Deep Time Explorers

A Year 4 (~9–10 years old) spin-off of the main GPlately + pyGMT tutorial suite, built around the same real EarthByte plate-reconstruction data — but with none of the install complexity. No pyGMT, no GPlately, no pygplates on the student's own machine: just numpy, pandas, matplotlib and ipywidgets.

Status: prototype, not yet classroom-tested.

Why this exists

A Year 4 teacher and EarthByte alumnus has been running his own ocean-current particle-tracking notebook with his class. That sparked the idea for a plate-tectonics sibling: same spirit (real code, real data, highly visual, age-appropriate), built around this repository's deep-time plate reconstructions instead.

An early version of this ran as live Jupyter notebooks in the browser (JupyterLite, below) — which turned out to be too cognitively demanding for a primary-school audience, and a real ipywidgets/JupyterLite bug meant some sliders silently stopped updating. The missions below are the fix: the same real data and the same mission structure, with every map baked in ahead of time instead of computed live, so there's no code, no cells, and no kernel that can get stuck.

Try the missions

Six short, self-contained web pages — pick one and start dragging the slider, nothing to install or run:

→ Open the missions (live once GitHub Pages is switched on for this repo — see 'Running it' below).

#	Mission	Big idea
1	Continents on the Move	A time slider redraws the continents — real paleo-elevation, real plate boundaries, and real plate-velocity arrows — from 300 Ma to today
2	The Jigsaw Continents	South America and Africa's coastlines fit together — Wegener's own evidence, plus a real mountain belt stitching them
3	Where the Ground Shakes	246 real earthquakes, switched on and off over today's real plate boundaries
4	Follow the Duck through Deep Time	A simplified 'duck' journey shows how an ocean gateway opening changes the shortest sea route
5	Postcard from Deep Time	Pick an Australian town and an age, get a postcard of where it used to sit
6	Fossil Climate Detectives	Real fossil and rock climate clues — coal swamps, desert dunes, crocodile fossils — plotted through six deep-time ages

These are built straight from the same reconstructions as the six research notebooks below — same coordinates, same datasets, no separate simplified copy of the science. The notebooks stay as the citable, inspectable 'how it's built' layer; the missions are what a classroom actually opens.

The six notebooks

#	Notebook	Big idea
1	01_Continents_on_the_Move.ipynb	A time slider redraws the continents — real paleo-elevation, real plate boundaries, and a second map of real plate-velocity arrows — from 300 Ma to today
2	02_The_Jigsaw_Continents.ipynb	South America and Africa's coastlines fit together — Wegener's own evidence, plus a real mountain belt stitching them
3	03_Where_the_Ground_Shakes.ipynb	Real earthquake locations plotted against plate boundaries, with ridges and subduction zones drawn on top of the dots for a clear view of both
4	04_Follow_the_Duck_through_Deep_Time.ipynb	A simplified "duck" journey shows how an ocean gateway opening changes flow — a deep-time cousin of an ocean-current particle-tracking activity
5	05_Postcard_from_Deep_Time.ipynb	Pick an Australian town and an age, get a postcard showing where it used to be
6	06_Fossil_Climate_Detectives.ipynb	Real fossil and rock climate clues — coal swamps, desert dunes, crocodile fossils — plotted as picture icons through deep time

Each notebook is self-contained, short, heavily commented, and ends with a "Try this next!" section for kids who want to push further.

What the maps look like

Every map in this suite is a real, properly-projected map, not a flat cartoon — and every background is built by reusing the main tutorial suite's own gplately.PlotTopologies plotting layer (the same pattern demonstrated in T01_Hello_Deep_Time.ipynb's Cartopy cell), not a from-scratch reimplementation:

• Notebooks 1, 3, 5 and 6 use a full-globe Robinson projection with a latitude/longitude graticule — the same kind of "properly curved" world map used in atlases.
• Notebooks 2 and 4 use a simple, zoomed-in latitude/longitude crop of just the region that matters (the Atlantic; the Southern Ocean).
• Every background shows real bathymetry and topography — how deep the ocean floor and how high the land actually stood at that point in time — from the Scotese & Wright (2018) PaleoDEM dataset, coloured with a hypsometric/bathymetric scheme in the same spirit as this suite's own house convention (T43_Geochem_Corrected_Paleo_ Elevation.ipynb uses GMT's "earth" cpt for the same purpose). Blues are ocean depth, green/tan/brown/white are land elevation — mid-ocean ridges and deep trenches are visible on the sea floor itself, not just implied.
• The continent outlines drawn on top of that DEM use the Scotese & Wright (2018) plate model itself (scotese_and_wright2018, via plate_model_manager) — not this suite's usual Zahirovic2022 — because that's the model the DEM was built from. A different model's coastlines don't land in the same place at a given age, especially in deep time; the continent outline has to come from the DEM's own plate model for the two to actually line up. Every background also has real plate boundaries drawn on it, reconstructed to that exact age, typed by kind using gplately's own plot_ridges / plot_transforms / plot_trenches methods: red for spreading ridges, orange for transforms, a thick dark-blue line for subduction zones — so Notebook 1 (which is about plates) actually shows the plates' edges, and the ridge/trench lines usually line up visibly with real bathymetric highs and lows underneath them. (Subduction zones use plot_trenches, not gplately's directional plot_subduction_teeth — this model's topology doesn't carry the resolved polarity that teeth need, so that method silently draws nothing for scotese_and_wright2018.) This only works for ages of 100 million years or more recent — the Scotese & Wright model's plate-boundary topology doesn't resolve any further back than that (its continent outlines do, across the whole 0-410 Ma range this suite uses) — so basemaps older than 100 Ma show a correctly-aligned coastline with no boundary lines on top.
• Notebook 1 also has a plate-velocity map: real direction-and-speed arrows for every age on its slider, straight from gplately's own plot_plate_motion_vectors method (the same one demonstrated in gplately's own official 04-VelocityBasics.ipynb example) — so kids can see which plates are racing along and which are barely moving, not just infer it from the boundary story. A reference arrow in the corner gives a real speed to check against, and every age uses the same arrow scale, so lengths are directly comparable across the whole time slider.
• Notebook 6's fossil "climate clues" are drawn as hand-made picture icons (a coal-forest conifer, a palm tree, a crocodile, a desert dune — plus a snowflake, defined but switched off by default, see below) built entirely from matplotlib shapes — no image files, no fonts, no internet connection needed, and they render identically on every computer. The map has its own legend row underneath it (not overlapping the globe) that reuses these exact same icon-drawing functions (not a separate matplotlib auto-legend) so what's in the legend always matches what's on the map.

A design note on Notebook 6's data

Notebook 6 uses the climate-sensitive-lithology dataset from Boucot, Xu & Scotese (2013) — the same dataset the main suite's T62_Boucot_Climate_ Sensitive_Lithologies.ipynb uses, simplified from its 15 rock/fossil categories down to a handful of kid-friendly climate types. A few things worth knowing:

• There's no literal "coral" category. Boucot's categories don't include coral reefs, so Notebook 6 doesn't have a coral icon — the crocodile icon (his "warm temperate" group, which also covers palms and mangroves) is the closest real match to "warm and wet." Real coral/reef occurrence data does exist in the main suite (see T57_Reef_Builders_Paleolatitude.ipynb and T60_PBDB_Paleobiogeography.ipynb, both PBDB-sourced) — folding that in as a real 6th icon is good future work, not done here.
• The "cold" (ice-age) category is in the data but not switched on by default. Across the six rock-clue ages, cold-climate points range from 0 to 21 out of totals in the hundreds — genuinely rare, and the basemap's elevation-based colouring doesn't depict ice sheets either, so a lone snowflake had little visual context to land on. draw_snowflake is still defined in the notebook; "Try this next!" invites kids to add "cold": draw_snowflake back into CLIMATE_ICON themselves. (The "warm temperate"/crocodile bin is similarly rare — 0 points at three of the six ages, 2–5 at the rest — but it stays switched on since a reflection question is already built around it; see the redrawn draw_crocodile note just below.)
• The crocodile icon has a jagged back ridge and a visible jaw line — the features that read most clearly as "crocodile" rather than a generic shape at small size — with the eye placed at the head end.
• Notebook 6 uses the same Scotese & Wright (2018) model/frame as its own basemap (mantle frame, anchor 0 — the same as Notebooks 1–5's basemaps), so the climate-clue points always land on the continent shown underneath them, at every age.
• The time slider goes all the way to today (0 Ma), on purpose without fake data. There's no real Boucot rock-clue data for the present day — that dataset exists specifically to infer a climate nobody could observe directly — so rather than invent a fake "today" data point, the 0 Ma frame shows an icon-free map with a short note explaining why: we don't need rock clues for a climate we can just look outside and see.

