North Carofina 9tme Library

Raleigh

^4. c

DOC.

Review and Evaluation of

Oil Spill Models

for Application to

North Carolina Waters

M 2 0 im

Bruce J. Muga

Consulting Engineer

Durham, NC

AUGUST 1982

North Carolina

Coastal Energy Impact Program

Office of Coastal Management

North Carolina Department of Natural Resources

and Community Development

CEIP REPORT NO. 31

To order:

Residents of North Carolina may
receive a single copy of a
publication free upon request.
Non-residents may purchase
publications for the prices
listed. Because of the
production costs involved, some
of the publications carry a
minimal charge regardless of
residency. Prices for these are
indicated in the price list as
being "for all requests".

When ordering publications
please provide the publication
number and title and enclose a
check made payable to DNRCD. For
a complete list of CEIP
publications - or to place an
order - contact:

Coastal Energy Impact Program
Office of Coastal Management
N.C. Department of Natural

Resources and Community

Development
Box 27687
Raleigh, NC 27611

Series Edited by James F. Smith
Cover Design by Jill Miller

REVIEW AND EVALUATION OF OIL SPILL MODELS
(for Application to North Carolina Waters)

by

Bruce J. Muga
Consulting Engineer
Durham, North Carolina

Prepared for

Office of Natural Resources Planning and Assessment
N. C. Department of Natural Resources and Community Development

The preparation of this report was funded by a Coastal Energy
Impact Program grant from the North Carolina Coastal Management
Program, through funds provided by the Coastal Zone Management
Act of 1972, as amended, which is administered by the Office of
Coastal Zone Management, National Oceanic and Atmospheric
Administration. The CEIP grant was part of NOAA grant
NA-79-AA-D-CZ097.

August 1982

CEIP Report No. 31

Digitized by the Internet Archive

in 2011 with funding from
State Library of North Carolina

http://www.archive.org/details/reviewevaluationOOmuga

Acknowledgements

This study was funded by a grant from the
Coastal Energy Impact Program (CEIP), Department of
Natural Resources and Community Development (DNRCD) .
In particular, the author wishes to acknowledge the
efforts of Mr. Jim Sm.ith , CEIP Coordinator, Office of
Coastal Management, DNRCD, and Mr. Roger Schecter, Chief,
Permit Information and Assistance Section, Office of
Planning and Assessment, DNRCD. Mr. Smith arranged for
this particular study to be undertaken following
completion of the objectives of an earlier study under
the original grant. Mr. Schecter served as Contract
Monitor and was very helpful in securing certain
references. The author also wishes to acknowledge the
help of Mr. Eric Vernon, Office of Marine Sciences,
Department of Administration, who provided some important
but difficult to obtain references.

Finally, the author appreciates the help of
Ms. Anne Taylor, Director of the Office of Planning and
Assessment, DNRCD. Ms. Taylor completed all of the
necessary but burdensome paperwork to ensure availability
of funds for the original grant objectives.

The encouragement, help, patience and under-
standing of all of the aforementioned personnel is greatly
appreciated.

Executive Summary

A review of the general status of oil spill
modeling is presented. The major processes of importance
to oil spill motion are advection, spreading, dispersion
and evaporation. The relative importance of each of the
processes is depicted on a time-length diagram. This
diagram is also employed to indicate those processes
that are most important to the various North Carolina
waters. The relative importance of these processes, in
effect, dictate the model requirements.

It was found that approximately twenty-five
distinct modeling efforts have been undertaken in recent
years. Summary information for each of these processes
are presented in the Appendix. A number of these models
were found to be suitable for adaptation to the offshore,
nearshore and inland and estuary waters of North Carolina.
For each category, those models that are best suited are
identified. Finally, some recommendations regarding
prediction of oil slick behavior in North Carolina
waters are presented.

CONTENTS

Chapter Page

1 Introduction , , . . 1

2 Background 4

Problem Formulation 7

Spreading and Dispersion

' Model Solution , 11

> 3 Review of Literature 17

[ Advection 18

j Wind 20

Surface Currents 23

r Subsurface Currents 25

\ River Currents 25

I; Waves 26

Spreading and Dispersion 26

Evaporation 32

I-

4 Evaluation of Oil Spill Models

for Use in North Carolina Waters .... 35

Inland Waters and Estuaries 44

Nearshore 46

Offshore 49

; 5 Summary and Conclusions 54

Recommendations 56

References 59

Appendix 60

List of Tables

Table Page

I Processes of Importance to the

Oil Spill and Transport Problem .... 6

II Spreading Laws for Various Regimes
According to Fay (1971) 12

III Oil Spill Models 19

IV Wind Field Model Types 21

V Types of Spreading Models 28

VI Classification of Evaporation

Submodels 33

VII Candidate Models for Application

to North Carolina Waters 43

List of Figures

Figure Page

1 Time-Length Scale Diagram for
Determination of Relative

Importance of Various Processes .... 15

2 Time-Length Scale Diagram for
Evaluating Relative Importance

of Various Processes 38

Chapter 1
Introduction

Recent interest in exploratory drilling for
hydrocarbon deposits off the coastlines of North Carolina
and neighboring states has disclosed the need to under-
stand a number of environmental risks attendant to this
activity. Among the most important of these risks is the
spillage of liquid hydrocarbons. Since risk is the
product of the probability of occurrence of an event
and its effect (should the event occur), the transport
mechanisms of spills is a topic of critical interest.

This topic has been the object of numerous
studies in recent years. These studies have resulted
in the development of various models for predicting the
movement of spills. Each of these models was developed
for a particular set of conditions. The result is that
no reliable prediction scheme has yet evolved which is
applicable to a general set of conditions. It is note-
worthy to mention that develocment of such a scheme is
a formidable task.

Therefore, the main focus of this study is an
evaluation of the methods for predicting the movement
of oil spills under the particular environmental
conditions that exist in North Carolina waters. Although

the idea for this study stems from the potential drilling
activity offshore, our interest in models for predicting
spill movements is not limited to these occurrences.
To a parallel degree, we are interested in methods for
predicting spills that occur in connection with the
transportation and transfer of hydrocarbons within and
adjacent to North Carolina waters. For example, methods
for predicting the movement of spills that occur in ports,
harbors and/or estuaries as well as along the adjacent
transportation routes of North Carolina, are also of
concern .

In response to the various needs, oil spill
models have evolved along two separate paths. The earliest
models were developed to predict one-of-a-kind events
occurring in real time. These are after-the-fact models
and depend upon reliable wind and current forecasts for
achieving even moderate success. Most frequently, these
models were roughly calibrated from real observations
taken during and following a spill. Such models, of
necessity, are program.med to accept real time data as
input. With the passage of time, other model types were
developed in connection with the evaluation of various
impacts. These models are much more sophisticated than
the earlier models and are designed to assist decision
makers in the environmental assessment and planning

processes. These models use the historical climatological
data as input and are programmed to generate sufficient
data to disclose all possible consequences of an event.
For other fundamental reasons, the former and
latter models tendea to be deterministic and probabalistic
in nature, respectively. This characteristic will be
examined in more detail in the following section.
Bishop (1979) has categorized the model types according
to Type I, Type II or Type III, which he describes as
follows:

"Type I models; Multiple trajectory models
for long-term strategic forecasts based on
archived data.

Type II models: Single event (highly structural)
models for specific day-to-day tactical forecasts
usually based on up-to-date data, and
Type III models: Type I or Type II models
implemented in a receptor (reverse) mode such
that one can project areas from which trajectories
would impact resources."

We see that Type III models are not really a
separate category but rather a unique application and use
of either Type I or Type II models. It is also interesting
to observe that the sequence of presentation is actually
the reverse of the historical development. In other words,
Type II models evolved and preceeded the development of
Type I models.

Chapter 2
Background

There is a substantial volume of literature
dealing with oil spills, most of it of recent origin.
Surprisingly, there have been only a few attempts to
review the literature in a systematic manner within a
comprehensive framework of the overall problem. The
work by Stolzenbach, et . al . (1977) is an important
reference in this regard. Most of the studies as
published in the open literature treat only one aspect
of the overall problem in an isolated context and often
under a narrow combination of circumstances that make it
difficult to apply the results to other situations with
reasonable confidence.

It is instructive to make a few preliminary
observations in connection with the treatment of the
oil spill problem. This is a general problem in the
field of fluid mechanics, the solution for which requires
a broad understanding of chemical and biological processes
The solution can be approached in one of two fundamental
ways; i.e., probabalistic or deterministic.

Because of the fact that (1) the environment
(i.e., winds, waves and currents) has an enormous
influence on the spill behavior, and (2) the environment

4

is a geophysical random process, one would expect that
the approach should be a probabalistic one. In other
words, the solution approach should be compatible with
the nature of the problem. On the other hand, there are
some features of the problem which are decidedly deter-
ministic. An example would be the sprea.ding of oil on a
quiescent sheltered body of water in the absence of any
winds, waves or currents. Also, the approaches are not
mutually exclusive since elements of deterministic methods
can be and are incorporated within probabalistic-based
approaches and vice-versa. In reviewing the literature,
it is important to recognize the existence of these two
fundamental solution methods.

In view of the realization that the probabalistic
nature of the problem derives from consideration of the
environmental influences (which have regional 'geographic'
dependency) and that the deterministic influence is
based on the physical mechanics of the problem, we can
identify two broad categories of processes which are of
importance to the oil spill and transport problem. These
processes are depicted in Table I , with some explanatory
notation .

The spreading and dispersion processes are, to
a large degree, geographic-independent, but are, neverthe-
less, dependent upon certain characteristics of the event

TABLE I

Processes of Importance to the Oil Spill
and Transport Problem

Process

Nature

Dependency

Spreading

Deterministic

Oil properties at ambient
temperature, largely site
dependent

Dispersion

Deterministic
(with probabalistic
elements)

Turbulence intensities in
water medium, weakly site
dependent

Advection (Drift)
(wind, currents, waves)

Probabalistic
(with deterministic
elements)

Strength of advecting
fluids, strongly site
dependent

Weathering and Other
Processes

biodegradation

dissolution

emulsif ication

evaporation

oxidation (photo)

patch (large-size)
breakup

sinking and
sedimentation

Probabalistic
and Deterministic

Various site dependent
and site independent
conditions

such as spill size, rate of spill and other physical and
chemical characteristics which describe the spill material
and receiving waters. In contrast, the advection (drift),
weathering and other processes are directly related to
and dependent upon conditions that characterize the site
environment at the time of the spill and immediately
thereafter. In other words, the processes which characterize
oil spill movement consist of both site-dependent and
site-independent ones. A brief review of the various
processes governing the mechanics of the problem is
presented in the following section.

Problem Formulation. Stolzenbach, et.al.(1977)
have formulated the essential boundary value problem for
oil slick transformations. The system of differential
equations were derived on the basis of some simplifying
assumptions and averaging approximations. The simplifying
assumptions seem to be well justified and are used to
some degree in all of the models that have been developed
and reported in the literature. Therefore, any error due
to the simplifying assumptions would not affect any
relative evaluation of the various models. With regard
to the averaging approximations, Stolzenbach, et.al. (1977)
states that it is justified "only for the analysis of the
oil motion and may not be a good approximation for studying
the diffusion of oil through the slick boundaries."

The continuity equation which expresses the
conservation of mass relationship appears as

9c. h 3uhc . 9vhc . „ „— „ .—

^J- + -^+-^= k. [/ (h|^) +^ (h|^)] -4, . -d,^. - R.h (1)
9t 3x 9y 1 9x 9x 9y 9y si bi i

The conservation of momentum relationship appears as

9u , — 9u , ~ 9u , w ^ ^ 9h bx , sx ,^^

^ ^ " ^ ^ ^ 97 = "^ ^T~^ 9^ - ^ ^ ^ (2a)

9v , — 9v , — 9u /__w_. 9h by , sy ,„, ^

The boundary condition on the slick boundary appears as

f=a-a-a (3)

n aw oa ow

In the above equations,

x,y = variables of moving coordinate system
with respect to origin at center of
mass of oil slick

u,v = components of oil particle velocities
relative to the center of mass of the
slick

C. = mass per unit volume of oil of the
ith oil fraction

h = local slick thickness

k. = molecular diffusion rate for ith oil
fraction within the slick

R. = rate of removal of the ith oil fraction
per unit volume

d) . = flux of the ith oil fraction outward
si

through the surface of the slick

4), . = flux of the ith oil fraction outward
through the bottom of the slick

p = average density of oil mass within the slick

p = density of water underlying the oil slick

t = time

T ,T = shear stresses on oil slick surface
sx ' sy

T, ,T, = shear stresses on oil slick bottom
bx ' by

g = acceleration due to gravity

f = net surface tension per unit length
of oil slick boundary

a ,a ,0 = surface tension of air-water, air-oil
aw' oa ow j -n ^ ■ ^ ^ -^ ■ ■^

and oil water interfaces, respectively

Equations (1) through (3) constitute, in a
simplified form, the boundary value problem of oil slick
transformations. In relating the physical processes
shown in Table 1 tc the various terms in the system of
equations, we can point out that the effects of weathering
(and other processes) are treated indirectly via the
concentration term C. for the different oil components
and directly by the last three terms of Equation (1).

9

We can also note that the motions (i.e., particle velocities
u and v) are given relative to the center of mass of oil in
the slick. Therefore, motions resulting from the advection
processes (appropriate for the center 6f mass of the slick)
must be superposed on the relative velocities to determine
the motions in an absolute sense.

In a preliminary review o:*" the literature, we
found that models for the spreading, dispersion and
advection processes have been developed and compared
with model and prototype data to a sufficient degree to
enable some reasonable conclusions and evaluations to be
made. Except for evaporation, this is not the case for
other weathering processes which remain incompletely
understood. Moreover, the interactions between weather-
ing and the other processes is even more poorly understood.
As a consequence, the scope of this study is limited to
an assessment of the spreading, dispersion and advection
models and their applicability to North Carolina waters.
In addition, some comments concerning evaporation models
and models of other weathering processes will be provided
but it is beyond the scope of the present study to consider
such models on the same level with spreading, dispersion
and advection models,

10

Spreading and Dispersion Model Solutions.
Returning to the fundamental system of equations, different
solutions can be obtained depending on the type of additional
simplifications that can be made. As indicated by
Stolzenbach, et . al . (1977) two broad categories of solutions
can be identified. These are the spreading and dispersion
solutions. The distinction between these two processes
(and solutions) depends upon the relative strength of the
external shear stresses acting on the oil slick. If the
external shear forces (i.e., the last two terms of
Equations (2)) are absent, or negligible, then the solution
corresponds to the spreading process since the resulting
motion depends upon the density differences and inter-
facial tensions. On the other hand, if the external
shear forces are very large, then the solution corresponds
to the dispersion process.

