distribution. Unit 1 contains rare large sub-angular fragments of weathered basalt of up to 16 mm diameter within a very poorly sorted muddy sand at the base and fragments of up to 7 mm diameter in the upper part dominated by sandy mud. Unit 2 consists of muddy sand, with the main mode between 170 and 148 μm, and another mode between 14 and 21 μm, exhibiting only subdued fining upwards above 25 cm depth, with an increase of 10% in mud content at that depth to a maximum of 54% mud at 21–20 cm depth, including 2.8% clay. There is a slight coarsening then fining upwards between 20 and 17 cm depth (Fig. S2), with a bimodal distribution still present. The overlying soil is coarser than Unit 2 (very coarse silt to very find sand), but is unimodal, except at 15–16 cm depth, where rare basalt fragments of 5 and 12 mm diameter occur, resulting in a trimodal distribution. Sediment on Hilo Beach (Fig. 7) was dominated by dark medium volcanic sand that also contained rounded very fine to medium gravel (main mode at 258 μm). 5.3. Geochemistry and magnetic susceptibility The elemental profiles (counts normalised over total counts per second (cps)) displayed distinctive changes as follows: Unit 1 is characterised by high counts of Mn, Fe and Rb associated with moderate counts of the rare earth elements lanthanum (La) and cerium (Ce), which coincide with a high magnetic susceptibility of up to 5000 S.I. (Fig. 9). The underlying and overlying peat shows high Mo Inc/Mo Coh, reflecting the high organic content, which also corresponds to high counts of Br and S, and low counts in Mn, Fe, Si, K, Ti, V, Cr, Ca, Sr, As, Rb, La and Ce. The less dense material is also clearly identified in the X-radiography (Fig. 9). There is a distinct change in a number of elements at 32 cm depth, at the base of Unit 2, with Si, K, Ti, V, Cr, Ca, Sr, As, Mn, Fe and Zr all showing a marked increase, coinciding with a small increase in magnetic susceptibility, while Br counts in particular are significantly lower (Fig. 9). All these elements exhibit small variations between 32 cm depth and the

Table 1 Fault parameters of the 1946 Aleutian and 1960 Chilean earthquakes as the sources of the tsunamis (Lon = longitude; Lat = Latitude).

1946 Aleutian 1960 Chile

Lon (°)

Lat (°)

Depth (km)

Length (km)

Width (km)

Slip (m)

Strike (°)

Dip (°)

Rake (°)

195.291 E 196.693 E 75.9 W

53.2800 N 63.6543 N 45.2 S

5 5 1

100 100 800

50 50 200

22.3 22.3 24.0

245 250 10

5 5 10

90 90 90

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

9

Fig. 9. Site SH-11 (core WP494a from Fig. 7). Stratigraphy, magnetic susceptibility (S.I.), X-radiograph, ITRAX counts (normalised over cps), Mo Ratio (a proxy for organic matter), and grain size distribution (with grain size divisions in μm – logarithmic scale). Unit 1 (U1) and Unit 2 (U2 – divided into Unit 2a (U2a) and Unit 2b (U2b)) are indicated. The radiocarbon age (cal BP) of sample Wk-44698 at 42–43 cm depth is also shown.

surface, with nevertheless a short-term decrease in counts of Si, K, Ti, V, Mn, Fe, Ca around 20 cm and 18 cm depth, as well as a small but distinctive increase of Si between 20 and 18 cm depth. The overlying sandy soil (above 17 cm depth) displays another small increase in magnetic susceptibility matched by small increases in Fe and Mn counts, as well as As, and decreasing counts in Sr and Zr. The upward decrease in magnetic susceptibility above 10 cm depth is matched by decreases in Mn, Fe and As, and an upward increase in organic matter (Mo Ratio) (Fig. 9). The variation in chemical composition between peat and coarser units in this core was also confirmed by PCA performed on the dataset, that indicates that 80% of the variance can be explained by the first three

principal components (PC1 = 61%, PC2 = 12.5%, PC3 = 6.5%) (Fig. 10a). The first principal component distinguishes the peat (between the lava flow and Unit 2), which exhibits high organic matter (Mo Inc/Mo Coh = Mo Ratio) and associated elements Br and S known to have a strong affinity for organic matter (e.g., Chagué-Goff, 2010), from Units 1 and 2 as well as the overlying soil (Fig. 10b). The second principal component is characterised by an inverse relationship between Mn, Fe, Rb, La and Ce, and the other elements, thereby distinguishing Unit 1 from Unit 2 and the overlying soil (Fig. 10c). The third principal component exhibits an inverse relationship between Sr, As, Zr, Fe and Rb (the score for V is almost nil) and the other elements (including organic

Fig. 10. Results of principal component analysis (PCA). (a) Percentage of variance for each principal component (numbers on the X axis indicate the principal components (PC1 to PC17)); (b) Eigenvalues for each element and MoR (=Mo Ratio) for the 1st principal component (PC1); (c) Eigenvalues for each element and MoR (=Mo Ratio) for the 2nd principal component (PC2); (d) Eigenvalues for each element and MoR (=Mo Ratio) for the 3rd principal component (PC3). See text for explanation.

