skip to primary navigation skip to content
 

The use of airborne LiDAR to characterise Mediterranean forest communities

The use of airborne LiDAR to characterise Mediterranean forest communities

Mediterranean semi-natural forest communities are amongst the most transformed, fragmented and threatened of landscapes worldwide and yet are also some of the least protected with extent of habitat conversion exceeding that of habitat protection. For example, in southern Portugal, the traditional landscapes of complex and dynamic agro-forestry cork oak (Quercus suber) mosaics (Figure 1, below) are threatened by fragmentation as a result of afforestation with fast growing timber-producing eucalyptus and pine plantations (Figures 2, 3 and 4). The need for the conservation of this habitat is increasingly recognised but this requires detailed habitat surveying and mapping. However, the complex topography of southern Portugal (typical of many Mediterranean regions) can make access for such surveys challenging, time-consuming and potentially costly, especially if repeat surveys are needed for monitoring purposes.

This work is based on the use of airborne laser scanning (LiDAR: Light Detection and Ranging) to characterise the 3-dimensional structure of the forest communities of southern Portugal, based on a field area in the Serra region of the western Algarve. LiDAR data were acquired in 2006 by the UK NERC Airborne Remote Sensing Facility and complemented by ground surveys in 2006, 2007 and 2009. The LiDAR data provide high spatial resolution with an average point density of 0.5 m-2, with height accuracy to about 0.1 m. Separate first and last returns are associated with each emitted laser pulse. Depending on the characteristics of the vegetation/terrain surface, some of the emitted energy is reflected by the first intercepting surface, whilst a proportion may penetrate that surface to be returned from a lower surface which, in the case of a forest stand, might be vegetation within or below the canopy surface or the ground. Using the last returns a Digital Terrain Model can be created, while the first returns can be used to establish a Canopy Surface Model. From these, the height of vegetation can be determined. LiDAR-derived structural attributes (Figure 5) and vegetation reflectance (Figure 6), as measured by the return intensity of the laser pulses, can then be used to discriminate between the different forest communities.

Linear discriminant analysis shows that the standard deviation of the intensity of canopy hits, the mean intensity of the vegetation first-only hits and the standard deviation of canopy height are the most important discriminating variables for identification of the dominant tree species in each forest stand. The deciduous cork oaks have a higher intensity than the coniferous pine species, while the intensities from eucalyptus vary according to stand age, which influences the degree of canopy openness and therefore the likely penetration of laser energy into the sub-canopy and gaps between trees. Canopy openness is also measured by the standard deviation of canopy height and the height of vegetation singular returns – those returns which derive from dense vegetation through which there is relatively little canopy penetration (see Maritime pine in Figure 6).

The ability of LiDAR to differentiate between forests stands of different species and to parameterise forest structure (e.g. canopy closure/openness, presence/absence/volume of understorey vegetation) means that it can be used to create habitat maps (for example the extract in figure 7), to measure above ground biomass and, in fire-prone ecosystems, to assess the understorey accumulation of inflammable woody material that potentially increases the incidence of fire.

index00.jpg
Figure 1 – Cork oak grove
index01.jpg
Figure 2 Eucalyptus plantation
index02.jpg
Figure 3 Maritime pine plantation
index03.jpg
Figure 4 Umbrella pine plantation
index04.jpg index05.jpg
index06.jpg index07.jpg

Figure 5 Typical examples of canopy height and return type by species


index08.jpg index09.jpg
index10.jpg index11.jpg

Figure 6 Typical examples of canopy intensity return type and height by species: red denotes vegetation first-only hits; blue denotes vegetation singular hits

index12.jpg

Figure 7 Extract from one LiDAR strip to the east of the Bravura reservoir. Colour variation denotes standard deviation of intensity: darker colours are plantations of eucalyptus and pine.