The Fiducial Reference Measurements (FRM) programme, implemented through ESA, is designed to facilitate robust in situ reference observations for the calibration and validation of satellite products, following the principles defined in the Quality Assurance Framework for Earth Observation (QA4EO). The programme emphasises metrological traceability and uncertainty analysis.
FRM4VEG is an ESA-founded project that applies these metrological principles to three variables of interest for vegetation: surface reflectance (SR), the fraction of absorbed photosynthetically active radiation (fAPAR) and canopy chlorophyll content (CCC), and identifies three missions of interest for validation: Sentinel-2, Sentinel-3, and Proba-V. Phase 1 of the project successfully developed and applied FRM methods to the application of field spectrometers for SR and indirect optical techniques including digital hemispherical photography (DHP) and chlorophyll meters for fAPAR and CCC.
Phase 2 of the programme (FRM4VEG2) advances the developed concepts by including new instruments, measurement platforms, and more detailed uncertainty characterisation. A key difference between the two phases is the collection of in situ SR data from a hyperspectral imager mounted on an unoccupied aerial vehicle (UAV), for which validation protocols and procedures are being developed. This is being achieved through two activities. One took place this summer, involving collection of validation data over Wytham Woods (UK), something which was not possible in phase 1.
The other, which will take place in 2022, is a Committee for Earth Observation Satellites (CEOS) endorsed intercomparison exercise of UAV-mounted hyperspectral imagers over an agricultural site in Barrax (Spain). Known as the Surface Reflectance Intercomparison Exercise for Vegetation (SRIX4VEG), the exercise will bring together teams from around the world with similar instruments to assess the need for a SR validation protocol from UAV-mounted instruments. SRIX4VEG will consider two scenarios: 1) where a pre-existing protocol is used as a basis for measurements, and 2) where each contributor is asked to collect the validation data according to their own judgement.
The Fiducial Reference Measurements (FRM) concept for validation of satellite-based products is a keystone of the ESA calibration/validation strategy. With a strong focus on metrology, it includes the need for traceability and fully characterized uncertainty budgets according to community-agreed-upon measurement protocols and uncertainty evaluation procedures. Within this strategy, the ESA funded Fiducial Reference Measurements for Vegetation (FRM4VEG) project is aiming to address the challenge of applying metrological principles to vegetation and surface reflectance product validation.
For vegetation product validation, an end-to-end uncertainty evaluation framework has been developed for two main variables: the fraction of absorbed photosynthetically active radiation (fAPAR) and canopy chlorophyll content (CCC), which makes use of leaf area index (LAI) in its calculation. The process involves quantifying the uncertainties associated with individual in situ reference measurements from different instruments (Digital Hemipherical Photographs, LAI-2200, AccuPAR, SPAD chlorophyll meter), and incorporating these uncertainties (and those of the high-resolution satellite image) within the upscaling procedure, which facilitates the validation (uncertainty estimation) and conformity testing of moderate resolution satellite products. fAPAR and CCC FRM covering several irrigated and non-irrigated crops (Barrax, Spain) and deciduous broadleaf forest (Wytham Woods, UK) were collected in 2018 and 2021 field campaigns.
This talk will present to the validation community, firstly, the latest achievements of the FRM4VEG project for vegetation variables, from ground measurements to up-scaled maps with per-pixel uncertainties estimates, with particular attention to the uncertainties contributed by different instruments and data processing methods. Secondly, we will use the FRM data to validate the Sentinel-3 OTCI product and two Sentinel-3 OLCI FAPAR products, the OGVI (or Instantaneous green FAPAR) provided by ESA, in addition to the Sentinel-3 FAPAR provided by Copernicus Global Land Service. Finally, conformity testing against user/mission requirements (using uncertainties in both satellite and in situ data) will be presented, and future validation needs addressed.
The HYPERNETS project is preparing the next generation of hyperspectral radiometer with associated pointing and acquisition system and network processing for validation of water and land surface reflectance derived from satellite missions.
Spaceborne optical imaging missions such as Sentinel-2 and Sentinel-3 are used routinely to provide data for environmental monitoring of water and land surfaces via products such as chlorophyll and suspended particulate matter concentrations in water and Fraction of photosynthetically active radiation absorbed by vegetation (FAPAR) and Land cover over land. Validation of the water and land surface reflectance products is needed to ensure reliability of all such satellite data products.