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

10

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

Fig. 11. Simulated waveform of the 1946 Aleutian tsunami in Hilo Bay at site TS3 in Fig. 8 (water depth = 10 m).

matter), which we tentatively attribute as indicating a different source for these elements and allowing us to distinguish Unit 2 from the overlying soil (Fig. 10d). 5.4. Chronology Radiocarbon dating of a peat sample taken immediately below Unit 1 (sample ID Wk-44698; 42–43 cm depth) returned a calibrated age of 431–289 cal BP (1520–1660 CE). The timing of deposition of Unit 2 was estimated using the distribution of As in correlation with the distribution of Si, K, Ti, Ca and Sr, which all exhibit a marked increase coinciding with the lower contact of Unit 2, while counts were much lower below that depth (Fig. 9). This therefore suggests a different source of the material composing Unit 2. Arsenic in particular can here be used as a temporal marker, as it is known to have been discharged from the Canec factory into Wailoa River Estuary and Waiakea Mill Pond from 1932 (Kelly et al., 1981), therefore implying that Unit 2 was deposited post 1932. 5.5. Modelling 5.5.1. The 1946 tsunami The first wave was 1 m high and arrived at Hilo Bay around 4 h 50 min after the earthquake (Fig. 11). The water level started to drop suddenly and reached the lowest (− 3 m) at 5 h 5 min. The peak of the second wave 5 h 10 min after the earthquake was the highest (1.7 m), while the model shows that the following waves (3rd to 6th waves) recorded peak water levels of 1.3–1.4 m. The simulated tsunami inundation reached beyond the historical inundation map of the 1946

tsunami (Figs. 12a, 2, respectively). Onshore in Hilo, the simulated tsunami height showed spatial variation ranging 2–4 m. The height was relatively lower (2–3 m) in and around Shinmachi and Waiakea Mill Pond than in the area east of the pond. Thus, the modelled tsunami height along the profile line that tracks the study area (Fig. 13) does not necessarily represent the higher tsunami inundation in Hilo. The modelled morphological changes showed remarkable erosion of the sandy beach and minor deposition around Shinmachi (Fig. 12b). Erosion is also significant in shallower parts of Hilo Bay, such as around the breakwater and near the base of the cliff on the west side of the bay (Figs. 8, 13). The depth of beach erosion far exceeded 0.15 m, with thin deposition (0–0.1 m) occurring inland at Shinmachi (Fig. 12b). It is worth noting that a considerable amount of sediment was deposited at the bottom of Waiakea Mill Pond. The profile showed that the tsunami height was 4 m on the beach and decreased to 2 m at inland locations (Fig. 13a). Beach erosion exceeded 0.8 m, and deposition was evident (~ 0.7 m) just offshore of the beach, due to transport by backwash (Fig. 13b). The modelled onshore deposit did not exceed 0.2 m along the profile line. Instantaneous snapshots of the simulation indicate that no sediment transport took place during the inundation by the first wave (Fig. S3a), while the second wave is responsible for the erosion and deposition of sediments at the survey sites in Shinmachi (Fig. S3b). Seafloor sand near the beach was entrained into the bore of the second wave, reaching an equivalent thickness of 0.05 m of suspended sand, and transported inland. When the tsunami bore entered the Wailoa River Estuary, a considerable amount of river bed sediments was additionally entrained into the flow (Fig. S3c), with the modelled equivalent thickness of the suspended sand exceeding 0.15 m. The suspended sand was transported toward the head of Waiakea Mill Pond, and deposited on the pond bottom and the area surrounding our survey sites (Fig. S3d). 5.5.2. The 1960 tsunami The tsunami in Hilo Bay started with minor water level oscillations (b0.1 m) that lasted for N1 h (Fig. 14; 13 h 15 min to 14 h 15 min). The first and second waves were relatively small, reaching a height of 0.4 and 0.6 m, respectively. A considerable drop in water level (− 2.7 m) occurred between the second and third waves. The third wave, which was the largest (2.7 m), arrived in Hilo Bay 15 h 45 min after the earthquake. The following waves did not exceed 1 m height, although the water level increased up to 2 m at the end of the simulation (Fig. 14; 17 h 15 min). The simulated tsunami height of the 1960 Chilean tsunami in Hilo Bay again showed spatial variation, but was generally larger than that of the 1946 Aleutian tsunami, reaching N4 m in most of the bay, and 5 m east of Waiakea Mill Pond (Fig. 15a). Tsunami amplification is evident in the eastern part of the bay and may be associated with the effect

Fig. 12. (a) Simulated height of the 1946 Aleutian tsunami in Hilo Bay (7 h 35 min after the earthquake); (b) Simulated morphological change by the 1946 Aleutian tsunami in Hilo Bay (7 h 35 min after the earthquake). Negative and positive values indicate gross depth of erosion and thickness of deposition, respectively.

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

11

deposition reached 0.2 m along the profile line, and was slightly greater than in the 1946 Aleutian tsunami. Instantaneous snapshots of the simulation indicate that the second wave did not cause sediment erosion nor deposition anywhere in Hilo Bay (Fig. S4a), and this also applies to the first wave. Transport of suspended sediment was however significant at the time of the third wave, which entrained small amounts of sand from the seafloor near the beach at Shinmachi (Fig. S4b; compare with Fig. S3b). It is worth noting that when the third wave entered the bay, it entrained a considerable amount of sediment that was transported into the bay near the head of the breakwater. However, most of the suspended sediment advected toward the east side of the bay. The subsequent processes of tsunami inundation and sediment transport were almost identical to those of the 1946 Aleutian tsunami. As the third wave inundating the beach of Shinmachi penetrated inland and entered Wailoa River Estuary, the flow entrained the river bed sand as suspended sediment with an equivalent thickness of more the 0.15 m (Fig. S4c). The suspended sand was transported toward the head of Waiakea Mill Pond, and deposited at the bottom of the pond and in areas surrounding our survey sites (Fig. S4d). 6. Discussion 6.1. Sedimentary and geochemical evidence of high energy events

Fig. 13. (a) Pre- and post-tsunami topography and tsunami height along the profile line (A–B) shown in Fig. 12; (b) Simulated morphological change along the profile line (A–B) shown in Fig. 12. Negative and positive values indicate gross depth of erosion and thickness of deposition, respectively.

of the breakwater. The simulated inundation area by the 1960 tsunami was also greater than that by the 1946 event. The spatial characteristics of morphological change caused by the 1960 tsunami are almost identical to those of the 1946 event. Considerable erosion of the beach and minor deposition inland characterise the sediment transport at Shinmachi (Fig. 15b). Erosion depth of the beach reached 0.15 m and deposition thickness ranged from 0 to 0.1 m inland in Shinmachi and the surrounding areas of Waiakea Mill Pond. Erosion occurred on the shallower seafloor areas, and sediments were distributed throughout Hilo Bay. The profile showed that the tsunami height was almost constant from the beach to the inland limit of inundation (Fig. 16a). Beach erosion again exceeded 0.8 m, and deposition up to 0.7 m occurred on the offshore side of the zone of erosion, indicating the effect of backwash (Fig. 16b). The thickness of onshore

Fig. 14. Simulated waveform of the 1960 Chilean tsunami in Hilo Bay at site TS3 in Fig. 8 (water depth = 10 m).

