[cig-commits] commit 2474 by gassmoeller to /var/svn/dealii/aspect
dealii.demon at gmail.com
dealii.demon at gmail.com
Fri Apr 11 06:01:49 PDT 2014
Revision 2474
Extended GPlates cookbook section and updated figure.
U trunk/aspect/doc/manual/cookbooks/gplates/gplates-comparison.png
U trunk/aspect/doc/manual/manual.tex
http://www.dealii.org/websvn/revision.php?repname=Aspect+Repository&path=%2F&rev=2474&peg=2474
Diff:
Modified: trunk/aspect/doc/manual/cookbooks/gplates/gplates-comparison.png
===================================================================
(Binary files differ)
Modified: trunk/aspect/doc/manual/manual.tex
===================================================================
--- trunk/aspect/doc/manual/manual.tex 2014-04-11 12:57:17 UTC (rev 2473)
+++ trunk/aspect/doc/manual/manual.tex 2014-04-11 13:01:47 UTC (rev 2474)
@@ -5014,17 +5014,18 @@
allows for the use of two- and three-dimensional models incorporating the
GPlates velocities. Since the output by GPlates is threedimensional in any case,
spect{} internally handles the 2D model by rotating the model plane to the
-specified orientation and projecting the plate velocities into this plane. The
+orientation specified by the user and projecting the plate velocities into this plane. The
user specifies the orientation of the model plane by prescribing two points that
define a plane together with the coordinate origin (i.e. in the current
formulation only great-circle slices are allowed). The coordinates need to be in
spherical coordinates $ heta$ and $\phi$ with $ heta$ being the colatitude (0
at north pole) and $\phi$ being the longitude (0 at Greenwich meridian, positive
eastwards) both given in radians.
+The approach of identifying two points on the surface of the Earth along with
+its center allows to run computations on arbitrary two-dimensional slices
+through the Earth with realistic boundary conditions.
-The relevant section of the input file is then as follows (you will need
-to adjust the first part of the exttt{Data directory} below to where your
-spect{} installation is located):
+The relevant section of the input file is then as follows:
egin{lstlisting}[frame=single,language=prmfile,escapechar=\%]
subsection Model settings
set Prescribed velocity boundary indicators = 1:gplates % \index[prmindex]{Prescribed velocity boundary indicators} \index[prmindexfull]{Model settings!Prescribed velocity boundary indicators} %
@@ -5036,51 +5037,100 @@
subsection Boundary velocity model
subsection GPlates model
- set Data directory = .../data/velocity-boundary-conditions/gplates/ % \index[prmindex]{Data directory} \index[prmindexfull]{Boundary velocity model!GPlates model!Data directory} %
+ set Data directory = $ASPECT_SOURCE_DIR/data/velocity-boundary-conditions/gplates/ % \index[prmindex]{Data directory} \index[prmindexfull]{Boundary velocity model!GPlates model!Data directory} %
set Velocity file name = current_day.gpml % \index[prmindex]{Velocity file name} \index[prmindexfull]{Boundary velocity model!GPlates model!Velocity file name} %
set Time step = 1e6 % \index[prmindex]{Time step} \index[prmindexfull]{Boundary velocity model!GPlates model!Time step} %
set Point one = 1.5708,4.87 % \index[prmindex]{Point one} \index[prmindexfull]{Boundary velocity model!GPlates model!Point one} %
set Point two = 1.5708,5.24 % \index[prmindex]{Point two} \index[prmindexfull]{Boundary velocity model!GPlates model!Point two} %
+ set Interpolation width = 2000000 % \index[prmindex]{Interpolation width} \index[prmindexfull]{Boundary velocity model!GPlates model!Interpolation width} %
end
end
\end{lstlisting}
-\paragraph{Comparing and visualizing 2D models.}
+In the model settings subsection the user prescribes the boundary that is supposed to
+use the GPlates plugin. Although currently nothing forbids the user to use GPlates plugin for other
+boundaries than the surface, its current usage and the provided sample data only make sense
+for the surface of a spherical shell (boundary number 1 in the above provided parameter file).
+In case you are familiar with this kind of modelling and the plugin you could however also use it to prescribe mantle
+movements extit{below} a lithosphere model. All plugin specific options may be set in the
+section~
ef{parameters:parameters:Boundary_20velocity_20model}. Possible options include the data directory
+and file name of the velocity file/files, the time step (in model units, mostly seconds or years depending on the
+`` exttt{Use years in output instead of seconds}'' flag) and the points that define the 2D plane.
