AN APPROACH TOWARDS WAVE CLIMATE . . .

JOURNAL of MARINE ENGINEERING

AN APPROACH TOWARDS WAVE CLIMATE STUDY IN

THE PERSIAN GULF AND THE GULF OF OMAN:
SIMULATION AND VALIDATION
Soheila Taebi 1 , Aliasghar Golshani 2 , Vahid Chegini 3
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

1- Research Fellow, Iranian National Center for Oceanography (INCO)

2- Research Assistant Professor, Iranian National Center for Oceanography (INCO)
3- Research Assistant Professor, Iranian National Center for Oceanography (INCO)

Abstract
This article describes the 11-year wave simulation (1992-2002) in the Persian Gulf and the
Gulf of Oman using the input data derived from European Center for Medium-Range Weather
Forecasts (ECMWF). The ECMWF 10 meter wind field and spectral wave boundary condition
at 18ْ N degree are input into one of the latest versions of numerical wave models (3rd
generation) after a few local modifications. Tropical cyclones during the last 30 years in the
northern Indian Ocean which affect the Gulf of Oman are regenerated and wave simulation for
individual cyclones is carried out. Open boundary of continuous hindcast is also modified in
cyclone periods. In-situ and satellite wind and wave data sets are used to evaluate the accuracy
of input wind and simulated wave fields. Extreme Value Analysis (EVA) is the next taken stage
in which the wave characteristics were calculated for different return periods. Similar analysis is
performed on the directional data to find out significance of storms in each direction. Finally, a
user-friendly engineering and management tool is developed and verified.
Keywords:Wave Modeling, Tropical Cyclones, Directional Extreme Value Analysis, Persian
Gulf, Gulf of Oman, Iranian Wave Atlas

1-Introduction
Wave, the most significant maritime global and regional scales to provide
phenomenon, due to its complicated and reasonable reconstructions of the past
stochastic behavior is known as one of waves condition. Since mid-1980s,
the most difficult phenomena in several major Numerical Weather
engineering studies. The effect of waves Prediction (NWP) centers (ECMWF,
on coastal and marine activities urges us U.S. Navy Fleet Numerical Meteorology
to identify the wave characteristics using and Oceanography Center (FNMOC),
field measurements, theoretical studies, and U.S. National Centers for
physical modeling, and numerical Environmental Prediction (NCEP)) have
simulations. Coastal and harbor operated global spectral ocean wave
engineers generally use these methods to models in real time and have
identify wave climate and the highest accumulated the analysis products to
probable wave characteristics as well as form preliminary estimates of the global
annual attributes of waves. For this wave climate. Navy Operational Global
reason, countries that take the advantage Atmospheric Prediction System
of contiguity to seas or great lakes have (NOGAPS) and the Navy Operational
developed a regular plan for studying Regional Atmospheric Prediction
wave and other marine phenomena. System (NORAPS) at FNMOC, generate
Thus, a couple of attempts have been a twice-daily suite of atmospheric
undertaken within the last few years on analyses and forecasts (Bayler and

Vol. 4/ No. 7/ Spring 2008 1

11/E
 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

Lewit, 1992). After the success of ERA- 226000 km2. Its average depth is 35 m
15 (Sterl et al., 1998), ECMWF and it attains its maximum depth of 100
performed their second reanalysis, ERA- m near its entrance- the Strait of
40 (Caires and Sterl, 2003). In parallel Hormuz. It is virtually surrounded by
with them, the American National arid land and connected to the Gulf of
Center for Atmospheric Research and Oman and Indian Ocean only by 60 km
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

