0 оценок0% нашли этот документ полезным (0 голосов)

12 просмотров8 страницNearshore wave climate is important for the assessment of coastal process and morphological changes. For port engineers, nearshore wave climate is crucial for the determination of operational conditions and downtime analysis. The nearshore wave climate is often derived by translation of the 10 to 30-year offshore time series to nearshore. Direct transformation of such large amount of observations is not feasible, since it requires an extremely long computational time. This paper describes an adv

Oct 28, 2019

© © All Rights Reserved

PDF, TXT или читайте онлайн в Scribd

Nearshore wave climate is important for the assessment of coastal process and morphological changes. For port engineers, nearshore wave climate is crucial for the determination of operational conditions and downtime analysis. The nearshore wave climate is often derived by translation of the 10 to 30-year offshore time series to nearshore. Direct transformation of such large amount of observations is not feasible, since it requires an extremely long computational time. This paper describes an adv

© All Rights Reserved

0 оценок0% нашли этот документ полезным (0 голосов)

12 просмотров8 страницNearshore wave climate is important for the assessment of coastal process and morphological changes. For port engineers, nearshore wave climate is crucial for the determination of operational conditions and downtime analysis. The nearshore wave climate is often derived by translation of the 10 to 30-year offshore time series to nearshore. Direct transformation of such large amount of observations is not feasible, since it requires an extremely long computational time. This paper describes an adv

© All Rights Reserved

Вы находитесь на странице: 1из 8

Matrix from Offshore Wave Data

Nguyen Thi Hai Ly†* and Nguyen Thanh Hoan††

† ††Haskoning Vietnam

National University of Civil Engineering Ltd., a company of Royal Haskoning

www.cerf-jcr.org

Dong Tam Ward, Hai Ba Trung District Lang Thuong Ward, Dong Da District

Hanoi, Vietnam Hanoi,Vietnam

ABSTRACT

Ly, N.T.H. and Hoan N.T., 2018. Determination of nearshore wave climate using a transformation matrix from

offshore wave data. In: Almar, R.; Almeida, L.P.; Viet, N.T., and Sall, M. (eds.), Tropical Coastal and Estuarine

Dynamics. Journal of Coastal Research, Special Issue No. 81, pp. 14–21. Coconut Creek (Florida), ISSN 0749-0208.

Nearshore wave climate is important for the assessment of coastal process and morphological changes. For port

www.JCRonline.org engineers, nearshore wave climate is crucial for the determination of operational conditions and downtime analysis.

The nearshore wave climate is often derived by translation of the 10 to 30-year offshore time series to nearshore.

Direct transformation of such large amount of observations is not feasible, since it requires an extremely long

computational time. This paper describes an advanced technique to quickly derive nearshore wave conditions from

offshore wave data. First, determination of matrix of boundary conditions in which matrix nodes are chosen to cover

wave conditions at offshore boundary. Then, simulations of wave for each condition as described in the selected

matrix of offshore conditions. The results are used to determine the matrix of wave conditions for any given

nearshore location. This matrix consists of factors that specify the relation between the offshore (model forcing) and

nearshore (model results) wave parameters. With this matrix, offshore wave observations can be translated to any

nearshore location. With this technique, a 30-year long time series of offshore wave measurement can be rapidly

(less than 2 days) translated to nearshore wave conditions. This paper discusses in detail the working principles of the

technique, as well important aspects such as the optimization of boundary conditions matrix.

ADDITIONAL INDEX WORDS: Wave study, wave transformation, wave climate, transformation matrix.

Nearshore wave climate is often characterized based on a time With this method, the offshore time series is analyzed to find

series of wave parameters for a sufficient long period of time which combinations of wave height, peak period, mean

(e.g., 10 to 30 years). Typically, the time series of nearshore direction, wind speed and wind direction exist for the considered

wave data are derived by transforming offshore hindcast wave data set. This is done by allocating to each parameter a set of

data to the nearshore. Spectral wave models such as SWAN or “bins” that cover the data range. The offshore time series is

MIKE 21 SW are often used to compute wave transformation. analyzed to find which of these possible combinations of

