Lidar Explained

This article continues the Elevation & Vertical Datums sub-hub. The /learn/digital-elevation-models-explained article surveyed the major DEM products; this article goes deeper on the technology behind modern high-resolution DEMs — what lidar does, how it works, and what its limits are.

What lidar is

A lidar instrument emits laser pulses, measures the time until each pulse's reflection returns, and computes distance from the time-of-flight. Multiplied across millions of pulses scanned over a target area, the result is a 3D point cloud — a set of XYZ coordinates representing the surveyed surface.

The name lidar was coined in the 1960s as a portmanteau of light and radar, on the analogy with radar (radio detection and ranging). Modern usage:

• Sometimes spelled LIDAR (the original acronym expansion).
• Sometimes LiDAR (a stylistic compromise).
• Increasingly lidar as a normal English word — the ASPRS, IEEE, and most modern technical literature use the lowercase form.

The fundamental measurement is a distance from sensor to target. Combining this with:

• The sensor's position (from GNSS).
• The sensor's orientation (from inertial measurement unit).
• The scan angle of the laser pulse.

...the system computes the absolute 3D position of each reflection point in a known coordinate system.

How the measurement works

The basic time-of-flight equation:

distance = (speed of light × time of flight) / 2

The factor of 2 accounts for the round trip (out and back). The speed of light is 299,792,458 m/s, so:

• 1 nanosecond round trip = 0.15 m
• 100 picoseconds round trip = 1.5 cm

Modern lidar electronics measure time of flight with sub-nanosecond precision — typically 100 ps or better, giving ranging precision of ~1 cm under ideal conditions.

Real systems have additional error sources:

• Beam divergence: the laser pulse spreads slightly with distance; the “point” on the ground is actually a small footprint.
• Atmospheric refraction: laser pulse paths bend slightly through density gradients.
• Sensor position uncertainty: GNSS position uncertainty propagates into point position.
• Time-tagging precision: synchronizing the laser trigger with the GNSS time tag.
• Surface reflectance variation: bright vs dark surfaces produce different return strengths.

Achievable accuracy: ±5–15 cm vertical for airborne systems; ±2–10 mm for terrestrial systems.

Lidar platform types

Airborne lidar (ALS)

Lidar mounted on an aircraft (fixed-wing or helicopter), looking downward and scanning across the flight path. The dominant source of high-resolution topographic data.

Typical specifications:

• Flying altitude: 500–3,000 m AGL.
• Pulse rate: 100,000–1,000,000 pulses per second.
• Point density: 0.5–100 points per m² on the ground.
• Swath width: 200–1,500 m per flight pass.
• Wavelength: typically 1064 nm (Nd:YAG laser) or 1550 nm (eye-safe).

Used for: topographic DEMs, forestry, flood mapping, power-line clearance, transportation corridor surveys, archaeology under canopy.

Terrestrial lidar (TLS)

Lidar mounted on a tripod or stationary platform. Operates at close range (tens of meters to a kilometer). Produces extremely dense point clouds (thousands to millions of points per square meter).

Typical specifications:

• Range: 1–1,000 m.
• Angular resolution: ~0.001° (a few arcseconds).
• Accuracy: ±2–10 mm at close range; degrades with distance.
• Point density: extraordinary at close range (often the resolution is limited by sensor angular resolution rather than scene complexity).

Used for: building information modeling (BIM), heritage preservation, accident reconstruction, forensic 3D scanning, archaeology, structural deformation monitoring.

Mobile lidar

Vehicle-mounted lidar, typically two or more units on a vehicle (forward, side-looking, or 360° rotating). Combines TLS-like density with airborne-like coverage along roads.

Typical specifications:

• Range: 100–500 m typically.
• Accuracy: ±2–5 cm.
• Scanning rate: 1,000,000+ points per second.

Used for: road surveys, mobile mapping (Google Street View / Apple Mappers cars), and increasingly autonomous vehicles — Waymo and many AV developers use lidar as a primary sensor for perception.

Spaceborne lidar

Lidar on satellites. Far higher altitude than airborne (hundreds of km), so footprints are larger and point density per area is lower, but coverage is global.

Current and recent missions:

• ICESat (Ice, Cloud, and land Elevation Satellite): NASA, 2003–2009. GLAS (Geoscience Laser Altimeter System) instrument. Monitored ice-sheet elevation changes.
• ICESat-2: NASA, 2018–present. ATLAS (Advanced Topographic Laser Altimeter System) instrument. Six beams, 532 nm green laser, designed primarily for ice-sheet monitoring but also used for forest canopy height and inland water levels.
• GEDI (Global Ecosystem Dynamics Investigation): NASA, 2018–present. Mounted on the International Space Station. Three lasers, eight beams, 1064 nm IR. Designed specifically for forest canopy height and biomass measurement; covers ±51.6° latitude (the ISS orbit limits).
• CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation): NASA/CNES, 2006–2023. Atmospheric profiling rather than surface elevation.