Based on the stratigraphy, grain size distribution, chemical composition and chronology, our study reveals that evidence for two or possibly three events has been preserved in the sedimentary record on the western shore of Waiakea Mill Pond, Hilo, Hawaii. Observations in the field allowed us to recognise two distinct units within peat and soil at a number of sites; a thin lower unit (Unit 1) dominated by very poorly sorted muddy sand but containing coarse fragments of weathered basalt, and a thicker fine-grained upper unit (Unit 2) (Fig. 7). Both units exhibit a sharp lower contact, although it is somewhat disturbed (Fig. 7), thus suggesting they were laid down by sudden events disrupting the accumulation of peat. While Unit 1 also displays a sharp upper contact with the overlying peat, the upper contact of Unit 2 is not as clear, although the colour change was clearly visible in core WP494c (Fig. 7). The geochemical signature of Unit 1 greatly differs from that of the peat in which it is intercalated, which exhibits high counts of Br and S, but also from that of Unit 2 (Figs. 9, 10b, c). While high counts of Fe and Mn occur, like in Unit 2 (despite some variations between both units) Rb counts are much higher in Unit 1. A number of elements have low counts in that unit, namely K, Ti, V, Cr, Ca, Sr, Zr and As. This is also confirmed by PCA (Fig. 10c). Unit 2, on the other hand, is characterised by high counts of Ca, Sr, Si, K, Ti, V, Cr, As, Mn, Fe and Zr, and moderate counts of La and Ce. It is however interesting to note that the chemical composition of the overlying soil is fairly similar to that of Unit 2, which appears to only be distinguished using PCA (Fig. 10d). Thus, not only the grain size distribution but also the geochemical composition, in association with the magnetic susceptibility of Unit 1 and Unit 2, infer different sources and mechanisms of emplacement, as also reported elsewhere in the literature (e.g., Chagué-Goff et al., 2017). A terrestrial source, most likely associated with a flood event in Wailoa River that occurred sometime around 1520–1660 CE, based on the 14C age, is suggested for Unit 1, which consists of very poorly sorted brown muddy sandy sediment containing sub-angular gravels of weathered basalt. Although the deposit was partly dominated by sand-sized particles, it did not contain any black volcanic sand like that observed on the beach of Hilo Bay, thus allowing us to discount the beach as a source for the sediment and thus strongly inferring an inland origin. The weathered basalt fragments are most likely to have been eroded from old lava flows of Mauna Loa along the Northeast Rift of the volcano, which extends to Hilo Bay (e.g., Lockwood and

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

12

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

Fig. 15. (a) Simulated height of the 1960 Chilean tsunami in Hilo Bay (17 h 15 min after the earthquake); (b) Simulated morphological change by the 1960 Chilean tsunami in Hilo Bay (17 h 15 min after the earthquake). Negative and positive values indicate gross depth of erosion and thickness of deposition, respectively.

Lipman, 1987; Buchanan-Banks, 1993). There is no record of this flood, as it predates the first historical record by Captain James Cook in 1778 who sailed into Hilo Bay but did not land, and the arrival of King Kamehameha in 1791 who then established his government in Hilo (Biography.com Editors, 2017). However, floods are a known occurrence in Hilo District, most frequently of the much larger Wailuku River, on the north side of downtown Hilo, although flooding of the Alenaio Stream and Wailoa River have also been reported during the 2000 floods associated with Tropical Storm Paul (Natural Hazards Hawaii, 2017). A similar deposit was also recorded at sites SH-9 and SH-7 further downstream (b50 and 100 m from site SH-11, respectively) on the western shores of Waiakea Mill Pond, and while we did not attempt to date these deposits, their occurrence suggests that flooding

Fig. 16. (a) Pre- and post-tsunami topography and tsunami height along the profile line (A–B) shown in Fig. 15; (b) Simulated morphological change along the profile line (A–B) shown in Fig. 15. Negative and positive values indicate gross depth of erosion and thickness of deposition, respectively.