Running a 30-year, 3-hourly dataset on a wave model means parameters occur, and using the bin mid-point as the

processing over 8x365x30 = 87,600 offshore boundary cases of representative value. For example, when analyzing a 30-year

significant wave height, peak wave period, wave direction, wind time series of the NOAA hindcast data at 130 N, 1100 E (off the

speed and wind direction. Direct transformation of such large coast of Bai Goc, Phu Yen Province, Vietnam, from 1/1/1979 to

amount of offshore wave conditions to the nearshore is not 1/1/2009), there are 190 events where the conditions fall within

feasible as the required simulation time is extremely long. If a the bins presented in Table 1. Instead of modelling the 190

standard 3-hour simulation takes 5 minutes (real world time), it individual wave and wind conditions, only one set is modelled,

would require 10 months for one desktop workstation to presented by the mid-points of the bins. The set of unique wave

transform 30-year offshore wave conditions to the nearshore. and wind events derived from the offshore time series is called

Indirect transformation can therefore be used to quickly translate the look-up table or matrix.

the long-term offshore time series to nearshore. Typically, a

reduction of factor 100 can be obtained without losing the Table 1. Example of bins used in the look-up table method.

accuracy (i.e. it would require 3 to 4 days to transform the whole

30-year time series from offshore to nearshore). Two methods Hm0 (m) Tp (s) Dir (° N) Uw (m/s) Wdir (° N)

that can be used to indirectly transform the offshore time-series 1 – 1.5 6-8 45 - 75 4-6 45 - 75

data set to nearshore are discussed in the following section.

____________________ Once the look-up table has been determined, all considered

DOI: 10.2112/SI81-003.1 received 31 September 2017; accepted in conditions (wave and wind parameters) are transformed to

revision 21 November 2017. nearshore by means of wave modelling (e.g., using SWAN

*Corresponding author: haily@dongchay.com model). This results in a nearshore look-up table for any

©

Coastal Education and Research Foundation, Inc. 2018

Transformation Matrix to Derive Nearshore Wave Climate 15

_________________________________________________________________________________________________

nearshore location. For example, when the set of conditions method will result in Hm0 of 0.12 m and 0.40 m at point A,

presented in Table 1 is modelled, the resulting nearshore wave respectively. This is certainly not realistic as the offshore wave

parameters (e.g., wave height, wave period, wave direction) are heights are almost the same but the nearshore wave heights are

assigned to all original offshore conditions that fall within the too different. Using the matrix interpolation method, the

bins listed (i.e. 190 events for this example). The method has resulting wave height Hm0 at point A would be 0.254 m and

been used by Fockert and Luijendijk (2011), Gunaratna (2012), 0.266 m, respectively.

Ruggiero et al. (2006), and was extensively validated by In the look-up table method, the coarser the bin size, larger

Leonard-Williams and Francis (2013). errors are expected. According to Leonard-Williams and Francis

(2013), the method works well for wave height which remains

Matrix Interpolation Method accurate even when coarse discretization parameters are used.

With this method, one needs to determine a matrix of However, wave period values are not generally well reproduced.

boundary conditions in which matrix nodes are chosen to cover By nature, interpolating result is less dependent on the

wave and wind conditions at offshore boundary. Simulations for coarseness of the matrix nodes (than in the look-up table). The

each case are performed and the results are used to determine robustness of the matrix interpolation method will be further

the matrix of wave conditions for any given point in the verified in this study.

nearshore. This matrix consists of factors that specify the The above discussion suggests that the matrix interpolation

relation between the offshore (model forcing) and nearshore method is considered more robust and less dependent on the

(model results) wave parameters. With this matrix, a wave choice of the matrix nodes (than the look-up table method) and

condition offshore can be translated to any nearshore location by therefore was chosen for further analysis in the present work.

means of interpolation. The method has been used in Hoan

(2016, 2017), Hoan and Lansen (2016), and Ly (2015). METHODS

Both methods limit the number of boundary conditions In this paper, the working principles of the technique of the

applied to a stationary spectral wave model. However, the look- matrix interpolation method are discussed in detail. In order to

up table method doesn’t preserve the uniqueness of the offshore validate the technique, nearshore time series of wave parameters

conditions (all conditions fall within the considered bins are determined by the direct and matrix (indirect) transformations

assigned the bin mid-point values). On the contrary, the matrix are compared. The wave model settings are identical for both

interpolation method keeps the offshore data unchanged. The direct and indirect transformations. SWAN model was used in

nearshore wave parameters are interpolated using the predefined stationary mode. The schematic of the two methods are shown

offshore matrix nodes and the derived nearshore matrix nodes in Figure 2.

