**1.2 Principles of lidar remote sensing**

When walking through a woodland on a sunny day, some of the sunlight reaches the ground through gaps between the foliage, woody branches and stems; some produces more diffuse light at the ground by transmittance though the foliage or reflection between different vegetation components and the ground, and some is absorbed by the intercepted surfaces. A proportion of the energy is reflected from these surfaces back towards the source. The same principles apply to lidar.

Lidar (Light Detection and Ranging) is an active remote sensing technology, which involves the emission of laser pulses from the instrument positioned on a platform, towards a target (e.g. woodland). Here, it interacts with the different surfaces it intercepts as outlined above (Figure 1). Features further from the sensor will intercept and reflect the laser energy back to the sensor later than those closer to it.

The area which is illuminated by the laser pulse is known as the lidar 'footprint'. The size of the footprint is determined by the laser divergence and the altitude/distance from the target of the lidar instrument. Whether the footprint is of large dimensions in the region of tens of metres from the altitude of a satellite sensor, tens of centimetres as generally produced from airborne platforms or several millimetres in the case of terrestrial laser scanners, the principles remain the same.

Interactions of the laser pulse with the vegetation depend on the wavelength of the emitted pulse and its reflectance, transmittance and absorption rates for each foliage, bark and background type (e.g. bare soil, litter, snow, etc). At wavelengths of 1064 nm (in the nearinfrared region of the spectrum and typical of many lidar systems used for vegetation analysis), reflectance and transmittance values may each be commonly ~45%.

The time for the reflected pulse echoes to be returned to the sensor is measured and, using the fact that the laser pulse travels at the speed of light, the total return distance travelled between the sensor and the intercepted surfaces can be calculated. The distance between the altimeter and the intercepted object is therefore half of this value (Baltsavias, 1999; Wehr *et al.*, 1999). This permits the three-dimensional reproduction of the Earth surface relief and above-surface object structures (e.g. vegetation, ice cover, atmospheric aerosols and cloud structure).

Very accurate timing is necessary to obtain fine vertical resolutions. Lidar time units are generally recorded in nanoseconds (ns), each being equal to approximately 15cm in one-way distance between the sensor and target. Time is measured by a time interval counter, initiated on emission of the pulse and triggered at a specific point on the leading edge of the returned pulse. This position is not immediately evident and therefore is set to occur where the signal voltage reaches a pre-determined threshold value. The steepness of the received pulse (rise time of the pulse) is a principal contributory factor to range accuracy and depends on the combination of numerous factors such as incident light wavelength, reflectivity of targets at that wavelength, spatial distribution of laser energy across the footprint and atmospheric attenuation (Baltsavias, 1999). The return pulse leading edge rise time is therefore formed by the strength of the return signal from the highest intercepted

Additionally, establishing allometric relationships between lidar and field measurements enables estimates to be extrapolated to stand, forest or national scales, which would not be feasible or very costly using field survey methods alone. Key aspects of biomass estimation

When walking through a woodland on a sunny day, some of the sunlight reaches the ground through gaps between the foliage, woody branches and stems; some produces more diffuse light at the ground by transmittance though the foliage or reflection between different vegetation components and the ground, and some is absorbed by the intercepted surfaces. A proportion of the energy is reflected from these surfaces back towards the

Lidar (Light Detection and Ranging) is an active remote sensing technology, which involves the emission of laser pulses from the instrument positioned on a platform, towards a target (e.g. woodland). Here, it interacts with the different surfaces it intercepts as outlined above (Figure 1). Features further from the sensor will intercept and reflect the laser energy back to

The area which is illuminated by the laser pulse is known as the lidar 'footprint'. The size of the footprint is determined by the laser divergence and the altitude/distance from the target of the lidar instrument. Whether the footprint is of large dimensions in the region of tens of metres from the altitude of a satellite sensor, tens of centimetres as generally produced from airborne platforms or several millimetres in the case of terrestrial laser scanners, the

Interactions of the laser pulse with the vegetation depend on the wavelength of the emitted pulse and its reflectance, transmittance and absorption rates for each foliage, bark and background type (e.g. bare soil, litter, snow, etc). At wavelengths of 1064 nm (in the nearinfrared region of the spectrum and typical of many lidar systems used for vegetation

The time for the reflected pulse echoes to be returned to the sensor is measured and, using the fact that the laser pulse travels at the speed of light, the total return distance travelled between the sensor and the intercepted surfaces can be calculated. The distance between the altimeter and the intercepted object is therefore half of this value (Baltsavias, 1999; Wehr *et al.*, 1999). This permits the three-dimensional reproduction of the Earth surface relief and above-surface object structures (e.g. vegetation, ice cover, atmospheric aerosols and cloud

