The integrated European Long-Term Ecosystem, critical zone and socio-ecological Research Infrastructure (eLTER RI) is now in the preparatory phase (started in 02/2020) of formal implementation as a pan-European Research Infrastructure. The eLTER RI will comprise about 250 eLTER Sites and eLTSER Platforms from over 20 participating European countries covering a vast majority of biogeographical regions. It will offer a wide range of European-level Central Services for different users, including access to sites and site data from multiple sources. eLTER RI will enable standardized observations and analyses of the environment in a holistic manner, encompassing biological, geological, hydrological and socio-ecological perspectives.

Among others eLTER considers the remote sensing community in general as one of its standard users. The reason is twofold: First, eLTER anticipates monitoring various variables which are of also of high relevance for Earth Observations (e.g. vegetation phenology, LAI, Biomass, soil moisture etc) (Zacharias et al. 2021), and enables open access to these data for third party usage (e.g. upstream service). Second, the strategic European priorities listed in the ESFRI roadmap provide a political framework for formalized RIs like the future eLTER RI. This enables the long-term coordination and sustainability of eLTER RI at both national and continental scales, and facilitates the co-design of services according to user needs.

We will share details of the current status and future eLTER RI and highlight how eLTER’s contribution to EO cal/val activities could be realized along the examples of currently pursued integration of standardised phenology into long-term Sentinel-2 (Copernicus) data series. Moreover, eLTER RI has apparent potential to provide upstream service for required standardised ground-truth measurements (including protocols, open access etc.) and for the usage of remote sensing data for eLTER users (downstream service). To achieve this, eLTER RI has initiated continuous collaboration between the remote sensing community including most prominently EO data/product providers such as Copernicus and the communities providing essential ground-truthing over time spans of decades This interface will be another aspect we will touch upon to foster the fruitful exchange and initiate discussions regarding the requirements and concrete actions from both sides.

[Zacharias, S. et al. (2021). Discussion paper on eLTER Standard Observations (eLTER SOs). Deliverable D3.1 EU Horizon 2020 eLTER PLUS Project, Grant agreement No. 871128.]

Land Surface Phenology (LSP) is stated in Skidmore et al (2021) as one of the 30 most important Remote Sensing enabled Essential Biodiversity Variables (RS-enabled EBVs) that are core in assessing and monitoring global biodiversity changes. Within our work, we access LSP characterized as regular, recurrent biological events depending on the annual vegetation activity of vegetated land surface observed from remote sensing. LSP is already widely used either regionally in high-resolution or globally in lower resolution (e.g., based on MODIS data) as a measure for terrestrial ecosystems’ responses to changes of environmental conditions or for characterising species composition and biodiversity of an ecosystem.
Having access to a global, high-resolution LSP data set would allow a homogenous and standardized use of information such as start of season or length of season. However, only since the launch of the Sentinel-2 satellites, with a temporal resolution of 5 days
globally, is a global LSP product on 10m resolution reliably feasible based on this dense
time-series. Due to the high amount of data (pixels and number of images per pixel) needed, a fully automated processing chain implemented on a cloud computing platform is crucial. Within this project, we thus implement a LSP algorithm for global and automated use including an automated validation and verification approach on the Google Earth Engine. Next to a flexible implementation, an accuracy measure is crucial for a reliable and correct use of the output data and the interpretation thereof. Therefore, a system for verification and validation is used resulting in quality layer as an interpreted measure for assessing the reliability of each individual pixel based on the input data and processing as well as an independent validation approach, e.g., with phenocams or alternative satellite data sets (e.g., MODIS LSP product).
The output should therefore be available and easily implementable for anyone needing such a data set on a transparent, open-source basis and thus is intended to provide the basis for a better use of remote sensing data for biodiversity conservation efforts and policy making within the EBV framework.