+The parameter `` exttt{Interpolation width}'' is used to smooth the provided velocity files by
+a moving average filter. All velocity data points within this distance are averaged to determine the
+actual boundary velocity at a certain mesh point. This parameter is usually set to 0 (no interpolation, use
+nearest velocity point data) and is only needed in case the original setting is unstable or slowly converging.
-The approach of identifying two points on the surface of the Earth along with
-its center allows to run computations on arbitrary two-dimensional slices
-through the Earth with realistic boundary conditions. Fig.~
ef{fig:gv-1} shows an example
-of three such computations the results of which are plotted simultaneously in a
-single three-dimensional figure.
+\paragraph{Comparing and visualizing 2D and 3D models.}
+The implementation of plate velocities in both two- and three-dimensional model
+setups allows for an easy comparison and test for common sources of error
+in the interpretation of model results. The left top figure in Fig.~
ef{fig:gv-1}
+shows an modification of the above presented parameter file by setting
+`` exttt{Dimension = 3}'' and `` exttt{Initial global refinement = 3}''.
+The top right plot of Fig.~
ef{fig:gv-1} shows an example of three independent
+two-dimensional computations of the same reduced resolution. The models were prescribed
+to be orthogonal slices by setting:
+egin{lstlisting}[frame=single,language=prmfile,escapechar=\%]
+ set Point one = 3.1416,0.0 % \index[prmindex]{Point one} \index[prmindexfull]{Boundary velocity model!GPlates model!Point one} %
+ set Point two = 1.5708,0.0 % \index[prmindex]{Point two} \index[prmindexfull]{Boundary velocity model!GPlates model!Point two} %
+\end{lstlisting}
+and:
+egin{lstlisting}[frame=single,language=prmfile,escapechar=\%]
+ set Point one = 3.1416,1.5708 % \index[prmindex]{Point one} \index[prmindexfull]{Boundary velocity model!GPlates model!Point one} %
+ set Point two = 1.5708,1.5708 % \index[prmindex]{Point two} \index[prmindexfull]{Boundary velocity model!GPlates model!Point two} %
+\end{lstlisting}
+
+The results of these models are plotted simultaneously in a single three-dimensional figure
+in their respective model planes. The necessary information
+to rotate the 2D models to their respective planes (rotation axis and angle) is provided by the
+GPlates plugin in the beginning of the model output. The bottom plot
+of Fig.~
ef{fig:gv-1} finally shows the results of the original \url{cookbooks/gplates-2d.prm}
+also in the three mentioned planes.
+
+Now that we have model output for otherwise identical 2D and 3D models with equal resolution and additional 2D output
+for a higher resolution an interesting question to ask would be: What additional information can be created by
+either using three-dimensional geometry or higher resolution in mantle convection models with prescribed boundary velocities.
+As one can see in the comparison between the top right and bottom plot in Fig.~
ef{fig:gv-1} additional resolution clearly
+improves the geometry of small scale features like the shape of up- and downwellings as well as the maximal temperature
+deviation from the background mantle. However, the limitation to two dimensions leads to inconsistencies,
+that are especially apparent at the cutting lines of the individual 2D models.
+Note for example that the Nacza slab of the South American subduction zone is only
+present in the equatorial model plane and is not captured in the polar model plane west
+of the South American coastline. The (coarse) three-dimensional model on the other hand
+shows the same location of up- and downwellings but additionally provides a consistent solution
+that is different from the two dimensional setups. Note that the Nazca slab is subducting eastward,
+while all 2D models (even in high resolution) predict a westward subduction.
+
+Finally we would like to emphasize that these models (especially the used material model)
+are way too simplified to draw any scientific conclusion out of it. Rather it is thought
+as a proof-of-concept what is possible with the dimension independent approach of
+spect{} and its plugins.
+
egin{figure}
- \includegraphics[width=0.48 extwidth]{cookbooks/gplates/gplates-comparison.png}
+ \includegraphics[width= extwidth]{cookbooks/gplates/gplates-comparison.png}
\hfill
- % \includegraphics[width=0.48 extwidth]{cookbooks/gplates/gplates-3d.png}
-
More information about the CIG-COMMITS
mailing list