Hydrodynamic Model

for Oil Spill Trajectory Prediction

Anukul Buranapratheprat1 and Supichai Tangjaitrong2

ABSTRACT
The paper described effect of tidal current on oil spill trajectory model. The model, which
incorporated the Ekman current, Stoke drift motion, background current and tidal current, has
been investigated. A field experiment for testing the model was carried out on March 30, 1995
around Ko Sichang, Thailand. Drift cards were used to represent oil slick. Observed and
computed positions of drift cards were compared. The results show that accuracy in trajectory
prediction is higher when tidal current computed by hydrodynamic model was included.

INTRODUCTION
Transportation of crude oil in the Gulf of Thailand has increased because of the industry development.
The number of tankers in shipping routes in the Upper Gulf particularly in Choa Phraya River and near
Laem Chabang Port in the Gulf of Thailand has been increasing for the past decades. These areas are not
only supporting tankers but also other ships; hence, increasing the opportunity of oil spill by ship collision.
Oil spill models for predicting trajectory at different time of crude oil is necessary for remedial measures if
the oil spill occurred. This paper presents the use of a hydrodynamic model to predict oil spill trajectory.
Results of the model are compared with field observations carried out earlier in March 1995.

OIL SPILL TRAJECTORY MODEL

The oil spill trajectory model is a numerical model to predict path of oil slick after spilling in the sea.
The basic concept of the model is that the motions of oil slick on the sea surface come from several forces,
and the trajectory is dominantly directed by the net force in that position and time. Oil drift mechanisms
are governed by the following contributors
1) motion relative to the water caused by wind (sailing effect),
2) motion due to the wave propagation (Stokes drift),
3) tidal currents,
4) permanent or semi-permanent current system on a large scale (background current), and
5) any other more or less stochastic motion of the water.
As the vector sum of the stochastic motions over a sufficient time period will approach zero, the
motion can be excluded from the model.

Sailing effect
For the sea of finite depth, the effect of earth’s rotation can deflect surface current from the wind
direction by an angle:

1
Department of Aquatic Science, Faculty of Science, Burapha University
2
Department of Marine Science, Faculty of Science, Chulalongkorn University

6D - 1
Session 6 on Ocean Transport Models and Marine Ecosystem Models

  h  h 
 sinh 2π  − sin 2π
d

d 
α = tan −1    
 sinh 2π h  h 
   + sin 2π 
 d  d  

where h = water depth, d = depth of frictional influence = (2A/f)1/2, A = vertical eddy viscosity, f =
Coriolis parameter = 2Ωsinφ, Ω = angular speed of the earth rotation and φ = latitude
In oil spill trajectory model, surface current is set equal to 1.3 – 1.4% of the wind speed. The direction
of the current depends on latitude. In area above 10οN and beneath 10οS, the current vector is deflected
with a constant angle α10 = 33ο to the right or left of the wind direction in northern or southern hemispheres
respectively. In area between 10οN and 10οS, the deflection angle (α) reduces linearly with latitude φ and
can be approximated as α = α10 φ /10

Stokes drift
Wave can generate current and mass transport in its direction. Such current, generated from the
motion of water particles that move in an almost closed circle, results in a very small displacement in the
direction of wave propagation. In shallow water, bottom friction flattens the particle motion hence
generating a larger displacement. The mass transport velocity or Stokes drift for shallow sea can be
calculated as
2
2
H
US = π   C
L
where Us = velocity of oil drifted by wave for shallow water, H/L = wave steepness (wave height over
wave length) and C = wave speed. Although Stokes drift has a small magnitude (1 – 2% of wind speed), it
should not be neglected in the oil spill model. For simplicity, the oil spill trajectory model has set this
motion to be parallel to the wind with 1.7% of the wind magnitude.