Most, if not all, of the spreading solutions
appear to be based on or related in some way to the work
of Fay (1969, 1971). Two sets of results have been
obtained which are based on certain idealistic assumptions.
The one-dimensional set of results is appropriate when it
is assumed that the slick spreads in one direction only.
The two-dimensional case is based on radial spreading.
A summary of the spreading functions for the two different
sets of results is given in Table II. These functions

11

TABLE II

Spreading Laws for Various Regimes
According to Fay (1971)

Regime
(Driving/Resisting Force)

Spreading Formula
One-Dimensional Radial

Gravitational/Inertial

L = k„

D = 2k

rP -P

iL

w

Gravitational/ Viscous

L = k.

p.rp - 1

(-^)AtVv^
w

H

D = 2k

• P -p 2" T 6

(-J^) V^t /v^

Surface Tension/Viscous

L = k„ ia^tVp 2 V

X.3 |_ W

D = 2k

3 I

a^tVp'v
w

w

where P
P
g

D
A
V
V

a

t

= density of water

= density of oil

= acceleration of gravity

= length of one-dimensional oil spill from fixed origin
to leading edge of oil slick solubility

= diameter of radial (axisymmetric) spill

= volume of oil per unit length normal to spill direction

= volume of oil

= kinematic viscosity of water

= interfacial tension

= elapsed time from initial spill

From various sources, the coefficients k may have the following values:

kij = 1.33

k-- = 1.14

_/0.98"l

■^r, -Sl.l2(

I1.45J

k^ =1.60

12

are presented with the unknown length (distance from
leading edge to center of gravity of spill) in an
explicit formulation. The other known or assumed
variables appear on the right-hand side (RHS) of the
equation .

The dispersion models result in the following
solution form: i

12 2

jn ^^^ I ^ y

X y

X

where

h = average spill thickness
m = total slick mass
p = density of oil

and a and a are related to the dispersion coefficients,

X y

respectively, as follows

9a 2 3a ^

^ = 2k and — ^— = 2k

at X 9t y

As noted by Stolzenbach, et . al . (1977) observations
of real oil slicks rarely fit the assumptions made in
formulating the spreading /dispersion models. Thus, compari-
sons between observations and predictions tend to be mixed.
As a result of this behavior, Stolzenbach, et . al . (1977)

13

conclude that

" 1. Both spreading and dispersion processes

may be important in determining the total
growth of the slick,

2. Existing techniques for estimating the
growth of surface oil slicks provide at
best only an order of magnitude estimate
of what the actual slick size will be.

3. Because of the complex and usually random
nature of the processes controlling slick
growth, it is unlikely that a significant
improvement in deterministic capability
will be possible. However, estimates of
the variance in slick sizes should become
more accurate as additional observations
are obtained.

4. The applicability of available spreading
and dispersion models should be judged
on a case by case basis in terms of the
site specific conditions."

The usefulness of the modeling work as well as
Stolzenbach ' s , et . al . (1977) review lies in their develop-
ment of a time and length scale diagram, as shown in
Figure 1 . Such a diagram enables one to determine the
relative importance of the various processes which affect

14

CO

•^ 0)

tfl

o

<u

CO

"e

/ "

^ .-H

■

CO

a;

•H

o

•U

C

CO

QJ

•H

0

cu

CO

o

t

•H

CO

/^

o

1

4-1

CO

iH

1

Cfl

ai

CJ

0)

1

c

o

r— 1

>

0)

i

1

•H

e

0

V4

P-.

o

>

1

(U

a>

•H

1

4-1

CO -^

CO

J-)

1

Q)

3 r^

/^

O

1

Q

O r^

f e

0)

/

•H CTn

o

>

/

!-i

^ .H

o

I — 1

TJ

/

O

CO

<

/

M-i

> '

/

CO

a

CO

/

c

e

m .-H

i

o

(0

0 CO

1

•H

!-i

/

u

4-1
Cfl

CO

CD 4J
O 0)

/

<u

e

3

•H

C

I— I

/

X.

cu

4-1

Q

CO *^

/

■u

4-1

o

4J ^

/

CO

CO

3

0)

!-( O

/

(U

>,^

i-H

O CO

i

L

S

CO

f^

CO
O

""

N

J=

0) I— 1

4-1

> 0

M

•H 4-1

C

4-1 C/^

0)

CO

I

tH 1-1

0)

Pi 4-1

e

U-l

•H

iw <;

H

O^-'

1-1

W

PS

t3

o

M'

O
O

15

oil spill motions. This diagram clearly shows that
spreading is important during the early stages (or,
alternatively, over shorter distances) of a spill
whereas the opposite is true of dispersion. Other
processes are also indicated. An example of the use
of such a diagram in the case of selecting appropriate
models for North Carolina waters will be indicated in
Chapter 4.

16

Chapter 3
Review of Literature

As noted in the introduction, there has been an
enormous volume of literature published within recent
years on the subject of oil spills. Approximately
200 references constituting roughly 15% of the available
number of articles have been consulted in connection with
this report. Only a small number of these references
are cited in the bibliography, however, since many
references deal with the topic in only a peripheral way.
From these references, about 2 5 distinct modeling efforts
have been identified. For each of these efforts, a summary
report has been prepared identified by the model name and
citing the primary reference and accompanying abstract
along with the reviewers comments. These summary reports,
one for each modeling effort are presented in the Appendix.
In this section, a summary of the reviews is presented so
that the reader can gain a panoramic view of the current
status of oil spill models, To reiterate, our main
emphasis is concerned with advection, spreading, and
dispersion processes and with evaporation. For reasons
discussed in the introduction, less emphasis is placed on
other weathering processes. Also, we are concerned with
application of these models to North Carolina waters.

17

For that reason, a number of Canadian sources were not
consulted since, in many cases, they deal with Artie
conditions.

A satisfactory means of classifying models is
to take note of the processes that the model considers.
This has been done by Stolzenbach, et . al . (1977) and is
presented herein as an update in Table III. Discussions
are presented in the following paragraphs for each of the
major processes that are of interest to this study.

Advection. Advection is probably the most
important process by which oil spills are transported
over any sensible distance, and is treated by every oil
spill model that was reviewed. Advection occurs as a
result of wind and currents and to an unknown extent ,
waves. Both wind and currents induce shearing stresses
on the top and bottom surfaces, respectively, of the oil
slick. These shearing stresses, in turn, cause the oil
slick to move. Because of the fact that the shearing
stresses are not, in general, uniform (or constant), the
oil slick experiences not only a net motion of its center
of mass but also differential motions of the oil slick
geometry about its center of mass. As a matter of fact,
the shearing stresses are, in the general case, time and
space dependent .

18

TABLE III

OIL SPILL MODELS
(Listed According to Date of Origination)

Processes Modeled

Advection
Advection
Advection
Advection
Advection
Advection
Advection

Model Name

Date

1.

U.S. Navy

1970

2.

WGD

1972

3.

Narragansett Bay

1973

4.

Tetra Tech , Inc.

1974

5.

CEQ

1974

6.

BOSTM

1975

7.

USCG

(New York Bight)

1975

8.

Delaware Bay

1976

9.

USCG

(New York Harbor)

10.

use/API

1977

11.

SLIKTRAK
(SLIKFORCAST,
OILSIM)

1977

12.

MOST

1978

13.

DPPO (Garver
and Williams)

1978

14.

URI

(Georges Bank)

1979

15.

Puget Sound

1979

16.

CAES

1979

17.

Riverspill

1979

18.

NWS/NOAA

1979

19.

USCG (Long
Island Sound)

1980

20.

SPILSIM

1980

21.

EDIS

1980

22.

PIC

1980

23.

OSTA

1980

24.

OSSM

1980

25.

DRIFT

1980

Spreading

Spreading, Dispersion

Spreading, Dispersion
Dispersion

Advection, Spreading, Dispersion
Advection, Spreading, Dispersion

Advection, Spreading, Dispersion,

Evaporation
Advection, Spreading, Dispersion,

Evaporation

Advection, Spreading, Dispersion
Advection, Spreading, Dispersion,

Evaporation
Advection, Spreading, Dispersion,

Evaporation
Advection, Spreading
Advection, Spreading, Dispersion,

Evaporation
Advection, Spreading
Advection, Spreading
Advection, Spreading, Dispersion

Advection
Advection
Advection
Advection
Advection
Advection

Spreading

Spreading, Dispersion

Spreading, Dispersion

Spreading, Dispersion,

Evaporation

Weathering processes other than evaporation
are not listed.

19

Wind. A substantial effort has been devoted to the
development of wind field and current submodels. One
of the distinguishing features of oil spill models is
the degree of realism that has been incorporated into
their wind and current submodels. In the case of wind
models, this degree ranges from the simplest and crudest
approach to the most elegant and sophisticated. At least
three different categories can be identified. These
include constant wind fields, time and spatially dependent
wind fields.

Table IV is a listing of the types of wind field
input that is programmed for oil spill models that were
reviewed. A majority of the models can be operated to
accept either measured (or real time data) or simulated
data. The measured data is, in general, time and spatially
dependent but this may not be a meaningful characteristic
if the grid size is large or if the elapsed time between
observations is long both relative to the volume of the
spill. The simulated data input capability is useful for
conducting contingency planning environmental assessment
studies .

As noted above, the simulated wind field submodels
can be highly variable in complexity. However, the degree
of complexity is not an assurance of suitability for a
particular site dependent oil spill. Even a simple

20

TABLE IV

WIND FIELD MODEL TYPES

Oil Spill
Model Name

Simulated

Other Measured

Constant Time Dependent Spatially!
Random Markov Dependent
Walk Chain

1.

U.S. Navy

2.

WGD

3.

Narragansett

Bay

4.

Tetra Tech, Inc.

5.

CEQ

6.

BOSTM

7.

USCG

(New York Bight)

8.

Delaware Bay

9.

USCG (New York

Harbor)

10.

use/API

11.

SLIKTRAK

(SLIKFORCAST,

OILSIM)

12.

MOST

13.

DPPO (Garver

and Williams)

14.

URI

(Georges Bank)

15.

Puget Sound

16.

CAES

17.

Riverspill

18.

NWS/NOAA

19.

USCG (Long

Island Sound)

20.

SPILSIM

21.

EDIS

22.

PIC

23.

OSTA

24.

OSSM

25.

DRIFT

X

X
X

X

X

X

X

X
X
X

X
X
X

X
X

X
X

X
X

X
X
X
X

X
X

X

X

21

constant wind field under a given set of circumstances
might be quite proper for predicting advection. Time-
dependent submodels consists of "random-walk" models,
for which there is no correlation between successive time
steps, and autoregressive models (Markov chain) for which
there is limited correlation in time between successive
time steps. Most natural processes such as wind are
continuous functions of time, thus, Markov chain models
appear to be the superior type of time-dependent simulation,
However, wind is also a spatially dependent function. A
number of models have been developed or are in the process
of development which simulate the w: nd on the basis of
its spatial dependence. These models are the most general
and represent the most sophisticated of all types of wind
simulation models. Finally, a few models employ wind
field models that do not fit any of the above categories.
They are listed under "other" on Table IV. In some cases,
there was insufficient information to categorize the wind
field submodel.

In nearly all cases (the CAES model being the
only exception) the determination of wind drift currents
follows the wind factor approach. In this approach, the
wind-induced surface current is assumed to be a fixed
percentage of the wind velocity taken at a height of
10 meters above the sea surface. Often a value of 3% is

22

assumed but this value can range from 2.0% to 4%. These
values have been estimated from numerous observations and
experiments. In at least two references, Madsen (1977)
and Huang (1979), show theoretically that this factor
should be slightly in excess of 3% for deep water. The
angle between the wind vector and the wind induced current
vector is known as the wind drift angle. This angle has
been taken to vary from 0 to approximately 20 depending
on latitudinal influences. Huang (1979) also shows that
the wind drift angle can vary from 0 to a maximum (which
depends upon the latitude) depending on the duration that
the winds have been blowing. This angle also depends
on the water depth and approach a value of 0 for very
shallow water.

Surface Currents. Although wind-induced drift currents
are important mechanisms in the transport of oil slicks,
they are not the only ones of importance. Therefore, we
must examine the role of residual and tidal currents as
well as waves. The residual currents include those
circulations due to thermohaline and geostrophic causes.
In general, the residual currents are most important for
open ocean spills in deep water whereas the tidal currents
are most important for near inshore and estuary (tidal)
spills in shallow confined waters. In many models, both
residual and tidal currents are considered together and

23

a procedure similar to the wind factor approach is employed
to estimate the surface drift current. The factor most
often used has a value of 56%. Many models are structured
to utilize information on currents as obtained from
prototype measurements or from estimates obtained from
circulation submodels. The circulation submodels are
often major components in the oil spill transport prediction
procedures.

The numerical circulation submodels are not true
three-dimensional models; rather they are two-dimensional
representations obtained by making certain simplifications.
Among others, these simplifications include the averaging
(in the vertical) of the component velocities. This
averaging presents some difficulty in estimating the
surface currents especially in regions where the wind and
tide are the dominant contributors to the circulation
pattern. Another difficulty is related to the model
boundaries especially where there is an interface with
major large scale circulations systems such as the
Norwegian current or the Gulf Stream. With the exception
of SLIKTRAK, none of the models attempt to model the
large scale circulation pattern. SLIKTRAK attempts to
model the effect of the Norwegian current by providing
for acceptance of predicted values of velocities
(presumably obtained from another submodel) at the model
boundaries .

24

The state-of-the-art consists of calibrating
the numerical circulation submodel with available field
measurements. The resulting currents predicted from
such a model are then combined with wind-induced drift to
estimate the advection.

Subsurface Currents. Because of the fact that oil
spills generally appear on the surface, little attention
has been directed to the advection by subsurface currents.
It is generally believed that subsurface advection is not
an important mechanism in oil slick transport. However,
it is known that advection of dispersed oil droplets occurs
as a result of subsurface currents. Two models include
this behavioral feature. One is the DPPO model (Garver
and Williams, 1978) and the other is the URI (Georges Bank)
model .

River Currents. Even though a large number of spills
occur in river estuaries or harbors where there is
substantial fresh water inflow as well as other currents,
this aspect of oil spill transport has received little
attention. One model, Riverspill, was developed specifi-
cally for river-type spills (Mississippi River) .
Considerable effort has been directed toward determination
of advection by river currents. Another model, USCG
(New York Harbor) has also been developed to consider
the fresh water inflows from the Hudson River. Both

25

models ha.ve been calibrated by comparison with numerous
field observations.

Waves . Advection by waves directly or by currents
generated by waves is one of the least understood aspects
of oil spill transport. Some authors believe that it is
an important mechanism but only three models (SLIKTRAK,
Delaware Bay and USC/API) have recognized the possibility
of considering its effect.

