Sun-Induced chlorophyll Fluorescence full spectrum retrieval and analysis of long-term time series

Remote sensing of Sun-Induced chlorophyll Fluorescence (SIF) represents a growing and promising area of research in support of the upcoming ESA’s FLEX (FLuorescence EXplorer) satellite mission. For this reason, the link between SIF and photosynthetic activity has been widely explored in the recent years, as tool to characterize and monitoring terrestrial ecosystems functioning.

 

The SIF detection is challenging because this faint signal (which represents only few percent of the total radiance) is over imposed on the light reflected from the Earth’s surface. Decoupling these two contributions is not trivial and dedicated algorithms are needed. For this reason, a novel SIF retrieval algorithm, named SpecFit, has been developed in order to retrieve the entire SIF spectrum in the entire wavelength interval in which chlorophyll fluorescence emission occurs (670-768 nm). This novel approach is able to disentangle SIF and reflectance contributions from the total radiance spectrum emerging from the top of canopy. Nevertheless, the further interpretation of the SIF spectrum in relation to plant photosynthesis is complicated by the fact that the SIF signal is strongly influenced by several biophysical parameters, such as canopy structure and chlorophyll content that affect the leaves/canopy radiation transfer and therefore the overall remote sensed signal. 

The proposed work aims to verify first the SpecFit algorithm robustness on both simulated and field data and, second to investigate the potential of novel fluorescence indexes defined from the SIF full spectrum.   

 

The algorithm accuracy has been tested on a set of simulated data, obtained by coupling MODTRAN (atmosphere) and SCOPE (canopy) radiative transfer models. Scatterplots between forward simulations and retrieved SIF showed R² close to 0.98 considering all the evaluated metrics, namely: maximum of the peaks in the red and far-red and SIF spectrum integral.

The temporal series acquired during the ESA’s ATMOFlex and FLEXSense campaigns organised in an agricultural area in Braccagni (Tuscany, Italy) were, instead, used to test the algorithm on experimental measures acquired with FLOX spectrometers, from February to August on different crops (forage, alfalfa and corn). For the first time, SIF spectra observed on different vegetation species at different growing stages are presented in this work and their consistency with SIF values estimated by the more consolidated and widely used Spectral Fitting retrieval Method (SFM) are presented. The relationship found shows a linear regression slopes close to 1, intercepts approximately equal to 0 and R² higher than 0.92 are all evidences of the SpecFit accuracy.  

 

The final step consists in analysing the temporal evolution of novel fluorescence indexes derived from the SIF spectrum. Specifically, SpecFit SIF evaluated at 760 nm and 687 nm and normalized by the retrieved spectrum integral (SIF_SpecFit/SIF_INT) were compared to the index SIF₇₆₀/SIF₆₈₇, the latter is a proxy of the chlorophyll content. SIF₇₆₀/SIF₆₈₇ and SIF₇₆₀/SIF_INT increase linearly during the growing season due to re-absorption processes that affect both the indexes. Conversely, an inverse relationship is found between SIF₇₆₀/SIF₆₈₇and SIF₆₈₇/SIF_INT because the contribute in the visible red wavelengths to the integral become weaker at increasing biomass content.