Silvilaser 2019 - Poster Presentations »
Using terrestrial and airborne laser scanning in the estimation of wood properties in standing timber
Changes in climate and silviculture are altering forest growth and wood formation globally, including changes in wood properties, such as wood density. These developments may affect the carbon sequestration in forests, as well as limit timber usage in long-term carbon storage in, e.g. construction. Because wood properties result from tree’s adaptation to the environment through tree crown, the crown and branching attributes could be used to model the underlying wood formation processes and wood properties. The emergence of three-dimensional terrestrial point clouds, such as those from terrestrial laser scanning (TLS), has made it possible to obtain detailed crown and branching data from standing timber. Moreover, increasingly dense airborne laser scanning (ALS) point clouds enable the delineation of individual tree crowns and the extraction of crown features that are suited for the generalization of detailed branching structures over larger spatial extent. Combining high-resolution remote sensing data sets could thus improve the conditions for forest managers and decision makers to account for the variability of wood formation and wood properties across variable scales. Here, we demonstrated such an approach with Scots pine (Pinus sylvestris L.) stands from various development stages in a landscape of boreal forests in Southern Finland. We recorded a sample of ca. 300 Scots pines across 27 stands with TLS, and obtained dense (~six pulses/m2) ALS data over the entire area of 2000 hectares. We reconstructed the branching structures of the TLS sample trees, and built Random Forest - models of select branching attributes with respect to the geometrical ALS features of respective tree crowns. We then used our data to predict branching structures for each tree over three Scots pine - stands within the area (not included in the other 27). We will evaluate the accuracy of our predictions by comparing the predicted branching variables against X-ray measurements of ca. 2000 trees harvested from the studied stands. By doing so, we aim at assessing the feasibilities of using a combination of TLS and ALS data to model the branching structures in such detail that the variabilities of various crown-dependent wood properties (e.g. wood density) could be estimated tree-specifically over a larger region. We will discuss the benefits, possibilities and challenges of including wood properties in remote sensing-based forest inventory data, from the point-of-view of informing the decisions regarding the management and use of forest resources.