A Quick Guide to LiDAR: Part 1- Theory
A concise article on important concepts and terms about LiDAR which are necessary to understand its working and usage.
What is LiDAR?
LiDAR stands for Light Detection and Ranging. It is a remote sensing method for measuring the exact distance of an object on the earth’s surface. LiDAR uses a pulsed laser to calculate an object’s variable distance from the earth’s surface.
LiDAR has 3 primary components —
- Scanner
- Laser
- GPS receiver
How does LiDAR work?
LiDAR works by emitting laser pulses on the earth’s surface and calculating the time it takes to return. The laser hits the objects on the earth’s surface, reflects off them, and is recorded by the sensor. The distance is calculated by the following formula:
Types of LiDAR Systems
There are two types of LiDAR systems:
1. Airborne LiDAR
As the name suggests, in an Airborne system, LiDAR is installed on helicopters or drones to collect data. It is further divided into 2 subcategories:
a) Topological LiDAR — It is used to derive surface models like Digital Terrain Model (DTM), Digital Surface Model (DSM), etc.
b) Bathymetric LiDAR — It is used to analyze elevation at shores, and detect objects at water depths. Bathymetric LiDAR uses green light of wavelength 532 nm since LiDAR laser (Infrared) cannot penetrate water.
2. Terrestrial LiDAR
Terrestrial LiDAR systems can be Mobile or Static. Mobile LiDARs are installed on moving vehicles like cars, boats, etc. whereas Static LiDARs are stationary and are installed on a tripod stand.
How do LiDAR measure trees, buildings, and ground?
As the laser travels towards the earth’s surface, it hits leaves, branches, and other objects. Some of the light is reflected by them while some travel to the ground. Therefore, one single pulse has multiple returns. The last returns are used to calculate the elevation of the ground. The first returns give us the elevation of the objects on the ground.
The height of an object is calculated by subtracting the elevation of the object from the elevation of the ground. In the case of a building or flat surface, all the laser pulses are reflected back with one single return and their standard deviation is almost zero. But the elevation of buildings is larger than that of the ground. This difference is used to tell whether the surface is ground or a roof-top of a building. If the ground around the building is not a flat surface, then we consider either the maximum of the ground elevation points or take the average of those points to compute the ground elevation value. In the case of trees, there are multiple returns and their standard deviation is a large value. Thus, we can differentiate between trees, buildings, and ground surfaces.
Additionally, the amount of energy returned to the LiDAR creates an intensity waveform. The strength of the intensity varies with the surface of the object from which it is reflected. For example, water absorbs almost all of the light. Therefore, it shows a very low intensity.
Below is an example of LiDAR height and intensity from the MUUFL Gulfport Dataset.
Specifications of LiDAR
- The wavelength of LiDAR lies in the Infrared range i.e., 780 nm- 1mm of the electromagnetic spectrum. Mostly, the wavelength of 1064 nm is used.
- The Pulse Repetition Frequency (PRF) of LiDAR can be high as 100 kHz or 150 kHz or even greater than that. PRF is defined as the number of pulses per second emitted by the laser.
- LiDAR is operated at an altitude ranging from 152 m to 914 m or 914 m to 2438 m above ground level (AGL).
- Spatial Resolution is the smallest area each pixel or point represents. For example, let’s say there are two LiDAR sensors, the first one has a spatial resolution of 1m and the second one has that of 5m. Then, the first sensor has a higher resolution as compared to the second one.
- The angle scan has a value between -90° and 90°.
Elevation Models
There are 3 Elevation models or Digital Elevation Models (DEM) —
1. DSM (Digital Surface Model)
A Digital Surface Model captures all the natural and man-made features on the earth’s surface. In short, DSM scans the surface as it is with each and every feature. DSM is calculated by considering all the first returns from the laser. It basically gives the surface of all objects.
2. DTM (Digital Terrain Model)
A Digital Terrain Model captures the ground excluding buildings, trees, and other features. It filters out all the non-ground features on the earth’s surface. A DTM is calculated by considering all the last returns from the laser.
3. CHM (Canopy Height Model)
A Canopy Height Model is given by subtracting DTM from DSM. It gives the height of trees from the ground.
How is LiDAR data stored?
LiDAR data is stored in files in standard LAS format.
A LAS file is an industry-standard binary format for storing airborne lidar data.
It is maintained by the American Society for Photogrammetry and Remote Sensing (ASPRS). The ASPRS codes can be found here:
The LAS file contains the data of the LiDAR survey in the header. The header contains the following: Point format, creation date, point count, scales, offsets, maxs, mins, number of points by returns.
Point format contains the following: X, Y, Z, intensity, return_number, number of returns, scan direction flag, edge of flight line, classification, synthetic, key point, withheld, scan angle rank, user data, point source id, gps time, red, green, blue, and reversible index.
If the data is colorized, then RGB (red, green, blue) values will also be present.
Below is an example of LiDAR las file data taken from the NEON data portal.
Conclusion
I presented all the theoretical aspects of LiDAR in a concise manner in this article. This information can provide a beginner with the necessary background to start working on LiDAR.
Next steps
In this Part 1, I talked about all the theoretical details of LiDAR. In Part 2, I will talk about visualizing and analyzing LiDAR using python. I will discuss how to display point cloud using Open3D and using the LiDAR data for the classification of landscapes.
Thanks for reading! I hope you found this article helpful.
Go Gators!🐊
