Here, we present one of the first datasets recorded by a drone-based prototype dual camera system based on narrowband high-quality optical filters, hereafter named SIFcam. The camera allows for sensing solar-induced chlorophyll fluorescence (SIF) at 760 nm (F760) with centimeter ground sampling distance. The system performance was evaluated against simultaneous observations of a wheat-bean mixed canopy carried out using a mobile ground-based non-imaging system FloX (R2= 0.84, RMSE = 0.17 mWm-2nm-1sr-1) and SIF imagery acquired by the airborne imager HyPlant (R2= 0.52, RMSE = 0.24 mWm-2nm-1sr-1). The analysis of near-infrared reflectance allowed us to quantify on average 95.4% contribution of F760signal originating from sun-lit pixels of investigated canopies. Additional multispectral imagery facilitated calculation of fractional vegetation cover (FVC) of >88%. Spatial patterns of sunlit pixels are systematically consistent, indicating the plausibility of the SIFcam measurements.

The retrieval of sun-induced fluorescence (SIF) from hyper-spectral imagery is an ill-posed problem that has been tackled in different ways. We present a novel retrieval method combining semi-supervised deep learning with an existing spectral fitting method. A validation study with in-situ SIF measurements shows high sensitivity of the deep learning method to SIF changes even though systematic shifts deteriorate its absolute prediction accuracy. A detailed analysis of diurnal SIF dynamics and SIF prediction in topographically variable terrain highlights the benefits of this deep learning approach.

In many remote sensing applications the measured radiance needs to be corrected for atmospheric effects to study surface properties such as reflectance, temperature or emission features. The correction often applies radiative transfer to simulate atmospheric propagation, a time-consuming step usually done offline. In principle, an efficient machine learning (ML) model can accelerate the simulation step. This is the goal pursued here in the context of solar-induced fluorescence (SIF) emitted by vegetation around the O2-A band using the spaceborne DESIS and airborne HyPlant spectrometers. We present an ML simulator of at-sensor radiances trained on synthetic spectra and describe its performance in detail. The simulator is fast and accurate, constituting a promising alternative to a full-fledged, lengthy radiative transfer code for SIF retrieval in the O2-A band with DESIS and HyPlant.

Recent advancements in unmanned aerial vehicle (UAV) technologies made it possible to monitor agricultural fields at higher spatial and temporal resolution than commonly possible by aerial and satellite surveys. Mapping crop variables such as leaf area index (LAI) and leaf chlorophyll content (LCC) from low-cost UAV-based multispectral cameras can deliver vital information about crop status to farmers and plant breeders. Retrieval of these variables using radiative transfer models (RTMs) has been widely studied in the satellite remote sensing community but is still not well explored in the UAV remote sensing community. This study aims to investigate the advantages of high spatial resolution UAV image data for retrieving LAI and LCC using RTM inversion. A breeding experiment consisting of soybean plots has shown that high-resolution imagery (0.015m) delivers better retrieval accuracy compared to coarser resampled image data. Particularly, biochemical parameters, such as LCC, benefit from high spatial resolution.

Demand for high spatial and temporal resolution measurements has triggered a rapid development of unmanned aircraft systems (UAS) for plant phenotyping and precision farming purposes. Similarly, recent progress in low-altitude remote sensing of solar-induced chlorophyll fluorescence (SIF) resulted in several studies aiming at the development of SIF proximal sensing approaches. Although first experimental results are promising, the requirements for reliable and repeatable measurements in agricultural experiments still constrain applicability of these platforms. In this study, we analyze current capabilities and potentials of SIF measuring UAS for operational use. We highlight existing challenges and outline how UAS SIF sensing could be used more frequently and reliably in precision agriculture applications in the near future.

Well radiometrically calibrated UAV-derived reflectance maps are important when analysing time series of vegetation canopies. In this paper, we assessed the quality of reflectance calibration of a multispectral camera system, MicaSense Dual, using two different methods: a single-panel approach offered by the camera manufacturer and an empirical line method. The results show a significant discrepancy between the reference reflectance measurements, and the single-panel approach in the NIR and the red edge bands. This discrepancy is especially pronounced for dark targets. The empirical line correction method has proven to be more accurate, yet for shaded and densely vegetated areas it has led to negative reflectance values in the visible bands. Hence, we argue that users should be aware of the caveats of both reflectance calibration pipelines when working with time-series UAV data.