In summary, advection is one of the most
important processes by which oil spills are transported.
The prediction of advection resolves into a determination
of the wind and current fields. From a knowledge of the
wind and current fields, a n\M:iber of numerical techniques
have evolved from which advection of oil spills can be
made. The role of subsurface currents is probably not
important in transport of large oil slicks but very
important in advection of oil droplets. The contribution
of waves to the advection process is very poorly understood.

Spreading and Dispersion. We consider
spreading and dispersion together since these mechanisms
are responsible for the growth of an oil slick after the
initial spill. During the initial stages, spreading is

Not to be confused with emulsif ication or oil-in-water
formulation of droplets.

26

clearly the most important of these mechanisms. Spreading
is dependent unon the oil properties at the time of the
spill whereas dispersion depends upon the external shear
forces acting on the oil spill surfaces. A number of
investigators (e.g., Stolzenbach, et . al . , 1977) have shown
that, for a given spill volume, after a certain time, the
dispersion process becomes more important than spreading.
There are essentially three distinct approaches
that are employed in the treatment of spreading and
dispersion. Table V presents a listing of the spreading
submodels that have been employed by the oil spill models
that were available for review. Most of the models that
treat spreading do so by using a method that is related
in some way to Fay's (1971) theoretical analysis of
spreading. Some models combine Fay's spreading analysis
with a diffusion approach. Other models consider the
growth of oil slicks to be dominated by diffusion aspects
and ignore altogether the role of the oil properties
such as surface tension. For example, the Delaware Bay
model substitutes a diffusion equation in lieu of the
third stage (viscous-surface tension) of Fay's (1971)
three-regime spreading model. As another example, the
use/API model eliminates the second stage (gravity-viscous)
of Fay's (1971) model altogether. As mentioned above,
a few models treat ; he growth of oil slicks solely by

27

TABLE V

TYPES OF SPREADING MODELS

Oil Spill
Model Name

Fay's

Spreading

Theory

Fay's
Spreading

with
Diffusion

Diffusion
Theory

Independent

Statistical

or Numerical

Treatment

Narragansett Bay
Tetra Tech, Inc.
BCSTM
USCG

(New York Bight)
Delaware Bay
USCG

(New York Harbor)
use/API
SLIKTRAK

(SLIKFORCAST,
OILSIM)
MOST
DPPO

(Carver and Williams)
URI

(Georges Bank)
Puget Sound
CAES

Riverspill
NWS/NOAA
USCG (Long

Island Sound)
EDIS
PIC
OSSM
DRIFT

X

X
X

X

7

X
X

X
X

X

X

28

means of a diffusion theory such as that developed by
Murray (1972) or a random Fickian diffusion model.
The rationale for this is based on observations of oil
spills that are continuously discharging (i.e. , versus
instantaneous spills) wherein it was concluded
(Murray, et.al. 1972) that the growth a short distance
from the source, was controlled by the horizontal eddy
turbulence rather than surface tension effects. For
example, the USCG models use Murray's turbulent diffusion
theory while DRIFT uses a random Fickian diffusion
approach .

One approach which is decidedly deterministic
is the NWS/NOAA model wherein a numerical solution of
the equations of motion (oil) is envisioned. At the
opposite end of the philosophical spectrum is the
Pudget Sound model which is best described as an indepen-
dent statistical (probabalistic) model. It is based on
the use of a system of empirical regression equations.
In this connection, all of the models that treat spreading-
dispersion with a diffusion theory in any way require
some evaluation of a set of diffusion/dispersion
coefficients. In general, these coefficients must be
obtained empirically or from published values which
have been obtained from experiments. In some cases,
where there are actual observations of oil spills, the

29

diffusion coefficients have been employed as calibration
coefficients. This is illustrated by the USCG (New York
Harbor) model.

From a theoretical viewpoint, those models which
treat spreading-dispersion by use of Fay's spreading theory
in combination with a diffusion approach are probably
superior to the other approaches. However, all of these
models do not treat spreading-dispersion in the same way
because there are some subtle, minor differences in their
procedures which may have profound consequences in the
results. As a general observation, none of the models
contain any provision for eliminating or evaluating the
numerical errors that may be present. This is an important
consideration in the treatment or diffusion/dispersion
because of the possible contamination of the real
physical dispersion with numerical dispersion. The latter
is merely an artifact of the computational scheme whereas
the former is real. This is clearly illustrated in the
case of the Tetra Tech, Inc. , model about which
Stolzenbach , et . al . (1977) has commented in some detail.

One interesting treatment of spreading-dispersion
is the Delaware Bay model in which the third phase of Fay's
three-regime spreading model is replaced by a compatible
diffusion model. The justification for making this
substitution stems from the observation that whereas the

30

third phase exists in laboratory situations (where the
background turbulence levels are low) , no such conditions
exist in the real world. Therefore, in the prototype case,
the higher turbulence levels require the existence of
a diffusion approach. This justification seems very
rational and plausible and is consistent with Murray's
et.al. (1972) observations and conclusions.

In summary, spreading-dispersion is an important
mechanism for the transformation and/or growth of oil
slicks. In the early stages, spreading is the dominant
process whereas after a critical time (for a given volume)
dispersion seems to predominate. Models which seem best
suited for a wide variety of time-dependent situations
are those which combine Fay's spreading theory with a
diffusion approach. Two extreme cases can be noted.
In confined nearshore areas, where the time and distance
scales are relatively short , a model which employs only
Fay's spreading approach would be quite valid. In a
similar way for regions where the time and distance
scales are relatively long, such as far offshore, a model
which employs only a diffusion theory would be valid.
In the case of the latter, an independent statistical
or numerical approach might also be considered. In
any of the diffusion-statistical- or numerical-based
submodels, the degree of numerical diffusion (or dispersion)
should be determined.

31

Evaporation. Although evaporation is known
to be an important factor in the f £,te of oil spills —
especially in the case of spills far offshore, it has
received less attention from modelers than the other
processes. Only six (6) of the models reviewed include
an evaporation submodel. Moreover, because evaporation
usually occurs in the presence of other weathering processes
such as photo-oxidation, dissolution, biodegradation and
sinking and sedimentation, it is difficult to make ..
reasonable evaluations concerning the appropriateness
of the evaporation submodels. Of the oil spill models
reviewed, there seem to be no essential differences in
the manner in which evaporation was estimated. In
general, an evaporation function or decay rate is derived
from the main factors which affect the evaporation.
These include: (1) the vapor pressures of the various
oil fractions, (2) the size of the spill (area and
thickness), (3) the evaporative mass transfer coefficient,
and (4) the environmental (climatic) conditions, including
wind speed.

All of the submodels employed a mass transfer
coefficient but some did not consider the spill area or
slick thickness in their evaluations. These are indicated
in Table VI. Also, the basis for the submodel can be
categorized as being empirical ; experimental- or
theoretically-related. These are also indicated in Table VI ^

32

TABLE VI

CLASSIFICATION OF EVAPORATION SUBMODELS

Oil Spill

Model
Reviewed

Type

Area
Considered

Slick
Thickness
Considered

use/API

SLIKTRAK (includes
SLIKFORCAST AND
OILSIM)

Theoretical
Experimental

X

DPPO (Carver

Empirical

-

and Williams)

URI

Empirical

X

(Georges Bank)

CAES

Experimental

X

DRIFT

Empirical

X

33

In summary, only a few models attempt to
evaluate evaporation in spite of its importance to the
ultimate fate of oil spills. Unfortunately, there is
insufficient data to permit a rigorous evaluation of
the various approaches .

34

Chapter 4

Evaluation of Oil Spill Models
for Use in North Carolina Waters

In order to select a suitable model or family of
models for predicting oil spill behavior in waters located
within and adjacent to North Carolina, it is necessary to
identify the physical characteristics (and associated
mechanisms) that (1) govern this behavior and (2) serve
as adequate descriptors of the referred-to waters. The
analysis carried out by Stolzenbach, et.al. (1977) and
summarized in Chapter 2 , provides a convenient means of
identifying these features, but it is not the only means.
An order of magnitude analysis of the governing equations
could also be carried out . But , it is very convenient
to use the work of Stolzenbach. Moreover, it provides a
clear visual representation of the conceptual framework.

Therefore, we can utilize Figure 1 which indicates
the major mechanisms of importance to oil spill movement
in terms of time and length scales. We can immediately
realize that two additional types of information are
needed to complete our analysis. One is a parameter that
describes North Carolina waters. For any given spill
locality, the parameter that seems natural is the
minimum distance from the spill source to shore

35

(i.e. , x-number of meters). This can be used to describe
localities in inland waters (in the Cape Fear River,
for example) or localities nearshore or localities far
offshore (an oil drilling platform, for example). For
estuaries, bays, inland waters and other confined or
semi-confined waters, these distances can be scaled from
any hydrographic chart. We can assume that a value of
from 0 to 500 meters would be a reasonable value to use
to describe these waters. To distinguish between near-
shore and offshore areas , we can adopt the arbitrary
criteria that offshore areas are those beyond which
shoreline impacts from tidal currents are virtually
non-existent. On this basis, we can make the following
descriptions for North Carolina waters in terms of
single-length parameter:

Inland waters and

estuaries 0 to 500 meters

Nearshore areas . . 0 to 1000 or 10,000 meters

Offshore areas .... >1000 or 10,000 meters

These values are not to be construed as actual
distances of any specific locality but are generally
representative of dista,nces that characterize the indicated
regions for the purpose outlined above.

The other source of information that is needed
is derived from climatic data for the indicated regions.
Again, precise values are not needed; only an order of

36

magnitude is required. For this purpose, we can assign
the following values of advection velocities (in the
direction of the distances indicated above) for the
three regions.

Inland waters and

estuaries 0.1 /sec - 1.0 /sec

Nearshore areas 0.01 /sec - 0.1 /sec

Offshore areas 0.001 /sec - 0.01 /sec

More precise values can be obtained from detailed
examination of the climatic data (wind and currents).
The above advective velocities are related only
indirectly to the actual wind-induced surface currents
and/or tidal or residual currents since we are interested
only in that component taken parallel to and along the
minimum trajectory path from spill source to impact
location .

These features can be illustrated by referring
to the length-time scale diagram shown in Figure 2. First,
consider a spill in an estuary or inland waterway where
the characteristic length is 500 meters and the
characteristic advection rate lies somewhere between
1.0 and 0.1 meters per second. We can see from Figure 2
that the total elapsed time from spill to impact is rather
short ranging from 425 to 4100 seconds. Figure 2 shows that
dispersion can be neglected and we also note (from other
information) that evaporation occurs along with other processes

37

4-1

a

CO

a.

Cfl

0)

o

o

•H

0)

6

•H
4J

■u
o

>

■V

n)

0)

cn

a

(0

CO

P
o

•H

rH

>

o

S-i

o

(U

o

cn

p!

0)

3

H

■u

+J

H /-s

■H

O r-

O

D-r--

o /

e a>

CD •.

> •

4J to

CO •

1-1 4J

a) cu

bO x:

c

o

•r-l

CO

■P

,0

CO

C

3

0)

>H

N

CO

r— 1

>

o

W

4J

w

M

o

)-l

<+-<

QJ

E u-i
CO <
u ^-^
■M
CO

•H cn

O cu

cn

0) cn

1—1 (U
CO O

o

CO

4-1

o

u

P4

CO
CiO 3
C O
0) -H

I CO
0) >

e

•H >4-l

H O

o

38

throughout the growth and transport of the slick.
However, evaporation and the other weathering processes
are still in their early stages of development by the
time of impact.

Therefore, modeling of oil spill behavior in
estuaries and inland waters should include the simulation
of advection, spreading and evaporation. From a knowledge
of additional details concerning the spill, such as spill
volume, and properties of the spill material as well as
detailed climatic conditions, we can evaluate the
relative importance of the growth (i.e. , spreading) or
transport (i.e., advection) activities in causing impacts.
For example, as a general trend for fixed advection rates,
the spreading activity increases in importance with
increases in spill volumes.

Finally, consider an offshore spill at a
distance greater than 10,000 meters from an impact
location and with characteristic advection rates
less than 0.01 meters per second. On this basis we
can determine that the impact time is greater than
830,000 seconds or 230 hours (almost 10 days). We
take note of the fact that dispersion is the dominant
process in the growth of the slick and that spreading
has been completed. Evaporation has been completed.
In some cases, advection may be an important mechanism

39

but advection by tidal currents is unimportant by definition.
Evaporation has been completed.

Therefore, modeling of spills in the offshore
region should include advection, dispersion and evaporation.
Dispersion is very important but advection by tidal currents,
in general, can be neglected.

Next, consider a spill in the nearshore region
(as defined above) such that the characteristic length has
a value of between 1000 and 10,000 meters and characteristic
advection rates of between 0.01 and 0.1 meters. Again,
using Figure 2 , we can determine that the elapsed time
from initial spill to impact varies from 7700 to 83,000
seconds for the 1000 meter distance, and from 83,000
to 830,000 seconds for the 10,000 meter distance.

In the first instance where the distance from
the spill source to impact is 1000 meters, we see that
spreading is almost complete and dispersion has not yet
become effective whereas in the second instance (where
the source-impact distance is 10,000 meters), spreading
has been essentially completed and dispersion has become
an important activity —remember that scales are logarithmic.
In spite of the apparent low advection rates, the transport
process (including advection by tidal currents) plays an
important role in determining the time of impact.
Evaporation has been active throughout and, in the case of

40

the longer impact times, essentially has been completed.
Other weathering processes have become significant.

From these observations, then, we may conclude
that modeling of oil spills in the nearshore region
should include provision for simulating advection,
spreading, dispersion and evaporation. Advection by
tidal currents should definitely be included in the
modeling scheme.

From the foregoing, we can list the modeling
requirements for the various regions.

Inland waters and estuaries Advection , spreading ,

evaporation

Nearshore areas Advection , spreading,

dispersion, evaporation

Offshore areas Advection, dispersion ,

evaporation

As stated in the Introduction, one of the goals
of this study is to evaluate and select models for the
prediction of oil spill behavior in North Carolina waters
As implicitly suggested, the possibility of identifying
a single model, while attractive on a superficial level,
is not realistic. Among other objections, the use of a
single model for solution of all cases of interest would
require use of the most complex and sophisticated model
for all cases. Therefore, it would be too expensive

41

for the simplest cases and might require input data
that is not even available. Thus, the approach has been
to classify the number of cases into the categories of
inland waters and estuaries, nearshore areas and offshore
areas. Within these categories, we have attempted to
identify families of models which would be appropriate.
In other words, there has been an attempt to gear the
model solution capability to the problem requirements.

