LIDAR Method. Comprehensive Model Selection, models were built and ranked by their back-transformed R2 values. However, unlike the methodology in Section 3.1.2, Intensity, and L-Moment metrics were not excluded from the comprehensive model selection process. Further review of the models with the highest R2 values resulted in selecting the model below.
LIDAR Method. The method used to build this linear regression model are described in further detail in Section 3.1, the LIDAR Method for Canopy Height.
LIDAR Method. 15 2.1.1 Digital Elevation Model Resolution 15
LIDAR Method. 72
10.1.1 Limiting Factors 72 10.2 Imagery Method 72 10.2.1 Accuracy Assessment 72 10.2.2 Limiting Factors 72 10.3 Recommendations 72
LIDAR Method. The development of channel locations from LIDAR is a standardized process, but involves making choices, all of which impact the final outcome. The general approach involves the following steps: develop a digital elevation model (DEM), perform a flow accumulation on the DEM, set a flow accumulation threshold to determine the perennial initiation point, and convert the result to a vector GIS dataset. Details of the specific processing performed for this project are available in Appendix A.
LIDAR Method. The method used to build these linear regression models is described in further detail in Section 3, the LIDAR Method for Canopy Height. Two models were developed to estimate crown diameter from LIDAR. The first used only metrics derived directly from the LIDAR data itself. The second could include radius or diameter values calculated from the individual tree objects (ITOs) created during the segmentation of the 6 ft. resolution canopy height model (CHM). The process of segmenting the canopy model into individual tree objects (i.e. portions of LiDAR point cloud assumed to represent individual trees) is described further in Appendix D. Two models were developed because it was believed that ITOs provide additional information about the trees on the plot, and could potentially improve the accuracy of the crown diameter model. However, segmenting a canopy height model and measuring diameters and radii of the resulting ITOs, is time consuming. The additional processing time, may outweigh the value of any additional accuracy. In Table 5 below, we describe the different crown size metrics that were calculated for each tree object. These were averaged for the trees objects on each plot. The plot averages were included as possible metrics in the regression models. The center of the tree, the high point, is the center of the cell with the highest height in the tree object, and can be considered the stem location. Each tree object has 16 vertices in the cardinal directions, at the cell centers nearest the edges of the crown. Distances from the high point to each vertex were calculated and used for the crown size metrics. Table 5. Diameter and radius metrics calculated for each tree object. Maximum Radius The longest radius from the high point (the furthest vertex). Minimum Radius The shortest radius from the high point (the closest vertex). Longest and Perpendicular Diameter The diameter based on averaging the longest transect with the transect perpendicular to the longest transect. NS/EW Average Diameter The diameter based on averaging the North/South transect and the East/West transect lengths; these may not be the longest transects. Average Radii Diameter The diameter based on averaging the lengths of all the radii from the high point to each vertex, and doubling the average. Crown Area Diameter The diameter based on treating the polygon as a circle and back- calculating the diameter from the circle’s area.
LIDAR Method. The methods described here were also used for Conifer/Deciduous Classification and Large Woody Debris. The field crew collecting plot data recorded whether each tree measured was alive or dead, allowing for plot level counts of snags. There were plots with no snags present, creating a non-normal distribution of counts shown in Figure 27 below.
LIDAR Method. Three methods were used to build stand density models from LIDAR. The first approach built a linear regression model based on individual tree objects (ITOs) from a segmented canopy height model. The second approach built a linear regression model using the method described in further detail in Section 3.1, the LIDAR Method for Canopy Height. The third method included the stratified bin of each plot, which indicates information about the height and cover values of each plot.
LIDAR Method. LIDAR was used to estimate the number of deciduous trees, following methods similar to the one described in Section 5.1 Snag Detection, LIDAR Method.
LIDAR Method. Vegetation class is a description of how a stand is progressing through a development process and how the individual trees are interacting with each other and their specific, local environment. There are many different schemes used for classifying vegetation into development stages, with possibly one, Xxxxxx and Xxxxxx (Xxxxxx & Xxxxxx, 1990), establishing the most commonly used terminology. As time has passed, schemes for classifying vegetation seem to have become more complicated, increasing the number of classes from the four described in Xxxxxx and Xxxxxx, to as many as six or eight. LIDAR is a tool that can help describe stand structure. Any method to estimate vegetation class from LIDAR must use the structure as a proxy for development progress. To this end, only classes that can be distinguished from one another structurally can be identified using LIDAR. This argues for a simple classification scheme. We chose a four class scheme based on Xxxxxx and Xxxxxx, but because the names of vegetation classes reflect stages of a biological process, and because LIDAR can only tell you structural characteristics, we propose a different naming scheme for these vegetation classes.