Multiple returns

A laser pulse can reflect off multiple surfaces on its way through vegetation. Modern lidar systems record each distinct return.

| Return | Typical source | Use | | ------ | -------------- | --- | | First | Top of canopy / building rooftop | DSM generation | | Intermediate | Mid-canopy / multi-story structure | Vegetation analysis | | Last | Bare ground (where reachable) | DTM generation |

Modern systems record 4–7 returns per pulse; research-grade full-waveform lidar records the entire time-amplitude signature, capturing fine canopy structure.

Workflow for separating canopy from ground:

1. Each return is classified by automatic algorithms (and refined by human QC).
2. ASPRS standard classes: 2 = ground, 3 = low vegetation, 4 = medium veg, 5 = high veg, 6 = building, 7 = low point noise, 9 = water, etc.
3. The ground-classified points form the bare-earth point cloud.
4. The bare-earth points are rasterized into a DTM at the desired resolution.

Even under dense forest canopy, enough laser pulses reach the ground (through canopy gaps) to support DTM generation. Typically a few percent of pulses reach the ground in dense forest — enough for 1 m DTM resolution.

Wavelengths

Lidar uses several specific laser wavelengths chosen for:

• Eye safety (regulatory).
• Atmospheric transmission (low absorption).
• Surface reflectance (need bright enough returns).

Common wavelengths:

| Wavelength | Source | Use | | ---------- | ------ | --- | | 532 nm (green) | Frequency-doubled Nd:YAG | Bathymetric (penetrates water); ICESat-2 | | 905 nm | InGaAs / GaAs laser diodes | Compact systems, autonomous vehicles | | 1064 nm | Nd:YAG (fundamental) | Traditional airborne lidar, GEDI | | 1550 nm | InGaAs diodes / fiber lasers | Eye-safe higher power; automotive lidar |

Bathymetric lidar uses green wavelengths to penetrate water; the depth limit depends on water clarity (typically 1.5× Secchi depth, up to ~50 m in very clear water). Topobathymetric lidar combines green (water penetrating) and near-IR (land surface) for seamless land-water mapping.

The ASPRS LAS format

LAS (originally for Laser, now a name) is the standard exchange format for lidar point clouds, published by ASPRS (American Society for Photogrammetry and Remote Sensing). Current version: LAS 1.4 (2013, with updates through 2019).

A LAS file contains:

• Header: file metadata, coordinate system, bounding box, scale factors.
• Variable Length Records (VLRs): extended metadata, including coordinate reference system WKT.
• Point records: one per lidar return.

Per-point data (LAS 1.4 standard):

• X, Y, Z coordinates (scaled integers; the header has scale and offset).
• Intensity (16-bit return strength).
• Return number (1, 2, 3, ...).
• Number of returns in this pulse.
• Classification (ASPRS standard class codes).
• Scan angle.
• GPS time.
• Optionally: RGB color, NIR, scan channel, user data.

LAZ is the compressed variant — typically ~10× smaller than LAS. Most modern tools (PDAL, LASTools, CloudCompare, QGIS, ArcGIS, Python laspy) handle both formats transparently.

For comparison: a 1 km² area scanned at 8 points/m² contains 8 million points. At 30 bytes per point in LAS, that's 240 MB; compressed as LAZ, ~24 MB.

National lidar programs

Multiple countries have ongoing nationwide lidar acquisition programs.

USGS 3DEP

3D Elevation Program, the US national lidar program managed by USGS. Goal: nationwide US coverage at Quality Level 2 (QL2) by 2030.

3DEP quality levels:

| Level | Vertical accuracy (RMSE_z) | Density | | ----- | -------------------------- | ------- | | QL0 | ≤ 5 cm | ≥ 8 pts/m² | | QL1 | ≤ 10 cm | ≥ 8 pts/m² | | QL2 | ≤ 10 cm | ≥ 2 pts/m² | | QL3 | ≤ 20 cm | ≥ 0.5 pts/m² | | QL5 | non-lidar source (photogrammetry) | various |

Coverage as of 2025: approximately 75% of the continental US at QL2 or better. Free download via the USGS National Map portal.

UK Environment Agency

The UK Environment Agency provides 1 m resolution lidar for England. Coverage is particularly dense in flood-risk areas; ~75% of England has lidar coverage. Free download.

Netherlands AHN

Actueel Hoogtebestand Nederland (Current Height Model of the Netherlands). The Netherlands has been fully lidar-mapped multiple times since 1997. AHN4 (2020–2022) covers the entire country at ~50 cm resolution. AHN5 acquisition ongoing.

Finland

Finland's nationwide 2 m DEM is lidar-derived. The National Land Survey of Finland (MML) acquires new lidar across the country every several years.

Other countries

• France (IGN LiDAR HD): ongoing nationwide acquisition.
• Germany (BKG): per-state lidar programs.
• Norway, Sweden, Denmark: national lidar programs at various stages.
• Australia (Geoscience Australia): partial national coverage.
• Japan (GSI): partial coverage for high-population-density areas.

The trend: by 2030, most developed nations will have nationwide lidar coverage.

Spaceborne lidar in depth

ICESat-2 and GEDI are the current operational spaceborne lidars.

