Every LiDAR point carries two coordinates that nearly everyone uses (X, Y, Z) and a third value that nearly everyone ignores (intensity). The ignored channel is sensor noise in some workflows and the actual deliverable in others. Knowing which is which is the difference between extracting road markings from a cloud you already have and re-flying because nobody told you it was possible.
If you open a LiDAR cloud in any reasonable viewer and toggle the colour mode to "by intensity", you see something most people never look at: a greyscale image of the ground built from the strength of each laser return. Road markings appear as bright lines; water appears as voids; vegetation has a distinctive grey-on-grey texture; pavement, gravel and concrete each have their own signature.
This is the second LiDAR channel — the intensity value every point carries alongside its X, Y, Z. It's recorded by every modern sensor, included in every LAS file, and mostly ignored downstream. This article is about when it matters, how to use it, and why treating it as a deliverable in its own right opens up products that the geometry channel alone can't deliver.
When a laser pulse hits a surface and reflects, the sensor measures the strength of the return — the fraction of emitted energy that came back. That measurement gets recorded alongside the geometric XYZ position as the intensity value.
LAS 1.4 records intensity as a 16-bit integer (0 — 65,535). Older LAS 1.0 — 1.2 formats used 8-bit (0 — 255). The numbers themselves are sensor-specific — same surface, different sensor, different number — but the relative intensity within a single capture is the meaningful information.
A bright return suggests a surface that reflected a lot of pulse energy at the laser's wavelength. A dark return suggests absorption, off-angle geometry, or distance from the sensor. The interpretation depends on what each of those means for your project.
Five factors affect the recorded intensity:
1. Surface reflectivity at the laser wavelength. Different surfaces have very different reflectivity in near-infrared (the wavelength most topographic LiDAR uses, typically 1064 nm). Vegetation reflects strongly in NIR (think how vegetation appears bright in infrared satellite imagery). Water and asphalt absorb. White concrete, marble, sand and snow reflect strongly.
2. Angle of incidence. A pulse hitting perpendicular to the surface returns more energy than the same pulse hitting at a shallow grazing angle. Edges of formation and steep batters show darker intensity than the flat surfaces either side.
3. Range from the sensor. Return energy falls off with the inverse square of distance (and atmospheric attenuation adds to this). Higher-altitude captures produce systematically dimmer returns than lower-altitude captures of the same surface.
4. Atmospheric attenuation. Humid air absorbs more NIR than dry air. Captures in different conditions produce systematically different intensities even from the same surface.
5. Surface texture, moisture and microstructure. Wet pavement returns less energy than dry pavement. Polished metal returns more than weathered metal. Fresh paint returns more than aged paint.
The interaction of these five factors is why raw intensity is useful within a single capture but unreliable for comparing different captures.
Raw intensity is what the sensor recorded. Calibrated reflectance is what you get after correcting for the things that aren't surface-dependent:
The result is a value that's closer to a true surface reflectance — comparable between captures, comparable between flights, comparable between sensors (in principle).
Most LAS deliveries ship raw intensity. Calibrated reflectance is a processing step — typically not free, sometimes available as an add-on, and worth specifying if your project needs between-capture intensity comparison.
Different topographic LiDAR sensors use different wavelengths, and the same surface looks different at each:
| Wavelength | Class | Vegetation | Water | Asphalt | Concrete | | ---------- | ----- | ---------- | ------ | ------- | -------- | | 532 nm | Green | Mid-bright | Some return | Mid | Bright | | 1064 nm | NIR | Very bright | Dark | Mid-dark | Bright | | 1550 nm | SWIR | Mid-dark | Dark | Dark | Bright |
The 1064 nm NIR wavelength dominates topographic LiDAR. Vegetation appears bright; water dark; asphalt mid-dark; concrete bright. This explains why intensity rasters from typical drone LiDAR look like greyscale infrared imagery — they essentially are.
532 nm green LiDAR is used for bathymetric work (it penetrates water; see bathymetric LiDAR). SWIR LiDAR is niche but useful for some material-discrimination applications.
