The software tool, ANUGA \cite{ON:modsim}, has been used to estimate 
the inundation extent
and associated water level at various points in space and time.
ANUGA has been developed by GA and the Australian National University
(ANU) to solve the nonlinear shallow water
wave equation using the finite volume technique\footnote{The finite volume
technique belongs to the class of computational fluid dynamic (CFD)
methods which is based on discretizing the study area in
control ''volumes''. The method ensures that conservation
of mass and horizontal momentum is satisfied for
each control volume.
An advantage of this technique is that the discretization
can be changed
according to areas of interest and that wetting and drying
is treated robustly as part of the numerical scheme.}. 
ANUGA is continually being refined and validated to ensure
the modelling approximations are as accurate as possible.
However, model sensitivity to errors in bathymetric data, 
frictional resistance of the seafloor and the size of the tsunamigenic event are not well understood and the topic of ongoing research. 
As such, the current results are preliminary.

The following information is required to undertake the 
inundation modelling;

\begin{itemize}
\item onshore and offshore elevation data (topographic and bathymetric data, 
see Section \ref{sec:data}),
\item initial conditions, such as initial water levels (e.g. determined by tides),
\item boundary conditions (the tsunami source as described in 
Section \ref{sec:methodology}), and
\item computational requirements relating to the mesh construction.
\end{itemize}
 
The initial conditions used for this scenario are MSL, HAT and LAT which were
defined in Section \ref{sec:data}.
The dynamics of
tidal effects (that is, the changes in water height over time for
the entire study area) are not currently modelled.
Sea floor friction will generally provide resistance to the water flow
and thus reduce the impact somewhat. However, limited
research has been carried out to determine 
the friction coefficients, and 
thus it has not been incorporated
in the scenario. The 
results are therefore likely to be over estimates.