ICESat-2 / ATLAS

NASA mission, launched September 15, 2018. The Advanced Topographic Laser Altimeter System (ATLAS):

• Wavelength: 532 nm (green; frequency-doubled Nd:YAG).
• Pulse rate: 10 kHz.
• Six beams: three pairs, with the pairs separated by ~3 km and beams within a pair by ~90 m. The pairing allows derivation of cross-track slope.
• Footprint: ~17 m diameter on the surface.
• Vertical precision: sub-centimeter over flat surfaces (averaged over many photons).

Primary mission: monitoring ice-sheet elevation changes over Greenland and Antarctica. Secondary products: inland water levels, forest canopy heights, ocean surface roughness.

GEDI

NASA mission on the ISS, launched December 5, 2018, operational from March 2019 (with a hiatus 2023-2024 when the instrument was stowed; returned to operations in 2024).

• Wavelength: 1064 nm.
• Eight ground beams: three lasers each illuminating two ground tracks (after dithering), giving eight footprints across track.
• Footprint: 25 m diameter.
• Coverage: ±51.6° latitude (ISS orbit inclination).

Primary mission: forest canopy height and biomass measurement; the resulting data feeds carbon accounting and climate models.

Applications

Topographic DEMs: the dominant use. Modern high-resolution DEMs come from lidar where available.

Forestry: canopy height, tree count, biomass estimation, gap analysis. Both airborne and GEDI feed forest inventories.

Flood modeling: precise DTMs are essential for high-resolution flood-inundation models. UK EA lidar covers most flood-risk areas for this purpose.

Power-line clearance: lidar-derived 3D models of transmission corridors check that vegetation isn't encroaching on power lines.

Transportation corridors: road, rail, and pipeline surveys use mobile and airborne lidar.

Building Information Modeling (BIM): terrestrial lidar produces 3D scans of existing buildings for renovation planning, heritage preservation, and documentation.

Archaeology: lidar reveals subtle ground features beneath vegetation — many ancient sites have been discovered through lidar-derived bare-earth DTMs. Major examples: lidar surveys of the Maya lowlands (Petén, Guatemala) revealed dense ancient urban networks invisible from satellite imagery.

Autonomous vehicles: forward-looking lidar provides 3D perception for self-driving cars (Waymo) and many ADAS systems. Automotive lidar is a major growth area; Velodyne, Luminar, Innoviz, and others ship production automotive sensors.

Snow depth: differencing summer and winter acquisitions gives snow depth maps; used for hydrological forecasting and avalanche analysis.

Common misconceptions

“Lidar is the same as radar.” Both use time-of-flight ranging, but lidar uses laser light (typically near-infrared or visible), radar uses radio waves. Wavelengths differ by 4–5 orders of magnitude. Lidar gives higher angular resolution; radar penetrates rain, fog, and vegetation better.

“Lidar can see through trees.” Partially. Individual laser pulses can't penetrate solid foliage, but gaps in the canopy let some pulses through to the ground. Repeated scanning samples enough gaps to build a DTM. In the densest evergreen forests, ground-return density drops dramatically.

“All lidar produces DEMs.” Lidar produces point clouds; DEMs are one possible derived product. Other products: DSMs, intensity images, canopy height models, building footprints, classified point clouds. The DEM is the most common output but not the raw measurement.

“Higher pulse rate means higher accuracy.” Higher pulse rate gives higher point density (more samples per area), not better ranging accuracy. Accuracy depends on timing precision and platform stability.

“Lidar replaces photogrammetry.” They're complementary. Lidar gives precise 3D structure; photogrammetry adds color and high-resolution imagery. Modern surveys often combine both — lidar for terrain, oblique photogrammetry for texture.

“Spaceborne lidar covers everywhere.” ICESat-2 has near-global coverage (89° N/S). GEDI is limited to ±51.6° latitude by ISS orbit inclination (no Arctic or Antarctic coverage).

“Lidar is too expensive.” Commercial airborne lidar acquisition is in the $0.10–1 per acre range for large areas — affordable for many applications. Consumer-grade lidar in the iPhone Pro and dedicated handheld scanners brings costs down further for small-area use. National free lidar programs (USGS 3DEP, UK EA, AHN) provide free data for many regions.

“LAS files only store lidar data.” The format is lidar-oriented but generic enough to store any geometrically-located point data — some projects use LAS to store photogrammetry-derived points, sonar-derived bathymetry, and other 3D measurements. The classification scheme is lidar-conventional but customizable.

“The first return is always the ground.” Opposite — the first return is the highest reflection, typically the top of vegetation or building rooftop. The last return is often the ground (where pulses penetrate to it). DTMs use the ground-classified points, which are typically last returns but not always.

“Lidar accuracy is uniform across an acquisition.” No. Accuracy degrades:

• Near the edge of the scan swath (oblique angles).
• Over reflective surfaces (water, polished metal).
• Over absorbing surfaces (dark roofs, fresh asphalt).
• Under canopy (ground returns sparse, accuracy degraded).
• In areas of strong GNSS multipath (urban canyons).

Quality control reports detail the spatially-varying accuracy of an acquisition.