ASSIMILATION OF AVHRR, GROUND WAVE RADAR AND RADARSAT SAR DATA INTO A COASTAL
CIRCULATION AND OIL SPILL MODELLING SYSTEM
Donald O. Hodgins
President, Seaconsult Marine Research Ltd.
8805 Osler Street, Vancouver, Canada V6P 4G1
Commission VII, Working Group 5
KEY WORDS: Remote sensing. Data Assimilation. Circulation Modelling. Oil spill modelling. Coastal zone protection.
ABSTRACT
Operational three-dimensional ocean circulation models have been developed for the western Canadian continental shelf and the
Georgia-Fuca inland waterway. Assimilation of near real-time SST data from AVHRR imagery using a nudging scheme, and
surface current measurements from SeaSonde HF radars with an error-dependent weighted blending method has led to significant
improvements in surface current prediction accuracy. An oil spill trajectory and weathering model has-been coupled into the
circulation model, utilizing the most recent current data for the advective calculations. RADARSAT SAR image classification,
combined with computerized editing tools, is used to monitor oil slicks and parameterize slick features for re-initializing the oil
spill model.
Model data are distributed via the Internet for assessment by users. The integrated SEACAST system provides
response organizations and monitoring agencies with timely, accurate information for decision making, taking maximum
advantage of near-real time data.
INTRODUCTION
Catastrophic and chronic oil spills in marine waters pose
major challenges for environmental protection and
management. Oil spill trajectory and fate models, coupled to
coastal circulation models, are used to plan and carry out
countermeasures for catastrophic spills. Similar models are
also used to backtrack smaller spills to ships and offshore
platforms to identify the responsible parties.
The accuracy and utility of such models is greatly improved
through assimilation of remotely sensed data. An operational,
integrated modelling system, SEACAST, has been
implemented for the west coast of Canada to provide
information for coastal protection and management, including
oil spills. A three-dimensional prognostic circulation model
(C3), which forms the core of this system, has been applied at
two different grid scales: 5-km over the continental shelf and
1-km in the coastal sea between Vancouver Island and the
mainland of Canada and the United States. An oil spill model
(SPILLSIM), coupled to the circulation model, provides
predictions of the dispersion, spreading and weathering of a
range of crude and distilled petroleum products. AVHRR
SST imagery is assimilated directly into the temperature field
of the hydrodynamic model in order to improve the accuracy
of the baroclinic circulation.
SeaSonde high-frequency ground wave radars are used to
collect real-time surface current maps with approximately 1-
km resolution over a broad expanse of sea. These data are
also assimilated into the circulation model over the radar's
field-of-view. The current data are obtained hourly and are
available to the modelling system within about 1 h of the end
of the measurement cycle. Current maps of this type increase
the accuracy of the modelled surface current fields, which is
particularly important in areas of complex flow and high risk
of spill.
Finally, classified images from Radarsat SAR are being used
to identify spill location, size and relative thickness of slicks.
This information is used to re-initialize the oil spill model at
the satellite image time.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
The SEACAST system and the use of near-real time imagery
for operational modelling in the context of managing and
mitigating oil spill damage in the coastal zone is described in
this paper. The discussion focuses on the relationship of the
different types of remote sensed data to the model
requirements and how these data are processed for
assimilation into a three-dimensional modelling system. The
method of distributing georeferenced SEACAST results to
spill response organizations over the Internet, as well as
global applications of this type of system are described.
OCEAN CIRCULATION MODEL
The circulation model is comprised of a three-dimensional
hydrodynamic model based on integrated. forms of the
Reynolds equations for turbulent flow and a fully coupled
transport-diffusion model based on conservation equations for
heat, salt and suspended solids. The layer-integrated forms of
the Reynolds equations have been derived for two conditions:
a surface layer defined by the conventional free-surface
kinematic boundary condition, superimposed upon a set of
fixed layers where the interfaces between layers are located at
specified depths below mean sea level. In the coastal model,
the surface layer thickness (typically 5 m) is sufficient to
contain the tidal variation in water level. The governing
hydrodynamic and transport-diffusion equations have the
form shown in equations (1) to (6), written in tensor notation
(Hodgins, 1976; Stronach et al., 1993).
In these equations the dependent variables are: uj =
horizontal velocity for i=1,2, vertical velocity for i=3, p =
density, © = dissolved or suspended substance, P; = mean
hydrostatic pressure gradient term, h = b-a, the layer
thickness at time t, p«u';u':» - Reynolds stress tensor, and
2e; J,k®jPuk = Coriolis force. Subscripts i,j,k denote
summation over indices 1,2 and 3. The layers are defined by
X3 —b(t) at the top of the layer and X3 — a at the bottom of
the layer; b is a function of time only for the top layer.
The turbulence closure models include a Mellor-Yamada
(1982) level-2 scheme for interfacial shear stress, which
incorporates a bulk Richardson number dependence, and the
Smagorinsky (1963) formulation for lateral shear stress,
dependent on local current shear.
427