At this stage, we can drop from further
consideration the U.S. Navy, WGD, CEQ, SPILSIM and OSTA
models, since they treat only advection. The WGD model
has been applied to offshore spills and the CEQ model has
been applied to both nearshore and offshore spills but
there are other models that include the other processes
as well. The same is true to a lesser extent of SPILSIM
which was developed for application to the special case
of the Great Lakes .

We can then reclassify Cin a subjective way)
those models that are suited for the three regions of
interest. This is shown in Table VII, where it will be
noticed that some models can be applied to more than
one region.

42

TABLE VII

CANDIDATE MODELS FOR APPLICATION
TO NORTH CAROLINA WATERS

Inland Waters and

Nearshore

Offshore

Estuaries

Narragansett Bay

USCG

USC/API

Tetra Tech, Inc.

(New York B

ight)

SLIKFORCAST

Delaware Bay

BOSTM

(SLIKTRAK)

USCG

(New York Harbor)

CAES
Delaware Bay

URI

(Georges Bank)

Puget Sound

use/API

DPPO

(Garver and Williams)

Riverspill
USCG

URI

(Georges Bank)

NWS/NOAA

(Long Island Sound)

DPPO

EDIS

PIC

(Garver and

Williams)

OSSM

EDIS

MOST

PIC

DRIFT

OSSM

DRIFT

43

Inland Waters and Estuaries. As shown in
Table VII, we have some eight (8) candidate models for
possible application to North Carolina.

The USCG (New York Harbor) model has been well
calibrated by a number of oil spills in New York Harbor
but it requires a comprehensive knowledge of the fresh
water inflows (Hudson River) and dispersion coefficients.
Fortunately for New York Harbor, thanks to the intense
interest over a long period of time, this information is
available. Essentially, the USCG (New York Harbor) and
USCG (Long Island Sound) models are empirical type models
which have the advantage of having been calibrated by
numerous observations and therefore have been validated.
Such information is generally not available for North
Carolina waters.

The Narragansett Bay model is a simplistic
easy-to-understand model that could be applied to
North Carolina inland waters. However, two changes
would have to be implemented. The wind drift angle
would have to be changed to something other than 20
and the spill-spill interaction effect would have to
be taken into account for non-instantaneous spills.
The latter may require a substantial reprogramming
effort .

Although the Tetra Tech, Inc., model has
been applied and calibrated to San Pedro Bay, California,

44

there are a number of reasons why this model should not
be considered for application to North Carolina waters.
These reasons are given in a review by Stolzenbach,
et.al. (1977), but the major reason is that the dispersion
is a numerical dispersion that is not related to the
physical dispersion, if it exists.

The PIC model is theoretically sound and a very
fundamental approach but the difficulty with this model
is that it has not been proven or validated.

The remaining three models (Riverspill,
Delaware Bay and Puget Sound) constitute the three
candidates for potential application to North Carolina
inland waters. All three models have more or less been
validated by comparison with actual spills or simulated
exercises. Riverspill was developed for the situation
on the Mississippi River and is the simplest of the three
models to implement. Although it is intended for the
situation where there is substantial river current ,
there is no reason, in principle, why it could not
handle the situation of tidal currents. Riverspill is
highly deterministic whereas the Puget Sound model is
statistical in nature.

The Delaware Bay model is also a deterministic
model and one that might prove to be better suited to
handle those cases of spills at estuary junctions with

45

the open ocean. Unlike the other two models, it includes
the advective drift due to waves as well as dispersion.
One of the difficulties in using the Delaware Bay model
is that the diffusion coefficients must be known or
assumed.

In summary, the following models seem well
suited for the prediction of oil spill behavior in
North Carolina inland waters and estuaries. These
are the Delaware Bay model, Puget Sound model and
Riverspill.

Nearshore. Some eleven (11) models reviewed
were selected as potential candidates for modeling spills
in the nearshore regions of North Carolina. For a niimber
of reasons, this is the most difficult region to model.
On the one hand, a very large area offshore must somehow
be considered and often this can only be done at a very
crude scale. On the other hand, in order to make
meaningful evaluations, modeling of details near coastal
impact locations requires a smaller, finer scale. The
result is that none of the models listed in Table VII
can be applied with confidence directly to North Carolina
waters without some modification.

The USCG (New York Bight) has been applied to
the study of spills off the Delaware and New Jersey coasts

46

but it has been only partially validated. Moreover, as
noted by Stolzenbach, et . al . (1977), many details are
lacking and some of the assumptions cannot be justified.

Neither the BOSTM or CAES model actually treat
dispersion (in a strict sense), and the BOSTM has the
additional disadvantage in that it does not consider
any of the weathering processes. For continuous spills,
the BOSTM model neglects spreading, whereas, the CAES
model neglects the spill-spill interaction effect.
However, results of both models have been applied to
prototype oil spills or to simulated spills.

If weathering processes were added to the
Delaware Bay model, it is possible that this model might
be adapted to the nearshore regions off the North Carolina
coast. Other comments regarding this model have been
provided in the previous section.

The USC/API model is a very comprehensive,
composite model , but the model as a whole has not been
validated by comparison with prototype spills. Moreover,
many of the component parts are somewhat dated. More
up-to-date information is available that could be used
in this model.

The EDIS model has been applied to the oil spill
in Campeche Bay (IXTOC-I) and to the Argo Merchant spill
but insufficient details are available to evaluate the

47

predictions. The model includes only advection and
spreading and therefore does not appear suitable for
the nearshore region off North Carolina.

The PIC model is still under development but
is very sound in a theoretical way and might be used
off North Carolina when the model has been completed and
validated. The OSSM model is still under development
although it has been applied to the IXTOC-I blowout in
the Gulf of Mexico.

From the foregoing discussion, we are left
with the following three models as the leading candidates,
These are the DRIFT, URI (Georges Bank) and DPPO (Garver
and Williams). DRIFT is a simple type model which takes
into account the tidal oscillations and which has been
applied to model a spill off Anglesey, North Wales.
The URI (Georges Bank) model is a substantially more
comprehensive model that has also been applied to a
prototype spill (i.e., Argo Merchant). The DPPO (Garver
and Williams) is the most comprehensive of the three
models, yet many of the submodel formulations are quite
simple and crude. Different versions of the DPPO model
have been applied to oil spill simulations in the Gulf
of Mexico.

In summary, modeling of oil spill behavior in
the nearshore region is a difficult task. None of the

48

models reviewed can be applied without some modification.
The following models seem best suited for this purpose:
DRIFT. URI (Georges Bank), and DPPO (Garver and Williams)
With slightly more modification, the Delaware Bay model
might also be a possibility.

Offshore. Recall that for the offshore
region, spreading and advection due to tidal currents
can, in general, be neglected. As a result of these
simplifications, modeling of oil spill behavior in the
offshore region tends to be somewhat less difficult than
in the nearshore region, all other factors being equal.

Models which are appropriate for the offshore
region are listed in Table VII. Comments regarding all
of the models except three have already been provided.
One of the deficiencies of the MOST model is that it
does not take into account evaporation. It has been
applied to the BRAVO blowout (North Sea) but the
agreement with observations needs substantial improvement
The NWS/NOAA model is still under development but when
completed will probably represent the most comprehensive
and fundamentally sound basis for predicting offshore oil
spills. The remaining model, SLIKFORCAST (SLIKTRAK) has
been employed to simulate the BRAVO blowout and the
predictions seem to compare favorably with observations

49

at different locations. SLIKFORCAST is a very comprehensive
simulation program which can be used for both contingency
planning and emergency tracking. The structure of the
program is in many ways similar to the DPPO model (Graver
and Williams ) .

In summary, SLIKFORCAST appears to offer the
greatest potential for modeling of oil spills in the
offshore region. However, the NWS/NOAA model, when
completed and validated, may be equally or better suited.
As secondary choices, the DPPO model (Garver and Williams)
and URI (Georges Bank) might also be considered.

All of the foregoing models require a certain
quality of climatological and oceanographic data for
input in order to obtain reasonable predictions. Nearly
all of the models employ a numerical approach utilizing
a given areal mesh of fineness ratio such that the
following two requirements are satisfied. One require-
ment is that the spacing dimension be small enough so
that all important and relevant details are known. In
practical terms, this translates into selecting a mesh
size that utilizes all quality environmental data. The
other requirement is that the area covered be large
enough so that boundary effects are minimized. The
latter introduces another restriction in that the mesh
spacing cannot be too small; otherwise, a system

5Q

involving an unmanageable number of equations must be
solved simultaneously at each time step. At the same
time, both the time step and the spacing must be
selected in order to minimize error propagation.

The data of most concern is that which affects
advection and dispersion. Advection is determined from
the wind field, surface currents and subsurface currents
and to a lesser extent by the wave field. Although each
oil spill model is affected somewhat differently by the
quality of the input data, the relative evaluation of
the models is not very sensitive to data which is
available for North Carolina waters. Therefore, only a
fevi general observations concerning this data (as obtained
from a review of the available open literature) are made.

As a general rule, wind information appropriate
for North Carolina inland waters and estuaries and for
the nearshore region is very poor. Wind data for the
offshore region is only marginally better. Wind data
for the inland waters and estuaries is based on extrapo-
lation of data collected from nearby airports whereas
that for the offshore region is based on ship reports.
Obviously, the geographic distribution is not uniform
since it is biased in favor of the more heavily traveled
shipping lanes.

51

The quality of current data for the inland waters
the estuaries is highly variable. Current due to tides in
some locations (i.e., Cape Fear River estuary) is adequate.
In other locations, such data is spotty. Currents due to
fresh water inflows is also highly variable.

The quality of current data for the nearshore
region is also highly variable. In the case of the offshore
region, the three-dimensional circulation pattern for
currents appears to be adequate for that portion south
of Cape Hatteras. At the same time, knowledge of the
circulation pattern for that portion north of Cape Hatteras
is only grossly known or, at best, uncertain.

The foregoing comments are based on a review of

a large number of references including but not limited to

those cited in connection with the Draft Environmental

Impact Statement, Proposed 1981 Outer Continental

Shelf Oil and Gas Lease Sale No. 56, Bureau of Land

Management Outer Continental Shelf Office, New Orleans,

Louisiana. Among those references reviewed were the

following in-house reports not available for general

distribution at this time (July, 1982).

1. Kantha, L.H., A.F. Blumberg, and G.L. Mellor
A Diagnostic Technique for Deducing the
Climatological Circulation as Applied to
the South Atlantic Bight, Report No. 69
Dynalysis of Princeton, Prepared for Bureau
of Land Management, U.S. Department of
Interior, Contract No. AA551-CT-9-32 , 1981.

52

2. Kantha, L.H,, A,F. Blumberg, H.J. Herring,
and G.L. Mellor, The Physical Oceanographic
and Sea Surface Flux Climatology of the
South Atlantic Bight, Report No. 70,
Dynalysis of Princeton, Prepared for Bureau
of Land Management, U.S. Department of
Interior, Contract No. AA551-CT-9-32 , 1981.

3. Blumberg, A.F. and G.L. Mellor, Circulation
Studies in the South Atlantic Bight with
Prognostic and Diagnostic Numerical Models ,
Report No. 71, Dynalysis of Princeton,
Prepared for Bureau of Land Management ,
U.S. Department of Interior, Contract

No. AA551-CT-9-32, 1981.

4. Blumberg, A.F., H.J. Herring, L.H. Kantha,
and G.L. Mellor, South Atlantic Bight
Numerical Model Application--Executive
Summary- — , Report No. 72, Dynalysis

of Princeton , Prepared for Bureau of
Land Management, U.S. Department of
Interior, Contract No. AA551-CT-9-32 , 1981.

These reports were made available to the author through

the efforts of Mr. Eric Vernon, Office of Marine Affairs.

In summary, there is a need for additional coastal

wind buoy station data and for improved knowledge of the

current circulation pattern offshore north of Cape Hatteras

53

CHAPTER 5
Summary and Conclusions

A brief history of the state-of-the-art in
oil spill modeling has been presented in Chapter 2,
along with an outline review of the fundamental problem
development as stated in mathematical terms. It has
been shown that the major processes which have greatest
influence on impacts resulting from oil spills are
advection, spreading.-, dispersion and evaporation, but
that all of these processes are not necessarily active
in all cases.

A review and classification of various models
as they are published in the open literature is presented
in Chapter 3. Detail reviews for each modeling effort are
presented in the Appendix. For each modeling effort, the
name or description of the model is presented along with
the primary reference(s) and corresponding abstract(s).
Reviewer's comments are also presented.

Chapter 4 contains an analysis establishing
the criteria for evaluating oil spill models for use in
North Carolina waters. This is followed by an evaluation
of leading model candidates for each of the regions into
which North Carolina waters may be classified.

54

It is concluded that the most appropriate
models for use in each of the regional classifications of
North Carolina waters are as follows:

Inland waters and estuaries: Delaware Bay Model, Puget

Sound Model and Riverspill

Nearshore areas: DRIFT, URI (Georges Bank), DPPO

(Garver and Williams).

Offshore areas: SLIKFORCAST and NWS/NOAA, when

completed and validated.

55

RECOMMENDATIONS

As a result of this study, a number of findings
and observations stand out which could easily be formulated
into a set of recommendations. Rather than present a
formal set of recommendations (which always reflect the
biases of the author) , the nature of the study is such
that it is believed to be more appropriate to simply
present the most important observations and let the
reader form his (or her) own recommendations. These are
as follows:

1. For North Carolina regions, there are
two major deficiencies which inhibit
application of the more sophisticated
models (including those now available
and under development) in order to
realize maximum benefits. One of the
deficiencies has to do with the data
base (climatic and oceanographic) which
is not sufficiently fine in order to
make detailed coastal evaluations of
oil spill impacts with reasonable
confidence. The other has to do with
the lack of detailed knowledge concerning
resource inventories (both living and

56

non-living) and their exposure to various
oil spills.

2. As a general observation, both short-term
and long-term weathering processes are
very poorly understood. This will require
a deliberate and sustained research effort
at the national level.

3. None of the models reviewed treat the
problem of continuous spills in a rigorous
manner. All of them make some assumptions
which introduce errors into the analysis.

4. There are a large number of models available
and under development. It seems more prudent
to adapt existing models to new situations
rather than embark on a new model development
program. This is true even for the nearshore
zone where , in general , most models are
inadequate. Adaptation of existing models

to North Carolina waters still remains a
challenging task, however.

5. It is unlikely that any one model can be
developed or adapted that has general
applicability to a wide variety of
situations. It is more likely that a
single family of models can be developed,

57

all with similar procedural characteristics,
to meet this need. Therefore, the attempt
to develop or to find a single model to meet
all needs is probably a fruitless endeavor.
As a by-product of this study, it was
observed that there is a tremendous
communication (or understanding) gap
between various factions of different
communities concerning the matter of
oil spills and their impacts. This
includes both federal and state
administrators and decision makers,
scientific community, industrial community,
local community leaders , concerned populace
and various state and local politicians.

