Synergistic use of multi- and hyperspectral remote sensing data and airborne LiDAR to retrieve forest floor reflectance

0574223 - ÚVGZ 2024 RIV NL eng J - Journal Article
Hovi, A. - Schraik, D. - Kuusinen, N. - Fabiánek, Tomáš - Hanuš, Jan - Homolová, Lucie - Juola, J. - Lukeš, Petr - Rautiainen, M.
Synergistic use of multi- and hyperspectral remote sensing data and airborne LiDAR to retrieve forest floor reflectance.
Remote Sensing of Environment. Roč. 293, AUG (2023), č. článku 113610. ISSN 0034-4257. E-ISSN 1879-0704
R&D Projects: GA MŠMT LM2023048
Institutional support: RVO:86652079
Keywords : understory * spectroscopy * radiative transfer * reflectance modeling * airborne laser scanning * sentinel-2 * prisma * hyperspectral
OECD category: Forestry
Impact factor: 13.5, year: 2022
Method of publishing: Open access
https://www.sciencedirect.com/science/article/pii/S003442572300161X?via%3Dihub

Forest floor vegetation can account for a notable fraction of forest productivity and species diversity, and the composition of forest floor vegetation is an important indicator of site type. The signal from the forest floor influences the interpretation of optical remote sensing (RS) data. Retrieval of forest floor reflectance properties has commonly been investigated with multiangular RS data, which often have a coarse spatial resolution. We developed a method that utilizes a forest reflectance model based on photon recollision probability to retrieve forest floor reflectance from near-nadir data. The method was tested in boreal, hemiboreal, and temperate forests in Europe, with hemispherical photos and airborne LiDAR as alternative data sources to provide forest canopy structural information. These two data sources showed comparable performance, thus demonstrating the value of using airborne LiDAR as the structural reflectance model input to derive wall-to-wall maps of forest floor reflectance. We derived such maps from multispectral Sentinel-2 MSI and hyperspectral PRISMA satellite images for a boreal forest site. The validation against in situ measurements showed fairly good performance of the retrievals in sparse forests (that had effective plant area index less than 2). In dense forests, the retrievals were less accurate, due to the small contribution of forest floor to the RS signal. We also demonstrated the use of the method in monitoring the recovery of forest floor vegetation after a thinning disturbance. The reflectance model that we used is computationally efficient, making it well applicable also to data from new and forthcoming hyperspectral satellite missions.
Permanent Link: https://hdl.handle.net/11104/0344567