The DESIS Hyperspectral instrument is approaching four years in operation at the MUSES platform on the International Space Station. DESIS operates in the VNIR range (400-1000 nm) and has some unique properties among spaceborne hyperspectral instruments like a narrow spectral sampling distance (2.55 nm), the use of a pointing unit and a LED equipped calibration unit. During this time, the instrument has been calibrated using a combination of vicarious calibration and calibration unit data. The calibration unit is used for the spectral calibration of the sensor. The analysis of the LED data has shown two different operation modes depending on the temperature gradient inside the instrument. Moreover, the data shows that the spectral calibration is stable within 0.1 nm (RMS) for each temperature gradient mode, with a separation between them of around 0.5 nm. The vicarious calibration is used for the radiometric calibration and uses as input Earth images over homogeneous areas and reference data from RadCalNet stations. Data from the homogeneous areas are used to update individual pixel coefficients in order to provide a uniform spatial and spectral sensor response. Later, RadCalNet data are used for an adjustment of the absolute calibration scale. With this approach the DESIS instrument can be calibrated within 4% (RMS) and absolute mean bias within 2% at wavelengths above 500 nm. The long-term analysis of the DESIS calibration data also shows that the sensor experiences a mean degradation of around 3.4%/year above 500 nm. Below 500 nm the instrument shows a stronger degradation (up to 20%/year at the shortest wavelengths) between the start of operations and July 2021, being stable after this date. Finally, the orthorectification process achieves a RMSE of 20 m in North and East directions with the automatic extraction of ground control points and comparison with reference images.

The Environmental Mapping and Analysis Program (EnMAP) hyperspectral satellite mission was successfully launched on 1st April 2022. The mission aims to monitor and characterise Earth’s environment in the spectral range from 420 – 2450 nm. The VNIR sensor provides 91 science channels ranging from 420 – 1000 nm with an average Spectral Sampling Distance (SSD) of 6.5 nm. While the SWIR sensor covers the range from 900 – 2450 nm with 131 channels and 10nm SSD. - The off-nadir pointing capability (up to 30 degrees) enables 5000 km to be monitored per day, with a swath width of 30 km and a spatial resolution of 30 m. The EnMAP satellite is equipped with several subsystems which allow periodic in-flight monitoring and calibration. The Full Aperture solar Diffuser Assembly (FADA) is used for absolute radiometric calibration. The On-Board Calibration Assembly (OBCA) is composed of 2 integrating spheres: one is coated with a doped diffuser material and is used for the spectral calibration; the second sphere, coated with a white spectralon, is used for Radiometric stability monitoring. Linearity LEDs are placed in front of the detector to monitor their linearity by measuring the response at constant illumination with increasing integration times. The Shutter Calibration Mechanism (SCM) allows for measurements with no light input to be performed in order to compute Dark Signal values and, in combination with Deep Space measurements, to compute any existing shutter emission in the SWIR range. This contribution will present a summary of the calibration activities performed during the EnMAP Commissioning Phase.

The Earth Mineral Dust source InvesTigation (EMIT) is a remote orbital visible shortwave infrared (VSWIR) imaging spectrometer - also known as a hyperspectral imager - measuring reflected solar radiance.   This talk describes the EMIT spectroradiometric calibration and on-orbit validation experiments. EMIT was launched to the International Space Station in July 2022, beginning a year-long mission to map the surface mineralogy of Earth's mineral dust forming regions. EMIT consists of just a handful of optical elements, no shutter, and no onboard calibration systems. Instead, its design philosophy emphasizes stability in order to enable calibration with vicarious targets. Notable achievements include successful on-orbit adjustments of FPA alignment with sub-micron precision, and spectral uniformity better than 98%. Optical artifacts are at least three orders of magnitude below signal.   EMIT demonstrates the high spectral fidelity necessary for the next generation of orbital imaging spectrometers. This research was performed by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. Copyright 2022 California Institute of Technology. We acknowledge support of the NASA Earth Science Division.

Availability of hyperspectral Earth Observation (EO) data is internationally recognised as key to address a number of the most important scientific and environmental challenges to monitor the status and understand the dynamics of our planet, and in parallel to support policy making. This need is broadly shared and several space agencies have already launched their hyperspectral EO satellites (e.g. the Italian Space Agency’s PRISMA mission, the DLR’s DESIS and EnMAP missions, etc.) and some others are in the process of designing and developing theirs (e.g. the NASA’s SBG mission and the ESA developed CHIME mission), targeting launch before 2030. The number of applications and activities that will benefit from the wide availability of hyperspectral EO data is remarkable and goes from (but not limited to) agriculture to food security; from raw material and soil monitoring to biodiversity and environment protection; from inland/coastal waters monitoring to forestry. However, broad user access to large amounts of hyperspectral EO data acquired with a multi-satellite and multi-Agency approach comes also with challenges that have to be tackled at early stages of the mission design process in order to be effectively faced. Indeed, provision of good quality EO data, calibrated and validated according to the latest technologies and internationally recognised methodologies, and harmonised between the several hyperspectral EO satellite missions, is a key element for ensuring the success of the mission. Now even more, with hyperspectral data acquired in parallel by several satellites and several Agencies, EO data quality becomes a key element for ensuring successful synergistic usage of the large amount of hyperspectral data available. At the same time, it ensures users’ confidence that the stated mission requirements and performance (both geometric, radiometric and spectral, and for the several EO product levels) are met and satisfied for the entire mission lifetime. Therefore, constant international coordination and cooperation between Space Agencies in the field of EO data quality and hyperspectral data calibration and validation (Cal/Val) methodologies and challenges becomes crucial. In this context, the proposed presentation aims at giving an overview of the pre-Flight and in-Flight Cal/Val technologies and methodologies currently considered and under development for the CHIME mission.

The Copernicus Hyperspectral Imaging Mission for the Environment (CHIME) is one of the High-Priority Candidate Missions (HPCM) endorsed by ESA for the expansion of the Copernicus Sentinel missions. CHIME will provide routine hyperspectral sampling of Earth surface reflectance over the solar spectral range (400-2500 nm) at a 30 m spatial resolution with a revisit of 22(11) days with one (two) satellite(s). CHIME observations will support EU- and related policies for the management of natural resources and assets providing a major contribution in the domains of raw materials and sustainable agricultural management with a focus on soil properties, sustainable raw materials development and agricultural services, including food security and biodiversity.Currently under phase B2 the development of the CHIME mission is performed by a consortium led by Thalès Alenia Space in France (as prime contractor) and OHB System AG in Germany (for the instrument). The Observation Performance Simulator (OPSI), is a software tool being developed by ACRI-ST under the management of the above partners as ATBD providers, to support the development and verification of the space segment as well as the development of the ground segment.The OPSI is devoted to simulate the instrument acquisition and its different acquisition modes (along with the platform behaviour), to prototype the corroborating ground segment processors which calibrate the payload measurements to TOA radiance (at L1b) and orthorectified TOA reflectance (at L1c) and to assess the instrument performance by comparing true and estimated parameters generated at different stages. In order to accomplish the above objectives OPSI is composed of an Instrument Performance Simulator (IPS), a Ground Processor Prototype (GPP) module and a Performance Assessment Module (PAM).This presentation is dedicated to the status of the OPSI Software System at its Preliminary Design Review, at which it offers to the leading consortium and to ESA a software tool able to simulate the radiometric aspects as well as to provide first mission performance figures to be expected from the instrument design.

.

PRISMA (PRecursore IperSpettrale della Missione Applicativa) is a demonstrative spaceborne mission, fully deployed by the Italian Space Agency (ASI). To support the calibration/validation activities of the PRISMA hyperspectral mission, ASI and the National Research Council (CNR) started in 2019 the PRISCAV project (Scientific CAL/VAL of PRISMA mission). The objective of PRISCAV is the comprehensive characterization of the performances of the PRISMA payload in orbit over different land-use types and the verification of the durability in time of the performances. Within this framework, PRISCAV created a network of instrumented sites (12) showing different land-use and surface settings (Snow; Sea; Inland and Coastal Water; Forest and Cropland) to perform independent and traceable in-situ and airborne Fiducial Reference Measurements (FRM) simultaneous to PRISMA acquisitions to assess the required performance of sensor, data products, and processors at the different levels. Ground-based and airborne campaigns were specifically designed to evaluate the performances of sensor radiometry and processors, and to validate Top-of-Atmosphere Level 1 Radiances and Bottom-of-Atmosphere Level 2 Reflectance standard products. To date, over 300 PRISMA acquisitions were collected over the target sites. Ground teams ensured a simultaneous land-use classification and an appropriate atmospheric characterization. This enabled a multiscale spectral matching with ground targets and the assessment of key parameters related to the spectral, spatial and radiometric performances of PRISMA over the mission duration till now, as well as their evolution with the different versions of the processors. The match up of in-situ and satellite-borne reflectances over representative land-use are presented herein. Results so far are highly promising, in line with the mission requirements, and confirm the potential of the PRISMA mission for the development of innovative products and new applications in the field of environmental monitoring and earth observation in general.

Hyperspectral remote sensing reflectance (Rrs) derived from PRISMA in the visible and infrared range was evaluated for two inland and coastal water sites using above-water in situ reflectance measurements from autonomous hyper- and multispectral radiometer systems. We compared the Level 2D (L2D) surface reflectance, a standard product distributed by the Italian Space Agency (ASI), as well as outputs from ACOLITE/DSF, now adapted for processing of PRISMA imagery. Near-coincident Sentinel-3 OLCI (S3/OLCI) observations were also compared as it is a frequent data source for inland and coastal water remote sensing applications, with a strong calibration and validation record. In situ measurements from two optically diverse sites in Italy, equipped with fixed autonomous hyperspectral radiometer systems, were used: the REmote Sensing for Trasimeno lake Observatory (RESTO), positioned in a shallow and turbid lake in Central Italy, and the Acqua Alta Oceanographic Tower (AAOT), located 15 km offshore from the lagoon of Venice in the Adriatic Sea, which is characterised by clear to moderately turbid waters. 20 PRISMA images were available for the match-up analysis across both sites. Good performance of L2D was found for RESTO, with the lowest relative (Mean Absolute Percentage Difference, MAPD < 25%) and absolute errors (Bias < 0.002) in the bands between 500 and 680 nm, with similar performance for ACOLITE. The lowest median and interquartile ranges of spectral angle (SA < 8◦) denoted a more similar shape to the RESTO in situ data, indicating pigment absorption retrievals should be possible. ACOLITE showed better statistical performance at AAOT compared to L2D, providing R2 > 0.5, Bias < 0.0015 and MAPD < 35%, in the range between 470 and 580 nm, i.e. in the spectral range with highest reflectances. The addition of a SWIR based sun-glint correction to the default atmospheric correction implemented in ACOLITE further improved performance at AAOT, with lower uncertainties and closer spectral similarity to the in situ measure- ments, suggesting that ACOLITE with glint correction was able to best reproduce the spectral shape of in situ data at AAOT. We found good results for PRISMA Rrs retrieval in our study sites, and hence demonstrated the use of PRISMA for aquatic ecosystem mapping. Further studies are needed to analyse performance in other water bodies, over a wider range of optical properties. Braga, F.; Fabbretto, A.; Vanhellemont, Q.; Bresciani, M.; Giardino, C.; Scarpa, G. M.; Manfè, G.; Concha, J. A.; Brando, V. E. 2022 ISPRS-P&RS https://doi.org/10.1016/j.isprsjprs.2022.08.009.