from Wailoa River was large enough to go over the banks of Waiakea Mill Pond and leave a deposit. The chemical and sedimentological composition of Unit 2 on the other hand suggests a marine or estuarine source, with in particular high counts in Sr and Ca most probably reflecting the occurrence of carbonates (e.g., Chagué-Goff, 2010; Chagué-Goff et al., 2017), which were absent from the peat underlying Unit 2 and from Unit 1. Indeed, major sources of sediment in the nearshore area in Hilo Bay include weathering of lava flows and soil on land, erosion of lava flows on the shoreline, and skeletons of marine organisms (Dudley, 1986). The increase in other elements (Si, K, Ti, V, Cr, Zr) most likely represents the mineralogical composition of the sediment that was transported inland. Ca can be attributed to the incorporation of carbonates in association with Sr, but also Ca-rich minerals, as reported elsewhere (e.g., Chagué-Goff et al., 2015, 2017), as highlighted in the elemental distribution in PC3, showing that the occurrence of Ca and Sr is not always linked (Fig. 10d). The sudden and marked increase in As suggests a marine or estuarine source for the sediment in Unit 2, as there was no As in the peat and Unit 1, while it is known that As had been discharged from the Canec plant between 1932 and 1963 into Waiakea Mill Pond. Arsenic has been found in subsurface sediments of Waiakea Mill Pond, Wailoa River Estuary and Hilo Bay and attributed to discharge from the Canec plant (Hallacher et al., 1985; Glendon-Baclig, 2007). Thus, As can also be used as a chronological and source material marker, suggesting that Unit 2 was deposited after 1932, and it originated from Hilo Bay and/or Wailoa River Estuary. The 1946 tsunami is therefore inferred as the most probable source for Unit 2 (or at least the 12 cm thick lower part of Unit 2), based on sedimentological, geochemical, chronological (As) data corroborating the well documented historical record of inundation of Shinmachi, which is also confirmed by the results of sediment transport modelling (Figs. 12, 13). The bimodal grain size distribution most probably suggests two sources of sediment for the deposit, as also reported in previous studies (e.g., Shi et al., 1995; Moore et al., 2007). The main sand fraction in Unit 2 (mode at 170 to 148 μm) is slightly finer than the dominant size fraction on Hilo Beach (258 μm), suggesting that the sand might originate from Wailoa River Estuary or be the result of decreasing flow speed and strength inland, which leads to generally reported inland fining of tsunami deposits (e.g., Morton et al., 2007; Chagué-Goff et al., 2011). This is supported by the results of sediment transport modelling that suggests that the sand might originate from Wailoa River Estuary, the beach and the nearshore area off the beach. The fine fraction (mode of 14–21 μm) is only slightly coarser than the size fraction of the peat in the area, and might therefore represent the fraction entrained as the tsunami wave moved inland over the marshy area surrounding Wailoa River Estuary. The contribution from local soil to tsunami deposits has

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

often been reported, resulting in a bimodal grain size distribution (e.g., Shi et al., 1995; Szczuciński et al., 2012). In the present study, the chemical signature of the deposit differs strongly from that of the peat (Figs. 9, 10b), thus inferring that the source of the fine fraction is most likely to be fine mud from Wailoa River Estuary. Unfortunately, no sediment could be collected from the estuary, and thus this assumption cannot be verified. A source further offshore can probably be excluded, based on the results of the study by Dudley (1986) who showed that sediment in Hilo Bay was mainly coarse sand and fine gravel, on the landward and seaward side of the breakwater, respectively. Closer inspection in the laboratory indicated that Unit 2 probably consists of two units (Unit 2A and Unit 2B), with a slight change in stratigraphy and colour at 20 cm depth. This was reflected in an increase in Si counts immediately above 20 cm depth, matching a decrease in magnetic susceptibility and decrease in As counts. There is however little change in grain size distribution, which is bimodal throughout the whole unit, although a slight coarsening upwards followed by a fining upwards is noted between 20 and 17 cm depth. It is difficult to ascertain whether Unit 2B represents a deposit left behind by one of the waves of the 1946 tsunami, or represents the deposit left behind by the 1960 event. It is possible that only a thin soil or peat layer formed between both events, although it was not visible, and could not be detected, as the minimum sampling interval for grain size analysis was 1 cm. The high resolution (1 mm) ITRAX analysis did not reveal a chemically-different layer either, such as autochthonous peat, but it might be due to the short time (14 years) between both events, which would have been too short to allow peat to accumulate. Another possible explanation for the lack of marked change between Units 2A and 2B is erosion by the 1960 tsunami waves, a common occurrence during inundation. However this assumption is not supported by the results of sediment transport modelling, which suggest that sediment was deposited in this area (Fig. 16b). Thus, Unit 2B is tentatively attributed to the 1960 tsunami, based on sedimentological and chemical data, and sediment transport modelling. 6.2. Sediment transport modelling Although the source models of the 1946 Aleutian and 1960 Chilean tsunamis are subjected to debate and adjustment because of the lack of buoy records (Tang et al., 2009), the numerical reconstruction of the inundations of Hilo due to the 1946 and 1960 tsunamis has been attempted a few times in the recent past. Tang et al. (2009) showed that the inundation area by the 1946 tsunami can be explained using the fault parameters determined by López and Okal (2006). Butler et al. (2017) approximated the López and Okal's 1946 tsunami source model using the unit source determined for the NOAA's Short-term Inundation Forecast for Tsunamis (SIFT) (Gica et al., 2008), and identified a suitable source to explain the observed tsunami heights in the Hawaiian Islands and other places in the Pacific. Tang et al. (2009) also investigated the inundation by the 1960 tsunami based on the source model by Kanamori and Cipar (1974), and showed that the reconstructed inundation area and wave height of the 1960 tsunami were smaller than that of the 1946 tsunami. It is however worth noting that the inundation area by the 1960 tsunami might have been slightly larger than that by the 1946 tsunami, since the historical inundation maps show the Waiakea headland was fully inundated in 1960, but only partially in 1946 (Fig. 2). In addition, the inundated area east of Waiakea Mill Pond is greater for the 1960 tsunami, although the inundation areas by both tsunamis were nearly identical west of the pond (Shinmachi). Cheung (2010) also carried out a numerical modelling of the 1946 and 1960 tsunamis and compared the tsunami inundation area and heights. The inundation areas by the 1946 and 1960 tsunamis were similar, and the water elevation reached 4–7 m for the 1946 tsunami and 4–6 m for the 1960 tsunami. Therefore, there is no conclusive evidence suggesting considerable differences in the size of the inundation area and wave heights of the 1946 and 1960 tsunamis in Hilo.

