Carbon Estimation of Old-Growth Bald Cypress Knees Using Mobile LiDAR

Titilayo T. Tajudeen^1*Leah Rathbun²Marcelo Ardón¹Helena Mitasova¹

• ¹North Carolina State University, Raleigh, United States
• ²Forest Service, United States Department of Agriculture, Hamden, Connecticut, United States

A rapid, reliable, cost-effective tree volume calculation is critical for estimating biomass and carbon sequestration. This estimation is vital for developing better carbon budgets for wetland ecosystems to assess current and future climate scenarios. Portable mobile light detection and ranging (LiDAR) systems such as the Apple iPad Pro sensor provide an efficient method for capturing 3D shapes of bald cypress (Taxodium distichum) pneumatophores, or "knees." The knee is a rounded conical structure growing above the water or land from the roots of bald cypress trees, usually a few feet away from the trunk. This study explores remote sensing techniques for mapping individual knees to eventually understand their significance in the carbon balance of forested wetlands. This project was conducted in the Three Sisters Swamp, part of the Black River Reserve in North Carolina, USA. The volume of individual tree knees was estimated using multiple geometric algorithms and compared to allometric estimates from traditional field measurements derived from the shape of a cone. Specifically, we used the convex-hull by slicing (C-hbS) and Canopy-Surface Height (CSH) algorithms to estimate the volume of individual knees after LiDAR data processing. The volume estimates from the CSH and C-hbS methods are higher than the allometric estimates due to the knees' natural irregular shape and concavities. The CSH method returned the largest volume values on average. The discrepancy in estimated volume between the allometric equation and the two algorithms became more pronounced with increasing knee height. The estimated aboveground mean biomass and carbon of the knees are 61.9 ± 23.4 Mg ha -1 and 32.83±12.38 Mg C ha -1 , respectively. The challenges of algorithmic methods include the time and equipment needed to process dense point clouds. However, they better capture irregularities in knee shape, ultimately leading to better estimates and an understanding of knee structure, which is currently poorly understood.