During the past 4 years, the PACO atmospheric correction program has been successfully implemented in the L2a processor in the ground segment of DESIS and EnMAP hyperspectral missions. The "DLR Earth Sensing Imaging Spectrometer" (DESIS) is a VNIR sensor mounted on-board of the International Space Station (ISS) and it has been operational since October 2019. DESIS acquires images of the Earth at user request with a swath of about 30 km and 235 bands with a Full Width at Half Maximum (FWHM) of 3.5 nm in the spectral range from 400 to 1000 nm. The recently launched EnMAP (Environmental Mapping and Analysis Program) is a German hyperspectral satellite mission that monitors and characterises Earth’s environment in the spectral range from 420 – 2450 nm with a FWHM of 6-10 nm in the VNIR and 7.5-11.5 in the SWIR. The off-nadir pointing capability (up to 30 degrees) allows 5000 km to be monitored per day with a swath width of 30 km at a spatial resolution of 30 m. The Ground Segment L2A processor for DESIS and EnMAP missions corrects the at-sensor received terrestrial reflection of the incident solar radiation for the effects of atmospheric constituents. It processes ortho-rectified Top-Of-Atmosphere (TOA) radiance scenes and generates the Bottom-Of-Atmosphere (BOA) ground reflectance spectral image cube, along with pixel-classification masks, Aerosol Optical Thickness (AOT at 550 nm) and Water Vapor (WV) maps. The same atmospheric correction program is used internally at EOC for processing multi-spectral data like Sentinel-2 and Landsat-8. This multi-sensor support minimizes the differences between the multi-spectral and hyperspectral ground surface reflectance. In this contribution we summarize the lessons learned on the validation of the uncertainty of the atmospheric correction, the hyperspectral L2A products and their consistency with the multi-spectral results.

The hyperspectral (HSR) sensors Earth Surface Mineral Dust Source Investigation (EMIT( of NASA and The Environmental Mapping and Analysis Program (EnMAP) of DLR were recently launched. These state-of-the-art sensors have joined the already operational HSR sensors DESIS (DLR), PRISMA (ASI), and HISUI (JAXA). The launching of more HSR sensors is being planned for the near future (e.g., SBG of NASA, CHIME of ESA and PRISMA-SG of ASI), and the challenge of monitoring and maintaining their calibration accuracy is becoming more relevant. We proposed two test sites (Amiaz Plain (AP) and Makhtesh Ramon (MR)) as CAL/VAL sites for spectral, radiometric, thematic and geometric quality checks. The sites are situated in the arid environment of southern Israel and are in the same coverage overpass. Both test sites have already demonstrated favorable results in assessing HSR sensors' performance and were chosen to participate in EMIT and EnMAP for a validation process. We first evaluated the feasibility of using AP and MR as CAL/VAL test sites with extensive datasets and sensors, such as the multispectral sensor Landsat, the airborne HSR sensor AisaFENIXK1, and the spaceborne HSRs DESIS and PRISMA. Field measurements were taken over time following solid protocols. The suggested methodology integrates reflectance and radiometric calibration test sites into one operational protocol. The method can highlight degradation in the spectral domain early on, help maintain quantitative applications, adjust the sensor's radiometric calibration in its lifetime mission, and minimize uncertainties of calibration parameters and evaluate the accuracy of the atmospheric correctors. A PRISMA sensor case study shows the complete operational protocol, i.e., performance evaluation, quality assessment, and cross-calibration between HSR sensors. These CAL/VAL sites are ready to serve as operational sites for other HSR sensors.

The CHIME-SBG Airborne Campaign carried out in summer 2021 has given the opportunity to develop operational procedures that deal with the CAL/VAL of spacebased hyperspectral imaging spectrometers. While some planning approaches have work satisfactorily, the mission has also shown the current limitations and upcoming challenges when striving for operational airborne CAL/VAL services. These services will need to be in place by the time the CHIME and SBG missions enter their commissioning phase. We have successfully demonstrated the synchronization of CAL/VAL mission constraints including weather, matchup calculations based on TLE propagation and target temporal constraints. An automated CAL/VAL procedure to compare airborne imagery to spectral ground control points was also implemented. Areas that need improvements include: a) closing the temporal gap between satellite overpass and airborne observation, b) spatial sampling considerations including the instruments point spread functions, c) BRDF models to compensate observation and sun angle differences, d) uncertainty budgets and propagations for all involved sensors and models. The above topic may also be seen under the light of the planned ESA TRUTHS mission, which aims at providing in-orbit cross CAL/VAL of other sensors in the solar reflected optical domain of the electromagnetic spectrum. The efforts within the TRUTHS mission preparation should thus be aligned with the CHIME and SBG efforts on CAL/VAL.

.

Satellite hyperspectral data provides high spectral resolution information about objects on the Earth, which is very useful for target detection, environmental monitoring, land resources exploration, disaster monitoring, precision agriculture, forestry surveying, and urban planning. In recent years, four Chinese hyperspectral missions have been launched since 2018, i.e., GaoFen-5 01(GF5-01), ZiYuan-1E (ZY1E), GaoFen-5 02 (GF5-02), and ZiYuan-1F (ZY1F). Since the launch of GaoFen-5 (GF5) satellite in 2018, Cal/Val activities for the Advanced Hyperspectral Imagers (AHSI) onboard GF5-02 and ZiYuan-1E (ZY1E) have been carried out to guarantee the required hyperspectral data quality. So, the calibration and performance assessment of AHSI sensor for GF5-02 and ZY1E will be presented in this workshop. The statue, basic instrument design, and the hyperspectral preprocessing system will be also introduced. A plenty of work have been done in ground pre-processing to generate the standard Level 1 product. It mainly comprises data format and auxiliary analysis, radiometric correction (including dark signal deduction, relative radiometric correction and bad pixel repairing) and geometric processing (including sensor splicing and band registration). Then, to radiometrically calibrate the AHSI sensor, synchronous ground measurement has been carried out at Dunhuang site in China during the on-orbit test. The surface reflectance as well as atmospheric parameters was measured in the field. In addition, the Baotou calibration and validation test site in China also provides operational high-accuracy and high-stability vicarious radiometric calibration condition. The Cal&Val of the AHSI sensor has been done over Dunhuang and Baotou sites, which will be presented in this presentation. For further evaluating the data quality of AHSI sensors, many other instrument parameters have also been tested during on-orbit test, such as, radiometric performance indicators (including signal to noise ratio, relative radiometric calibration accuracy and dynamic range), geometric performance indicators (including band registration accuracy and positioning accuracy), and spectral performance indicator (including spectral resolution, spectral accuracy, and smile effect, et al.). The results of performance assessment show the good and stable on-orbit radiometric, spectral, and geometric performance of AHSI sensor, and the imaging data is reliable enough to be applied in quantitative applications. Some quantitative application tests have also been carried out, such as the validation of surface reflectance, vegetation / soil / water quality products et al. Some evaluation cases will also be shown in this presentation.

Historically, imaging spectroscopy missions have depended on airborne instrumentation, which limits observation in the spatial and temporal domains, yet soon there will be numerous space-borne spectroscopy missions. To provide end-to-end support for the traceability of these space-borne missions, we have coupled the LPJ-wsl dynamic global vegetation model (DGVM) with the canopy radiative transfer model, PRO4SAIL, to generate globally gridded spectra across the visible to shortwave infrared (VSWIR) range (400-2500 nm) at a daily time-step. LPJ-wsl variables are modified to estimate required PROSAIL parameters, which include leaf structure, Chlorophyll a+b, brown pigment, equivalent water thickness, and dry matter content. Using reflectance data from canopy imagers mounted on towers and air and spaceborne platforms, we have compared simulated spectra with a boreal forest site, a temperate forest, managed grassland, and a tropical forest site. We also compare simulated normalized differenced vegetative index (NDVI) with the commonly used NASA MODIS NDVI product. We find that canopy nitrogen and leaf-area index are the most uncertain variables in translating LPJ-wsl to PROSAIL parameters, but at first order, LPJ-PROSAIL successfully simulates surface reflectance dynamics. Future work will optimize functional relationships required for improving PROSAIL parameters and the continued development of LPJ-PROSAIL to represent improvements in leaf area index, leaf water content, and canopy nitrogen. The LPJ-PROSAIL model can support missions such as NASA’s Surface Biology and Geology (SBG) and higher-level modeled products.

Availability of hyperspectral Earth Observation (EO) data is internationally recognised as key to address a number of the most important scientific and environmental challenges to monitor the status and understand the dynamics of our planet, and in parallel to support policy making. This need is broadly shared and several space agencies have already launched their hyperspectral EO satellites (e.g. the Italian Space Agency’s PRISMA mission, the DLR’s DESIS and EnMAP missions, etc.) and some others are in the process of designing and developing theirs (e.g. the NASA’s SBG mission and the ESA developed CHIME mission), targeting launch before 2030. The number of applications and activities that will benefit from the wide availability of hyperspectral EO data is remarkable and goes from (but not limited to) agriculture to food security; from raw material and soil monitoring to biodiversity and environment protection; from inland/coastal waters monitoring to forestry. However, broad user access to large amounts of hyperspectral EO data acquired with a multi-satellite and multi-Agency approach comes also with challenges that have to be tackled at early stages of the mission design process in order to be effectively faced. Indeed, provision of good quality EO data, calibrated and validated according to the latest technologies and internationally recognised methodologies, and harmonised between the several hyperspectral EO satellite missions, is a key element for ensuring the success of the mission. Now even more, with hyperspectral data acquired in parallel by several satellites and several Agencies, EO data quality becomes a key element for ensuring successful synergistic usage of the large amount of hyperspectral data available. At the same time, it ensures users’ confidence that the stated mission requirements and performance (both geometric, radiometric and spectral, and for the several EO product levels) are met and satisfied for the entire mission lifetime. Therefore, constant international coordination and cooperation between Space Agencies in the field of EO data quality and hyperspectral data calibration and validation (Cal/Val) methodologies and challenges becomes crucial. In this context, the proposed presentation aims at giving an overview of some of the Cal/Val technologies and methodologies currently considered and under development in the ESA Ground Segment for the CHIME mission.

Operational Calibration and Validation (Cal/Val) is required to ensure the quality of and build confidence in Copernicus data. However, current Cal/Val activities are limited and insufficiently harmonized between different missions. The objective of the Copernicus H2020 Cal/Val Solution (CCVS) project is to define a holistic solution for all Copernicus Sentinel missions to overcome current limitations both for current and upcoming Sentinel-missions. This includes improved calibration of currently operational or planned Copernicus Sentinel sensors and the validation of Copernicus core products generated by the payload ground segment. CCVS started with an overview of existing calibration and validation sources and means, identified gaps in the current cal/val practise and is proposing long-term solutions to address the currently existing constraints in the Cal/Val domain. An objective is also to exploit existing synergies between the missions. The analysis performed within the CCVS project is based on experience from many experts in the Cal/Val domain and on feedback from different working groups gathering European Space Agencies, Copernicus Services, measurement networks and International partners. This presentation will give an overall assessment of Copernicus Cal/Val maturity in the optical mission component both for sensor calibration and characterization and for product quality. Required developments in terms of technologies and instrumentation, Cal/Val methods, instrumented sites and dissemination service are addressed. One of our findings is the need for ground-based hyperspectral reference measurements in particular to prepare the validation of CHIME.

.

.

The SBG Space-based Imaging Spectroscopy and Thermal pathfindER, or “SISTER”, is prototyping algorithms and workflows for a select set of proposed SBG products across all classes of algorithms, including terrestrial ecosystems, aquatic/coastal ecosystems, hydrology (snow and ice) and geology. These algorithms are being implemented in a modular workflow that allows flexibility to update algorithms as they are refined. As well, the approach utilizes data processing rules that translate across products, but also can be customized to the needs of specific products. Finally, our prototyping activities are being developed within an emerging on-demand processing framework that may be necessary, as it is possible that not all products will be able to be produced for all imagery collected. The on-demand approach may enable the production of alternative versions of standard products based on user needs. In this talk, we illustrate some of the prototype products developed by SISTER using PRISMA and DESIS imagery.

The PRISMA SCIENZA programme supports and promote the scientific use of hyperspectral data, acquired from the PRISMA satellite, by the Italian User Communities (Universities, public research bodies and industries). The main objective is to support the full data exploitation of the PRISMA mission and strategically promote the development of Italian know-how in the hyperspectral remote sensing sector.

