Sun-induced chlorophyll fluorescence and biophysical variables analysis in a forest ecosystem

This contribution aims at analysing the distribution of biophysical variables and sun-induced chlorophyll fluorescence (F) retrieved from high-resolution hyperspectral images acquired over a mixed forest ecosystem. The airborne data were acquired on June 16th, 2013 over the Hardt forest, a mid-latitude plain forest located in Alsace (France), using the imaging spectrometer HyPlant, developed by Forschungszentrum Jülich in collaboration with SPECIM Spectral Imaging Ltd. HyPlant is composed of a hyperspectral sensor covering the visible, near-infrared and shortwave infrared part of the spectrum, and of a narrowband sensor specifically designed for F retrieval, that covers the 670-780 nm spectral region with a sub-nanometer resolution. The images were elaborated in order to obtain three products: (i) a thematic map of the distribution of the main species over the study area (i.e., hornbeam (Carpinus betulus L.), oak (Quercus robur L., Quercus petraea (Matt.) Liebl.), maple (Acer campestre L.), linden (Tilia L.), pine (Pinus L.) and larch (Larix decidua Mill.)), through a machine-learning classification method; (ii) maps of biophysical variables (e.g., chlorophyll, LAI maps) through inversion of radiative transfer models; (iii) a map of F at the oxygen absorption bands through the innovative Spectral Fitting Method (SFM). A field campaign was conducted concurrently with the overflight to collect ground measurements to validate all the products: (i) forest species composition and crown condition were assessed by forest experts in 22 sampling units of 20_20 m; (ii) pigment concentration and LAI were determined in the same sites by means of laboratory extraction and hemispherical photography acquisition; (iii) top-of-canopy red and far-red F was measured over the main forest species using high-resolution spectroradiometers manually operated from a mobile hydraulic platform. The validated products obtained from HyPlant airborne data were used to compare and analyse the spatial variability of F and biophysical variables in relation to species composition and forest structure. Results showed that different species (mapped with an overall accuracy of 75%) are characterised by different F values: pine and linden are characterised by the lowest fluorescence emission, while the other broadleaves show higher values. The F emission intensity analysed in relation to pigment content and biomass pointed out that these relationships are not unique. In highly dense and green canopies, where the biophysical variables tend to saturate, F shows an inter-species variability, highlighting the potential of using F maps in combination with reflectance-derived biophysical products in order to improve our understanding of complex ecosystem's dynamics