(for a given point). Therefore, the resulting wave is better

correlated to the offshore conditions as it should be. Figure 1 is

used to further explain the differences of the two methods.

Assume that two bins for the offshore wave height are

considered, [0.0 m to 0.5 m] and [0.5 m to 1.0 m]. The bin mid-

points are 0.25 m and 0.75 m, respectively. This two wave

conditions are transformed to nearshore, resulting in a wave

height of 0.12 m and 0.40 m at a given point A. Find Hm0 at

point A if the offshore wave height Hm0 is of 0.40 m (assume

that other parameters such as wave periods and directions are

unchanged). As Hm0 = 0.40 m falls in the first bin [0.0 m-0.5

m], the representative value of this bin (0.25 m) is used,

resulting in Hm0 = 0.12 m at point A. If the matrix interpolation

method is used, the wave height at point A would be 0.204 m (as

a result of interpolation).

transformations.

transformation, referred here as “baseline” conditions, were used

Figure 1. Illustration of the look-up table method. to validate the results obtained by matrix transformation method.

Below the followed steps are described:

To get a better understanding of the differences between the • Transformation of the full offshore time series to

two methods, let’s consider another situation. If the offshore nearshore using SWAN model to obtain the “baseline”

wave height Hm0 = 0.49 m and 0.51 m, using the look-up table nearshore conditions.

16 Ly and Hoan

_________________________________________________________________________________________________

• Analysis of the offshore wave data to determine a matrix coeﬃcient of the rate of dissipation), γ=0.73 (the breaker index);

of boundary conditions in which matrix nodes are chosen bottom friction JONSWAP CFJON = 0.038 m2s−3; white

to cover wave conditions at offshore boundary. capping activated; triads activated and quadruplets deactivated

• SWAN simulations of the nearshore wave conditions (no wind forcing). A regular computational grid with a mesh

based on the matrix of offshore wave conditions. size of 50 m (for both x- and y-coordinate) was used. Apart from

• Creation of the transformation matrices from the matrix the spatial grids, the user must specify the spectral grid.

of wave transformation calculations for various In this study, the directional resolution was set to 100. The

nearshore locations. lowest and highest frequencies of 0.03 Hz and 1 Hz were used

• Translation of the time series of offshore waves to respectively. The number of frequencies was determined

nearshore using the transformation matrices. automatically by SWAN using an increment factor of 1.1

• Comparison of “baseline” and “matrix derived” between two consecutive frequencies.

nearshore wave data. Initial test runs indicated that simulation time for a typical

wave transformation takes about 13 minutes. This duration

For this study, the 31-year time series of the NOAA hindcast makes the use of the 2D model impracticable in the present

data at 13° N, 110° E (from 1/1/1979 to 1/1/2010) was chosen as study. Given that the coastline is straight and the cross-shore

the offshore wave data, due to the large range of wave height profile is uniform, the 1D model is expected to provide similar

(up to 9.5 m), wave period (up to 20 s) and wave direction. The results. In addition to this, the less demanding computational

wave roses are shown in Figure 3. effort required by the 1D model allow to perform a large number

of simulations.

A 1D SWAN model was setup using the same model settings

as for the 2D model. To compare 1D and 2D model results, a

number of offshore wave conditions are simulated:

• Run 1: Hm0 = 2.0 m, Tp = 8 s; MWD = 45° N;

• Run 2: Hm0 = 2.0 m, Tp = 8 s; MWD = 60° N;

• Run 3: Hm0 = 2.0 m, Tp = 8 s; MWD = 900 N.

Figure 5 shows the development of significant wave height

(Hm0), mean wave period (Tm01) and mean wave direction

(MWD) along the middle cross-section (located at Y = 50 km in

Figure 3. Wave roses at 13° N, 110° E (NOAA hindcast). Figure 4) of the 2D model and those obtained from the 1D

model. The horizontal axis in Figure 5 shows the distance from

the shoreline (where X = 0) towards the sea (pointing East).

