Satellite remote sensing offers great promise to monitor land surface processes across a range of spatial and temporal scales, but integrating multiple sources of remote sensing into a single platform remains a challenge. Here we present BESS as a flexible and scalable platform to monitor ecological processes from plot to the global scales. BESS is based on first principles in atmospheric and canopy radiative transfer, canopy photosynthesis, transpiration, and leaf energy balance which are forced with multiple satellite remote sensing data. Therefore, BESS provides the flexibility to add new modules. In the default setting, BESS produces maps of canopy photosynthesis and evaporation at 1 km, daily scale since 2000. We present three recently added modules: BESS-Rad, BESS-Rice, and BESS-SiF. BESS-Rad provides global maps of shortwave, PAR and diffuse PAR at 5 km, daily scales since 2000, which were validated with data of more than 150 stations. BESS-Rice simulates rice growth development and yield formation by allocating canopy photosynthesis into grain, leaf, stem and root and was evaluated at six site-years in Korea. BESS-SiF computes canopy escaping SiF by coupling the Farquhar photosynthesis model with an established leaf model of SiF and an escape ratio scheme under the two-leaf canopy model, which was tested on tower-based, landscape and global scale SiF datasets. BESS is highly scalable. As one example, we present BESS based carbon and water flux maps at 30 m resolution through image fusion in crop landscapes. We are incorporating more modules into BESS and increasing spatial and temporal resolutions with more satellite datasets, which could serve as one tool for monitoring our Living Planet.

Understanding and predicting ecosystem functions is a major challenge in ecological and environmental sciences. A critical step is to understand how current dynamic vegetation models can grasp the spatial and temporal variability in ecosystem processes. In this context, remote sensing data at a range of spatial scales from proximal observations to global extent can detect essential changes in plant traits, biodiversity and ecosystem functioning, providing a method for scaling-up. However, there are still methodological and technical constraints that hamper a systematic incorporation of remote sensing products in ecosystem models, including scalability and multi-source data integration issues.
In this contribution, the main achievements of the Training on Remote Sensing for Ecosystem modelling (TruStEE) project, funded by Horizon 2020 Marie Curie ITN 2016, are presented. TRuStEE was established in 2016 to create a European Network on the use of the still under-exploited potential of novel remote and proximal sensing techniques to face grand societal challenges as the increasing pressure of environmental change on ecosystem functioning and land-atmosphere interactions. In particular, specific objectives addressed in the project are:
i) to exploit the recent technological development in the field of proximal and remote sensing (e.g. Unmanned aerial vehicles, fluorescence sensors) and Earth Observation data from new satellite missions (e.g. ESA-FLEX) for the description through dynamic products of the ecosystem functioning and the prediction of ecosystem services;
ii) to improve the understanding and modelling of biosphere-atmosphere interactions and ecosystem services allowing the exploration of inter-linkages, synergies and feedbacks between ecosystem functional properties (EFPs) and plant traits (PTs) measurable from remote observation.
These main topics are developed in the individual PhD projects of twelve early stage researchers involved in the TRuStEE training network.
The more promising and useful PTs and EFPs retrievable by remote sensing techniques were identified based on a literature review. Protocols for field data and drone data acquisition were drafted and shared; an assessment of the optimal strategies for characterizing PTs and EFPs from remote sensing was achieved through the use of a wide range of sensors and platforms.
For the first time, complementary, overlapping information products were acquired and processed at a variety of spatial, spectral, temporal and angular resolutions. An innovative spectrometer for vegetation optical property acquisition was deployed on flux towers acquiring data in an automatized operation and used for manual bidirectional measurements using a goniometer prototype. Rotary wing drones with optical and thermal sensors and airborne imaging sensors were flown with a spatial resolution of centimeters to few meters. Scaling approaches from drone to airborne and medium resolution satellite images (e.g. from Landsat-8 and Sentinel-2 and 3) were also performed in different natural and agricultural ecosystems using empirical and physically based approaches.
The link between PTs, EFPs and the vegetation optical properties was analysed using the complete datasets of remote sensing data mentioned above, collected on the same focus areas by the multiple platforms at various altitudes and view angles.

Acknowledgments
The project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 721995.

