Numerical assessment of bathymetric changes caused by the 2004 Indian Ocean tsunami at Kirinda Fishery Harbor, Sri Lanka

Coastal Engineering 81 (2013) 67–81

Contents lists available at ScienceDirect

Coastal Engineering journal homepage: www.elsevier.com/locate/coastaleng

Numerical assessment of bathymetric changes caused by the 2004 Indian Ocean tsunami at Kirinda Fishery Harbor, Sri Lanka D. Prasanthi Lanka Ranasinghe a,⁎, Kazuhisa Goto a, Tomoyuki Takahashi b, Jun Takahashi c, Janaka J. Wijetunge d, Takeshi Nishihata e, Fumihiko Imamura a a

International Research Institute of Disaster Science, Tohoku University, Sendai 980-8579, Japan Faculty of Safety Science, Kansai University, Japan Tohoku Electric Power Co., Japan d Department of Civil Engineering, University of Peradeniya, Peradeniya, Sri Lanka e Penta-Ocean Construction Co. Ltd., Japan b c

a r t i c l e

i n f o

Article history: Received 21 August 2012 Received in revised form 8 July 2013 Accepted 11 July 2013 Available online 20 August 2013 Keywords: Bathymetric change Erosion Kirinda Sediment transport Wind wave 2004 Indian Ocean tsunami

a b s t r a c t Thus far various numerical models have been developed and improved to aid understanding of the sediment transport process due to tsunamis. However, the applicability of these models for the field-scale bathymetric change remains a major issue due to the scarcity of measured bathymetric data immediately before and after tsunamis. This study focuses on assessing the applicability of the sediment transport model by comparing the model results with measured bathymetry data obtained one month before and two months after the 2004 Indian Ocean tsunami at Kirinda Fishery Harbor, Sri Lanka. Obtained model results were compared with measured data along four different transects. In particular, similar to the measured data, the model reproduced the bed level change at the harbor mouth well, although it shows some discrepancy on bathymetric change along the shoreline, which is directly affected by littoral drift. Therefore, it is noted that the divergence of reproducing the local bathymetry change is due to the normal wind wave effect on measured data and the model limitations. Hence we included the wind wave effect in modeled data and the discrepancy between measured and modeled data was reduced. Furthermore, the modeled bed level change indicates a dynamic behavior in terms of the net variation during the tsunami flow, such that deposition dominates in the inflow and erosion dominates in the backflow. Both bed level variation and the suspended load concentration reveal that the large amount of eroded sediment attributable to tsunami waves was in suspended form and was deposited in the nearshore area after the water fluctuation had abated. The model results further indicate that eroded sediment at the initial depth deeper than 11 m might be brought by the incoming tsunami waves and deposited in the nearshore area where the depth is shallower than 7 m. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The Indian Ocean tsunami (IOT) on December 26, 2004, caused enormous destruction to life and property in many countries bordering the Indian Ocean, with more than 35,000 deaths being recorded in southern, eastern and northern Sri Lanka (Dahanayake and Kulasena, 2008; Srinivasalu et al., 2007). Although Sri Lanka is located far from the tsunami source, the country is vulnerable to tsunamis because it faces directly towards the northern section of the Sunda subduction zone. The bathymetry configuration of an arrow continental shelf offshore with a sudden drop in water depth from approximately

⁎ Corresponding author. E-mail addresses: [email protected] (D.P.L. Ranasinghe), [email protected] (K. Goto), [email protected] (T. Takahashi), [email protected] (J. Takahashi), [email protected] (J.J. Wijetunge), [email protected] (T. Nishihata), [email protected] (F. Imamura). 0378-3839/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.coastaleng.2013.07.004

150–200 m to 3000 m also increases its vulnerability (Hettiarachchi and Samarawickrama, 2005). Therefore, the wave energy that was transmitted over the shelf came directly toward the land because the shelf is not wide enough for significant energy dissipation (Hettiarachchi and Samarawickrama, 2005; Pattiaratchi, 2005). While the destructive power of tsunamis is very well known, tsunamis can also cause significant morphological change through erosion and deposition caused by tsunami-induced currents and flows (Dahanayake and Kulasena, 2008; Gelfenbaum and Jaffe, 2003; Imamura et al., 2008). However, relative to the onshore study of tsunami deposits, the impact of the tsunami on the offshore bathymetry is poorly understood simply because of the scarcity of pre- and post-tsunami bathymetric data. Numerical models of sediment transport can be used effectively to develop a better understanding of the offshore bathymetric changes caused by a tsunami. To explore the sedimentary process of onshore and offshore sediment transport, various models have been proposed (Apotsos et al., 2011a,b; Asai et al., 1998; Fujii et al., 1998; Jaffe and Gelfenbuam, 2007; Nishihata et al., 2005; Takahashi et al., 1993, 2000).

68

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

Early models predicted sediment transport using formulas based only on the bed load transport rate, which was assumed to be proportional to the Shields number raised to a power (Takahashi et al., 1993); however, it was also recognized by these authors that the suspended load should not be neglected. Later, Kobayashi et al. (1996), using experimental data, proposed a formula estimating the bed load transport rate to be proportional to the Shields number raised to the power of 1.5. Fujii et al. (1998) and Takahashi et al. (2000) suggested that including suspended load entrainment and deposition would improve the accuracy of the prediction of bottom topography changes due to tsunamis. Furthermore, Takahashi et al. (2000) were able to validate the formulas used in the model using a water tank experiment. Takahashi et al. (1993, 2000) numerically investigated the bathymetric changes in the inner bay at Kesennuma City, Japan, caused by the 1960 Chilean tsunami using bathymetric data from four years before and one month after the tsunami. However, opportunities to test the applicability of the model to the field-scale phenomena are very rare because of the scarcity of bathymetric data just before and after the tsunami. At Kirinda Harbor, Sri Lanka, shallow water bathymetry (b7 m depth) was measured in November 2004 and again in February 2005 (Japan International Cooperation Agency et al., 2006), thereby capturing any bathymetric changes caused by the 2004 tsunami. We note, however, that the post-tsunami bathymetry was recorded two months after the tsunami and does not capture precisely the post-tsunami conditions, as natural littoral transport processes were ongoing in that time period (LHI, 1985). Nevertheless, this data provides us with a unique opportunity to test the applicability of sediment transport models and to understand tsunami-induced bathymetric changes both onshore and offshore. A number of studies (Goto et al., 2011; Kihara and Matsuyama, 2010; Nishihata et al., 2006; Takahashi et al., 2009) have focused on the bathymetric changes near Kirinda Harbor, and these reports will be reviewed in Section 3. However, still there are some issues in results comparison with measured data that could not be solved in those previous studies. For instance, previous studies were not heavily focused on the wind wave effect on measured bathymetric data while model results were compared. In this study, we apply a tsunami sediment transport model proposed by Takahashi et al. (2000), which includes the bed load rate, the suspended load rate and the exchange load rate, to test its applicability to the field-scale phenomena and to help understand the process of real-scale bathymetric change by the tsunami. In contrast to previous works, we discuss in detail the dynamic behavior of bathymetric changes during the tsunami flow and wind wave induced bathymetric change after the tsunami event.

