Aerial photography has been the basis of resource assessments since the start of human flight, but more intensely since the Second World War. Space photography, on the other hand, has been in use since the 1970s. Orthocorrected photographs, i. e., where the effects of terrain or topographical variations have been removed, can be used as a planimetric source of information, as all 3D distortions are removed. It is the use of photogrammetry that is of particular interest in biomass estimates. Not only can areas of the resource be mapped, but a height value can be estimated for standing biomass via stereo-photogrammetry.

Stereo-photogrammetry provides a measure of parallax, which can be interpreted as the positional difference between the base and the top of an object, due to changes in viewing geometry. By collecting points on the ground surface (terrain) and the

top of all objects on the terrain (surface) and performing an interpolation on these points, digital terrain models (DTM) and digital surface models (DSM) are products of the digital photogrammetry process.

Subtraction of a DTM from a DSM results in the height of an object above the terrain at each chosen point, as shown in Fig. 2.5. The digital orthophoto, which is also a product of the digital photogrammetry process, also provides a basis of measuring crown diameters. Although trained analysts can estimate individual tree sizes, which are useful for landscape-level analyses, this process must be automated to increase analyst efficiency. Automatic extraction of crown area, DBH, and biomass at the tree level is a relatively new type of image-analysis capability and associated techniques, which rely on delineating a tree crown from background soil and vegetation, and adjacent trees. A GIS layer of polygons, where the polygon area represents the crown area, can then be produced (Maltamo et al. 2003). Crown diameters are related to stem diameters for a given species and site, which means that both height and DBH could be estimated from the products of photogrammetry, with accuracy depending on the scale or resolution of the photography and the natural variability of the variables in the population.

Images that are suitable as input to the digital photogrammetry process range from satellite products to unmanned aerial vehicle (UAV) photography, which is becoming increasingly popular. For instance, individual tree canopies were mapped from 0.1 m aerial imagery to estimate carbon in a tropical forest in Belize, while canopy structure attributes can be estimated from images with resolution better than 4 m (Chambers et al. 2007; Greenberg et al. 2005). Figure 2.6 shows a graphical representation of the interaction between object size, in this case a tree canopy, and spatial resolution (pixel size) of some common satellite sensors. It is important to note, however, that (i) there is a trade-off between spatial resolution and temporal resolution (revisit time), where sensors with high spatial resolutions, e. g., IKONOS can take years to revisit the same spot on Earth; (ii) higher spatial resolutions generally imply lower spectral resolutions (broader bands); and (iii) inventory models that are developed at one spatial resolution, e. g., high spatial resolution IKONOS imagery, should ideally be coupled to and scaled with lower spatial resolution sensors, such as the Landsat suite of sensors. This approach effectively enables a tree-to-stand-to-landscape type inventory approach.