Periodic Reporting for period 4 - FREEDLES (From needles to landscapes: a novel approach to scaling forest spectra)
Okres sprawozdawczy: 2022-11-01 do 2024-04-30
Monitoring forests using satellite data is essential for understanding what is happening to them. Accounting for vegetation structure, such as the clumping of foliage into shoots or crowns, has for a long time been the largest remaining challenge in modeling scattered and absorbed radiation in complex vegetation canopies like forests. Clumping controls the radiation regime of forest canopies and influences the signal measured by optical satellite sensors, yet it has been poorly understood and quantified. The FREEDLES project developed a method for quantifying foliage clumping in forests using detailed 3D structure and spectral reflectance data. The FREEDLES approach contrasts with many other current developments that favor increasing complexity in canopy radiation models. Leveraging this scientific understanding, the project developed a spectral scaling approach that connects the spectra of vegetation at different hierarchical levels, from needles and leaves to crowns, stands, and landscapes. This approach was successfully tested with detailed reference measurements in both laboratory and natural conditions in European temperate, hemiboreal, and boreal forests. It was then applied to interpret forest characteristics from multispectral and hyperspectral satellite images.
A significant effort has been dedicated to collecting empirical data for developing and testing the new spectral scaling approach. We have carried out field and flight campaigns in multiple forest biomes (in Finland, Estonia and the Czech Republic), representing different types of forest structures and tree species. In addition, we have developed laboratory measurement methods e.g. for spectral properties of needles and leaves and for multiangular spectral measurements of trees. Another important line of development has been the use of modern LiDAR point cloud data to characterize the 3D structure of trees (esp. silhouette areas and clumping of crowns). So far, we have published 17 peer-reviewed journal articles, and are actively working on several manuscripts which will be submitted soon.
In the FREEDLES project so far, we have made progress beyond the state of the art in e.g. the following aspects:
1) We have designed the world's first experiment for measuring multiangular spectra of single trees.
2) We have performed the world's first test of the photon recollision probability (spectral invariants) theory for single trees.
3) We have collected significant spectral libraries of different components of vegetation (e.g. leaves, needles, understory vegetation and lichens).
4) We have used LiDAR point cloud data to estimate new concepts describing the structure of trees (incl. tree crown silhouette areas).
Currently, we are working on testing the photon recollision probability based model for real forests in natural outdoor conditions. We are using field and remote sensing data from all our forest sites (Finland, Estonia, the Czech Republic) for this purpose. We are also continuing the work on using LiDAR point clouds to characterize 3D structure of forests, and plan to connect this line of development conceptually to photon recollision probability based modeling. Finally, we aim to show previously unexplored links between vegetation structure and spectra, e.g. how understory spectral properties depend on overstory structure in different biomes.
1) We have designed the world's first experiment for measuring multiangular spectra of single trees.
2) We have performed the world's first test of the photon recollision probability (spectral invariants) theory for single trees.
3) We have collected significant spectral libraries of different components of vegetation (e.g. leaves, needles, understory vegetation and lichens).
4) We have used LiDAR point cloud data to estimate new concepts describing the structure of trees (incl. tree crown silhouette areas).
Currently, we are working on testing the photon recollision probability based model for real forests in natural outdoor conditions. We are using field and remote sensing data from all our forest sites (Finland, Estonia, the Czech Republic) for this purpose. We are also continuing the work on using LiDAR point clouds to characterize 3D structure of forests, and plan to connect this line of development conceptually to photon recollision probability based modeling. Finally, we aim to show previously unexplored links between vegetation structure and spectra, e.g. how understory spectral properties depend on overstory structure in different biomes.