(a)

R1 Sri Lanka R2

Indian Ocean

(b) R2

R3 R4

R5

R6

(c)

R6

Kirinda Harbor

2. Study area The Kirinda Fishery Harbor is located on the southeast coast of Sri Lanka, as shown in Fig. 1. The harbor was constructed in 1985, and beach width changes of the order of 60 m were observed even before the construction phase due to the high littoral transport rates common along this coastline (LHI, 1985). During the southwest monsoon (April–October) the dominant wave direction is from the southwest, forcing a north-easterly directed littoral transport. In the northeast monsoon (December–March) the wave direction is reversed and waves from the northeast push sediment down the coast towards the southwest. Harbor function was suspended in June 1986 due to blockage of the harbor entrance by sediment accumulation. Several measures were proposed to address the problem (JICA, 1989), however none were successful. By 2004, several modifications were introduced, such as additional breakwaters (main breakwater (BW-A), secondary breakwater (BW-B) and northern breakwater (BW-C)) and groins (indicated in Fig. 2a) and additional dredging of the harbor entrance. The dredged material was used to create sand hills on shore with heights of 6–8 m above MSL. Despite these measures, the problem of harbor siltation persisted until the 2004 IOT, which caused considerable scour at the harbor entrance.

Fig. 1. Nested grids: (a) Regions 1 and 2; (b) Regions 2, 3, 4 and 5; (c) Region 6; R2, R3, R4, R5 and R6 denote Regions 2, 3, 4, 5 and 6, respectively. The green points in Fig. 1b indicate the tsunami wave heights extracted for locations in Colombo, Hambantota and Kirinda. The positive values represent the altitude, and the negative values represent the water depths.

3. Review of previous studies on sediment transport by the tsunami at Kirinda Harbor Applying tsunami sediment transport model to the real scale phenomenon is a challenging issue. Many researches were focusing to the Kirinda Harbor to assess the tsunami impact as they could

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

(a)

69

(b)

A4

A3

A2

A1

Fig. 2. Measured bathymetries in Kirinda Harbor: (a) November 2004, which represents the pre-tsunami condition, and (b) February 2005, which represents the post-tsunami condition. BW-A, BW-B and BW-C denote Breakwater A, B and C, respectively. Area-C is the common area that is selected to perform the bathymetric change calculation; P1 is the water level (shown in Fig. 6) extracted location at the depth of 6.3 m near the BW-A. The green square is the building at which the inundation height was observed. Blue triangles (A1, A2, A3 and A4)are data extracted points and results are shown in Fig. 10. Contours are in 1 m intervals in both figures, and black dotted lines demarcate the zero and 5 m contours.

compare their numerical results with measured pre- and post-tsunami bathymetries. Among the previous studies conducted on Kirinda Harbor, Nishihata et al. (2006) applied their sediment transport model to consider the bathymetric changes in the harbor. They considered two options: depth uniform eddy diffusivity and linearly increasing diffusivity, which yields less diffusivity near the sea bottom. Based on their results, they concluded that both shear stress-based models mentioned above could successfully predict the severe scour at the harbor mouth (Nishihata et al., 2006). However, this study was limited to the area along a single transect for comparison with the measured and calculated bathymetric changes. Takahashi et al. (2009) also used a sediment transport model to evaluate both the nearshore and the offshore bathymetric changes due to the tsunami and found that the sediment at water depths shallower than 50–60 m was eroded and that the eroded sediment was accumulated along the shoreline during the first runup wave. Based on this model simulation, which was performed until two hours after the tsunami waves hit the harbor, their results were compatible with the measured data in terms of the change in the total sediment volume. Kihara and Matsuyama (2010) performed a three-dimensional model simulation study on Kirinda Harbor to identify the relationships between the tsunami-induced flow and the sediment transport. They noted that local scour and a large amount of suspended sediment are generated around the breakwater heads as the tsunami flow passes around them. Furthermore, they found that the suspended sediment is deposited near the centers of vortices induced by the flow. Their results were compatible with the measured data for the landward side of BW-A, but there was some disagreement between the model and the measurement on the seaward side. Another study was conducted by Goto et al. (2011), who also selected the same area of the Kirinda Fishery Harbor. In that study, they mainly discussed the result of the inundation model and its relevance to bathymetric change. After the detailed validation of the model results,

they concluded that the results were useful for understanding the sediment transport process at Kirinda Harbor. Some of the limitations and the issues that still remain in results comparison are summarized as below. • Although previous studies applied various sediment transport models, no work has discussed the processes of tsunami-induced sedimentation and erosion in detail or the wind wave action on measured bathymetric data. • Although numerical model result was compared with the observed results along single transect by Nishihata et al. (2006), proper verification needs to be done using detailed spatial measured data. • Temporal variation in bed level change in the tsunami profile would not be considered and model simulations were limited to the 2 h from the earthquake (Takahashi et al., 2009). Therefore, sediment transport model simulations need to be done until the water level fluctuation was getting stabilized after a tsunami event so that the bed level change remains as constant. • The model results obtained by Kihara and Matsuyama (2010) were compatible with the measured data at the places that are not directly facing to the open sea. It raised the further questions why model results could not reproduce the better results at the open sea area. Is there any other factors effecting on bed level change other than the tsunami and is it due to wind wave induced bathymetry change on measured post tsunami bathymetry? If the wind wave is the fact short-term recovery could happen after the tsunami that would include in data measured several months after the tsunami. Therefore, considering the above issues mentioned our study is focused on the applicability of the sediment transport model to a realscale phenomenon while also considering the wind wave effect on the bathymetry. We further discuss the improvement of results in comparison with previous studies. We discuss the bathymetric changes in the nearshore area and its temporal variation during the tsunami flow.

70

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

4. Numerical methods 4.1. Bathymetry data A six-level nested grid system was constructed from available topographic and bathymetric data sets for the model. Fig. 1a indicates the larger domain region 1 and region 2, while the rest of the domains are shown in Fig. 1b (R2–R6). Grid resolutions over the various domains ranged from 1862 m in the largest grids to 7.6 m in the innermost grids (Fig. 1c). These domains were created using GEBCO 1 arc-minute topography/bathymetry data, local bathymetric charts of southeast Sri Lanka, and the land elevation data from the Shuttle Radar Topography Mission (SRTM) (Rabus et al., 2003). The November 2004 bathymetry data (Japan International Cooperation Agency et al., 2006), which has spatial resolution of 20 m covering the nearshore region from −7 m MSL to +8 m MSL, was used while making a finer grid domain and it was merged with topographic data collected by Nishihata et al. (2006) (Goto et al., 2011; Takahashi et al., 2008). In this study, we modified the fine-scale grids used by Goto et al. (2011) by additionally including some local features such as the lagoon indicated in Fig. 2a. The bathymetric data measured in February 2005 with the same spatial resolution of 20 m within the range from −8 m MSL to +8 m MSL was used to represent the posttsunami condition (Fig. 2b). The common area (Area-C indicated in Fig. 2a) is selected for all the comparisons and calculations as the February 2005 bathymetry is limited to that area. 4.2. Tsunami propagation and inundation model In this study, the numerical modeling of tsunami inundation follows that of Takahashi et al. (2008) and Goto et al. (2011). Therefore, the inundation process of the tsunami is similar to their results. Nevertheless, we describe the detailed methodology and results for the tsunami inundation process because the results of the sediment transport model are greatly affected by the tsunami inundation process. We used a tsunami propagation and inundation model based on the linear and nonlinear shallow water wave equations, as described by Goto and Ogawa (1982). The equations are solved using the staggered leapfrog finite difference method described by Goto et al. (1997). The linear model was used in the largest domain while the nonlinear model was used in the smallest domains. Water surface elevations and depth-averaged velocities computed in each grid were used as the boundary condition for the next smaller grid. The initial tsunami wave was based on the source model of Takahashi et al. (2008) with the initial sea floor deformation computed using the method of Mansinha and Smylie (1971). The model of Takahashi et al. (2008) was shown to effectively reproduce tide gauge recordings and field observations from Sri Lanka. The fault parameters for this source are listed in Table 1, and Fig. 3 shows the initial wave generated by the source model. 4.3. Sediment transport model We applied the model of Takahashi et al. (2000), which includes both bed load and suspended load transport layers. It is assumed that sediment within each layer is moving in the direction of flow, with

