Terrain preview · provenance
A landscape with a source.
What this preview represents
This is a modern reference surface, assembled from elevation surveys of different ages and resolutions. It is not a reconstruction of any archaeological period. An optional Malta–Sicily overlay adds a dated sea-level scenario; the underlying terrain remains modern.
Mapzen Terrain Tiles on AWS supply elevation in Terrarium PNG format. Tiles are requested directly from the public elevation-tiles-prod bucket as the visitor navigates. No account or API key is used. Tile availability depends on that external service.
Terrain geometry and shaded imagery use the same elevation tiles. The application resamples each mesh to 65 × 65 vertices, stops requesting new source detail at tile level 12, and applies a fixed northwest hillshade to an elevation-based color palette. Green does not establish vegetation; cream does not establish desert. Shading is illustrative, not a calculation of historical sunlight.
Negative elevations are clamped to zero for the displayed mesh. Sea color approximates water using non-positive elevations. This can incorrectly color inland depressions and coastal cells, and is not a hydrological water mask. It cannot support shoreline reconstruction, flood extent or land-bridge claims. Polar caps outside the Web Mercator coverage (approximately 85° latitude) have no elevation tiles.
Height emphasis defaults to 2× and can be changed to true scale (1×) or 3×. It changes the display only. Archaeological markers use generalized positions from the shared atlas, not surveyed excavation coordinates. At close range they may appear above terrain to remain selectable.
Elevation attribution
Mapzen / Tilezen compilation, served through AWS Open Data. The upstream attribution and license register and source documentation describe the constituent datasets. The global application credits the full compilation, rather than claiming that every tile uses every source:
- ArcticDEM: DEMs created from DigitalGlobe imagery and funded under National Science Foundation awards 1043681, 1559691 and 1542736.
- Australia: © Commonwealth of Australia (Geoscience Australia) 2017.
- Austria: © offene Daten Österreichs – Digitales Geländemodell (DGM) Österreich.
- Canada: contains information licensed under the Open Government Licence – Canada.
- Europe: produced using Copernicus data and information funded by the European Union – EU-DEM layers.
- Global ETOPO1: U.S. National Oceanic and Atmospheric Administration.
- Mexico: INEGI, Continental relief, 2016.
- New Zealand: Copyright 2011 Crown copyright (c) Land Information New Zealand and the New Zealand Government (All rights reserved).
- Norway: © Kartverket.
- United Kingdom: © Environment Agency copyright and/or database right 2015. All rights reserved.
- United States 3DEP and global GMTED2010 / SRTM: courtesy of the U.S. Geological Survey.
Interactive rendering uses CesiumJS, version 1.146.0, under its Apache 2.0 license. Runtime assets are built from the pinned package. This preview does not use Cesium ion or Cesium World Terrain.
Malta–Sicily: the first changing-shoreline layer
Open the Ice Age shoreline scenario ↗
Status: bathymetric sea-level scenario, not a validated palaeogeographic reconstruction. We use a native 15 arc-second subset of GEBCO 2019, with the esl series and named esl minus sigma / esl plus sigma columns in Blinkhorn’s published workbook. The curve derives from Lambeck et al. (2014). It is an equivalent/global sea-level proxy, not a locally corrected Maltese relative sea-level curve.
The overlay covers 13–15.75°E and 35.5–37.5°N. Its 660 × 480 cells are about 370 × 460 m here. GEBCO combines measured and inferred bathymetry and assumes mean sea level, with heterogeneous input datums. Our derived PNG quantizes elevations to 0.125 m for storage; this does not imply that measurement accuracy. GEBCO Compilation Group (2019), doi:10.5285/836f016a-33be-6ddc-e053-6c86abc0788e. Data accessed through CEDA/BODC; public-domain reuse with acknowledgment, without endorsement.
What changes when you move through time
The shoreline uses the center of the evidence window, rounded to 100 calendar years BP (before 1950). The valid interval is 21,000–200 BP; no values are extrapolated outside it. Linear interpolation provides a viewing step, not 100-year dating precision. Evidence markers still represent their complete published ranges.
| Model date (BP) | Sea-level proxy | Named ±sigma inputs |
|---|---|---|
| 21,000 | -134.0 m | -135.6 to -132.4 m |
| 14,200 | -93.8 m | -94.7 to -92.8 m |
| 13,000 | -71.2 m | -72.1 to -70.4 m |
| 8,500 | -17.7 m | -18.0 to -17.4 m |
Ochre marks cells that are connected to the sea at the zero-metre reference but exposed at the selected level. Pale lines show the present reference coast; turquoise shows the scenario coast. A four-neighbour flood-fill starts at the regional boundary, so isolated inland depressions are not simply painted as ocean. Narrow channels and connections remain sensitive to grid resolution and crop boundaries. The overlay is a footprint on modern relief, not a newly excavated ancient terrain surface.
What the amber band does—and does not—mean
The band shows where the two named ±sigma sea-level inputs produce different land/water classifications. It does not include bathymetric error, local land movement, glacial isostatic adjustment, erosion or sediment deposition. It is not a total shoreline confidence envelope. The author code uses these ±sigma columns in its landscape analysis, while the article’s sea-level plot uses ±2sigma. We preserve those different conventions.
Independent evidence stays separate
Rossi et al. (2025) infer a western LGM palaeocoastline from seismic evidence at present seabed depths of 119–131 m; the eastern extent remains uncertain. Those feature depths are not an error band around this global curve. The paper’s vector share did not provide downloadable resources in this access attempt. Related EMODnet features still need an exact source match, so no purported seismic line is drawn here.
The approximately 14,200 BP disconnection in Blinkhorn et al. (2026) and approximately 13,000 BP account in Scerri et al. (2025) remain distinct source-attributed estimates. They are not the endpoints of one confidence interval. Our native-grid visualization is not an exact replication of the population model’s projected/resampled grid. A land bridge is environmental context; it does not document a human crossing.
What would improve this reconstruction
Next: verify and import the published seismic coastline as a separate interpreted layer, compare a regional relative sea-level model with the global proxy, and assess vertical land motion and post-Ice-Age seabed changes. Ice sheets, vegetation and river courses require their own dated datasets.
Each new landscape layer must declare its source/version, extent, valid age interval and time convention, datum, resolution, method, uncertainty, license and reconstruction status. Archaeological date uncertainty remains separate from landscape-model uncertainty.