Tidal current
Tidal current is considered not important in the open sea where the net drift of water mass during a
tidal cycle is very small. For shallow water, however, several effects such as bottom friction, shorelines,
and Coriolis effect will affect the tidal ellipse and could generate a relatively larger drift.
A hydrodynamic model is needed in order to estimate the tidal current. There are several models
developed for the Gulf of Thailand during the past decades (see Table 1). This study adopted the concept
developed by Banpapong et al. (1985) and modified the boundaries to match the configuration of the study
area. The momentum and the conservation of mass equations of the models are (Buranapratheprat, 1997):
r
∂Q r r r r
+ ( f k × Q ) + gD∇ξ = TS − TB
∂t
∂ξ r r
+∇ •Q = 0
∂t
r
Q : mass transport per unit width, ξ : water elevation, D : density related parameter
where r r
TS : surface stress, TS : bottom stress

6D - 2 The Second OMISAR Workshop on Ocean Models

 Hydrodynamic Model for Oil Spill Trajectory Prediction

Table 1. Hydrodynamic models of the Gulf of Thailand

Area Technique Authors
Upper Gulf of Thailand 2D; finite difference; tidal-driven Vongvisessomjai et al., 1978
Upper Gulf of Thailand 2D; finite element; tidal-driven Liengcharensit, 1979
Gulf of Thailand 3D finite difference; wind-driven Nelasri, 1981
Gulf of Thailand 3D; finite difference; wind-driven Sojisuporn, 1984
Gulf of Thailand 3D; finite difference; wind-driven Sripanyawitchya, 1988
Upper Gulf of Thailand 2D; ADI finite difference; tidal-driven Charuskumchornkul, 1988
Upper Gulf of Thailand 2D; finite element; tidal-driven Chokechalermwat, 1990
South China Sea 2D; finite difference; wind-driven Azamy et al., 1991
Gulf of Thailand 2D; finite difference; wind-driven Archevarahuprok and Wongwises, 1994
Upper Gulf of Thailand 3D; finite difference; density-driven Sojisuporn, 1994
Gulf of Thailand 3D; finite difference; density-driven Snidvongs and Sojisuporn, 1997

Background currents
Background currents are the large current system that appears to be permanent or semi-permanent.
Magnitude and direction of background current cannot be calculated by wind vector like Ekman current and
Stoke drift. Therefore, they are usually estimated by averaging the current data over a long period of time
so as to remove the directly wind induced current and the tidal motion.

FIELD EXPERIMENT AND MODEL OPERATION

Field experiment
The drift cards experiment was exercised near Ko Sichang (Figure 1) at 6.00 hrs. on March 30, 1995.
Approximately 200 pieces of laminated postcards were released at latitude 13°08′ N and longitude 100°50′
E. Wind speed and direction at the released time were measured. After that, the motions of drift cards on
the sea surface were tracked and the positions recorded (with a GPS) every 3 hours. Wind data were also
measured at the new positions of the cards. The field experiment finished at 15.00 hrs., because it was hard
to trace the drift card for the next time-step (16.00 hrs.).

Gulf of Thailand

Figure 1. The study area.

The Second OMISAR Workshop on Ocean Models 6D - 3

Session 6 on Ocean Transport Models and Marine Ecosystem Models

Model operation without tidal current

The study used the simplified model developed by Hang et al. (1989) for prediction the oil spill
trajectory. Basic input for the model included background current, real-time observed wind, location and
time of oil leak. Tidal currents were estimated from a time-dependent, normal mode hydrodynamic model.
The model was originally developed by Banpapong et al. (1985) and was adjusted to fit the computational
domain for the Gulf of Thailand by Buranapratheprat (1997).
The background current was digitized from seasonal data compiled by Siripong (1985). The data
were temporally interpolated from seasonal to monthly using yearly and half-yearly Fourier components.
Then, the data were spatially interpolated from the available 1.0 degree resolution to grid point of 0.1
degree resolution. Current vectors in grids next to shore were adjusted to parallel along coast.
The observed wind, position and time when oil was spilled over the sea must be inputted explicitly.
The wind data were used to calculate the sailing effect or Ekman current and Stokes drift. The Ekman
current was vector summed with the background current. Since the study area is located in the monsoon
region, the seasonal variations of the currents are considerable and are reflected in background current like
Ekman current.
Next, the net current from Ekman and background current was vector summed with Stokes drift. The
new position of oil slick in the sea was then calculated from this solution. (The oil slick was considered as a
piece of floating object that cannot spread or disperse.) The model predicted the oil position every 3 hours
by using 6-hour observed wind data. Figure 2 shows results of the model comparing with the observed
position of the drift cards.