The upcoming launch of the L-band Synthetic Aperture Radar (SAR) satellite mission Radar Observing System for Europe L-band SAR (ROSE-L) will enable multi-frequency SAR observations when combined with existing C-band satellite missions (e.g., Sentinel-1). Due to the different penetration depths of the SAR signals, multi-frequency SAR offers great potential for field-scale agricultural monitoring and the estimation of soil and plant parameters. The SARSense campaign, conducted between June and August 2019 at the Selhausen agricultural test site near Jülich, Germany, has yielded a comprehensive dataset that includes both air- and space-borne C- and L-band SAR data, extensive in-situ field measurements of soil and plant parameters as well as unmanned aerial systems (UAS)-based multispectral and thermal infrared measurements and cosmic neutron sensing observations. The study provides both, an insight into the strengths and limitations of the acquired dataset as well as an analysis of the different behaviour of C- and L-band backscattering on changing soil moisture and plant parameters for taproot crops and cereals.

The connection between solar-induced fluorescence (SIF) and vegetation gross primary productivity is being widely investigated across spatial, temporal, and biological scales, including: a) studies at the leaf [1], [2], plant canopy [2]–[4] or satellite pixel scale [5], [6], b) temporally with studies spanning from diurnal [7] to seasonal scales [1], [3], [5], and b) biologically with studies covering various plant functional types (PFTs), e.g., crops [4], [7], deciduous [8] or evergreen forests [1], [3], in response to different sources of stress.

Solar-induced fluorescence (SIF) has become a promising remote sensing parameter to quantify actual photosynthesis beyond the ‘greenness' measurements. Despite the great advances in instrumentation to measure canopy SIF, we are still at the beginning of having concepts to quantitatively relate SIF to actual rates of photosynthesis. In this article, we discuss the three elements that are crucial to scale canopy SIF measurements to leaf function, namely (i) canopy structure and its bio-chemical composition determining light absorption, (ii) the functional status of photosynthetic light conversion and fluorescence emission under non-steady state conditions, and (iii) the re-absorption and scattering of the fluorescence signal within the canopy.

The impact of elevated$[\text{CO}_{2}](\mathrm{e}[\text{CO}_{2}])$in on yield, biomass (BM) and chlorophyll fluorescence (ChlF) was analyzed in three genotypes of common beans (Phaseolus vulgaris L.), a key food-security crop. Active- and passive-sensed ChlF traits acquired by the Light-Induced-Fluorescence-Transient (LIFT), Moni-Pulse-Amplitude-Modulation (MoniPAM), and Fluorescence Box$(\text{FloX})$instruments were compared. Total biomass increased for all genotypes under$\mathrm{e}[\text{CO}_{2}]$, but their biomass partitioning significantly differed. The highest yielding genotype under$\mathrm{e}[\text{CO}_{2}]$also showed the highest photosynthetic activity according to different active-sensed ChlF methods. Furthermore,$\mathrm{e}[\text{CO}_{2}]$resulted in earlier senescence, which was detected by either satellite- or FloXderived Normalized Difference Vegetation Index (NDVI). Moreover, we observed a significant agreement between MoniPAM- and LIFT-measured ChlF data$(\mathrm{R}^{2}=0.89, p= 0.02)$, as well as between SIF and FloX measurements$(\mathrm{R}^{2}= 0.62,p=0.03)$.

The Copernicus Hyperspectral Imaging Mission for the Environment (CHIME) is in preparation to carry a unique visible to shortwave infrared spectrometer. CHIME will globally provide routine hyperspectral observations to support new and enhanced services for, among others, sustainable agricultural and biodiversity management. The mission shall provide Level 1B, 1C and 2A products, as well a set of downstream products related to the different environmental applications, such as the quantification of vegetation traits. In this context, this work presents the first hybrid retrieval models for the operational delivery of vegetation properties products. Within ESA's CHIME end-to-end (E2E) simulator study, 13 leaf and canopy trait models were developed as part of the L2B vegetation (L2BV) module. The E2E framework functions as a simulated reality that enabled to test and improve the algorithms. The models were further tuned and validated against campaign data using active learning methods. As a proof of concept, the prototype retrieval models were applied to both hyperspectral airborne (HyPlant) and spaceborne (PRISMA) imagery that were first resampled to CHIME band settings. Among the provided vegetation products, it led to a first space-based canopy nitrogen content map over a heterogeneous landscape. The obtained CHIME-like L2BV traits maps demonstrate the feasibility to routinely deliver a collection of next-generation vegetation products across the globe.

In summer 2019 the SARSense campaign was held in Jülich, Germany, to provide insights into the potentials and specifications of the ESA Copernicus candidate mission ROSE-L (Radar Observation System for Europe). ROSE-L will consist of two satellites that carry a polarimetric L-band SAR. Since the L-band signal can penetrate through many natural materials such as vegetation, dry snow and ice, the mission will provide additional information that cannot be gathered by the Copernicus Sentinel-1 C-band SAR mission. The overall objective of the SARSense 2019 campaign is to analyze the mission design concerning its potential for agricultural monitoring services including target applications such as soil moisture monitoring, irrigation management, crop type discrimination, food security and precision farming. The SARSense in situ measurements of soil moisture, soil temperature, vegetation properties, UAS-based multispectral and thermal mapping, as well as the airborne SAR observations are presented as well as strategies for soil moisture retrieval and first analysis.

