RICS Draft Global Guidance Note: Earth observation and aerial surveys, 6th edition

RICS Draft Guidance Note: Earth observation and aerial surveys, 6th edition

5 LiDAR

LiDAR instruments are active sensors that emit and receive a laser pulse from an aerial platform, deducing ranges or distances to the terrain by measuring the time taken for the laser pulse to return. Coupled with GNSS/IMU technologies, these ranges are used to compute the 3D position of each laser pulse, forming a LiDAR point cloud (see section 5.4.1) that accurately depicts the terrain below the sensor. Modern LiDAR sensors carry a medium format RGB camera, enabling imagery to be captured simultaneously.

LiDAR technologies have established themselves as the predominant method of obtaining accurate 3D data from aerial surveys and offer advantages over photogrammetric methods such as the ability to capture data at night, during the winter, under trees and irrespective of solar angle.

Bathymetric LiDAR systems are equipped with a laser (frequently referred to as a green laser) operating at wavelengths capable of penetrating inshore waters down to depths of approximately 40m. The result is an accurate 3D model of the sea floor.

Developments in LiDAR technology are underway: Geiger mode LiDAR (GML), single photon LiDAR (SPL) and flash LiDAR are examples of developments that are currently being refined.5 GML and SPL LiDAR have a sensor design that is based on a focal plane array of pixels, as opposed to the single pixel that is used in traditional LiDAR technology, known as linear LiDAR. These technologies divide a single pulse into hundreds of thousands of sub-pulses, providing potentially higher point densities. GML and SPL also require less laser energy, enabling them to be flown at higher altitudes while maintaining point densities and reducing costs. Flash LiDAR, while offering higher point densities, is a relatively high energy system, making it more suitable for low altitude applications. It is unclear whether any of these developments will gain the market penetration that linear LiDAR currently enjoys; linear LiDAR still offers advantages in vegetation penetration and accuracy.

5.1 Key considerations

5.1.1 Point density

Laser point density is a key metric when commissioning LiDAR data. Expressed simply as the number of points per metre squared (ppm2), laser point density is a measure of the spatial resolution of the LiDAR data. The higher the point density, the higher the spatial resolution, and therefore the more detail that will be visible in the data.

The point density is dependent on the laser pulse rate frequency (PRF, also known as the pulse repetition rate or PRR) and the flying speed. These two variables control the rate of data capture. The PRF is simply the number of times the laser fires every second. The higher the PRF, the higher the achievable point densities. High specification lasers can fire at a rate of 2MHz. Slowing down the flying speed of the aircraft will increase the point densities that can be achieved. The flying speed should be slow enough to meet the requirements of the specification, but fast enough to maintain an efficient number of flying hours.

The scan rate is the rate at which the data is captured as the laser is directed back and forth (with the use of a mirror), perpendicular to the direction of travel of the aircraft over the terrain. It is not uncommon for the scan rates of high-end systems to operate up to 600 lines/sec. The scan rate controls the laser point spacing along the aircraft path.

Point density is also influenced by the terrain itself. Areas of water will absorb the laser energy, recording very few points. Vegetation will also absorb a larger amount of the laser energy than manmade surfaces. It is not uncommon for LiDAR specifications to specify point densities and accuracy requirements with respect to hard surfaces.

The choice of LiDAR system should be determined by the eventual use for which the data is being commissioned. For example, a flood risk analysis on a large river catchment may require a point density of around 8ppm2. This type of project is better suited to the larger LiDAR instruments, where the laser is powerful enough to provide an adequate number of returns from high altitudes, keeping flying time to a minimum and reducing cost. Corridor engineering applications such as for road, rail and power line corridors tend to be much smaller but require higher point densities. LiDAR instruments that are 'eye safe' at low altitudes are more suitable for this type of work.

5.1.2 Footprint area

Data coverage is influenced by:

  • the instrument laser scan angle
  • the laser power level and
  • the flying altitude above the terrain.

More powerful lasers can operate at higher altitudes. The laser scan angle, also referred to as the field of view (FOV) of the instrument, controls the footprint of the laser data capture area on the ground. An FOV of 60 refers to 30 of coverage either side of a nadir line drawn directly below the aircraft. As the altitude increases, the footprint on the ground increases and the point density decreases, as shown in Table 5.

