Analysis of ocean wave characteristic in Western Indonesian Seas using wave spectrum model

Understanding the characteristics of the ocean wave in Indonesian Seas particularly in western Indonesian Seas is crucial to establish secured marine activities in addition to construct well-built marine infrastructures. Three-years-data (July 1996 1999) simulated from Simulating Waves Nearshore (SWAN) model were used to analyze the ocean wave characteristics and variabilities in eastern Indian Ocean, Java Sea, and South China Sea. The interannual or seasonal variability of the significant wave height is affected by the alteration of wind speed and direction. Interactions between Indian Ocean Dipole Mode (IODM), El Niño Southern Oscillation (ENSO) and monsoon result in interannual ocean wave variability in the study areas. Empirical Orthogonal Functions (EOF) analysis produces 6 modes represents 95% of total variance that influence the wave height variability in the entire model domain. Mode 1 was dominated by annual monsoon and has spatial dominant contribution in South China Sea effected by ENSO and Indian Ocean influenced by IODM. Java Sea was influenced by Mode 2 which is controlled by semi-annual monsoon and IODM. A positive (negative) IODM strengthens (weakens) the winds speed in Java Sea during the East (West) season and hence contributes to Mode 2 in increasing (decreasing) the significant wave in Java Sea.


Introduction
Various activities on seas, either for sea transportation activities, fisheries, marine resources exploration, and also development in marine sector are very sensitive to weather and sea condition changes.Two hundred sixty cases of ship accidents were due to natural factor increased with years [1].Among various natural factors, one that greatly influences marine activities is wave, and therefore, in the marine meteorological services, in addition to wind information, wave information is the most important part that should exist in every kind of marine weather information [1].The existence of extreme high tides could threaten the safety on sea and may result in great losses.
The wave conditions in the waters are strongly influenced by wind variability as the dominant force of ocean waves [2].Indonesia is an area affected by the monsoon wind cycle.The monsoon interacts with an interannual phenomenon such as El Niño-Southern Oscillation (ENSO) and Indian Ocean Dipole Mode (IODM); affects regional and local wind circulation patterns in Indonesian waters [3].Characteristics of waves and the influence of interannual phenomenon such as ENSO and IODM to wave variability in the waters can be determined from long-term wave measurements.Some sources of data buoy measurement, Voluntary Observing Ship (VOS), and altimeter satellite measurement can be utilized as a basis for determining wave climate.In addition, the wave hindcasting data using model is also can be used as an alternative data.
Characteristics and variability of waves in the waters of western Indonesia, including the South China Sea, Indian Ocean, and Java Sea are addressed in this study.The wind field in the mentioned three waters is strongly influenced by the relative position of the sun against the earth and it is influenced by the IODM system.This study addresses two main issues: (1) how the dominant variability pattern can influence the variability of the wave, (2) how it interacts with the interannual phenomenon of ENSO and IODM.This paper is organized as follows: the study area is introduced in Sect.2, and the details of data used and the methodology are presented in Sect.3. Section 4 presents the results and discussion of the study, and the conclusions are given in Sect. 5.

Study area
This study focusses on the territorial waters of western Indonesia between 20°N -23°S and 95°E -120°E.The boundary covers the eastern Indian Ocean, South China Sea, and Java Sea.Domains are further extended towards the Indian Ocean to accommodate the perfect wave formation state in the region.Figure 1

