What is LiDAR data?

LiDAR data is the set of 3D points produced by a laser scanner that times the return of each pulse to measure distance. Each point has x, y, z coordinates plus intensity and a classification such as ground, building, or vegetation. Airborne LiDAR data maps terrain and structures at centimeter to decimeter accuracy.

What LiDAR data is

LiDAR stands for Light Detection and Ranging. A scanner emits laser pulses, a sensor records the time each reflection returns, and distance follows from the speed of light.

A GNSS receiver and an inertial unit fix the scanner’s position and orientation, so each return becomes a georeferenced point.

The raw product is a point cloud, a file of millions to billions of points with no grid and no connectivity. That is the property that defines LiDAR data. It samples surfaces directly, in three dimensions, at whatever density the flight delivered.

A single pulse can return several times, first from a canopy top and last from the ground below it. Multiple returns are why LiDAR data maps terrain under forest.

Airborne LiDAR data covers regions from aircraft or drones. Terrestrial and mobile scanners cover buildings and road corridors from the ground. Spaceborne instruments such as ICESat-2 and GEDI sample the Earth in profiles along the orbit track.

What LiDAR data does

LiDAR data supplies the elevation surface under most modern maps of terrain, buildings, and forest structure.

The USGS 3D Elevation Program (3DEP) is building a LiDAR-derived elevation dataset of the United States at 2 points per square meter or better. Its 1 m DEMs feed FEMA flood mapping.

Foresters compute canopy height by subtracting the ground surface from the first-return surface, then estimate biomass and fuel load from the result.

Utilities fly transmission corridors each year to measure conductor clearance to vegetation and ground.

Insurers and assessors extract building height, roof shape, and roof pitch from classified building points.

Coastal agencies compare surveys from different years to measure dune erosion and cliff retreat in centimeters.

Each use starts from the same file and diverges at the classification step, where points are labeled ground, vegetation, building, water, or noise.

How LiDAR data works

LiDAR data is stored in LAS, the ASPRS binary standard, or LAZ, its lossless compressed form. Cloud Optimized Point Cloud (COPC) is a LAZ variant with a spatial index so clients read one region over HTTP.

Each point record holds x, y, z, intensity, return number, number of returns, scan angle, GPS time, and a classification code.

ASPRS class	Meaning
1	Unclassified
2	Ground
3, 4, 5	Low, medium, high vegetation
6	Building
9	Water
7, 18	Low and high noise

Density sets what the data can resolve. USGS quality level QL2 requires at least 2 points per square meter; QL1 requires 8. A QL1 survey resolves individual roof planes and utility poles.

Vertical accuracy for airborne LiDAR data is typically 5 to 10 cm RMSE on open ground, worse under canopy and on steep slopes.

Processing runs in stages. Points are classified, then gridded. Ground points produce a digital terrain model, first returns produce a digital surface model, and the difference is a height model of everything above ground.

Tile size sets the working unit. A 1 km tile at 8 points per square meter holds 8 million points, so regional work means thousands of tiles processed in parallel.

LiDAR data in Wherobots

WherobotsDB works with LiDAR data in two forms: point tables and the rasters derived from them.

Point clouds convert from LAS or LAZ to GeoParquet with PDAL, one row per point, and load as a Havasu table with an ST_PointZ geometry. From there ST_3DDWithinand ST_3DDistance run 3D proximity queries, and a spatial join with ST_Intersectsassigns every point to a parcel or building footprint.

Derived rasters need no conversion. A 1 m 3DEP DEM or a canopy height model stored as GeoTIFF loads with RS_FromPath as an out-db raster, so pixels are read only when a query touches them.

RS_ZonalStats returns mean or maximum height per building polygon. RS_MapAlgebra subtracts a terrain model from a surface model across two rasters to produce a height model in one statement.

Because raster and vector data share one engine, a building footprint, its LiDAR-derived height, and its parcel record join in the same query.

Where LiDAR data does not exist, RasterFlow’s Meta CHM v1 model estimates canopy height in meters from 60 cm NAIP imagery. The result is a raster that RS_ functions read like any other.

-- Building heights from a LiDAR-derived surface model (DSM) and terrain model (DTM)
WITH height AS (
  SELECT RS_MapAlgebra(dsm.rast, dtm.rast, 'D',
           'out = rast0[0] - rast1[0];', -9999.0) AS rast
  FROM lidar.dsm_1m dsm
  JOIN lidar.dtm_1m dtm ON dsm.tile_id = dtm.tile_id
)
SELECT b.id,
       RS_ZonalStats(h.rast, b.geometry, 1, 'max') AS height_m
FROM height h
JOIN wherobots_open_data.overture_maps_foundation.buildings_building b
  ON RS_Intersects(h.rast, b.geometry);

Related terms

• Digital elevation model: the raster most LiDAR surveys are turned into
• Raster vs vector data: why points and derived grids are handled differently
• Remote sensing: the discipline LiDAR belongs to as an active sensor

Read more from Wherobots

• Estimating canopy height with RasterFlow
• Raster support in Havasu
• El Niño 2026, atmospheric rivers, and California’s burn scars: one query, two data models
• Loading geospatial data with Wherobots

Load LiDAR-derived elevation rasters into WherobotsDB on the free tier at cloud.wherobots.com.