settling of sediment from the suspended layer to the bed load layer and entrainment of sediment from the bed load layer to the suspended load layer. Within each layer, the sediment fluxes are conserved and the exchange load is included to provide the interaction between these two layers (Fig. 4). In this model, the exchange load rate is defined as the balance of the rising load and the settling load. Hence, this model can give the transition of the suspended load without assuming the equilibrium condition of rising and settling load (Takahashi et al., 2000). The conceptual diagram of the proposed model is shown in Fig. 4 and the relevant notations are described next to the following equations, and the governing equations developed based on the model are as follows. Eqs. (1) and (2) are the continuity equations in the bed load layer and the suspended load layer, respectively. ∂hB 1 ∂qBx ∂qBy þ þ þ wex 1−λ ∂x ∂t ∂y

! ¼0

ð1Þ

∂Cs M ∂Cs N ∂C h þ þ wex þ s s ¼ 0 ∂t ∂x ∂y wex ¼ εh

ð2Þ

∂C −w0 C; ∂h

ð3Þ

Where ρs is the density of sand particles, Cs is the mean concentration of the suspended load, CB is the mean concentration of the bed load, C is the concentration at the boundary of both layers, qBx and qBy are the bed load rates in the x and y directions, respectively, hs is the depth of the suspended load layer, hB is the depth of the bed load layer, M(=uD) and N(=vD) are the flow fluxes in the x and y directions, respectively, D is the total water depth, u and v are the flow velocities in the x and y directions, respectively, λ is the porosity, wo is the settling velocity, wex is the exchange load rate, εz is the vertical diffusion coefficient and hB is the bed level. To determine the momentum equations for the bed load and the exchange load that could be applicable for tsunamis, a hydraulic experiment, which includes a large volume head tank to create a large tractive force similar to those generated in tsunamis, was conducted. Based on that experiment, Takahashi et al. (2000) found empirical equations such that the bed load transport rate qB and the non-dimensional exchange load rate wex could be expressed as follows. qB ¼ 21

qffiffiffiffiffiffiffiffiffiffi 3=2 sgd3 τ

wex ¼ 0:012

ð4Þ

pffiffiffiffiffiffiffiffi 2 sgdτ −w0 Cs ;

ð5Þ

Where τ* is the Shields number, s is the relative density of the sand particles, g is the gravitational acceleration and d is the particle diameter. Takahashi et al. (2000) further found that the exponent of bed load formula coincided with the other bed load formulas, similar to the results of Meyer-Peter and Muller (1948), while the exponent of the exchange load formula can be derived based on the Bagnold (1957) formula. Hence, Takahashi et al. (2000) successfully validated their formulas used in the model.

Table 1 Fault parameters for the tsunami source model by Takahashi et al. (2008).1, 2,…., 6 are six fault segments that are used to generate the tsunami source model. Segment

Dislocation(m)

Length(km)

Width(km)

Depth(km)

Strike(°)

Dip(°)

Slip(°)

Longitude(°)

Latitude(°)

1 2 3 4 5 6

14 17 20 12 12 7

200 125 180 145 125 380

150 150 150 150 150 150

10 10 10 10 10 10

323 335 340 340 345 7

15 15 15 15 15 15

90 90 90 90 90 90

94.4 93.32 92.87 92.34 91.88 91.57

3.03 4.48 5.51 7.14 8.47 9.63

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

71

6 5 4 3 2 1

Fig. 3. The initial wave generated by the source model proposed by Takahashi et al. (2008). 1, 2,…, 6 are the fault segments mentioned in Table 1.

The sediment transport model is applied to Region 6 (Fig. 1c), which is the smallest domain, to identify bathymetric changes in the harbor caused by the tsunami. The extracted still water depth, water level and flux flow from the tsunami inundation model and the average satiated suspended sediment concentration are used as the initial conditions for the sediment transport model. As shown in Fig. 5, at each time steps bed profile was updated with the bed level change induced by the tsunami and updated the tsunami flow based on that profile, is used in the next time steps so that bed level and the flow will keep change throughout the simulation. At the boundary, the sediment cannot enter the system but can exit the system. The exchange load rate

will change accordingly during the topographic change calculation to maintain the average satiated suspended load concentration throughout the suspended layer. A flow chart for the calculation of the topographic changes is shown in Fig. 5. A uniform grain size of 163 μm is assumed based on field data collected by Goto et al. (2011), who noted a grain size of 136–163 μm around the harbor. In addition, the Manning roughness coefficient (0.025), settling velocity (0.0173 m/s), satiated suspended load concentration (0.05), specific gravity of the sediment (2.59) and porous ratio (0.4) are used as the basic parameters in the model. Before selecting the average particle size of 163 μm, a sensitivity analysis was performed

Sea Surface

Sea Bed Fig. 4. The conceptual diagram for the model proposed by Takahashi et al. (2000). All of the notations shown in the diagram are explained in Section 4.2.

72

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

locations available in the Southern coast of Sri Lanka in the north east monsoon period, which is effective in the months of December to February (Gunaratne et al., 2011). We use the bed elevation data taken from the tsunami sediment transport model after 360 min after the earthquake as the initial bathymetry data for the wind wave sediment transport model. In other words, this initial bathymetry is already included the tsunami induced bathymetry change in the harbor. Therefore, the output of the wind wave sediment transport model will contain both tsunami and wind wave effects on Kirinda bathymetry.

Initial condition Still water depth Water level Flux Flow The average SSC K=1 Topographic Discharge

change

5. Numerical model results

Flux flow Tractive

5.1. Tsunami propagation and inundation

Bed load rate formula

Exchange load rate formula

Topographic Change Calculation

Boundary condition (Bed load rate and average SSC at the boundary)

Mass conservation equation for suspended load Mass conservation equation for bed load

Yes

K< KE No

K: Number of time steps KE: Total number of time steps SSC: Suspended sediment

End Fig. 5. Model flow chart for the topographic change calculation (after Takahashi et al., 1999).

for several particle sizes and settling velocities that lie within the measured values. This value was chosen because it best reproduces the scour at the harbor mouth, which is the main feature of the bathymetric change in Kirinda Harbor presented by many studies (Kihara and Matsuyama, 2010; Nishihata et al., 2006).