Skidmore, A. K., Coops, N. C., Neinavaz, E., Ali, A., Schaepman, M. E., Paganini, … (2021). Priority list of biodiversity metrics to observe from space. Nature Ecology and Evolution. https://doi.org/10.1038/S41559-021-01451-X

The ever-increasing pressure on tropical forests implies the need for better characterisation of their properties. This information is crucial for the delineation and protection of the most vulnerable regions. The remote sensing community has reacted by developing new methods for biodiversity variables estimation as potential contributions to the Essential Biodiversity Variable (EBV) framework. Methods based on spectral variation hypothesis and linking spatial heterogeneity of optical imagery with different components of biodiversity show good potential for wide scale biodiversity monitoring. These spectral based methods allow for both alpha (local diversity) and beta (community distribution) diversity estimates at a regional scale. They also make possible the upscaling of biodiversity monitoring results based on costly field surveying efforts. At this time, it is crucial to show that the results from these new remote sensing-based biodiversity indices are robust to atmospheric correction artefacts.
Sentinel-2 satellite data are particularly relevant for vegetation monitoring: the spectral information acquired from Earth surface integrates vegetation chemical, structural and taxonomic information at high spatial resolution, with a five-day revisit period. The exploitation of such data strongly depends on its spatial and temporal consistency after correcting the spectral information from atmospheric effects. Such a consistency is critical for tropical forest monitoring, where cloud cover strongly reduces the availability of operable data, and regional monitoring may require multiple acquisitions.
In this study, we evaluated the stability of the Level-2A (bottom of atmosphere) reflectance products obtained from four different atmospheric correction methods - Sen2cor, Maja, Overland and LaSRC – as well as spectral indices and biodiversity products computed from these reflectance data. When considering pixels with constantly high NDVI values, we made the hypothesis that dense tropical forest should not undergo drastic reversible changes over a short period of time. We selected five Sentinel-2 acquisitions from a forested area located in northern Cameroon over five weeks and compared reflectance data, spectral indices and biodiversity indices derived from the four atmospheric correction methods.
Our results showed that the temporal consistency in reflectance, vegetation indices and biodiversity products over the five-week time step was strongly dependent on the choice of the atmospheric correction method. The temporal stability of L2A-reflectance, spectral indices and diversity indices obtained strongly varied depending on the atmospheric correction method. We found only moderate agreement when comparing the results between methods, even when comparing the aggregated mean values over the full period. Validation is in preparation to allow the comparison between biodiversity patterns with ground observations.
These results highlight the importance of considering the atmospheric correction method used in tropical forest studies. Both reproducibility and validity of biodiversity estimation methods will benefit from systematic explanation and justification of the atmospheric correction method used.

Ice-free areas within the northern Antarctic Peninsula region are found along the coast lines and are mainly influenced by glacial, periglacial and paraglacial processes. Although rock outcrops, sediments and water bodies are the dominant surface covers in such areas, there is a concentration of fauna and flora where mosses and lichens are the main vegetation cover and bird and seal colonies have their breeding grounds. Numerous studies have shown that these ice-free areas contain fragile ecosystems with a potential biodiversity that is influenced by an active hydrologic cycle during the austral summer period. Remote sensing data and techniques are ever more useful for characterizing and monitoring biodiversity and ecosystems location. The objective of this work was to identify different indicators of biodiversity using remotely sensed data within the South Shetland Islands.
The ice-free area of Byers Peninsula in the western part of Livingston Island has an extension of 60 km2, being the largest ice-free area in the archipelago, and is designated as an Antarctic Specially Protected Area (ASPA 126). This area contains a particularly wide diversity of species in comparison to other areas in Antarctica. In this case, biodiversity indicators include spectral features of vegetation covers, surfaces areas saturated with melt water and water bodies, and areas under the influence of penguins and elephant seal colonies that are determined from data obtained with different optical satellite sensors and with spectroradiometer measurements in the field and laboratory. The field and laboratory spectra are used as reference spectra that validate the signature and for the interpretation of the data obtained from the multispectral sensors. Combining the spectral information obtained from the field and lab spectra and using the multispectral spatial resolution, the aim is to obtain an improved characterisation of the indicators. This information is used to simulate information obtained from current and upcoming high spectral resolution satellite sensors such as PRISMA and EnMAP.
Multispectral Landsat-8 and Sentinel-2 data were obtained on the 23/03/2018 and 20/12/2019, respectively. Furthermore, field spectroradiometer measurements with an ASD FieldSpec3 and samples for laboratory analysis were taken during the campaign carried out in February of 2018. The spectral measurements were taken with a contact probe in the field and with a turntable in the laboratory. A digital elevation model was created from topographic information available for the region and used for determining the present fluvial drainage system of the different watershed. Preliminary results from the multispectral sensors for the vegetation show an extensive distribution of moss on the upper Holocene raised beaches. The moss in vigour is easily identified, but this will be reduced depending on the snow cover that affects the vegetative cycle. Areas dominant with lichens are less well identified due to their structure and density. Water saturated sediments as well as water bodies are well identified providing an overview of the different extensions of the numerous watersheds. Bird and elephant seal colonies are also determined due to the guano that is present in such areas. All these surface covers act as indicators of higher biodiversity. This is shown by soil processes evolving under a vegetation cover where for example microbiological communities are active in the soil. These latter communities also play an active role within the animal colonies. Furthermore, available water for seasonal nutrient dynamics and particle transport throughout the watersheds are key in activating biogeochemical processes.