It can be observed that the results are almost identical for the

1D and 2D models regardless of incoming wave directions.

Therefore, the 1D model can be used for this study without

affecting the model results and findings. However, it should be

noted that in practice where the coastline is not straight and the

cross-shore profiles are not uniform, the use of 2D model is a

mandatory. To transform the 31-year offshore time series to

nearshore, 90520 simulations were performed using the 1D

model.

SWAN modelling (in stationary mode). Initially a 2D model was

setup for this purpose. A straight coastline with uniform cross-

shore profile is assumed. A cross-shore profile at Bai Goc is

used as a representative profile. The model domain and

bathymetry are presented in Figure 4. The offshore waves are

imposed at the eastern boundary where the water depth is about

60 m below mean sea level (MSL). For simplicity, wind forcing

was excluded, which is the case for simulation of swell waves.

A uniform water level of 0.0 m was used for all simulations.

The SWAN settings that were used for this study are as Figure 5. Comparison of 1D (red) and 2D (blue) model results

below: third-generation mode; breaking α=1.0 (proportionality for run 1 (top), 2 (middle) and 3 (bottom).

Transformation Matrix to Derive Nearshore Wave Climate 17

_________________________________________________________________________________________________

20

15

n (%)

10

0

0 1 2 3 4 5 6

Significant wave height Hs (m)

nodes of significant wave height (Hm0), peak wave period (Tp),

and wave direction (MWD). The histogram of the significant

wave height and peak wave period is presented in Figure 6, in

which n is the percentage of the total number of occurrences in

the interval 0.25 m for wave height.

The occurrence of wave parameters is important for

optimizing the matrix nodes. In this case, a higher resolution of

the matrix is recommended for wave height ranging from 0.5 m

to 2.0 m. In this study, in order to investigate the impact of the

matrix node resolution on the model results, different

combinations of matrix nodes were considered. In practice,

water level could play an important role on nearshore wave

conditions. In such cases, water level variation must be

considered in the matrix nodes. In this study, a fixed water level

of 0.0 m MSL was used for all simulations to quantify the Figure 7. Wave roses at point B (top), C (middle) and E

performance of the matrix transformation method, quality (bottom).

indices were used to compare the nearshore “baseline” and

“matrix derived” time series:

• RMS error: root mean square error between the

nearshore “baseline” time series and the “matrix

derived” time series;

• Bias error: between the nearshore “baseline” time series

and the “matrix derived” time series and;

• Correlation coefficient (ρ): is the correlation coefficient

between two stochastic variables. The correlation

coefficient reflects the degree to which the variation of

the first is reflected in the variation of the other variable.

The “baseline” results are extracted at several output locations

with water depth varying from 2.0 m (A), 3.0 m (B), 4.5 m (C),

6.7 m (D), 8.5 m (E) and 11.0 m (F). The nearshore “baseline”

wave conditions are presented in Figure 7 (wave roses for point Figure 8. Probability of exceedance of Hm0.

B, C and D) and Figure 8 (probability of exceedance of Hm0).

A range of wave conditions, described by a matrix in terms of

For all combinations within the matrix (20 wave height x 19

wave height (Hm0), wave period (Tp), peak wave direction

wave period x 47 wave direction = 17860 combinations),

(Dir) was determined to represent the offshore wave climate.

SWAN calculations were carried out. For each nearshore

Computations were made for all combinations of the following

location, the results of all these simulations were combined in a

parameter values:

transformation matrix which was used for the transformation of

• Hm0 = from 0.1 to 9.5 with steps of 0.5 (in m); the offshore time series to nearshore based on multi-linear

• Tp = from 2 to 20 with steps of 1 (in seconds); interpolation between the values at the nodes of the matrix. The

• Dir = from 0 to 230, with steps of 5 (in relative to the results are referred to as “matrix derived” nearshore time series.

North). Figure 9 shows a comparison of nearshore “baseline” and

18 Ly and Hoan

_________________________________________________________________________________________________

B for the year 2009. The quality indices (i.e. Bias, RMS, ρ) were

computed using 31 years of records, from 1/1/1997 to 1/1/2010.

