Size and Accuracy Considerations
Updated October 18, 2017
This article applies to:
- Terrain 3D
- Terrain Forestry
- RoadEng Civil
- RoadEng Forestry
The 32-bit version of Terrain is limited to approximately 5 million points. The 64-bit version of RoadEng® can handle more points depending on the speed of the user’s CPU processor and amount RAM, 15 million points is reasonable.
Interpolating the LiDAR into regular grid format is not recommended, because this creates additional points by interpolation (lost accuracy). For accuracy, it is better to work with the raw data points.
When importing LiDAR data, it is very important to group points together instead of making each point an individual feature. Doing this saves memory, because if a feature is attached to many individual points, you will end up with thousands of features, and this will take a long time to load. By grouping points into features, the memory used for that feature is 'shared' by many points.
It is common to have data sets with hundreds of millions of points (well exceeding the recommended maximum of 10 million points). This limitation is generally not a problem for most corridor projects, if points outside the area of interest are thinned. Consider a relatively large road project say 20 kilometers (~ 12 miles). Assume that your LiDAR horizontal resolution is 1 meter (3 feet) and that you have identified a corridor that is 200 meters (~656 ft.) wide along a preliminary alignment. This yields about 4 million data points.