Coverage

Point density

Output

FOV ()

Flying height (m)

PRF (Hz)

Scan rate (lines/sec)

Aircraft velocity (km)

Point density ppm2

Capture rate (km2/sec)

60

1,300

1 x 106

200

160

8.1

0.12

30

1,300

1 x 106

200

160

17.4

0.06

30

2,800

1 x 106

200

160

8.1

0.12

60

1,300

1 x 106

200

125

10.4

0.1

60

1,300

1.25 x 106

200

125

12.9

0.1

Table 5: Relationship between FOV, flying height, PRF, scan rate, aircraft velocity, point density and capture rate.

Reducing the LiDAR footprint on the ground by decreasing the FOV will increase the point density if the laser is fired at the same rate and the aircraft speed is maintained. This will also reduce the capture rate, as less terrain will be covered in the same time period. Increasing or decreasing the flying height while maintaining the FOV will influence the coverage area.

An increased point density can be achieved by simply slowing down the aircraft velocity at the expense of reducing the capture rate. Increasing the PRF will also improve the point density while maintaining the capture speed.

5.1.3 Directly geo-referencing LiDAR imagery

LiDAR data is scanned line by line as opposed to being captured in a single frame. The scanner head position should be accurately directly geo-referenced. All LiDAR instruments should have an integrated GNSS/IMU navigation system.

The location of each individual laser pulse at the time of capture is described using three coordinates (easting, northing and height) and three rotations (omega, phi and kappa) around the three principal instrument axes. These systems do require access to GNSS base station data, which can either be data that is captured specifically for the project or data that comes from a continuously operating network.

The camera positions can be output in the coordinate system of the client's choice.

A LiDAR system should also be calibrated after a new installation or at a regular interval of one month during a large project. The calibration procedure involves a special calibration flight at two different altitudes.

The calibration flight is used to deduce the angular difference between the LiDAR sensor and the aircraft coordinate systems to ensure alignment. This provides the angular misalignments, which are then applied to the LiDAR data when producing the laser point cloud (see section 5.4.1).

Ground control points are not normally required for this procedure but are useful for quality control of the captured data. Evidence should be provided by the contractor that the GNSS and associated IMU are calibrated on a test area at regular intervals. This is particularly true when the LiDAR components have been removed and reinstalled on the aerial platform.

5.1.4 Calibration

The LiDAR unit should have a factory calibration certificate, valid for a two-year period, prepared by the instrument manufacturer. Several internal sensor parameters should be measured and compared against the values at the time of manufacture. The sensor model should be adjusted accordingly to maintain its accuracy.

5.2 Flying and coverage

5.2.1 Flight lines and overlap

Professional flight planning software is used by contractors to consider all the factors affecting point density and will enable coverage of the client's AOI in as few flight lines as possible. The flight planning is completed using an underlying DTM, which takes into consideration the effects of the local topography of the area.

Flight line planning for LiDAR sensors follow the same principles as those for vertical aerial photography detailed in section 4.2.1. The best results are achieved by organising the coverage of an area in straight flight lines that are as level as possible. Coverage of corridor features such as for transport infrastructure can be achieved via the use of additional flight lines to capture the bends in the most economical manner.

Overlaps between flight lines are usually at between 15% and 35% and are necessary to ensure that there are no gaps. Cross strips are also frequently flown to assist in the data processing by tying blocks of flight lines together.

5.2.2 Acceptable quality limits

The following list is intended to act as a set of AQLs to provide guidance on the subjective topic of LiDAR data quality. The prevailing weather and atmospheric conditions, which are outside of the control of the contractor, are the most important factors that affect the data quality, and therefore the AQLs. The client and contractor should work closely together to ensure a mutually acceptable result.

  • Point density and point cloud accuracy specifications should be met.
  • Full coverage should be achieved.
  • There should be a good match between flight runs.
  • The LiDAR should only be flown in good conditions, in the absence of rain, cloud, atmospheric haze, snow and flooding.
  • The LiDAR may be flown at any time when the weather conditions are suitable to achieve the specified standards of data quality, except where special time constraints are defined.