Curriculum links (NSW / Australian Curriculum v9)

• NSW Science and Technology K–6, Stage 2, Earth and Space — natural processes that change Earth's surface (currently framed around erosion; plate tectonics is the more dramatic companion story).
• AC9S4U02 (Year 4 Science) — the water cycle, including movement of water through the ocean — the direct peg for Notebook 4.
• AC9S4I04 (Year 4 Science Inquiry Skills) — constructing and using representations (tables, graphs, visual models) to show patterns — what every notebook here actually does.
• AC9TDI4K03 (Digital Technologies, Years 3–4) — recognising that the same data can be represented differently. Note: real Python/Jupyter is above the formal Years 3–4 Digital Technologies expectation (which targets visual programming) — pitch this as science/numeracy enrichment, not core DT coverage.

Running it

One-time setup (a grown-up does this once)

The student notebooks need a handful of small data files that don't exist yet in a fresh checkout — they're generated from the real plate model using export_data.py. From a terminal, in your existing gplately conda environment:

conda activate gplately
python export_data.py

This is designed to sit as a sibling folder to the main GPlately-pyGMT_tutorials checkout on disk — if it finds ../GPlately-pyGMT_tutorials/gplately_data, it reuses the Zahirovic2022 model already cached there (no re-download). If that folder isn't there (e.g. this repo has been cloned somewhere else entirely), it falls back to its own local gplately_data/ cache and downloads the model once (~50 MB). It also downloads a second, smaller model — scotese_and_wright2018 (~12 MB) — into this repo's own local gplately_data/ cache regardless of whether the sibling folder is found; that one's used only for drawing continents/plate boundaries on the basemaps, to match the plate model the PaleoDEM itself was built from (see "What the maps look like" above). Either way, output goes to data/. It also fetches a small real earthquake catalogue from the public USGS service (falls back to a short built-in list if there's no internet at the time). Re-run it any time you add a town to the list inside the script.

New requirement for the map backgrounds and Notebook 6: the script also needs the real Scotese & Wright (2018) PaleoDEM grids and the Boucot 2013 lithology tables, both of which already live in the main suite's large-data archive at ../GPlately-pyGMT_tutorials/zenodo_data/paleoDEM_ScoteseWright2018/ and ../GPlately-pyGMT_tutorials/data/Boucot2013_Lithology_Data_Tables/. If you're running this from a fresh checkout of the main suite, make sure you've unzipped its Zenodo companion archive first (see that repo's README) — export_data.py will raise a clear error naming the missing folder if it can't find these.

cartopy, xarray and netCDF4 (used only by this script, never by the student notebooks) are already listed in the main suite's environment.yml.

For students

Once data/ is populated, each notebook runs standalone with nothing beyond a normal Jupyter install (pip install ipywidgets if it isn't already there). No conda environment, no GPlately, no cartopy, no internet connection needed.

Zero-install notebook lab (the research layer, not the classroom one)

For anyone who wants to run the actual research notebooks in a browser — not the missions above, the real ipywidgets-driven notebooks — this suite is also packaged as a static JupyterLite site: the same six notebooks running entirely in the browser on Pyodide (a full Python interpreter compiled to WebAssembly). numpy, pandas, matplotlib and ipywidgets all run exactly as they do in a normal notebook — no server, no accounts, no local install, nothing to break.

This is not the recommended path for a classroom — see 'Try the missions' above for that, and 'Why this exists' for why. It's here for anyone who wants to see, tweak, or re-run the real analysis code itself.

