Written and maintained by CASRAI Editorial Board
Published
On this page
GPlates is open-source desktop software for reconstructing and visualising tectonic plates and associated geological data through time. It combines plate-motion models with geographic information system capabilities. For research, the key distinction is between the software, the reconstruction model it runs, and the evidence preserved with its outputs.
Short answer
GPlates helps researchers examine where geological features were located and how they moved relative to other features in the past. Its interactive globe and maps connect geographical observations to a reconstruction through geological time. Geological, geophysical and paleogeographic data can be associated with simulated plates, allowing their changing relationships to be investigated. The project’s official overview describes this combination of reconstruction, GIS functionality and raster visualisation.
A useful research record therefore goes beyond the final map. The rotation model, feature collections, reference frame and output type explain what the map represents. This guide brings those distinctions together in an evidence checklist and a data table of reconstruction records. They help a research group distinguish a change in its scientific inputs from a change in how those inputs are displayed.
Definition and scope
A plate reconstruction calculates probable past positions, orientations and motions from relative plate positions inferred from other data. That definition, given in the GPlates introduction, is consequential: a reconstruction expresses an interpretation of geological history. The animation is a way to examine that interpretation through time, rather than an independent observation of the past.
The software’s scope includes manipulating reconstructions, attaching data to moving plates and producing paleogeographic maps. Its stated goals also include connecting plate kinematics to geodynamic models. Those roles make it relevant to Earth-science research groups working with spatial observations and geological time. A group can use the same framework to consider the positions of mapped features and their relationships to a model of plate movement.
GPlates runs on Windows, Linux and macOS. Both GPlates and the pyGPlates library are distributed under the GNU General Public License, version 2. The application, its programming interface and the geological datasets remain distinct parts of the work: identifying the application alone does not identify the data or model used in a particular reconstruction.
How GPlates is used in practice
A motion model connects features to geological time
The Reconstructions manual explains motion through plate IDs and finite rotations. Features sharing a plate ID move together. The ID’s meaning is determined by the rotation-file author; it is not a universal identifier for a physical plate. Finite rotations describe displacement using an axis and angle. A total reconstruction pole combines a moving ID, a fixed ID, a time and a rotation.
These relative motions form a hierarchy leading to an anchored plate ID. The anchor defines the fixed reference for the reconstruction, and GPlates allows the user to change it. Geological times are expressed in Ma, meaning millions of years before present. In a methods description, the reconstruction age and anchor therefore belong alongside the model identity, rather than being left implicit in a figure.
Layers process data rather than merely naming files
The Layers manual distinguishes feature collections from processing layers. A file holds a collection of features. A layer accepts inputs and generates an output, which can be displayed or supplied to another layer. Reconstructed geometries, for example, combine feature information with a reconstruction-tree input containing the rotations.
This distinction is especially useful with rasters. A georeferenced raster can remain stationary as reconstruction time changes. Reconstructing it requires the appropriate inputs: static polygons divide the image into pieces that follow their plate rotations. An age-grid input can add per-pixel age comparisons. The presence of an image on the globe, by itself, therefore does not demonstrate that the image has been reconstructed.
The export determines what a collaborator receives
The Export manual separates screenshots, reconstructed geometries, colour rasters and numerical rasters. Reconstructed geometry can be exported in formats including Shapefile and GeoJSON. Numerical raster export retains floating-point values, while a colour raster may represent numerical data after a palette has been applied. An image prepared for a figure and data prepared for further analysis consequently serve different purposes.
For a downstream raster workflow, record which export was produced and how its spatial extent and resolution were configured. CASRAI’s Rasterio guide to geospatial raster data provides related vocabulary for handling the resulting files. Exporting a visually clear map does not, on its own, document the numerical values another analyst would need.
Types and related concepts
GPlates desktop and pyGPlates
GPlates desktop exposes reconstruction functions through an interactive graphical interface. PyGPlates exposes GPlates functionality through Python. The official pyGPlates introduction describes the choice in terms of interaction and flexibility: a script can incorporate reconstruction within a wider research pipeline and access functionality at different levels of detail.
External pyGPlates runs with a Python interpreter and does not require the GPlates desktop executable. This makes the distinction useful when documenting an analysis. A methods statement should identify whether the work depended on a saved desktop project or a Python script, and identify the inputs either one used. The two interfaces are ways to access reconstruction functionality; naming one does not remove the need to describe the geological model.
The Python documentation also distinguishes high-level operations for common tasks from lower-level access that permits more detailed control. Its example contrasts reconstructing a dataset with restricting reconstructed features to a particular region. That difference matters when describing a scripted result: the description should say what was calculated, rather than treating the library name as a complete account of the analysis.
GPlately and analysis of reconstructed data
The official GPlates overview describes GPlately as a Python package for reconstructing points, lines, polygons and rasters through geological time. It also supports interrogating plate kinematics, comparing plate-motion models and plotting reconstructed outputs. These related tools make the suite broader than the desktop interface. A publication that names the package actually used helps readers distinguish interactive visualisation from scripted reconstruction and subsequent analysis.
A reconstruction model and a reconstruction service
A model supplies the reconstruction information; a service makes reconstruction functions available remotely. The GPlates Web Service model documentation states that its models contain at least a rotation model and static polygons. Additional components and geological time coverage differ between models. It also recommends specifying the model name explicitly because the service’s default can change.
For a research team, that is a reason to record the model selection in an analysis record rather than describing an output simply as a GPlates result. The data table below compares the records associated with the model, layers, project and export. It compiles their distinct roles from the official manuals rather than ranking models or claiming that one is universally appropriate.
