Using Lidar Data to Predict Photo Interpreted Attributes

Abstract

Large datasets and robust workflows exist for both the photo interpretation and Lidar methodologies. Both methodologies have unique and non-overlapping strengths in describing forest conditions. Bridging the data products of these methodologies would expand the capabilities for both approaches. To my knowledge no previous studies have attempted to evaluate the comparability of photo interpretation datasets and Lidar data products. In this study I attempted to develop methods that incorporate Lidar products into the photo interpretation process, and evaluate the comparability of Lidar products to photo interpretation attributes. I evaluated correlations between photo interpretation attributes and logical analog Lidar data products. I developed models to predict photo interpretation attributes using Lidar data products and predictor variables. I summarized photo interpretation attributes and equivalent Lidar predicted attributes using watershed scale spatial pattern metrics to evaluate the substitutability of the datasets in a mid-scale analysis scenario. Models of photo interpretation attributes describing Overstory Canopy Cover performed better than those describing characteristics of Understory Canopy Cover. Models performed poorly in exact matching of photo interpretation classification, but frequently predicted classes within one class of observed values for ordinal metrics. Comparisons of watershed scale summary metrics produced mixed results.