Why terrain resolution matters

How much ground per value

Raster resolution describes the spacing of the terrain grid.

A raster elevation model divides an area into cells or stores values at regular posts. Resolution commonly refers to the linear size of each cell or the spacing between those positions. A five-metre grid and a fifty-metre grid make very different choices about how often height is represented.

The value may describe a sample at a cell centre, an average across the cell or another product-specific calculation. The technical specification must say which. In every case, features smaller than the spacing cannot be explicitly described as independent cells. They may influence a value, disappear between samples or be generalised into the surrounding surface.

Two separate properties

A fine grid can still contain inaccurate heights.

Accuracy asks how close a modelled position or elevation is to an accepted reference. Resolution asks how finely the model is divided. Storing a value every metre does not prove that each value is correct to a metre, or to any other unstated tolerance.

Source measurements, sensor calibration, classification, coordinate transformation, interpolation and terrain type all contribute to error. Vegetated or steep ground may behave differently from open, level ground. Accuracy statements also use defined statistical measures and test conditions, so compare the full specification rather than a headline number.

What the grid can lose

Narrow landforms are vulnerable to sampling and alignment.

A broad hillside can be represented convincingly by a relatively coarse grid because its height changes gradually. A narrow ridge, gully, embankment or cutting may fall between cells and become lower, shallower or wider than it is on the ground. Sharp cliffs can turn into ramps.

Route alignment adds another layer. A line placed just to one side of a narrow crest can sample the slope rather than the top. Increasing the number of points along that route does not help if every new point still reads the same coarse or misaligned terrain model.

• Broad landformCoarser data can retain overall mountain, valley and watershed shape.
• Narrow featureLocal banks, hollows and ridge crests need spacing and source detail suited to their width.
• StructureBridge, tunnel and wall behaviour depends on surface definition and processing, not resolution alone.
• Route profileSmall missed features can change a local gradient or accumulated ascent calculation.

Smaller pixels, same evidence

Interpolation can smooth a model without discovering new terrain.

Software often resamples a grid so it matches another dataset, projection or display. Nearest-neighbour, bilinear and other methods assign values to the new cells using existing values around them. The result may render more smoothly and align with an output grid, but it does not add a measurement of the missing ridge or gully.

This distinction matters when a website displays a dense mesh or high-resolution image. The output pixels and triangles can be much smaller than the native terrain spacing. Visual detail from lighting or texture should not be mistaken for additional elevation evidence.

Resampling changes the arrangement of known values. It cannot recover a feature absent from the source.

Fit for purpose

Match source detail to the decision, then retain the caveats.

1. Feature

   Name the smallest landform that matters.

   A regional terrain illustration and a local drainage study do not need the same grid.

2. Surface

   Confirm ground, canopy or another model.

   Fine resolution on the wrong surface is still the wrong evidence.

3. Quality

   Read accuracy, date and datum with spacing.

   Keep every property attached to its published definition.

4. Result

   State what processing changed.

   Interpolation, smoothing and resampling should remain visible in the interpretation.

The 3D route map guide follows terrain data into a render, and the beacen Guide covers using those views in route preparation.

Choose source detail for the smallest landform that affects the decision. Dense output pixels or triangles cannot replace terrain evidence absent from the source.