Print

Airborne laser scanning (also known as airborne lidar) is an efficient method of defining the shape of the ground beneath an aircraft. But it can be done well or badly, depending on the choices one makes. How Lidar Works Essentially, when undertaking a lidar survey, the aim is to direct a stream of laser points through at least one rapidly rotating mirror so as to create a pattern of points left and right of an aircraft travelling at perhaps 220 kph. The aircraft speed and attitude, and the mirror angle at the time each point is transmitted and received, plus the transit time of each point is recorded.

RELATED ARTICLES Laser Scanners (Part 1): Understanding the Technology Wal Mayr, Monday 22 Dec 2008

The co-ordinates of each reflected point are computed from this data– sometimes many returns for each pulse. Ground and non-ground points are separated and each swathe is adjusted to obtain a homogeneous dataset. Finally, the points are transformed into the client’s nominated system.

Data is routinely delivered with an absolute vertical accuracy of 100 mm and a horizontal accuracy of 300 mm. I wonder how we do it sometimes, but we do!

A number of variables make a difference to the quality of the data that is captured.

Variables

All these factors must be considered before finalising the design of a survey:

• Pulse repetition rate
• Flying height
• Multi-pulse
• Flying Speed
• Scan Angle
• Scan frequency
• Beam divergence

Pulse Rate

The pulse rate of an aerial laser scanner is a key differentiator. To a first approximation, the higher the pulse rate the better. A high pulse rate is a useful attribute when defining complex terrain or cityscapes.

However, there are downsides which should be considered during project design. Perhaps the most significant of these is the fact that since the energy output from the system is finite, the higher the pulse rate, the less energy is emitted in each pulse. The less energy emitted, the less the likelihood that a return signal will be received by the aircraft.

Experience shows that the ground beneath dense vegetation is often modelled more reliably with a lower pulse rate.

Point Spacing

Regarding point spacing, it is important to consider the survey objectives, the shape of the ground and the non-ground features, as well as the ground cover, when deciding how many points per square metre should be specified. The issue should be more about how many points are needed – either on the ground or in the aboveground data – to achieve an agreed outcome, rather than how many points it is possible to emit.

First generation scanners achieved eye safety by automatically stopping the laser if the detected laser beam approached dangerous levels.But many of the latest scanners are designed to be eye-safe, and still operate at a low level. One way they achieve this is with a larger beam divergence. This creates a bigger footprint on the ground, and thus puts less energy into every square centimetre. But the larger the footprint, the more likely it is there will be significant variations in height within the footprint. The result is decreased horizontal accuracy, and decreased vertical accuracy on sloped terrain. The return signal from each footprint comes from the part of it with the highest reflectivity.

Flying low produces closely spaced points, but it also requires more swathes to cover a given area. This leads to longer and more expensive data acquisition. Moreover, the additional swathes increase the likelihood of mismatches between runs, making it more difficult to adjust every swathe to achieve a homogeneous dataset.

Multi-pulse

Scanners now have the ability to emit and track more than one pulse at a time. This is known as multi-pulse. Older airborne laser scanners were only able to track one pulse at a time. They recorded each pulse, and did not emit the next pulse until the first had been reflected and received back at the aircraft. As the available pulse rate increased, it became possible for more than one pulse to be in the air at once at higher flying heights.

The development of a timing chip that could differentiate sequential pulses helped eliminate errors caused by the return signal being confused with the return from the next, or previous pulse. Laser scanners with multi-pulse capability can fly higher – wider swathes, fewer swathes required, less building shadowing – and still produce closely spaced reflected points.

But remember, higher flying heights mean lower vertical accuracy, larger footprints and increased positional uncertainty. Yet such reduced vertical accuracy, perhaps 30 or 50 mm, may be acceptable, particularly if the acquisition time and acquisition cost is reduced by 25 per cent.

Accuracy Versus Cost

While the accuracy of airborne laser scan data is important, it is also important that potential users carefully consider how the data is to be used, and the level of accuracy required. There is little point in designing a survey to achieve a vertical accuracy of, say 100 mm, when the ground is covered with dense bushes or clumps of grass. Remember that laser light passes between leaves, tree limbs, etc. It does not pass through these objects!

Over-specifying a survey leads to significantly increased costs, and may not achieve the desired result anyway. High data accuracy, even in vegetated terrain, can be achieved using other measurement tools. Where these costs are prohibitive, a laser scan survey may be the most cost-effective way of achieving the required dataset.

Part 1 of this series makes an interesting point regarding the difference in point density across the track compared with along it. Minor changes in the shape of small objects may be missed if reflected points are too widely spaced. Most airborne scanners today allow the pulse pattern to be modified so that along and across track distances more closely match.

The issue of the pattern made by the points on the ground raises some interesting issues. Do they need to be in a grid? Is this more a modelling software requirement than a need for ground definition?

Experience has shown that if the average point spacing specified for a laser scan survey is sufficient to model the ground to the required accuracy, the act of interpolating points to derive a grid does not degrade the result. If the random points are spaced such that the ground shape is not recorded, interpolation could degrade the result even more.

Conclusion

Airborne laser scanning enables great variation in data density and accuracy. This means that suppliers using photogrammetry, lidar or other such products to produce spatial data need to communicate with their clients to fully understand their expectations, and the limitations of the technology. Once this is done, an appropriate solution can be proposed. Laser scanning is only one.

ASM Newsletter is distributed by e-mail every week. It reports on events in the spatial industry in Asia and events outside the region which will impact on Asia.