Figure 2. Computed (solid circle) and

observed (solid square) positions of drift
cards on March 30, 1995.

Model operation with tidal current

This operation was similar to the previous operation except that tidal current was vector summed to
the net current (of Sailing or Ekman current, background and drift currents) before it was used to calculate
the slick positions. The tidal current was estimated from the simplified hydrodynamic model discussed
earlier. Time derivative was represented by the standard centered differences. For space derivative, a finite
difference of the spherical coordinate system was employed.

6D - 4 The Second OMISAR Workshop on Ocean Models

 Hydrodynamic Model for Oil Spill Trajectory Prediction

The multi-operational alternating direction implicit (ADI) algorithm developed by Lendertse (1967)
was used for time integration of the finite difference equations. During the odd time steps, elevation (ψ) and
velocity in the east-west direction (U) were computed implicitly along lines of constant latitude, then
velocity in the north-south direction (V) was computed explicitly. During the even time steps, ψ and V
were computed implicitly along lines of constant longitude, and U was computed explicitly.
Parameter shown in Table 2 was used to obtain a series of tidal current as shown in Figure 3. Figure 4
shows results of the model comparing with the observed position of the drift cards.
(a) (b)

(c) (d)

Figure 3. Calculated tidal current on March 30, 1995 (a) at 6.00 hrs (b) at 9.00 hrs (c) at 12.00
hrs (c) at 15.00 hrs.

The Second OMISAR Workshop on Ocean Models 6D - 5

Session 6 on Ocean Transport Models and Marine Ecosystem Models

Figure 4. Computed (solid circle) and

observed (solid square) positions of drift
cards on March 30, 1995, after including
tidal current in the computation.

Table 2. Parameters used in the tidal current model

Components Variable Values

Tidal Constituents K1 Amplitude = 22.7 cm phase = 195.2°
at the upper point O1 Amplitude = 15.9 cm phase = 150.6°
(Campong Som) M2 Amplitude = 13.7 cm phase = 5.3°
S2 Amplitude = 9.1 cm phase = 111.0°
Tidal Constituents K1 Amplitude = 48.2 cm phase = 66.0°
at the lower point O1 Amplitude = 26.7 cm phase = 11.9°
(Kuala Trengganu) M2 Amplitude = 29.4 cm phase = 207.5°
S2 Amplitude = 11.3 cm phase = 284.6°
Surface (wind) stress K 1.1×10-6 (dimensionless constant)
Bottom stress k 2.5×10-3 (dimensionless constant)

RESULTS
The predicted position of drifted cards for the computation without tidal current was significantly
different from the observed data. The model showed that oil slick would move northward only. In the
observation, however, drift cards displaced toward a southwest direction during the first 3 hours of the
observation. The wind data revealed that the difference between measured and computed positions
especially during the first three hours were influenced by the background current. The magnitude of wind
speed at 12 m/s with direction 195 degrees from north could generate Stokes drift at 0.204 m/s in the same
direction of the wind, and Ekman current at 0.168 cm/s at 48 degrees from north. If there was a net current
moving to the observed position of the drifted cards, the magnitude of background current must be higher
than 0.168 m/s in the opposite direction of Ekman current (i.e., at 228 degrees from north).
This diagnosis showed that the difference might not come from background or residual current, since

6D - 6 The Second OMISAR Workshop on Ocean Models

 Hydrodynamic Model for Oil Spill Trajectory Prediction

such strong magnitude of background current is impossible to occur in general. Therefore, the difference in
the position of drifted cards should come from the instantaneous or tidal current that was not included in the
computation.
The predicted position of drifted cards by the model including tidal current showed better
correspondence with the observed data. The model predicted that during the first 3 hours, drifted cards
displaced southward (while the actual movement was in the southwest direction). After that, the predicted
and observed positions were very near especially at 15.00 hrs. This agreed well with the tidal current model
where it expected ebbing during 6.00 to 12.00 hrs and flooding started after that. Very strong flooding was
observed at 15.00 hrs which is consistent with the predicted positions of drifted cards.
Error in prediction by the oil spill trajectory model after including the computed tidal current could be
generated for two reasons. Firstly, Ko Sichang was not considered in the both the hydrodynamic and
trajectory models. The shear stress from Ko Sichang could retard the current flow and may change the
current direction. Therefore the oil spill trajectory model should be tested in several areas to confirm this in
the future. Second cause could be the grid size deployed. Although the oil spill trajectory model gives
output at one-minute resolution, it reads current data at much finer resolution (at 0.1 degree or 6 minutes).
It is clear that this error can come from the inaccuracy in the input position data. The error, then, can
propagate along the computational processes.

