Silvilaser 2019 - Poster Presentations »
A Comparison of airborne ALS, photogrammetrically-derived, and IfSAR data for forest Inventory in Western Washington State, USA
Airborne Laser Scanning (ALS) technologies have enabled and continue to refine our ability to quantify forest structure across landscapes. They are generally regarded as the most accurate and precise source of digital terrain information. In many parts of the world ALS is relied upon as the primary source for forest inventory information. ALS point cloud densities continue to increase leading to a more detailed description of forest resources while acquisition costs remain stable or decrease. However, the overall cost of ALS remains substantial, and repeat acquisitions over large areas are rare. Alternative information sources from which to derive structure information, including airborne photogrammetrically-derived point clouds and radar data products, have been suggested as potentially more cost effective for some applications. Recently, special attention has been given to point clouds generated from imagery acquired using pushbroom instruments operated from higher altitudes. Wall-to-wall pushbroom imagery is acquired on a three-year cycle across the conterminous Unites States and thus offers the potential to provide information useful for detecting and quantifying changes in forest conditions. We performed a rigorous comparison of information derived from high-density ALS, frame and pushbroom imagery, and radar against traditional forest inventory field measurements. The study was conducted in the Hood Canal area in the State of Washington, United States. Hood Canal comprises a variety of physiographic and forest structure conditions. Because photogrammetry-, and to a certain extent radar-derived information is often limited to the upper parts of tree canopies, we used the ALS-derived terrain surface to assess vegetation height. We found that the utility of photogrammetry and radar data products changes substantially with the scale of the analysis. For forest inventory parameters highly correlated with tree height, such as tree volume and biomass, the efficiency gains, defined as reduction in estimate uncertainty, realized by using wall-to-wall remotely sensed information ranged from a factor of four for LiDAR to two and a half for pushbroom imagery data. In general, the utility of radar data was lower, especially over steep slopes. The value of photogrammetrically-derived data and radar in support of tactical decision making and operations was less clear, owing primarily to tree omission errors in areas with low canopy cover or clumped forest structure.