Sources
Methodology, data and scientific limitations
The atlas combines three different scientific inputs: modern relief, a sea-level stack and an ice-sheet reconstruction. It does not solve a coupled Earth-system model.
Terrain and exposed land
NOAA ETOPO1 ice-surface elevations are sampled every 10 arc-minutes and stored as signed 16-bit metres. Terrain above the selected global sea-level threshold is drawn as land; modern submerged terrain above that level is highlighted as exposed shelf.
The calculation omits land deformation, erosion, sediment changes, regional sea levels and ocean-connectivity tests. Present ice-surface elevation is not ancient subglacial terrain. Narrow channels and local flooding dates are not reliably resolved.
References: NOAA — ETOPO1 Global Relief Model
Sea-level processing
The 0–30 ka portion of the Spratt–Lisiecki stack is normalized by subtracting +8.49 m, its 0 ka estimate. Intermediate values are linearly interpolated between 1,000-year samples. The original uncertainty values are retained in the provenance file; the display is not a pointwise uncertainty analysis.
BP here uses 1950 as the reference year. The formal Holocene boundary is about 11,700 years before 2000, or 11,650 BP. The map’s era label uses that converted boundary; its 100-year slider still cannot resolve the transition precisely.
References: Spratt & Lisiecki (2016) — A Late Pleistocene sea level stackNOAA Paleoclimatology — Spratt and Lisiecki sea-level datasetInternational Commission on Stratigraphy — Holocene GSSP
Ice margins and thickness
ICESHEET margin polygons use the pinned repository revision recorded in the ice provenance file and are simplified by 0.035°. For North America, Eurasia, Antarctica and Patagonia, the outlines are rasterized to a 0.25° display grid. Signed distances to their edges are blended linearly between the surrounding 2,500-year snapshots to animate growth and retreat. This geometric interpolation is not a physical ice-flow model, does not add scientific evidence and can differ from the thickness grid. If animation is unavailable, the nearest source outline is shown.
Thickness comes from the corrected April 2021 PaleoMIST global 1° minimal-scenario NetCDF file. We select the nearest node and nearest time, clamp tiny negative gridding artifacts to zero, store whole metres and display values rounded to 10 m. Missing data are distinct from zero. This is grounded ice thickness, not floating sea ice.
References: Gowan et al. (2021) — A new global ice sheet reconstruction for the past 80,000 yearsICESHEET 2.0 — published margin files and codeGowan (2019) — PaleoMIST reconstruction dataset, corrected April 2021 (CC BY 4.0)
Model disagreement and interpretation
The Spratt–Lisiecki sea-level series is independent of PaleoMIST. Combining them does not produce a self-consistent reconstruction of ice loading, land deformation and ocean volume. The published debate about PaleoMIST’s sea-level implications and the authors’ reply are both relevant background.
References: Yokoyama et al. (2022) — Towards solving the missing ice problem and the importance of rigorous model data comparisonsGowan et al. (2022) — Reply to the missing ice commentary
Licenses and reproducibility
The PANGAEA thickness dataset is licensed CC BY 4.0 and attributed to Evan J. Gowan and its associated publication. ICESHEET code and margin materials include the repository’s GPL license in the data directory. The provenance files record the source URLs, transformations and source-file checksum for thickness.
Use the downloadable data registry and provenance links below to identify exactly which inputs produced the map. The displayed coastlines and thickness values are educational estimates, not navigation, engineering or future flood-risk data.
References: Gowan (2019) — PaleoMIST reconstruction dataset, corrected April 2021 (CC BY 4.0)ICESHEET 2.0 — published margin files and code