CONCLUSIONS
Results of the study lead to the following conclusions: The prediction of oil spill in the Gulf of
Thailand will be more accurate if the computed tidal current is included in the model.

REFERENCES
Archevarahuprok, B. and P. Wongwises, 1994. Numerical Simulation of the Monthly Mean Sea Current in the Gulf of
Thailand. In IOC-WESTPAC 3rd International Scientific Symposium, 22-29 November, Bali, Indonesia.
Azamy, A.R., Y. Isoda, and T. Yanaki, 1991. Sea Level Variation due to Wind Around West Malaysia. Memoirs of the
Faculty of Engineering. Ehime University, 12(2): 148-161.
Charuskumchornkul, S., 1988. Oceanographical Circulation in the Upper Gulf of Thailand. Master’s Thesis, the
Asian Institute of Technology, Thailand.
Chokchalermwat, W, 1990. Current Pattern Analysis in the Upper Gulf of Thailand. Master’s Thesis, Chulalongkorn
University, Bangkok, Thailand..
Bunpapong, M., R.O. Reid and E. Whitaker, 1985. An Investigation of Hurricane-Induced Forerunner Surge in the
Gulf of Mexico. Research conducted through Texas A&M Research Foundation Project 4667. Texas A&M
University.
Buranapratheprat, A., 1997. Hydrodynamic Model for Investigation of Oil Spill in the Gulf of Thailand. Master’s
Thesis, Department of Marine Science, Chulalongkorn University, Bangkok, Thailand.
Hang, O., P. Evensen, and E.A. Martinsen, 1989. Oil Models for the South China Sea. Technical Report No. 70. Det
Norslse Meteorologiske Institutt.
Leendertse, Jan J., 1967. Aspects of a Computational Model for Long-Period Water-Wave Propagation.
(Memorandum RM-5294-PR. United Air Force.
Liengchalernsit, W., 1979. Mathematical Models for Hydrodynamic Circulation and Dispersions of Selected Water
Quality Constituents with Applications to the Upper Gulf of Thailand. Doctoral Dissertation, the Asian Institute
of Technology, Thailand.
Siripong, A., 1985. The Hydrography of the South China Sea and the Gulf of Thailand: Wave, Tide and Currents.
United Nations Environment Programme. Unpublished manuscript. 253 pp.
Snidvongs, A. and P. Sojisuporn, 1997. Numerical Simulations of the Net Current in the Gulf of Thailand Under
Different Monsoon Regimes. In the 1st Technical Seminar of the Interdepartment Collaborative Research
Program, 24-26 November, Bangkok, Thailand.
Sojisuporn, P., 1984. Computer Simulation Model for Wind Driven Current in the Upper Gulf of Thailand. Master’s
Thesis, Chulalongkorn University, Thailand.

The Second OMISAR Workshop on Ocean Models 6D - 7

Session 6 on Ocean Transport Models and Marine Ecosystem Models

Sojisuporn, P., 1994. Density-Driven and Wind-Driven Current in the Upper Gulf of Thailand. In IOC-WESTPAC
3rd International Scientific Symposium, 22-26 November, Bali, Indonesia. 374-385.
Sripunyawitchya, S., 1988. Mathematical Model of Wind-Driven Circulation I the Gulf of Thailand. Master’s Thesis,
Chulalongkorn University, Thailand.
Vongvisessomjai, S., A. Arbhabhirama, and Y. Fuh, 1978. A Mathematical Model of Oil Spill Movement: Upper Gulf
of Thailand. Research Report No. 73. The Asian Institute of Technology, Thailand.

6D - 8 The Second OMISAR Workshop on Ocean Models