the National Centers for Environmental wide Strait of Hormuz. Qishm Island
Prediction (NCEP/NCAR) produced a borders the north side of the strait. The
global reanalysis of the surface wind Gulf of Oman is northwestern arm of the
from 1958-1997 which was used to force Arabian Sea, between the eastern part of
the spectral ocean wave model (Cox and the Arabian Peninsula and Iran. It is
Swail, 2001). Although they provide about 370 km wide and 545 km long.
valuable sources for global Strong winds characterize most desert
oceanographic studies, their low areas including those bordering the
resolution for coastal engineering and Persian Gulf. The Shamal blows mainly
environmental purposes requires national from the North West in the Northern
and regional centers to prepare high- parts of the Persian Gulf, but tends to
resolution datasets with special attention veer to North as one approaches to the
to the areas of their interests. South East. Because wave and surface
In this regard, a forty-year wave hindcast currents are related to "Shamal" wind,
study for the Arabian Sea, emphasizing they are directed mainly towards the
on the wave climate of Al-Ashkharah in South East (Purser and Seibold, 1973).
Oman, was performed by Baird & In contrast, Gulf of Oman is mainly
Associates (Dibajnia, 2002). They used a affected by seasonal winds called
second generation (2G) spectral wave monsoons.
model based on the WAVAD model The first attempts for studying the wave
(Resio, 1981) to simulate wave climate characteristics for Iranian coastlines
of the Arabian Sea. They also addressed began by deploying some wave
tropical cyclones separately. Ocean measurement stations in deep waters of
weather and DHI Water & Environment Iranian seas. Together with this basic
jointly did a comprehensive metocean data, other data sources such as different
study of the Persian Gulf called short period measurements of wave
PERGOS. It was based on hindcast of a parameters, satellite data, and results of
continuous 20-year period (1983-2002) large scale numerical simulations enable
and 108 storms of various dates between us to pursue a hindcast project in Iranian
1961 and 2002. Seas including the Caspian Sea in the
The oceanic region compromised of the North, and the Persian Gulf and the Gulf
Persian Gulf, Strait of hormuz and the of Oman in the South titled Iranian Seas
Gulf of Oman is one of the most Wave Modeling (ISWM).
important waterways in the world. In The main goal of ISWM is to identify
peak periods, one ship passes the Strait the wave climate in Iranian seas using
of Hormuz every 6 minutes, and the results of wave simulations. This is
approximately 60% of the world marine performed by employing the latest
transport of oil comes from this region version of wave numerical models (3rd
(Reynolds, 1993). The Persian Gulf is a generation) and the tuning with in-situ
marginal sea measuring some 1000 km measurements and available satellite
in length and 200-300 km in width, data. If successfully validated, this data
covering an area of approximately set can be used to compute extreme

12/E
2 Vol. 4/ No. 7/ Spring 2008
 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING

waves statistics. The ISWM results in the driving wind speed will result in
present the required wave parameters for an error of at least 20% in the hindcast
different applications in harbor and wave height (Weisse and Feser, 2003).
coastal engineering fields. Iranian Therefore, obtaining the wind field may
National Center for Oceanography be the most problematic task of wind-
(INCO), in association with DHI Water wave prediction. The following wind
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

& Environment undertook this project. sources were evaluated for possible use
This paper is organized as follow. First, in generating the long term wave
the methodology and general approach hindcasting.
towards the wave hindcast study are For long period simulations, however,
briefly described. The in-situ and additional requirements regarding the
satellite-measured wind speed and quality of the wind fields have to be
direction and wave height data sets are made. For instance, the wind field
then introduced. The data has been used should be free of any artificial trends.
for model tuning and result evaluation. They should be homogeneous in time
EVA results which are of great and for coastal applications, their spatial
importance in this research have been and temporal resolution should be high
completely discussed. Derivative enough to resolve the relevant
products and conclusions are provided in topographic features in the coastline in
the remaining parts. order to obtain appropriate wave/surge
model simulations. While global
2- Wave hindcast methodology reanalyses provide a useful product for a
The purpose of the study is to provide variety of studies (e.g., climate studies),
a reasonable reconstruction of the wave their spatial and temporal resolution
conditions and the wave climate over the remains too coarse for many
past 11 years, from 1992 to 2002, for the environmental applications, such as
Iranian coastal area. Hindcast is ocean or wave modeling in coastal areas
completed by individual modeling of (Weisse and Gayer, 2000).
cyclones coming from north east of Among the identified wind fields,
Indian Ocean during 1975-2004. The NCEP/NCAR products have been
following approach has been adopted for discarded because of their low spatial
the Persian Gulf and the Gulf of Oman. resolution for coastal modeling and not
covering the required modeling period
2-1- Atmospheric/Ocean forcing (1992-2002), regardless of their
It is accepted that the 3rd generation accuracy. Besides, based on the data
wave models are satisfactory reliable for assessment, the ERA-40 wind speeds
many practical applications. Therefore, compare better with the in-situ and
for reliable wave prediction, it is altimeter observations than the
essential to have reliable input wind NCEP/NCAR wind speeds (Caires et al.,
fields. The results of any numerical wave 2002). The other source, U.K. Met
hindcast study depend heavily on the Office data set, does not cover the
quality of the wind data used to drive the simulation period. The only remaining
model and are, therefore, only as good as source is ECMWF 6 hourly 10 m wind
the input data (Hubertz et al., 1991). For field. Both operational and reanalysis
instance, for a fully developed sea, wave data sets were purchased and despite the
height approximately scale with the non-homogeneous output of operational
square of the wind speed (Tolman, 1998) model, it was chosen due to the better
which implies that an error of about 10% agreement with measured data in coastal

