How Bilby works
Terrain score and suggested build area
A suitable patch matters more than the steepest corner
Bilby’s terrain overview looks for connected ground that may suit building. The suggested area is a screening result from elevation and slope cells, boundary information and any applied setback mask. It is not a legal building envelope or the excavation calculation for the house you place in the planner. Steep ground elsewhere need not ruin a useful building area.
Coverage and area assumptions
The current overview needs at least 80% valid elevation and slope coverage inside the analysed geometry. Otherwise it returns no score. Slopes are percentages, not degrees: a 10% slope rises 1 m over 10 m horizontally. Slope cells are smoothed with a 3 × 3 mean for scoring.
| Input | Unit | Origin |
|---|---|---|
| Elevation and derived slope | m / % | Terrain dataset cells |
| Plot/analysis mask | m² and cell membership | Boundary or estimated geometry |
| Setback envelope | m and mask | Available rule or disclosed assumption |
| Cell resolution and valid coverage | m / % | Source metadata and calculated coverage |
| Selected candidate geometry | m² | Derived connected region |
Candidate selection and the 100-point result
Connected candidates use eight-neighbour connectivity. A core below 12% slope can expand into cells below 15%; below 8% is preferred. Estimated suitable area sums valid cells below 15% inside the available setback envelope, including disconnected pieces. The selected candidate is one connected region; a representative display box targets about 400 m² and is approximate.
Candidate rank combines area 35%, slope quality 30%, consistency 20% and compactness 10%, divided by 95% because access geometry is not included. Area factors are 1 at ≥800 m², 0.9 at ≥600, 0.75 at ≥400, 0.55 at ≥250 and 0.18 below. Plots under 800 m² can use a better proportional factor: 1 at ≥70%, 0.85 at ≥55%, 0.7 at ≥40%, 0.45 at ≥25%, otherwise 0.18.
The public result adds rounded candidate quality (up to 55), local relief/consistency (20), surrounding terrain context (15) and geometry quality (10 for boundary geometry, 5 for estimated area). No candidate or a candidate below 250 m² caps the result at 34; median candidate slope ≥15% also caps at 34.
Slope, relief and context details
Descending quality factors are 1 / 0.82 / 0.55 / 0.28 / 0.08 across successive limits. Slope quality combines the shares below 8% and 12% (35% and 25%), median slope with limits 5/8/12/15% (25%), and 90th-percentile slope with limits 8/12/15/18% (15%). Consistency combines slope spread limits 2.5/5/8/12 percentage points (60%) and relief limits 1.5/3/5/8 m (40%). Compactness is scaled from the region’s area/perimeter shape, limited to 0.15–1.
The separate 20-point relief contribution weights relief 55% and slope spread 45%. The context starts at 15; deductions are 7 for steep adjacency ≥60%, 4 at ≥30%, 2 above 10%, otherwise 0. A candidate ≥400 m² touching the plot edge gets no contextual deduction. Without an adequate candidate the deduction is 10. Labels start at 85 Excellent, 70 Favourable, 55 Moderate adaptation, 35 Significant adaptation; lower means Complex constraints.
Fictional flat plot
Consider a fully covered square 30 × 30 m plot, 1 m cells, constant elevation, zero slope, a real boundary and no setback exclusion. The 900 m² connected candidate has full area, slope, consistency and compactness factors: rank 100. Quality contributes 55, relief 20, context 15 and geometry 10: 100. If only the geometry source changes to estimated area, the same terrain gives 95. With valid coverage below 80%, neither result is published.
The selected house pad is a different calculation
The house-groundworks path evaluates actual candidate pad positions and cut/fill quantities. Its internal terrain ranking weights pad geometry 40%, earthmoving 25%, retaining 15%, variability 10% and access 10%; missing access uses 65. A fictional set of four 100-point components and access 65 gives 96.5, rounded to 97. That internal rank supports placement and groundworks; it is not a replacement formula for the overview score. Moving the house changes pad work without necessarily changing plot-wide terrain evidence.
What the suggestion cannot establish
Boundary uncertainty, coarse data and unmodelled access affect interpretation. The 5/10 geometry contribution is part of this particular score; it does not replace source coverage or the separate confidence of selected-area earthworks. A rectangular display box can contain cells outside the exact candidate. Verify access, legal setbacks, drainage, rock and retaining design before treating a suggested patch as buildable.
Method and review
Overview method terrain-score-v3-boundary-setback and selected-area callers reviewed 18 October 2026. The source/resolution/date shown in the terrain details identify the actual elevation data. Lantmäteriet describes national elevation information; Bilby defines the ranking and thresholds. Lantmäteriet
The review date covers this explanation and its implemented method. Price and data dates are stated separately in the article.