Where GPlates matters for Earth-science labs
For an Earth-science laboratory or research group, GPlates concerns the computational interpretation of geological data. Its relevance here is the connection between observations, reconstruction inputs and shareable outputs. The following evidence checklist is an editorial synthesis of the official documentation, intended to make those connections inspectable.
- Model identity: retain the model name and the actual rotation and feature files used. The Web Service documentation establishes that models can differ in components and time coverage, and that a default is subject to change.
- Computational state: retain the project or script that describes how the inputs were connected. The desktop manual defines a session as loaded feature collections together with layers and their settings; pyGPlates provides a scripted route to reconstruction.
- Transferable files: include the project dependencies when sharing desktop work. The project manual explains that moving a project to another computer requires its associated data files, with their relative paths maintained, and any colour palettes used by its layers.
- Output meaning: identify whether the deliverable is reconstructed geometry, a numerical raster or a rendered image. The export documentation treats these as different data products.
These records support data provenance: a collaborator can relate an output to the inputs and processing choices behind it. They also make a discussion of computational reproducibility concrete. The practical issue is whether another analyst has enough information to examine the same calculation, including the files on which a saved project depends.
Illustrative example: comparing two reconstructions
Consider an illustrative research comparison, rather than a reported experiment. A group examines the same mapped features at one chosen reconstruction age using two alternative rotation models. Its comparison record identifies the feature collection, age, anchor and rotation inputs for each result. That record distinguishes the intended model change from other differences in the calculation.
The group also identifies the exported product. Two screenshots can support visual discussion, whereas a comparison involving reconstructed coordinates needs the corresponding geometry outputs. If the desktop projects are shared, the referenced data travel with them. This example makes no prediction about how far any feature would move: the manuals establish the input and output distinctions, while a numerical result would require the specified models and an actual calculation.
Common misunderstandings
An animation identifies its geological model
Appearance does not provide that identification. The software can operate on supplied reconstruction data, while the Web Service documents multiple models with different components and time ranges. Preserve the model identity with the result. Agreement between pictures is a different question from agreement between the inputs that produced them.
A project file contains the entire project
A saved project records the names and locations of associated data files. The Projects and Recent Sessions manual explains why transferring the project file alone can leave dependencies behind. It recommends copying the project and its data while preserving their relative directory structure. Missing files can also mean missing layers when the session is restored.
For collaborative work, distinguish the saved session from the collection of files needed to restore it. A project that opens on its creator’s computer has demonstrated access to that computer’s files; that does not establish that a recipient has received those files.
A coloured map preserves the underlying measurements
The export documentation distinguishes palette-coloured raster output from numerical raster output. A colleague who needs numerical values needs a data product suited to that purpose. A screenshot records a view, so its role in a research package differs from the role of reconstructed geometries or floating-point raster data.
Dateline wrapping changes only how a line looks
GPlates documents a stronger consequence: wrapping can split a line or polygon into multiple geometries when it crosses the dateline. Reloading the exported file does not reverse that split. Record the wrapping choice when geometry structure matters to subsequent analysis; a display-oriented export option can affect the structure of the delivered data.
Sources
- GPlates: official software overview. Checked 2026-11-15.
- GPlates User Manual: Introduction to GPlates. Checked 2026-11-15.
- GPlates User Manual: Reconstructions. Checked 2026-11-15.
- GPlates User Manual: Layers. Checked 2026-11-15.
- GPlates User Manual: Projects and Recent Sessions. Checked 2026-11-15.
- GPlates User Manual: Export. Checked 2026-11-15.
- pyGPlates documentation: Introduction. Checked 2026-11-15.
- GPlates Web Service: Plate Tectonics Models. Checked 2026-11-15.
Data: GPlates reconstruction records and their roles
Original cross-reference compiled from the GPlates reconstruction, layer, project, export and Web Service manuals. It distinguishes documented artefact roles and the evidence to retain when comparing or sharing reconstructions. The evidence-record column is editorial synthesis of these documented dependencies.
| Record | Documented role | Evidence to retain | Primary source |
|---|---|---|---|
| Rotation model | Defines relative plate rotations in a hierarchy leading to an anchored plate ID. | Rotation files and anchored plate identity. | https://www.gplates.org/docs/user-manual/Reconstructions/ |
| Layer configuration | Processes feature collections and other layer outputs into reconstructed results. | Connected feature collections and reconstruction-tree inputs. | https://www.gplates.org/docs/user-manual/Layers/ |
| Saved desktop project | Stores a session with references to associated data files. | Project file, associated data and used palettes with relative paths preserved. | https://www.gplates.org/docs/user-manual/Projects_and_Recent_Sessions/ |
| Exported product | Distinguishes reconstructed geometry, screenshot, colour raster and numerical raster outputs. | Export type, output files and relevant export settings. | https://www.gplates.org/docs/user-manual/Export/ |
| Web Service model | Supplies rotations and static polygons; other components and time coverage vary. | Explicit model name rather than reliance on a changeable default. | https://gwsdoc.gplates.org/models/ |
Sources
- GPlates User Manual: Reconstructions (accessed 2026-11-15)
- GPlates User Manual: Layers (accessed 2026-11-15)
- GPlates User Manual: Projects and Recent Sessions (accessed 2026-11-15)
- GPlates User Manual: Export (accessed 2026-11-15)
- GPlates Web Service: Plate Tectonics Models (accessed 2026-11-15)
Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.