The reason to care about intensity is what it enables:
White paint has reflectivity 3-5× higher than aged asphalt at NIR wavelengths. Linemarking on a captured road shows up as bright linear features in the intensity raster — extractable as 3D polylines with minimal manual digitisation.
For road corridor projects (see the corridor mapping article), the linemarking deliverable that would otherwise require photogrammetry or manual digitisation comes essentially free from the intensity channel.
Conductors — aluminium, steel-reinforced — return strongly in NIR. Intensity-based filtering can isolate conductor returns from the broader corridor cloud more reliably than position-based filtering alone. For catenary modelling (see this article), intensity is a useful second filter.
Water absorbs NIR almost completely — water surfaces produce very low intensity returns (typically only specular reflections from surface ripples). Intensity-based water masking is more reliable than geometric detection for thin water bodies (creeks, drainage features, ponds).
Different vegetation species have measurably different NIR reflectivity. With calibrated reflectance and ground-truthed training data, machine-learning classifiers can discriminate between species at point level — pine vs eucalypt, deciduous vs evergreen, healthy vs stressed. Niche but valuable for vegetation-management (see this vertical) and ecology projects.
Building roofs come in distinct intensity signatures — Colorbond, tile, slate, membrane, solar panels each return differently. For asset inventory across council areas or insurance datasets, intensity-driven roof classification is production-ready.
On civil sites, distinguishing bare earth from compacted gravel from sealed pavement from concrete pad is sometimes geometrically ambiguous (all are flat). Intensity differentiates them clearly, helping classification of the same cloud into more granular ground-cover classes.
Three things worth knowing about working with intensity:
Sensor-to-sensor consistency is poor. Raw intensity from one sensor model is not directly comparable to another. A multi-vendor project that needs unified intensity products needs explicit calibration to a common reference (typically reflectance, not raw intensity).
Same surface, different captures, different intensities. Atmospheric conditions, angle of incidence and range all change between captures. Multi-temporal intensity comparison requires calibrated reflectance and consistent capture geometry — harder than it sounds.
Intensity is per-return, not per-pulse. A pulse with multiple returns records intensity separately for each. The first return typically has higher intensity than subsequent returns from the same pulse (more energy left for the first surface). This matters for canopy intensity products — first-return canopy and last-return ground have systematically different intensity properties.
The emerging frontier: sensors that fire multiple wavelengths simultaneously and record intensity at each. The result is a true spectral signature per point — not just one intensity value but three or four across NIR / green / SWIR.
Multi-spectral LiDAR is still niche in drone work (sensor mass and power matter), more common in aircraft systems. The applications that benefit:
For most drone topographic LiDAR projects in 2026, multi-spectral is research-grade rather than production-ready. Worth knowing the landscape; not yet worth specifying.
The clean intensity specification:
A clean scoping note: "Intensity included in LAS deliverable (per-return). First-return intensity raster at 0.5 m grid, 16-bit GeoTIFF. Raw intensity is fine — single-capture project."
That sentence eliminates the most common intensity-related delivery confusion.
Intensity is the second LiDAR channel everyone has, nobody asks for, and a meaningful fraction of projects could exploit. Raw intensity is useful within a single capture; calibrated reflectance enables between-capture comparison.
Six application categories actually benefit at production scale (road markings, conductor identification, water masking, species discrimination, roof classification, paved-vs-bare). Multi- spectral LiDAR extends this further but is still niche for drone work.
If your project would value any of these, specifying intensity at scoping costs nothing and unlocks deliverables the geometry channel alone can't produce.
Tell us the deliverable goal (linemarking, conductor isolation, water masking, material classification). We'll specify intensity products at scoping so you receive what the second channel can actually produce — not just the geometry default.
Where the intensity channel earns its keep — road linemarking extraction at scale from the same capture that produces the corridor DTM.
The other 'second-channel' story — how a different wavelength changes what the sensor can see, and what that enables.