Read from a source
A pixel or published value supplied by a dataset. “Measured” still carries that source's resolution, date, uncertainty and known artifacts.
Data · accuracy · terms
What the land
can tell us.
LANTAW is an exploratory visualization built from real datasets, calculated geometry and clearly marked reconstructions. It is designed to make landscapes understandable—not to replace a survey, an official hazard map or current agency advice.
Scene-specific citationsVisible model disclosuresOpen source codeNo hidden precision
The language we use
Four words prevent false precision.
Calculated from evidence
A mesh, contour, color class or spatial join produced from source data. It is reproducible, but it is not a new field measurement.
A disclosed approximation
A form added where the available source cannot resolve it—such as unsurveyed bathymetry, tiny islets, cloud decks or eruption animation.
Editorial interpretation
Labels, routes and narration chosen to help a visitor understand a place. These are researched and cited, but remain human framing.
Dataset ledger
What enters the renderer.
Coverage differs by scene, so each terrain has its own source panel. This ledger explains the recurring datasets and the exact role they play across LANTAW.
measured
NASA SRTM via Mapzen Terrain Tiles on AWS
The renderer decodes Terrarium RGB pixels into elevation and displaces a geographic mesh. Philippine land coverage is primarily SRTM-derived.
Scale & vintage
SRTM was acquired in 2000. Its global one-arc-second product has postings approximately every 30 metres.
Do not infer
Exact summit height, property-level slope, engineering grade or present-day surface change.
visual context
Esri World Imagery
Satellite and aerial imagery is draped over the elevation mesh. It provides recognizable visual context, not additional terrain height.
Scale & vintage
A global mosaic from Esri, Maxar, Earthstar Geographics and the GIS User Community. Source, capture date and resolution vary by location and zoom.
Do not infer
A single observation date, guaranteed alignment or current land cover. Seams and tone differences may cross tiles.
scene-specific
EOxCloudless Sentinel-2 and USGS Landsat
Pag-asa uses EOX's cloudless 2024 Sentinel-2 mosaic as a broad base. Pinatubo uses a 25 March 1990 Landsat 5 scene for its before-eruption view.
Scale & vintage
Sentinel-2's visible bands are 10-metre class; the EOX product is a processed annual mosaic. Landsat 5 imagery is historical and public domain through USGS.
Do not infer
Live conditions. A cloudless mosaic combines observations, while a Landsat scene records one historical acquisition.
measured model output
DOST–UP DREAM / Phil-LiDAR 1 via LiPAD
The urban scenes request 10-metre flood-hazard layers for 5-, 25- and 100-year rainfall return periods and sample the published low, medium and high categories at building locations.
How to read it
A 5-year event has a 20% annual probability; 25-year, 4%; 100-year, 1%. These are probabilities, not schedules.
Do not infer
Current risk at an individual building. The legacy layers omit later construction and flood-control changes and were still being harmonized with DENR-MGB maps.
derived estimate
OpenStreetMap + Google Open Buildings 2.5D
OpenStreetMap supplies building footprints. LANTAW samples Google's 2023 building-height raster to extrude them; unmatched footprints receive a simple size-based fallback.
Scale & accuracy
The height model has an effective resolution near 4 metres. Google reports 1.5 m mean absolute error only on evaluation in North America, Europe and Japan.
Do not infer
Surveyed height. Google warns of alignment drift, tiling artifacts and potentially larger errors in the Global South and for tall buildings.
reference geometry
geoBoundaries gbOpen PHL ADM1
The homepage archipelago outline is simplified from the Philippines first-level administrative boundary dataset.
Metadata
Boundary year 2020. The source record names NAMRIA, PSA and OCHA Philippines; the source license is CC BY 3.0 IGO.
Do not infer
A legal boundary determination. The outline is a navigation graphic, deliberately simplified for display.
More mesh triangles do not create new measurements.
LANTAW interpolates the source grid for a smooth surface. High-resolution imagery can look sharper than the elevation beneath it, but it does not make the terrain survey-grade.
A basemap is a mosaic, not a live camera.
Imagery providers combine different sensors and acquisition dates. LANTAW treats those pixels as context and keeps event claims tied to dated sources.
A hazard class does not predict one building's outcome.
Risk also depends on exposure, vulnerability, defenses, rainfall and present conditions. Urban scenes are educational overlays, never operational advice.
Use & terms
Explore freely.
Decide carefully.
Intended use
LANTAW is for education, cultural interpretation and geographic exploration. It is not a substitute for professional surveying, engineering analysis, navigation, evacuation planning, insurance assessment or an agency-issued hazard map.
No warranty
The project is provided as-is. Source datasets may contain voids, offsets, outdated observations, classification errors or service interruptions. Verify material decisions with the responsible agency and current local records.
Third-party rights
LANTAW's source code is MIT-licensed. Data, imagery and rendered derivatives retain their providers' licenses, attribution requirements and usage restrictions; the project does not relicense those underlying materials.
Flood material
Urban scenes include copyrighted Phil-LiDAR Program & DREAM material displayed with source credit. Provider terms govern reuse. Those scenes explicitly remain prototypes and are not for operational use.
Corrections
If a label, source, data vintage or modeling disclosure is wrong, open an issue in the public repository with the scene name and a primary source. Traceable corrections are part of the project.