Abstract
In forest ecosystems the loss of diversity at the genetic, species and landscape levels is a huge social, environmental and economical loss that undermines different ecosystem services providing a multitude of benefits to humans. The assessment and mapping of the forest biodiversity is one of the biggest challenge of our time. Earth observation represent a key instrument for this purpose, collecting rapidly and economically vast quantities of environmental data at multiple spatial and temporal resolutions. The main objective of this thesis is to study new methodologies and hypothesis to assess tree species diversity in different forest ecosystems using different remote sensing data. In particular, in the first part of the thesis, we tested the Spectral Variation Hypothesis (SVH) in different forest ecosystems to understand the relationship between tree species diversity and spectral heterogeneity (SH) of different optical satellite data. For this purpose, different indices and optical data have been tested through a single and multi-temporal approach. The results showed that the SVH could be an interesting tool for the assessment of tree species diversity in forest ecosystems. The outcomes showed that the relationship between SH and tree species diversity is seasonal- and ecosystem dependent (in terms of species composition). The spectral and spatial resolution of the optical data also influences the above mentioned relationship. In the second part of the thesis LiDAR data have been used to transfer the concept of the SVH to structural heterogeneity in order to understand the relationship between the variation in forest tree height and species diversity. The approach has been named "Height Variation Hypothesis -HVH-" and tested in different forest ecosystems. The results showed that the HVH could be used as a proxy of forest tree species diversity. Forest density and spatial resolution of the LiDAR data are the main factors that influences the HVH. The last part of the thesis aimed to understand which are the biophysical parameters (BP), retrieved through the inversion of different radiative transfer models, that drives the SVH in the alpine coniferous forest. The results showed that the SVH, in the considered ecosystem is mainly influenced by the spectral response of some BP (Brown pigments in primis and also Carotenoids and Chlorophill). Concluding, the variability in the values of remotely sensed data (spectral, traits variability or height heterogeneity) have been analyzed and tested in this thesis, as a proxy to assess tree species diversity in forest ecosystems, highlighting its strengths, limitations and possible applications.