Vegetation products are core components of proposed data suites for both SBG and CHIME. Proposed products for SBG include vegetation traits (e.g., foliar nitrogen, leaf mass per area, chlorophyll, canopy water content) as well as fractional area of green vegetation, non-photosynthetic vegetation (NPV) and substrate. In addition, there is considerable demand from the applications community for vegetation composition classification. Similarly, CHIME high priority prototype products include leaf and canopy nitrogen, water and pigment content, leaf mass per area and leaf area index (LAI). There is substantial overlap among products for the two missions, which offers opportunities for synergistic development of these data products, especially when it comes to calibration and validation as well as algorithm development and testing. Differences in planned approaches between CHIME and SBG for vegetation products relate in part to underlying thematic motivations, e.g. agriculture and food security for CHIME and biodiversity and ecosystem function for SBG. As such, it is likely that some algorithmic approaches will differ between missions. Here, I briefly discuss: 1) the suite of algorithmic approaches that are under consideration for vegetation products for the two missions, 2) the challenges that both missions will face in calibration and validation, and 3) additional considerations related to product definition and units of measurements that have consequences for product harmonization. Nevertheless, given that we are in the early stages of algorithms prototyping for both CHIME and SBG, it makes sense for the two missions to collaborate to the extent possible now on inter-comparisons, as it is likely that non-expert user communities will be interested in using products from both missions and will want clearcut information on how the products compare.

The current Copernicus mission, and especially the high-resolution multispectral data from Sentinel-2 is already being used for agricultural and food security applications in an operational context. These services span from Smart Farming services showing qualitative differences within fields to large-scale indicators of agricultural production, that support predicting food security on country and continental scale. But multispectral data are limited in that they don’t show individual spectral absorption bands that characterize the plant bio-chemical parameters. Those are necessary to go from observation of crops to a deeper understanding of causes, and from qualitative description to quantitative assessment. Coping with climate change is one of the greatest challenges of the 21st century. As a major economic sector and CO2 emitter, agriculture is a major polluter and thus part of the problem. On the other hand, agriculture can make a significant contribution to reducing CO2 levels in the atmosphere by binding CO2 through special cultivation methods and storing it in the soil through targeted humus build-up. This potential has been recognized and there are numerous initiatives and business models worldwide, e.g. climate farming, Carbon Farming, regenerative agriculture, climate certificates or sponsorships. In this presentation, the possibilities of improved derivation of relevant surface properties (topsoil humus content, nutrient uptake of vegetation over the vegetation period, irrigation support, tillage and crop residues) using the analysis of hyperspectral Earth Observation data will be illustrated. In will be analysed which information products could be derived from the hyperspectral satellite data that have not yet been made available, or have not been made available in the same quality and automation capability before, and that can be used for Smart Carbon Farming. For the first time, satellite missions such as EnMAP or PRISMA are making available hyperspectral EO data that, thanks to their high spectral resolution, allow the analysis of both narrow and shallow absorptions, even in the SWIR range. This in turn allows the derivation of complex agricultural variables. Recent results suggest e.g. that hyperspectral data can be used to measure canopy nitrogen content, or the amount of non-photosynthetically active vegetation, which supports the mapping of the carbon cycle in simulation systems. It is important to note that these measurements are only possible with hyperspectral data on a physical basis, since the relevant absorption regions in the SWIR overlap each other and therefore have to be separated with high spectral resolution.

Non-photosynthetic vegetation (NPV) biomass has been identified as a priority product for the upcoming Hyperspectral Imaging Mission for the Environment (CHIME). Being an essential part of the total organic carbon within the biosphere, the spatiotemporal knowledge of non-photosynthetic (lignocellulosic) plant material within the system of canopy and soil helps to understand carbon fluxes or drought effects. Estimation of NPV is crucial for all terrestrial ecosystems, but especially for agriculture where the presence of crop residues (CR) plays an important role in tillage management. Indeed, NPV and CR are important for soil health, erosion control, and attempts to increase soil carbon stocks for climate change mitigation. While the fractional coverage of CR has been studied extensively, only little attention has been paid to the quantification of biomass of nonphotosynthetic plant material. In this survey, we summarize past attempts to quantify NPV or CR biomass from Earth observation data with a particular focus on hyperspectral data exploitation. In addition, we propose to implement efficient workflows toward the quantification of NPV within operational delivery of next-generation global products. Recently, significant advances in physically-based radiative transfer models (RTM) have been achieved with the inclusion of the carbon-based constituents (CBC) in the leaf optical properties model PROSPECT-PRO. The spectral signal can be upscaled to the canopy level through the 4SAIL radiative transfer model. Based on this, we propose the use of a “hybrid strategy” for the retrieval of NPV cropland biomass from spaceborne hyperspectral missions. More precisely, we pursue Gaussian process regression (GPR) algorithms due to their appealing intrinsic property to provide uncertainty information along with estimates, which is outstanding compared to alternative ML methods. First NPVbiomass-GPR models were successfully validated, and mapping capability was demonstrated on PRISMA hyperspectral images acquired over agricultural areas in Germany. A hybrid workflow for mapping cropland NPV is currently under investigation within the framework of the CHIME mission. With the launched PRISMA and EnMAP missions, we expect an exciting opportunity to further test, improve and optimize the retrieval strategy, also taking into account more advanced RTMs and/or alternative retrieval strategies. In the near future, the synergy between next-generation (Sentinel) sensors and spaceborne imaging spectrometers should be explored through joint analysis of multi-mission acquisitions along with in-situ data collection to allow exhaustive model validation activities. In this way, it ought to be possible to achieve accurate quantification of nonphotosynthetic cropland biomass as a next-generation routine product from space.

In preparation of new-generation imaging spectrometer missions and the accompanying unprecedented inflow of hyperspectral data, optimised models are needed to process routinely hyperspectral data into relevant surface attributes, such as vegetation traits. Hybrid models, combining the generality from radiative transfer models (RTM), and the flexibility of machine learning algorithms are preferred. Hence, a critical review of the existing models as developed within CHIME’s E2E L2b Vegetation Retrieval Module is required. To do so, we evaluated the earlier developed hybrid models based on Gaussian process regression (GPR) against alternative machine learning regression algorithms (MLRAs) and for different noise scenarios. It was attempted to assess the robustness of the developed models, as well whether alternative MLRAs (quantile random forests: QRF, neural networks, kernel ridge regression) would lead to superior hybrid retrieval models. At the same time, for each of the analysed MLRAs associated uncertainties are provided, e.g. through bootstrapping techniques. The uncertainties allow us to assess the robustness of the algorithms, e.g. how the model responds to added noise, but also how the model behaves to different images in space and time. The systematic analysis for leaf and canopy variables led to the following main findings: GPR was evaluated as the best performing algorithm for most of the variables in the absence of noise. It justifies the selection of GPR as the core algorithm in the hybrid model development. Yet, given that GPR was not always top-performing, it implies that other MLRAs deserve to be considered as backup. In case the spectral data becomes irregular and noisy, GPR stays top performing with low added noise levels (e.g. up to 5%). However, with increasing added noise, i.e. 10% or higher, GPR degraded for the majority of the variables. Instead, QRF emerged as a most robust alternative for most of the variables.         The uncertainty maps provide insights into the robustness of the models. For instance, GPR uncertainties worsen with increasing noise, while QRF uncertainties stay largely invariant with increasing noise levels.    Summarising, the conducted analysis suggests that GPR stands as a benchmark algorithm in hybrid models development. GPR typically performs superior in ideal situations with low noise levels. However, it is wise to consider backup models, especially in case data appears to be noisy, e.g. due to imperfect atmospheric correction. In such a situation, QRF emerged as the most appealing alternative algorithm.

Climate change is placing unprecedent pressures on agricultural grasslands. Monitoring the impacts of these pressures, e.g. on biomass production, is necessary for national reporting and for designing climate change adapted management strategies. This is particularly relevant for meeting a number of international policy targets, such as those defined by the Common Agricultural Policy (CAP) or Sustainable Development Goals (SDG). Fine-scale and dense-temporal multispectral time series provided by the Sentinel and Landsat programs are cornerstones for quantifying SDG- or CAP-relevant indicators. Recent scientific imaging spectroscopy missions (e.g. PRISMA, EnMAP) complement these data with high spectral resolution imagery. Findings from these precursor missions provide insights into how future operational hyperspectral (e.g. CHIME, SBG) and superspectral (e.g. next generation Sentinel, Landsat) missions will advance current applications regarding improved process understanding. Understanding feedbacks between climate change and grassland vitality is specifically relevant for Central Europe, where grasslands have been heavily impacted by droughts. We here present a grassland drought monitoring framework that was developed with Sentinel-2-based fractional cover time series of green vegetation, non-photosynthetic vegetation (NPV) and soil. Through analyzing PRISMA data, we further elaborate on how future hyperspectral or superspectral missions will advance such frameworks. While the quantification of critical parameters will be improved, novel possibilities for retrieving grassland traits directly related to biomass quantity or quality will be discussed. We find that hyperspectral data is imperative to improve NPV-soil separation. This directly translates into grassland trait quantification, e.g. by avoiding over- or underestimation of estimates when spectral differences are subtle. However, we also find that temporal resolution is core to tackle dynamic processes in grassland regimes, such as swift drought impacts. Where management is intense, different triggers of changing grassland vitality can only be captured with temporally dense data. It is hence likely that both future hyperspectral and superspectral missions will play a crucial role in better characterizing different grassland traits globally.

Canopy nitrogen (N) content has been identified as one of the CHIME High Priority Products. However, there are still open aspects and questions to be discussed. First, a clear definition of remotely sensed N quantities is missing. Typically, nitrogen comes as leaf N content (mass of N per leaf area) or canopy N content (CNC, mass of N per ground area) or N concentrations (mass of N per mass dry matter). However, some authors use the term “N uptake” for N content or N concentration, which may not be in line with the definition of physiologists, as N uptake refers to a dynamic process with assimilation and N translocation. Second, different methods are available for estimating plant N from VNIR/SWIR hyperspectral satellite systems: Direct measurement of leaf or canopy nitrogen is not possible and thus a suite of indirect approaches have been developed: Indirect N estimation through N-Cab (total chlorophyll) link, indirect N estimation through LAI-CNC link, direct empirical N estimation or direct physically-based via protein absorption features. Hereby, physically based and hybrid approaches using the PROSPECT-PRO model appear as the most promising methods. All methods have their advantages and limitations, which need to be discussed. Third, for efficient N monitoring, remote sensing time series are required, which can mainly be provided by multispectral systems. However, only hyperspectral bands are able to provide the required information for estimating (protein-related) vegetation N content. Hence, the retrieval accuracy when using multispectral information is expected to be much lower. Thus we identified the need to robustly quantify the decrease in N estimation accuracy from frequently available Sentinel-2 compared to recent spaceborne hyperspectral systems. Fourth, are satellite-based N products provided at the right time to support agronomists? Decreasing the environmental effect of N fertilization in intensive agrosystems requires matching N inputs to crop needs at the right physiological stage. To handle issues related to revisiting frequency and cloudiness, data from CHIME and SBG (and possibly multispectral systems like Sentinel-2) should be combined. Finally, would the N product alone be sufficient for agricultural management? It is essential to know the exact fertilization timing and required rates to avoid plant stress and overfertilization. Ideally, a simulation of spectral-temporal signatures across the optical spectrum (updated with hyperspectral observations) combining crop growth and canopy reflectance models would be required to integrate all sensor data and derive important physiological knowledge.

Thorough assessments of ecosystem stress responses to environmental change must consider the cascade of vegetation adaption mechanisms acting at different temporal and spatial scales. Sensitive and comparable observations across scales from the leaf and plant scale, the canopy or agricultural field scale, towards whole ecosystems and the globe are consequently needed [Damm et al 2018]. Besides measurements of multiple parameters related to plant biochemistry, canopy structure, plant physiology, leaf temperature and plant morphology, measurements of abiotic environmental drivers are needed to quantify vegetation processes (e.g. photosynthesis and related gas exchange) and understand the many factors that may cause ecosystem stress and limit the potential plant functioning. No single satellite mission is able to provide all the necessary plant, soil and atmosphere parameters at the scales needed, whereas multi-mission observations at different spatial and temporal scales can provide a deeper understanding of the feedbacks and interactions between ecosystems and environmental change. Remote sensing measurements that cover optical (multi- and hyper-spectral), thermal, and microwave domains can provide many of the information at relevant scales needed to monitor ecosystem stress. Implementing and evaluating remote sensing based assessment schemes of ecosystem stress additionally requires well characterized validation sites that allow the measurement of multiple vegetation parameters and abiotic environmental factors, and can support the Cal/Val needs of current and upcoming satellite missions [Niro et al 2021]. This contribution aims to outline the development of thematic ecosystem stress validation sites in Switzerland, that can help tease apart the processes driving ecosystem stress at the necessary temporal and spatial scales. We show the extensive field and airborne campaigns conducted at these sites, and how these data can be used to understand and monitor water resources and energy budgets, to identify key challenges in remote sensing of ecosystem stress, to validate remote sensing products, and to develop state-of-the-art reconstructions for modelling and remote sensing signal understanding. Current observational gaps for parameters related to ecosystem stress will also be identified and future plans for extending the instrumentation and data catalogue addressed. References A. Damm et al (2018) “Remote sensing of plant-water relations: An overview and future perspectives”, Journal of Plant Physiology, 227 F. Niro et al (2021) “European Space Agency (ESA) Calibration/Validation Strategy for Optical Land-Imaging Satellites and Pathway towards Interoperability”, Remote Sensing, 13

.

