Contours tell you how high the ground is. A slope map tells you how steep it is. Both come from the same elevation grid, but they are read differently. Everyone knows that tightly spaced contours mean steep ground; if that is the only cue, every reader arrives at a different answer. The SLOPE panel replaces the eyeballing with numeric bands rendered as flat colour.
This is an explanation of what the panel computes, why the bands sit where they do, and what the panel cannot tell you. The last part matters most — do not read a grade off this diagram and put it in a code-compliance argument.
The panel takes the same elevation grid the TERRAIN panel uses and differentiates it: the elevation difference east–west and north–south at every grid point gives a gradient, and the gradient's magnitude is the slope. Elevation comes from AWS Terrain Tiles, which over the United States are built from USGS 3DEP. We use elevation relative to the site centre only, not absolute; the reasons are on the Data sources page.
For US sites the panel reports percent grade, with breaks at 5, 10, 15 and 25 percent. Five bands result.
| Band | How to read it |
|---|---|
| under 5% | Effectively flat. This is also the ADA threshold below which a walking surface is not treated as a ramp (1:20). |
| 5–10% | Gentle. A ramp with handrails (ADA maximum 1:12, 8.33%) lives in this band. Grade can be absorbed by steps and a split level. |
| 10–15% | Sloping. Building placement and the driveway begin to be dictated by the ground. |
| 15–25% | Steep. Retaining walls, cut and fill become premises of the scheme rather than details. |
| 25% and up | Very steep. Many jurisdictions restrict or forbid development on slopes in this range (hillside ordinances). |
The Korean edition of the same panel uses degrees (5·10·15·20°), because Korean development-permit rules are written in degrees. Same map, different unit — the same reason contours are labelled in metres there and feet here.
The first version came out looking like leopard skin. Unmistakably flat ground was covered in blotches averaging five degrees. The cause is noise in the elevation data: slope is the derivative of elevation, so small noise in the heights is amplified in the gradient. The light smoothing we use for contours (1.5 grid cells) was fine for contours and hopeless for a derivative.
Rather than guess a smoothing width we measured it, on three flat sites and three hilly ones in Seoul, watching the mean slope as the width changed.
| Smoothing | Flat 1 | Flat 2 | Flat 3 | Hill | Mountain | Hillside suburb |
|---|---|---|---|---|---|---|
| 16 m | 6.2° | 5.6° | 5.2° | 23.0° | 37.2° | 17.5° |
| 52 m (adopted) | 2.5° | 2.5° | 2.6° | 20.2° | 33.5° | 12.6° |
| 104 m | 1.4° | 1.2° | 1.6° | 13.4° | 23.6° | 8.5° |
Around 50 m is the hinge. The floor noise on flat ground halves; the real slopes on hilly ground survive. Widen to 100 m and the flat sites get cleaner still, but the mountain drops from 37° to 24° — that is erasing slope that exists. So the smoothing width is fixed at 50 m (about 165 ft).
One technical decision worth recording: the width is fixed in metres, not grid cells. Grid spacing changes with the analysis radius (6.2 m at 300 m radius, 16.7 m at 800 m). Fixed in cells, the same smoothing would mean different things at different radii, and the same site would change when you only changed the radius.
In Korea the same elevation source carries artificial errors in dense city centres: compared with an independent DEM at random points, downtown Seoul ran a median 10 m high (up to 48 m), and a non-existent 120 m hill appeared in the middle of a flat district. Tall buildings appear to have leaked into the terrain. Hilly ground was fine.
Over the United States the tiles are derived from 3DEP, which is largely lidar-based, so we expect fewer of these artefacts. We have not measured this in US cities. Treat a sudden steep blotch in a flat downtown as a data artefact first and terrain second, the same way you would in Seoul.
Hillside and grading ordinances define slope in specific ways — average slope over a parcel computed by a prescribed formula, or slope measured between contours at a set interval on a survey. This panel is a point-wise slope smoothed to 50 m from a public DEM. Where a parcel has a steep part and a flat part, the panel shows both; an ordinance's average shows one number. The two are not interchangeable.
Contours are drawn thinly over the five colour bands. Without them a slope map does not tell you which way is up — a 15% slope facing south and one facing north mean opposite things for a building. If you pick an accent colour, the ramp keeps its lightness and saturation and only the hue follows, so the bands stay distinct.
There is no aspect (facing direction) panel yet. "South-facing slopes only, in the accent colour" would be one more plate; whether to add it is undecided.
On a sloping site this panel is the first button of the layout. Which edge the access road touches and how steep that edge is decides where a parking ramp can go and how long it must be; where the steep ground lies decides where a building can stand at all. Make those judgements from colour bands here, then confirm direction with the contours — rather than the other way round.
Overlay it with sun hours. South-facing slopes cast short shadows and north-facing ones long, and the two plates explain each other.