= GPlates Tutorial: Velocity Fields == Authors - Sabin Zahirovic^1^ - Kara J. Matthews^1^ - R. Dietmar Muller^1^ ^1^EarthByte Research Group, School of Geosciences, University of Sydney, Australia == Aim This tutorial is designed to teach the user how to generate velocity fields for topologically closed polygons. These closed and dynamic polygons must be created in *GPlates* first. See the "Topology Tools" section in the user documentation for more information on creating closed dynamic polygons. Velocity output can be used as plate kinematic input for geodynamic modelling codes. *GPlates* is capable of directly generating time-dependant velocity fields for the mantle convection modelling code CitcomS. *GPlates* can also be linked to the TERRA mantle convection code. Importantly, users can extract the required data from the velocity output and make it compatible with their own particular modelling code. All of the required files for this tutorial, including the dynamically closing plate polygons, are contained in the accompanying data bundle. == Included Files The data bundle for this tutorial, *Velocity Fields*, can be located http://www.earthbyte.org/Resources/GPlates_tutorials/Velocities_Tutorial/SampleData/Velocity_Fields_Tutorial_Data.zip[here]. Create a folder named +Velocity_Fields+, unarchive the datasets into this folder, and remember the location so you can access the datasets later. See http://www.earthbyte.org/Resources/earthbyte_gplates.html[] for EarthByte data sets. == Background .What is a mesh? What is it used for? A mesh is an equally spaced grid of co-ordinate points. The distance between the points is equal on a spherical surface such as the Earth^1^. The use of a mesh is critical in linking plate kinematics and geodynamic models. For example, *CitcomS* (a mantle convection model) uses a static mesh that measures the velocity of the tectonic plates through time at each node. *GPlates* is capable of generating time-dependant velocity fields for CitcomS. *GPlates* can also be linked to the TERRA mantle convection code. ============================ ^1^In fact the shape of the Earth is not spherical, but rather an ellipsoid. However, the deviation from sphere to ellipsoid is small -- and for the purpose of modelling, the sphere is the simplest case. Modelling requires millions of computations, which would be much slower if the actual ellipsoid shape of the Earth was implemented. As a result, the sphere is the closest estimate requiring simpler computations. ============================= image::velocity_tut_image_1.JPG[] _________ *Figure 1:* CitcomS 9-mesh global cap and node distribution _________ image::velocity_tut_image_2.JPG[] _________ *Figure 2:* CitcomS 9-mesh node distribution _________ .CitcomS Mesh Specifications CitcomS meshes can be either regional or global. A global mesh is composed of 12 diamond-shaped "caps" numbered from 0 to 11. The density of the mesh nodes can be adjusted. The global distribution of "caps" can be seen in Figure 1. The rectangular projection does not preserve the uniform distance between the nodes, whereas an orthographic projection is a better representation of the inter-nodal distance (Figure 2). The CitcomS manual is an extensive and useful document containing additional information. It is available http://www.geodynamics.org/cig/software/packages/mc/citcoms/[here]. image::velocity_tut_image_3.JPG[] _________ *Figure 3.* CitcomS Mesh Cap 0 with 9 X 9 nodes (Left), and Caps 0 to 11 global distribution (Right) _________ In Exercise 2 we will make each cap comprise of 9 X 9 nodes (Figure 3). This is a reasonably coarse mesh, especially for mantle convection modelling purposes. Mesh caps are composed of n^2^ nodes (n being the number of points along each side of the cap), and so processing the velocity of denser grids increases proportionally to the square of the node numbers along each cap side (See Figure and Table 4). .Nodes [width="75%",cols="3^,^3,^3,3^",options="header"] |========================================================= |Nodes per cap side |Nodes in One Cap |Total number of Mesh nodes |Increase in global density from 9X 9 cap |9 |81 |972 |0% |10 |100 |1200 |23% |11 |121 |1452 |50% |12 |144 |1728 |77% |... |... |... |... |50 |2500 |30000 |2990% |========================================================= image::velocity_tut_image_4.JPG[] _________ *Figure 4.* The increase in mesh density follows the square law, meaning that computation time of velocities increases significantly with increasing mesh density. _________ == Exercise A - Generating a Mesh File *GPlates* currently supports the generation of global mesh files. These mesh files consist of 12 caps that cover the globe (see CitcomS Mesh Specifications above). Each point in the mesh is a "sampling" location of velocity, for the tectonic plate in which it is located. This is a CitcomS standard. Ultimately the following steps will allow you to create velocity fields as input for mantle convection modelling in CitcomS. Information can be extracted for other purposes, CitcomS modelling is merely the example used here. The density of the meshpoints can vary, and is dependent on your application. For the purpose of this tutorial, we will use a coarse mesh of 9x9 nodes. - Open *GPlates* - *Reconstruction -> GenerateMeshCap ...* (Figure 5) image::velocity_tut_image_5_2.JPG[] _________ *Figure 5.* How to open the Generate Mesh Caps window from the main menu. _________ - Our mesh file will have a 9x9 node resolution, so enter +9+ into the *nodeX* box (9 will automatically be entered in the *nodeY* box). See Figure 6. image::velocity_tut_image_6_2.JPG[] _________ *Figure 6.