Finding an appropriate satellite image as simultaneous as possible with the sampling time campaigns is challenging. Fusion can be considered as a method of integrating images and obtaining more pixels with higher spatial , spectral and temporal resolutions. This paper investigated the impact of Landsat 8-OLI and Sentinel-2A data fusion on prediction of several toxic elements at a mine waste dump. The 30 m spatial resolution Landsat 8-OLI bands were fused with the 10 m Sentinel-2A bands using various fusion techniques namely hue-saturation-value (HSV), Brovey, principal component analysis (PCA), Gram-Schmidt (GS), wavelet, and area-to-point regression kriging (ATPRK). ATPRK was the best method preserving both spectral and spatial features of Landsat 8-OLI and Sentinel-2A after fusion. Furthermore, the partial least square regression (PLSR) model developed on genetic algorithm (GA)-selected laboratory visible-near infrared-shortwave infrared (VNIR-SWIR) spectra yielded more accurate prediction results compared to the PLSR model calibrated on the entire spectra. It was hence, applied to both individual sensors and their ATPRK-fused image. In case of the individual sensors, except for As, Sentinel-2A provided more robust prediction models than Landsat 8-OLI. However, the best performances were obtained using the fused images, highlighting the potential of data fusion to enhance the toxic elements' prediction models.

A critical hydrological soil attribute is the water infiltration rate (WIR) to the soil profile due to its effect on the runoff regime, leaching, soil erosion, and water availability for both plants and groundwater. Whereas the topography is a regional parameter that affects the WIR mainly due to the slope and landscape conditions, the WIR is highly affected by soil moisture, organic matter, soil mineralogy, soil texture, and soil sealing. Amongst these factors, the WIR is highly affected by the interface between the atmosphere and the pedosphere that is controlled by the soil seal condition that is evolving from raindrop energy, from biological substances (known as biogenic crust), and from mineral alteration (known as fire-driven crust). Negative effects (low WIR) are due to hydrophobic organic substances and fine particles of soil minerals that transport to the soil surface. Positive effects (high WIR) are due to stabilization of the soil aggregation processes that increase the water permeability to the soil and are a result of the interaction between cementation agents at the soil (such as, soil texture, and minerals). Assessing the WIR level in advance and under high and wide spatial coverage may help farmers to reduce soil water runoff and soil erosion process by applying wise agrotechnical practices (e.g. no-till). In this study, we first apply a proof-of-concept activity to demonstrate that the WIR can be estimated via proximal spectral sensing on a pixel-by-pixel basis using airborne hyperspectral data. This first exercise used unmanned aerial vehicles) across the VIS-NIR region with a hyperspectral sensor (Cubert). The results showed that the WIR can be mapped in a clayey agriculture field in Southern Italy with very reasonable accuracy (R2=0.76). The second experiment of this study was to apply a similar methodology on sandy soils exploiting the entire VIS-NIR-SWIR spectral region data acquired from a hyperspectral airborne sensor (Hyspex) in Northern Greece. The WIR maps that were generated from both exercises in several agriculture fields from Italy and Greece were validated on the ground level yielding reasonable results (R2=0.59). Recently the spectral based model was applied onto PRISMA data southern Israel and demonstrated how WIR maps can estimate erosion risks over large area. The success to exploit the hyperspectral technology for soil practices as shown in this study goes beyond the traditional soil attributes assessment from reflectance spectroscopy, paving the way for more innovative ideas in soil remote sensing practices based on spectroscopy.

Advancing user and institutional skills to use Earth Observation (EO) data does not only involve the use of current EO constellations, sensors, data, facilities & platforms, products, services; but also, their improvement, reshaping and development. The new generation of satellite sensors supports more comprehensive identification of key monitoring Essential Variables (EVs) and gives insight into the environmental system discontinuities and emergent properties. This study is part of the CHIME Requirements Consolidation Study (RCS) and PRISMAScienza OVERSEE project and the results have been strengthened with the collaboration of SBG (NASA/JPL), PRISMA Second Generation (ASI) and CHIME (ESA) research teams. In addition, the relevant study results are feeding the CHEES ESA project aimed at developing an end-to-end CHIME performance simulator that requires also the testing of existing algorithms to derive L2B products. In particular, hyperspectral sensors, like CHIME and PRISMA, show a new opportunity to exploit the existing algorithms, improving image-based detection of land cover types, in order to enhance product service portfolio. The results are obtained applying sub-pixel classification techniques and the added value offered by higher spectral resolution of hyperspectral sensors. The spatial variability of topsoil texture is retrieved using the linear spectral mixture analysis (LSMA), an image-based algorithm that breaks down the hyperspectral information into fractional cover abundance within each pixel. The membership degree of each cover type to each of the surface’s physical reflectances is assessed via correlation models based on the spectral dimensionality of the topsoil texture classes. Furthermore, the improvement of the reliability of vegetation related to Essential Climate Variables (ECVs) estimation, such as FCover and FAPAR, is expected, considering that these variables depend on the canopy structure, density and typology.

Soil is a fundamental natural resource and accurate maps of its properties are essential for sustainable management of the land at all scales and by a large range of stakeholders. Digital Soil Mapping (DSM) is an established approach to derive these maps, by linking ground observations with environmental layers describing the soil forming factors. Remote sensing provides useful (pattern) data to describe the soil forming factors. In the past, mainly optical and thermal data have been used. Radar data is currently showing promising results as well. Hyperspectral remote sensing has proven very useful for mapping vegetation traits and land cover. Preliminary results show that mapping of soil properties will be enhanced when using hyperspectral remote sensing, also when considering permanently vegetated soil, because of the more specific mapping of soil forming factors and as a tool for direct estimation of soil properties of the soil surface. This study will provide an overview of DSM performance results using a range of remote sensing data and outline the challenges for the future. This includes likely advantages and disadvantages or suitability of hyperspectral and other imagery for (digital) soil mapping. The availability, quality and (level of) standardisation or harmonisation of reference soil data or ground observations is pivotal to the success of deriving soil property or soil function information from spaceborne imagery. The study will also address the challenges and necessity of sufficient and sufficient quality reference soil data for calibration purposes, data sharing, exchange and serving of results and reference data. For the latter it will describe current developments in soil data repositories and sharing in Europe as developing in the EU Soil Observatory, EJP SOIL and other projects.

Soil organic carbon (SOC) is an indicator for the balance between the input from biomass production and mineralization of organic matter, and hence for the effects of the three pillars of conservation agriculture: i) maximizing crop cover, ii) minimizing soil disturbance and iii) increasing the variation in crop rotations. With the rapid expansion of conservation agriculture, there is need for cost effective methods to monitor its effects and its permanence. With the increasing amount of freely available satellite data, recent studies have focused on stabilizing the soil reflectance by building reflectance composites using time series of images. An exposed soil composite from Sentinel‑2 imagery for southern Belgium and the Northern part of The Netherlands (covering the spring periods of 2016‑2021) showed that spectral indices are efficient to remove soils in unfavorable conditions i.e. wet soils and soils covered by crops and crop residues. The SOC prediction maps show that the uncertainty of prediction decreases when the number of scenes per pixel increases, and reaches a minimum when at least six scenes per pixel are used (mean prediction interval ratio of all pixels is 12.4 g C kg‑1, while mean SOC predicted is 14.1 g C kg‑1). We applied these SOC prediction models to investigate the differences between fields under conservation agriculture in Belgium (n=307) and paired fields under traditional agriculture. The conservation agricultural fields have a higher SOC content (from 7 to 19 % depending on the agricultural region). The majority (90 %) of paired (conservation/traditional) did not overcome the minimum detection limit of the SOC map. However, using bootstrapping we were able to produce a distribution of the SOC predictions for each field and 72 % of the fields showed a higher SOC content under conservation agriculture. Regarding the management practices, there is a strong demand for the use of remote sensing for detecting cover crops and tillage practices. Time series of NDVI analyzed in Google Earth Engine already showed that for 70% of the conservation fields were correctly classified. Although remote sensing offers a clear added value in detecting the length of the cover crop growing season as well as a proxy for the return of biomass to the soil, the distinction between reduced tillage and inversion tillage proves more difficult in Belgium.

Environmental DNA (eDNA) has transformed the way the composition of ecological communities, especially microbial communities, are being used for biodiversity monitoring, by allowing thousands of species to be identified in situ from a sample of only a few grams of soil, water, or plant material. Despite this paradigm shift, the spatial distribution of microbiological biodiversity, and its critical role in ecosystem functioning, remains largely unknown due to sparse in situ observations. Here, we describe a fundamentally different approach to biodiversity monitoring by coupling eDNA community composition from in situ plots with imaging spectroscopy to rapidly predict microbiological diversity across landscapes. We examined whether the diversity of microorganisms in soil could be estimated by plant ecosystem functions derived from advanced remote sensing imagery. We show for the first time that combinations of microbiological families can be mapped using image spectroscopy reflectance when combined with eDNA.

Imaging spectroscopy is increasingly used to characterize soil chemical and physical parameters at different temporal and spatial resolutions based on laboratory, field, and remote sensing data. Soil properties such as soil texture, mineralogy, inorganic and organic carbon content can be estimated using spectroscopic analyses and represent crucial parameters e.g., for assessing soil carbon storage, soil fertility, as well as the soil degradational status. Large-scale mapping of the soil surface makes imaging spectroscopy attractive to soil scientists by reducing time-consuming and costly field campaigns and supporting regular soil monitoring activities. Especially with the current and upcoming next generation of hyperspectral satellite sensors, new capabilities for the production of high-quality maps of soil surface properties at larger spatial scales and with an increased temporal resolution will become available. However, limitations for spectral modelling exist due to considerably spectral influence of dynamic surface condition such as fractional vegetation cover, varying soil moisture and roughness, as well as the general lack of harmonized reference data. This contribution presents several case studies conducted in the frame of the EnMAP pre-operational phase, as well as the ESA Worldsoils project, which demonstrate the potential and limitations of imaging spectroscopy for mapping soil properties and soil degradation. In particular, the presentation covers a) the exploitation of new advances in sensor technology combining the VNIR-SWIR and LWIR spectral range for soil property mapping, b) the mapping and evaluation of the soil degradational status and its impact on plant vitality and grain yields, c) challenges of spectral soil modelling on larger spatial scales due to disturbance effects.

Understanding how soil properties vary between and within agricultural fields allows for more efficient use of resources, improving agronomic and environmental management. The relationship between the soil reflectance spectrum in the optical domain and the topsoil properties have led to the development of promising data-driven or physical-based methods of estimating soil properties. In this contribution, we evaluate the ability of satellite PRISMA images to estimate the soil organic carbon, clay, sand and silt content. To investigate this, a test was carried out using topsoil data collected in Italy at two experimental sites. Firstly, a new laboratory soil library created in this study was resampled to the PRISMA bands and machine learning algorithms applied to estimate four soil properties: soil organic carbon content, sand, silt and clay. Then, bare soil reflectance data were obtained from two experimental areas in Italy, using airborne images and ground-measured spectra sensing a similar area. The airborne and ground data were used to calibrate the estimation models of soil properties employing the machine learning algorithms implemented in the ARTMO toolbox. Finally, the best performing models for each parameter were applied to real PRISMA data. Using laboratory spectra, the Gaussian Process Regression technique provided the best models for the retrieval of the four soil properties with a coefficient of determination ( r2) between measured and estimated values of 0.88 for soil organic carbon, 0.82 for sand, 0.79 for clay and 0.69 for silt. At airborne level, satisfactory results were also achieved. The developed models were applied on independent PRISMA images collected over the experimental sites in 2020 and 2022 to evaluate their robustness and exportability. The results obtained in this study demonstrate the ability of PRISMA images to accurately retrieve topsoil properties from space using machine learning retrieval schemes.