Very accurate timing is necessary to obtain fine vertical resolutions. Lidar time units are generally recorded in nanoseconds (ns), each being equal to approximately 15cm in one-way distance between the sensor and target. Time is measured by a time interval counter, initiated on emission of the pulse and triggered at a specific point on the leading edge of the returned pulse. This position is not immediately evident and therefore is set to occur where the signal voltage reaches a pre-determined threshold value. The steepness of the received pulse (rise time of the pulse) is a principal contributory factor to range accuracy and depends on the combination of numerous factors such as incident light wavelength, reflectivity of targets at that wavelength, spatial distribution of laser energy across the footprint and atmospheric attenuation (Baltsavias, 1999). The return pulse leading edge rise time is therefore formed by the strength of the return signal from the highest intercepted

analysis), reflectance and transmittance values may each be commonly ~45%.

from satellite, airborne and terrestrial lidar systems are outlined below.

**1.2 Principles of lidar remote sensing** 

source. The same principles apply to lidar.

the sensor later than those closer to it.

principles remain the same.

structure).

surfaces within the footprint. This will vary with the nature of the surface; flat ice sheets producing abrupt returns with fast leading edge rises and multilayered, complex vegetation creating broad returns (Harding *et al.*, 1998; Ni-Meister *et al.*, 2001).

Fig. 1. Representation of the interception of foliage, bark or ground surfaces by an emitted laser pulse. At each surface, some energy is reflected, transmitted (in the case of foliage) or absorbed.

The location of every returned signal to a known coordinate system is achieved by precise kinematic positioning using differential GPS and orientation parameters by the Inertial Measurement Unit (IMU). The IMU captures orientation parameters of the instrument platform such as pitch, roll and yaw angles. Therefore, the GPS provides the coordinates of the laser source and the IMU indicates the direction of the pulse. With the ranging data accurately measured and time-tagged by the clock, the position of the returned signal can be calculated.

### **1.3 Full waveform and discrete return systems**

A waveform is the signal that is returned to the lidar sensor after having been scattered from surfaces that the laser pulse intercepts. Full waveform lidar systems record the entire returned signal within an elevation range window above a background energy noise threshold. An example of this from NASA's Geoscience Laser Altimeter System (GLAS; Section 2) is shown in Figure 2 (left). The scene shows a two-storey Douglas Fir canopy (*Pseudotsuga menziesii*) on a gentle slope of 4.9°. Typically, for vegetated surfaces on relatively flat ground, a bimodal waveform is produced.

Lidar Remote Sensing for Biomass Assessment 7

Distribution patterns of footprints differ between lidar systems. Lidar profiling involves the systematic location of footprints at intervals along the sensor's path on the ground. These may be contiguous such as the Portable Airborne Laser System of Nelson *et al.*, 2003 (PALS), or placed at discontinuous distances along the ground track in the case of NASA's Geoscience Laser Altimeter System, GLAS (Schutz *et al.*, 2005). This generally permits the sampling of extensive areas, however requires a means to extrapolate biophysical parameter

Laser scanning, obtained from an airborne platform, occurs perpendicular to the direction of travel and generally produces a dense distribution of small footprints. Swath width and footprint density are dependent on the altitude and speed of the aircraft plus the scan angle applied. A scanning mirror directs laser pulses back and forth across the flightline causing data to be captured typically in a sawtooth arrangement. The maximum off-nadir scan angle for the instrument can be customised according to the needs of each campaign. Narrower scan angles improve the chances of each shot penetrating dense vegetation canopies and of the sensor receiving a returned pulse from the ground as there is greater likelihood of a clear path through the canopy to the ground. The usual practice is to create an overlap of flightlines similar to photogrammetric surveys of ~60%. Multiple flightlines can then be combined to provide full coverage of the desired area by means of specialised software. Small footprint

laser scanning is generally operated at the forest scale, largely due to cost implications.

patterns on the ground, density and size of individual laser footprints.

As discussed above, lidar sensors can be operated at different scales from different altitudes and different viewing perspectives in relation to the target surface; from above in the case of satellite and airborne systems and from below or to the side for terrestrial laser scanners. Lidar instrument specifications therefore vary considerably, combining different sampling

Nelson *et al*'s (2008; 2003; 2004) PALS is an example of a small footprint, discrete return, lidar profiler which is operated from an aircraft. Its innovative and portable design has permitted sampling and vegetation parameter estimation at regional scales throughout the world and may be considered a predecessor to the satellite lidar profiling sensor discussed

The Laser Vegetation Imaging Sensor (LVIS) is an experimental lidar instrument developed at NASA Goddard Space Flight Center (GSFC, 2010). It is a full waveform, scanning lidar that emits a 1064 nm laser beam at a pulse repetition rate of 100-500Hz. LVIS can operate at an altitude in excess of 10 km and this offers the capability of producing swaths up to two km wide and medium-sized footprints of 1-80 m diameter (Blair *et al.*, 1999; Dubayah *et al.*,

Until relatively recently, small footprint lidar were almost exclusively restricted to discrete return systems within the commercial and operational sector whilst full waveform instruments remained a research and development tool. It should be noted that recent advances in data storage capacity are beginning to open opportunities for small footprint, full waveform scanning systems, however to date, software to process such data is not

By necessity, this chapter cannot attempt to fully present all combinations of lidar specifications. Readers should note that the multiple vegetation applications of lidar data lead

**1.4 Lidar footprint distribution patterns** 

estimates for areas where data were not acquired.