Figure 10 shows a comparison of wave roses based on

“baseline” and “matrix derived” wave data at location C

(location deeper than point B). Offshore wave events with wave

direction larger than 230° N are excluded from the analysis.

The comparison shows that the “baseline” and “matrix

derived” nearshore waves are almost identical. This behavior is

found for all considered points with the water depth ranging

from 2.0 m to 11.0 m (see Table 2). This result suggests that

even for shallow water area, where wave breaking occurs, the Figure 10. Wave roses of nearshore “baseline” (left) and “matrix

matrix interpolation method is still valid. derived” (right) wave height versus wave direction at point C.

The good match between “baseline” and “matrix derived”

nearshore waves confirms the validity of the matrix Robustness of Matrix Interpolation Method

transformation method. It should be noted that the applied In the previous analysis, the matrix nodes of wave direction

matrix nodes have high resolution (i.e. 0.5 m for the wave were chosen starting from 0° to 230° with grid resolution of 5°.

height, 1 s for wave period and 50 for wave period), as a result of Different grid resolutions ranging from 10° to 60° were used to

a large number of model simulation. In case of 2D modelling, form the matrix. Other parameters such as wave height (Hm0)

such a large simulation number is not practical. and wave period (Tp) are kept unchanged. Wave direction grids

In practice, it is important to optimize the matrix nodes such are chosen as follow: Dir = [0:dx:230], where dx is the grid

that the accuracy remains high while simulation time is still resolution. For instance, if the grid resolution dx = 30° is chosen,

acceptable. As a rule of thumb, the matrix nodes should be wave directions considered in the matrix nodes would be 0, 30,

chosen finer where more wave events occur and coarser where 60, 90, 120, 150, 180, 210 degrees. Based on each set of matrix

there are less wave events. To investigate the impact of matrix nodes, different nearshore time series were determined for

node choice on the nearshore wave transformation, different location A to F. This time series were compared against the

combinations of matrix nodes were considered. This new set of “baseline” nearshore data to obtain the Bias, RMSE, and ρ. The

analysis is discussed in the next section. analyses were performed for the whole 31-year dataset. The

results of these analyses are presented in Figure 11, showing the

sensitivity of the RMSE on the choice of grid resolution of wave

direction. It can be seen that the nearshore “matrix derived”

wave conditions are quite robust regardless of the choice of grid

resolution of wave direction. Nearshore wave directions are

more sensitive to the choice of grid resolution of wave direction,

but the RMSE range is quite acceptable for wave direction.

In general, the higher grid resolution of the matrix nodes, the

better the nearshore wave results would be (i.e. smaller the

RMSE). However, Figure 11 also shows that the nearshore wave

heights are better predicted for grid resolution of 40° than for

350. This can be explained by the fact that the interpolation

results are also dependent on node values. The node values in

these cases would be:

• dx = 35°: Dir = 0, 35, 70, 105, 140, 175, 210 and 230;

• dx = 40°: Dir = 0, 40, 80, 120, 160, 200 and 230.

As most of offshore wave directions are closer to 40° N than

35° N, the interpolating results for the grid resolution of 40° are

slightly better.

Similar analyses were performed to investigate the impact of

grid resolution of significant wave height (Figure 12) and peak

wave period (Figure 13). The results show that in general the

“matrix derived” nearshore wave periods are unresponsive to the

choice of grid resolution of wave height and wave period.

Nearshore wave directions are more sensitive to grid choices of

both wave height and wave period. But keep in mind that for the

Figure 9. Comparison of nearshore “baseline” and “matrix wave direction, a RMSE within a few degrees is generally

derived” wave parameters at point B. acceptable.

Transformation Matrix to Derive Nearshore Wave Climate 19

_________________________________________________________________________________________________

Table 2. Quality indices quantifying the comparison of 31-year simulation time is still acceptable. The matrix nodes should be

“baseline” and “matrix derived” wave data at various points. defined with high resolution where more wave events occur and

coarser where wave events are less frequent. The results

A B C D E F presented in this study can be used as a rule of thumb for the

selection of matrix nodes.

Bias 0.01 0.00 0.00 -0.01 0.00 0.01

Sign. wave

height (m)

Peak wave period

(s)

direction (°N)

Peak wave

derived” nearshore wave results.