13

Cheung (2010) pointed out that some of the observed heights of the 1946 and 1960 tsunamis are questionable. It is unclear whether the observation included some uncertainties, such as splash. In addition, numerical modelling does not account for the collapse of the breakwater that happened in the 1960 tsunami (but also partially during the 1946 tsunami). This may widen the gap between the observation and simulation. In this study, the simulated inundation area and tsunami height are larger for the 1960 tsunami. Since the simulated inundation area of the 1946 tsunami shows better agreement with the observed flooded area (Figs. 12a, 2, respectively), the inundation area of the 1960 tsunami is likely to be overestimated. However, the simulated tsunami heights (~ 5 m; Fig. 15a) were smaller than the observed ones (4.0–10.7 m; Fig. 2). If the simulated inundation area is adjusted to the observation, the tsunami height will be further underestimated, and vice versa. This is also applicable to the 1946 tsunami. The simulated tsunami heights (~4 m; Fig. 12a) are lower than the measured heights (5.2–7.9 m; Fig. 2), but the simulated and observed inundation area was comparable. Model adjustment for the tsunami height will result in an overestimation of the inundation area. The sediment transport modelling showed that the modelled ‘small’ tsunamis can explain the deposition of sand in the study area with a thickness reaching up to 0.2 m (Figs. 13b, 16b), which is comparable to or even greater than the observation. It is therefore highly likely that the model adjustment for the tsunami height will cause an overestimation of the sediment deposition. Considering the uncertainties in the tsunami source models and the observations, a detailed comparison of the heights and inundation areas by the 1946 and 1960 tsunamis represents a big challenge. Discrepancies between the observation and simulation remain with regards to the tsunami heights. However, comparison of the inundation area and sediment deposition suggests the adjustment for the tsunami height may not improve the modelling results. The current results of the numerical modelling have some implications for the tsunami behaviour and relevant characteristics of sediment transport in Hilo Bay. The second wave was the highest in the simulation of the 1946 tsunami, although the third wave was the highest in the observation. This difference may be ascribed to a factor not considered in the current modelling, such as a submarine landslide as an additional tsunami source (e.g., Okal and Hébert, 2007) or the storm waves that had affected the Hawaiian coasts at the time of the 1946 tsunami (Macdonald et al., 1947). The simulated time series of the 1946 tsunami suggested that a significant decrease in the water level was followed by the largest wave (Fig. 11). The time history of the water level by the modelling of the 1960 tsunami demonstrated that the first and second waves were relatively small (~ 0.5 m), and a considerable decrease in the water level (N2.5 m) occurred between the second and third waves (Fig. 14). The third wave, about 3 m high, was the largest, and the height of the following waves ranged from 1 to 2 m over the next 1.5 h. According to Eaton et al. (1961), the first and second waves were small (1.2–2.7 m) and the ocean receded 7 ft (2.3 m) below normal level. The third wave was the highest and it reached 6.1 to 10.8 m in elevation. The simulated behaviour of the 1960 tsunami in Hilo Bay is thus comparable to the observation. Our modelling data suggest that the 1946 and 1960 tsunamis at the coast of Shinmachi likely behaved in a similar way during the arrival of the largest wave, which followed a sea-level drop. Sand entrainment can be enhanced by a preceding decrease in sea level, since the shear velocity is increased under the condition of decreased thickness of the flow. The magnitude of the sea-level drop and the size of the largest wave control the amount of sediments to be transported; however, in the present study, it is difficult to determine which tsunami transported more sediment from the seafloor, because of the uncertainties in the modelling. Meanwhile, the general behaviour of the 1946 and 1960 tsunamis on land is almost identical. When the tsunamis entered the Wailoa River Estuary and Waiakea Mill Pond, bottom sediments were entrained into the tsunami flow and redeposited in and around the

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

14

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx

water bodies. It is likely that the onshore tsunami behaviour in Hilo Bay is less sensitive to the difference in the wave characteristics at sea, in terms of sediment transport. The modelling results suggest that the tsunami deposit found in the area previously occupied by Shinmachi might have originated from the nearshore seafloor, beach and bottom sediments of the water bodies. 6.3. Accommodation space Limited accommodation space has been suggested (e.g., Oliveira et al., 2008; Nichol et al., 2010) to explain the lack of or poor sedimentary evidence for tsunami inundation. The sedimentary sequence preserved in Wailoa River State Park, which does not exceed 150 cm (including peat and soil), is an example of what is seen elsewhere on Hawaii. Recent Holocene lava flows have resulted in limited accommodation space, so that few tsunami deposits, despite the known history of repeated tsunami inundation, have been identified during geological investigations. It is acknowledged that not all tsunamis leave a visible sedimentary evidence (Chagué-Goff et al., 2016; Judd et al., 2017) and the nature of tsunami deposits depends, at least partly, on sediment availability. Nevertheless, it is expected that sites in low-lying areas inland of sandy beaches are more likely to have preserved some evidence for tsunami inundation. This might also explain the thin deposit attributed to the 1960 tsunami in Shinmachi, compared with the thicker 1946 one, although the availability of source material was similar. On Hawaii, tsunami deposits have been recorded at only a few sites, in valleys on the north-northeast side of the island. They include Pololū Valley (Fig. 1c), where evidence for the 1946 and 1957 Aleutian tsunamis has been reported (Chagué-Goff et al., 2012). Further recent investigations in Pololū Valley and Waipio Valley (Fig. 1c) have also revealed the existence of deposits attributed to these recent tsunamis and older events (unpublished data). A record of tsunamis might also be preserved in Waimanu Valley (Fig. 1c), although research is yet to be carried out at this site. A reconnaissance survey around the island has however revealed that deposits left behind by high energy events, tsunamis or storms, are rare, most likely due to the limited accommodation space and lack of suitable depositional environments. Nevertheless, coarse clast deposits attributed to tsunamis and storms have been identified on a basalt platform near Apua Point, SE Hawaii (Fig. 1c) (Richmond et al., 2011). 7. Conclusions Using sedimentological, geochemical and geochronological analyses, as well as sediment transport modelling, we were able to find the geological evidence of the 1946 and 1960 tsunamis, which had destroyed the community of Shinmachi, Hilo. While only a discontinuous and patchy record was preserved in Wailoa River State Park, most likely reflecting anthropogenic disturbance, two coarse units were however found within peat and soil at three sites on the western shore of Waiakea Mill Pond. The geochemical composition and grain size characteristics of these anomalous layers clearly indicate different sources and depositional processes. The lower thin unit composed of weathered basalt fragments within sandy mud, is inferred to be a river flood deposit that occurred around 1520–1660 CE, most probably due to a tropical storm. The upper anomalous unit, composed of fine volcanic sand and fine silt, is characterised by the appearance of arsenic, calcium and strontium, suggesting a marine or estuarine source from Wailoa River Estuary. Arsenic is here used as a geochronological and source material marker, as it is known to have been discharged in this area between 1932 and 1963. This suggests that this unit was deposited post 1932, and was probably transported inland by the 1946 Aleutian tsunami, as confirmed by results of sediment transport modelling. A thin (3 cm) layer above this unit, characterised by a subtle change in colour, chemistry and grain size distribution, probably represents the deposit left behind by the 1960 Chilean tsunami, in line with results of sediment