4.4. Wind wave model Although the analysis of the measured data was performed soon before (November 2004) and after (February 2005) the 2004 IOT (Japan International Cooperation Agency et al., 2006), we cannot exclude the possibility that the monsoon wind waves, in addition to the tsunami, affected the measured bathymetry in February 2005. We carried out wind wave sediment transport modeling of tsunami-influenced bathymetry for the duration of December 2004 to February 2005. The same validated model used by Nishihata et al. (2009) is used to grasp the bathymetric change due to wind waves. This model is able to calculate the local and alongshore sediment transport rates, considering both wave and wave-induced currents. Further, we preliminary applied the model for the Kirinda Harbor for November 2005 to February 2006 period (Ranasinghe et al., in prep.). The obtained current distribution and the sediment movement by the model were well comparable with wind wave effect on bathymetric change in NE monsoon period (LHI, 1985, 2005). As recorded wave data at Kirinda is limited for short period, we used the transformed wave data from long period wave recording

The inundation model is well validated by Takahashi et al. (2008) and Goto et al. (2011), and our results are fundamentally similar to theirs, as summarized below. • Because the exact arrival time at the Kirinda Fishery Harbor is unknown, arrival times at Colombo and Hambantota (Fig. 1b) were compared with model predictions. According to Nishihata et al. (2009) and Inoue et al. (2007), the first tsunami wave was observed at Hambantota at approximately 9:10–9:20 AM and 9:22 AM (local time), respectively, while the model predicted that the time for the peak of the first wave would be 9:12 AM, which is 133 min after the earthquake. Though observed arrival time at Colombo was 9:50 AM (Niroshinie, 2008) and 9:48 AM (Inoue et al., 2007), the result according to the model is 9:45 AM. However, the time that was required for the tsunami to extend from Hambantota to Colombo is quite similar to the observed value, even though the modelpredicted time is much earlier than the observed data. • The model results show that the tsunami wave approached Kirinda Harbor from the southeast. The first peak of the first and largest tsunami wave reached Kirinda Harbor 130 min after the earthquake (Fig. 6), occurring at approximately 9:09 AM local time. According to Shibayama et al. (2005), Nishihata et al. (2006) and Inoue et al. (2007), tsunami inundation heights at the buildings in front of Kirinda Harbor (Fig. 2a) were observed to be 6.9–9.3 m, 8.4 m, and 8.7 m, respectively. Showing a better agreement with the observed values, this model predicted a maximum inundation depth of 7.3 m near the same buildings. • The tsunami overtopped all of the breakwaters and hence inundated the entire harbor, reaching as far inland as the lagoon area located on the north side of the harbor basin (Fig. 7b). The first incoming flow was mainly directed toward the lagoon through the opening between BW-B and BW-C and over the natural dune at the north side of BW-C. As a result, the area of maximum inundation spread up to 800 m between BW-A and the groin, 500 m between BW-B and BW-C and 1 km on the northside of the BW-C. • The return flow was mainly concentrated between BW-A and BW-B and to the north of the groin (Fig. 7c). According to the model results, the tsunami return flow was concentrated toward the breakwater heads, especially through the harbor mouth, and toward the head of the groin, which agrees closely with the findings of Nishihata et al. (2006). In this way, we used the inundation model results as an input for the sediment transport model to identify the tsunami-induced bathymetric change. 5.2. Bathymetric changes Our modeling results suggest that the first incoming wave eroded the sea bottom sediments at a depth range of −11 m (MSL) to −25 m (MSL), which is the offshore boundary of the smallest region. Nearshore and onshore bathymetric changes in region 6 are discussed as follows

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

73

Fig. 6. Water level variation (red solid line) at Point P1 indicated in Fig. 2a, and net volume change of sediment (dotted purple line) within the area shown in Fig. 2a; blue dots denote times after the earthquake at which the erosion/depositionvolume shown in Table 2 was calculated.

and flow dynamics associated with those changes will be discussed in the next section. • As illustrated in Fig. 7, the first incoming flow deposited a considerable amount of sediment along the coastline, which was transported from offshore. Moreover, the model results demonstrate that the return flow was unable to move the accumulated sediment along the shoreline (between BW-A and the groin and north of the BW-C) left by the incoming flow because the existing sand hills along the BW-A and the north side of the BW-C directed the return flow through the lowland (Fig. 7c and d). • With the first incoming flow, the harbor mouth, which was completely closed before the tsunami attack, began to erode and back rushing created about 5 m erosion at the mouth. These eroded sediments were transported by progressive tsunami waves and deposited in the harbor basin. • The return flow, which creates the scour at the harbor mouth, directed along the breakwaters resulted in a large amount of scour at the heads (Fig. 7). The same phenomenon is described by Nishihata et al. (2006) such that the backward-rushing tsunami flow tends to concentrate along coastal structures, causing extensive bed erosion. Moreover, the area between BW-B and BW-C showed considerable erosion by the tsunami return flow, which passed through the lowland lagoon area. The results were also showed that maximum erosion occurs at the point where the sediment transport rate becomes the maximum gradient. This in turn suggests that spatial acceleration of the flow causes the major erosion. That is probably the reason why we have the maximum erosion at the harbor mouth and not in between BW-A and BW-C where the maximum flow velocities are recorded (Fig. 7b). • Fig. 8 illustrates the temporal variation of bathymetric change during the second to fifth tsunami waves. It also suggests that the net change remained almost the same during the latter waves, namely the fourth and fifth (Fig. 8d, e and f). The same phenomenon can be seen in Fig. 11, which shows the time series of the bed level change and the elevation in four selected points. Hence, the model results obtained at the time after the fifth wave of the tsunami profile were selected for comparison with the measured data. A common area (Area-C) was selected to calculate the net deposition and erosion for both measured data and the modeled results (Fig. 2a). The analysis of the measured data in Area-C shows that the total deposition and total erosion in that particular area were 2.8 × 105 and 1.4 × 105 m3, respectively. The model

prediction values for total deposition and total erosion after the fifth tsunami wave were 3.1 × 105 and 1.7 × 105 m3, respectively. 5.3. Wind wave effect Fig. 9 shows the bathymetric change due to wind waves during the two months after the tsunami event. As the dominant wave direction in this north east monsoon period is towards the southwest, wind waves were able to move the sediments deposited by tsunami at the nearshore area of BW-C and those transported sediments were deposited near the harbor mouth. Furthermore, it can be seen that a considerable amount of nearshore sediment was moved near the shoreline above the BW-C. Similar behavior can be observed between BW-A and the groin, with the erosion of some sediments at the nearshore area because of the wind wave action. During these two months, maximum deposition was modeled as 1.5 m near the harbor mouth and maximum erosion of 2 m occurred near the BW-C. 6. Discussion 6.1. Observed tsunami damage at Kirinda Harbor At approximately 9:15 AM, the tsunami wave hit the Kirinda Harbor creating a maximum inundation depth of 8.7 m observed at a building near the harbor (Inoue et al., 2007), and the inundation area extended up to 500 m from the shoreline (Nishihata et al., 2006; Shibayama et al., 2005). After the tsunami attack, there was excessive erosion at the breakwaters and the groin heads, as well as at the harbor entrancewith a maximum value of 3 m recorded. Except near to those structure heads, high sediment concentration was observed along the shore line. Furthermore, observed data illustrated maximum sedimentation of 4 m near BW-C and considerable deposition inside the harbor, which was expected to be deposited with the tsunami incoming flow. Measured data also revealed that the ratio between the erosion and deposition volumes of sediments for the Area-C illustrated in Fig. 2a was 0.52 (based on Fig. 10a). 6.2. Flow dynamics created by the tsunami Our model results indicated that the tsunami overtopped all of the harbor structures with the 6 m wave height when it hit the Kirinda Harbor

