Changeset 7451


Ignore:
Timestamp:
Aug 30, 2009, 4:45:41 AM (11 years ago)
Author:
jakeman
Message:

john: update tsunami validation paper

Location:
anuga_work/publications/boxing_day_validation_2008
Files:
7 edited

Legend:

Unmodified
Added
Removed
  • anuga_work/publications/boxing_day_validation_2008/appendix.tex

    r7450 r7451  
    2121\end{figure}
    2222
     23\begin{figure}[ht]
     24\begin{center}
     25%\includegraphics[width=6cm,keepaspectratio=true]{sensitivity_f0_0003_depth}
     26%\includegraphics[width=6cm,keepaspectratio=true]{sensitivity_f0_03_depth}
     27\includegraphics[width=\textwidth,keepaspectratio=true]{figures/sensitivity_friction}
     28\caption{Model results for different values of Manning's friction
     29  coefficient shown to assess sensitivities.
     30  The reference inundation extent for a
     31  friction value of 0.01 is shown in Figure
     32  \protect \ref{fig:reference_model} (left).  The left and right images
     33  show the inundation results for friction values of 0.0003 and
     34  0.03 respectively. The inundation extent increases for the lower
     35  friction value while the higher slows the flow and decreases the
     36  inundation extent. Ideally, friction should vary across the entire
     37  domain depending on terrain and vegetation, but this is beyond the
     38  scope of this study.}
     39\label{fig:sensitivity_friction}
     40\end{center}
     41\end{figure}
     42
     43%\begin{figure}[ht]
     44%\begin{center}
     45%\includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_f0_0003_speed}
     46%\includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_f0_03_speed}
     47%\caption{The maximal flow speeds for the same model parameterisations
     48%  found in Figure \protect \ref{fig:sensitivity_friction}. The
     49%  reference flow speeds are shown in Figure \protect
     50%  \ref{fig:reference_model} (right).}
     51%\label{fig:sensitivity_friction_speed}
     52%\end{center}
     53%\end{figure}
     54% John: I do not think we need to show sensitivity to flow speeds
    2355
    2456
     
    4375
    4476
    45 \begin{figure}[ht]
    46 \begin{center}
    47 \includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_minus10cm_speed}
    48 \includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_plus10cm_speed}
    49 \caption{The maximal flow speeds for the same model parameterisations
    50   found in Figure \protect \ref{fig:sensitivity_boundary}. The
    51   reference flow speeds are shown in Figure \protect
    52   \ref{fig:reference_model} (right).}
    53 \label{fig:sensitivity_boundary_speed}
    54 \end{center}
    55 \end{figure}
     77%\begin{figure}[ht]
     78%\begin{center}
     79%\includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_minus10cm_speed}
     80%\includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_plus10cm_speed}
     81%\caption{The maximal flow speeds for the same model parameterisations
     82%  found in Figure \protect \ref{fig:sensitivity_boundary}. The
     83%  reference flow speeds are shown in Figure \protect
     84%  \ref{fig:reference_model} (right).}
     85%\label{fig:sensitivity_boundary_speed}
     86%\end{center}
     87%\end{figure}
    5688
    5789\begin{figure}[ht]
     
    78110\end{figure}
    79111
    80 
    81 \begin{figure}[ht]
    82 \begin{center}
    83 %\includegraphics[width=6cm,keepaspectratio=true]{sensitivity_f0_0003_depth}
    84 %\includegraphics[width=6cm,keepaspectratio=true]{sensitivity_f0_03_depth}
    85 \includegraphics[width=\textwidth,keepaspectratio=true]{figures/sensitivity_friction}
    86 \caption{Model results for different values of Manning's friction
    87   coefficient shown to assess sensitivities.
    88   The reference inundation extent for a
    89   friction value of 0.01 is shown in Figure
    90   \protect \ref{fig:reference_model} (left).  The left and right images
    91   show the inundation results for friction values of 0.0003 and
    92   0.03 respectively. The inundation extent increases for the lower
    93   friction value while the higher slows the flow and decreases the
    94   inundation extent. Ideally, friction should vary across the entire
    95   domain depending on terrain and vegetation, but this is beyond the
    96   scope of this study.}
    97 \label{fig:sensitivity_friction}
    98 \end{center}
    99 \end{figure}
    100 
    101 \begin{figure}[ht]
    102 \begin{center}
    103 \includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_f0_0003_speed}
    104 \includegraphics[width=6cm,keepaspectratio=true]{figures/sensitivity_f0_03_speed}
    105 \caption{The maximal flow speeds for the same model parameterisations
    106   found in Figure \protect \ref{fig:sensitivity_friction}. The
    107   reference flow speeds are shown in Figure \protect
    108   \ref{fig:reference_model} (right).}
    109 \label{fig:sensitivity_friction_speed}
    110 \end{center}
    111 \end{figure}
  • anuga_work/publications/boxing_day_validation_2008/conclusion.tex