The intended use of the LiDAR may impose limitations upon times of flying. See project constraints in section 3.3.

5.3 LiDAR accuracy and resolution table

Preparation of Table 6 is based on the American Society for Photogrammetry and Remote Sensing (ASPRS) document: ASPRS Positional Accuracy Standards for Digital Geospatial Data, edition 1, version 1.0., 2014, pg. A7.

The accuracy values in the table are dependent on:

  • the flying altitude and
  • the GNSS positional and IMU angular rotation errors of the equipment used.

Laser ranging and timing errors are also considered.6 Many other factors may also affect accuracy and resolution. Therefore, the values quoted can only be referenced as achievable.

Platform

Height AGL

Achievable accuracy RMSE (m)

Achievable resolution (ppm2)

m

Ft

Plan X,Y

Height Z

UAV

30

100

+/- 0.019

+/- 0.012

140

UAV

122

400

+/- 0.06

+/- 0.05

51

Helicopter

260

853

+/- 0.03

+/- 0.03

100

Helicopter

400

1,312

+/- 0.04

+/- 0.03

48

Fixed wing

500

1,640

+/- 0.04

+/- 0.03

30

Fixed wing

725

2,379

+/- 0.06

+/- 0.04

20

Fixed wing

1,300

4,265

+/- 0.10

+/- 0.05

8

Fixed wing

2,600

8,530

+/- 0.20

+/- 0.10

2

Fixed wing

5,000

16,404

+/- 0.39

+/- 0.15

1

Table 6: Achievable accuracy and resolution values for LiDAR sensors

The UAV flying altitude of 400ft represents the highest altitude at which a UAV can be operated in the UK without the approval of an operational safety case.

For comparison purposes, the LiDAR FOV was maintained at 60, keeping a high degree of coverage. LiDAR resolutions were calculated from first principles. The laser PRF values vary with altitude as lasers with a high PRF capability currently tend to be larger and heavier and can therefore only be carried by fixed wing and helicopter platforms. The differences in platform velocity at different altitudes were also considered when calculating the LiDAR resolution.

5.4 LiDAR deliverables and products

5.4.1 LiDAR point cloud

The basic deliverable is the LiDAR point cloud, which is made up of individual laser data points that are fully geo-referenced in 3D in the client's choice of coordinate system and usually cut into 1km squares.

LAS - or the compressed version, LAZ - is the most frequently used format for LiDAR data. It is an internationally used standard format, maintained by the ASPRS, which facilitates data classification and storage of metadata.7

The client may specify the data format, data compression and data transfer medium. As with aerial photography, LiDAR data at high point densities can require large volumes of space, so clients are likely to specify a format that can be easily incorporated into their archive system.

5.4.2 Metadata

Metadata may be specified for LiDAR or for any other LiDAR products as described in the following sections.

The LiDAR LAS format promotes the easy management and exchange of metadata. Examples within this format are:

  • date and time flown
  • GPS time
  • geographic reference
  • flying height
  • coordinate system
  • scan angle
  • number of laser returns and
  • laser intensity.

5.4.3 Digital terrain/surface models

The LiDAR point cloud can be processed to extract a ground class and a second class containing all points above the ground. The ground class can be extracted separately to create a DTM of the AOI. A DSM can be created by combining the ground class and the above ground class into a single file.

5.4.4 Classified point clouds

Classified point clouds further categorise the data into separate groups. The points above the ground, for example, can be classified into:

  • buildings
  • hard surfaces
  • lampposts
  • power supply lines and
  • areas of low, medium and high vegetation.

It is also common to remove temporary movable objects such as cars and classify these as noise. This enables further specialist analysis of the data. The LAS format sets out several standard classes within this specification.

Where an RGB camera has been flown simultaneously with a LiDAR instrument, it is possible to assign the RGB colour value from the camera to each individual LiDAR point in the point cloud. These colourised point clouds offer a more realistic representation of the AOI.

5.4.5 Mapping

Topographic mapping in 3D vector format can be extracted from LiDAR point clouds. The linework is digitised from the data in 2D, frequently with the use of a simultaneously captured imagery layer. Heights are then assigned to the 2D feature strings by draping the linework onto the 3D point cloud. This approach has been successful for high accuracy engineering applications.