74

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 7. Bed level variation and current velocity during the first hour after the tsunami hit Kirinda Harbor:(a) 125, (b) 130, (c) 138, (d) 159, (e) 178 and (f) 186 min after the earthquake. The color map shows the bed level change, and the arrows represent the velocity vectors. T-A, T-B, T-C and T-D are the selected transects A, B, C and D, respectively.

(Fig. 11), including the breakwaters and the groin, and the current velocity generated in lowland area was about 5 m/s, which is comparatively higher than that in other areas (Fig. 7b). Furthermore, it shows that during the first return flow, high current speed was generated in the flow, and flow was mainly concentrated in a narrow strip toward the breakwaters and the groin (Fig. 7b). According to Fig. 11 (see the graphs At point A1 and A2), the tsunami back flow was unable to pass through the points A1 and A2. This channel effect was enhanced by artificial and natural sand hills and hence the back-rushing flow was directed through the lowland area. This may have caused increased current velocity and the turbulence in the tsunami back flow. As this high turbulence would increase the suspended sediment concentration in the tsunami flow, the inundation model results further emphasize the importance of including the suspended load calculation in the sediment transport model.

Furthermore, the large scours at the harbor entrance and at the heads of structures could have been caused by the high velocity of the concentrated back flow. The funneling of the tsunami wave energy caused by the bay effect is primarily responsible for the increase in wave height as the tsunami waves move in on such partially confined coastal formations (Wijetunge, 2009a,b) and the tsunami wave decreases in magnitude and spreads out slowly while it propagates through the bay areas. Therefore, in addition to back flow effects, which were also determined in previous studies (Goto et al., 2011; Nishihata et al., 2006; Takahashi et al., 2009), the phenomenon described above would also have enhanced the erosion, which had been noted at the headlands and at the artificial coastal structures, such as the breakwaters. Similarly, high sedimentation occurred in the bay areas because of the diffracted tsunami inflow into the bays as it spread

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

75

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 8. Temporal variation of bed level change: (a) 231, (b) 268, (c) 296, (d) 314, (e) 338 and (f) 347 min after the earthquake; the color map shows the bed level change.

throughout the bay area slowly, allowing the material brought in with the tsunami inflow to deposit.

higher, respectively, than the measured volumes. Hence we did further analysis, selecting four transects (Fig. 7a) near the Kirinda Harbor.

6.3. The effect of the tsunami on bathymetric changes 6.3.1. Total change The calculated ratio for the modeled erosion to deposition volume within the Area-C was 0.54 (Fig. 10b). Here, the bathymetric change after the fifth wave (349 min after the earthquake) was considered for the calculation because the water level variation seemed to be stable during the latter half of the fifth wave. Therefore, our model results comparison shows the better agreement than that of in other studies like Takahashi et al. (2009) which was limited their simulation only up to 2 h from the earthquake. Although this ratio comparison between the measured data and the model results shows a good agreement, the model calculated erosion and deposition volumes are 22% and 10%

6.3.2. Spatial variation Fig. 12 shows the bathymetric change 349 min after the earthquake calculated by the model and the measured bathymetric change from November 2004 to February 2005 throughout transects. Out of the four transects, transect B, which represents the harbor mouth, has a correlation coefficient (R2T) of 0.917 for the measured and the modeled datasets, indicating a better agreement between them (Fig. 12B). However, in transects A, C and D, a quantitatively good agreement between the measured and the modeled data is not observed, even though there is a qualitative agreement in the bed level change. This discrepancy can be seen especially at depths shallower than −3 m MSL (referring to the initial bathymetry).

76

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

These differences between the measured and the model result shown in the bed level change might be primarily attributable to two conditions. One is the normal wind wave effect on measured bathymetry and second is the model assumptions and limitations.

Fig. 9. Bathymetric change due to wind waves: the dashed line is the shoreline in February 2005 and the pink line is the computed shoreline 349 min after the earthquake.

(a)

6.3.3. Temporal variation By calculating the cumulative erosion and deposition with respect to the initial bathymetry at the times when the tsunami wave was at its peak, trough, and the zero crossing at the point P1 (Fig. 2a), which is located near the harbor entrance, the dynamic variation in the sediment movement was identified. Fig. 6 shows the water level variation at point P1 during several repetitive tsunami waves and Table 2 shows the calculated erosion and deposition volumes in Area-C at the times, which are shown in Fig. 6. However, unavailability of measured bathymetric data during the tsunami profile restricts the comparison of modeled bathymetric data with observed values. Until the second wave, the bathymetry changes in Kirinda Harbor were shown to be dominated by erosion, regardless of the dynamic variation in the bed level during the incoming and return flows, indicating massive erosion during the first return flow with an erosion and deposition ratio of 10 (Table 2). The model also suggests that the incoming flows brought the sediments with the tsunami flow and deposited them inland and onshore (Fig. 7b), while eroding or moving some of the deposited material back to the nearshore with the return flows (as an example: the area between BW-A and BW-C in Fig. 7b and c). However, after the third wave, the bathymetric changes were heavily dominated by deposition, and the bathymetric changes seemed to be stabilized by the end of the fourth wave (Table 2, Figs. 6 and 8). Fig. 11, which includes the time series of the bed level change and elevation at four points, also clearly explains the above phenomenon. Suspended load concentration associated with the tsunami profile also explains the same phenomenon described above. Fig. 13 shows that the suspended load concentration in tsunami flow is at its maximum during the 1st return flow and it gradually decreases as water level variation reduces. In other words, a large amount

(b)

No data from wind wave

(c)

Fig. 10. Variation in measured bathymetries from November 2004 to February 2005 (left), modeled bathymetric change in Kirinda Harbor 349 min after the earthquake due to the tsunami (middle), and modeled bathymetric change due to the tsunami and wind wave effects (right). In the left-hand figure, the black line is the shoreline in November 2004, and the dashed line is the shoreline in February 2005. In the right-hand figure, the black line is the initial shoreline before the tsunami, and the pink line is the computed shoreline 349 min after the earthquake.

