Forest canopy height retrieval using LiDAR data, medium-resolution satellite imagery and kNN estimation in Aberfoyle, Scotland

Abstract

The most common form of forest stand parameter estimation is undertaken based on field-based forest inventory data. Parameter estimations are implemented at a range of scales depending on the user needs and requirements. Due to time and cost constraints, these approaches are sample based, with measurements collected using sampling frames, which tend to be sparsely distributed when inventories are carried out across a country (e.g. national forest inventories) or region (Matthews and Mackie, 2006). Commonly, forest managers require detailed information on individual stands in order to plan silvicultural management strategies. Such information is typically derived from a stand-based inventory, where detailed measurements from several plots are collected within each stand, providing more precise forest parameter estimates. In recent decades, forest practitioners have increasingly integrated remote sensing data as a means of assisting and complementing forest inventories. Aerial photography has been used extensively to estimate forest attributes, which include species identification, tree height, crown diameter and the delineation of forest boundaries (Tuominen and Pekkarinen, 2005). Top height is an important forest parameter that is used in the UK and Ireland to estimate Yield Class (annual production potential) and as an indicator of standing biomass. Using remote sensing, tree height can be measured using high-resolution stereoscopic aerial photos, with photogrammetric techniques to measure the lengths of shadows projected onto level open ground (Kovats, 1997). However, these techniques are dependent on a number of factors that include level open ground and the need to determine the sun elevation and latitude, which consequently limits the operational use of these techniques. However, a more effective approach to determine tree heights in forests is by using parallax measurements from stereoscopic photographs (Schut and Van Wijk, 1965). More recently, research has been conducted that combines space-borne satellite imagery with forest inventory Forest canopy height retrieval using LiDAR data, medium-resolution satellite imagery and kNN estimation in Aberfoyle, Scotland