Open the notebook lab (live once GitHub Pages is switched on for this repo — see below).

.github/workflows/deploy-jupyterlite.yml rebuilds and redeploys this site automatically on every push to main, straight from the same six root notebooks and the data/ folder above — there's no separate hand-maintained copy of the notebooks to keep in sync. The only one-time step is turning Pages on: Settings → Pages → Source: GitHub Actions. After that the link above goes live and stays current automatically.

Data files (generated, not hand-edited)

File	Used by	Contents
data/basemap_<age>Ma.png (×13: 0, 15, 50, 80, 100, 150, 155, 200, 240, 250, 300, 330, 410 Ma)	1, 3, 5, 6	Real bathymetry+topography, real plate boundaries (ridges/transforms/subduction zones), full-globe Robinson projection + graticule
data/basemap_subduction_0Ma.png	3	Same Robinson basemap/extent as basemap_0Ma.png, but with ONLY the subduction-zone line drawn (no ridges/transforms) — the backdrop for Notebook 3's earthquake map, so the vector ridge/subduction overlays plotted on top aren't competing with a busier background
data/subduction_zones_0Ma.csv	3	Real subduction-zone (trench) line geometry as plain lon/lat + Robinson x/y, one row per vertex, grouped by segment_id (~38 disconnected lines) — plotted in the notebook itself, in dark blue, on top of the earthquake dots, so the line can't be buried by markers drawn after it
data/ridges_0Ma.csv	3	Real mid-ocean-ridge (spreading boundary) line geometry, same shape as subduction_zones_0Ma.csv (~26 segments) — plotted alongside it, in red, on top of the earthquake dots
data/velocity_<age>Ma.png (×7: 0, 50, 100, 150, 200, 250, 300 Ma)	1	Real plate-velocity arrows (direction + speed) at each age, plain continents/coastlines/boundaries background, full-globe Robinson projection
data/basemap_atlantic_<age>Ma.png (0, 300 Ma)	2	Same, zoomed to the Atlantic region
data/basemap_southern_ocean_<age>Ma.png (0, 60 Ma)	4	Same, zoomed to the Southern Ocean
data/land_<age>Ma.geojson (×7, ages 0–300 Ma)	—	Simplified land polygons; kept for reference, no longer used by any notebook now that real elevation backgrounds exist
data/towns_through_time.csv	5	Reconstructed positions (lat/lon + Robinson x/y) of ~11 Australian places at each age, using the same scotese_and_wright2018 model + reconstruction as the basemap it's plotted on
data/notable_earthquakes.csv	3	M ≥ 7.5 earthquakes since 1970 (USGS) with Robinson x/y, or a small static fallback
data/fossils_through_time.csv	6	Boucot 2013 climate-sensitive lithology points, reconstructed with the same scotese_and_wright2018 model as the basemap, at 6 ages, with climate bin + Robinson x/y
data/duck_path.csv	4	The duck's ocean-only shortest-path route (Dijkstra over real PaleoDEM elevation, land excluded by construction) at 2 ages, with lat/lon + Robinson x/y per step
data/postcard_<town>_<age>Ma.png	5	Whichever postcard(s) you generate by running Notebook 5 — saved alongside the rest of this suite's generated pictures rather than at the top level of the repo

License

The notebooks and scripts in this repository are released under the MIT License. The underlying scientific datasets (below) are each under their original source's own terms, not this repository's license.

Attribution

Developed by the EarthByte Group, University of Sydney, in the same spirit as the main GPlately-pyGMT tutorial suite — see that repository's own CONTRIBUTORS.md.

Paleo-elevation data: Scotese, C.R. & Wright, N. (2018). PALEOMAP PaleoDEM Elevation Models. Zenodo. https://doi.org/10.5281/zenodo.5460860

Climate-sensitive lithology data: Boucot, A.J., Xu, C. & Scotese, C.R. (2013). Phanerozoic Paleoclimate: An Atlas of Lithologic Indicators of Climate. SEPM Concepts in Sedimentology and Paleontology No. 11. https://doi.org/10.2110/sepmcsp.11