Vol. 4/ No. 7/ Spring 2008 13/E

3
 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

and islands stations and the higher is only 60 km (0.54 degree) there, which
spatial resolution which is an obligation requires high resolution data for input
for modeling the narrow strait of and output parameters. A brief review of
Hormuz (Figure 1). The minimum width global atmospheric models is given in
Table 1.
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

Fig.1- General view of the study area and wind/wave stations

Table 1- Spatial and temporal resolution of global wind fields

in the Persian Gulf and the Gulf of Oman

Resolution
Production Time step
Organization Time Period Lat. × Long.
Type (Hour)
(degree)
NCEP/NCAR Reanalysis 1957-1998 1.25 × 2.5 6
1 Jan.1985-30 Apr.1985 1.875 × 1.875

1 May 1985-16 Sep. 1991 1.125 × 1.125

17 Sep.1991-20 Nov. 2000 0.56 × 0.56

Operational 6
ECMWF
21 Nov. 2000- 31 Jan. 2006 0.35 × 0.35

1 Feb. 2006 up to now 0.22 × 0.22

Reanalysis 1957-2002 1.125 × 1.125 6

U.K MET OFFICE Operational 1999 up to now 0.55 × 0.833 6

4
14/E Vol. 4/ No. 7/ Spring 2008
 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING

The wave parameters on the open It has been extensively tested and
boundary of the Gulf of Oman can be verified at a range of sites throughout the
obtained from ECMWF Global wave world. The wave spectrum is resolved in
model, both spectral and parametric data, 16 directional bins (22.5ْ angular
and U.K. Met Office parametric wave bandwidth) and 25 frequency bins
data. A few short periods were simulated (Df/f=0.1).
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

using these boundary data and the results Hourly output is adopted on a regular
comparisons with wave measurements grid spacing of 0.125ْ in latitude and
showed that a better agreement can be longitude and the modeling results in
achieved using ECMWF spectral wave smaller bays and estuaries are also
data. available via unstructured output files.
Due to the importance of EVA for Test runs and model tuning was done
marine and coastal studies, cyclones, separately for two basins (Persian Gulf
which might have the return period and Gulf of Oman). White Capping
longer than 11-year simulation period, parameters, wave breaking parameters
have been considered separately. Thus, and bottom friction coefficient are
cyclones specifications including their selected by changing them in their
time, maximum wind speed, and trace normal range to obtain best agreement
within latest 30 years (1975-2004) were for both wave height and period. Wind
obtained from Unisys Weather. fields are interpolated at 3 hours
intervals, while the model execution
2-2- Wave modeling time step is 300 seconds. Time step is
The wave model used for this study is automatically getting smaller by the
Mike 21 Spectral Wave model, a 3rd model in the fine mesh near the Iranian
generation spectral wind-wave model coasts. Output wind and wave
based on unstructured meshes (Figure 2). parametric data up to 64ْ E and 22ْ N are
DHI Water & Environment (2003) has archived for users at 6 hourly intervals at
developed this model that simulates the all model grid points, while directional
growth, decay and transformation of spectra are archived at 6 hourly intervals
wind-generated waves and swells in every 0.125ْ in Iranian coasts and every
offshore and coastal area. 0.25ْ in other areas.

Fig. 2- Unstructured mesh and open boundary on 18ْ N

Vol. 4/ No. 7/ Spring 2008 5

15/E
 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

3- Validation data 3-3- Satellite

Both input and output data are Sea winds and waves from Topex and
supposed to be validated against accurate QuikSCAT instruments are used for
measured data. Although we tried to comparing with ISWM winds and
choose the best available wind field and waves. Topex data were obtained from
model set-up, input data and model World Wave Atlas (WWA) along 10
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