For water reflectance, the AERONET-OC network is currently the main source of radiometric validation data. This network is mature and operational but only multispectral. A new radiometer, the HYPSTAR®, has been designed and is being tested to provide finer spectral resolution (3nm FWHM for the range 380-1020nm) at lower cost. The new radiometer is mounted on a commercially-available pan-tilt unit and is controlled by a rugged PC with purpose-built electronics and software. Data is transmitted in near-real time for standardised processing and distribution.
For land surface reflectance measurements, there are few automated systems. The new instrumentation that has been designed and is being tested within the H2020/HYPERNETS project will provide land surface reflectance measurements at high spectral resolution including spectral coverage extended to 1700nm (with 10nm FWHM resolution) and with angular/spatial variability.
The progress of the H2020/HYPERNETS project will be summarised briefly here. In situ data from the validation sites (7 in Nov 2021, 14 expected by May 2022) will be presented. The in situ data will be compared with water and land surface reflectance data from various satellites, including Copernicus missions Sentinel-2 and Sentinel-3 but also hyperspectral satellite data (e.g. PRISMA) and metre-scale satellite data (Planetscope Doves).
A growing number of High Resolution (HR) and Very High Resolution (VHR) optical satellites have been launched in the last years, and more will be launched in the next years. Together with the benefits that this increased availability of Earth Observation (EO) data can bring, there is a number of other aspects to be considered, like the increased need of harmonized calibration and validation (cal/val) methodologies and of common cal/val sites also dedicated to VHR missions. Several cal/val networks are in place nowadays (e.g. RadCalNet, AeroNet, Hypernets, cal/val sites set-up by national space agencies or institutions, etc.), but each site and network is usually dedicated to a specific task and need, with little flexibility.
Therefore, the European Space Agency (ESA) is investigating the idea of setting-up a common playground, i.e. the so-called Cal/Val Park, where to explore current and new cal/val methodologies for HR and VHR optical satellite sensors, for both testing and operational purposes. The Cal/Val Park is meant to be flexible enough to be used for both multi-spectral and hyperspectral missions, scalable enough to accommodate new needs and new EO missions that may be launched in the next years, and open to be used by both the “institutional space” and the “commercial and New Space” players. The Cal/Val Park concept is currently in the definition phase and is meant to be a multi-Agency effort.
Copernicus is the European Union Earth Observation program, dedicated to monitor our planet and its environment, giving free access to remote sensing data and derived Earth Observation products. For proper use in environmental monitoring and scientific applications, it is fundamental to guarantee high quality and consistency of these satellite derived products. One of the possibilities to ensure product quality is to perform quantitative comparisons of satellite derived products with matching in situ observation. Two options can then be considered for ground data sources: through intensive field campaigns or making use of permanent ground stations deployed and maintained on the long term. In the first case, a large variety of variable can be assessed, but logistical challenges and financial resources limit in time and space the products validation. More over meteorological constrains often limit the number of data that can actually be used for Earth Observation products. The second option is from far the most cost effective although it is not yet possible to cover all ground variables with permanent field deployment.
To achieve these objectives of systematic and long-term data validation, the Ground-Based Observations for Validation (GBOV) service has been implemented, facilitating the use of observations from operational ground-based monitoring networks and their comparison to EO products. The service is guaranteed through 3 different components:
• Collection of multi-year ground-based observations (Reference Measurements - RMs) of high relevance for the understanding of land surface processes from more than 80 sites. These RMs are then upscaled to generate Land Products (LPs), in order to validate the Copernicus products. In particular, the LPs distributed through the GBOV portal are: Top of Canopy Reflectance (ToC-R), surface albedo, Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Available Radiation (FAPAR), Fraction of Covered ground (FCover), Surface Soil Moisture (SSM) and Land Surface Temperature (LST).
• Upgrade of existing sites with new instrumentation or establishing entirely new monitoring sites to close thematic or geographical gaps.