.

Under a climate change and growing population scenario, improving water productivity will be pivotal for tackling freshwater scarcity. Accurate estimations of evapotranspiration (ET) and crop productivity at appropriate spatiotemporal scales would allow optimizing irrigation and provide diagnostic tools for soil conservation practices. This will increase agricultural sustainability hence ensuring farmers’ livelihoods in a changing climate. Improving ET and yield estimations through remote sensing would provide vital information to monitor, mitigate and plan for the effects of prolonged droughts on food production. The retrieval of plant biophysical traits are key inputs needed for accurate estimates of ET and crop productivity. With the new PRISMA mission, we have the opportunity to exploit hyperspectral data acquisitions at 30m to improve biophysical traits estimation. Particularly, these data could prove useful to separate green and non-photosynthetically active (i.e. senescent) vegetation biomass, appreciating that non-photosynthetic vegetation play an important role in energy, water, carbon and nutrient cycling. In this EOAfrica Explorers project we will make use of a thermal-based energy balance model model (TSEB), to evaluate ET with inputs from the scientific ECOSTRESS and PRISMA missions, as an exploration of the capabilities for future operational satellite missions in deriving improved ET and yield/biomass products. This will also allow acquiring a better understanding of water use efficiency of cultivated landscapes. Both intermediate (ET, crop traits) and final products (crop productivity and yield) will be evaluated against available in situ measurements collected outside Africa, in already running long-term experimental sites (Majadas de Tietar FLUXNET site in Spain and other ICOS sites), as well as in African sites in collaboration with African Early Adopters (Ministry of Agriculture of Burkina Faso and the Department of Water and Sanitation of Botswana). Together with CSIR and University of Pretoria of South Africa, we foresee to exploit existing crop yield and irrigation reports that are acquired by those institutions to validate project outputs.

Tree-grass ecosystems (TGEs) roughly occupy 1/6th of the Earth’s surface and play a pivotal role at both local and global scales, providing crucial ecosystem services. Semi-arid grasslands, often present in TGEs, are the most dominant contributor to the global carbon fluxes’ trends and inter-annual variability. Therefore, monitoring the ecosystem functional properties (EFPs) of these highly complex landscapes is of utmost importance to adequately understand the effects of climate change and the subsequent impact on global ecosystem services. Novel remote sensing hyperspectral sensors and techniques provide new opportunities to better assess EFPs, including biodiversity and their relations to water/carbon fluxes at various spatial and temporal scales. Despite their global relevance, Earth observation models and algorithms often misrepresent the unique structural and functional characteristics of TGEs, where high-quality ground data is necessary to evaluate, calibrate and inform space-borne models. This study focused on estimating plant traits and diversity of the herbaceous understory of a Mediterranean TGE located in the experimental station of Majadas de Tiétar, Spain. Long-term vegetation biophysical and spectral measurements (2009-2022) have been collected in this station along with ancillary data from eddy-covariance systems, automated optical and thermal sensors (Phenocam, multispectral and hyperspectral sensors), plus airborne hyperspectral images and vegetation traits and diversity sampling. In this study, we concentrate on the estimation of plant traits (e.g., specific leaf area (SLA), Canopy water content (CWC), leaf area index (LAI), chlorophyll content) and diversity metrics (e.g., Shannon, Evenness, Functional Dispersion (FDis)) using hyperspectral data acquired with an ASD Fieldspec 3 portable spectroradiometer (400-2500 nm). The relationships between plant traits and diversity metrics with spectral reflectance factors were explored by means of vegetation indices featuring different band combinations, along with machine learning (ML) techniques exploiting the entire optical domain. A high correlation (r2 > 0.7) was achieved against LAI, SLA, and CWC, especially for the ML models, whereas it was overall lower for the diversity indices. However, we found that functional diversity metrics, particularly FDis, correlated with key plant traits through a multiple regression model (r2 > 0.5, rRMSE = 20%). These results indicate the potential of hyperspectral data in estimating functional diversity through the retrieval of plant traits significantly related to plant diversity, relevant for the application of new space-borne hyperspectral sensors such as CHIME, PRISMA and EnMAP.

The CHIME & SBG airborne AVIRIS-NG Europe campaign 2021 was the opportunity to collect ground information on various sites across Europe to validate remote sensing products derived from imaging spectroscopy. Two French sites corresponding to Mediterranean vegetation were included during the airborne campaign: Pic Saint Loup and Puechabon. The tolerance and adaptation to recurrent drought of Quercus ilex and Quercus pubescens, two of the main species found in this Mediterranean ecosystem, is particularly important to understand in the context of global warming. During the campaign, vegetation biophysical properties including leaf area index and leaf traits were collected from various individual of Quercus ilex and Quercus pubescens. Optical properties and chemical traits were measured from leaf samples collected from individuals distributed over the two sites covered by airborne acquisitions, and leaf area index was measured with a LAI2000. This field data collection was part of a set of campaigns performed during spring, summer and fall 2021, aiming at monitoring the influence of drought on Mediterranean ecosystems. Additional data including LiDAR acquisitions (with UAV) and field spectroscopy of soil and surfaces were also collected. Leaf spectroscopic measurements measured in the lab were analyzed and leaf traits including pigment content, water content and leaf mass per area were estimated using PROSPECT model inversion. A set of traits were also measured using destructive measurements or alternative optical measurements (SPAD and Dualex). Leaf traits were accurately estimated for both species. At canopy scale, hybrid methods combining physical modeling and machine learning were also applied in order to estimate vegetation traits from both Sentinel-2 time series and airborne imaging spectroscopy. The ground measurements are currently compared to canopy scale estimates, and the complementarity between Sentinel-2 time series, airborne imaging spectroscopy and available spaceborne imaging spectroscopy data, including PRISMA and DESIS, is also investigated for a set of traits. 3D physical modeling based on LiDAR data is in preparation to compare the influence of the level of details used in the modeling stage.

Atmospheric Carbon uptake by plants is a key process feeding the entire earth's biosphere. The idea of measuring the net amount of C exchanged by vegetation (Net Ecosystem Exchange (NEE) = Gross Primary Productivity (GPP) - Ecosystem Respiraton (Reco)) is therefore extremely attractive. Satellite remote sensing has been repeatedly trying to contribute to achieve such a goal at both the global or regional scales by providing more or less indirect estimations of one term of the equation: GPP. Indeed, the fraction of Photosynthetic Active Radiation (PAR) which is absorbed by green surfaces (APAR) can be seen as a good proxy of GPP as it is the amount of Sun Induced Fluorescence (SIF) which is emitted by the vegetation. But while satellite remote sensing is progressing towards the reality objective of measuring gross photosynthesis from space under stady-state conditions of clear skies (see the ESA-EE-8 FLEX, mission), detailed studies at the plant and canopy scales, are revealing that the idea of measuring estimating GPP exclusively during sunny periods only has serious limitations. This presentation will summarize the most recent findings in plant science showing that in many circumstances the behaviour of the photosynthetic apparatus under fluctuating light conditions is constraining GPP more than steady-state photosynthetic rates under periods of constant and high irradiance (clear sky periods). Not by chance new plants with higher photosynthetic rates could be recently created by modifying the speed of the relaxation of photochemical protection mechanisms during transition between illuminated and shade conditions rather than enhancing photosynthetic rates under constant illumination. Indeed, the indirect quantification of GPP from space simply based on clear-sky reflectance or even SIF remains illusory. Nevertheless the conclusions of this presentation will highlight why those measurements are anyhow of importance and will finally contribute to understand part of the complexity of earth's biosphere processes.

In 2019, the NERC Field Spectroscopy Facility (Edinburgh, UK) expanded its capability to include UAV based sensors in addition to its pool of ground based spectroscopic instrumentation, developing the "FSF UAV Suite". With capital investment from UKRI, the suite has been developed to provide a number of dedicated drone and sensor platforms, which – due to the modular nation of the UAVs and sensors used – allow for novel, custom solutions, tailored to the needs of the facility’s user base. The suite consists of a fleet of drones ranging in their payload capacities, several multispectral cameras matched to Sentinel-2 and WorldView-3 spectral bands, thermal imaging sensors, and a 350-2500 nm range UAV mounted hyperspectral imager. The new instrumentation is complemented with our pool of ground based spectroscopic instrumentation, including hyperspectral field spectrometers, reflectance targets, and a pool of autonomous sun photometers measuring atmospheric properties. In addition, in 2022, the facility acquired further capital investment to further expand our optical dark laboratory to enhance the calibration and quality assurance of the UAV sensors within the suite. Expanding on already previously developed radiometric and spectral calibration workflows, the new laboratory equipment allows for accurate determination of sphere uniformity effects (critical when considering radiometric calibration when using radiance spheres), stray light effects, BRDF effects when using relfectance targets, and enhanced spectral calibration routines. In this presentation, we discuss the workflow of the FSF UAV suite with regards to our hyperspectral imager when used in conjunction with aircraft measurements or satellite overpasses for vegetation reflectance measurements, using previous campaigns as example case studies. We will discuss the process for the radiometric and spectral calibration of the sensor, and how this follows similar stages to airborne imager calibrations; the development of pre-flight workflows and the challenges this can present when paired with aircraft measurements or satellite overpasses; data acquisition, and the importance of ground based supplementary measurements, particularly with regards to the local atmospheric conditions; and finally, routines for the processing of acquired hyperspectral imagery, and evaluating this in conjunction with aircraft or satellite acquired data.

Over the last decade Unmanned Aerial Vehicles (UAVs) have become an ubiquitous tool for environmental monitoring, robustly filling the observational void between ground/near-surface and satellite sensing. This is largely due to sensor miniaturisation, which has allowed employment of sensors and multi-sensor arrays that were previously restricted to satellite and airborne platforms. UAV imaging spectroscopy, in particular, provides a promising opportunity to retrieve vegetation traits from individual to regional scales, starting from physiological status at individual plant level up to field-level characterisation of crop traits. Such high spatial level of detail not only allows development of precision agriculture practices, but can also provide a valuable insight into spatial variability across cultivated areas. As such, UAV-derived products can provide more comprehensive ground reference than spatially-limited field surveys. In this contribution we will highlight capabilities of UAV imaging spectroscopy, focusing on crop traits retrieval in perennial ryegrass to support assessment of fodder quality and in seed potatoes related to disease detection. Recommendations for upscaling of crop trait retrieval approaches from UAV to the satellite level and relevance of time-series methods for anomaly detection will be highlighted.

Forest ecosystems provide numerous environmental goods and services, such as conservation of biological diversity, climate regulation and nutrient cycling. Given the global importance of forests and the expected increase in threats, the implementation of sustainable forest management is vital to our future. Remote sensing approaches can provide key information on forest traits at multiple temporal and spatial scales. Here, we investigate the potential of hyperspectral data collected by the new generation PRecursore IperSpettrale della Missione Applicativa (PRISMA) hyperspectral sensor of the Italian Space Agency for plant trait estimation in mixed forest ecosystems. In particular, we exploit PRISMA data to develop and test a hybrid retrieval workflow for forest trait mapping. The hybrid retrieval scheme consisted in the use of physically based radiative transfer models for the forward simulation of a set of spectral responses as a function of the model input variables, and in the use of machine learning regression algorithms to learn the relationships between the simulated spectra and the model input variables. The model trained on the simulated dataset was then applied to the real remotely sensed spectra for estimating the traits of interest. Finally, the accuracy of the proposed retrieval scheme was evaluated against ground data collected in correspondence of PRISMA overpasses. Based on their ecological importance in terms of plant functioning, the plant traits investigated at leaf and canopy level were: leaf chlorophyll content (LCC), leaf water content (LWC), leaf mass area (LMA) and leaf area index (LAI). Strong correlations were found between measured and predicted values of the LCC and LAI. Slightly worst results were achieved for LMA and LWC. Overall, the estimated LCC and LAI showed value distributions within the ranges expected based on the field measurements. The LCC and LAI maps showed some similarities, but they were not totally correlated. Based on the tree species functional composition obtained from a previous classification performed using airborne data, we observed that LAI showed a larger inter-species variability. Also, the LCC did not differ significantly in regeneration and mature stands, while LAI was higher in the regeneration stands. The results obtained in this study demonstrated that the retrieval of a broad set of leaf and canopy traits from space using hybrid retrieval schemes is feasible, paving the way for future operational algorithms for the routine mapping of vegetation traits from spaceborne sensors.