parameters still need to be assessed and tracks (Figure 3). A series of data control
modified. tests have been developed with special
emphasis on track consistency and with
3-1- Synoptic station a fair degree of manual inspection and
Synoptic stations are meteorological control. Root-Mean-Square Error
stations which record different (RMSE) after using calibration function
atmospheric parameters including wind has been calculated 0.30 m for the
speed and direction regularly every 3 significant wave height comparing with
hours. Regarding the length of recording in-situ data from NOAA, Norwegian Sea
data and accuracy of measurements, and South Pacific (Oceanor, 1999). The
several stations located along Iranian sea winds scatterometer on QuikSCAT
coasts and inside Iranian islands were satellite started operating in July 1999
chosen for assessing the wind fields and continues through the present. The
(Figure 1). grided QuikSCAT Level 3 data up to
2003 at a resolution of 0.25ْ × 0.25ْ was
3-2- Buoy obtained from the JPL sea winds project
Two buoys, whose data are used in (http://podaac.jpl.nasa.gov/). Separate
this study, come from Islamic Republic maps are provided for daily ascending
of Iran Meteorological Organization and descending passes. The project
(IRIMO) and are located in Bushehr and requirement for wind speed and direction
Chabahar. They are located in 27 and 17 is 2 m/s RMSE for wind speeds from 3
m depth and have measured wave to 20 m/s and 10% for wind speeds up to
characteristics for 2~3 years 30 m/s and 20° for wind speeds from 3 to
discontinuously. Asaluye buoy belongs 30 m/s, respectively (Podaac, 2001).
to National Iranian Oil Company Three versions of QuikSCAT/SeaWinds
(NIOC) operated for one year at 7.5 m wind data (L2B, DIRTH, RSS) were
depth, covering a few months of collocated with buoy observations
simulation period. The measurements; operated by the National Data Buoy
including significant wave height, mean Center (NDBC), Tropical Atmosphere
wave direction and wave period are Ocean (TAO), as well as, Pilot Research
available hourly or at 3 hours intervals Moored Array in the Tropical Atlantic
from three buoys (Figure 1). These (PIRATA) project, and the Japan
measurements have gone through some Meteorological Agency (JMA).
quality controls. Observations that According to Ebuchi et al. (2002), wind
deviate more than 6 times the standard speeds and directions observed by
deviation of the monthly data from its QuikSCAT agree well with the buoy
mean, or more than 2 times the standard data. The RMSE of the wind speed and
deviation of the monthly data from the direction are 1.01 m/s and 25ْ,
previous observation, are identified as respectively.
outliers and removed from the data.

16/E Vol. 4/ No. 7/ Spring 2008

 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING

Strait of Hormuz. On the other hand,

wave boundary condition on 18ْ N is
underestimated at the time of tropical
cyclones coming from the Indian Ocean
and is modified by performing a
supplementary model. Such deficiency
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

also exists in the NCEP/NCAR

reanalysis surface winds over the
Fig. 3- Topex tracks in the Persian Gulf and
equatorial Indian Ocean (Goswami and
the Gulf of Oman Sengupta, 2003). Thus, wind fields are
generated for individual cyclones from
4- Validation of ISWM 1992 to 2002 and the cyclone modeling
While ECMWF wind field shows covering additional area from 10ْ N to
generally good agreement with offshore 18ْ N is performed using ECMWF
synoptic stations and satellite parametric wave data on 10ْ N (Figure
measurements, coastline wind speeds are 4). Then, the spectral wave data on 18ْ N
underestimated due to the rectangular are extracted from a supplementary
land-sea mask which is used in ECMWF model and merged into the spectral
atmospheric model. Comparative time boundary condition for the main model.
series and exceedence diagrams have The main model is calibrated in selected
been drawn for several points and data periods and after an 11-year wave
assimilation and increasing factors were simulation, the output is evaluated
applied at limited points. For best usage against validation data. Cyclone
of assimilated data, ECMWF wind field simulation is not limited to a continuous
is interpolated into 3 hours time step and hindcast period, but is also extended to
synoptic measurements are assimilated 30 years from 1975 to 2004. EVA results
into background wind field. Regarding of individual cyclones, which are of
non-homogeneity of data specially great importance, are discussed in next
caused by changing the model grid parts.
spacing from 0.56ْ to 0.35ْ in Nov. 2000,
time dependent factors are applied in the

Fig. 4- Supplementary modeling area for cyclone periods

Vol. 4/ No. 7/ Spring 2008 17/E
 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

Figure 5 shows typical time series of the Gulf of Oman, respectively.

significant wave height (Hs) and Mean Recorded data are not simultaneous;
Wave Direction (MWD) for Bushehr and therefore time series do not cover the
Chabahar buoys in the Persian Gulf and same period.
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