transport modelling. The lack of a distinctive thicker deposit expected there might be due to the limited accommodation space, a feature common on the island of Hawaii. Although the island is known to have been repeatedly hit by tsunamis from around the Pacific, there are only a few places in valleys on the Kohala coast in NE Hawaii and Shinmachi, where fine-grained tsunami deposits are preserved. Our study demonstrates that geochemical signatures combined with grain size characteristics provide information about the sediment source, thus allowing the distinction and identification of deposits of various origins, including that of an earlier river flood, as well as the 1946 and 1960 tsunamis. While the exact numerical reproduction of the tsunami and deposit features was difficult, the general behaviour of the tsunami waves and sediments shown by the modelling was comparable with the geochemical and sedimentological data. Acknowledgements This work was supported by the Japan Society for Promotion of Science (JSPS), KAKENHI Grant Number 26302007. We would like to thank Andy Bohlander and Jim Wilson for their assistance in acquiring the necessary permits to carry out our scientific research in the Wailoa River State Recreation Area, which includes the site of Shinmachi. Thanks are also due to Barbara Muffler from the Pacific Tsunami Museum for providing high resolution photos, Volker Roeber from Tohoku University for bathymetry data of Hilo Bay, and Claire Kain from UNSW for writing the code for PCA with R studio. Finally, we would like to thank the two anonymous reviewers for their constructive comments which helped improve the manuscript. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.sedgeo.2017.09.010. References Anon, 1902. Tidal Wave in Hawaii. The Opelousas Courier:p. 2 (January 11th). http:// chroniclingamerica.loc.gov/lccn/sn83026389/1902-01-11/ed-1/seq-2/#date1= 1902&index=1&rows=20&words=HAWAII+TIDAL+WAVE&searchType= basic&sequence=0&state=&date2=1902&proxtext=tidal+wave+in+hawaii&y= 15&x=13&dateFilterType=yearRange&page=1. Bhumbla, D., Keefer, R., 1994. Arsenic mobilization and bioavailability in soils. In: Nriagu, J.O. (Ed.), Arsenic in the Environment, Part 1: Cycling and Characterization. Wiley & Sons, New York, pp. 51–82. Biography.com Editors, 2017. Kamehameha I Biography.com. https://www.biography.com/ people/kamehameha-i-9359827, Accessed date: 6 April 2017. Bissen, M., Frimmel, F.H., 2003. Arsenic—a review. Part I: occurrence, toxicity, speciation, mobility. CLEAN – Soil, Air, Water 31, 9–18. Blott, S., Pye, K., 2001. GRADISTAT: a grainsize distribution and statistics package for the analysis of unconsolidated sediments. Earth Surface Processes and Landforms 26, 1237–1248. Buchanan-Banks, J.M., 1993. Geologic Map of the Hilo 7 1/2′ Quadrangle, Island of Hawaii. US Geological Survey. Buchanan-Banks, J.M., Lockwood, J.P., Rubin, M., 1989. Radiocarbon dates for lava flows from Northeast Rift Zone of Mauna Loa Volcano, Hilo 7 1/2′ Quadrangle, Island of Hawaii. Radiocarbon 31, 179–186. Butler, R., Walsh, D., Richards, K., 2017. Extreme tsunami inundation in Hawai'i from Aleutian-Alaska subduction zone earthquakes. Natural Hazards 85, 1591–1619. Chagué-Goff, C., 2010. Chemical signatures of palaeotsunamis: a forgotten proxy? Marine Geology 271, 67–71. Chagué-Goff, C., Schneider, J.-L., Goff, J.R., Dominey-Howes, D., Strotz, L., 2011. Expanding the proxy toolkit to help identify past events - lessons from the 2004 Indian Ocean Tsunami and the 2009 South Pacific Tsunami. Earth-Science Reviews 107, 107–122. Chagué-Goff, C., Goff, J., Nichol, S.L., Dudley, W., Zawadzki, A., Bennett, J.W., Mooney, S.D., Fierro, D., Heijnis, H., Dominey-Howes, D., Courtney, C., 2012. Multi-proxy evidence for trans-Pacific tsunamis in the Hawai'ian Islands. Marine Geology 299-302, 77–89. Chagué-Goff, C., Goff, J., Wong, H.K.Y., Cisternas, M., 2015. Insights from geochemistry and diatoms to characterise a tsunami's deposit and maximum inundation limit. Marine Geology 359, 22–34. Chagué-Goff, C., Chan, J.C.H., Goff, J., Gadd, P., 2016. Late Holocene record of environmental changes, cyclones and tsunamis in a coastal lake, Mangaia, Cook Islands. Island Arc 25, 333–349. Chagué-Goff, C., Szczuciński, W., Shinozaki, T., 2017. Applications of geochemistry in tsunami research: a review. Earth-Science Reviews 165, 203–244.