Novel measurement techniques to estimate sun-induced chlorophyll fluorescence (F) together with the hyperspectral reflectance (R) have become available in recent years. A relevant question is whether the combined use of these measurement techniques can help to quantify the productivity of different crop varieties. We investigated the F and R of two varieties of Soybean (Glycine max (L.) Merr.), a chlorophyll-deficient (MinnGold, chlorophyll (Cab) of 9.52±0.98 μg cm-2) and a non-chlorophyll deficient (Eiko, Cab of 46.49±5.59 μg cm-2) variety, in order to explain similarities and differences in photosynthesis and light use efficiencies. We were particularly interested in explaining the light absorption and fluorescence emission at photosystem level, which are both indicators of photosynthesis.
Our measurements of top of canopy (TOC) F and R, and canopy level gross primary productivity (gas exchange chambers) showed that despite the higher Cab of the Eiko both varieties present similar canopy-scale gross primary production, and the ratio of carbon uptake to light absorption ratio was higher in the MinnGold. We further retrieved parameters for the ‘Soil-Canopy-Observation of Photosynthesis and Energy fluxes’ model (SCOPE) from leaf and (TOC R, and used SCOPE to correct F for reabsorption of fluorescence in the canopy. The emission of fluorescence in the photosystems appeared to be similar in both varieties. This indicates that the higher light absorption in the Eiko as compared to the MinnGold is compensated by higher non-photochemical quenching (NPQ) due to light saturation. Our findings demonstrate that the combination of F and hyperspectral R can help better understand and quantify the light reactions of photosynthesis in crop varieties.

Remote and proximal sensing applications, especially with the advent of fine resolution spectroradiometers, have the most untapped potential in deriving plant traits (PTs) and ecosystem functional properties (EFPs) from ecosystem level observations. These allow for characterizing physiological processes and their responses to the environment. Along with traditional reflectance-based approaches, sun-induced chlorophyll fluorescence (F) - a portion of energy absorbed by chlorophyll and reemitted at longer wavelengths (red and far-red) - is increasingly employed for monitoring variations in plant biochemical and physiological functions.
Long-term continuous ground-based spectroscopy measurements represent a novel approach for monitoring canopy temporal development. Paired with simultaneous carbon flux data from eddy covariance towers, they can provide better understanding of the link between vegetation indices and F and photosynthesis. Disentangling the phenological, fast-response physiological, structural and directional components of time-series of optical indices and F can help to retrieve plant traits and ecosystem processes characterised by different dynamics. However, this decoupling is still a challenging task, and availability of continuous data of different types of canopy will facilitate it.
Therefore, the aim of our research is the decoupling of the slow and fast temporal dynamics of time-series of optical indicators of plant physiology (e.g. Photochemical Reflectance Index, PRI) using spectral data collected continuously with high resolution field spectrometers. Moreover, this can help to identify the temporal windows more suited to retrieve ecosystem processes characterised by different dynamics. The developed time-series decomposition was tested on synthetic dataset created with the SCOPE model that includes the xanthophyll cycle activation/deactivation routine to account for non-photochemical quenching (NPQ) effect. Also, the technique was applied on a real dataset, collected in Las Majadas (Mediterranean tree-grass ecosystem). This site is equipped with an eddy covariance flux tower measuring carbon and water fluxes between the ecosystem and the atmosphere, and a hyperspectral instrument (FloX, JB Hyperspectral Devices UG), specifically designed to measure F760 and F687 and visible-near infrared reflectance.
The technique, based on Fourier analysis, allowed to distinguish physiological response of PRI from phenological cycles, which was validated by comparison with fast-response components of photosynthesis, obtained from the eddy covariance tower data. The variation attributed to the directional effects, was compared with field multi-angular reflectance data, acquired with the FloX system coupled with a goniometer device.
The obtained results provide new insights into understanding of the temporal dynamics of optical signals related to PTs and EFPs in different terrestrial ecosystems.

Co-authors: Atherton J, Rajewicz P, Riikonen A, Grebe S, Weiwei L, Zhang C, Aalto J, Alonso L, Aro E-M, Bäck J, Bendoula R, Burkart A, Chen H, Erkkilä K-M, Féret J-B, Fernandez-Marin B, Frankenberg C, Garcia-Plazaola JI, Gastellu-Etchegorry J-P, Hakala T, Hartikainen S, Häkkänen H, Honkavaara E, Ihalainen J, Julitta T, Junker L, Kolari P, Kooijmans LMJ, Kusnetsova V, Levula J, Loponen MH, MacArthur A, Magney T, Mäkipää R, Malenovsky Z, Maseyk K, Mõttus M, Näthe P, Neimane S, Oksa E, Osterman GB, Rascher U, Robinson I, Robson M, Sabater N, Solanki T, Tikkanen M, Vesala T.