Fig. 5- Comparison of ISWM results versus buoys

18/E Vol. 4/ No. 7/ Spring 2008

 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING

These time series describe different wave

CC =
(
∑ xi − x yi − y)( ) (4)
climates in these basins. A large variation
of wave height in Bushehr station implies
( ) (
2
∑ xi − x ∑ yi − y )
2

dominance of wind-waves in the Persian

Gulf in all seasons. In contrast, seasonal In all these formulae, the xi 's represent
variation of wave climate in Chabahar the observation, the yi 's represent the
ISWM product, x and y are the average
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

station is quite clear. Wave height in

summer period is relatively high with a and n is the number of observations.
constant direction because of A global view of the wave quality can be
southwestern monsoon. The only sharp obtained by comparing the ISWM wave
peak of Hs in this period is generated by data with buoy data in Iranian coasts and
a tropical cyclone dated 1~9 June 1998. several satellite points along Topex
Wave climate in the Gulf of Oman shows tracks in the Persian Gulf and the Gulf of
a different pattern in winter season. Wave Oman. Standard statistics have been
height and direction from October to calculated for all data and for higher
February is influenced by the local winds waves (Hs>0.5 m) in Table 2. Since a
rather than swells. In general, simulated wide range of values have been obtained
Hs tracks the buoy observation. However in different projects, two well-known
typically, the lowest waves tend to be references from Cox and Swail (2001) in
slightly overestimated. Global Reanalysis of Ocean Waves
The differences between ISWM wave (GROW) based on NCEP/NCAR wind
height product and the observations are fields, and Caires et al. (2002) in ERA-40
also quantified by computing some project are considered for comparison.
standard statistics such as Bias, RMSE, All the derived values are within the
Scatter Index (SI), and Correlation acceptable range and the ISWM accuracy
Coefficient (CC). increases by moving toward higher wave
heights in nearshore locations.
Bias = y − x (1) Another parameter, which is used
specifically in this project, is the Average
RMSE =
1
∑ (y i − xi )
2 (2) Relative Error (Liu and Frigaard, 2001).
n
1 yi − xi (5)
1
(( yi − y ) − (xi − x ))2
ARE (%) = ∑ × 100
n
∑ (3)
n xi
SI =
x

Table 2-Eror values in ISWM project

Buoys in nearshore Satellites in offshore Range from Cox Range from Caires
Wave Wave and Swail (2001) et al. (2002)
Error value All wave All wave
heights > heights > for wave height in for wave height in
heights heights GROW ERA-40
0.5 m 0.5 m
Bias (m) 0.06 0.04 -0.07 -0.10 -0.32 – 0.85 -0.44 – -0.02
RMSE (m) 0.22 0.24 0.35 0.38 n/a 0.31 – 0.71
CC 0.86 0.71 0.85 0.83 0.67 – 0.93 0.82 –0.95
SI 0.39 0.25 0.40 0.30 0.17 - 0.60 0.13 – 0.32

Vol. 4/ No. 7/ Spring 2008 19/E

 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

Due to the importance of simulation of Figure 7 shows the scatter and quantile
big waves, especially for the EVA which plots of the Topex significant wave
is part of this study, this parameter has height observations and the
been calculated for waves higher than 1 corresponding ISWM data on more than
m. ARE is 12% for nearshore locations 1600 wave records over 7 points in
and 19% for satellite offshore that shows Figure 3. The ISWM underestimates
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

good agreement between model and some of the high peaks of the significant
measurements. ARE is reduced to 8% wave height and shows good
and 18%, in nearshore and offshore correspondence with the observations at
respectively, for waves higher than 1.5 m. low sea states. It was quite predictable
The scatter plots and quantile plots help from negative bias in satellite which is
visualizing the differences and decreasing by moving toward higher
deficiencies of the data set. Figure 6 wave heights. Negative bias implies that
shows excellent agreement between the average of modeled values is smaller
buoys and ISWM wave heights. As it was than measured ones.
shown in buoy comparison, there is a
tendency in nearshore output to
overpredict slightly the lowest sea states.
The same feature exists in GROW and
ERA-15 projects (Cox and Swail, 2001).

(a)

(a)

(b)
Fig. 7- Scattering (a) and quantile (from 1 to
99%, b) wave height (m) comparisons of
ISWM and Topex measurements

It should be noted that the wind field

(b) modification and the model tuning were
Fig. 6- Scattering (a) and quantile (from 1 to
performed to capture the best results in
99%, b) wave height (m) comparisons of
ISWM and buoy measurements Iranian coastline. This procedure has

