Using 3-dimensional point clouds to improve characterizations of tree stems across scales in boreal mixedwood forest stands
Christopher Mulverhill, Nicholas C. Coops, Piotr Tompalski, Joanne C. White, Peter L. Marshall, Todd Bailey
The extent and complexity of boreal forests can make undertaking inventories difficult – particularly across the diverse species, age gradients and disturbance regimes that exist in mixedwood stands, one of the most common ecosystem types in Canada’s boreal forest (Drapeau et al. 2000). Individual tree dimensions are critical components of inventories which provide a basis for the remainder of a forest inventory and inform wood quality parameters. Tree location or detection, measurements of the diameter at breast height (DBH), height, taper and branching structure are useful in determining tree characteristics. Attributes such as the location, diameter at breast height (DBH), and height of a tree, are fundamental to a variety of other measurements, such as tree volume or by group measures such as a stem size distribution (SSD), which represents the relative frequency of tree sizes in a given area (Taubert et al. 2013). SSDs can contribute to the assessment of a tree’s merchantability or utilization for harvest (Landsberg et al. 2005).
Complete characterization of a forest requires the estimation or measurement of trees in all stands in the area of interest. The advancement of remote sensing technologies and development of associated methodologies has made it easier to survey forest attributes quickly and effectively. Broad-scale estimation requires operation from the air, while detailed plot- and tree-level estimates typically come from ground-based sensors. The most common sensor for large scale forest inventories due to its ability to accurately and continuously characterize large areas is Airborne Laser Scanning, or ALS (Hudak et al. 2009). ALS point clouds are used to generate descriptive metrics characterizing height, volume, and biomass on larger scales (Næsset 2002), as well as finer scale descriptions such as crown dimensions, and vertical and horizontal canopy structure (Coops et al. 2007). A developing technology applied to finer scales of inventory is digital terrestrial photogrammetry (DTP), which is an inexpensive alternative to ground-based laser scanning. DTP uses images to create a detailed point cloud which can be used to characterize stem attributes and used as inputs for harvesting and wood quality assessment. Combined, ALS and DTP can be used to characterize stems at multiple scales, from measuring their individual characteristics to assessing patterns of their distribution across the landscape. This research describes new methods for estimating individual stem attributes and modifies existing techniques to assess tree attributes across scales in a 700,000-ha boreal mixedwood forest near Slave Lake, Alberta.
First, at stand scale, this research uses ALS to characterize SSD as simple (unimodal) and complex (bimodal) structural types. Various stand characteristics, measured on plots and predicted with ALS, were assessed for their ability to identify the plots as either unimodal or bimodal. Once the best ALS-based metric for identification was determined, it was used to categorize plots for estimation of SSD parameters by ALS – a Weibull model for unimodal stands and a Finite Mixture Model (FMM) for bimodal stands. Field-based SSD on classified plots were used as response data for prediction with a suite of ALS metrics.
Figure 1. Workflow of analysis applied to estimating mixedwood SSDs
Next, on the tree scale, the accuracy of DTP point clouds in characterizing stem measurements, such as DBH, location, and upper stem diameter was evaluated. This research uses two RICOH THETA S cameras, each capable of capturing 360˚ images (Ricoh Company Ltd. 2017) mounted on a telescoping pole capable of taking images from 1.3m to 5m high. DTP-based estimates of DBH and upper stem diameter are compared to ground measurements and overall accuracies were related to acquisition conditions such as light intensity, stem density, and understory to determine how the accuracies of point clouds were influenced by these conditions. Preliminary results indicate the success of DTP for modelling individual stem attributes and suggest that accurate estimates of DBH, height, and taper can be made from DTP point clouds in mixedwood forest stands.
Figure 2. Example of tree stem derived from DTP
The difference in structures and stem forms among stands requires detailed, landscape-level information to guide the fitting and modeling process. At larger scales differentiation by ALS allowed structurally appropriate SSDs to be fit to respective stands and allowed for more robust characterizations of SSD than using a single model for the entire study area. For forest professionals who rely on both detailed tree-level measurements and stand-level information, being able to quickly and effectively characterize the forest at both scales provides crucial insights that lead to more informed localized management decisions and more accurate ecological understanding of stem attributes and stand structure in complex habitats.
Coops, N.C., Hilker, T., Wulder, M.A., St-Onge, B., Newnham, G., Siggins, A. and Trofymow, J.T., 2007. Estimating canopy structure of Douglas-fir forest stands from discrete-return LiDAR. Trees, 21(3), p.295.
Drapeau, P., Leduc, A., Giroux, J.F., Savard, J.P.L., Bergeron, Y. and Vickery, W.L., 2000. Landscape‐scale disturbances and changes in bird communities of boreal mixed‐wood forests. Ecological Monographs, 70(3), pp.423-444.
Hudak, A.T., Evans, J.S. and Stuart Smith, A.M., 2009. LiDAR utility for natural resource managers. Remote Sensing, 1(4), pp.934-951.
Landsberg, J., Mäkelä, A., Sievänen, R. and Kukkola, M., 2005. Analysis of biomass accumulation and stem size distributions over long periods in managed stands of Pinus sylvestris in Finland using the 3-PG model. Tree physiology, 25(7), pp.781-792.
Næsset, E., 2002. Predicting forest stand characteristics with airborne scanning laser using a practical two-stage procedure and field data. Remote sensing of environment, 80(1), pp.88-99.
Ricoh Company Ltd (2017). RICOH THETA S.
Taubert, F., Hartig, F., Dobner, H.J. and Huth, A., 2013. On the challenge of fitting tree size distributions in ecology. PloS one, 8(2), p.e58036.