58

References
(In addition to those documented in text and appendix)

Bishop, J.M. , Editor, Proceedings of Workshop on

Government Oil Spill Modeling, V/allops Island,
Virginia, November 7-9, 1979.

Fay, J. A., "The Spread of Oil Slicks on a Calm Sea,"
Oil on the Sea, Edited by D.P. Hoult , Plenum
Press, New York, pp. 53-63, 1969.

Fay, J. A. , "Physical Processes in the Spread of Oil on

a Water Surface , " Proceedings of Joint Conference
on Prevention and Control of Oil Spills , "
Sponsored by API, EPA, USCG , Washington, D.C.,
June 15-17, 1971.

Huang, N. , "A Simple Model of Ocean Surface Drift Current,"
Proceedings , Workshop on the Physical Behavior
of Oil in the Marine Environm.ent , Princeton, N.J.,
May 8-9, 1979.

Madsen , O.S., "Wind Driven Currents in an Infinite
Homogeneous Ocean , " Journal of Physical
Oceanography , 1977,

Murray, S.P., "Turbulent Diffusion of Oil in the Ocean,"
Limology and Oceanography , Vol. 27 (5),
pp. 561-660, 1972.

Stolzenbach, K.D., O.S, Madsen, E.E. Adams, A.M. Pollack

and C.K. Cooper, A Review and Evaluation of Basic
Techniques for Predicting the Behavior of Surface
Oil Slicks, Report No. 222, MIT, Cambridge,
Massachusetts, 1977.

59.

APPENDIX

6.0

APPENDIX

This appendix contains summaries for each of
the modeling efforts identified during the course of this
investigation. The original publications, which describe
and provide details on the modeling efforts, have been
collected and reviewed in all but three cases. For each
effort, the summaries list the model name or description
followed by the primary reference(s) and by the author's
own abstract or one synthesized by the reviewer. The
reviewer's own concise comments follow the abstract.

In the cases of the three exceptions (BOSTM,
use/API, and MOST), the abstract is not given and the
reviewer's comments are based on the numerous comments
provided by other reviewers or isolated comments appearing
in the open literature. The BOSTM and USC/API models have
been reviewed comprehensively by Stolzenbach, et . al . ( 1977) .
Unfortunately, in spite of a deliberate and dedicated
search by several of the area librarians, the primary
references appropriate for these three models could not
be obtained in a reasonable period of time. In one case,
the reference was simply out-of-print and in another case,
permission to release the reference due to proprietary
restrictions has not yet been obtained.

61

Finally, the 1981 Oil Spill Conference is
scheduled for release in September 1982 and will contain
additional material relevant to these models.

The order in which the modeling efforts are
presented is a historical one with the earliest developed
models appearing first.

62

Model Name or Description

U. S. Navy

Reference

Webb, L.E., R. Taranto and E. Hashimoto,
"Operational Oil Spill Drift Forecasting
Seventh U.S. Navy Symposium on Military
Oceanography, Annapolis, Maryland, 1970

Abstract

Recent incidents of large scale oil pollution
have had catastrophic effects on the ecology, recreation
and wild life resources of coastlines and coastal waters
and caused worldwide public opinion to be very sensitive
to this problem. Fleet Weather Centrals should have the
capability to provide movement forecasts for oil spills
from naval vessels or other sources, identifying the
area of maximum threat and permitting containment action
to be taken most efficiently. A tentative method is
presented using surface current parameters based on
operational data available at most Navy Fleet Weather
Centrals .

The surface current parameters influencing
oil spill drift used in this proposed forecast method
are permanent currents, wind drift, geostrophic and tidal
currents. A basic operational forecast using the vector
sum of the pertinent parameters is presented. Modifica-
tions of the basic forecast method due to the location
of the spill (open water, restricted water) are explained,
The method presented should provide an acceptable
forecast as an aid in effective control or containment
of oil spillage.

Reviewer's Comments

This is one of the earliest models proposed

and is a very simple advection-only model. The advection

is taken to be the vectorial sum of the various currents

which are assumed to consist of permanent geostrophic,

tidal and wind-drift types. This model is one of those

reviewed by Stolzenbach, et . al . (1977).

63

Model Name or Description

Warner, Graham, Dean (WGD) Model

Reference

Warner, J.L., J. W. Graham and R. G. Dean,
Prediction of the Movement of an Oil Spill
on the Surface of the Water," Paper No. 1550
Preprint of the Fourth Annual Offshore
Technology Conference, Houston, Texas, May 1972

Abstract

A calculation procedure is described to represent
the displacement of a surface oil spill under the action of
time-varying surface wind stress components; predictions using
this technique are carried out for the 1970 Arrow spill in
Chedabucto Bay, Nova Scotia.

The procedure presented is based on a convolution
integral of the x- and y- components of a time-varying wind
stress and is approximated in summation form for computer
application. For an impulsive wind stress, it is demonstrated
that the summation form is in agreement with an analytical
solution by Fredholm.

The Arrow oil spill occurred on 4 February, 1970
and during the first week of March, oil was observed on the
beaches of Sable Island some 120 miles Southeast of the spill
location. Meteorological data in the form of reduced
geostrophic winds have been published by Neu. These winds
were utilized to predict the oil spill trajectory based on:
(1) a surface velocity equal to 3% of the wind speed and
aligned with the wind direction, and (2) a surface velocity
determined from the convolution procedure presented in this
paper. It was found that the convolution procedure predictions
provided better agreement with the available information relating
to the spill transport to Sable Island. The importance of an
adequate description of prevailing surface currents and
vertical eddy viscosity is illustrated.

Reviewer's Comments

This model has been reviewed by Stolzenbach,
et , al . (1977) who notes that this is purely an advection model
since it neglects spreading, dispersion and other processes
such as evaporation. The model does not provide for inclu-
sion of coastline boundaries; therefore, its application is

64

limited to far offshore regions where wind induced currents
dominate the movement mechanisms. The model has been applied
to the Arrow oil spill (Nova Scotia) where, after calibration
via the eddy viscosity parajneter , the measured trajectory
was satisfactorily simulated.

65

Model Name or Description

Narragansett Bay (or Premack and Brown) Model

Reference

Premack, J., and G.A. Brown, "Prediction
of Oil Slick Motions in Narragansett Bay,"
Proceedings of the Joint Conference on
Prevention and Control of Oil Spills,

American Petroleum Institute, Washington, D.C.
1973

Abstract

In the development of meaningful oil spill
contingency plans, it is of great value in establishing
the response to a spill emergency to have predictions
of oil slick motions once the spill occurs. In an attempt
to evaluate some of the present technical literature on
oil spill motion, a calculation was made for the oil spill
motion which occurred in Narragansett Bay in September,
1960, when the tanker P.W. Thirtle ran aground and emitted
about 24,000 barrels of Bunker C oil over a 12 hour period
before the successful abatement of the source was completed.

The existing literature on oil slick spreading
was reviewed and the work of Fay was chosen to represent
the slick's spreading characteristics. The existing
literature on the oil slick drift was reviewed and the
work of Teason , et . al . , was used to establish the drift
motion under the influence of current and wind actions.
An available numerical hydrodynamic model of Narragansett
Bay was used to calculate the current characteristics in
the vicinity of the spills during the period of interest.
Appropriate wind data were combined with current data
in order to obtain the important hydrodynamic and meteoro-
logical conditions. Since no comprehensive theory exists
at the moment for oil slick spreading and drift, a simple
model was taken in which the 24,000 barrels were emitted
from the source in the form of 12 hourly discharges of
2000 barrels each. These individual spills were then
handled on the basis of the available spreading theory
and the drift motion calculated as described above.
Although this is a crude approximation, it does give an
estimate of the location and area magnitude of the spreading
as a function of time after the spill.

The predicted results were compared with
documentation of this spill as presented in the Providence
Journal . The overall slick motion as calculated by this
procedure was in good agreement with the arrival times
of the spill in Newport Harbor and other places in
Narragansett Bay and with the overall surface area involved

66

in the spill. This example of the calculation of the
oil slick motion in an estuary at least gives some
confidence to oil spill contingency planners that
numerical calculations can be made for use in planning
response and abatement to spills.

Reviewer's Comments

This model has been reviewed by Stolzenbach,

et . al , , (1977) who note that this model treats advection

(both wind and current) and spreading. Advection due

to wind is considered to be directed at a wind drift angle

of 20° clockwise from the wind vector. The modified wind

induced motion is then added vectorially to the tidal

currents and both are superimposed on the slick spreading.

Spreading is treated via Fay's (1971) equations. The

model does treat the case of spills that occur with time

(as contrasted with instantaneous spills). This is

accomplished via a uniform discretization of the oil spill

volume. However, the resulting subspills are treated

independently of each other via simple superposition.

That is, the spill-spill interaction effect is neglected.

The model has been applied in one case to the Thirtle

oil spill where some limited comparisons have been made.

67

Model Name or Description

Tetra Tech, Incorporated

Reference

Wang, S. , and L. Huang, "A Niimerical Model for
Simulation of Oil Spreading and Transport and
Its Application for Predicting Oil Slick Movement
in Bays." Tetra Tech, Inc. , Sponsored by
U.S. Coast Guard Research and Development Center.

Abstract

A computer model for simulating oil spreading
and transport has been developed. The model can be utilized
as a useful tool in providing advance information and this
may guide decisions for an effective response in control and
clean-up once an accidental spill occurs.

The spreading motion is simulated according to the
physical properties of oil and its characteristics at the
air-oil-water interfaces. The transport movement is handled
by superimposing the spreading with a drift motion caused by
winds and tidal currents. By considering an oil slick as a
summation of many elementary patches and applying the
principal of superposition, the model is capable of predicting
the oil size, shape, and movement as a function of time after
a spill originates.

Field experiments using either cardboard markers
or soybean oil to simulate a spill were conducted at the
Long Beach Harbor. Computer predictions showed good agreement
with the field tracers. In order to accommodate the model in
local port offices, two hardware candidates are proposed.

Reviewer's Comments

This model considers the combined effects of

advection and spreading. The numerical scheme that is

employed is intended to simulate the simultaneous

effects of advection, spreading and dispersion. Both

wind and tidal current advection are considered. The

differential spatial effects are modeled by use of elemental

cells. However, the procedure is based on a linear

68

superposition of subspills (in the spatial sense) and the
effect of neighboring subspills is not considered. In
other words, during each time step an individual cellular
subspill is considered to react independently of its
neighbors. This is considered to be a serious weakness
by Stolzenbach , et . al.(1977), who has reviewed this
procedure and who also points out that the numerical disper-
sion that results from this approach is not directly related
to any physical dispersion.

The model considers only instantaneous spills,
but, because of the fact that each cellular subspill is
considered to be "new" at the beginning of each time step,
there is no reason, in principle, why time-dependent spills
could not be considered.

The model is intended for use where spreading
and advection are the dominant processes. Therefore, its
application is limited to mostly enclosed water bodies
such as harbors and bays and protected near-shore areas
having relatively short time and length scales. To derive
maximum benefit, it is further limited to those situations
where detailed environmental data is available.

The model has been field tested and an example
using Long Beach harbor (San Pedro Bay) is illustrated.
Some five different simulated spills using tracers were
conducted to validate the model. A computer code is
included in the Appendix of the above reference.

69

Model Name or Description

Council on Environmental Quality (CEQ)

Reference

Stewart, R.J., J . W. Devanney , III, and W. Briggs
Section III, "Oil Spill Trajectory Studies for
Atlantic Coast and Gulf of Alaska," Primary,
Physical Impacts of Offshore Petroleiim Develop-
ment, Massachusetts Institute of Technology,
Sea Grant Program, MITSG 74-20, Report to
Council on Environmental Quality, April, 1974

Abstract

Oil spills can be transported many miles from
the site of an accident by the action of wind, waves and
currents. The purpose of this study is to obtain insight
into the likely behavior of oil spill trajectories
emanating from each of the thirteen potential Atlantic
Outer Continental Shelf (OCS) production regions and each
of the nine potential production areas in the Gulf of
Alaska as identified by the U.S. Geological Survey. In
addition, the likely behavior of oil spills emanating from
three potential nearshore terminal areas. Buzzards Bay,
Delaware Bay and Charleston Harbor are examined in greater
detail. Major emphasis in all of these analyses is placed
on the probability of a spill coming to shore, the time
to shore and in the case of the terminal (areas) analyses,
the wind conditions at the time the spill first reaches
shore .

Reviewer's Comments

This model has been reviewed in detail by

Stolzenbach, et.al, (1977). It is essentially an advection

model with two versions. One version can be applied to

nearshore areas and the other version is intended for

application to offshore areas. Advection by wind, tidal

currents and 'residual' currents are considered. Application

of the model requires a knowledge of the tidal currents ,

residual currents (in the case of the offshore version) and

wind measurements appropriate for the region under study.

70

Model Name or Description

Batelle Oil Spill Transport Model (BOSTM)

Reference

Ahlstrom, S.W., A Mathematical Model for
Predicting the Transport of Oil Slicks in
Marine Waters, Batelle Pacific Northwest
Laboratories, Richland, Washington, 1975

Abstract (not available)

Reviewer's Com.ments

This model has been reviewed by Stolzenbach,
et . al . (1977) and numerous references to it appear in the
open literature. From these reviews and citations, it
appears that the model treats advection, spreading and
dispersion but does not treat weathering. The CAES model
appears to be based on the BOSTM model with the addition
of weathering so that the CAES comments also apply to the
BOSTM model.

Inherent in the Discrete Parcel Random Walk
method is the assumption that each parcel (subspill)
behaves independently of the other parcels. Stolzenbach,
et.al. (1977) points out that this assumption has not been
proven and that it gives rise to a numerical dispersion
which may not be realistic. Moreover, the model employs
some arbitrary empirical factors whose origin is
uncertain ,

71

Model Name or Description

U.S. Coast Guard (New York Bight)

Reference

Miller, M.C., J.C. Bacon and I.M, Lissauer,
A Computer Simulation Technique for Oil Spills
Off the New Jersey-Delaware Coastline, DOT
U.S. Coast Guard, Washington, D.C. 1975

Abstract

Predictions for the movement of oil slicks
and their impact implications along the shoreline of
New Jersey and Delaware were determined for two potential
deepwater ports and two potential drilling sites. A hydro-
dynamic numerical model for the New York Bight area was
coupled with a wind generating model to produce temporal
patterns of concentration of oil. Shoreline impact
determinations were made for the four spill sites for
the average winter storm conditions and average summer
high pressure systems generated by the model.

Reviewer's Comments

This model has been reviewed by Stolzenbach,

et . al . ( 1977) , who points out that many details are lacking

in the report. The model consists of essentially a numerical

hydrodynamic system which utilizes a wind field generating

submodel to predict fluid particle velocities. The latter,

in turn, are employed to yield the oil spill trajectories.

