The application of ALS in forest system investigations largely depends on the quality and quantity of field reference data, especially with the currently applied ALS-based inventory technique, which is an area-based inventory where field reference determines the output of each raster cell based on non-parametric estimates [5].Reference data for forest inventories are conventionally collected by utilizing expensive and labor-intensive manual measurements [6,7]. Both destructive and non-destructive methods can be applied in field data collection. Destructive measurements are typically accurate but highly expensive and are not applicable or acceptable in many cases, such as in urban forests and in conservation areas. Non-destructive measurements are typically carried out by utilizing simple tools, such as calipers and measuring tape.

Tree attributes that can be collected are limited to those measurable at reasonable cost and accuracy.Important tree attributes needed in forest mapping and ecosystem services include the diameter at breast height (DBH), location, tree height, tree species, leaf area index (LAI), the height of the first living branch, stem curve, stem volume and biomass. Many attributes, such as the stem curve, are not practically measurable using simple tools. Allometric models can be used to determine attributes that are not directly measurable but can be derived from other basic measurements. The applicability of allometric functions, however, is limited to particular climatic, geographic and silvicultural conditions.

The use of inappropriate allometric models can lead to large estimate errors, as was shown previously [8]. In general, more automated and cost-effective techniques are needed to provide tree attribute data for forest ecosystem services, especially for forest inventories.Recently, TLS has been proven to be a promising technique in forest field measurements. The first commercial TLS system was built by Cyra Technologies (acquired by Leica in 2001) in 1998, and the first papers related to plot-level tree attribute estimation were reported in early 2000s [9�C15]. Currently, TLS has shown to be feasible for collecting basic tree attributes at plot-level, such as DBH and tree position [16�C21]. By reconstructing tree stem, it is possible to derive high-quality stem volume and biomass estimates comparable in accuracy with the best national allometric models [22,23].

New TLS systems are continuously being developed, with the size and weight of the laser scanners decreasing rapidly. Dual-wavelength TLS is currently being studied within forest ecology [24]. The Brefeldin_A use of dual wavelength laser pulses makes it easy to separate leaf returns from returns of stems, branches and the ground, which is convenient for forest ecosystem services.