19.1 LiDAR Fundamentals

Key takeaways

• LiDAR (Light Detection and Ranging) emits laser pulses and measures return times to build dense 3D point clouds.
• Airborne, terrestrial, and mobile LiDAR serve different scales.
• Classical products: DTM, DSM, canopy height, buildings, power lines.

Introduction

Once a specialty technology used for forestry and flood modelling, LiDAR now covers entire countries at metre or better resolution. It's become the default for high-precision terrain, infrastructure, and vegetation analyses. This lesson covers the fundamentals.

How it works

A LiDAR sensor emits a short laser pulse and measures the time for the pulse to return. Combined with the sensor's precise position (GNSS) and orientation (IMU), this gives the 3D coordinates of the reflecting surface.

• Pulses per second: 100 000 – 1 000 000 typical.
• Range: cm accuracy at close range; tens of cm at airborne distance.
• Wavelength: 1064 nm (near-IR) for topographic LiDAR; 532 nm (green) for bathymetric.

Return types

A single laser pulse may reflect off multiple surfaces (tree canopy, understorey, ground). Modern sensors record multiple returns per pulse:

• First return — top of canopy, rooftops.
• Intermediate returns — branches, intermediate vegetation.
• Last return — typically ground (if the pulse penetrates).

Some sensors (full-waveform) record the entire return signal, not just discrete peaks — enabling finer analysis.

Point-cloud structure

A LiDAR dataset is a set of 3D points, typically carrying:

• x, y, z — coordinates.
• intensity — strength of the return (surface reflectivity proxy).
• return number and number of returns — which return of the pulse, out of how many.
• classification — ground, vegetation, building, water, etc. (per ASPRS standard).
• GPS time.
• scan angle.
• Optional RGB from a co-registered camera.

Formats

• LAS — text-friendly specification; binary file.
• LAZ — compressed LAS (~10× smaller); the de facto distribution format.
• COPC — Cloud Optimised Point Cloud — LAS reorganised for cloud streaming.
• Entwine EPT — octree-structured cloud point cloud format.

Platforms

Airborne LiDAR (ALS)

Aircraft / helicopter at 500–2000 m altitude. Covers 10 km² – 10 000 km² per mission. Density 1–50 points/m². Used for:

• National elevation datasets (USGS 3DEP, AHN, Denmark DHM).
• Large-area forest structure.
• Flood modelling.

Terrestrial LiDAR (TLS)

Tripod-mounted stationary scanner. Very dense (1000+ points/m²), very accurate (mm). Used for:

• Architectural surveys.
• Cave / quarry mapping.
• Archaeology.
• Site inspections.

Mobile LiDAR (MLS)

Vehicle-mounted, scanning as it moves. Used for:

• Road corridor mapping.
• Utility inventories.
• Street-level detail.

UAV / drone LiDAR

Drone-mounted. Good for small areas needing sub-10 cm resolution.

Spaceborne LiDAR

• ICESat-2 (2018) — NASA, global coverage with sparse footprints.
• GEDI on ISS (2019–2024) — canopy structure over tropical and temperate forests.

Coverage patchy; density low compared to airborne but globally consistent.

Processing workflow

1. Classify returns into ground, vegetation, buildings, etc. (automatic tools + manual QA).
2. Filter to remove noise and outliers.
3. Grid to rasters — DTM from ground, DSM from first returns.
4. Derive products — canopy height model, slope, buildings, power lines.
5. Vectorise where needed — building footprints, tree centroids.

Tools:

• LAStools — fast, commercial/free hybrid.
• PDAL — open-source pipelines.
• CloudCompare — interactive point-cloud editor.
• WhiteboxTools — hydrology-focused.
• ArcGIS / QGIS — built-in LiDAR support.

Accuracy

Typical airborne LiDAR:

• Vertical: 5–15 cm at 1–2 m point spacing.
• Horizontal: 10–30 cm.

Terrestrial LiDAR:

• Vertical and horizontal: mm-level over short distances.

Always report both specifications from the acquisition metadata.

Density considerations

More points = better detail but more storage:

• 1 pt/m² — good enough for 1 m DEM; typical free public LiDAR.
• 10 pt/m² — captures individual trees, ridgelines.
• 100 pt/m² — captures power lines, building details.
• 1000+ pt/m² — architectural detail.

Public LiDAR sources

• USGS 3DEP (USA).
• AHN (Netherlands; nationwide ≤5 cm).
• DHM (Denmark; 40 cm).
• Environment Agency LiDAR (UK).
• SwissALTI3D (Switzerland).
• Toposwiss, NLS Finland, etc.

Many are free under permissive licences.

Self-check exercises

1. Why does LiDAR give you both a DSM and a DTM from the same flight?

Pulses reflect off multiple surfaces — the first return comes from the top (canopy, rooftop), later returns come from layers below, and last returns often reach the ground. Gridding first returns gives a DSM; gridding ground-classified returns gives a DTM. Subtracting yields canopy or building heights.

2. What's the main format difference between LAS and LAZ?

LAZ is compressed LAS — identical data but typically 5–10× smaller on disk. Most tools read both transparently. Distribute and archive as LAZ; convert to LAS for tools that only accept the uncompressed format. Modern tools (PDAL, LAStools, CloudCompare) handle LAZ natively.

3. Why might spaceborne LiDAR (ICESat-2, GEDI) not replace airborne LiDAR?

Spaceborne coverage is sparse — GEDI samples discrete footprints (~25 m) with gaps between, not continuous coverage. Airborne LiDAR scans every square metre. Spaceborne is excellent for global monitoring of coarse structural parameters (canopy height trends); airborne is essential for site-level analysis where every tree or roof matters.

Summary

• LiDAR = laser scanning, producing dense 3D point clouds.
• Multiple returns let you separate ground / vegetation / buildings.
• LAZ is the distribution format; PDAL / LAStools are the processing tools.
• National programmes deliver free high-resolution LiDAR in many countries.

Further reading

• PDAL documentation.
• LAStools documentation.
• NOAA Digital Coast — LiDAR 101 introductory guide.
• ASPRS LAS Specification.