**1.5 Lidar system configurations** 

in Section 2.

2010; GSFC, 2010).

readily available.

The beginning and end of the waveform signal above the background noise threshold are represented by the upper and lower horizontal blue lines respectively (mean noise + 4.5σ in the case of GLAS). Amplitude of the waveform (x axis) represents both intercepted surface area at each elevation plus the reflectivity of the surfaces at the emitted wavelength (1064nm).

The gradient at the beginning of the signal increases slowly initially due to the relatively small surface area of foliage and branch elements at the uppermost canopy. As the energy penetrates down through the canopy, the waveform amplitude increases as more features are intercepted, before decreasing towards the base of the tree crowns. A small peak, which corresponds to a shorter tree, can be observed above the abrupt, narrow peak, which is returned from the ground. Below the ground surface, the signal can be seen to trail off gradually. This relates to both a gentle slope found at this site plus the effect of multiple scattering between features within the scene, which serves to delay part of the signal that is returned to the sensor.

Due to the complex waveform signal which is produced, this is often simplified using Gaussian decomposition of the waveform (Figure 2, left). Representing the waveform as the sum of the Gaussians, smoothes the signal yet allows a means of retaining and identifying the dominant characteristics of the signal for easier interpretation.

Small footprint lidar systems can produce dense sampling of the target surface. The returned signal is also in the form of a waveform, however with discrete return systems, only designated echoes within the waveform are recorded. These can be the first and last returns, or at times, also a number of intermediate points. The amalgamation of these returns from multiple emitted lidar pulses allows the scene to be reconstructed as a 'point cloud' of geolocated intercepted surfaces. This is seen in Figure 2, right, which illustrates the same location as seen within the GLAS waveform. The small footprint lidar point cloud can be interpreted more intuitively as a dominant upper storey of approximately uniform height and a single tree of lower height at the centre of the scene. Points are coloured with respect to their elevation.

Fig. 2. Example of a waveform produced by a large footprint lidar system (left) and a discrete return lidar point cloud (right) for a coincident area. Location: Forest of Dean, Gloucestershire, UK

The beginning and end of the waveform signal above the background noise threshold are represented by the upper and lower horizontal blue lines respectively (mean noise + 4.5σ in the case of GLAS). Amplitude of the waveform (x axis) represents both intercepted surface area at each elevation plus the reflectivity of the surfaces at the emitted wavelength

The gradient at the beginning of the signal increases slowly initially due to the relatively small surface area of foliage and branch elements at the uppermost canopy. As the energy penetrates down through the canopy, the waveform amplitude increases as more features are intercepted, before decreasing towards the base of the tree crowns. A small peak, which corresponds to a shorter tree, can be observed above the abrupt, narrow peak, which is returned from the ground. Below the ground surface, the signal can be seen to trail off gradually. This relates to both a gentle slope found at this site plus the effect of multiple scattering between features within the scene, which serves to delay part of the signal that is

Due to the complex waveform signal which is produced, this is often simplified using Gaussian decomposition of the waveform (Figure 2, left). Representing the waveform as the sum of the Gaussians, smoothes the signal yet allows a means of retaining and identifying

Small footprint lidar systems can produce dense sampling of the target surface. The returned signal is also in the form of a waveform, however with discrete return systems, only designated echoes within the waveform are recorded. These can be the first and last returns, or at times, also a number of intermediate points. The amalgamation of these returns from multiple emitted lidar pulses allows the scene to be reconstructed as a 'point cloud' of geolocated intercepted surfaces. This is seen in Figure 2, right, which illustrates the same location as seen within the GLAS waveform. The small footprint lidar point cloud can be interpreted more intuitively as a dominant upper storey of approximately uniform height and a single tree of lower height at the centre of the scene. Points are coloured with

Fig. 2. Example of a waveform produced by a large footprint lidar system (left) and a discrete return lidar point cloud (right) for a coincident area. Location: Forest of Dean,

the dominant characteristics of the signal for easier interpretation.

(1064nm).

returned to the sensor.

respect to their elevation.

