EP002-0002
Carbon Stored as Large Wood in the Mackenzie River Delta, Northwest Territories, Canada
Abstract:
In this work, we use remote sensing and field work to quantify carbon stored as large wood in the MRD. We measure the distribution of large wood deposits by applying machine learning classification algorithms to very high-resolution satellite imagery of the region. Algorithms tested include Support Vector Machines (SVM) and Convolutional Neural Networks (CNNs), resulting in accurate wood capture in the landscape. Using targeted field work, we transform wood extents into wood volume and carbon stock. We find extensive wood deposit sites across the delta that have residence times spanning decades to centuries, making them long-term carbon storage pools. Measured wood coverage over the delta results in 26e6 m3 of wood stored with carbon storage estimates of greater than 6600 Mg-C. Results from field work also suggest extensive wood deposits under the canopy that cannot be captured by imagery, making total estimated carbon stored as large wood over 13e3 Mg-C. We compare these results to carbon stocks found in other landscapes.