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

At Point A1

10

0.8

6

0.6

4

0.4

2

0.2

0

0.0

-2

-0.2

-4

-0.4 -0.6 180

220

260

300

340

At Point A2

1.0

8

0.8

6

0.6

4

0.4

2

0.2

0

0.0

-2

-0.2

-4

-0.4

-6 100

-0.6 140

180

220

260

300

340

At Point A3

10

1.0

8

0.0

6

-1.0

4

-2.0

2

-3.0

0

-4.0

-2

-5.0

-4

-6.0

-6 100

Bed Level Change/m

140

10

Elevation/m

1.0

8

-6 100

77

-7.0 140

180

220

260

300

340

At Point A4

10

Elavation

5.0

Bed level change

8

4.0

6

3.0

4

2.0

2

1.0

0

0.0

-2

-1.0

-4

-2.0

-6 100

-3.0 140

180

220

260

300

340

Time after the Earthquake/min Fig. 11. The modeled bed level change (green solid line) and the elevation (red solid line) at four selected points A1, A2, A3 and A4 shown in Fig. 2a.

of eroded sediment during the 1st return flow was located as suspended load and was deposited as water level variation stabilized. According to Goto et al. (2011), there was less erosion of the sand hills or sand dunes compared to the deposited sediment volume around the harbor. Therefore, after the overall tsunami effect, a positive sediment balance was achieved in the considered area (Area-C in Fig. 2a), suggesting a sediment supply from outside of that area. 6.4. Significance of wind wave effect According to LHI (1985) and transformed wave conditions at Kirinda 15 m depth, the predominant wave direction is east and northeast from December to March. Therefore, waves coming from these directions would have had an effect on the February 2005 bathymetry, which

Fig. 12. Measured and modeled bed level changes and initial bed level along the transects A, B, C and D. The purple line shows the bed level change based on November 2004 (N04) and February 2005 (F05). The red dotted line shows the modeled bed level change due to the tsunami 349 min after the earthquake and the dark green line shows the modeled bed level change due to both the tsunami and wind wave effects. The light blue line shows the initial bed level in November 2004 bathymetry. R2T and R2T + Ware the correlation coefficients between measured data and modeled bed level change due to the tsunami and the combined effect of the tsunami and wind waves, respectively.

was measured two months after the tsunami. The current velocity distribution obtained from the inundation and propagation model (Fig. 7) and the water level variation at selected points (Fig. 11) clearly show that the tsunami back flow did not overtop the higher land but concentrated on the lower land, creating a channeling effect. In other words, the back flow, which contributed significantly to the erosion, was not passing through the shallower area in transect D but rather through the shallow area of transect A. Therefore, this result suggests that erosion occurred near the shore line in transact D and deposition occurred at shallower depths in transect A, which were enhanced by the normal wind-induced littoral drift rather than the tsunami. Fig. 10c shows the bathymetric change due to both tsunami and wind waves. This figure clearly illustrates that sediment deposited at the nearshore by the tsunami was moved to the shoreline. Another

78

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

Table 2 Deposition, erosion and net volume change within the area indicated in Fig. 1d, with respect to initial bathymetry at the time when tsunami waves were at their peak (P), trough (T) and the zero crossing (Z) at the point P1; E/D is the ratio of erosion volume to deposition volume; the minus sign indicates erosion. Time after the earthquake (min)

Description

Deposition (m3)

Erosion (m3)

Net (m3)

Ratio (E/D)

121 130 138 159 178 186 220 231 243 268 279 287 296 300 306 314 324 338 345 347 349 352 360

Z P Z T Z P Z T Z P Z T Z P Z T Z P Z T Z P

104 31,462 25,949 32,207 53,019 73,632 127,388 93,653 131,940 199,319 206,964 150,550 177,786 220,950 229,243 221,611 220,620 288,032 302,021 304,804 305,867 282,429 303,936

20 67,835 260,455 269,682 198,173 143,980 136,887 180,061 164,670 140,895 164,146 210,822 199,218 173,472 187,416 187,386 191,164 169,486 167,627 165,981 166,649 182,694 184,672

84 −36,373 −234,507 −237,475 −145,153 −70,347 −9499 −86,408 −32,730 58,424 42,819 −60,272 −21,432 47,478 41,827 34,225 29,456 118,546 134,394 138,823 139,218 99,735 119,264

– 2.16 10.04 8.37 3.74 1.96 1.07 1.92 1.25 0.71 0.79 1.40 1.12 0.79 0.82 0.85 0.87 0.59 0.56 0.54 0.54 0.65 0.61

1st Wave

2nd Wave

3rd Wave

Matsuyama (2010), the model result had some inconsistencies with the observed data at the open sea area (between BW-A and groin). As shown in Fig. 12 (see transect D), the results at the open sea area is improved in our model most likely because we considered the posttsunami recovery process by the wind wave that are largely affected to the open ocean area. Further, bed level change along the harbor mouth also gives the better results in this study than Nishihata et al. (2006). This in turn suggests that modeled maximum bed erosion by tsunami would have been recovered by approx. 1.5 m by the posttsunami wind wave and consequently the measured bed level at that point is consistent with the modeled value in our study (see transect B in Fig. 12). These lines of results suggest that the sediment transport models proposed by Kihara and Matsuyama (2010) and Nishihata et al. (2006) are also well applicable to explain the real scale tsunami events although the effect of the wind wave should be included and the model simulation should be continued until the bed level change by the tsunami become stable. 6.6. Implications for future research

4th Wave

5th Wave

important finding is that tsunami-induced maximum scour of 4.5 m at the harbor mouth was reduced by 1.5 m deposition due to wind waves, and finally the maximum depth at the mouth remained at 3 m due to the combined effect of tsunami and wind waves, and a similar value was also recorded in measured data (Fig. 12). We calculate the correlation coefficients (R2T + W) for measured data bathymetric change and the modeled bathymetric change due to tsunami and wind waves. As the correlation coefficients for results compared along transects were improved, as shown in Fig. 12, we infer that the wind wave effect on the bathymetry during two months of period was significant. We compare the difference between measured (2005 Feb) data and the modeled data by tsunami sediment transport model (2004 Dec) and by wind wave sediment transport model (2005 Feb) separately to understand the interpretation of local bed level changes by the models (Fig. 14). The difference between measured and modeled data has been reduced significantly at the open sea areas especially at the area between BW-A and groin and, the area near to BW-C head when the wind wave effect is included. Further, disagreement between measured and tsunami model data at harbor mouth area also reduced and give a similar results as measured data after considering the wind wave effect. Both Fig. 14a and b show more than 3 m difference in bed level when it compares to measured data around the groin. This is because we have lack of measured data in this area which led us to interpolate available data to get the smoothed bathymetry for the tsunami sediment transport model. Therefore, model gives unusual erosion at the groin because of the spatial acceleration of the flow at the steeper slope, and this misinterpretation is progress to the wind wave model as well. Apart from that, the area above the BW-C still shows some quantitative disagreement with the measured data even though the model results of that area also improved considerably when the wind wave effect is considered. Further, our wind wave model results may also be given lower values than measured as we use the transformed wave data which are less in magnitude than actual wave data at Kirinda. 6.5. Results comparison with other studies Under this section we discuss how the model results were improved when we include the wind wave effect. In the study by Kihara and

