**4.1 Characteristics**

The small and portable terrestrial lidar systems can be mounted on a static tripod or transported on a moving vehicle and therefore can be easily taken into the field. GPS measurements also allow these scans to be geolocated. In contrast to the viewing perspective from above provided by satellite and airborne sensors, terrestrial lidar provides a clear view of the tree stem, understorey and ground surface (Figure 7).

This measurement of a relatively small area within viewing distance of the scanner can be considered to replicate field plot measurements, however additionally provides an understanding of context which would not be possible from field data. The upward looking approach often leads to difficulty in detecting tree tops, however representation of tree stems, ground surface roughness and understorey vegetation offer a level of detail which cannot be retrieved using airborne instruments.

This approach causes only the side of stems facing the scanner to be detected in any one scan and also obscures the view of trees which are behind those closer to the instrument. Therefore

Lidar Remote Sensing for Biomass Assessment 17

Aside from forestry management objectives, the use of terrestrial lidar can potentially complement traditional field data collection by improving the efficiency and accuracy of survey approaches. Whereas the time required to measure the trees within an inventory plot by a team of surveyors may be quite considerable, the combination of lidar scans of a few minutes each could substantially reduce this whilst providing additional contextual

Terrestrial laser scanning measurements are restricted to small area sampling, similar to typical field data collection. However, this permits the plot to be 'revisited' visually and analytically for multiple purposes without returning to the field, allowing the scene to be reconstructed to enable trees to be placed in the context of their immediate surroundings. For management purposes, this could be invaluable to determine optimum thinning or harvesting times or to assess growth trends against model predictions through the

Once diameter distributions within measured plots are calculated, in the same way as with field approaches, allometric relationships allow wider stand-level forest attributes to be inferred. If applicable species groupings are known, general DBH-based regression equations can be applied to estimate forest stand-level biomass. For sites within the USA, Jenkins *et al.*, 2003, 2004 present diameter-driven allometric equations for biomass of North American species whilst for Europe, a similar resource is provided by Zianis *et al.*, 2005.

Forest growth models such as the Ecosystem Dynamics model, ED, (Hurtt *et al.*, 2004) and the Tree and Stand Simulator, TASS, (Goudie & Stearns-Smith, 2007) enable forest growth scenarios to be predicted. Remote sensing analysis can be used as a valuable tool to provide observational inputs to models and in order to produce detailed inventories for long-term

Airborne lidar stand level analysis can be used to produce statistically-derived model inputs. This approach is being undertaken as part of the NASA Carbon Monitoring System (NASA, 2010) using an interpolated surface 80th percentile canopy height model as an input to the ED model (e.g. Hurtt *et al.*, 2004). This method can potentially be applied across large areas and could be achieved with relatively low density lidar data such as might be acquired

Alternatively, Suárez, 2010, used a tree list generated from individual tree delineation as baseline inventory data from which to predict future scenarios and demonstrate processes at work within stands using the distance-dependent model TASS. These processes include competition, establishment of a dominance hierarchy and recovery from catastrophic events such as wind damage or thinning. This means that the biological principles behind such models adapt them to local conditions, unlike empirical models and suggesting wider

The temporal dimension provided by the TASS simulations provides a valuable insight into the long-term effects of each stand intervention or natural disturbance. Not only growth increments, but timber products can also be predicted with this method. In addition, management practices can be balanced by the constraints introduced by the future risk of wind damage. The scale of analysis and the possibility of creating future scenarios contribute to a substantial reduction in the level of uncertainty associated with forest management.

information which could not be achieved with field measurements.

measurement of diameter increments.

**5.1 Forest growth models** 

for regional or national campaigns.

application may be possible.

scenario modelling.

**5. Further applications of lidar remote sensing** 

the combination of several scans is often required to accumulate adequate information for analysis of the scene within a plot. Low vegetation and heavy branching can affect the quality of the data and in some studies were removed before the site was scanned. This difficulty is especially important with the trees closest to the scanner as they cause the most occlusion.

Research into the detection and reconstruction of stems (e.g. Huang *et al.*, 2011) and branches (e.g. Bucksch & Fleck, 2011) is a current area of development which is of great interest. For forestry purposes this would permit the quality of timber to be determined more easily, such as stem straightness, branch number and branch angle. The ability to reconstruct the tree geometry from terrestrial laser scanning is unprecedented. Whilst for the purposes of forest inventory, the ability to detect the ground surface, height above the ground along the tree stem and to determine the size of the stem allows diameter at breast height to be directly measured, which is one of the most fundamental operational parameters collected by foresters in the field.

Fig. 7. Below-canopy plot sampling using terrestrial laser scanning. Images courtesy of Forest Research, UK.