* The Generate Mesh Caps window enables you to define the resolution and saving destination of your mesh file. _________ - Enter the destination where you would like to save the mesh files (it is suggested that you make a directory called +Velocity_Tutorial+) -> *Generate* 12 mesh files with 9x9 nodes have been saved into the directory you specified. These files have also been loaded into *GPlates*. The files are named 9.mesh.X.gpml, where X is the cap number, ranging from 0-11. You can see that they are loaded by opening the 'Manage Feature Collections' window (*File -> Manage Feature Collections*). In the next exercise we will visualise and export plate velocities. == Exercise B - Rotations and Dynamically Closing Polygons *GPlates* uses rotation files to reconstruct geometries through time. The geometry features are a set of intersecting lines, each assigned a Plate ID and thus move according to the information in the rotation file. These geometries can be used to create a set of dynamically-closed plate polygons. The result of this is that the surface of the Earth is split into a discrete number of tectonic plates to cover the temporal span of the plate kinematic model. The velocity of each plate through time will be tracked in *GPlates* by the mesh files we created in the previous exercise. At each time interval in *GPlates* the mesh nodes are assigned Plate IDs according to the plate polygon on which they are located. These velocities can be used as boundary conditions for mantle convection models, including CitcomS and TERRA. In this way *GPlates* provides a link between plate kinematics and mantle dynamic processes. In order to generate and export plate velocities through time, *GPlates* expects a rotation file and a file of dynamically-closing plate polygons, along with our mesh files. The sample data contains all the necessary input files. For more information please visit the *GPlates* User Documentation online. - Click *File -> Manage Feature Collections* - Click *Open File* and navigate to the folder containing the geometry and rotation files. We will select the following geometry (GPML) and rotation (ROT) files: . +Topological_plate_boundaries_hybridhs_20100415.gpml+ . +Topological_plate_boundaries_hybridhs_20100225.rot+ The boundaries and the associated polygons are now displayed. In this model absolute plate rotations are a combination of a moving hotspot reference frame (O'Neill et al., 2005) for 0-100 Ma and a fixed hotspot ref frame from 100-140 Ma (Muller et al, 1993). *GPlates* automatically displays the velocity vectors based on the dynamically closing polygon and rotation files (Figure 7). That means your data MUST contain topological polygons created in GPlates with assigned Plate IDs, as it reads the rotation file and calculates velocity vectors on the fly for each mesh node. The latest GPML and ROT files can also be downloaded from the following website: http://www.gps.caltech.edu/~gurnis/GPlates/gplates.html. Note: filenames may differ. image::velocity_tut_image_7_2.JPG[] _________ *Figure 7.* The *GPlates* globe displaying polygon outlines and velocity vectors. _________ You may want to load up a coastline file to help identify the regions of the world. In Figure 7, the red arrows correspond to the velocity vectors of the Indo-Australian Plate. The velocity vectors indicate both the magnitude and direction of motion. There are two alternative models available for comparison: One is entirely based on a fixed hotspot model (Muller et al., 1993): . +Topological_plate_boundaries_fixhs_20100415.gpml+ . +Topological_plate_boundaries_fixhs_20100225.rot+ The second is a hybrid model based on a moving hotspot model from 0-100 Ma (O'Neill et al., 2005) and a paleomagnetic model from 100-140 Ma (Torsvik et al., 2008): . +Topological_plate_boundaries_pmag_movhs_20100415.gpml+ . +Topological_plate_boundaries _pmag_movhs_20100225.rot+ You can only investigate one model at a time, so you need to unload one data/rotation file and load the next to investigate differences. However, the plate velocity display is quite useful to try to assess shortcomings of one or the other model. You can reconstruct the globe and see how the velocities change through time if you like. == Exercise C - Exporting Velocities *GPlates* can export animations in a number of formats. For the purposes of creating velocity fields, only one is of interest to us. The output files generated will be in GPML format. Note that CitcomS requires simple ASCII text files of the velocity fields as input, therefore if you plan to use CitcomS you will need to convert the GPML files to ASCII. - *Reconstruction -> Export...* (Figure 8) image::velocity_tut_image_8_2.JPG[] _________ *Figure 8.* Navigating the main menu to open the Export Animation window. _________ - We will export velocities from 10 Ma to present-day, with an increment of 1 Myr per frame. Therefore, all you need to change in the 'Range' box (top) is the 'Animate from' time to 10.00 (Figure 9). image::velocity_tut_image_9_2.JPG[] _________ *Figure 9.* The Export Animation window enables you to set the temporal parameters of your export. See top section entitled 'Range'. _________ - We must now specify what files we wish *GPlates* to generate -> click *Add*. - In the 'Add Export' window select Colat/lon Mesh Velocities and then the GPML format (Figure 10) -> *Add -> Close* (close the 'Add Export' window as we do not wish to generate any additional files. image::velocity_tut_image_10_2.JPG[] _________ *Figure 10.* The Add Export window enables you to choose which files GPlates will generate for your time interval. _________ - Choose the target directory where the output will be created and then click Begin. The files will now be generated. To save time we have only selected 10 Myr. The velocity files are now saved in your selected target directory. A velocity file is generated for very cap, every 1 Myr (as this was the interval chosen). == References Muller, R.D., Royer, J.-Y. and Lawver, L.A., 1993. Revised plate motions relative to the hotspots from combined Atlantic and Indian Ocean hotspot tracks. Geology, 16: 275-278. O'Neill, C., Muller, R.D. and Steinberger, B., 2005. On the uncertainties in hotspot reconstructions, and the significance of moving hotspot reference frames. Geochemistry, Geophysics, Geosystems, 6, Q04003, doi:10.1029/2004GC000784, 1-35. Torsvik, T., Muller, R.D., Van der Voo, R., Steinberger, B. and Gaina, C., 2008. Global Plate Motion Frames: Toward a unified model. Reviews in Geophysics, 46, RG3004, 1-44.