A regular monitoring of plants is inevitable to ensure an effective production and to reduce yield losses, for example, caused by different diseases. Infected plants show a visual effect shortly after inoculation. These effects can be understood as anomalies, which do not occur in healthy plant stocks. For automation of harvesting or spraying it is important to recognize anomalies to ensure an on-time reaction by the farmer or breeder. However, these anomalies differ largely in their appearance and a representative model is generally too complex to be learned. Our main objective is reconstruction-based anomaly detection by all-convolutional autoencoder (all-CAE), which combines convolutions with the architecture of an autoencoder (AE). To achieve our objective, we use an hourglass all-convolutional encoder-decoder architecture to create a highly compressed representation in the middle layer. Moreover, we compare different types of noise as regularizer. In our experiments, the method is tested on images of grapes acquired in a vineyard. We show that all-CAE are suitable for anomaly detection and that unnatural noise (salt) shows the best results.

The contribution presents the development and testing of a fluorescence retrieval scheme based on the ESA's FLuorescence EXplorer mission concept. The algorithm employs on a coupled surface-atmosphere forward model at oxygen absorption bands: i) the atmospheric effect is computed by MODTRAN5; ii) the surface reflectance and fluorescence are modeled by means of the Spectral Fitting approach. The algorithm, previously tested on numerical simulations, was further implemented and optimized to process real observations collected by the FLEX airborne demonstrator HyPlant. The retrieval scheme has been tested on a number of flight lines collected in several locations, different ecosystems types, atmospheric conditions and instrument observation conditions. For the first time, this work shows the capability of retrieving canopy fluorescence from real airborne observations by means of a physically-based algorithm as envisaged for FLEX. The results achieved on the large core data sets of imageries show the consistency of the physical retrieval algorithm for a wide range of scenarios and fluorescence values are in line with ground observations.

Reliable measurements of Sun Induced Fluorescence (SIF) require a good instrument characterization as well as a complex processing chain. In this paper, we summarize the state of the art SIF retrieval methods and measurements platforms for field phenotyping. Furthermore, we use HyScreen, hyperspectral-imaging system for top of canopy measurements of SIF, as an example of the instrument requirements, data process, and data validation needed to obtain reliable measurements of SIF.

Field phenotyping conceptually can be divided in five pillars 1) traits of interest 2) sensors to measure these traits 3) positioning systems to allow high-throughput measurements by the sensors 4) experimental sites and 5) environmental monitoring. In this paper we will focus on photosynthesis as trait of interest, measured by remote active fluorescence in `BreedFACE', an infrastructure for field phenotyping under elevated CO2(eCO2). The sensor used is the Light Induced Fluorescence Transient (LIFT) device mounted on a manual operated field4cycle positioning system. In direct response to eCO2at 600ppm for winter wheat, NDVI and photosynthetic efficiency (Fq`/Fm') did not change whereas the QA reoxidation efficiency (Fr2/Fm) decreased at eCO2. This study further demonstrates the completion of BreedFAce with an example of proximal sensing of photosynthetic traits by the LIFT.

Although traditional remote sensing systems based on spectral reflectance can already provide estimates of the “potential” photosynthetic activity of terrestrial vegetation through the quantification of total canopy chlorophyll content or absorbed photosynthetic radiation, the determination of the “actual” photosynthetic activity of terrestrial vegetation requires information about how the absorbed light is used by plants, such as vegetation fluorescence, using very high spectral resolution spectroscopy in the range 650-800 nm. The Fluorescence Explorer (FLEX) mission, selected in November 2015 as the 8th Earth Explorer by the European Space Agency (ESA), carries the FLORIS spectrometer, with a spectral resolution of 0.3 nm and a spatial resolution of 300 m, with a swath of 150 km. The FLEX mission is designed to fly in tandem with the Copernicus Sentinel-3 satellite, in order to provide all the necessary information to disentangle emitted fluorescence from the background reflected radiance, and to allow proper interpretation of the fluorescence spatial and temporal changes in relation to photosynthesis dynamics, accounting for non-photochemical energy dissipation and canopy temperature effects.