Model
Simulating Wave Nearshore (SWAN) is a thirdgeneration wave model for obtaining realistic estimates of wave parameters in coastal areas, lakes and estuaries from given wind, bottom and current conditions.However, SWAN can be used on any scale relevant for windgenerated surface gravity waves.The model is based on the wave action balance equation with sources and sinks [5,6].
In this study, hindcasting wave was performed by using model of SWAN with nonstationary mode where the input of wind field always change every step of time.The physical processes activated in this simulation are development of wind wave, quadruplet nonlinear interaction, dissipation due to white-capping, basic friction, and breaking waves.In this study, the initial spectrum used in the model is the initial spectrum of Joint North Sea Wave Project (JONSWAP).Meanwhile, the discrete wave propagation in the geographic space uses the BSBT (backward in space and backward in time) scheme.The formulation of the wind source/development of wind wave and white-capping function employed in the model is described in [7,8] and the basic friction function performed in the study followed the formulation in [8,9].
Here are the SWAN model configurations applied to this study: Resolution : 1/6° x 1/6° Frequency : Frequency with lowest frequency 0.04 Hz and highest 1 Hz Space-θ : 180 mesh in-space θ for a full circle 360° Domain : 20°N -23°S and 95°E -120°E with resolution 1/6° x 1/6° The model runs for 3 years (July 1, 1996, 00 00 UTC to June 30, 1999, 18 00 UTC), with a spin-up period (from calm state) for the first 7 days.Model output is the significant wave height and direction of propagation.

Emphirical Ortogonal Function
Empirical Orthogonal Function (EOFs) technique aims at finding a new set of variables that capture most of the observed variance from the data through a linear combination of the original variables.EOFs have been introduced in atmospheric science since the early 50's [10,11].EOF techniques are deeply rooted in statistics and EOFs was introduced as principal component analysis (PCA) [12] and it is analyzed details [13,14].
To obtain the main patterns of wave fields variability in this current study, EOFs was performed in determining the variation of significant wave height (Hs) values in the model domain either temporally or spatially.The time series of the EOFs principal components (PCs) are normalized to have unit variance, so that the corresponding EOFs represent the typical variability of the data in their original units.

Wind and Significant Waves Height (Hs)
The model results show a linier relationship between the higher of wind speed and the higher of Hs.The direction of wave propagation is strongly influenced by the wind direction.Waves generally move in the direction of the wind, except in areas with considerable refraction effects such as waters near the coast.The seafloor indicated by significant depth as found in the South China Sea (SCS) causes significant wave heights in the area to be relatively smaller as the wave energy is dissipated by increasingly larger friction and steep bathymetry changes can cause breaks.
The significant wind direction and wave height of the model in the west season (DJF) is shown in Figure 2

Results of EOF Analysis
The result of significant wave height simulation affirms significant variation of Hs to time at the three waters.
Figure 4 shows the mean significant wave heights in the model domains.The highest average Hs up to 1.75 m is in Indian Ocean, followed by South China Sea up to 1.3 m.The average Hs value in Java Sea is lower than 1 m, smaller than in South China Sea and Indian Ocean due to a narrower and closed water in Java Sea.Moreover, wind speed passing through Java Sea is not as big as in South China Sea and Indian Ocean.
The EOF analysis in this study uses the monthly average of Hs value and yields 6 modes representing 95% of the total variance as shown in

Results of EOF Analysis Test
The EOF analysis produces modes that affect significant wave height variability in the model area.The EOF analysis test has been done by comparing the number of multiplication products between the spatial pattern value and the temporal pattern with the input data (not shown).

Seasonal Wind and Significant Wave Height
Furthermore, based on the results of EOF analysis, in this study, domain model can be divided into 6 areas that are considered to have the characteristics of wave variability.The divided area is illustrated in Figure 11.The wind field and the average Hs for each month for 3 years at the six consecutive areas are shown in Figure 12 and Figure 13.The wind moves from the southeast in Area 1 throughout the year.The maximum wind occurs during east season (JJA) with a great average speed ranging from 6.46 m/s to 7.90 m/s.This corresponds to the maximum average value of simulated Hs results.The minimum average Hs in Area 1 occurs during west season with values between 1.17 m to 1.26 m and corresponds to the lowest wind velocity conditions in Area 1.There is a difference in the wind field pattern between Area 1 and 2. In Area 2, the wind moves from the southwest during west season with an average Hs value of 1.0 m.The wave height is significant up to 1.30 m to 1.90 m during west season, whereas during east season the significant wave height reaches 0.90 m to 1.0 m.In Area 6, the dominant wind moves from the northwest from May to September with a significant wave height of 0.56 m to 0.87 m.During west season, the wind moves from the northeast with a significant wave height of 0.34 m to 0.44 m.