• Implementation and maintenance of a database for the distribution of the Reference Measurements and the corresponding Land Products, available through the website https://land.copernicus.eu/global/gbov. GBOV data access is completely free, after registration and acceptation of the terms of use and the data policy.
Initially designed to serve the Global Land Service, GBOV products are used by a worldwide community The purpose of this presentation is to provide a status of four year service achievement and the perspectives.
In recent decades, numerous scientific field-instrument networks have been developed in different contexts and for several purposes. SpecNet, Aeronet, FluxNet, or similar are just a few examples. One promising area of research involves the use of automatic field spectrometers, operating in the visible and near-infrared spectral domain, for the purpose of long-term and unattended monitoring of their targets. The applications of field spectroscopy are manifold: from vegetation studies (plant phenology, plant physiology, and phenotyping) to the study of water quality, the atmosphere, and the optical properties of snow/glaciers.
In recent years, JB Hyperspectral Devices, with the support of the scientific community, has developed autonomous instruments, capable of operating continuously in the field in a wide range of harsh climatic conditions (from the equator to the poles) along with dedicated open-source software to process and manage the collected data.
These instruments acquire punctual, high-temporal-resolution hyperspectral data. Thanks to a robust and standardized data processing chain (based on R: A language and environment for statistical computing), they return time series of radiometric parameters as well as optical properties related to the monitored targets. JB devices are made to synchronously acquire upwelling and downwelling radiance, optimizing the acquisition speed according to light conditions and acquiring the dark current signal at each measurement cycle. For example, the FloX (Fluorescence BoX), features a high-performing spectrometer (FWHM: 0.3 nm, SSI 0.15, SNR 1000) and allows continuous measurement of solar-induced chlorophyll fluorescence (SIF) emission. Furthermore, continuous measurements of spectral down-welling and up-welling radiance, using an additional spectrometer in the VIS-NIR range, allows for the computation of reflectance and different spectral indices. Another example of JB’s instruments is the RoX (Reflectance BoX), which is a simplified version of the FloX. It contains only the VIS-NIR spectrometer and therefore it is designed to measure reflectance and spectral indices.
To further utilize the data collected with these devices, JB is currently in the process of making these instruments compatible with the latest flux data collection and processing systems operated in typical flux stations (e.g. SmartFlux from LI-COR). This approach is aimed to help with accurate synchronization and improved compatibility of various types of data collected by different devices at flux stations. Moreover, a collaboration with existing consolidated flux networks (e.g. FluxNet) is ongoing to explore better coupling between flux, optical, and remote sensing measurements. The possibility to integrate JB devices into flux networks, via a standardized procedure of configuration, set up, data processing, and different data product levels are further evaluated.
In the view of satellite cal/val activities, the need of multiple instruments distributed around the globe is furthermore introduced. A specific study, conducted on a subset of nine RoX and FloX field spectrometer systems, deployed across the northern hemisphere in Europe, US, Africa and China was conducted, with the purpose being to validate ESA’s Seninel-2 time-series. The instruments were installed above agricultural fields, broadleaf forest, savannah and alpine pasture for timeframes ranging from several months to up to three years, continuously measuring hyperspectral reflectance data in the visual/NIR range between 350nm and 900nm. Spatially temporal clustered time-series were validated using the ground-truth for clear-sky conditions. Our results suggest a good agreement with R² larger than 0.5 between field spectrometers and satellite bands. Vegetation indices agree very well with R² larger than 0.7, e.g. NDVI with R² =0.89 between Sentinel and field spectrometer subsets across all sites. Our results suggest that RoX and FloX monitoring field spectrometers are a valid ground truth for known vegetation targets. In terms of reflectance bands, they are in good agreement with the Sentinel-2 satellite time-series and in terms of Vegetation Indices, they are in very good agreement.
In this contribution, we present an overview of the state-of-the-art JB hyperspectral devices with applications ranging from SIF, water, atmosphere and snow studies. Finally, point field spectrometers are put forth to promote the potential for a future ground-based network of devices with multiple purposes: from investigating ecosystem properties to the validation of satellite products. In this way, field spectrometers can contribute to guiding the application of ground-based remote sensing practices towards an improved insight into vegetation, water and snow across scales, as well as bridging information between ground and satellite.