20/E Vol. 4/ No. 7/ Spring 2008

 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING

produced a good agreement for Iranian value series. Different distribution

nearshore locations, but the accuracy is functions and fitting methods are
unknown for other coasts. This evaluated to find the risk of extreme
deficiency can be treated using wind and events in the study area. Truncated
wave measurements of southern Gumbel shows the best goodness of fit
coastlines in model tuning and statistic, less standard deviation in
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

performance in future and the ISWM uncertainty calculations and smooth

results would be hopefully an invaluable spatial pattern for both wind speed and
source for regional coastal engineering wave height. Figure 8 shows 100-year
and management. wave height and standard deviation of
simulated values with Monte Carlo
5- EVA Results and discussion method in the output area. Details of
Long term wind and wave distribution and simulation method are
characteristics for 5, 20, 50 and 100 provided in Appendix A.
years return periods are calculated based
on fitting a theoretical probability
density function to the observed extreme

Fig. 8- 100-year Hs (top) and corresponding standard deviation (bot.) in the Persian Gulf and the
Gulf of Oman based on an 11- year continuous wave hindcast

Vol. 4/ No. 7/ Spring 2008 21/E

 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

There are recommendations such as 15 Extreme calculation is verified by means

or 30 years data to have a reliable 100 of EVA analysis on available
year wave height estimate, but there is measurements. Satellite data from Topex
no specific work or paper regarding the cover the study area in hindcast period, but
minimum length of storm wave data for 10 days time step for data recording does
estimating the design wave height with not provide an appropriate wave field for
extreme calculations. Thus, Extreme
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

the return period of 100 years or so. It is

well known that the width of confidence values are calculated on buoys
interval of the 100 year wave height measurements and corresponding values
decreases with the increase in the data are obtained from the model. Bushehr and
Chabahar buoys with a minimum of two
length and the number of storm events
years of recording data are selected for
(Goda, personal communication).
comparisons. Table 3 shows EVA results
Therefore, it is recommended to consider from recorded and simulated data as well
a range of wave height (i.e. as correlation coefficient between
Hs ± 2 × Standard deviation) rather than a extracted values and corresponding values
single value for 100 years return period. from truncated Gumbel distribution.
Similar analysis was performed on 41 Coefficients are satisfactory for fitness of
individual simulated cyclones, occurred selected distribution to extreme data.
between 1975 and 2004 in the Gulf of Extreme values would be more applicable
Oman. According to Figure 9, maximum in engineering if directions of coming
100-year wave height increases from 5.5 wind and wave in different return periods
to 8 m while there is a slight decrease in are specified. The results would be helpful
wave height in Iranian coasts. Therefore, for engineering purposes such as
EVA results of cyclone modeling in optimizing marine structures layout,
offshore area are assimilated into the design and construction. This demand was
main data base, derived from 11-years met by dividing wind and wave data into
simulation. eight 45ْ directions. Figure 10 shows 100-
year wave height in dominant directions in
the Persian Gulf and the Gulf of Oman
with special attention to Iranian coasts. It
is interesting to know that all values from
directional analysis are smaller than values
from non-directional analysis. Wave
height in dominant directions e.g. North
West, West and South East directions
which are shown in this Figure are at most
Fig. 9- 100-year Hs in the Gulf of Oman based equal to wave height in non-directional
on a 30-year individual cyclone simulation analysis given in Figures 8 and 9.

Table 3- Comparison between EVA results from buoys and ISWM data

Location Bushehr Bushehr Chabahar Chabahar

Parameter Buoy ISWM Buoy ISWM
2-year Hs(m) 2.47 2.29 1.95 2.03
5-year Hs(m) 2.65 2.42 2.03 2.17
10-year Hs(m) 2.77 2.51 2.08 2.27
CC 0.98 0.98 0.93 0.95

22/E Vol. 4/ No. 7/ Spring 2008

 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

Fig. 10- 100-year Hs in the Persian Gulf and the Gulf of Oman in dominant directions

6- Derivative products
The main product of ISWM consists IWA program. Wind and wave rose also
of a sophisticated GIS-based program can be drawn in this program. It should
called Iranian Wave Atlas (IWA) which be mentioned that IWA consists of the
covers all the results of the simulation in wave modeling results in the Caspian
the Persian Gulf and the Gulf of Oman Sea that is not discussed in this paper.
(Figure 3). IWA is a comprehensive Figure 11 shows the interface of IWA
database for engineering and software. Users can view geographical
environmental purposes. This program information of the grid points on the
provides wave parameters including map. Data can be uploaded by selecting
significant wave height (Hs), wave points and extracting time series. Simple
periods (Peak wave period and T02), statistical analysis can also be executed
mean wave direction for swells, wind in this software. Explanations about each
waves and total waves separately, as button are given in the help of software.
well as wind speed/direction in each
point, using on-line display position
(lat/long) of the pointer. Parametric data
are provided up to 5 meters depth at
0.125 degree spatial resolution and 6
hours temporal resolution and the wind
and wave EVA results are available at
the same points. The wave energy
spectra are archived at 0.125 degree
resolution for nearshore and 0.25 degree
resolution for offshore areas. Many
statistical parameters such as minimum,
maximum, average, monthly and
quarterly average, frequency and Fig. 11- A view of IWA software
probability density can be easily
calculated for each point/parameter in

