exploring fluorescence and pigment reflectance as methods to estimate photosynthesis with remote sensors
Nastassia Rajh Vilfan is a PhD student in the research group Water Resources. Her supervisor is prof.dr.ing. W. Verhoef from the faculty of Geo-Information Science and Earth Observation.
Almost any food chain on this planet begins with plants: by harvesting the energy of the sun, they provide food and oxygen for us all. Even more, they do so by incorporating the inorganic carbon (CO2) from the air, converting it into organic compounds. With the climate changing and the human population rising, it is crucial to develop efficient methods that can monitor vegetation photosynthetic efficiency, in order to track the global carbon and improve food production.
Plants absorb much of the available light, and the remaining light is either reflected or transmitted back into the environment. How much light a plant can absorb under certain conditions depends on its photosynthetic capacity. When the plant is exposed to limiting conditions, for example, reduced soil moisture or lack of certain nutrients, such stress negatively affects the amount of light that can be used for photosynthesis.
These subtle changes in plant absorption can be detected with remote sensors in the visible and near-infrared part of the electromagnetic spectrum. For decades, we have been observing vegetation from the leaf to global scales, with leaf probes, aircrafts and satellites; and lately also with affordable and fast developing innovations, such as drones. The key aspect of vegetation remote sensing is its ability to non-destructively monitor plant photosynthetic activity and health, detecting stress before significant damage in plant tissueshas even occurred.
Two indicators have been proven particularly valuable as estimators of terrestrial photosynthesis: Chlorophyll Fluorescence (ChlF) and Photochemical Reflectance Index (PRI). ChlF emanates directly from the photosystems and II, while the PRI is linked to the xanthophyll cycle effect, effectively dissipating excess absorbed energy as heat: Together with photochemistry they form a balance between dissipation and utilization of absorbed light. Both ChlF and PRI can be detected from leaf to satellite scales, and their potential to track photosynthetic activity of vegetation has led to selection of the FLuorescence EXplorer (FLEX) as the eighth Earth Explorer mission of the European Space Agency.
To correctly interpret the remotely sensed information, models are needed. Physical or radiative transfer models can explain how light propagates through leaves and canopies, while models for photosynthesis can explain the biochemical utilization of the available energy. The two types of models can be combined, promoting our understanding of how the changes in optical properties of vegetation are linked to the process of photosynthesis. The objective of this study was to extend a leaf radiative transfer model to include both ChlF and PRI, and explore their potential as methods for remote sensing of leaf photosynthesis. The main objective was achieved as a combination of three consecutive steps. At each step, the model performance was evaluated and validated using various datasets collected over the course of the study.
First, a leaf radiative transfer model Fluspect (Fluspect-B) was developed, which simulates leaf ChlF, reflectance and transmittance spectra. The existing PROSPECT model and its concept of a compact leaf were used as a starting point. Fluspect calculates the emission of ChlF on both the illuminated and shaded side of the leaf, with incident light and ChlF quantum efficiencies (h) for the two photosystems provided as the input parameters. To solve the differential equations for the radiative transfer within the leaf, an efficient doubling algorithm is used. Due to the simplicity of these equations, Fluspect offers a high computational speed. The results show, that Fluspect simulations can closely match the observed ChlF spectra, especially for ChlF measured under natural illumination. Most of the variability in ChlF and reflectance of different leaves could be explained from differences in leaf pigment contents, amount of water and leaf thickness, while h was shown to hold potential additional information.
In the next step, Fluspect-B was extended to include the changes in green reflectance as caused by the xanthophyll cycle effect (Fluspect-CX). The xanthophyll cycle is an interconversion of three xanthophylls belonging to a carotenoid pigment group: violaxanthin, antheraxanthin and zeaxanthin. Violaxanthin de-epoxidation provides a sink for the excess absorbed energy in a process called non-photochemical quenching (NPQ) of chlorophyll fluorescence. The changes in the de-epoxidation state (DEPS) of xanthophyll pigments can be observed as changes in the leaf absorption of light with wavelengths between 500 to 570 nm. The leaf is said to be unstressed, when DEPS = 0, and fully stressed when DEPS = 1. The idea of Fluspect-CX is to use in vivo specific absorption coefficients for two extreme states of carotenoids, representing the two extremes of the xanthophyll de-epoxidation, and to describe the intermediate states as a linear mixture of these two states. The ’photochemical reflectance parameter’ (Cx) quantifies the relative proportion of the two states. Cx was estimated from reflectance and transmittance measurements of various datasets, and the retrieved Cx correlated with measured xanthophyll DEPS. Moreover, the results indicated a clear relation between Cx and NPQ, important for the last step of this study.
In the last step, Fluspect-CX was coupled to an extended photosynthesis model, able to explain the relationship between fluorescence, photosynthesis and heat dissipation at the leaf level. The two models were linked by means of ChlF and photochemical reflectance: outputs of the biochemical model, fluorescence efficiency and NPQ, could be linked to _ and Cx, respectively. By inverting the combined model, a new method was developed for the estimation of the maximum photosynthetic capacity (V cmax) parameter from leaf ChlF and reflectance information. V cmax was estimated from hyperspectral measurements of CO2 and light response curves measured on sugar beet and barley leaves. The method can correctly estimate the magnitude of V cmax, when compared to the values estimated from gas exchange measurements. Using a coupled model instead of empirically derived relations among spectral and photosynthetic information opens up new ways to study the link between leaf radiative transfer and underlying biochemical processes.
As an addition, Fluspect-CX was incorporated into the Soil Canopy Observation, Photochemistry and Energy fluxes (SCOPE) model to scale the processes to the canopy level. Preliminary results indicate that the directional and physiological effects on the canopy reflectance could be separated. Including photochemical reflectance into SCOPE provides the foundation for the future studies of these effects on the canopy, as well as airborne and potentially satellite scales. Since both leaf models, Fluspect and Biochemical, are an integrated part of the model SCOPE, a scheme for V cmax estimation, similar to the one developed for the leaf in this study, could be devised for the canopy and higher spatial scales. SCOPE has been used as an ’end-to-end simulator’ in the FLEX/Sentinel 3 Tandem Mission Photosynthesis study, making the results of this dissertation particularly relevant for upcoming FLEX satellite mission. By including the two promising spectral indicators into a leaf radiative transfer model, and being able to harness the information they provide by coupling them to a model for photosynthesis, this study provides an important advancement in the remote sensing of vegetation.