    r7450 r7451  
    66utilises the uniquely large amount of observational data for model
    77comparison obtained during, and immediately following, the
    8 Sumatra--Andaman tsunami of 26 December 2004. Unlike the small
    9 number of existing benchmarks, the proposed test validates all three
    10 stages of tsunami evolution - generation, propagation and
    11 inundation. In an attempt to provide higher visibility and easier
     8Sumatra--Andaman tsunami of 26 December 2004. The proposed benchmark is intended to aid validation of tsunami inundation, which is the most important stage
     9of tsunami evolution. However individual tests are presented to
     10facilitate model evaluation for the generation and propagation
     11phases as well. In an attempt to provide higher visibility and easier
    1212accessibility for tsunami benchmark problems, the data used to
    1313construct the proposed benchmark is documented and freely available at
    1414\url{http://tinyurl.com/patong2004-data}.
    1515
    16 This study also shows that the tsunami impact modelling methodology
    17 adopted is credible and able to predict inundation extents with reasonable
    18 accuracy.  An associated aim of this paper was to further validate the
    19 hydrodynamic modelling tool \textsc{anuga} which is used to simulate
    20 the tsunami inundation. Model predictions
    21 matched well the geodetic measurements of the Sumatra--Andaman earthquake,
    22 altimetry data from the \textsc{jason}, eye-witness accounts of wave
    23 front arrival times and flow speeds and a detailed inundation survey
    24 of Patong Bay, Thailand.
     16 An associated aim of this paper was to further validate the
     17\textsc{ursga--anuga} tsunami modelling methodology employed by Geoscience
     18Australiawhich is used to simulate the tsunami inundation.
     19This study ashows that the tsunami  modelling methodology adopted is credible
     20and able to predict detailed inundation extents with reasonable accuracy.
     21Model predictions matched well a detailed inundation survey
     22of Patong Bay, Thailand as well as altimetry data from the \textsc{jason},
     23eye-witness accounts of wave front arrival times and onshore flow speeds.
    2524
    2625A simple sensitivity analysis was performed to assess the influence of
     
    2827the presence of buildings and other structures on the model
    2928predictions. Of these three, the presence of buildings was shown to
    30 have the greatest influence on
    31 the simulated inundation extent. The value of friction and small
     29have the greatest influence on the simulated inundation extent. This result
     30indicates that the influence of human-made structures should be included,
     31where possible in any future studies. The value of friction and small
    3232perturbations in the waveheight at the \textsc{anuga} boundary have
    3333comparatively little effect on the model results.
  • anuga_work/publications/boxing_day_validation_2008/data.tex

    r7450 r7451  
    1616model validity. In fact for non-physics based models it may not be possible
    1717 to validate the generation and propagation phases of tsunami evolution.
    18 For physics-based models evaluation of the model during the generation and
    19 propagation phases is still useful. If a model is physics-based one should
    20 ensure that all physics are being modelled accurately. Moreover evaluation
    21 of all three stages of tsunami evolution can help identify the cause of any
    22 discrepancies between modelled and observed inundation. Consequently in this
    23  section we present data not only to facilitate validation of inundation but
    24 to also aid in assessment of tsunami generation and propagation.
     18However, for physics-based models evaluation of the model during the generation
     19 and propagation phases is still useful. If a model is physics-based one
     20should ensure that all physics are being modelled accurately. Moreover
     21evaluation of all three stages of tsunami evolution can help identify the
     22cause of any discrepancies between modelled and observed inundation.
     23Consequently in this section we present data not only to facilitate
     24validation of inundation but to also aid the assessment of tsunami
     25generation and propagation.
    2526
    2627\subsection{Generation}\label{sec:gen_data}
     
    153154high quality field measurements are also required. For the proposed
    154155benchmark a high resolution topography data set and a high quality inundation
    155  survey map from the (FIXME(John): what data set was used to generate the topogaphy? RICHARD )
     156 survey map from the
    156157Coordinating Committee Co-ordinating Committee for Geoscience Programmes
    157158in East and Southeast Asia (CCOP) (\cite{szczucinski06}) was obtained
     
    231232\includegraphics[width=\textwidth,keepaspectratio=true]{figures/gauges.jpg}
    232233\caption{Location of timeseries extracted from the model output. FIXME(John):
    233 should we combine inundation map with gauages map?}
     234should we combine the inundation map with the gauages map?}
    234235\label{fig:gauge_locations}
    235236\end{center}
     