Chlorophyll-a fluorescence is increasingly being applied to investigate the dynamics of photosynthesis at unprecedented scales, via the estimation of solar-induced fluorescence (SIF) from increasing numbers of tower, airborne or satellite platforms. The connection between SIF and photosynthesis originates at the photosystem level via light absorption and subsequent energy partitioning between photochemistry, regulatory heat dissipation, and chlorophyll fluorescence emission, but the two signals are intricately connected at the leaf and canopy scales too. At the leaf level, diurnal variations in ChlF can relate to changes in photochemical or non-photochemical quenching, antenna and pigment dynamics, or chloroplast movements, whereas seasonal variations can include sustained changes in photochemical and non-photochemical quenching related to the photoinhibition of reaction centres or the adjustment in the protein-pigment structure of the photosystem. At the canopy level, variations in ChlF can reflect changes in total leaf area, canopy structure, species-specific phenology, dynamics of background, or changes in illumination or sun-target-sensor geometry. At the pixel level, atmospheric scattering and attenuation will also influence the fluorescence signal requiring correction. Clearly, while some of these controlling factors can further strengthen the link between SIF and photosynthesis, others can decouple it. Mutiscale datasets spanning from the leaf to the satellite pixel level and from the short-term to the long-term will be needed to quantitatively assess how the link between fluorescence and photosynthesis propagates across scales. We here present results from a recent multiscale campaign conducted in Southern Finland during the winter-to-summer transition in an “evergreen” boreal forest. The goal of the Florescence Across Space and Time (FAST) campaign was to characterize the processes that control the ChlF signal dynamics at each scale, establishing a comprehensive dataset for multiscale hypothesis and model validation.
Measurements expanded from the molecular to the satellite pixel level and from the picosecond to the seasonal scale, including five species and two different canopy positions, providing a unique record of the multiscale spring recovery of photosynthesis in a boreal forest. Amongst others we recorded or quantified: Fluorescence lifetime, ChlF spectra, Fast OJIP kinetics, PSI and PSII activity, total leaf absorption, foliar pigments, photosynthetic proteins, photosynthetic gas exchange, carbonyl sulphide (COS), volatile organic compounds (VOCs), total leaf absorption, canopy SIF and spectral reflectance, ecosystem CO2, water, COS, and VOC fluxes, as well as drone-based measurements of ecosystem SIF from 30 to 500 m distance to match with OCO-2 SIF obserbations. We here report the newest results from the FAST campaign which emphasize the spatial and temporal variability in the coupling between fluorescence and photosynthesis across species and canopy light gradients. The roles of Photochemical (PQ) and Non-photochemical Quenching (NPQ), light absorption, foliar pigment contents, PSII:PSI stoichiometry, photosystem structure, species and light environment are discussed.

Remote sensing of Sun-induced chlorophyll fluorescence is a topic of growing interest for the scientific community thanks to its inherent link with vegetation photosynthetic activity. Within this context, a number of activities has been carried out by ESA towards the future FLEX mission, the first mission specifically conceived for fluorescence retrieval from space. Among these, several measurements campaigns were performed using hyperspectral devices specifically designed for passive fluorescence retrieval from ground and airborne platforms. These activities set the basis for the selection of FLEX as the 8th Earth Explorer mission.
In February 2018, ESA started the ATMOFLEX project, with the main aim of collecting Sentinel-3 measurements and co-located ground observations characterizing the state of the atmosphere for an extended period of time. Within this context, several sites have been instrumented with state-of-the-art hyperspectral spectroradiometers (FloX; J&B Hyperspectral Devices, Germany) continuously measuring over different vegetated targets. The FloX houses two high resolution spectrometers covering the visible and near infrared spectral regions, one being specifically optimized for fluorescence retrieval, measuring the incoming irradiance and upwelling radiance from the vegetation. The measurement sites span across Europe (Italy, Germany, France, Spain, and Switzerland), in a variety of natural and agricultural ecosystems. Besides the ground measurements, an intensive airborne campaign was carried out between June and August 2018 flying the airborne FLEX demonstrator HyPlant sensor over all sites.
In this contribution we exploit the feasibility of consistently retrieving plant traits and Sun-induced chlorophyll fluorescence from canopy level FloX and HyPlant measurements, through the numerical inversion of a light version of the SCOPE model. The optimization approach was adapted for working with the high-frequency (~1-3 minutes) time series produced by the FloX, as well as the spatially resolved HyPlant imagery. Within this context, a strategy for optimal retrieval of plant traits at daily scale is discussed too.
The full spectrum of canopy F, the fluorescence quantum efficiency, as well as the main plant traits that control light absorption and reabsorption, were retrieved concurrently. Improvements and challenges of this method compared to other retrievals are discussed, together with the potential of applying a similar retrieval scheme to the reconfigured “FLEX mode” data acquired with the recently launched Sentinel-3B during the commissioning phase.