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010

C. Chagué et al. / Sedimentary Geology xxx (2017) xxx–xxx Cheung, K.F., 2010. Hawaii tsunami mapping project: data sources, procedures, and products. Final Report for Hawaii Inundation Maps (FOUO) (65 pp.). Cheung, K.F., Bai, Y., Yamazaki, Y., 2013. Surges around the Hawaiian Islands from the 2011 Tohoku Tsunami. Journal of Geophysical Research: Oceans 118, 5703–5719. Cox, D.C., Morgan, J., 1977. Local Tsunamis and Possible Local Tsunamis in Hawaii. 77. Hawaii Institute of Geophysics, University of Hawaii (135 pp.). Dinis, P., Castilho, A., 2012. Integrating sieving and laser data to obtain bulk grain-size distributions. Journal of Sedimentary Research 82, 747–754. Dudley, W., 1986. A Baseline Study of the Geochemistry and Sedimentology of Nearshore Marine Sediments in Selected Areas off the Island of Hawaii. Department of Planning and Economic Development, Business and Industrial Development Division, Ocean Resources Branch, Honolulu (42 pp.). Dudley, W.C., Lee, M., 1998. Tsunami!. 2nd edition. University of Hawaii Press, Honolulu (375 pp.). Dudley, W.C., Stone, S.C., 2000. The Tsunamis of 1946 and 1960 and the Devastation of Hilo Town. Donning Press, Virginia Beach, VA (64 pp.). Eaton, J.P., Richter, D.H., Ault, W.U., 1961. The tsunami of May 23, 1960, on the island of Hawaii. Bulletin of the Seismological Society of America 51, 135–157. Fletcher, C.H., Grossman, E.E., Richmond, B.M., Gibbs, A.E., 2002. Atlas of natural hazards in the Hawaiian coastal zone. Geologic Investigations Series I-2761. U.S. Geological Survey, Denver, CO (182 pp.). http://geopubs.wr.usgs.gov/i-map/i2761/. Folk, R.L., Ward, W.C., 1957. Brazos River bar: a study in the significance of grain size parameters. Journal of Sedimentary Research 27, 3–26. Fujii, Y., Satake, K., 2013. Slip distribution and seismic moment of the 2010 and 1960 Chilean earthquakes inferred from tsunami waveforms and coastal geodetic data. Pure and Applied Geophysics 170, 1493–1509. Gica, E., Spillane, M.C., Titov, V.V., Chamberlin, C.D., Newman, J.C., 2008. Development of the forecast propagation database for NOAA's Short-term Inundation Forecast for Tsunamis (SIFT). NOAA Technical Memorandum OAR PMEL-139 (89 pp.). Glendon-Baclig, C.E.-A.K., 2007. Distribution and Inorganic Speciation of Arsenic in Waiakea Mill Pond and Wailoa River Estuary, Hawaii Island. Faculty of the Tropical Conservation Biology and Environmental Science. University of Hawaii at Hilo, Hilo (114 pp.). Goto, C., Ogawa, Y., Shuto, N., Imamura, F., 1997. IUGG/IOC Time Project, numerical method of tsunami simulation with the leap-frog scheme. IOC Manuals and Guides. UNESCO, Paris (130 pp.). Guyard, H., Chapron, E., St-Onge, G., Anselmetti, F.S., Arnaud, F., Magand, O., Francus, P., Mélières, M.-A., 2007. High-altitude varve records of abrupt environmental changes and mining activity over the last 4000 years in the Western French Alps (Lake Bramant, Grandes Rousses Massif). Quaternary Science Reviews 26, 2644–2660. Hallacher, L.E., Kho, E.B., Bernard, N.D., Orcutt, A.M., Dudley, W.C.J., Hammond, T.M., 1985. Distribution of arsenic in the sediments and biota of Hilo Bay, Hawaii. Pacific Science 39, 266–273. Hosok, O., 2010. Cultural Analysis of the eArly Japanese Immigration to the United States During the Meiji to Taisho Era (1868–1926). (Ph.D. thesis). Graduate College, Oklahoma State University, USA. Iida, K., 1994. A study of the residence and birthplace of the Japanese in Hawaii - 1885 and 1929. Japanese Journal of Human Geography 46, 85–102 (in Japanese with English abstract). Johnston, D., Pettersson, R., Downes, G., Paton, D., Leonard, G., Pishief, K., Bell, R., 2008. Developing an effective tsunami warning system: lessons from the 1960 Chile earthquake tsunami for New Zealand coastal communities. Kōtuitui: New Zealand Journal of Social Sciences Online 3, 105–120. Judd, K., Chagué-Goff, C., Goff, J., Gadd, P., Zawadzki, A., Fierro, D., 2017. Multi-proxy evidence for small historical tsunamis leaving little or no sedimentary record. Marine Geology 385, 203–215. Kanamori, H., 1977. The energy release in great earthquakes. Journal of Geophysical Research 82, 2981–2987. Kanamori, H., Cipar, J.J., 1974. Focal process of the great Chilean earthquake May 22, 1960. Physics of the Earth and Planetary Interiors 9, 128–136. Kelly, M.A., Nakamura, B.S., Barrere, D.B., 1981. Hilo Bay: A Chronological History; Land and Water Use in the Hilo Bay Area, Island of Hawaii. Bernice P. Bishop Museum, Honolulu, Hawaii, USA. Lander, J.F., Lockridge, P.A., 1989. United States tsunamis (including United States possessions) 1690–1988. NOAA Pub. 41-2. Department of Commerce: National Oceanic and Atmospheric Administration (265 pp.). Lander, J.F., Whiteside, L.S., Lockridge, P.A., 2003. Two decades of global tsunamis. Science of Tsunami Hazards 21, 3–88. Lipman, P.W., Moore, J.G., 1996. Mauna Loa lava accumulation rate at the Hilo drill site formation of lava deltas during a period of declining overall volcanic growth. Journal of Geophysical Research 101, 11631–11641.