Vol. 4/ No. 7/ Spring 2008 23/E

 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

7- Conclusions • Directional EVA results do not

The ISWM project has successfully exceed non-directional values in
generated a comprehensive set of wave these basins and provide an
data for the Persian Gulf and the Gulf of applicable data source for
Oman basins for an 11-year continuous economic marine design.
period (1992-2002) and 41 storms of • The research output which is
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

various dates between 1975 and 2004 in accessible via IWA program will
the Gulf of Oman. prove invaluable for design and
• The approach is based on driving operational purposes and
modified operational ECMWF environmental managements.
wind field over the whole basins. The results can be evaluated
Open boundary of the Gulf of against new altimeter and in-situ
Oman was obtained from the observations. Although
same center and the big swells satisfactory results are obtained
generated by cyclones were at this stage, accuracy can be
assimilated in the observed time improved by carrying out a local
by the satellite with the wind modeling and higher
appropriate strength. Tuning was resolution wave hindcast. Wave
generally performed for Iranian forecasting can be added to
coasts and the offshore area and future project as well.
southern coasts were at lower
interest. Acknowledgements
• The statistical parameters show This research is partially funded by
good agreement between ISWM Iranian Port and Ship Organization
results and validation data (PSO). The authors are grateful to
especially in term of bias. All the Morten Rugbjerg from DHI Water &
parameters have been compared Environment for his technical supports.
with the values derived from
reference global hindcasts and Appendix A
are in acceptable range.
• ISWM results show excellent Gumbel Distribution
agreement with nearshore The Gumbel distribution is given as
measurements in Iranian coasts, 1 x −ζ x −ζ
but the offshore data are a bit f ( x) = exp[− − exp(− )] (A.1)
α α α
underestimated for big waves.
• Test runs and related x −ζ (A.2)
F ( x) = exp[− exp(− )]
investigations in EVA show that α
the most appropriate statistical
distribution for both wind and Where f(x) is the probability density
wave is truncated Gumbel. The function and F(x) is the cumulative
wind speed and wave height over distribution function or the probability of
the two basins have been non-exceedance, ζ is location parameter
calculated for 5, 20, 50 and 100 and α the scale parameter.
years return periods. The 30-year A truncated Gumbel distribution for
cyclones in the Gulf of Oman are modeling exceedances above the
considered in the EVA and have threshold level x0 can be defined by g(x)
shown strong effect in the
offshore area.

24/E Vol. 4/ No. 7/ Spring 2008

 AN APPROACH TOWARDS WAVE CLIMATE . . . JOURNAL of MARINE ENGINEERING

and G(x) as probability density function References

and cumulative distribution function. 1-Bayler, G. ,and H. Lewit. (1992): “The
navy operational global and regional
f ( x) (A.3) atmospheric prediction system at the fleet
g ( x) =
1 − F ( x0 ) numerical oceanography center”. Weather
and Forecasting, 7(2): 273–279.
f ( x) − f ( x 0 )
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