Analysis of Annual Variations
Based on the results of EOF analysis, it is known that wave variability in domain model is dominantly influenced by Mode 1 and Mode 2. In the previous section also has been explained that a monsoon with a period of 12 months plays a role in Mode 1 and a semi-annual monsoon plays a role in Mode 2. However, there is a discrepancy in the temporal pattern at certain intervals and it is associated with the influence of the interannual phenomenon of IODM and ENSO.
Mode 1 gives the dominant influence in the Indian Ocean and South China Sea with different phase variability.Thus, the analysis was performed separately for both waters and selected a point that has the largest spatial pattern value in the Indian Ocean and South China Sea.In addition, the average monthly wind vector supports the analysis.From Figure 11, point A (South China Sea) and point B (Indian Ocean) are chosen to illustrate interaction between IODM and ENSO to Mode 1 and point E represents Java Sea is picked to capture the interaction of the interannual phenomena to Mode 2. During west season, the mean value of significant wave height at point A (h1,A) (South China Sea) at interval 2 is smaller compared at interval 1.This is due to the weakening of monsoon winds during the El-Niño incident.At interval 3, the value of h1,A is greater due to the strengthening of the wind during La Niña event.To conclude, ENSO phenomenon contributes in Mode 1 at point A (South China Sea).As mentioned above, point E is selected to represent Java Sea where contribution of Mode 2 is dominant in affecting wave variability rather than Mode 1.In July to September (JAS) at interval 2, winds in the Java Sea move westward and it is reinforced by a positive IODM which drive stronger winds westward around Java Sea.Meanwhile, in December to February (DJF) at interval 2, the wave at point E becomes smaller than at interval 1.This is associated with a movement of winds eastward in Java Sea which is reduced by the positive IODM.

Conclusion
Empirical Orthogonal Functions (EOF) analysis produces 6 modes represents 95% of total variance that influence the wave height variability in the entire model domain.Annual monsoon plays a role in Mode and has spatial dominant contribution in South China Sea and Indian Ocean.Java Sea was influenced by Mode 2 which is controlled by semi-annual monsoon.ENSO effect was identified in South China Sea and IODM effect was captured at Indian Ocean in Mode 1 and at Java Sea in Mode 2. A positive (negative) IODM strengthens (weakens) the winds speed in Java Sea during the East (West) season and hence contributes to Mode 2 in increasing (decreasing) the significant wave in Java Sea

Fig. 1 .
Fig. 1.Domain model and bathymetry profiles . The significant monthly wind and wave height is the result of MATEC Web of Conferences 147, 05001 (2018) https://doi.org/10.1051/matecconf/201814705001SIBE 2017 the reduction of the wind and Hs for each month for 3 years.The x and y direction components of the wind field and Hs are averaged separately to determine the mean direction of wind movement and significant wave propagation.In December, maximum winds are located at South China Seas (latitude 10°N -20°N) and in January and February months winds in the South China Seas weaken.The direction of wave propagation is also in accordance with the direction of the wind.In narrow and closed waters, Hs tends to be relatively lower than Hs in wide and open waters, as large and open waters support perfect waveforms and the inclusion of wave energy from other places through wave propagation phenomena.

Fig. 2 .
Fig. 2. Averaged of (a) the wind field (m/s) and (b) the significant wave height (m) in December-January-February (DJF)

Figure 10 Fig. 10 .
Fig. 10.Five test points of the EOF analysis results

Figure 14 -
Figure 14 -16 expose the plot of Ocean Nino Index (ONI), Dipole Mode Index (DMI), spatial and temporal pattern and the averaged monthly wind field at point A, B, and C, respectively.Interval 1 indicates a normal IODM and ENSO states, index 2 (interval 2) reveals the time interval of positive El-Niño and IODM events, and index 3 (interval 3) affirms the time interval of La-Niña events.

MATEC
Fig. 14.ONI and DMI (top panel), spatial and temporal pattern (middle panel) and the average monthly wind field (bottom panel) at point A (South China Sea)