Gloucestershire, UK

#### **1.4 Lidar footprint distribution patterns**

Distribution patterns of footprints differ between lidar systems. Lidar profiling involves the systematic location of footprints at intervals along the sensor's path on the ground. These may be contiguous such as the Portable Airborne Laser System of Nelson *et al.*, 2003 (PALS), or placed at discontinuous distances along the ground track in the case of NASA's Geoscience Laser Altimeter System, GLAS (Schutz *et al.*, 2005). This generally permits the sampling of extensive areas, however requires a means to extrapolate biophysical parameter estimates for areas where data were not acquired.

Laser scanning, obtained from an airborne platform, occurs perpendicular to the direction of travel and generally produces a dense distribution of small footprints. Swath width and footprint density are dependent on the altitude and speed of the aircraft plus the scan angle applied. A scanning mirror directs laser pulses back and forth across the flightline causing data to be captured typically in a sawtooth arrangement. The maximum off-nadir scan angle for the instrument can be customised according to the needs of each campaign. Narrower scan angles improve the chances of each shot penetrating dense vegetation canopies and of the sensor receiving a returned pulse from the ground as there is greater likelihood of a clear path through the canopy to the ground. The usual practice is to create an overlap of flightlines similar to photogrammetric surveys of ~60%. Multiple flightlines can then be combined to provide full coverage of the desired area by means of specialised software. Small footprint laser scanning is generally operated at the forest scale, largely due to cost implications.

#### **1.5 Lidar system configurations**

As discussed above, lidar sensors can be operated at different scales from different altitudes and different viewing perspectives in relation to the target surface; from above in the case of satellite and airborne systems and from below or to the side for terrestrial laser scanners. Lidar instrument specifications therefore vary considerably, combining different sampling patterns on the ground, density and size of individual laser footprints.

Nelson *et al*'s (2008; 2003; 2004) PALS is an example of a small footprint, discrete return, lidar profiler which is operated from an aircraft. Its innovative and portable design has permitted sampling and vegetation parameter estimation at regional scales throughout the world and may be considered a predecessor to the satellite lidar profiling sensor discussed in Section 2.

The Laser Vegetation Imaging Sensor (LVIS) is an experimental lidar instrument developed at NASA Goddard Space Flight Center (GSFC, 2010). It is a full waveform, scanning lidar that emits a 1064 nm laser beam at a pulse repetition rate of 100-500Hz. LVIS can operate at an altitude in excess of 10 km and this offers the capability of producing swaths up to two km wide and medium-sized footprints of 1-80 m diameter (Blair *et al.*, 1999; Dubayah *et al.*, 2010; GSFC, 2010).

Until relatively recently, small footprint lidar were almost exclusively restricted to discrete return systems within the commercial and operational sector whilst full waveform instruments remained a research and development tool. It should be noted that recent advances in data storage capacity are beginning to open opportunities for small footprint, full waveform scanning systems, however to date, software to process such data is not readily available.

By necessity, this chapter cannot attempt to fully present all combinations of lidar specifications. Readers should note that the multiple vegetation applications of lidar data lead

Lidar Remote Sensing for Biomass Assessment 9

Fig. 3. Multiple GLAS laser campaigns sampling overlaid on a GoogleEarth image of Tambopata, Peru. Missing data are found where dense cloud prevents sufficient energy

Within each footprint, if the top of the canopy is assumed to be the start of the waveform signal (upper horizontal blue line, figure 2 left), the accuracy with which the signal returned from the vegetation can be identified depends on the ability to identify a representative ground surface within the waveform. Methods to achieve this have included the use of an independent DTM to account for terrain slope within lidar footprints (Lefsky *et al.*, 2005; Rosette *et al.*, 2008) or using Gaussian decomposition of the waveform to locate a peak corresponding to the ground surface (Rosette *et al.*, 2008; Sun *et al.*, 2008a; Sun *et al.*, 2008b). Vegetation height can therefore be estimated as the elevation difference between the start of the waveform signal and the identified ground surface estimated within the waveform. The

Most commonly, waveform indices of Height of Median Energy (*HOME*) or relative height percentiles above ground (*RHi*) are calculated using the cumulative energy distribution within this region of the waveform returned from vegetation. More recently, Lefsky *et al.*, 2007, devised an alternative method of estimating a vegetation height parameter which accounts for terrain and vegetation roughness using the waveform leading and trailing

The sampling measurements produced within the satellite lidar footprints are typically combined with coincident field measurements of biomass. This enables regression equations to be developed using waveform metrics to estimate biomass for the areas

from penetrating to the Earth's surface and returning to the sensor.

edges rather than isolating the signal returned from vegetation.

studies above report RMSE as 3+ metres.

**2.2 Applications for biomass estimation** 

to wide-ranging variations in sensor design and characteristics as outlined above. In the descriptions within sections 2-4, an example of a satellite sensor is used to demonstrate the principles of large footprint, full waveform profiling data, whilst airborne and terrestrial lidar instruments are used to provide examples of small footprint, discrete return laser scanning.
