LiDAR: From Point Cloud to Building Footprints

This tutorial turns a raw LiDAR flight into analysis-ready products: clean the cloud, classify ground and buildings, derive height and surface models, and pull out individual trees. It uses the LiDAR tools (desktop app — the tools read and write local files).

Desktop app required The LiDAR tools run locally in the desktop app. Convert ASCII clouds with Ascii To Las first if your source is not already LAS/LAZ.

1. Load and trim the cloud

1. If your source is ASCII, run Processing → Whitebox Toolbox → LiDAR → Ascii To Las.
2. Run Processing → Whitebox Toolbox → LiDAR → Clip Lidar To Polygon to cut the cloud to your area of interest — everything downstream gets faster.

Download a USGS survey clip

For a quick extract without a local LAS/LAZ file, open Plugins → USGS LiDAR and choose EPT (AWS Open Data). Search the map extent (or draw a smaller rectangle), select one or more surveys, and choose Download Clipped COPC. The remote service limits each request to five square miles; the selected EPT URLs are sent to the service to produce a .copc.laz download. The Save PDAL pipeline action also saves a portable .pipeline.json artifact containing the Web-Mercator bounds, selected surveys, and the readers.ept/writers.las stages. Keep that artifact with your project when you need a reproducible local clip or want to run the pipeline with PDAL later.

The USGS panel streams point clouds in the browser. The generated file download uses the browser download flow; in the desktop app it opens the system browser so the Tauri webview does not retain a hidden download. No service token is stored in the project or included in error messages.

2. Clean the noise

Run Processing → Whitebox Toolbox → LiDAR → Filter Lidar Noise to drop the birds, balloons, and sensor artifacts that float above or sink below the scene. Check the point count before and after; a few percent removed is normal.

3. Classify ground and buildings

1. Run Processing → Whitebox Toolbox → LiDAR → Improved Ground Point Filter to separate the terrain returns.
2. Run Processing → Whitebox Toolbox → LiDAR → Classify Lidar to label the remaining points.
3. Run Processing → Whitebox Toolbox → LiDAR → Classify Buildings In Lidar to flag roof planes.

Use Colourize Based On Class to render every class in its own color and eyeball the classification before trusting it.

4. Derive the height model

• Run Processing → Whitebox Toolbox → LiDAR → Height Above Ground to normalize the cloud — every point now carries its height above the terrain, which is what building detection and tree analysis actually need.
• Run Processing → Whitebox Toolbox → LiDAR → Lidar Digital Surface Model for the top surface, and a second pass on ground-only points for the bare-earth DEM.

5. Extract features

• Processing → Whitebox Toolbox → LiDAR → Individual Tree Detection returns a point per tree with crown height — useful for canopy audits.
• Processing → Whitebox Toolbox → LiDAR → Las To Shapefile exports the classified building points as vectors you can style and label like any layer.

6. Present the products

Stack the DSM over the bare-earth DEM, add a Lidar Hillshade for relief, and publish the set with a print layout.

Next steps

• Individual Tree Segmentation when you need crown areas, not just stem points.
• Lidar Contour for engineering-grade contours from the bare-earth cloud.