Characterizing Spatial Pattern and Heterogeneity of Pine Forests in North Carolina’s Coastal Plain using LiDAR

Abstract

Remote sensing tools that directly characterize canopy structure would be beneficial for management activities and conservation planning. LiDAR (Light Detection And Ranging) is such a tool, as an active remote sensing technology that provides fine-grained information about the three-dimensional structure of ecosystems across a broad spatial extent. This project assesses the feasibility of using the state-wide North Carolina Floodplain Mapping LiDAR dataset to differentiate between the structural components of evergreen forest types in North Carolina’s coastal plain. Vertical structure and spatial patterns of vertical structure were quantified using geospatial measures such as semivariogram/ correlograms, lacunarity analysis, and correlation length. LiDAR-derived metrics were also created for comparison with standard field-based measurements of stand structure. I found that LiDAR is capable of measuring canopy variation and can differentiate between the structural characteristics of evergreen forest types. Also, the N.C. LiDAR has potential for use as a surrogate for field measurements when collection is not feasible due to time, labor, or financial constraints. In addition, the project examined LiDAR’s use as a screening tool in the identification of suitable habitat for the federally endangered red-cockaded woodpecker (Picoides borealis). I used Maximum Entropy (Maxent), an inductive modeling algorithm for presence data only, to create a spatial species distribution model using LiDAR-derived variables in addition to more typical geospatial variables. The Area (AUC) under the Receiver Operating Characteristic (ROC) curve was analyzed for increases in predictive power with additions of variables. Results suggested that the addition of LiDAR-derived variables to habitat models improved their predictive power, resulting in a test AUC increase from 0.923 with standard spatial variables only, to a test AUC of 0.951 with LiDAR-derived variables added. The success of this project has important implications for natural resource management and conservation planning, especially given that the LiDAR dataset is publicly available and covers the entire state of North Carolina.