72

Model Name or Description

Delaware Bay (Wang, Campbell, Ditmars) Model

Reference

Wang, H., J . R. Campbe] 1 , and J.D. Ditmars,
Computer Modelling of Oil Drift and Spreading
in Delaware Bay, CMS-RANN-1-76 , Ocean Engineering
Report No. 5, University of Delaware, Newark,
Delaware, 1976

Abstract

A generalized two-dimensional model and an
interactive computer code for the determination of the
oil slick dispersion are presented. The model considers
two mechanisms to dominate, spreading and dispersion.
Drifting refers to the drift of the center of mass of the
slick and is influenced by winds, water currents and the
earth's rotation (Coriolis force). Spreading refers to
the spread of the oil slick with respect to its center
of mass and a function of first, a gravitational force;
second, a viscous force; third, a diffusion process.
Relevant details and user options of the computer code
are discussed. An important adjunct of the computer code
is the capability for interactive graphics. This paper
concludes with a comparison between this model and field
data taken in Delaware Bay.

Reviewer's Comments

This model has been reviewed by Stolzenbach,

et.al. (1977) who notes that this is the only model that

includes the effect of advection due to waves. The model

treats advection due to wind and current also, as well as

spreading dispersion. Spreading is treated by using a

modified approach to Fay's (1971) equations. The first

two phases (gravity-inertia and gravity-viscous) are

treated unchanged. However, in lieu of the third phase

(viscous-surface tension), the modellers substitute a

73

Fickian diffusion relation which requires evaluation of a
dispersion coefficient. The authors state that the reason
for this substitution lies in the fact that while the
third phase exists in the case of small scale laboratory-
experiments, its existence is questionable in the proto-
type (or field) experiments. Therefore, the influence
of dispersion completely dominates behavior beyond the
second phase. The model results have been compared with
some field tests of drifting as carried out in Delaware Bay

74

Model Name or Description

U.S. Coast Guard (New York Harbor)
Reference

(a) Lissauer, J.M., "A Technique for Predicting
the Movement of Oil Spills in New York Harbor,"
U. S. Coast Guard Research and Development
Center, Groton , Connecticut (1974)

(b) Kollmeyer, R.C. , and M.E. Thompson , "New York
Harbor Oil Drift Prediction Model," Proceedings
of the Joint Conference on Prevention and
Control of Oil Spills, American Petroleum
Institute, Environmental Protection Agency,

and U. S. Coast Guard, Washington, D.C. (1977)

Abstract (Reference (b))

An operational predictive oil slick movement
model is developed and applied to New York Harbor. This
computer simulation employs hourly tidal currents for all
stages of the tide as input data, combined with river flow
and continuous wind data to predict oil slick movement.
Slick spreading and transport is accomplished using a
conservative form of the dif fusion/advection equation.
The shape of the slick as well as its possible separation
into multiple slicks due to current divergence is
predicted. Time steps on the order of three minutes and
grid spacing of 200 meters allow short term, small scale
slick position and shape predictions xo facilitate quick
response for location of sites for containment or cleanup
activities. Hindcasting features allow for possible
source location of the initial spill.

Reviewer's Comments

Reference (a) has been reviewed by Stolzenbach,
et.al. (1977). Reference (b) is an update and an enhance-
ment of the work begun by Lissauer (1974). Advection is
by wind, tidal currents and fresh water inflows from the
Hudson River. In reference (a), spreading followed
Fay's (1971) model. However, in reference (b), spreading

75

follows Murray's (1972) turbulent diffusion approach.
Both approaches require a moderate amount of field data
under a variety of conditions in order to calibrate the
model. Stolzenbach, et.al. (1977) notes that although
the model can be applied to other harbors, estuaries
and bays, its usefulness is hampered by the data require-
ments including knowledge of the fresh water inflows.
He also points out that the accuracy drops off in areas
close to shore.

As reported in reference (b), the model has
been extensively calibrated by a physical model
(U.S. Waterways Experiment Station, Vicksburg, Mississippi)
and by numerous observations in New York Harbor. Both the
time step interval and grid size appear to be small enough
to minimize numerical errors but error analysis has not
been conducted.

76

Model Name or Description

use/API
Reference

(a) Kolpack, R.L., N.B. Plutchak and R.W. Stearns,
Fate of Oil in a Vfater Environment, Phase II -

A Dynamic Model of the Mass Balance for Released
Oil , Chapter 19, "Settling" , American Petroleum
Institute, Publication No. 4313, Washington, D.C,
1977.

(b) API, Publication No, 4212

(c) API, Publication No. 4213

(The above publications are out-of-stock)
Abstract (not available)
Reviewer's Comments

This model has been reviewed by Stolzenbach,
et.al. (1977). It is known that the model includes
advection, spreading and weathering with particular
attention on the weathering processes.

77

Model Name or Description

SLIKFORCAST (with OILSIM and SLIKTRAK)
Reference

(a) Audunson, T., V. Dalen , J. P. Mathisen,

J. Haldorsen and F. Krogh , "SLIKFORCAST -
A Simulation Program for Oil Spill Emergency
Tracking and Long Term Contingency Planning,"
Exploration and Production Forum, London, England,
1980

(b) Det Norske Veritas, et.al., OILSIM - Oil Spill
Simulation Model Phase 1, Veritas Report No. 77-441
Det Norske Veritas Research Division, Norway, 1977

(c) Blaikley, D.R., G.F.L. Dietzel, A. W, Glass, and
P.J. Vankleef, "SLIKTRAK - A Computer Simulation
of Offshore Oil Spills, Cleanup, Effects and
Associated Costs, Proceedings of the 1977 Oil
Spill Conference, American Petroleum Institute,
Environmental Protection Agency and U.S. Coast Guard,
1977

Summary (Reference (a))

The oil spill simulation program Slikforcast is
an integrated program system where the main components are
a deterministic oil spill simulation program, a statistical
oil spill simulation program., a hydrodynamical model for
tidal currents, a da,ta preprocessor, and various output
routines. Wind data and data on residual currents must be
supplied from outside sources. The wind data may be from
(i) meteorological wind models with winds on a gridded
net; (ii) from time series of wind from one or several
observational stations; (iii) from statistical wind data
from given locations. The data preprocessor is used to
arrange the data on a unified format suitable for the
simulation models.

Simulation of oil spill drift and fate may be
carried out using either the deterministic or statistical
model or using the two simulation models in a coupled mode.
In the latter case, the results from the deterministic
simulations are used as starting conditions for the
statistical model. This permits oil spill forecasting
both on a short term and long term basis. The deterministic
model is interactive allowing for continuous updating of the
simulation results against field observations.

78

The program package may be used both for contin-
gency planning and for oil spill forecasting and hindcasting
during an emergency situation. Great care has been taken to
obtain a general program structure suitable for implementation
in different geographical regions of interest. The access to
sufficient and relevant ambient data and boundary conditions
is, however, paramount to the quality of the simulation results,

Abstract (Reference (c))

The reasons are introduced for the development of
a siumlator sufficiently simple to enable weather data normally
acquired for E & P operations to be used. "SLIKTRAK" ,
developed by Shell , applies a slick description and combat
concept , developed within the E & P forum for well blowouts
in the North Sea, but applicable to other areas. This concept
includes costs for cleanup, damages and effects of phenomena
such as evaporation and natural dispersion. These factors are
based on industry experience and vary primarily with sea
conditions .

The computer program simulates the continued
creation of an oil spill and applied weather data to predict
movements of each day's spillage for successive days at sea
and quantities of oil left after each day until the oil
either disappears or reaches a coastline.

Cumulative probability curves for the oil volumes
cleaned up, oil arriving at specified shores, total costs,
etc. , are produced by random selection of input variables
such as well location, weather data, the possibility of well
bridging, etc. , and repetition of simulated spill incidents
over a large number of cycles. Trace plots of individual
spills may also be generated.

In association with the E & P Forum's position
as technical advisers to the North West European Civil
Liability Convention for Oil Pollution Damage from Offshore
Operations, a study based on the North Sea areas has been
made. These results and further developments of the program
are discussed.

Reviewer's Comments

References (a), (b) and (c) described a main

program and two subroutines for simulating oil spills,

respectively. Reference (b) deals with deterministic

79

trajectory analysis. Reference (c) is the subroutine that
deals with the statistical trajectory analysis. This is a
general oil spill simulation program designed specifically
for accidents (particularly blowouts) in the North Sea
but it is intended for application to different geological
areas. It simulates wind from measurements or from historical
data. It also simulates dispersion into the water column
and evaporation but does not simulate biodegradation and
photooxidation because these are very slow relative to other
phases. This oil spill simulation model is not designed for
and not sufficiently accurate for detailed coastal effect
evaluation. However, both short term deterministic and
long term probabalistic oil spill forecasts are possible.

80

Model Name or Description

MOST

Reference

Paily, P.P. , and B.N. Rao, MOST-The Model for
Oil Spill Transport, Hazelton Environmental
Laboratory, Northbrook, Illinois, 1978

Abstract (not available)

Reviewer's Comments

This model treats advection, spreading and

dispersion but no weathering. Both wind drift and

surface currents are apparently treated. Spreading

is treated by Fay's (1971) approach combined with

diffusion. The model has been applied to the BRAVO

blowout (North Sea) but apparently overestimated the

slick size.

Model Name or Description

Deepwater Ports Project Office (DPPO) Model
(LOOP, Inc., and SEADOCK, Inc.)
Also referred to as SEADOCK Model

References

(a) Williams, G.N., R. Hann and W.P. James,
"Predicting the Fate of Oil in the Marine
Environment," Proceedings of the Joint
Conferen-^e on Prevention and Control of

Oil Spills held in San Francisco, California.
American Petroleum Institute, Washington, D.C.,
1975.

(b) SEADOCK Deepwater Port Final Environmental
Impact Statement, Volume III. Prepared by
Arthur D. Little, Inc., Cambridge, Massachusetts,

(c) LOOP Deepwater Port License Application,
Volume I. Prepared by Arthur D. Little, Inc. ,
Cambridge, Massachusetts.

(d) Carver, D.R. and G.N. Williams, "Advancement

in Oil Spill Trajectory Modelling," Proceedings
of Oceans '78, Marine Technology Society,
Washington, D.C., 1978,

Abstract

The continuing increase in offshore crude oil
production and transportation is presently coupled with
widespread concern over protection of the environment. This
has led to interest by members of the petrochemical community
in the development of suitable m.ethods for quickly predicting
the probable transport path and coastal impact time of a
possible oil spill in the offshore environment. An oil slick
simulator of the stochastic trajectory type has been enhanced
in an effort to provide such methods.

In particular, four areas will be addressed.
First , both the Fay radial spreading equations and a long-
term dispersive algorithm are now provided. Additionally,
the slick may now assume an elliptical shape. Secondly,
wind and current forcing functions may now be generated with
the use of Markov state transition matrices as well as through
the use of input tim.e histories. Thirdly, the model will
adaptively calculate coefficients for the transport equations
in an attempt to correct for deficiencies in the particular
equations used. Finally, the model will now execute in a
real-time m,ode with an on-line computer terminal to allow
tracking to occur during an actual spill.

82

Reviewer's Comments

Stolzenbach, et . al . (1977) reviewed reference (a)
above which is referred to as the SEADOCK model as well as
predecessor reports to references (b) and (c) which he refers
to as the DPPO model. Actually, all four references taken
together constitute a continuing effort to predict the
probable transport path and coastal impact time of an oil
spill occurring in the offshore region. Reference (d)
describes the most recent enhancements to this model but ,
unfortunately, many of the details are not provided. There-
fore, some details are assumed to be similar to those reported
in references (a), (b) and (c).

The m.odel objectives dictated the need to follow
a probabalistic approach, but the model incorporates various
levels of deterministic elements. In addition, an option
exists for operating the model in a deterministic mode.
The Fay (1971) equations are used for modeling spreading
and advection is modeled for both wind and current by using
the available data for the region. Garver and Williams state
that although the vector combination approach to slick
advection is poorly supported, the behavior of actual oil
spills does not justify the use of a more rigorous modeling
method. They do, however, use a weighted average approach
to evaluate the proportionality constants and provision is
made to incorporate new values as more data becomes available.

83

An algorithm for modeling long-term dispersion
is provided. This algorithm is based on a very limited
number of observations but does permit the slick to assume
an elliptic shape. In addition, the subsurface transport
processes are modeled.

The movement of the slick is tracked for a
specified duration of time via employment of actual data
(if measured) or of simulated data (based on a Markov
chain model) if data is not available. Thus, the model
is suitable for both environmental assessment and planning
studies as well as for operational purposes during
emergencies and clean-up. The latter is known as the
adaptive tracking feature.

Certain submodels that account for weathering
and degradation of the slick are also included. These
follow the first order model of Moore, Dwyer and Katz
(1973) for evaporation and dissolution and arbitrary
empirical functions for emulsif ication and sedimentation.

In summary, this model represents a very
comprehensive approach to the problem of predicting oil
spill trajectories in the offshore region and even
though many of the submodels are elementary, the
fundamental stochastic approach is will justified.

84

Model Name or Description

University of Rhode Island (URI) Georges Bank

Reference

Cornillon, P.C., M.L. Spaulding, and K. Hansen
"Oil Spill Treatment Strategy Modeling for
Georges Bank," Proceedings of the 1979 Oil
Spill Conference, Los Angeles, California,
American Petroleum Institute, Environmental
Protection Agency, and U.S. Coast Guard.

Abstract

As part of a larger project assessing the
environmental impact of treated versus untreated oil spills,
a fates model has been developed which tracks both the
surface and subsurface oil. The approach used to spread,
drift and evaporate the surface slick is similar to that
in most other oil spill models. The subsurface technique,
however, makes use of a modified particle-in-cell method
which diffuses and advects individual oil/dispersant droplets
representative of a large number of similar droplets. This
scheme predicts the time-dependent oil concentration distri-
bution in the water column, which can then be employed as
input to a fisheries population model. In addition to
determining the fate of the untreated spill, the model also
allows for chemical treatment and/or mechanical cleanup of
the spilled oil. With this capability, the effectiveness of
different oil spill control and removal strategies can be
quantified.

The model has been applied to simulate a 34,840
metric ton spill of a No. 2-type oil en Georges Bank. The
concentration of oil in the water column and the surface
slick trajectory are predicted as a function of time for
chemically treated and untreated spills occurring in April
and December. In each case, the impact on the cod fishery
was determined and is described in detail in a paper by
Reed and Spaulding presented at this conference.

Reviewer's Comments

This model considers spreading, advection

dispersion and evaporation. Spreading is based on a

modified version of Fay's (1971) theoretical development.

