Written by: Mackenzie Mills
As more and more individuals independently collect 3D data using drones equipped with Lidar sensors, it is important to discuss the accuracy of the data. Analysis performed using point cloud data is only as accurate as the source data. When working with 3D data you must worry about not only accuracy in the x and y directions, but the elevation, or z, component. This z value usually contains the most variation or error. While GPS receivers provide accurate horizontal information, they can struggle with vertical accuracy. RTK or PPK systems can help improve location information, but can be prohibitively expensive to include in a drone set up. An alternative would be to survey ground control points for use in post processing to help improve the accuracy of your point cloud.
To check for vertical error and correct it, those surveyed ground control points can be used with Global Mapper’s Lidar QC tool.
This Lidar QC tool, included in the Lidar Module of Global Mapper, compares control points with nearby points in the point cloud layer to measure the elevation difference between the two. Before the point cloud is adjusted, the tool provides a report showing the compared points and elevation difference between the point cloud and control points. After reviewing these metrics, you can then choose to apply the adjustment to the point cloud layer.
The adjustment of the point cloud is determined by interpolating a best fit surface for the area of the point cloud based on the elevation differences calculated and reported by the Lidar QC tool. This method allows the point cloud to be adjusted accurately even if there are differing degrees of vertical error in different areas of the cloud.
To look closer at the process and options available in the Lidar QC tool, we will work through an example.
To begin, ensure a point cloud layer and a layer of 3D ground control points have been loaded into a Global Mapper workspace.
Opening the Lidar QC tool from the Lidar Toolbar, select the point cloud layer and control point layer to use. You do have the option to use only selected 3D points as control points instead of using a whole layer of point features.
The maximum distance parameter in the Lidar QC setup determines how far from each control point Lidar returns will be considered and compared to the control point elevations.
The maximum point cloud returns to consider determines how many Lidar points will be compared to each control point. After the maximum number of point cloud points is reached for each control point, the program stops the comparisons even if all the returns within the maximum distance have not been considered.
Clicking the OK button will generate the Feature Measurement Information seen below. To generate the feature measure values for each control point, the Lidar QC tool uses inverse distance weighting (IDW) to determine the point cloud elevation at the coordinate location of each control point. The reported statistics show the point cloud elevation (LIDAR_ELEV), control point elevation (ELEVATION), and the elevation difference (ELEV_DIFF) for each control point.
This is where you could stop if your elevation differences are within your accepted error margin. Additionally, the feature measure metrics reported by the Lidar QC tool are added as attributes to the control point vector features in Global Mapper so you can look up and reference these values later.
If the elevation difference exceeds your accepted error, you can choose to Fit Lidar to Control Points from the Feature Measurement Information dialog. This will use an interpolated best fit surface determined by comparing the point cloud to the control points to vertically adjust your point cloud layer(s) to better fit the control points. As you can see in the path profile below, the adjusted point cloud is several meters above the original
Although we have been discussing this tool in the context of Lidar data, you can use it with any point cloud, such as those constructed from drone images. By ensuring the accuracy of your point cloud, you will have a more accurate terrain grid, contour lines, and other analyses derived from the quality-controlled point cloud layer.