The coefficients used in the bed load and rising load formulas in this model were independent of the grain size. Moreover, based on a hydraulic experiment, Takahashi et al. (2011) suggested that both coefficients vary according to the size of the sand grains; therefore, they updated the empirical equations (Eqs. (4) and (5)) with the upgraded coefficients. Using these updated coefficients and the grain size distribution as a variable in the domain would be a further step to incorporating the local bathymetric effects that could not be achieved in this study. Additionally, the cut-off depth of the calculation of the bed level change had to be limited to overcome computational problems. Hence, the cut-off depth was limited to 5 cm because some instability was created when the model calculated the bed level change in a shallower depth. If the total water depth was less than the cut-off depth in a grid, that particular grid was excluded from the bed level change calculation for the current time step. When the cut-off depth was increased further, the resultant bed level change became less and less as the number of omitted grids was also increased. On the other hand, when the cut-off depth was further decreased, a blow-up problem occurred. Therefore, we had to fix the cut-off depth at 5 cm for this simulation. This limitation is also a reason why the tsunami model could not reproduce the bed level change at shallower depths. Furthermore, this model assumed a unique size for the sand grains, which vary in actual conditions. This assumption allows the model to overestimate the erosion when the actual grain size is higher than what was assumed and vice versa when the actual grain size is less than the assumed value. The limitations and assumptions used in the model could also account for the difference in the measured and model values for bed level change. Further, if there is a gap between measured bathymetry data and the tsunami event and the wind wave effect is significant in considered area wind wave effect on measured bathymetry need to be taken in to consideration while the tsunami sediment transport model is performed. 7. Conclusion Our work brings to light two important results. First, we check the applicability of the sediment transport model by Takahashi et al. (2000) using data measured pre- and post-tsunami within a couple of months of the tsunami event. The calculated erosion and deposition ratio for the Kirinda Harbor using the measured data is 0.52, whereas the same ratio obtained by the model after the fifth tsunami wave, when the dynamic bathymetric changes seem to have stabilized, is 0.54. However, the model calculated erosion and deposition volumes are 22% and 10% higher, respectively, than the measured volumes. Furthermore, while the bed level change across the harbor mouth shows good agreement with the measured data, there are some discrepancies

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

79

(a)

(b)

(c)

(d)

Fig. 13. Variation of suspended load concentration (SLC): (a) 138, (b) 268, (c) 306, (d) 349 min after the earthquake of the 1st return flow, 3rd incoming flow, 4th return flow and 6th incoming flow, respectively; the color map shows the SLC.

in the bathymetric change comparison at the locations directly affected by wind wave-induced littoral drift. On the other hand, wind wave model results illustrate that sediments deposited by the tsunami in the nearshore area were moved towards the shoreline due to wind waves. Finally, the wind wave model of tsunami-influenced bathymetry can reduce the difference between measured and model data by proving the influence of wind wave-induced littoral drift during two months. Hence, the results predicted by the sediment transport model, which include the concept of an exchange load rate, are compatible with the measured data, and results were further improved by including wind wave effects during two months after the tsunami event. Second, we clarified the processes of erosion and sedimentation during the tsunami profile using the sediment transport model. The onrushing waves were the main source of sedimentation, and the deposition of sediment occurred especially along the shoreline. A backward-

rushing flow was directed toward the breakwater head through the lowland areas and contributed to moving some eroded sediments from onshore to nearshore. However, the presence of the sand hills and sand dunes behind the shore prevented the back flow from passing through them; hence it was unable to erode the accumulated sand along the shoreline. A detailed analysis of bed level change in the vicinity of the harbor, which has a depth range of −7 m to +8 m MSL, found that the first backward-rushing wave caused the majority of the erosion measured by the bathymetry, although dynamic changes in the deposition and erosion were observed during subsequent inflow and back flow, respectively. Because the net change ultimately consisted of deposition after all of the tsunami waves had hit, it could be inferred that the sediment eroded during the backward-rushing flows would be located in the suspended load and deposited in the nearshore area, where the depth is shallower than 7 m, when the flow fluctuation became calm.

80

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81

(a)

(b)

Fig. 14. Difference between measured and modeled bed level data (a) without wind wave effect (b) with wind wave effect. Ash color area shows the land and the dashed line is the shoreline in February 2005.

Post-tsunami re-arrangement of the bathymetry by the wind wave during December 2004 to February 2005 is compatible with measured data in February 2005. Moreover, the correlation coefficient of modeled and measured data is increased when the wind wave effect is considered. Therefore, we infer that the applicability of tsunami sediment transport model of Takahashi et al. (2000) is successfully verified for the evaluation of bathymetric changes attributable to tsunamis in a real scale tsunami event using the Kirinda bathymetric data. Acknowledgments We would like to convey our sincere thanks to the Japan International Cooperation Agency (JICA) for providing the bathymetric data from November 2004, February 2005, November 2005 and February 2006. Furthermore, we extend our gratitude to the Ceylon Fishery Harbor Corporation (CFHC) for providing the old (before 2004) bathymetric data for this research and Penta-Ocean Construction Co. Ltd., Japan for providing the numerical code of the wind wave model. This research was supported by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (no. 22241042) for the field survey and data analysis and from the Japan Nuclear Energy Safety Organization for the numerical modeling. We also convey our gratitude to all reviewers for their valuable comments which helped to improve this manuscript. References Apotsos, A., Gelfenbaum, G., Jaffe, B., 2011a. Process‐based modeling of tsunami inundation and sediment transport. Journal of Geophysical Research 116, F01006. http:// dx.doi.org/10.1029/2010JF001797. Apotsos, A., Gelfenbaum, G., Jaffe, B., Watt, S., Peck, B., Buckley, M., Stevens, A., 2011b. Tsunami inundation and sediment transport in a sediment limited embayment on American Samoa. Earth-Science Reviews. http://dx.doi.org/10.1016/j.earscirev. 2010.11.001. Asai, D., Imamura, F., Shuto, N., Takahashi, T., 1998. Estimated tsunami heights and sand transport at Iruma Izu in the 1984 Tokai earthquake. Proceedings of Coastal Engineering, JSCE 45, 371–375 (in Japanese). Bagnold, R.A., 1957. The flow of cohesionless grains in fluids. Philosophical Transactions of the Royal Society of London 249.