.

Fossil fuel production activities (oil and gas extraction, coal mining) are responsible for a large fraction of anthropogenic methane emissions. These emissions often happen as methane plumes emitted from so-called point sources, namely small surface elements rleasing relatively large volumes of gas. The detection and elimination of methane point sources have been identified as “low hanging fruits” to reduce the concentration of greenhouse gases in the atmosphere. Satellites offer a unique capability for global monitoring of methane emissions. The retrieval of methane from space measurements typically relies on spectrally-resolved measurements of solar radiation reflected by the Earth's surface in the shortwave infrared (SWIR) part of the spectrum (~1600–2500 nm). The Sentinel-5P TROPOMI mission allows to monitor methane emissions around the globe at regional scales with a daily resolution, but lacks the spatial resolution needed to pinpoint single point emissions. Imaging spectrometers, such as PRISMA, EnMAP, EMIT or the Chinese GF5-02 AHSI, sample the 2300 nm region with tens of spectral channels and a typical spatial resolution of 30-m, which can be exploited for the detection of point emissions and the attribution to particular emitting elements. As of August 2022, PRISMA is the only 400–2500 nm imaging spectrometer with potential for high resolution methane mapping currently accessible to the international science community, which will soon change thanks to the advent of EnMAP and EMIT. In this contribution, we will present an overview of the state-of-the-art in methane mapping with satellite imaging spectroscopy missions. This will include the description of methane retrieval and emission flux rate quantification methods, and the presentation of different studies using available data (PRISMA and AHSI, and potentially EnMAP and EMIT) to detect and quantify methane point emissions in several regions of the world, including oil and gas extraction fields in Algeria and Turkmenistan and coal mines in China. We will also discuss the current and near-future scenarios for the detection and quantification of methane point sources using imaging spectroscopy data.

Atmospheric correction is one of the critical steps when processing satellite optical data. Its main task is the conversion of the measured top-of-atmosphere radiance into surface reflectance by compensation of the atmospheric absorption and scattering effects o nthe electromagnetic radiation. Atmospheric correction is particularly challenging for imaging spectroscopy mission, where hundreds of narrow spectral channels are affected by gaseous absorptions. This is the case of ESA’s FLEX Earth Explorer mission, where an accurate processing in the deep O2 bands is of major importance for the achievement of its mission objectives. In this context, many issues are been considered in a prototype processor to improve the quality of the inverted surface reflectance and retrieval of vegetation solar-induced fluorescence. In this presentation we aim at giving an overview of the aspects to be considered for atmospheric correction of hyperspectral measurements. Here we will discuss about instrumental aspects (smile effect, spectral response, spectral convolution), atmospheric aspects (vertical profile, aerosol modeling) and technical (atmospheric look-up table generation and interpolation). This presentation will be given with the perspective of the FLEX mission but emphasizing the issues that are also expected to be found in broader band missions such as CHIME, PRISMA or SBG. We expect that these considerations will help to improve the atmospheric correction spaceborne imaging spectroscopy missions.

Imaging spectroscopy has brought disruptive technology to the study of greenhouse gas emissions over the last decade. The ability to identify, quantify, geolocate, and attribute individual carbon dioxide and methane plumes with meter-scale resolution over areas from 100 – 10,000 km2 enables unprecedented insights. Anthropogenic emissions from the energy, waste management, and agricultural sectors are now easily distinguished and quantified. Emissions from complex facilities may be identified and geolocated to individual stacks or infrastructure components. Super-emitter behavior has been identified in every sector. Imaging of wetlands methane emissions has similarly transformed our understanding methane hotspots across natural landscapes. Large-scale surveys of the Arctic-boreal region have shown how these emissions hotspots correlate with geomorphology and recent disturbances. We will illustrate these points with examples from airborne and space-based retrievals.

.

In this study we analyse the complementarity between datasets acquired with different ground resolution distance. Merging the information acquired by portable spectro radiometer (nominally referred to “point”) with those derived by remote sensed data (nominally referred to pixel with different GSD), we analyse the improvement in accuracy in the surface classification from space. During the last 2 years hyperspectral images such as ASI-PRISMA and AVIRIS-NG have furnished a suitable datasets enabling systematically validation activities. Since 2018 ground campaigns have been deployed on Parco Naturalistico delle Biancane (PNB) which represents a suitable natural laboratory to provide validation for ground reflectance identification which represent key parameter for many retrieval algorithms used to characterize outcropping surfaces. The PNB belongs to the Lardarello Travale geothermal fields, characterized by boraciferous region, which has been exploited by human activity for low enthalpy energy generation. Steam is now completely utilized for energy production, and the hydrothermal manifestations are now present in limited areas only. The surface geology of PNB is characterized by the outcrops o marl, limestone and jasper which characterize the most frequent lithologies. The PNB area represents an interesting site for testing the capability of classification algorithm in retrieving lithology or mineralogy related spectra in medium resolution satellite hyperspectral sensors like PRISMA and high resolution sensor like AVIRIS-NG. A comparison among the radiometry of PRISMA imagery and the spectra acquired by portable spectra radiometer has been performed before to use the ground truth for classification purposes. AVIRIS-NG is an acronym for the Airborne Visible InfraRed Imaging Spectrometer - Next Generation. AVIRIS-NG measures the wavelength range from 380 nm to 2510 nm (similar to ASD and PRISMA) with 5 nm sampling and a GSD varying according the flight elevation ranging from 0.3 m to 4.0 m. AVIRIS NG data over PBN was acquired in parallel to the filed campaign hold on 4th May 2021. Despite the ASI-PRISMA data used for comparison was acquired on 9th April 2020, PNB surface geology can be considered temporally invariant therefore this data has been choosen for the good weather conditions not available for dataset of PRISMA 2021. The procedures used to process and classify the different datasets are based on ENVI implemented algorithm.

The availability of temporally coherent multispectral and hyperspectral data from remote sensing instruments such as satellite platforms and Unmanned Aerial Systems (UAS) opens the way for numerous Earth Observation applications related to the urban and natural environments. Coupling the above with in-situ observations, such as spectral signatures collected during dedicated field campaigns using spectroradiometers, is critical for providing profound insight of the optical and thermal properties of surface elements, as well as the corresponding earth processes at larger scales. Τhe availability of spectral libraries is disproportionately limited compared to satellite observations, while few of them represent materials observed under conditions similar to imagine observations from satellites and UAS platforms. The increasing availability of high resolution multispectral and hyperspectral data such as the PRISMA and EnMAP missions, the hyperspectral sensors on the ISS, the new planned missions of CHIME (ESA) & SBG (NASA), along with the systems that can be onboard of UAS systems require high quality and updated spectral observations in a standardized and well documented approach under the FAIR data principles. This study presents the first spectral library created with the Spectral Evolution RS-3500 handheld spectroradiometer, which provides spectral information in the Ultraviolet (UV), Visible (VIS), Near-Infrared (NIR) and SWIR (Short-wave infrared) spectrum in the range of 350-2500 nm for the urban materials available for construction material in city of Heraklion, Crete, a typical Mediterranean city. A hierarchic classification scheme has been designed based on the abundance of material and cover that occupy the city. The samples are classified as building materials, which include roofs of various materials (tiles, metal surfaces, concrete, other) and façade samples, ground materials, vegetation and miscellaneous. The spectral signature acquisitions and the library architecture have been designed to assist coupling with satellite observations by ensuring similar acquisition times with satellite overpasses, which are now focused on the Landsat 8/9 and Sentinel 2, as well as UAS acquisitions, both nadir and façade by providing façade samples, which are indispensable for studying the thermal properties of buildings. Next step will be to incorporate the temporal variability of the selected sites for a year-around spectral signature behavior along with the use in analysis of UAS-based hyperspectral data, aiming at a complete mapping of the urban materials of Heraklion, making robust geospatial datasets for use in emissivity corrections of thermal observations, optimum designations of urban planning along with targeted activities for the removal of hazardous materials.

Urban surface fabric identification and mapping is one of the most challenging tasks of the Earth Observation community. Surface fabric information is critical to a plethora of applications in the domains of urban planning and urban climate. The abundance of different man-made materials used in recent years, as well as, the mixed pixel phenomenon, create a large spectral variability, which are indistinguishable for multispectral sensors and require higher spectral resolution signatures for their determination. In this work, the development of an urban hyperspectral library has been designed using the handheld spectroradiometer Spectral Evolution RS3500. The library includes representative materials from the urban area of Berlin city, Germany and Heraklion city, Greece. Spectral Evolution RS3500 measures the surface reflectance in the spectral region from 350nm to 2500nm with 1nm spectral resolution (2151 total spectral bands). The experimental set-up used, allows the extraction of the hyperspectral signatures for a homogenous surface with a 7cm diameter field of view. Preprocessing steps include the removal of the atmospheric channels and use the respective spectral response function. The data were assigned to the multispectral bands of the Copernicus Sentinel 2A MSI sensor, which covers the range from 443nm to 2190nm with 12 spectral bands from which the cirrus band was excluded. Two Sentinel-2A L2A images were selected for image classification, one for Heraklion and one for Berlin. The 20m and 60m bands were downscaled to 10m spatial resolution using Super Resolution toolbox in ESA SNAP ver9. The images were classified using X-SVM algorithm, which was trained by the above mentioned assigned resampled hyperspectral library. The overall accuracy exceeded 87% in both cities.

Imaging spectroscopy is a proven technology for the systematic mapping of the mineralogical footprints of hydrothermal systems at various scales, from core scanning and ground-based imaging to airborne platforms. Spaceborne multispectral remote sensing data have also been successfully used for target generation and mapping of alteration minerals during the early stages of exploration. With the advent of spaceborne imaging spectroscopic data, exploration geologists can take a step forward and reveal the architecture and physicochemistry of the underlying mineral system. Through the use of imaging spectroscopic data, one can map the physicochemical gradients and zonation within alteration assemblages and provide vectors toward prospective mineralized zones contributing to a significant area reduction and operational efficiency. To illustrate this capability, a porphyry Cu-Au deposit called Shadan was thoroughly studied using airborne HyMap and spaceborne simulated EnMAP hyperspectral data at 5 and 30 m ground sampling distances, respectively. Shadan is a typical porphyry deposit with near-perfect zonation located in the volcanic belts of eastern Iran. We used the feature extraction methodology for the accurate identification and quantification of alteration minerals in this area. The chemical composition of a number of minerals including white micas’ Tschermak substitution was mapped by tracking the wavelength minimum of the diagnostic absorption feature using the polynomial fitting technique. The applied method revealed a rich variety of mineralogic products, including white mica abundance, composition, and crystallinity, kaolinite abundance and crystallinity, amphiboles (actinolite) abundance and composition, ferrous/ferric iron abundance and composition, together with biotite, jarosite, gypsum, chlorite-epidote, and tourmaline abundance maps. The spectral products together with bulk rock geochemistry and geophysics helped understand the footprints of fluid flow, phases of alteration and mineralization, the temperature of alteration events, and the chemistry of the fluids comprising their oxidation-reduction state and acidity. This study demonstrated that imaging spectroscopic data coupled with advanced processing algorithms could be utilized to delineate the architecture and physicochemistry of alteration minerals in porphyry copper systems and pinpoint the most promising zones for follow-up exploration activities.