G ( x) = (A.4)
1 − F ( x0 ) 2-Caires, S., A. Sterl, J. R. Bidlot, N.
Graham , and V. Swail. (2002):
“Climatological assessment of reanalysis
Monte Carlo Simulation ocean data”. In: Proceeding of 7th
In Monte Carlo Simulation the International Workshop on Wave
standard deviation is obtained by Hindcasting and Forecasting, October 21-25,
randomly generating a large number of 2002, Banff, Alberta, Canada, 1-12.
samples that has the same statistical
characteristics as the observed samples. 3-Caires, S., and A. Sterl. (2003):
The algorithm can be summarized as “Validation of ocean wind and wave data
follow using triple collocation”. Journal of
Geophysical Research, 108(C3), 3098,
1. Randomly generate a set of m
doi:10.1029/2002JC001491.
data points from considered
distribution using the estimated 4-Cox, A. T. , and V. R. Swail. (2001): “A
parameters, i.e. global wave hindcast over the period 1958-
xi = F −1 (ri , θ ) i = 1,2,3,..., m (A.5) 1997: Validation and climate assessment”.
Journal of Geophysical Research, 106(C2):
2313-2329.
Where ri is a random generated
number between 0 and 1, θ is 5-DHI Water & Environment. (2003):
calculated parameters for “Mike 21 Spectral Wave Module: User
distribution such as location and Guide”, DHI Water & Environment ,
scale, m is equal to sample size Denmark, 84 pp.
used in extreme analysis before.
6-Dibajnia, M. (2002): “Wave climate study
Then, wave height for selected
for Al Ashkharah – Oman”. In: Proceeding
return period ( xt for t years) is of 5th International Conference on Coasts,
calculated for m data. Ports and Marine Structures, October 14-17,
2. Step 1 will be repeated k times. 2002, Ramsar, Iran, 119-122.
The standard deviation st of the T-
7-Ebuchi, N., H. C. Graber ,and M. J.
year event estimate are then given Caruso. (2002): “Evaluation of wind vectors
by observed by QuikSCAT/SeaWinds using
2
1 k ocean buoy data”. Journal of Atmospheric
s = ∑ ( xt( j ) − xt )
2
(A.6)
t
k j =1 and Oceanic Technology, 19: 2049-2062.

8-Goswami, B. N. ,and D. Sengupta. (2003):

Investigations suggest that Monte Carlo “A note on the deficiency of NCEP/NCAR
based estimates of mean and standard reanalysis surface winds over the equatorial
deviation of the T-year event estimator Indian Ocean”. Journal of Geophysical
saturate at sample size in the order of Research, 108(C4), 3124, doi:
k=10000. 10.1029/2002JC001497.

9-Hubertz, J. M., D. B. Driver, and R. D.

Reinhard. (1991): “Wind waves on the Great

Vol. 4/ No. 7/ Spring 2008 25/E

 Iranian Association of
Naval Architecture & Marine Engineering
JOURNAL of MARINE ENGINEERING

Lakes: A 32 year hindcast”, Journal of 15-Reynolds, R. M. (1993): “Physical

Coastal Research, 7(4): 945-967. Oceanography of the Gulf, Strait of Hormuz,
and the Gulf of Oman-Results from the Mt
10-Liu, Z., and P. Frigaard. (2001): Mitchell Expedition”. Marine Pollution
“Generation and Analysis of Random Bulletin, 27: 35-59.
Waves”. Aalborg University Press,
Denmark, 76 pp. 16-Sterl, A., G. J. Komen, and P. D. Cotton.
Downloaded from marine-eng.ir at 3:03 +0330 on Saturday October 23rd 2021

(1998): “Fifteen years of global wave

11-Oceanor. (1999): “World Wave Atlas: hindcasts using winds from the European
User's Manual Version 2”, Oceanor, Center for Medium-Range Weather
Norway, 33 pp. Forecasts reanalysis: Validating the
reanalyzed winds and assessing the wave
12-Physical Oceanography Distributed Active climate”. Journal of Geophysical Research,
Archive Center. (2001): “SeaWinds on 103(C3): 5477-5494.
QuikSCAT Level 3(JPL SeaWinds Project):
Guide Document”, Available online at: 17-Tolman, H. L. (1998): “Validation of
http://podaac.jpl.nasa.gov:2031/DATASET_D NCEP's ocean winds for the use in wind
OCS/qscat_L3.html, 39 pp (accessed August wave models”, Global Atmosphere and
12, 2004). Ocean System, 6(3): 243-268.
18-Weisse, R., and F. Feser. (2003):
13-Resio, D. T. (198): “The estimation of
“Evaluation of a method to reduce
wind-wave generation in a discrete spectral
uncertainty in wind hindcasts performed
model”, Journal of Physical Oceanography,
with regional atmosphere models”, Coastal
11(4): 510-525.
Engineering, 48(4): 211-225.
14-Purser, B. H. ,and E. Seibold. (1973):
19-Weisse, R. ,and G. Gayer. (2000): “An
“The principal environmental factors
approach towards a 40-year high resolution
influencing holocenes sedimentation and
wave hindcast for the Southern North Sea”.
diagenesis in the Persian Gulf”. Pp.1-9 in
In: Proceeding of 6th International
The Persian Gulf, Purser, B. H., editor,
Workshop on Wave Hindcasting and
Springer-Verlag, Berlin, New York.
Forecasting., November 2000, California,
USA, 1-7.

26/E Vol. 4/ No. 7/ Spring 2008