Dahanayake, K., Kulasena, N., 2008. Geological evidence for paleo-tsunamis in Sri Lanka. Department of Geology, University of Peradeniya. Science of Tsunami Hazards 27 (2), 54–61. Fujii, N., Oomori, M., Takao, M., Kanayama, S., Ootani, H., 1998. On the deformation of the sea bottom topography due to tsunami. Proceedings of Coastal Engineering, JSCE 45, 376–380 (in Japanese). Gelfenbaum, G., Jaffe, B., 2003. Erosion and sedimentation from the 17 July, 1998 Papua New Guinea tsunami. Pure and Applied Geophysics 160, 1969–1999. Goto, C., Ogawa, Y., 1982. Numerical Calculation Method Using the Leapfrog Method, Data of Civil Engineering Department. School of Engineering, Tohoku University (52 pp., (in Japanese)). 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. Goto, K., Takahashi, J., Oie, T., Imamura, F., 2011. Remarkable bathymetric change in the nearshore zone by the 2004 Indian Ocean tsunami: Kirinda Harbor, Sri Lanka. Geomorphology 127 (1-2), 107–116. Gunaratne, P.P., Ranasinghe, D.P.L., Sugandika, T.A.N., 2011. Assessment of Nearshore Wave Climate off the Southern Coast of Sri Lanka. ‘ENGINEER’ the journal of Institute of Engineers Sri Lanka, vol. XXXXIV, No. 02 33–42. Hettiarachchi, S.S.L., Samarawickrama, S.P., 2005. University of Moratuwa, Sri Lanka Experience of the Indian Ocean Tsunami on the Sri Lankan coast. International Symposium Disaster Reduction on Coasts. Monash University, Melbourne, Australia, pp. 2–3. Imamura, I., Goto, K., Ohkubo, S., 2008. Numerical model for the transport of a boulder by tsunami. Journal of Geophysical Research 113 (C01008), 1–12. Inoue, S., Wijeyewickrema, A.C., Matsumoto, H., Miura, H., Gunaratna, P., Madurapperuma, M., Sekiguchi, T., 2007. Field survey of tsunami effects in Sri Lanka due to the Sumatra–Andaman Earthquake of December 26, 2004. Pure and Applied Geophysics 164, 395–411. Jaffe, B.E., Gelfenbuam, G., 2007. A simple model for calculating tsunami flow speed from tsunami deposits. Sedimentary Geology 200, 347–361. Japan International Cooperation Agency, 1989. The study on sand Drift in the southeastern Coast of Sri Lanka Final report, Japan. Japan International Cooperation Agency, PADECO Co. Ltd., Nipon Koei Co. Ltd., Overseas Agro-Fisheries Consultants Co. Ltd., 2006. Recovery, rehabilitation and development project for tsunami affected area of southern region in the Democratic Socialist Republic of Sri Lanka: final report (pp. 8-63-71). Kihara, N., Matsuyama, M., 2010. Numerical simulations of sediment transport induced by the 2004 Indian Ocean tsunami near Kirinda port in Sri Lanka. Proceedings of the International Conference on Coastal Engineering, No 32, Shanghai, China. Kobayashi, A., Oda, Y., Fujii, N., 1996. Study on sand movement caused by the tsunami. Proceedings of Ocean Engineering 43, 691–695 (in Japanese). Lanka Hydraulic Institute Ltd., 1985. Note on Siltation Conditions and Proposed Investigations. 1–10. Lanka Hydraulic Institute Ltd., 2005. Wave climate & Sediment Transport (Kalpitiya – Oluvil)-Final report prepared for: Coastal Resources Management Project Component B: Institutional Strengthening.

D.P.L. Ranasinghe et al. / Coastal Engineering 81 (2013) 67–81 Mansinha, L., Smylie, D.E., 1971. The displacement fields of inclined faults. Bulletin of Seismological Society of America 61, 1433–1440. Meyer-Peter, E., Muller, R., 1948. Formulas for bed-load transport. Proceedings of the 2nd IAHR Congress, Stockholm. Niroshinie, M.A.C., 2008. Modeling the tsunami wave propagation. CIB W89 International Conference on Building Education and Research BEAR. Nishihata, T., Tajima, Y., Moriya, Y., 2005. Study on the dynamic topography change due to tsunami attacks at Kirinda port—December 26, 2004 earthquake tsunami of Indian Ocean. Annual Journal of Coastal Engineering 52, 1386–1390 (In Japanese). Nishihata, T., Tajima, Y., Moriya, Y., Sekimoto, T., 2006. Topography change due to the Dec. 2004 Indian Ocean tsunami — field and numerical study at Kirinda Port, Sri Lanka. Proceedings 30th International Conference on Coastal Engineering, ASCE, pp. 1456–1468. Nishihata, T., Goto, K., Tajima, T., Takahashi, T., Imamura, F., 2009. A study for the sediment transport by the 2004 Indian Ocean tsunami along the natural coast of Hambantota, Sri-Lanka. Coastal Dynamics 2009. Pattiaratchi, C., 2005. Tsunami impacts on Sri Lanka—lessons for disaster reduction on coasts. International Symposium Disaster Reduction on Coasts. Monash University, Melbourne, Australia, pp. 1–2. Rabus, B., Eineder, M., Roth, A., Bamler, R., 2003. The shuttle radar topography mission—a new class of digital elevation models acquired by spaceborne radar. Photogrammetry and Remote Sensing 57, 241–262. Ranasinghe, D.P.L., in preparation. Evaluation of the bathymetric change due to tsunami, and its recovery process by normal wind waves in Kirinda Fishery Harbor Sri Lanka, Doctoral Thesis, Tohoku University, Japan, pp. 51–54. Shibayama, T., Okayasu, A., Wijayaratna, N., Sasaki, J., Suzuki, T., Jayaratne, R., 2005. The 2004 Sumatra earthquake tsunami, tsunami field survey in southern part of Sri Lanka. Annual Journal of Coastal Engineering, JSCE 52, 1401–1405 (in Japanese).

81

Srinivasalu, S., Thangadurai, N., Switzer, A.D., Mohan, V.R., Ayyamperumal, T., 2007. Erosion and sedimentation in Kalpakkam (N Tamil Nadu, India) from the 26th December 2004 tsunami. Marine Geology 240, 65–75. Takahashi, T., Imamura, F., Shuto, N., 1993. Numerical simulation of topography change due to tsunamis. Proceedings IUGG/IOC International Tsunami Symposium, Wakayama, pp. 243–255. Takahashi, T., Shuto, N., Imamura, F., Asai, D., 1999. A movable bed model for tsunami with exchange rate between bed load layer and suspended layer. Annual Journal of Coastal Engineering 46, 606–610 (in Japanese). 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 International Conference on Coastal Engineering, ASCE, pp. 1508–1519. Takahashi, J., Goto, K., Oie, T., Yanagisawa, H., Imamura, F., 2008. Inundation and topographic change due to the 2004 Indian Ocean tsunami at the Kirinda port, Sri Lanka. Annual Journal of Coastal Engineering, JSCE 55, 251–255 (in Japanese). Takahashi, J., Goto, K., Imamura, F., 2009. Numerical analysis for the sediment transport by the 2004 Indian Ocean tsunami at the Kirinda Fishery Harbor, Sri Lanka. 3rd International Conference on Estuaries & Coasts, Japan, pp. 1–9. Takahashi, T., Kurokawa, T., Fujita, M., Shimada, H., 2011. Hydraulic experiment on sediment transport due to tsunamis with various sand grain size. Journal JSCE, Series B2 67 (2), 231–235 (in Japanese). Wijetunge, J.J., 2009a. Field measurements and numerical simulations of the 2004 tsunami impact on the south coast of Sri Lanka. Ocean Engineering 36, 960–973. Wijetunge, J.J., 2009b. Field measurements and numerical simulations of the 2004 tsunami impact on the east coast of Sri Lanka. Pure and Applied Geophysics 166, 593–622.