LiDAR Terrain & Geohazard Analysis

The problem

A regional geohazard and infrastructure risk assessment needed reliable bare-earth terrain products across 580 km² of steep, vegetated, mountainous ground — where photogrammetry fails and existing elevation data was far too coarse to map slope instability.

This project processed a high-density airborne LiDAR survey (average 8 pts/m²) into the assessment's primary inputs: a 0.5 m bare-earth DEM, 11 terrain derivative rasters, and a classified geohazard susceptibility index for slope instability and debris-flow initiation zones.

Raw LAZ point clouds were processed through a LAStools pipeline to generate classified ground, vegetation and structure returns. Bare-earth DEMs were hydro-conditioned and used to derive slope, aspect, curvature, roughness and topographic wetness index (TWI) surfaces. All products were validated against independent GNSS survey benchmarks and delivered via ArcGIS Online.

Methodology

1. Point cloud tiling & classification — Raw LAZ tiles classified using LAStools (lasground, lasclassify) into ground, low/medium/high vegetation, buildings and noise; visual QC in CloudCompare.
2. Bare-earth DEM generation — Ground-return TIN interpolated to 0.5 m raster (las2dem); pit-filled and edge-artifact corrected across tile boundaries.
3. Terrain derivatives — Slope, aspect, multi-directional hillshade, plan/profile curvature, TPI, TRI and TWI computed using ArcPy; all co-registered to a common grid.
4. Vegetation height model — Canopy Height Model (CHM) = DSM − DEM; used to characterise vegetation mass loading on unstable slopes.
5. Geohazard susceptibility index — Slope angle, curvature, TWI and lithology combined in a logistic regression model trained on a mapped landslide inventory; output classified into 5 susceptibility classes.

Tools & stack

Outcomes

• Delivered 11 terrain derivative rasters and a geohazard susceptibility map within a 10-week project schedule.
• Susceptibility model identified 14 previously unmapped debris-flow initiation zones confirmed as moderate-to-high risk by field inspection.
• Bare-earth DEM achieved RMSE of 0.08 m against 120 independent GNSS check points (Class 1 accuracy per ICSM Guidelines).
• Outputs incorporated into the client's infrastructure corridor risk assessment, informing re-routing of a proposed access road around two high-susceptibility zones.