“matrix derived” nearshore wave results.

(> 6.5 m deep) are found unresponsive to the grid resolution of

wave height (RMSE of Hm0 < 0.15 m). However, for shallower

locations (water depth smaller than 4.5 m), the RMSEs of Hm0

can be as larger as 0.3 m. The test results suggest that the matrix Figure 13. Impact of grid resolution of Tp on the “matrix

nodes should be optimized such that the accuracy of nearshore derived” nearshore wave results.

“matrix derived” wave parameters remains high while

20 Ly and Hoan

_________________________________________________________________________________________________

Application of Matrix Interpolation Method approximation is considered appropriate for this study area as

In practice, the local bathymetry is often so complex that the offshore wind and wave are strongly correlated. A fixed water

use of 2D wave model is mandatory. In addition, other level of 1.5 m above CD was used for all simulations. Total

parameters such as wind speed, wind direction and water level number of simulations was 1890.

should be taken into account. In such cases, the number of It should be noted that all combinations between those

simulations defined by the matrix can become an issue. For parameters in Table 3 are used, although some of them will be

instance, if 3 water levels, 7 wave heights, 6 wave periods, 9 unlikely. In practice, a wave height and wave period scatter plot

wave direction, 5 wind speeds and 12 wind directions are can be used to eliminate those conditions that are not used in the

considered in a matrix, the total number of simulations would be interpolation (because no wave events exist close to that nodes).

68040. Engineering judgement should be made to minimize the In case of sea wave, wave steepness should be used instead of

number of simulations while keeping the accuracy of nearshore wave period (in such case, there is no combination such as wave

wave results acceptable. For instance, if the impact of water height of 8 m and wave period of 2 s).

level variation is expected insignificant (e.g., the output points

of interest are at relatively deep water), a constant water level Table 3. Range of offshore boundary conditions.

can be used without affecting the model results. Similarly, if the

wind and the wave are strongly correlated (i.e. sea waves), the Z (mCD) Hm0 (m) Tp (s) Dir (° N) Uw (m/s)

input wind speed can be chosen in associated with the wave +1.5 0.2 2 20 0

height and the wind direction is assigned with the wave +1.5 1.0 5 35 5

direction. This would reduce the number of simulations +1.5 2.0 8 50 10

significantly. +1.5 3.5 10 65 20

+1.5 5.0 18 85 30

In the following section an example is shown for the Port of +1.5 8.0 26 130 40

Dung Quat, Quang Ngai province, Vietnam. The objective is to

illustrate the use of the matrix interpolation method in practice

and to confirm the validity of the method in 2D model with

complex bathymetry, geometry and model inputs. Figure 14

shows the port of Dung Quat and output location A (-40 m Chart

Datum - CD) and B (-14 m CD) which are used for comparison

of “baseline” and “matrix derived” wave parameters.

To derive the wave conditions at the nearshore area, a

curvilinear SWAN model was used. The grids were developed

within a local coordinate UTM 49N system.

heights at location A (top) and B (bottom).

and 80 m at the site. The domain and grid resolution of the

model are considered appropriate to properly capture the

characteristics of the study area. The bottom depths at

computational grid points for the computational grid were

interpolated from the survey and C-map data.

The offshore hindcast data were obtained from the NOAA

global wave model for point 15.5° N, 109.5° E. The boundary

conditions that define the matrices are described in terms of

wave height (Hm0), direction (Dir), peak period (Tp) and wind

speed (Uw) as presented in Table 3. The wind direction is Figure 16. Comparison of “baseline” and “matrix derived” wave

assumed equal to the wave direction offshore. This period at location A (top) and B (bottom).

Transformation Matrix to Derive Nearshore Wave Climate 21

_________________________________________________________________________________________________

The SWAN model settings are similar to the above tests water level) which 2D modelling is required. This is

except that quadruplets are activated as wind forcing is included. confirmed by the application at Port of Dung Quat

Besides the “matrix derived” nearshore wave parameters at example.

location A and B, “baseline” nearshore wave parameters were • Typically, a reduction of factor 100 can be obtained for

obtained by performing direct transformation of the offshore the total simulation time (e.g., 10 months to 4 days)

wave for the period from 6th to 31st October 2015. without losing the accuracy (as has been proved in Table

Comparisons of “baseline” and “matrix derived” wave 2). This makes it possible to utilize high grid resolution

parameters are presented in Figures 15, 16, and 17. In general, which otherwise cannot be used if direct transformation

the RMSE are small, indicating good agreements between is used. As a result, model accuracy for shallow area or

“baseline” and “matrix derived” values. complex bathymetry can be increased.

