Environmental Engineering Reference
In-Depth Information
available for practical evaluation of how well (or poorly) any particular model will
predict spill transport or how cumulative effects of uncertainty should be evaluated.
This hole in our knowledge is the motivating focus of our work.
This chapter presents an overview of a generic oil spill forecast system (Sect.
5.2
),
followed by a discussion of salient characteristics of oil spill models (Sect.
5.3
),
sources of uncertainty (Sect.
5.4
), model parameters affecting uncertainty (Sect.
5.5
),
a proposed systems-level approach for producing a range of oil spill forecasts
(Sect.
5.6
), a multi-model system for updating forecasts (Sect.
5.7
), discussion of
uncertainty evaluation methods (Sect.
5.8
), and methods for integrating probability
estimates within oil spill models (Sect.
5.9
).
5.2 Oil Spill Forecast Systems
Predicting the fate and transport of an oil spill requires a system of models, including
forecast models for wind, water currents, waves, oil advection/dispersion, and the
weathering processes that alter oil properties [
24
]. A monolithic model that predicts
all driving/response processes is simply impractical to build and maintain, so opera-
tional models generally use forecast models for wind, waves, and currents originally
designed for other purposes. An efficient operational system requires automated link-
ing of models (i.e. output from one model is input to another), along with integration
of real-world observational data (e.g. [
27
,
47
]). For rapid use in emergency opera-
tions, an oil spill forecast system also benefits from a user interface displaying the
model predictions as a geo-referenced visualization that can be readily interpreted
by oil spill response personnel. These components can be generally structured as in
Fig.
5.1
with three different computational modules, (i) geophysical forcing, (ii) oil
transport and chemistry, and (iii) visualization; which are linked to COS data and
known (or estimated) information about the oil spill source [
46
]. Note that the spill
forecast system illustrated in Fig.
5.1
has a unidirectional flow of data: there are no
feedbacks from the oil spill model to the geophysical forcing models. However, we
know that surface oil can affect wave development and the transfer of wind energy
into the water, which in turn affects local surface currents and near-surface turbu-
lence. Given the uncertainties in present modelling systems, it is likely that such
feedback effects are of minor consequence, but this remains an area where (to our
knowledge) there have been no clear quantitative evaluations of these phenomena or
applications within operational models.
5.2.1 Oil Spill Data
An oil spill model requires data for the spill location, event time, spilled volume, oil
type, fraction of oil at the water surface, and information on the ocean conditions that
can affect the near-field behaviour. Most oil spill transport models are designed to
Search WWH ::
Custom Search