Application of Low-Cost Digital Elevation Models to Detect Change in Forest Carbon Sequestration Projects

MacDicken, Kenneth Glenn

Winrock International Institute For Agricultural Development

关键词: Biomass;    Classification;    Forests;    Simulation;    Trees;   

DOI  :  10.2172/922194
RP-ID  :  None
RP-ID  :  FC26-05NT42437
RP-ID  :  922194

美国|英语

来源: UNT Digital Library

【 摘 要 】

This two-year study evaluated advanced multispectral digital imagery applications for assessment of forest carbon stock change. A series of bench and field studies in North Carolina and Ohio tested aerial assessments of forest change between two time periods using two software packages (ERDAS and TERREST) for Digital Elevation Model (DEM) creation, automated classification software (eCognition) for canopy segmentation and a multiple ranging laser designed to improve quality of elevation data. Results of the DEM software comparison showed that while TERREST has the potential to produce much higher resolution DEM than ERDAS, it is unable to resolve crucial canopy features adequately. Lab tests demonstrated that additional laser data improves image registration and Z-axis DEM quality. Data collected in the field revealed difficult challenges in correctly modeling the location of laser strike and subsequently determining elevations in both software packages. Automated software segmentation of tree canopies provided stem diameter and biomass carbon estimates that were within 3% of comparable ground based estimates in the Ohio site and produced similar biomass estimates for a limited number of plots in the Duke forest. Tree height change between time periods and canopy segmentation from multispectral imagery allowed calculation of forest carbon stock change at costs that are comparable to those for ground-based methods. This work demonstrates the potential of lower cost imagery systems enhanced with laser data to collect high quality imagery and paired laser data for forestry and environmental applications. Additional research on automated canopy segmentation and multi-temporal image registration is needed to refine these methods for commercial use.