The results show that nearshore wave conditions are similar to

Water level variations are important for shallow water area. In

the offshore waves. This was explained by the fact that the

practice, water level variations should be considered and can be

considered points are at relatively deep water and that wind

included in the transformation matrices as an independent

forcing helped maintaining the wave energy from offshore to

variable. It is important to note that for any wave transformation

nearshore.

study, the model results should be validated against

measurements regardless of the method (direct or indirect) used.

ACKNOWLEDGMENTS

We would like to thank the National University of Civil

Engineering (Vietnam) for the financial support (under the

framework of project 156-2017/KHXD-TĐ). Many thanks go to

Royal Haskoning DHV for providing facilities and software

used to conduct this study.

LITERATURE CITED

Fockert, A.D. and Luijendijk, A., 2011. Wave Look-up Table

Building with Nature. Delft, The Netherlands: Deltares,

Technical Note, Ref: 1002337-002-ZKS-0001, 50p.

Gunaratna, P., 2012. A wave transformation matrix approach for

establishing nearshore wave climate in the Dubai coastal

Figure 17. Comparison of “baseline” and “matrix derived” peak zone. Proceedings of the 8th International Conference on

wave direction at location A (top) and B (bottom). Coastal and Port Engineering in Developing Countries

(Chennai, India), 11p.

Hoan, N.T., 2016. Kuala Tanjung Wave Modelling Study,

CONCLUSIONS

Indonesia. Hano, Vietnam:. Haskoning DHV Vietnam Co.

This paper discusses possible techniques that can be used to

Ltd., Report VN1038R01D02, 34p.

quickly obtain nearshore wave conditions. The two methods that

Hoan, N.T., 2017. Karimun Ferry Terminal Development -

can be used are (i) the look-up table method and (ii) the matrix

Wave Modelling Study, Indonesia. Hanoi, Vietnam:

interpolation method. The matrix interpolation method is

Haskoning DHV Vietnam Co. Ltd., Report VN1419R02D01,

recommended due to its robustness and its higher accuracy (than

39p.

the look-up table method) in determining nearshore wave

Hoan, N.T. and Lansen, J., 2016. Dung Quat Refinery

conditions. Following can be concluded from this study:

Expansion Project - FEED Services for Marine Facilities -

• The study has shown that the matrix interpolation Wave modelling Study, Vietnam. Hanoi, Vietnam:. Haskoning

method works very well. The quality indices presented in DHV Vietnam Co. Ltd., Report VN1083REP004D02, 33p.

Table 2 clearly show a perfect match between the Leonard-Williams, A. and Francis, A., 2013. Development of a

“baseline” and “matrix derived” wave parameters Methodology Utilising Look-up Tables to Accelerate Spectral

regardless of the water depths. Even for a very shallow Wave Model Run Times. Met Office - Weather Science,

location (e.g., point A with a water depth of 2 m), the Technical Report No. 571, 31p.

method still works well. Ly, N.T.H., 2015. Feasibility Study of Tra Ly Navigation

• The study also confirms that the matrix interpolation Channel - Nearshore Wave Modelling Study. Hanoi, Vietnam:

method is robust and less sensitive to the choice of Consultancy Company Limited of University of Civil

matrix nodes. To optimize the number of simulations, the Engineering, Report TLY/FS/Vol.09/BD/Modelling, 88p.

matrix nodes should be chosen finer where more wave Ruggiero, P.; List, J.; Hanes, D., and Eshleman, J., 2006.

events occur and coarser where there are less wave Probabilistic shoreline change modelling. Proceedings of the

events. 30th International Conference on Coastal Engineering (San

• The matrix interpolation method works well for Diego, USA), pp. 1-14.

situations with complex bathymetry, geometry and