Observing and understanding Earth’s processes are more important than ever as the demand for resources and the human impact on the planet are skyrocketing. Hyperspectral mineral mapping is a crucial Earth observation (EO) tool with manifold applications, including understanding geological processes (e.g., for waste storage, geothermal energy), green- and brownfield mineral exploration, monitoring mining activities (e.g., grade control, monitoring of tailing dams), characterizing human-made mineral deposits (e.g. tailings, contaminations and mine drainage precipitates) and monitoring the local and global impact of mining on the environment. EO enables digital archiving of mineralogic information and drives the transition from traditional maps to digital twins of the Earth’s surface. However, it also implies specific requirements that future spaceborne missions will need to meet in order to support the above applications:   Scale and orientation: Geological features of interest span a wide range and may require the simultaneous interpretation of cm-scale features (e.g. veins, fractures) and regional scale variations in mineral composition (e.g. alteration halos). At the same time, geological outcrops are often obliquely oriented and may be obscured if only nadir data are collected. The integration of data collected from different vantage points and at different scales (space-, air-, drone-borne, terrestrial) is critical for meaningful analysis. For a great part of applications (e.g. monitoring mining activities), additional temporal coverage is crucial. Coordinated acquisition of multi-mission data, established processing platforms, and careful corrections are required to enable this framework.   Spectral range: Mineralogically relevant information is contained (often exclusively) in confined spectral ranges, which are in the visible and shortwave, but also mid- and longwave infrared range. Especially the latter must not be forgotten if the full portfolio of hyperspectral mineral mapping is to be achieved. A careful selection of relevant spectral regions and adapted spectral resolution could help to reduce data load and improve spatial resolution. Processing and validation: Tremendous progress has been made towards machine learning assisted processing of hyperspectral datasets. Nevertheless, developments too often rely on simple and small benchmark datasets. Large scale, mineralogically relevant datasets struggle with heterogeneous, scale dependent classes and often subjective geological interpretation. We recently established three reference sites in Europe that integrate reference data from different scales and technologies as part of the EU-funded INFACT project. We need to continue this effort and engage the community to provide both large-scale benchmarked datasets as well as architectures suitable for scalable machine learning approaches.

Urban ecosystems achieve to host more than 50% of the global population with a trend to increase year by year. Historically, even from the Middle Age, living in cities means access to wealth and prosperity, better education, and more working opportunities. Globalization, economic, cultural, and socio-political reasons (e.g., due to working places, quality of life, transportation, friends etc.) make that trend to increase. Projections for the year 2055 estimate that more than 10 billion will live in urban and rural landscapes. Landscape changes and the conversion to buildup areas of various shapes and dynamics requires a regular monitoring for the assessment and estimation of urban processes and their direct and indirect impacts on humans, nature, and the climate. Quite recently the necessity for the knowledge of the surface materials of the urban and rural areas has been highlighted linked with needs related to urban climate modelling. Further, the knowledge of urban material is important in studies linked with the thermal behavior and thermal comfort of citizens since true urban surface temperature requires emissivity correction based on material information and urban form. The PRISMA sensor by ASI is exploited here to disentangle the three components of the urban landscape of Heraklion city, an urbanscape that has complex urban forms. Spectral unmixing to quantify the key components of the complex area of Heraklion, named Vegetation-Impervious-Soil, has been applied to a L2D imagery. Preprocessing steps include the coregistration of the imagery with the VNIR aerial photography, the removal of the atmospheric bands and the bad lines bands through manual process. Endmembers have been selected manual using the basemap. Regression-based unmixing using Random Forests and the synthetic mixture development of EnMAP Box has been used to quantify the contribution of each component. Validation data have been created manual using the same basemap, but in areas different from the once where endmembers have been selected. Vegetation and Impervious maps show a good performance when accuracy assessment has been done (r2=0.71, RMSE=0.15 for V and r2=0.83, RMSE=0.15 for I), however Soil shows poorer results with RMSE=0.37 and r2=0.56. Abundance maps delineate the spatial distribution of the targets. Soil is a “problematic” key component of urban areas due to the seasonal dynamics where vegetative surface turn into non-photosynthetic vegetation and soils. Fixed targets with the components of urbanscape through the year can support further developments in the hyperspectral domain allowing dynamic mapping of the complex urban landscape.

.

The current surge of high spectral resolution Earth observation missions, such as PRISMA, EnMAP, HISUI, EMIT, and their successors, breaking the ground for more operational hyperspectral endeavors, such as SBG, CHIME and PRISMA-Second-Generation, will lead to unprecedented streams of spectral data. Due to the generic nature of contiguous spectral measurements on one hand and the universality of physics with respect to the interaction between electromagnetic energy and matter on the other, data acquired by different imaging spectrometers can potentially be combined, thus unlocking synergies between different missions. The advantage of combining data from multiple hyperspectral missions mostly lies in increasing temporal observation frequency as the various missions all acquire data of similar characteristics. These include GSDs, bandwidths, band numbers etc., which all are not varying significantly between the different instruments, while differences mostly are limited to SNRs and/or individual data processing approaches. With vegetation, and especially vegetation that is managed in seasonal cycles as part of agricultural production, representing one of the temporally most dynamic surfaces of the Earth, increased observation density is extremely valuable for agricultural applications. After almost 40 years of demonstrating the power of hyperspectral spatial measurements for vegetation monitoring in laborious field experiments and scarce airborne campaigns around the globe, the community now faces the challenge of converting multitemporal hyperspectral data into information products, which should globally provide reliable, quantitative guidance for land surface management decisions. Following this path, we have learned that locally constrained purely empirical retrieval approaches generally contradict the global applicability of spaceborne instruments. At the same time, we must accept that the physics and the chemistry associated with growth processes and land-atmosphere exchange are extremely complicated and thus cannot adequately be traced by simplified models. One of the main scientific goals of the community therefore should be to include relevant light-interacting processes into invertible reflectance models, thus improving the agreement between modelled and measured spectra. Using these models, we should aim to identify the most relevant agricultural variables that can best be accessed via spectroscopy and should engage in the challenge of providing harmonized time-series of these variables. As agriculture is one of the major contributors of greenhouse gas emissions and the role of agricultural land as long-term carbon sink is vividly discussed under the term “Carbon Farming”, quantifying vegetation carbon content via a hybrid retrieval scheme will be presented as an example for relevant agricultural variable retrieval from spaceborne spectroscopy data.

Intact peatlands are essential carbon sinks; at the same time, they are a crucial source of carbon emissions when drained. More than 90% of natural peatland areas have been drained in the federal state of Mecklenburg-Western Pomerania, Germany, which makes them the most significant source of carbon emissions in this less industrialised region. Thus far, various initiatives have been taken for peatland restoration, e.g., rewetting of fen peatlands. In the past, drained peatland areas were used as cropland or grassland; hence alternative strategies for agricultural usage are required. Paludiculture, i.e. agriculture under permanently wet conditions, is one way to utilise these areas after rewetting. Planting Phragmites spp. (Reed) and Typha spp. (Cattail), for example, may present one climate-friendly and economically valuable approach. Still, it is necessary to monitor the success of rewetting efforts for paludiculture because unwanted variations of water tables may hinder the positive effects. Remote sensing is well suited for monitoring rewetting activities. Given the need to accurately separate different vegetation types, the availability of multitemporal hyperspectral information may contribute to such monitoring. We investigated the possibilities of mapping peatland vegetation communities within a complex landscape along the Peene River valleys using multitemporal PRISMA (April, June and August) imagery. Different combinations of acquisition dates were used to map cover fractions of different peatland vegetation types using the regression-based unmixing approach in the EnMAP-Box. A highly accurate classification from fine-scale AVIRIS-NG data served as quantitative reference information. Results achieved with different combinations of PRISMA data were compared to results from multispectral data (Landsat-8/Sentinel-2) from similar acquisition dates. The multi-date PRISMA datasets produced the best results (MAE = 16.4%), followed by results from the single-date PRISMA images and combined multispectral datasets. A class-by-class evaluation showed that different peatland vegetation is best characterised at different acquisition dates. This underpins the importance of multi-date image acquisition. The observed benefits of the multitemporal, hyperspectral information are expected to increase further when adding more observations, e.g. from DESIS or EnMAP, or when more frequent and operational CHIME or SBG data become available. This way, monitoring of peatland rewetting is expected to become more accurate and relevant in the future.

High spectral resolution images provide an efficient tool for evaluation of vegetation chemical and structural composition and photosynthetic function. Consistent measurements of surface reflectance and vegetation traits are required to assess the seasonal dynamics in vegetation function and compare current and past seasonal trends and spatial patterns. The Surface Biology and Geology High-Frequency Time Series (SHIFT) Spring 2022 campaign provides a novel opportunity for evaluation of vegetation health and photosynthetic function across seasonality, using a high spectral and temporal resolution reflectance time series, as anticipated from the forthcoming SBG and CHIME hyperspectral missions. We used the data from the AVIRIS-NG SBG Spring 2022 campaign to compare the differences in spectral properties, leaf and canopy biophysical parameters and carbon allocation through the spring green-up for California live oak (Q. agrifolia), representative of the local ecosystems. Leaf samples were collected from the top sunlit portions of the canopy. In the lab we measured leaf reflectance and transmittance, leaf photosynthetic pigments, and specific leaf weight and area. For each date we collected AVIRIS NG canopy reflectance using for each site regions of interest consisting of 20-30 pixels. We used the Soil Canopy Observation Photosynthesis Energy (SCOPE) biophysical model and vegetation indices to compare among sites the derived gross primary productivity (GPP) and canopy traits. Leaf chlorophyll (Cab) varied with leaf stage, with the younger leaves having lower Cab, which was captured by both leaf in situ measurements and Fluspect simulations. Measured and predicted Cab values were strongly correlated, however the model underestimated the higher Cab levels. To derive canopy spectral and bio-physical traits we used the AVIRIS NG reflectance with the radiative transfer module of SCOPE in an inversion. Canopy traits varied by site and date and captured the differences in photosynthetic potential among the sites. To compare the photosynthetic activity at the sites across the spring, we used SCOPE in forward simulations to estimate canopy gross primary productivity (GPP) and absorbed photosynthetically active radiation (APAR). GPP and APAR varied by site and declined for all sites during the season. Our preliminary results demonstrate the value of high-density hyperspectral reflectance time series for monitoring the trends in vegetation function and productivity. The derived canopy traits demonstrated the change in functionality throughout the season, and the response of the canopies to the two rain events. Further work will compare the traits derived at leaf and canopy level. Using local canopy temperature and PAR data, we will simulate GPP and APAR to develop a more complete picture of vegetation function for each date and site. Future research will evaluate the applicability of the findings to other sites and species

Future optical remote sensing will combine space-borne imaging spectroscopy observations with data-driven machine learning (i.e., deep and active learning) to transform the optical signals into functional traits (e.g., leaf foliar pigments or water contents), informing us about forest ecosystem services. To be successful, the approach requires a robust and comprehensive knowledgebase for i) achieving an accurate training process, and ii) understanding what types of satellite optical inputs (e.g., spectral transformations, principal components, or vegetation indices) are most suitable. Coupled leaf and canopy radiative transfer models (RTMs), simulating interactions of electromagnetic radiation within forest canopies, are suitable virtual environments to generate such a knowledgebase. Radiative transfer through individual plant leaves is dominated by one-dimensional (1D) semi-empirical models PROSPECT (Ferét et al., 2021) and Fluspect (Vilfan et al., 2018). Both models require only a few inputs, and their universal calibration satisfies needs for simulating leaf optical properties of many broadleaf species. Still, are their accuracy and precision sufficient to facilitate retrievals of leaf biochemical traits of any forest type (i.e., tropical, sub-tropical, temperate, and boreal)? Three-dimensional (3D) Monte-Carlo (MC) ray tracing models (Kallel, 2020) are more detailed and input demanding, but, in theory, can simulate a wide variety of forest leaf optical properties. The main hurdles in their broader use are i) a limited calibration of optical parameters, ii) scarce species-specific 3D leaf representations, and iii) a missing extensive validation. As at the leaf level, radiative transfer at the scale of whole canopies is frequently performed with 1D models (e.g., SAIL; Verhoef, 1984), which simulate a mono-species horizontally homogeneous canopy with structural, optical, and biochemical variability only in the vertical dimension. Although 1D RTMs were not designed for applications in forest environments, more suitable 3D RTMs (e.g., DART; Gastellu-Etchegorry et al., 2017) are often dismissed from operational use as i) computationally demanding (i.e., having high requirements on hardware and long simulation times), and ii) requiring preparation of complex representations of 3D virtual canopies. Nevertheless, recent advances in MC light transport modelling, using open-source ray-tracers like LuxCoreRender or Mitsuba (Wang et al., 2022), and semi-automatic construction of 3D virtual canopies from open-access terrestrial and airborne laser scanning point clouds of trees and forest stands (Yin et al., 2022) are removing these classical 3D RTM impediments. This contribution will present recent advances and still existing challenges in the operational use of 3D RTMs for mapping forest biochemical traits from future space-borne imaging spectroscopy observations.

