November 8, 2022

Troubleshooting Lidar Ground Classification

Written by: Amanda Lind


The automatic classification tools in Global Mapper make point cloud classification easy! In this blog, I will discuss a few lidar classification tips we share with users who submit data to the technical support box ( Often, it’s a case of using the tools in the correct order or adjusting the settings to fit your point cloud’s structure. The lidar classification tools will work on just about any point cloud, including aerial lidar, terrestrial lidar, or photogrammetric point clouds. For terrestrial lidar, try the Segmentation tool.

When we, humans, look at a point cloud, we can easily distinguish buildings apart from the ground, powerlines, or other landmarks. Computers, however, must identify objects in point clouds based on their attributes and structure. Global Mapper’s Automatic classification tools are each built to look for specific types of structures in the data. For example, the Ground Classification tool will look for characteristics in points such as areas of low elevation change, the last return, etc. To further tailor this process, each Automatic Classification tool has settings that let you tell Global Mapper what to expect with your particular point cloud. Adjusting these settings accordingly can help you take full advantage of the Automatic Classification tools. 

Terrestrial lidar at ground level
Structures in this point cloud are easily identified visually.

The classification tools are loosely made to be run in a specific order: noise, ground, buildings and vegetation, powerline, and finally, powerpole. The order is important because the tools look for data that has already been classified. For example, the building and vegetation classification tool looks for points at a certain distance above the ground. The power pole classification tool can use power lines to help distinguish poles from trees. 

Noise Classification 

First, run the point cloud through the Noise classification tool. This tool will help to remove outliers that would otherwise negatively influence the other classification tools. Noise points are points that don’t represent the surface or structures in the data. These can sometimes be birds that flew between the sensor and the ground, rain that fell throughout the dataset, error from light reflecting off of a windshield, or something else.   

There are two parts to the Noise classification dialogue. The elevation options on the bottom of the dialogue box will automatically classify all data based on an elevation range. These elevation settings are optional. 

Base bin size is also referred to as Neighborhood. This is how far from each point Global Mapper should search for other points when identifying noise. Points that are isolated will be classified as noise. See how this was applied to rain across a Texas town in the screenshot below. The reflection of the resin caused noise to appear below and above the surface, as well as between the buildings. The noise classification tool classified all of those points. 

Swipe to see how the rain was successfully classified as noise in this USGS lidar data.

Ground Settings 

Before adjusting the settings, it’s always good to run the data, or part of the data, with the default settings to see what structures of points are being missed. The default settings work well with most data, but just like the other classification tools, you sometimes need to adjust the settings to tell Global Mapper what to expect with your data. Some of these settings help Global Mapper differentiate between ground and non-ground points, such as Max Building Width. There are two settings that are frequently missed: the Minimum Height Departure and the Expected Terrain Slope.  

Minimum Height Departure from the Local Mean setting tells Global Mapper how thick your ground layer is supposed to be. This value can be measured in the Path Profile tool. Draw your path across a bare ground portion of your data, and measure how thick the cloud is. Lower this value to keep it from creeping up the base of trees or buildings.

Swipe to see how adjusting height departure improved the ground classification.

The Expected Terrain Slope asks for your dataset’s highest degree of slope. This setting can be measured in the Path profile tool. In point clouds with steep slopes in certain areas, you may need to process the cloud in sections based on the settings needed to address each area. 


These settings adjustments should get you started and enable you to get better results from the automatic classification tools for ground points. Stay tuned for future blogs covering the other classification tools:  buildings, vegetation, powerline, and powerpole, and segmentation.  

Contact for more specific help in classifying your data. More information about Global Mapper lidar classification tools and settings can be found here in the knowledge base

For full lessons on how to manage your data in Global Mapper, consider enrolling  in an online, hands-on training class

For blog topic requests, please email

Companies using Blue Marble’s geospatial technology