In November 2015, the FLuorescence EXplorer (FLEX) was selected as the eighth Earth Explorer mission of the European Space Agency. The tandem mission concept will provide measurements at a spectral and spatial resolution enabling the retrieval and interpretation of the full chlorophyll fluorescence spectrum emitted by the terrestrial vegetation. This paper provides a mission concept overview of the scientific goals, the key objectives related to fluorescence, and the requirements guaranteeing the fitness for purpose of the resulting scientific data set. We present the mission design at the time of selection, i.e., at the end of project phase Phase A/B1, as developed by two independent industrial consortia. The mission concepts both rely on a single payload Fluorescence Imaging Spectrometer, covering the spectral range from 500 to 780 nm. In the oxygen absorption bands, its spectral resolution will be 0.3 nm with a spectral sampling interval of 0.1 nm. The swath width of the spectrometer is 150 km and the spatial resolution will be 300 × 300 m-2. The satellite will fly in tandem with Sentinel-3 providing different and complementary measurements with a temporal collocation of 6 to 15 s. The FLEX launch is scheduled for 2022.

The Fluorescence Explorer (FLEX) mission has been recently selected as the 8thEarth Explorer by the European Space Agency (ESA). It will be the first mission specifically designed to measure from space vegetation fluorescence emission, by making use of very high spectral resolution imaging spectroscopy techniques. Vegetation fluorescence is the best proxy to actual vegetation photosynthesis which can be measurable from space, allowing an improved quantification of vegetation carbon assimilation and vegetation stress conditions, thus having key relevance for global mapping of ecosystems dynamics and aspects related with agricultural production and food security. The FLEX mission carries the FLORIS spectrometer, with a spectral resolution in the range of 0.3 nm, and is designed to fly in tandem with Copernicus Sentinel-3, in order to provide all the necessary spectral / angular information to disentangle emitted fluorescence from reflected radiance, and to allow proper interpretation of the observed fluorescence spatial and temporal dynamics.

Pre-symptomatic drought stress prediction is of great relevance in precision plant protection, ultimately helping to meet the challenge of "How to feed a hungry world?". Unfortunately, it also presents unique computational problems in scale and interpretability: it is a temporal, large-scale prediction task, e.g., when monitoring plants over time using hyperspectral imaging, and features are `things' with a `biological' meaning and interpretation and not just mathematical abstractions computable for any data. In this paper we propose Dirichlet-aggregation regression (DAR) to meet the challenge. DAR represents all data by means of convex combinations of only few extreme ones computable in linear time and easy to interpret.Then, it puts a Gaussian process prior on the Dirichlet distributions induced on the simplex spanned by the extremes. The prior can be a function of any observed meta feature such as time, location, type of fertilization, and plant species. We evaluated DAR on two hyperspectral image series of plants over time with about 2 (resp. 5.8) Billion matrix entries. The results demonstrate that DAR can be learned efficiently and predicts stress well before it becomes visible to the human eye.

Plant photosynthesis mediates about 60Gt of the carbon uptake by vegetated ecosystems and is considered to be a critical component of the terrestrial carbon cycle. Photosynthesis is, however, a highly adaptive process and lack of knowledge on its dynamic causes uncertainties in current carbon budgets. New remote sensing approaches allow measuring the chlorophyll fluorescence signal (FS) and hold the potential to directly assess ecosystem photosynthesis and related carbon assimilation rates (i.e., gross primary production (GPP)). This study provides one of the first spatial investigations of GPP at local/regional scale. FS was retrieved from data of the new imaging spectrometer APEX (Airborne Prism EXperiment) and used to model GPP. Spatial variations of GPP were investigated for five major ecosystems in Switzerland. Results of this study are considered as important, e.g., for the development of ESA's (European Space Agency) FLEX (FLuorescence Explorer) mission, for investigating functional ecosystem responses to environmental properties, or for evaluating common ecosystem monitoring approaches (e.g., eddy flux tower).

Early stress recognition is of great relevance in precision plant protection. Pre-symptomatic water stress detection is of particular interest, ultimately helping to meet the challenge of “How to feed a hungry world?”. Due to the climate change, this is of considerable political and public interest. Due to its large-scale and temporal nature, e.g., when monitoring plants using hyper-spectral imaging, and the demand of physical meaning of the results, it presents unique computational problems in scale and interpretability. However, big data matrices over time also arise in several other real-life applications such as stock market monitoring where a business sector is characterized by the ups and downs of each of its companies per year or topic monitoring of document collections. Therefore, we consider the general problem of embedding data matrices into Euclidean space over time without making any assumption on the generating distribution of each matrix. To do so, we represent all data samples by means of convex combinations of only few extreme ones computable in linear time. On the simplex spanned by the extremes, there are then natural candidates for distributions inducing distances between and in turn embeddings of the data matrices. We evaluate our method across several domains, including synthetic, text, and financial data as well as a large-scale dataset on water stress detection in plants with more than 3 billion matrix entries. The results demonstrate that the embeddings are meaningful and fast to compute. The stress detection results were validated by a domain expert and conform to existing plant physiological knowledge.