    296297   the eye-witness videos, that fall within the bounds obtained from
    297298   the videos.
     299 \item reproduce the \textsc{jason} satellite altimetry sea surface
     300   anomalies (see Section~\ref{sec:data_jason}),
    298301 \item reproduce the vertical deformation observed in north-western
    299302   Sumatra and along the Nicobar--Andaman islands (see
    300303   Section~\ref{sec:gen_data}),
    301  \item reproduce the \textsc{jason} satellite altimetry sea surface
    302    anomalies (see Section~\ref{sec:data_jason}),
    303304\end{itemize}
    304305
  • anuga_work/publications/boxing_day_validation_2008/introduction.tex

    r7450 r7451  
    117117 localised and highly detailed spatially distributed assessment of
    118118modelled inundation. To the authors knowledge it is also the first benchmark to
    119 assess model inundation under influenced by numerous human structures. Eye-witness videos also allow the qualitative assessment of onshore flow patterns.
     119assess model inundation under influenced by numerous human structures.
     120Eye-witness videos also allow the qualitative assessment of onshore flow
     121patterns.
    120122
    121123An associated aim of this paper is to illustrate the use of this new
    122 benchmark to validate a dedicated inundation model called
    123 \textsc{anuga} used by Geoscience Australia. A description of
    124 \textsc{anuga} is given in Section~\ref{sec:models} and the validation
     124benchmark to validate the three step modelling methodology employed by
     125Geoscience Australia to model tsunami inundation. A description of the model
     126components is provided in Section~\ref{sec:models} and the validation
    125127results are given in Section~\ref{sec:results}.
    126128
     
    132134However, model uncertainty should not be ignored. Section
    133135~\ref{sec:sensitivity} provides a simple analysis that can
    134 be used to investigate the sensitivity of model predictions to model
    135 parameters.
     136be used to investigate the sensitivity of model predictions to a number
     137of model parameters.
  • anuga_work/publications/boxing_day_validation_2008/paper.tex

    r7450 r7451  
    5353propagation and a detailed inundation survey of Patong city, Thailand
    5454to compare model and observed inundation. Furthermore we utilise this
    55 benchmark to further validate the hydrodynamic modelling tool
    56 \textsc{ursga--anuga} which is used to simulate the tsunami
     55benchmark to further validate the \textsc{ursga--anuga} modelling methodology
     56 used by Geoscience Australia to simulate the tsunami
    5757inundation. Important buildings and other structures were incorporated
    5858into the underlying computational mesh and shown to have a large
  • anuga_work/publications/boxing_day_validation_2008/results.tex

    r7450 r7451  
    22This section presents a validation of the modelling practice of Geoscience
    33Australia against the new proposed benchmarks. The criteria outlined
    4 in Section~\ref{sec:checkList} are addressed for each of the three stages
    5 of tsunami evolution.
     4in Section~\ref{sec:checkList} are addressed.S
    65
    76\subsection{Generation}\label{modelGeneration}
     
    4140not surprising, since the original slip model was chosen
    4241by~\cite{chlieh07} to fit the motion and seismic data well.
    43 
     42Consequently the replication of the generation data has little meaning for
     43the model structure presented in Section~\ref{sec:models}. But for
     44uncalibrated source models or source models calibrated on other data
     45this test of generation would be more meaningful.
    4446%
    4547%This does demonstrate, however, that \textsc{edgrn} and our modified version of
     
    129131\end{center}
    130132\end{figure}
    131 FIXME (Jane): This graph does not look nice. The legend URS Model should
    132 be URSGA model.
    133 
    134 \subsection{Inundation}
     133FIXME (Jane): This graph does not look nice.
     134
    135135After propagating the tsunami in the open ocean using \textsc{ursga},
    136136the approximated ocean and surface elevation and horisontal flow
     
    153153region in Patong Bay. The coarse resolution was chosen to be
    154154commensurate with the model output from the \textsc{ursga} model
    155 (FIXME - this has to be clearly stated in ursga section) RICHARD
    156155while the latter was chosen to match the available resolution of topographic
    157156data and building data in Patong city.
     
    188187reasonable.
    189188
    190 Maximum onshore inundation depth was computed from the model
     189\subsection{Inundation}
     190The \textsc{anuga} simulation described in the previous section and used to
     191 model shallow water propgation also predicts
     192inundation. Maximum onshore inundation depth was computed from the model
    191193throughout the entire Patong Bay region and used to generate
    192194a measure of the inundated area.
     