15

Lockwood, J.P., Lipman, P.W., 1987. Holocene eruptive history of Mauna Loa volcano. US Geological Survey Professional Paper 1350, pp. 509–535. López, A.M., Okal, E.A., 2006. A seismological reassessment of the source of the 1946 Aleutian ‘tsunami’ earthquake. Geophysical Journal International 165, 835–849. Macdonald, G., Shepard, F., Cox, D., 1947. The tsunami of April 1, 1946, in the Hawaiian Islands. Pacific Science 1, 21–37. Mansinha, L., Smylie, D.E., 1971. The displacement fields of inclined faults. Bulletin of the Seismological Society of America 61, 1433–1440. Moore, A.L., McAdoo, B.G., Ruffman, A., 2007. Landward fining from multiple sources in a sand sheet deposited by the 1929 Grand Banks tsunami, Newfoundland. Sedimentary Geology 200, 336–346. Morton, R.A., Gelfenbaum, G., Jaffe, B.E., 2007. Physical criteria for distinguishing sandy tsunami and storm deposits using modern examples. Sedimentary Geology 200, 184–207. Muffler, B., the Pacific Tsunami Museum, 2015. Hawai'i tsunamis. Images of America. Arcadia Publishing, Charleston, USA. Munger, S., Cheung, K.F., 2008. Resonance in Hawaii waters from the 2006 Kuril Islands Tsunami. Geophysical Research Letters 35, L07605. https://doi.org/10.1029/ 2007GL032843. National Geophysical Data Center/World Data Service (NGDC/WDS) 2017. Global Historical Tsunami Database. National Geophysical Data Center, NOAA. https://doi.org/ 10.7289/V5PN93H7 (accessed 7 April 2017). Natural Hazards Hawaii, 2017. Natural Hazards Big Island. University of Hawai'i at Hilo, Flood Hazard Gallery http://hilo.hawaii.edu/~nat_haz/floods/gallery.php, Accessed date: 20 May 2017. Nichol, S., Chagué-Goff, C., Goff, J., Horrocks, M., McFadgen, B., Strotz, L., 2010. Geomorphology and accommodation space as limiting factors on tsunami deposition: Chatham Island, southwest Pacific Ocean. Sedimentary Geology 229, 41–52. Okal, E.A., Hébert, H., 2007. Far-field simulation of the 1946 Aleutian tsunami. Geophysical Journal International 169, 1229–1238. Oliveira, M., Andrade, C., Freitas, M., Costa, P., 2008. Using the historical record and geomorphological setting to identify tsunami deposits in the southwestern coast of Algarve (Portugal). 2nd Tsunami Field Symposium, Abstract Book, Ostuni, Italy, pp. 85–88. Richmond, B.M., Watt, S., Buckley, M., Jaffe, B.E., Gelfenbaum, G., Morton, R.A., 2011. Recent storm and tsunami coarse-clast deposit characteristics, southeast Hawai'i. Marine Geology 283, 79–89. Shepard, F.P., MacDonald, G.A., Cox, D.C., 1950. The tsunami of April 1, 1946. Bulletin of the Scripps Institution of Oceanography 5, 391–528. Shi, S., Dawson, A.G., Smith, D.E., 1995. Coastal sedimentation associated with the December 12th, 1992 tsunami in Flores, Indonesia. Pure and Applied Geophysics 144, 525–536. Soloviev, S.L., Go, Ch.N., 1984. Catalogue of tsunamis on the eastern shore of the Pacific Ocean. Canadian Translation of Fisheries and Aquatic Sciences No. 5078. Canadian Institute for Scientific and Technical Information (293 pp. (Original title: Katalog tsunami na vostochnom poberezh'e tikhogo okeana. Nauka Publishing House, Moscow, 204 pp, 1975)). Sperazza, M., Moore, J.N., Hendrix, M.S., 2004. High-resolution particle size analysis of naturally occurring very fine-grained sediment through laser diffractometry. Journal of Sedimentary Research 74, 736–743. Sugawara, D., Takahashi, T., Imamura, F., 2014. Sediment transport due to the 2011 Tohoku-oki tsunami at Sendai: results from numerical modeling. Marine Geology 358, 18–37. Szczuciński, W., Kokociński, M., Rzeszewski, M., Chagué-Goff, C., Cachão, M., Goto, K., Sugawara, D., 2012. Sediment sources and sedimentation processes of 2011 Tohoku-oki tsunami deposits on the Sendai Plain, Japan — insights from diatoms, nannoliths and grain size distribution. Sedimentary Geology 282, 40–56. Takahashi, T., Shuto, N., Imamura, F., Asai, D., 2000. Modeling sediment transport due to tsunamis with exchange rate between bed load layer and suspended load layer. Proceedings of International Conference of Coastal Engineering, pp. 1508–1519. Takahashi, T., Kurokawa, T., Fujita, M., Shimada, H., 2011. Hydraulic experiment on sediment transport due to tsunamis with various sand grain size. Journal of Japan Society of Civil Engineers, Series B2 (Coastal Engineering) 67, 231–235. Tang, L., Titov, V.V., Chamberlin, C.D., 2009. A Tsunami Forecast Model for Hilo, Hawaii. US Department of Commerce, National Oceanic and Atmospheric Administration, Pacific Marine Environmental Laboratory (54 pp.). Yamashita, K., Sugawara, D., Takahashi, T., Imamura, F., Saito, Y., Imato, Y., Kai, T., Uehara, H., Kato, T., Nakata, K., Saka, R., Nishikawa, A., 2016. Numerical simulations of large-scale sediment transport caused by the 2011 Tohoku Earthquake Tsunami in Hirota Bay, southern Sanriku Coast. Coastal Engineering Journal 58:1640015. https://doi.org/ 10.1142/s0578563416400155.

Please cite this article as: Chagué, C., et al., Geological evidence and sediment transport modelling for the 1946 and 1960 tsunamis in Shinmachi, Hilo, Hawaii, Sedimentary Geology (2017), https://doi.org/10.1016/j.sedgeo.2017.09.010