Short answer

Terrain risk screening overlays the physical factors that make a location hard or dangerous to build on — steep or unstable slopes, flood-prone ground, weak geology, landslide susceptibility — so you can rank candidate sites before committing survey and engineering budget. The core method is a weighted overlay of reclassified factor layers, all forced onto a common CRS, resolution, and suitability scale. The result is decision support, not a hazard assessment: it tells you where to look harder and where a specialist must take over, and it is only as trustworthy as its documented weights and source data.

Hazard, susceptibility, and risk are three different things

Sloppy terminology produces misleading maps, so fix the vocabulary first.

  • Hazard is the damaging process itself — a landslide, a flood, ground subsidence — usually with an implied magnitude and frequency (e.g. the 1% annual chance flood).
  • Susceptibility is the spatial likelihood that a process can occur at a location, based on the conditioning factors (slope, lithology, soil), independent of when. Most GIS terrain screening produces susceptibility, not true probabilistic hazard.
  • Risk combines hazard with exposure (what is there — people, assets, the proposed facility) and vulnerability (how badly it would be affected). Risk = hazard × exposure × vulnerability, conceptually.

A screening map that colours steep slopes red is a susceptibility layer. Calling it "risk" implies you have accounted for what is exposed, which you usually have not at the screening stage. Keep the labels honest on the legend.

The factor layers that matter

Start from a quality DEM (SRTM at 30 m for reconnaissance; national lidar at 1–5 m where available — resolution drives everything downstream).

Terrain derivatives:

  • Slope — the single most important factor. Derive with gdaldem slope dem.tif slope.tif or QGIS Slope. Typical thresholds: <5° benign, 5–15° manageable, 15–30° costly, >30° often excluded.
  • Aspect — matters for solar, snow retention, and slope stability (wet north-facing slopes in the northern hemisphere).
  • Curvature / roughness — convex/concave form and surface texture flag instability and earthworks cost.

Hydrology:

  • Flow accumulation and distance to drainage — derive flow direction/accumulation (QGIS r.watershed via GRASS, or Strahler/SAGA tools) to find channels and convergence zones.
  • Flood hazard zones — use authoritative regulatory products where they exist (FEMA flood maps under the US National Flood Insurance Program; national equivalents elsewhere). Do not substitute a quick DEM-derived wetness index for a regulatory flood zone.

Geology and ground:

  • Lithology / soil — weak, expansive, or liquefiable units; karst.
  • Distance to faults — buffer mapped faults; seismic and instability proximity.
  • Landslide susceptibility / inventory — existing inventories (e.g. USGS landslide resources) and conditioning-factor models.

Exposure / access:

  • Proximity to roads, power, water; protected areas and other constraints.

A weighted-overlay workflow

The standard screening method is a weighted linear combination. Worked example for a small facility:

  1. Assemble layers and normalise. Reproject every layer to one projected CRS (the local UTM zone) and resample to a common cell size with gdalwarp -t_srs EPSG:32631 -tr 10 10 -r bilinear in.tif out.tif. Mismatched CRS or resolution silently corrupts an overlay.
  2. Reclassify each factor to a common suitability scale (say 1 = best, 5 = worst). For slope: 0–5°→1, 5–15°→2, 15–25°→3, 25–35°→4, >35°→5. Use QGIS Reclassify by Table or gdal_calc.py. Continuous distance layers can be scaled the same way.
  3. Assign weights reflecting relative importance. For a flood-sensitive site you might use slope 0.30, flood zone 0.30, landslide susceptibility 0.20, distance to fault 0.10, geology 0.10 (weights summing to 1).
  4. Combine. Weighted sum in the Raster Calculator: 
    ("slope_rc@1"*0.30) + ("flood_rc@1"*0.30) + ("landslide_rc@1"*0.20)
+ ("fault_rc@1"*0.10) + ("geology_rc@1"*0.10)
    ArcGIS users have the dedicated Weighted Overlay / Weighted Sum Spatial Analyst tools.
  5. Apply hard exclusions separately. Some constraints are pass/fail, not weighted — inside a regulatory floodway, slope >45°, protected land. Mask these out rather than letting a good score elsewhere average them away.
  6. Rank candidate sites by their summary score and zonal statistics, and carry the layer-by-layer reasoning into the report.

Where data and an expert support it, more rigorous methods exist — AHP for defensible weights, or statistical/ML susceptibility models (frequency ratio, logistic regression) validated against a landslide inventory — but the weighted overlay is the honest screening baseline.

Common pitfalls and why they happen

  • Calling a screening overlay a final risk assessment. The overlay omits site-specific geotechnical, hydraulic, and engineering analysis. It exists to narrow options; treating it as a verdict invites liability and bad decisions.
  • Mixing official hazard zones with model outputs unlabelled. A FEMA floodway and a DEM-derived wetness index are not the same evidence class. Blending them into one ramp hides which boundaries are regulatory and which are indicative.
  • Ignoring source scale and data age. A 1:250,000 regional geology map cannot resolve a site-scale boundary; a decade-old flood map may predate channel changes or new mapping. Scale and date belong in the metadata and the caveats.
  • CRS / resolution mismatch in the overlay. Combining a 30 m DEM derivative with a 250 m layer, or layers in different CRSs, produces misaligned cells and meaningless sums. Normalise first.
  • Equal weights by default. Slope and flood usually dominate site viability; weighting everything the same buries the decisive factors. Justify the weights.

QA and validation

Before the screening output informs a decision: confirm every layer shares one CRS, resolution, and extent, and aligns at the edges; record the reclassification thresholds and weights so the map is reproducible; keep observed events, susceptibility models, and regulatory zones in separate, labelled layers; sanity-check the result against known problem ground in the area; and state explicitly what the map can and cannot decide, with a clear handoff point to specialist review for anything touching safety, permitting, engineering, insurance, or public communication.

Bathyl perspective

We build terrain screening as transparent decision support: documented weights, separated evidence classes, hard exclusions kept distinct from weighted factors, and an explicit boundary where qualified specialists take over. The value is early clarity on which sites are worth deeper, costly investigation — and an audit trail that survives the handoff to the engineers who make the binding calls.

Related reading

Sources