From the Ground Up
Defining LiDARAt first surveyors and mappers were not overly impressed with LiDAR collection. Conventional surveys and traditional photogrammetry were generally more cost competitive and provided higher accuracies than could be attained with LiDAR. But the increases in the capabilities of LiDAR that have occurred over the last several years tell a different story. LiDAR today has application in a diverse group of areas, including city and county mapping, airport obstruction analysis, transmission line surveys, timber inventories, quantity surveys and bathymetric surveys.
LiDAR systems use three powerful measurement technologies: a laser for distance, airborne GPS (ABGPS) for the three-dimensional position of the LiDAR system and an inertial measurement unit (IMU) for the three-dimensional rotation of the system. Internally, the laser emits high-energy light pulses that are directed to a series of mirrors and an optical scanner that oscillates to direct individual laser points in a swath (the width of ground coverage from one pass of the aircraft) perpendicular to the line of flight of the aircraft.
LiDAR actually operates similarly to a reflectorless total station, with the significant differences that LiDAR systems are mounted on a moving aerial platform and operate at tens of thousands of points collected each second. Each pulse of light is precisely measured by high-speed electronics as it is transmitted, and again as its return reflection is detected. The distance to the ground is accurately determined by using the constant speed of light plus time-of-flight calculations. Subsequently, the accurate three-dimensional position of the ground is determined when this time-of-flight distance is coupled with the aircraft position (i.e., GPS data) and rotation of the LiDAR unit (i.e., IMU data), along with the internal sweep angle of the laser.
All returns from the laser are logged and post-processed using specially designed software. Increases in the capabilities of the hardware and software platforms used to process the data are an important part of the growth of today's systems. It is essential to recognize that the collection of information on the ground includes features above the surface of the earth that would not be collected in a traditional mapping project. For example, the raw returns from LiDAR will include data on tops of buildings, vehicles, trees and other cultural features. This data needs to be removed from the dataset for most traditional surveying and engineering projects.
A Brief HistoryLiDAR first became commercially available in 1993 when Optech Inc. (Toronto, Ontario, Canada) introduced a 2-kilohertz (kHz) system. This often-quoted numerical rating represents what is known as the repetition rate, or collection frequency, of the laser. A rate of 2 kHz represents 2,000 pulses of light per second from the laser, and therefore the collection of 2,000 ground points for each second of LiDAR operation.
The repetition rate has grown significantly since the early days of LiDAR and has been instrumental in the improved capabilities of the system. LiDAR platform repetition rates grew to 5 kHz in 1995, 10 kHz in 1998 and 25 kHz in 1999.
The rates available on today's top-of-the-line systems are truly impressive in my view. Leica Geosystems (Atlanta, Ga.) and Optech are the dominant players in the worldwide LiDAR systems market. Leica currently offers an 83 kHz platform, and at the International LiDAR Mapping Forum held in Orlando in February 2004, Optech unveiled its new 100 kHz platform.
The earliest systems did not have the capability to measure multiple returns from the same burst of light from the laser. But by 1995, the ability to measure a first and last return from the laser was implemented-a significant addition to LiDAR systems. We only have to look at what happens in typical vegetation to understand why. When the beam of light from the laser strikes the top of a tree some of the energy is returned to the LiDAR unit. This is measured as a first return and provides information relating to the elevation of the tree canopy. Normally, some of the energy will continue on through the trees. It may subsequently strike a branch, reflecting some of the energy back to the LiDAR unit, which is then measured as a second return. Some of the remaining energy may make it onto the ground where all remaining energy is reflected and measured as a last return. Today's systems are able to measure first, second and last returns, and even first, second, third and last returns.
Value in the NoiseWhat was once considered "noise" in the data is now becoming valuable information. LiDAR was first recognized as an efficient way of collecting elevation information on the earth's surface. Any features above this bare earth surface were typically removed from the data and discarded. For example, returns collected on the tops of buildings, in treetops, on utility poles or streetlights, and on the tops of vehicles were removed during the data processing, leaving an accurate terrain model.
However, there is a lot of value in the cultural features found in this raw data. In some instances, clients are more interested in the model of these cultural features than in the ground surface itself. More often, the data on the cultural features is captured in addition to the ground data, which creates a detailed model of the entire built environment. The hardware and software providers have recognized the value of this additional information and are tailoring their products to meet the demands of both the LiDAR service providers as well as the end users of the data.
A great deal of research is underway to develop software routines for automatic extraction of rooftop data from LiDAR data. The accuracy of this extraction feature is a function of several things, especially the resolution of the LiDAR data. More density in the data obviously results in more accuracy in the extraction of rooftop data or other cultural features. While planning has always been critically important to the success of LiDAR projects, planning takes on even more importance when non-standard features are desired from the LiDAR collect.
Other valuable information in LiDAR "noise" is the collection of above-ground transmission lines. Electric utility companies can benefit greatly from gaining detailed information on all of their transmission lines as well as learning about the heights of the trees that may be in the vicinity of their lines. This information allows the utility companies to monitor clearances above and below these lines. Moreover, it allows them to perform risk analysis relating to the impact of fallen trees.
Gaining a great deal of information on something as small as a 1-inch diameter transmission line can be particularly challenging. Reliable information can be collected, however, by making the cross-track spacing of the LiDAR data less than the illuminated footprint of the LiDAR on the ground. In this scenario, each pulse of light from the LiDAR overlaps the adjoining pulse, and therefore almost guarantees a hit on the transmission line on each pass of the laser.
The Importance of AccuracyAccuracy is an important consideration for any project, and while the accuracies available from LiDAR have greatly improved over the years, LiDAR is not appropriate for every project. Both Optech and Leica publish elevation accuracies at the 1-sigma level (one standard deviation) of 15 centimeters, which is about 6 inches (accuracy is dependent on altitude). Statistically assuming a normal distribution of the data, this means that nominally two-thirds of the elevation points collected with LiDAR should fall within 6 inches of their true elevation on the ground.
Under the same statistical assumptions, 95 percent of the data points should fall within two standard deviations, or in this case within 12 inches. Finally, 99.7 percent of the data points should fall within three standard deviations, or 18 inches.
But the elevation accuracy, while the most often quoted number, is only half of the accuracy story. In addition to the elevation error, there is a certain amount of error in the horizontal location of the elevation points collected with LiDAR. Generally, this error is less than 1/2,000 of the altitude of the aircraft during the collection. Therefore, a maximum horizontal displacement of the ground points of 1 foot could be expected when flying 2,000 feet above the ground. Similarly, an error of 2 feet could be expected at an altitude of 4,000 feet above the ground.
Actual project experiences providing LiDAR services are somewhat better than these published figures. But detailed planning is required, including careful consideration of the strength of the GPS satellite configuration during flight, the maximum distance realized from the ground-based GPS unit (used in the post-processing) to the aircraft, and the field of view for data collection. It is strongly recommended that LiDAR users have an open discussion with their LiDAR service providers regarding accuracy expectations before moving forward on any project.