    241243\rho_{in}=\frac{A(I_m\cap I_o)}{A(I_o)}
    242244\end{equation}
    243 representing the ratio $\rho_{in}$ of the observed
    244 inundation region $I_o$ captured by the model $I_m$. Another useful
     245representing the ratio of the area of the observed
     246inundation region $I_o$ and the area of the observed inundation region
     247captured by the model $I_m$. Another useful
    245248measure is the fraction of the modelled inundation area that falls
    246249outside the observed inundation area given by the formula
     
    262265missing data in the field survey data itself. The impact of some of
    263266these sources of uncertainties are is investigated in
    264 Section~\ref{sec:sensitivity}
     267Section~\ref{sec:sensitivity}.
    265268
    266269\subsection{Eye-witness accounts}
     
    284287\begin{figure}[ht]
    285288\begin{center}
    286 \includegraphics[width=\textwidth,keepaspectratio=true]{gauges_hotels_depths.jpg}
    287 \includegraphics[width=\textwidth,keepaspectratio=true]{gauges_hotels_speed.jpg}
     289\includegraphics[width=\textwidth,keepaspectratio=true]{figures/gauges_hotels_depths.jpg}
     290\includegraphics[width=\textwidth,keepaspectratio=true]{figures/gauges_hotels_speed.jpg}
    288291\caption{Time series obtained from the two onshore locations, North and South,
    289292shown in Figure \protect \ref{fig:gauge_locations}.}
  • anuga_work/publications/boxing_day_validation_2008/sensitivity.tex

    r7450 r7451  
    22\section{Sensitivity Analysis}
    33\label{sec:sensitivity}
     4The numerical models used to simulate tsunami impact
     5are computationally intensive and high resolution models of the entire
     6evolution process will often take a number of days to
     7run. Consequently, the uncertainty in model predictions is difficult to
     8quantify as it would require a very large number of runs.
     9However, model uncertainty should not be ignored. The aim of this section is
     10not to provide a detailed investigation of sensitivity but to rather
     11illustrate that changes in important parameters of the \textsc{usrga--anuga}
     12model  produce behaviour consistent with the known physics and that
     13small changes in these parameters produce bounded variations in the output.
     14
    415This section investigates the effect of different values of Manning's
    516friction coefficient, changing waveheight at the 100 m depth contour,
    617and the presence and absence of buildings in the elevation dataset on
    7 model maximum inundation. The reference model is the one reported in
    8 Figure~\ref{fig:inundationcomparison1cm} (right) with a friction coefficient of 0.01,
    9 buildings included and the boundary condition produced by the
     18model maximum inundation.
     19
     20The reference model is the one reported in
     21Figure~\ref{fig:inundationcomparison1cm} (right) with a friction coefficient of 0.01, buildings included and the boundary condition produced by the
    1022\textsc{ursga} model.
    1123
     
    2032we simulated the maximum onshore inundation using a Manning's
    2133coefficient of 0.0003 and 0.03. The resulting inundation maps are
    22 shown in Figure~\ref{fig:sensitivity_friction} and the maximum flow
    23 speeds in Figure~\ref{fig:sensitivity_friction_speed}. These figures
    24 show that the on-shore inundation extent decreases with increasing
     34shown in Figure~\ref{fig:sensitivity_friction}
     35% and the maximum flow speeds in Figure~\ref{fig:sensitivity_friction_speed}.
     36 The figure, along with Table~\ref{table:inundationAreas},
     37shows that the on-shore inundation extent decreases with increasing
    2538friction and that small perturbations in the friction cause bounded
    2639changes in the output. This is consistent with the conclusions of
     
    3346The effect of the wave height used as input to the inundation model
    3447\textsc{anuga} was also investigated.
    35 Figure~\ref{fig:sensitivity_boundary} indicates that the inundation
     48Figure~\ref{fig:sensitivity_boundary} and  Table~\ref{table:inundationAreas}
     49indicate that the inundation
    3650severity is directly proportional to the boundary waveheight but small
    3751perturbations in the input wave height of 10 cm appear to have little
     
    4963The presence or absence of physical buildings in the elevation model was also
    5064investigated.
    51 Figure~\ref{fig:sensitivity_nobuildings} shows the inundated area and
    52 the associated maximum flow speeds
    53 in the presence and absence of buildings. It
    54 is apparent that densely built-up areas act as
    55 dissipators greatly reducing the inundated area. However, flow speeds
    56 tend to increase in passages between buildings.
     65Figure~\ref{fig:sensitivity_nobuildings} shows the inundated area
     66%and the associated maximum flow speeds
     67in the presence and absence of buildings. From
     68Table~\ref{table:inundationAreas} it is apparent that densely built-up
     69areas act as dissipators greatly reducing the inundated area.
     70This result suggest that, when possible the presence of human-made structures
     71should be included into the model topography. Furthermore this result also
     72indicates that simply matching point sites with much lower resolution meshes
     73than used here is an over simplification. Such simulations cannot capture the
     74fine detail that so clearly affects inundation.
     75%However, flow speeds tend to increase in passages between buildings.
    5776 
    5877
Note: See TracChangeset for help on using the changeset viewer.