With the increased availability of sensors and data, mainly from unmanned aerial and space platforms, attention to imaging spectroscopy has been renewed. The advantage of imaging spectroscopy data over the existing multispectral imagery is its richer data content. For many target types, and especially vegetation, most variables of interest affect more than one spectral channel or band. For a proper interpretation of imaging spectroscopy data, the numerous vegetation parameters affecting the reflectance spectrum can only be untangled with the help of physical modeling. Furthermore, as with existing multispectral data, monitoring the dynamic state of a vegetation canopy requires multitemporal imagery recorded under different illumination and atmospheric conditions, further discouraging the use of ad hoc empirical approaches. VTT is working on new sensors suitable for hyperspectral nano- and microsatellites. When employed as a constellation, these instruments can provide frequent hyperspectral observations of selected targets. We present the theoretical capabilities of such systems from hardware and algorithmic viewpoints and demonstrate the potential of hyperspectral monitoring of vegetation targets with algorithms making use of the so-called theory of spectral invariants, which allows to minimize the effects of canopy structure and view geometry on the observed reflectance spectrum.

.

The study and monitoring of water bio-physical parameters and the management of aquatic ecosystems are crucial to cope with the current state of inland water degradation. Not only does water quality monitoring support management decision making, it also provides vital insights to better understand changing structural and functional lake processes. Remote sensing has been widely recognized as an essential integrating technique for water quality monitoring, thanks to its capabilities to utilize both historical archive data for thousands of lakes as well as near-real time observations at multiple scales. To date, most of the applications developed for inland waters were, however, developed using multispectral sensors or ocean colour satellites. This situation is about to change with the recent and upcoming spaceborne imaging spectrometers being deployed from experimental satellites. With this contribution we aim to present the exploitation of data gathered from currently orbiting hyperspectral sensors (mostly the ASI-PRISMA) to retrieve water quality parameters and bottom type mapping. To the aim we will update on the state-of-the-art as well as we will provide demonstration with imaging spectroscopy data acquired across different aquatic ecosystems distributed globally, encompassing deep clear lakes, shallow waters colonized by aquatic vegetation and river dammed reservoirs. The unique value of spaceborne imaging spectroscopy will be presented in synergy with multispectral and ocean color sensors as we believe that cooperation among different classes of satellite missions is a key approach for remote sensing of aquatic ecosystems. Some of the challenges for exploiting imaging spectroscopy will be also presented according to the developments of ongoing R&D projects (e.g. H2020 PrimeWater, H2020 HYPERNETS); among the others we will mention the requirements for an improved atmospheric correction, the need of expanding hyperspectral validation sites and training the user community. This would enable the integration of datasets and amplify the value of information strands too often left separate.

A group of 38 experts specializing in hyperspectral remote-sensing methods for aquatic ecosystems attended an interactive Euromarine Foresight Workshop at the Flanders Marine Institute (VLIZ) in Ostend, Belgium, June 4–6, 2019. The objective of this workshop was to develop recommendations for compre- hensive, efficient, and effective laboratory and field programs to supply data for development of algorithms and validation of hyperspectral satellite imagery for micro-, macro- and endosymbiotic algal characterization across the globe. The international group of researchers tackled how to develop global databases that merge hyperspectral optics and phytoplankton group composition to support the next generation of hyperspectral satellites for assessing biodiversity in the ocean and in food webs and for detecting water quality issues such as harmful algal blooms. This presentation will summarise the programmatic recommendations that were formulated during the workshop on topics such as: how to better integrate optics into phytoplankton monitoring programs; approaches to validating phytoplankton composition with ocean color measurements and satellite imagery; new database specifications that match optical data with phytoplankton composition data; requirements for new instrumentation that can be implemented on floats, moorings, drones, and other platforms; and the development of international task forces. Full paper: Dierssen, H.; Bracher, A.; Brando, V.; Loisel, H.; Ruddick, K., 2020 . Oceanography. https://doi.org/10.5670/oceanog.2020.111.

Most of the applications and studies on inland and coastal waters still use multispectral satellites as only very few hyperspectral spaceborne sensors provide(d) enough signal to noise and large dynamic range for observing water targets. A highly dynamic aquatic ecosystem also requires high spatial resolution imaging sensors with frequent revisit times. All these complex mission requirements still cannot be met by a single mission reinforcing the need to combine data from different missions synergistically. Water colour retrievals can be significantly improved and extended with the use of Imaging Spectroscopy, particularly in terms of phytoplankton composition, which is a proxy for assessing environmental changes implicating ecosystem changes. In this presentation, we show examples of current (TypSynSat Project) and past efforts (SynSenPFT Project, Loza et al. 2017) to synergistically combine multispectral and hyperspectral retrievals of phytoplankton functional types. We explore the potential of DESIS hyperspectral data for water colour applications by intercomparing DESIS, Sentinel-2 MultiSpectral Instrument (S2-MSI) and Sentinel-3 Ocean and Land Colour Instrument (S3-OLCI) retrievals of remote sensing reflectance, Chl-a and Phytoplankton Functional Types at the largest German inland water body Lake Constance. Furthermore, we discuss opportunities and challenges of multi- and hyperspectral missions for advancing water colour remote sensing, including joint validation campaigns, algorithms and in situ measurements intercomparisons.

The water-leaving radiance is directly driven by the respective concentrations of optically active water constituents (OAWC) and by their inherent optical properties (IOPs). Nevertheless, the relationship between remotely measurable water reflectance and the IOPs is still to be better elucidated in moderately to very turbid waters. One of the goals of this study was to reassess the hyperspectral IOPs-reflectance forward model over a wide range of water turbidity, hydrosol optical properties while accounting for the polarized nature of light in the atmosphere-interface-water system. A special focus was paid to evaluate the role of the viewing geometry (sun and viewing angles, and relative azimuth angle between Sun and sensor) and to provide the uncertainty attached to such widely used forward model. Based on this model, the performances of OAWC-retrieval algorithms are evaluated for both hyperspectral and multispectral approaches. A specific inversion scheme was applied to a series of in situ data sets of moderate to highly turbid waters including microalgae-dominated waters. Results showed the need to consider the actual multimodal size distribution of sediment and spectrally dependent refractive index to accurately reproduce hyperspectral observations. Conversely, those findings demonstrate the sensitivity of the measured reflectance to size distribution, thus providing a framework for size distribution retrieval from space. Based on those results, we argue that physically based analysis of the signal remains a fundamental step to gain more genericity and applicability of suspended sediment retrieval algorithms enabling to reconcile the exponentially increasing number of regional algorithms.

.

Decades of satellite, airborne, and ground observations clearly show increased melting of snow and ice. This increased melting of cryosphere cover makes Earth more absorptive of sunlight and moves enormous volumes of stored water from land surfaces, raising sea level and changing water availability to large populations. However, the distribution of forcings controlling this accelerated melting is poorly known. Atmospheric warming from greenhouse gases (GHG) is contributing to this acceleration but its magnitude is uncertain due to uncertainties in the controls on the dominant contributor to annual melt (90-95% of the net flux), absorbed sunlight, itself controlled by snow albedo. With increased solar absorption from GHG feedbacks on albedo and increased loading of light absorbing particles, warming and melt commence markedly earlier in the snow season. Despite this crucial role of albedo, sparse measurements have kept us from understanding the global distribution of controls on albedo, snow grain size and radiative forcing by light absorbing particles. In turn, this void has prevented us from accurately modeling cryosphere processes worldwide. While multispectral optical remote sensing has given us access to indications of the controls on snow albedo and in turn climatic feedbacks of the cryosphere in the Earth system, the uncertainties in these retrievals inhibit the proper quantification relative to GHG forcing. The accuracies required to understand the relative roles of changes in albedo and GHG radiative forcing in changing snow warming and melt require the spectral resolving capacity of the VSWIR imaging spectrometer. Through our NASA mission concept SIRFA that was nearly selected for flight, we had little margin with the orbital and field-of-view constraints of cost-limited mission to meet the temporal science requirements. With the implementation of VSWIR imaging spectrometer missions (PRISMA, EnMAP, EMIT, SBG, and CHIME), some in coordinated duty cycles, and the spectrometers on the operational Airborne Snow Observatories, the temporal challenges of SIRFA can be lessened and the snow energy balance science markedly improved. However, such an improvement requires consistency among the snow albedo/grain size/radiative forcing by impurities products among. Such assurance of consistency among our spaceborne and airborne products is the subject of this work. Objectives 1. Determination of consistency in spectral hemispherical-conical reflectance factor among the spaceborne and airborne imaging spectrometers. 2. Establishment of consistency in our algorithms for snow albedo, snow grain size, and radiative forcing by light absorbing particles in snow for SBG and across these missions and operational platforms.

Imaging spectroscopy (IS) can provide important information on the state and dynamics of the global cryosphere. These data allow the retrieval of several physical properties of the surface such as albedo, snow/ice grain size, liquid water content, and concentration of light-absorbing particles. The recent launch of the PRISMA mission (April 2019) opened interesting perspectives for the quantitative estimation of snow and ice properties from satellite IS data. In this contribution, we present results from research activities aimed at evaluating the quality of PRISMA products (L1 as Top-Of-Atmosphere Radiance, and L2D as Surface Reflectance) for studying the cryosphere. Furthermore, we present some preliminary results for the retrieval of snow and ice parameters in polar areas. The calibration and validation activities were accomplished in predefined periods, which were representative of the surface condition during the season, i.e. fresh and aged snow. The study was performed in the European Alps making use of two test sites located at different altitudes: Torgnon (2160 m) and Plateau Rosa (3500 m). Field reflectance measurements were collected in both sites using the Spectral Evolution spectrometer, which operates in the 300-2500 nm domain. Atmospheric Aerosol Optical Thickness (AOT) was measured as well. At the Torgnon site, an automatic system for continuous monitoring of spectral reflectance in the visible to near-infrared (VNIR) provided additional field data to compare with PRISMA observations. Field spectra were propagated to Top-Of-Atmosphere with MODTRAN and then compared with L1 PRISMA products. L2 PRISMA products were validated by using a direct comparison with both field data and by an additional intercomparison with a different retrieval method based on Optimal Estimation and Sentinel-2 data. The agreement between the in situ measurements and satellite data is generally good, and for both L1 and L2 products the mean absolute difference is around 5%. Underestimation of radiance and reflectance for wavelengths below 500nm was observed both for fresh and aged snow. A further preliminary analysis was also conducted regarding the retrieval of snow and ice parameters in polar areas. In particular, we analysed PRISMA scenes acquired in South-West Greenland and over the Nansen Ice Shelf (East Antarctica). In both cases, we obtained reliable estimation of snow and ice parameters such as albedo, grain effective radius, liquid water content, and concentration of impurities and algae. Although the preliminary results are encouraging, further analyses are needed to validate these retrievals with field data.

The presentation reviews the state of the practice in solving for snow properties from spectroscopy—fractional snow cover, snow grain size and shape, concentration and optical properties of light absorbing particles, liquid water content, and snow water equivalent for shallow snowpacks—accounting for effects of the atmosphere, terrain, and vegetation on the signal measured by a spaceborne spectrometer. A variety of algorithms address the forward problem, estimating snow spectral reflectance based on properties of the snowpack, with the main differences owing to calculation of the bihemispherical reflectance or the angularly dependent reflectance. The inverse problem, using spaceborne remote sensing to retrieve the snow properties that govern albedo, addresses discontinuous snow, local and long-distance transport and deposition of light absorbing particles, forests and topography that shelter and obscure the snow, and consideration of surface roughness.

.