Climate change and human activities are rapidly impacting biodiversity around the globe, posing concerns on the loss of ecosystem function and the critical goods and services they provide. Though, the conservation efforts are hindered by the limited availability of spatially and temporally contiguous information on key aspects of biodiversity. Remote sensing encompasses a wide array of tools capable of providing consistent observations across space and over time. In particular, hyperspectral imaging spectroscopy may advance the assessment of biodiversity due to its spectral dimensionality. The recent and upcoming availability of spaceborne hyperspectral sensors may pave the path towards an integrated global biodiversity monitoring system, providing the possibility to overcome the limitations of the ground-based and airborne hyperspectral sensors in terms of spatial and temporal coverage.
In this contribution, we explore the potential of the new generation PRecursore IperSpettrale della Missione Applicativa (PRISMA) hyperspectral sensor of the Italian Space Agency for biodiversity estimation in forest ecosystems located in different geographic contexts. Emerging techniques based on vegetation functional traits as well as on spectral information content for estimating the ecosystem heterogeneity were applied to PRISMA images acquired on contrasting forest ecosystems in Europe in the summer of 2021. The selected sites for this study are a plain mid-latitude mixed forest (Hardt forest) located in Alsace (France) and mountain forests along an altitudinal gradient located in the Gran Paradiso National Park in the western Alps (Italy). The functional diversity patterns obtained were analysed based on the expected drivers of the vegetation heterogeneity in the considered study areas and on complementary in-situ data. We found that the main factors determining the biodiversity patterns are the species composition and the forest structure and maturity stage in the plain temperate forest, while the morphological and altitudinal factors play a fundamental role in the mountain alpine forests. Future directions and possibilities offered by the newly available streams of Earth Observation data are also discussed in this contribution.

Biodiversity change has many dimensions and occurs at a range of spatial and temporal scales rendering traditional in situ biodiversity monitoring inadequate for regular global monitoring. Direct human observation of species and their traits is both expensive and time consuming, with poor consistency in spatial and temporal coverage when upscaled to country or global level. This poses a challenge as UN Convention on Biodiversity (CBD) obligates its 196 parties (the EU included) to monitor the 20 Aichi targets, such as target-5 to halve the rate of loss of natural habitats by 2020 and target-15 to restore 15% of degraded ecosystem by 2020. Remote sensing continually surprises by 'seeing the unexpected' in wavelengths ranging from the optical (for instance, seeing the biochemical content of plants4) to shortwave- and thermal-infrared (e.g. seeing how water content varies within a tree canopy and beyond to radar (seeing flooding under trees). In this presentation we demonstrate how the relative abundance of bacterial family taxa in different ecosystems (boreal, tundra and savanna) can be mapped using hyperspectral imagery remote sensing. The technology to achieve this has been developed using hyperspectral imagers combined with eDNA profile data of microbiological taxa. Environmental DNA (eDNA) can rapidly profile taxonomic groups at order, family, and genera levels across the entire tree of life using barcoding methods that use a short genetic marker in an organism's DNA to match it with DNA belonging to a known species. An innovation here is to go from point data (eDNA) to interpolating continuous genetic composition biodiversity products over large geographical areas and then to understand how biodiversity in general and ecosystem function in particular respond to biotic conditions, competition, and stress. A key question is whether such genetic composition biodiversity products and Essential Biodiversity Variables mapped using remote sensing are indeed relevant to policy indicators. The improved biodiversity information will more completely inform our progress in meeting CBD Aichi targets and UN Sustainable Development Goals, as well as generate more accurate biodiversity maps required to further improve predictions from cutting-edge ecological and climate models.