Speaker
Description
Understanding temporal variations in L-band forest backscatter remains challenging because the conventional SAR observations represent a mixture of canopy, branch, trunk and ground scattering contributions from multiple trees within a single resolution cell. High-resolution radar tomography provides an opportunity to separate these contributions and investigate forest dynamics at the scale of individual trees.
In this study, we explore the capability of tower-based L-band SAR tomography at all polarization combinations to identify and monitor individual trees in a boreal forest. Measurements were acquired by the BorealScat-2 radar tower located in the Svartberget Experimental Forest in northern Sweden. The tower features a horizontally moving antenna array that enables coherent SAR processing and reconstruction of 3D forest backscatter tomograms. Measurements are acquired every 30 minutes during the growing season. Due to acquisition times of several minutes, tree motion causes temporal decorrelation during windy periods, and focused tomographic reconstructions are therefore only obtained under low-wind conditions.
L-band tomographic observations from two growing seasons are analyzed together with meteorological data and stem water content measurement acquired at three heights of a few trees. The high spatial resolution of the tomograms enables extraction of vertical reflectivity profiles for selected trees and the assessment of their temporal variability across polarization channels.
Results show that observations acquired during calm conditions allow clear identification of individual trees within the tomogram and reveal temporal variations in their vertical reflectivity profiles. Diurnal backscatter variations as well as longer-term responses to changing environmental conditions are observed. Differences in the temporal behavior between trees are further investigated using LiDAR-derived structural information to assess the influence of local canopy structures and line-of-sight conditions.
The results demonstrate the potential of tower-based L-band SAR tomography to resolve and monitor individual trees within a dense boreal forest and provide new insight into the processes governing temporal L-band backscatter variations and their relation to stem water content variations.