Where Our Elevation Data Comes From and How Accurate It Is

Last updated 2026-10-21 · MapElevation

Which datasets are currently active

The public point lookup currently uses Copernicus GLO-30 through terrain tiles provided by Mapterhorn. The original global dataset is a digital surface model with approximately 30-metre resolution. The application samples the global source at tile zoom 12 and labels the result Grade C. The Copernicus DEM collection description provides the authoritative source context, while Mapterhorn attribution describes the terrain delivery project and underlying data credits.

The basemap is separate from elevation data. OpenFreeMap supplies a map style and vector-map infrastructure, with OpenMapTiles and OpenStreetMap attribution. A new road label or a detailed building outline on that basemap does not add original samples to the terrain grid. The elevation map combines these layers for orientation, while its result card identifies the source behind the height estimate.

How to use this page

1. Identify the dataset and model reported by the tool.
2. Check original spacing, source reference, and the selected coordinate.
3. Separate a decoding check from an independent accuracy comparison.
4. Keep unresolved reference, coverage, or licensing questions explicit.

How our A–D grades are defined

The grades are this site’s source categories, not an official certification or a universal vertical-error specification. They help distinguish broad classes without implying that every point has a known survey error. Coverage and source metadata still matter. The following definitions describe the intended classification system, including future categories that are not yet enabled.

| Grade | Source category | Current point-lookup status | | ----- | --------------------------------------------------- | -------------------------------------------- | | A | National DTM with source resolution at or below 5 m | Planned regional upgrade | | B | National DTM around 10 m | Planned regional upgrade | | C | Copernicus GLO-30 global DSM | Active global policy | | D | GLO-90 global category | Defined for future fallback; not active here |

US 3DEP and Spain IGN are planned as later regional work. Japanese national terrain requires licensing confirmation before activation. Neither a country name nor the map’s current zoom is enough to claim that a higher grade is in use. A future upgrade must have verified coverage and a complete description of its model, reference, resolution, and licence, with a truthful fallback label when another source actually supplies the result.

Resolution is different from vertical accuracy

Original spacing describes how the source samples represent the surface across an area. Tile pixels, screen pixels, displayed decimal places, and the size of a selected marker describe other things. Resampling can create more display pixels without adding independently observed terrain. A closer view is useful for selecting a location, but it does not turn the underlying source into a finer survey.

Vertical accuracy concerns how reported heights differ from an appropriate reference measurement. It can vary with terrain, surface objects, processing, and the comparison method. This site does not publish a universal plus-or-minus guarantee for every point. The elevation finder preserves coordinates for user interpretation, while the source label reminds you that a precisely written position can still be sampled from a coarse or imperfect model.

DSM versus DTM

A DSM represents the visible or modeled surface, which can include vegetation and structures. A DTM aims to describe ground after appropriate processing. The distinction is especially relevant around buildings, trees, and narrow features. A surface-model value can be higher than local bare ground without the decoding algorithm being wrong. Conversely, a lower reading is not proof that every obstruction has been removed or every ground detail captured.

The USGS explanation of DEM and DSM terminology provides background on these model types. An address elevation finder adds another uncertainty: the address coordinate can lie on a road range or beside the intended building. Model type and location matching should therefore be reviewed together rather than treating a house address as proof of a building-specific measurement.

Vertical datums and unit conversions

GLO-30 uses the EGM2008 geoid reference. A GPS altitude, a local tidal datum, NAVD88, and NGVD29 are not interchangeable labels. A valid comparison needs compatible height systems, the intended physical measurement, and an appropriate transformation where one is available. NOAA’s geoid information and NOAA VDatum explain the reference concepts and supported transformation context.

The height above sea level guide translates these distinctions into everyday map use. The BFE lookup preserves the flood source’s original datum and units instead of subtracting its value from a global DSM reading. Changing metres to feet is a unit conversion; it does not resolve a reference mismatch or change a surface pixel into a surveyed building measurement.

How terrain tiles are sampled

For a point query, the browser identifies the tile and pixel corresponding to the selected coordinate at the configured source zoom. It downloads the tile if needed, decodes its encoded height, checks for unusable data, and returns the value with source metadata. A bounded in-memory cache can reuse recently decoded terrain tiles. It does not store a permanent history of exact user addresses.

The current terrain encoding is Terrarium, read from the configured Mapterhorn source. Decode and display choices are separate from source accuracy: successfully reading a value only establishes what is stored at that sample. The contour map generator derives lines from the modeled grid, and the sea level rise map uses a coarser bounded calculation grid for its illustrative scenario. Neither is an independent survey of the point result.

Known limitations and verified comparisons

Urban surfaces, vegetation, steep terrain, narrow barriers, location error, and source resampling can all affect interpretation. A visual agreement between hillshade and a height card is not independent validation, because they may use the same underlying source. A published city altitude is not a valid check of a specific point unless its coordinate and physical meaning are known.

The PRD contains a Miami Beach comparison that motivated the source distinctions. Its exact coordinate, original observations, and datum treatment have not yet been reproduced as a complete example record for this release. The individual figures are therefore not republished here as verified measurements. The example TODO list requires a reproducible location and source record before numerical claims are added. No blanket vertical accuracy claim is inferred from a small number of comparisons.

Geocoding and FEMA data limitations

US address matching uses the Census Geocoder, which can derive positions from address ranges. Optional worldwide matching uses Geoapify when configured. The current elevation tool instead requests a browser position after user action; that position has its own horizontal uncertainty. The terrain lookup is independent of which input method supplied the coordinate.

The flood zone lookup queries FEMA NFHL and preserves missing or unavailable states. The flood zone codes guide explains category interpretation, including Zone X and Zone AE. A certificate and LOMA review can require certified building or ground information beyond this site’s surface data. No point query is represented as an official property determination.

Data licences and attribution

Derived terrain use includes the Copernicus attribution reproduced below, together with Mapterhorn and the basemap credits. The Copernicus licence is the governing source for its attribution requirements. Additional regional datasets must bring their own verified rights and credits when activated. Planned data is not silently added to the active-source table.

Updates and corrections

The date on this page records an actual editorial update. Rebuilding or deploying unchanged text does not refresh it to imply new data. Terrain-source changes, coverage corrections, and verified example additions require corresponding metadata and explanatory updates. The full first-version limitations remain visible until the relevant capability has been implemented and checked.