LiDAR Processing

LiDAR point clouds are the highest-resolution elevation data source available for most practitioners. WbW-QGIS exposes the full Whitebox LiDAR pipeline — from quality assurance through ground classification, surface modelling, and height normalisation — directly in the QGIS Processing Toolbox.

This chapter walks through a complete bare-earth and canopy-height workflow starting from a raw LAS/LAZ file.

Key Concepts

• Point cloud: A set of 3-D coordinates (X, Y, Z) plus attributes (intensity, return number, classification, scan angle, etc.) acquired by laser scanning from airborne or terrestrial platforms.
• LAS/LAZ: The industry-standard binary format for point clouds. LAZ is a losslessly compressed variant. COPC (Cloud-Optimised Point Cloud) is a tiled LAZ variant for efficient streaming.
• Classification codes: Numeric labels assigned to points indicating surface type (1 = unclassified, 2 = ground, 3–5 = vegetation, 6 = building, etc. per ASPRS convention).
• DTM: Digital Terrain Model — a raster surface interpolated from ground-classified points only.
• DSM: Digital Surface Model — a raster surface from first returns, representing the tops of all objects (vegetation, buildings).
• CHM: Canopy Height Model — DSM minus DTM, representing object height above ground.
• Height above ground (HAG): Per-point elevation relative to the interpolated ground surface. Enables classification of vegetation returns by height tier.

End-to-End Workflow: DTM, DSM, and Canopy Height Model

Inputs

Step 1 — Point Cloud Quality Check

Processing Toolbox → Whitebox Workflows → LiDAR → LiDAR Point Stats

Review the report for:

• Point density (pts/m²)
• Classification distribution (% ground, vegetation, unclassified)
• Z range and intensity histogram
• Scan angle range (should be ±20° for most airborne missions)

Step 2 — Thin High-Density Files (Optional)

For files > 50 pts/m², thinning reduces processing time with minimal accuracy loss.

Processing Toolbox → Whitebox Workflows → LiDAR → LiDAR Thin

Step 3 — Classify Ground Points

If the input file has all points as unclassified (class 1), classify ground returns before surface modelling.

Processing Toolbox → Whitebox Workflows → LiDAR → LiDAR Ground Point Filter

For complex terrain (steep slopes, dense vegetation), increase radius to 4.0 and reduce minimum slope to 2.0.

Step 4 — Build DTM from Ground Points

Processing Toolbox → Whitebox Workflows → LiDAR → LiDAR IDW Interpolation

Alternatively, use LiDAR TIN Gridding for faster interpolation on uniformly distributed clouds.

Step 5 — Build DSM from First Returns

Processing Toolbox → Whitebox Workflows → LiDAR → LiDAR IDW Interpolation (second pass)

Step 6 — Canopy Height Model

Subtract DTM from DSM using the QGIS Raster Calculator, or use the dedicated CHM tool.

Processing Toolbox → Whitebox Workflows → LiDAR → Canopy Height Model

Apply a stretch from 0 to the 98th percentile height value. Negative cells (< 0) indicate DTM–DSM interpolation artefacts; clamp to 0 in post-processing.

Step 7 — Height Above Ground Normalisation

Assign a HAG value to every point for per-return vertical stratification analysis.

Processing Toolbox → Whitebox Workflows → LiDAR → Height Above Ground

The tool sets the Z coordinate of each point to its height above the interpolated ground surface. Ground points are set to 0.

Python Console Equivalent

import processing

cloud = '/data/cloud.laz'

# Step 1: stats / QA
processing.run('whitebox_workflows:lidar_point_stats', {
    'input': cloud,
    'output': '/data/cloud_stats.html',
})

# Step 3: ground classification
processing.run('whitebox_workflows:lidar_ground_point_filter', {
    'input': cloud,
    'radius': 2.0,
    'min_slope': 5.0,
    'max_slope': 85.0,
    'terrain_type': 'Normal',
    'output': '/data/cloud_classified.laz',
})

# Step 4: DTM
processing.run('whitebox_workflows:lidar_idw_interpolation', {
    'input': '/data/cloud_classified.laz',
    'parameter': 'elevation',
    'returns': 'Last',
    'classes_included': '2',
    'weight': 2.0,
    'radius': 2.5,
    'min_points': 3,
    'resolution': 0.5,
    'output': '/data/dtm.tif',
})

# Step 5: DSM
processing.run('whitebox_workflows:lidar_idw_interpolation', {
    'input': '/data/cloud_classified.laz',
    'parameter': 'elevation',
    'returns': 'First',
    'classes_included': '',
    'weight': 2.0,
    'radius': 2.5,
    'min_points': 3,
    'resolution': 0.5,
    'output': '/data/dsm.tif',
})

# Step 6: CHM
processing.run('whitebox_workflows:canopy_height_model', {
    'input': '/data/cloud_classified.laz',
    'dtm': '/data/dtm.tif',
    'resolution': 0.5,
    'output': '/data/chm.tif',
})

# Step 7: HAG
processing.run('whitebox_workflows:height_above_ground', {
    'input': '/data/cloud_classified.laz',
    'output': '/data/cloud_hag.laz',
})

print("LiDAR pipeline complete.")

Common Pitfalls

Validation Checklist

• Point stats report shows expected classification distribution (> 5 % ground points).
• DTM has no large NoData holes in vegetated areas.
• DTM is smooth with no pits deeper than 1–2 m.
• DSM ≥ DTM across the entire overlap extent.
• CHM values are 0 on roads and bare ground.
• HAG-normalised cloud has all ground points at Z ≈ 0 (±0.1 m).