85

Ocean currents were simulated by averaging estimates
obtained from a long-term study of surface drifter returns.
The model envisions that detail information from ocean
surface transport or continental shelf circulation studies
will be available as input. Advection is then simulated
by the vectorial addition of ocean and wind-induced currents
Evaporation is simulated by application of a numerical
solution to the procedures developed by Wang, Yang and
Hwang (1976).

The subsurface transport of the dispersed oil
is simulated by a three-dimensional mass transport equation
which is solved by a "quasi Monte Carlo technique."
Although it is stated that extensive model testing was
performed in order to assure that the sub-process models
correctly interfaced with one another, it appears that the
model contains a mixture of numerical as well as physical
dispersion .

The model has been applied to four different
versions of a simulated spill. Unfortunately, there are
no comparisons with field observations.

86

Model Name or Description

Puget Sound Model

Reference

Karpen , J. , and J. Gait, "Modeling of Oil
Migration in Puget Sound," Proceedings of
Oceans ' 79 , Marine Technology Society,
Washington, D. C.

Abstract

The oil migration is simulated with a mixed
Lagrangian-Eulerian model. The movement of the oil is
modeled with Lagrangian point masses or Lagrangian elements
(LE). Time dependent dispersion is applied to the
individual LE ' s . Eulerian current and wind fields are used
to advect the LE ' s .

Several scenarios for hypothetical spills in the
Straits and upper Puget Sound are shown. The model accurate-
ly predicted the migration of the diesel oil spill at
Ancortes, Washington.

The model is general enough so that submodels
from currents and winds developed for other regions can
be easily integrated into the simulation.

Reviewers Comments

A model najned the Puget Sound Model was reviewed

by Stolzenbach, et . al (1977), but the model described in the

above reference is an entirely different one than that

reviewed by Stolzenbach. The model simulates advection

and the combined effects of spreading and dispersion.

Spreading, using the theoretical approach developed by

Fay (1971) was not employed. Instead the available data

was used to determine a. spill length scale as a function of

time. The spill length corresponding to the spill size

was then used in a modified random walk scheme to simulate

87

the combined spreading and dispersion. It is stated that
this approach is comparable to diffusion by continuous
movements for discrete particles or Fickian diffusion.
The advection for wind and tidal currents are then super-
imposed on the combined spreading and dispersion.

Most of the data used in the regression
equations to determine the spill length, size and time
relations do not consider spill rates. There is some
question how this technique can be used in situations
where spill rate is an important factor. Also, there
may be other factors (i.e., boundary conditions) that
effect the spill length-time relationship.

88

Model Name or Description

Canadian Arctic Environment Service (CAES )

Reference

Venkatesh , S., S.H. Sahota and A.S. Rizkalla,
"Prediction of the Motion of Oil Spills in
Canadian Arctic Waters," Proceedings of the
1979 Oil Spill Conference, Los Angeles,
California, American Petroleum Institute,
Environmental Protection Agency, and
U.S. Coast Guard (1979)

Abstract

An oil spill movement prediction m.odel operating
as part of a real-time Environmental Prediction Support
System for the Canadian Beaufort Sea has been developed.
The present version of the model considers spills only in
open waters, that is, the sea surface is considered to be
ice free. The model has been partially verified with
data obtained from oil simulation experiments conducted
in the Bay of Fundy , off the coast of Canada during the
months of August and September 1978. With the use of
observed winds, the model-predicted locations of "Orion"
buoys used to simulate the motion of oil on water, agreed
fairly well with their observed locations. These verifi-
cation tests also pointed out the need for high resolution
surface wind forecasts — essential data for computing wind-
driven water currents which move the oil.

Reviewer's Comments

This model uses the numerical technique known

as the Discrete Parcel Random Walk method. This method

is identical to that used by Ahlstrom (1975). In a sense,

this model is an adaptation of Ahlstrom' s model (BOSTM) .

Spreading follows Fay's theory. Of the weathering processes

only evaporation and emulsif ication are considered because

of their predominance in the first few days of slick

formation. Treatment of both evaporation and emulsif ication

89

follows the procedures developed by Mackay and Leinonen
(1977). Advection is considered to be due only to wind-
induced surface currents and is based on the work of
Madsen (1977). The expression for steady surface drift
becomes a very simple relation.

The computational method assumes that the sub-
spills act independently of each other. Therefore, the
effect of neighboring subspills is neglected as is the
spill-spill interaction effect. The model has been applied
to four separate simulation experiments conducted in the
Bay of Fundy.

90

Model Name or Description
Riverspill

Reference

Tsahalis, D.T., "Contingency Planning for
Oil Spills: Riverspill - A River Simulation-
Model," Proceedings, 1979 Oil Spill Conference,
Los Angeles, California, American Petroleiim
Institute, Environmental Protection Agency,
and United States Coast Guard (1979)

Abstract

A simulation model, Riverspill, has been
developed for the prediction of transport, spreading, and
associated land contamination of oil spills on rivers. The
effects of volume and type of oil, use of Oil Herder or
not, type, location, and time of occurrence of the oil spill,
geometry, and hydrographic characteristics of the river and
wind speed and direction are taken into account. The model
is capable of operating in either deterministic or stochastic
mode. When the model is used in the deterministic mode, it
predicts the path and associated land contamination of a
specific oil spill as a function of time. When the model
is used in the stochastic mode, it estimates the probability
that an oil spill will be transported into a specific region
after an accidental discharge within another specified
region. In the present study, the model is specifically
applied to the lower Mississippi River. However, the model
is general and can be applied to any river. The predictions
of the model are in very good agreement with the observed
behavior of actual oil spills on the Mississippi River.

Reviewer's Comments

This model was developed specifically for river

flows. Spreading follows Fay's (1971) theory, in which all

three regimes are modeled. Advection due to wind and current

is modeled by some derived relationships which depend on the

direction of the wind velocity vector relative to the

current vector. Some experimentally d9termined coefficients

91

are employed in the derived relationships. Cross flows
(i.e. , secondary) are also considered. The model has been
applied to two accidents on the Mississippi River. The
agreement between observed and model predicted deposition
sites ranged from good to excellent.

92

Model Name or Description

National V/eather Service (NWS ) Model
References

(a) Hess, K.W., " The National Weather Service Oil
Spill Motion Forecasting Program," Workshop

on the Physical Behavior of Oil in the Marine
Environment , Princeton, University, Princeton,
New Jersey, May 8-9, 1979

(b) Hess, K.W. , and C.L. Kerr, "A Model to Forecast
the Motion of Oil on the Sea, Proceedings of the
1979 Oil Spill Conference, Los Angeles, California.
American Petroleum Institute, Environmental
Protection Agency and U.S. Coast Guard,
Washington, D.C. 1979

Abstract (From Reference (a) Text)

The purpose of this model is to provide an
operational method to forecast the movement of oil spilled
in the ocean. The model utilizes the National Weather
Service (NV/S) meteorological forecasting capabilities to
provide scientific prediction of the mediiim or long-range
(up to 5 days) future oil behavior. In this time range,
oil motions are strongly influenced, if not dominated,
by wind effects (wind stress, wind-driven ocean currents,
wind waves and evaporation). Forecasting using numerical
models of the atmosphere give better results for this time
range than the use of either a statistical (climatology)
or a presistence approach.

The present model does not use the trajectory
and point-mass simplifications of oil behavior; instead
a two-dimensional approach is employed to predict the
spatial variation of oil thicknesses. The oil spill model
consists of three major segments, relies heavily on finite
difference techniques and is driven by output of NWS
atmospheric models. The three separate parts are: (1) a
model of the atmospheres near-surface planetary boundary
layer, (2) a dynamic model of the upper mixed-layer of the
ocean, giving surface currents, and (3) a dynamic model of
two-dimensional variable thickness oil distribution on the
sea. In general, the lirst two segments provide input to
the third in the form of stresses at the upper and lower oil
boundaries. The most important simplifications of the
present model are neglect of tide and wave effects and absence
of weathering. Reference (b) provides additional details
on this model which is under continuing development.

93

Abstract (From Reference (b))

A model to forecast the motion of oil spilled
on the surface of water was established by combining
separate models for the motion of oil, the motion of
water, and the motion of air. The model for the motion
of oil is based upon the hydrodynamic equations as they
apply to oil on water. This model requires information
at both the lower and upper boundaries of oil. At the
oil lower boundary, the information is obtained from a
model for the motion of water. This model is formulated
by combining Ekman dynamics and continuity for the upper
mixed layer of the sea. At the oil upper boundary, a
model for the motion of air provides the required
information. This model is based upon an analysis of
output obtained from one of the National Weather Service's
multi-level atmospheric models. A number of case studies
demonstrate the features of the separate models and the
composite oil spill model.

Reviewer's Comments

This is a very comprehensive model which takes
advantage of the operational wind forecasts made available
by the National Weather Service. Separate models for the
motion of oil , water and air are combined into a composite
model which is undergoing continuing development and
improvement . The model has been applied in preliminary
form to the Argo Merchant spill. The model results in
com.parison with observations indicate the potential
applications and limitations of this model. The model
approach appears well suited for making real time fore-
casts of oil motion in locations far offshore and
especially where wind is the dominant driving force.

94

Model Name or Description

U.S. Coast Guard (Long Island Sound)

Reference

Kollmeyer, R,C. , "Long Island Oil Spill Drift
Prediction Model," Proceedings of Workship on
Government Oil Spill Modeling, November 7-9,1979
Wallops Island, Virginia, NOAA , Washington, D.C.
February 1980

Abstract

Because of the sensitive nature of the Long Island
area and the complex nature of the surface currents in the
Sound, the On-Scene Coordinator (OSC) must be able to predict
oil spill trajectories accurately in order to deploy cleanup
equipment and to protect sensitive areas. Present oil spill
movement prediction models are inadequate for this need and
are not readily accessible to the OSC.

The goal of the project is the production of a
reaH;ime prediction model that will forecast the movement
and spread of an oil spill in Long Island Sound. Upon
command, the model will, for a given period, produce a
time series of charts displaying the location, shape and
concentrations of the oil spill.

The model will be constructed on the computer
facilities provided by the Department of Computer Science,
U.S. Coast Guard Academy. Captain R.C. Kollmeyer will be
the overall project coordinator. A close liaison will be
maintained am.ong all interested and contributing units.
At present, the following Coast Guard units are involved:

1. Coast Guard Academy

2. Commander, CG Group, Long Island Sound

3. USCG Research and Development Center, and

4. USCG Oceanographic Unit.

The oil drift mechanisms to be modeled will
include the following:

1. predicted tidal currents,

2. Stokes drift, considering both
duration and fetch limitations

3. leeway of oil slick caused by the wind, and

4. wind drift currents of the surface layer.

95

A relatively sophisticated computer graphics
output is desired in the form of segment maps showing the
spill, its location, areal size, and concentration
gradients. These maps would be produced on a time basis,
showing the oil spill's predicted location every 15 minutes
from the time of th^ sDill.

The preparation of the tidal current data set
will include the use of overlays for the NOS tidal current
charts to allow their transfer to tho model matrix. A scaling
program (inverted smoothing) will then determine the currents
for all other points on the matrix for each of the 13 current
charts available. Currents along the shore boundaries will
be made zero unless other information is available.

A planned program of verification and testing
will be developed at part of the mod^^l completion. Proce-
dures will be proposed by which small oil spills in
Long Island Sound may be monitored and used in a hindcast
mode. In addition, a testing program will be drafted that
would use oil drift simulators which can be tracked and
compared with model predictions.

Reviewer's Comments

The above reference simply describes a program

being undertaken to develop a model with certain specific

goals and features. It likely will be similar to the

model developed by USCG for New York Harbor.

96

Model Name or Description

SPILSIM
References

(a) Pickett, R.L., "A Surface Oil Spill Model for
the Great Lakes , Proceedings of the V/orkshop
on Government Oil Spill Modeling,

November 7-9, 1979, Wallops Island, Virginia
NOAA, February 1980

(b) Boyd, J. D. , "A Surface Spill Model for the
Great Lakes," Great Lalces Environmental Research
Laboratory, Ann Arbor, Michigan, 1979

Abstract (Edited from Reference (a))

This model was developed as an operational
forecast model for the movement of surface-pollutant spills
on the Great Lakes with special emphasis on oil spills. Oil
spills are of particular concern because of their environ-
mental impact and the substantia.l quantities of oil transported
on the Great Lakes, both as cargo and as fuel. The Great Lakes
Environmental Research Laboratory (GLERL) undertook this
modeling effort because a number of models were being developed
for oceanic spills but none were being developed for those on
the Great Lakes. The resulting model, SPILSIM, is a batch
oriented model derived form oil spill models from the Canada
Center for Inland Waters (CCIW) and NOAA's Pacific Marine
Environmental Laboratory (PMEL) . It predicts the movement
of an insoluble surface spill anywhere within the Great
Lakes, given spill size and location and surface currents
and winds in the area of interest. Modifications to make
the model interactive are straightforward.

Reviewer's Comments

This is a simple model that simulates advection

by wind and currents.

97

Model Name or Description

Reference

Environmental Data and Information Service
(EDIS) Model

Bishop, J.M. , A Climatological Oil Spill
Planning Guide, No. 1, The New York Bight.
NOAA, Washington, D.C., 1980

Abstract

Over the past 4 years, the Environmental Data
and Information Service (EDIS) of NOAA has developed a
multiple trajectory oil spill model. The model is based
on the archived wind and current data available within
EDIS's National Climate Center (NCC) and National
Oceanographic Data Center (NODC). The model was
originally designed for assessment of possible environ-
mental impact due to construction of a deepwater port
off the Texas coast, but has been used effectively in
the Argo Merchant and Compeche oil spills for a rapid
estimate of the impact areas. The successful use of
this model for climatological assessments of large ocean
spills leads to the conclusion that this type of
climatological forecasting technique should be a part
of our overall response to major oil spills.

Although the utility of this approach has been
shown in these two examples of large open-ocean oil spills,
a better application of this climatological (Type I) model
is in contingency planning (prespill resource allocation).
In this mode, one can map most probable imnact zones
over known local resources (biological or economic). Such
a use has been initiated in a recent EDIS publication
produced for use by the 3d Coast Guard District contingency
planners, couples key environmental data (both physical
and biological) with climatological oil spill trajectory
forecasts.

Reviewer ' s Comments

This model is an enhancement of the DPPO model

but in an empirical -simplified direction rather than in

the fundamental-comprehensive direction taken by Garver

98

and Williams (1978). Advecticn due to wind and current is
determined using measured data. These, in turn, are coupled
with Fay's spreading theory to predict the spill trajectories

99

Model Name or Description

Particle-In-Cell (PIC) Model

Reference

Tingle, A.G., "Perturbation Analysis of the
New York Bight," Proceedings of the Workshop
on Government Oil Spill Modeling, November 7-9,
1979, V/allops Island, Virginia, NOAA,
Washington, D.C., February 1980.

Abstract

The physical transport of pollutants, their
modification by the coastal food web, and transfer to men
are problems of increasing complexity on the continental
shelf. In an attempt to separate cause and effect, a
computer modeling technique is applied to problems involving
the transport of pollutants as one tool in assessment of
real or potential coastal perturbations. Approaches for
further model development of the biological response within
the coastal marine ecosystem are discussed. Our present
perturbation analyses consist of 1) a circulation submodel,
2) a simulated trajectory of a pollutant particle within
the flow field, and 3) a time-dependent wind input for each
case of the model. The circulation model is a depth-
integrated, free surface formulation that responds to wind
stress, bottom friction, the geostrophic pressure gradient,
the coriolis force, and the bottom topography. The transport
diffusion model is based on Lagrangian mass points, or
"particles" moving through a Eulerian grid. The trajectories
of material moving on the surface and in the water column
are computed. It has the advantage that the history of each
is known. With these models, we have been able successfully
to: 1) reproduce drift card data for determining the
probabilities of a winter oil spill beaching within the
New York Bight, 2) analyze the source of floatables
encountered on the south shore of Long Island in June 1976,
and 3) predict the trajectory of oil spilled in the Hudson
River after it had entered the New York Bight Apex. For
future analyses, the shallow water model can be modified
or replaced with a numerical model that contains a more
sophisticated parameterization of the physical circulation.
Also, the particle-in-cell model can be modified to
explicitly include chemical reactions and interactions
with the biota. A model should be used in the context
of the level of resolution or aggregation that is known
about the ecosystem and the management decision to the made.
Models are used also as an aid in selecting situations
that merit further analysis with more comprehensive ecologi-
cal reasoning.

100

Reviewer's Comments

This model is based on a numerical hydrodynamic
model of the area under study. One of the difficulties
in applying models of this type to other regions is the
determination of the model boundaries and treatment of the
boundary cells (free, fixed, no-slip, reflecting), A
disadvantage in the use of models of this type is the
required cost of computation — even on the largest, most
sophisticated computers.

101

Model Name or Description

Reference

The U.S. Geological Survey Oil Spill Trajectory
Analysis (OSTA) Model

Samuels, W.B., "The USGS Oil Spill Trajectory
Analysis Model," Proceedings of the Workshop
on Government Oil Spill Modeling, November 7-9, 1979,
Wallops Island, Virginia, NOAA, Washington, D. C,
February 1980.

Abstract

The USGS oil spill trajectory analysis model
(OSTA) is used to calculate the probability of oil spills
occurring and contacting environmental resources and sections
of the coastline. A grid system is superimposed on the
study area with a maximum of 480 miles on a side. The
dimension of the grid cell is variable depending on the size
of the study area. Locations of environmental resources
proposed and existing lease tracts, and oil transportation
routes (pipeline and tanker) are determined by their
positions in the model's grid system. Data from different
map projections can be digitized and fitted into the model's
grid system by coordinate conversion subroutines. A maximum
of 31 categories of resources and up to 100 segments
(2 different sets) of the coastline can be included in the
analysis.

Oil spills are simulated in a Monte Carlo fashion.
Typically, 500 simulated oil spills are launched per season
from each launch point (platform location, pipeline, or
tanker route). Spills are transported by monthly currents
and by winds sampled from wind transition matrices. These
matrices, composed of 41 wind velocity states, are based
on historic wind records. They are constructed for each
season for up to six wind stations. Surface ocean currents
are incorporated in a deterministic manner by representing
monthly current fields in the model's grid system. The
spill movement algorithm consists of computing the vector
sum of a wind and current vector for successive 3-hour
increments. Each grid cell in the path of the spill is
checked for the presence or absence of each environmental
resource. Spill movement ends in one of three ways:
1) the spill contacts land, 2) the spill decays at sea,
or 3) the spill moves off the map.

102

Conditional probabilities of contact are reported
for 3-, ] 0- , and 30- day travel times. Oil spill occurrence
is treated as a Poisson process, in which the exposure
variable is the volume of oil produced or transported.
Scenarios outlining proposed oil production and transportation
are constructed for various alternatives. The overall risks
are determined by combining spill occurrence probabilities.
Recent applications of the model have been: Sale 55 (Northern
Gulf of Alaska, Sale 53 (Northern and Central California),
and Sale 46 (Kodiak Island).

Reviewer's Comments

This model uses a vectorial approach of wind and

current to simulate advection. The use of input data is

very similar to the DPPO model. The movement of the slick

is tracked for a specified duration of time via employment

of actual data or of simulated data (based on a Markov chain

model) if real data is not available. This model has been

used for environmental assessment in connection with a

number of offshore lease sales (e.g., Alaska, Gulf of

Mexico, Southern California and North-, Mid- and South

Atlantic).

103

Model Name or Description

On-Scene Spill Model (OSSM)

Pacific Marine Environmental Laboratory

NOAA, Seattle, Washington

Reference

(a) Torgrimson, G.M., and J. A. Gait, "An On-Scene
Spill Model for Pollutant Trajectory Simulation,"
Workshop on the Physical Behavior of Oil in the
Marine Environment , Princeton University,
Princeton, New Jersey

(b) Gait, J.A. , and G.M. Torgrimson , "On-Scene
Trajectory Modelling for the U.S. Response
to the IXTOC-1 Blowout," Proceedings of the
Workshop on Government Oil Spill Modeling,
November 7-9, 1979, Wallops Island, Virginia,
NOAA, Washington, D.C., February 1980.

Abstract

(a) The Pacific Marine Environmental Laboratory's
Hazardous Materials Scientific Support Team (PMEL/HMSST)
has developed a data management system for the rapid
simulation of pollutant trajectories in the marine
environment. The On Scene Spill Model (OSSM) has been
designed to function at several different levels of
complexity depending on user requirements and available
environmental data. OSSM may be utilized as an environ-
mental assessment tool to predict pollutant trajectories
and coastal impact areas, given hypothetical pollutant
spills. General cautions about interpreting the output
are included. Planned modifications and additions to
the existing program are listed. The data management
system was used to document and predict pollutant
trajectories during the PECK SLIP oil spill off the
east coast of Puerto Rico on December 19, 1978.

(b) During the spill event associated with the
IXTOC I well blowout, the U.S. National Contingency
Plan was activated. In anticipation of this, the NOAA
Hazardous Materials Response Project requested the
Hazardous Materials Scientific Support Team (for physical
processes) to supply trajectory information through the
scientific support coordinator. This was done beginning
June 12, 1979. During the summer, this plan required a
variety of activities including: (1) collecting available

104

background information, (2) carrying out observational
field programs, (3) coordinating data and information
presented by other researchers (Federal, State and private)
and (4) analyzing trajectories.

During the spill, trajectory models were used
in basically three different forms. The first was in a
long-term statistically controlled form (strategic) for
planning call up and retirement of scientific or cleanup
units. The second was a short-term localized forecast
(tactical) for input into day-to-day planning. The third
was in a receptor mode that identified danger zones that
could impact scientific high-valued regions. This, in
turn, was used to develop optimal mapping strategies for
overflights .

The basic model for IXTOC I studies was a new
version of OSSM (On-Scene Spill Model) incorporating a
number of advanced features. In addition, several
auxiliary programs to analyze circulation data were also
used for the first time in a real spill situation. Both
hindcasts and forecasts from the models have provided
useful input to the overall response program..

Reviewer's Comments

This model is undergoing continuing development
and improvement. The model is designed to be utilized
as a real-time spill trajectory model, Advection is
handled in vector fashion from both wind and current.
Diffusion is currently modeled via a "Monte Carlo" approach
to the governing diffusion equations. Other algorithms are
being considered which would more accurately simulate spread-
ing and diffusion. Additional details will be provided in
the 1981 Joint Conference on Prevention and Control of
Oil Spills which will be available in September, 1982.

105

Model Name or Description

DRIFT

Reference

Hunter, J.R., "An Interactive Computer Model

of Oil Slick Motion," Oceanology International '80,

Abstract (from text)

An interactive computer model has been developed
to predict the motion of an oil slick on the sea surface
which is particularly suited to the shallow waters of the
sea shelf. Wind-induced drift and spreading are considered
to be the most important mechanisms in this motion. The
wind-induced drift is modeled via an empirical relationship
and the spreading, mixing and decay are modeled using Monte
Carlo techniques.

The model has been applied to an area of Irish
sea near Anglesey, U.K., and has predicted at least one
recent spill successfully.

Reviewer's Comments

This model is a classical deterministic approach

which incorporates the major mechanisms of importance to the

prediction of the motion of an oil spill in the area for

which it was designed. One of the underlying assumptions

in this model is that spreading is caused by turbulence

and lateral velocity shears. As a consequence, this model

should not be used to predict oil spill behavior during

early stages following the initial spill. The justification

for the approach that has been employed is based on a few

observations of very large spills wherein it was noted

that spill diameter increases with the first power of

time. With proper imput , the model appears to provide

106

reasonable predictions. However, for predictions nearshore ,
the model is very sensitive to errors in certain input
information. Therefore, use of the model in an operational
mode requires continuous updating of input information.

107

STATE LIBRARY OF NORTH CAROLINA

3 3091 00748 0361

CEIP Publications

1. Hauser, E. W. , P. D. Cribbins, P. D. Tschetter, and R. D. Latta.
Coastal Energy Transportation Needs to Support Major Energy Projects

in North Carolina's Coastal Zone. CEIP Report iH. September 1981. $10.

2. P. D. Cribbins. A Study of OCS Onshore Support Bases and Coal Export
Terminals. CEIP Report #2. September 1981. $10.

3. Tschetter, P. D., M. Fisch, and R. D. Latta. An Assessment of
Potential Impacts of Energy-Related Transportation Developments on
North Carolina's Coastal Zone. CEIP Report #3. July 1981. $10.

4. Cribbins, P. S. An Analysis of State and Federal Policies Affecting
Major Energy Projects in North Carolina's Coastal Zone. CEIP Report
#4. September 1981. $10.

5. Brower, David, W. D. McElyea, D. R. Godschalk, and N. D. Lofaro.
Outer Continental Shelf Development and the North Carolina Coast:
A Guide for Local Planners. CEIP Report #5. August 1981. $10.

6. Rogers, Golden and Halpern, Inc., and Engineers for Energy and the
Environment, Inc. Mitigating the Impacts of Energy Facilities: A
Local Air Quality Program for the Wilmington, N. C. Area. CEIP
Report #6. September 1981. $10.

7. Richardson, C. J. (editor). Pocosin Wetlands: an Integrated Analysis
of Coastal Plain Freshwater Bogs in North Carolina. Stroudsburg (Pa):
Hutchinson Ross. 364 pp. $25. Available from School of Forestry,
Duke University, Durham, N. C. 27709. (This proceedings volume is for
a conference partially funded by N. C. CEIP. It replaces the N. C.
Peat Sourcebook in this publication list.)

8. McDonald, C. B. and A. M. Ash. Natural Areas Inventory of Tyrrell
County, N. C. CEIP Report #8. October 1981. $10.

9. Fussell, J., and E. J. Wilson. Natural Areas Inventory of Carteret
County, N. C. CEIP Report #9. October 1981. $10.

10. Nyfong, T. D, Natural Areas Inventory of Brunswick County, N. C.
CEIP Report #10. October 1981. $10.

11. Leonard, S. W., and R. J. Davis. Natural Areas Inventory for Pender
County, N. C. CEIP Report #11. October 1981. $10.

12. Cribbins, Paul D., and Latta, R. Daniel. Coastal Energy Transporta-
tion Study: Alternative Technologies for Transporting and Handling
Export Coal. CEIP Report #12. January 1982. $10.

13. Creveling, Kenneth. Beach Communities and Oil Spills: Environmental
and Economic Consequences for Brunswick County, N. C. CEIP Report
#13. May 1982. $10.

CEIP Publications

14. Rogers, Golden and Halpern, Inc., and Engineers for Energy and the
Environment. The Design of a Planning Program to Help Mitigate Energy
Facility-Related Air Quality Impacts in the Washington County, North
Carolina Area. CEIP Report #14. September 1982. $10.

15. Fussell, J., C. B. McDonald, and A. M. Ash. Natural Areas Inventory
of Craven County, North Carolina. CEIP Report #15. October 1982.
$10.

16. Frost, Cecil C. Natural Areas Inventory of Gates County, North
Carolina. CEIP Report #16. April 1982. $10.

17. Stone, John R., Michael T. Stanley, and Paul T. Tschetter. Coastal
Energy Transportation Study, Phase III, Volume 3: Impacts of Increased
Rail Traffic on Communities in Eastern North Carolina. CEIP Report #17.
August 1982. $10.

19. Pate, Preston P., and Jones, Robert. Effects of Upland Drainage on
Estuarine Nursery Areas of Pamlico Sound, North Carolina. CEIP
Report #19. December 1981. $1.00.

25. Wang Engineering Co., Inc. Analysis of the Impact of Coal Trains
Moving Through Morehead City, North Carolina. CEIP Report #25.
October 1982. $10.

26. Anderson & Associates, Inc. Coal Train Movements Through the City of
Wilmington, North Carolina. CEIP Report #26. October 1982. $10.

27. Peacock, S. Lance and J. Merrill Lynch. Natural Areas Inventory of
Mainland Dare County, North Carolina. CEIP Report #27. November 1982.
$10.

28. Lynch, J. Merrill and S. Lance Peacock. Natural Areas Inventory of
Hyde County, North Carolina. CEIP Report #28. October 1982. $10.

29. Peacock, S. Lance and J. Merrill Lynch. Natural Areas Inventory of
Pamlico County, North Carolina. CEIP Report #29. November 1982. $10.

30. Lynch, J. Merrill and S, Lance Peacock. Natural Areas Inventory of
Washington County, North Carolina. CEIP Report #30. October 1982.
$10.

31. Muga, Bruce J. Review and Evaluation of Oil Spill Models for Applica-
tion to North Carolina Waters. CEIP Report #31. August 1982. $10.

33. Sorrell, F. Yates and Richard R. Johnson. Oil and Gas Pipelines in
Coastal North Carolina: Impacts and Routing Considerations. CEIP
Report #33. December 1982. $10.

34. Roberts and Eichler Associates, Inc. Area Development Plan for Radio
Island. CEIP Report #34. June 1983. $10.

35. Cribbins, Paul D. Coastal Energy Transportation Study, Phase III,
Volume 4: The Potential for Wide-Beam, Shallow-Draft Ships to Serve
Coal and Other Bulk Commodity Terminals along the Cape Fear River.
CEIP Report #35. August 1982. $10.