Open Earth Monitor — Global Workshop 2024

Welcome plenary by Steffen Fritz - International Institute for Applied Systems Analysis (IIASA)

Please add an abstract as soon as possible

The Open-Earth-Monitor Project - Coordinator of the Open-Earth-Monitor project and director of the Opengeohub Foundation

Please provide an abstract as soon as possible

Approximate five billion hectares (38%) of global land area is used for agricultural system, contributing significantly to the loss of biodiversity and having a substantial impact on water resources and greenhouse gas emissions of the World. Aiming to support multi-scale environmental policies and decision making process, several land monitoring systems / products were launched in the last years, including WorldCereal, GLaNCE, Dynamic World, UMD GLAD GLCLUC, GLC_FCS30D and Global Pasture Watch. Even though all these systems / products have different advantages, limitation, constraints and resolutions (thematic, spatial and temporal), in general they have a high potential to be combined to support different land cover and land use applications at global, national and local scale. Here we present a framework able integrate global monitoring systems / products in an automated, flexible and reproducible way, taking advantages of new technologies as cloud-optimized formats and cloud services. We demonstrated it integrating different crop and pastures classes in seamless monitoring system for the tropics, allowing the users to define their own area of interest, harmonization rules and overlap criteria. The implementation is publicly available in scikit-map (https://github.com/openlandmap/scikit-map) and all input layers publicly accessible through SpatioTemporal Asset Catalog (STAC).

Eddy covariance (EC) systems are commonly used to measure the net exchanges of energy, water carbon dioxide (CO2) and other trace gasses between the ecosystems and the atmosphere. Such measuring systems have been established in different ecosystems and climate regimes across the globe, thereby providing invaluable ground information to understand ecosystem dynamics at global scale. Although the number of EC stations installed worldwide (e.g. FLUXNET sites) are constantly growing with time, their spatial distribution is limited in comparison to the vast complexity of land ecosystems. Furthermore, EC towers track the exchange of energy and matter from an area (often referred to as a footprint) that spans some few hundred meters around and upstream of the measurement site (the so-called fetch), and which can vary according to meteorological conditions. Remote sensing (RS) and in-situ flux datasets are commonly combined to upscale the exchanges of carbon and energy at a global scale (e.g. the FLUXCOM project), as well as for calibration and validation activities. The challenge to do this correctly lies in trying to link the footprints of the EC measurements to those of the satellite measurements, a task that is often disregarded or oversimplified. In this study we designed a methodological approach within the Open-Earth-Monitor (OEMC) project to estimate dynamically the match (or mismatch) between some likely proxies of EC footprints (approximated as circles with radius from 50 to 200 meters) and the footprints of (coarse) spatial resolution RS time series. To quantify the degree of mismatch we collect Sentinel-2 images at 10 meters resolution for several EC sites over Europe. Then, we compute the kNDVI vegetation index for all the sites masking clouds and cloud shadows. We also define proxies for different pixel sizes of satellite data ranging from 500 meters to 1500 meters radius around the tower. To compare the EC footprints with the Satellite pixel resolution we compute the Jensen-Shannon index that quantifies the amount of information (in terms of kNDVI) shared between both scales at every available time step. As a result, we provide initial recommendations of when in the year the sites are more suitable to be matched with satellite data according to the surrounding phenology. We expect these will open the possibility to correct biases in future upscaling fluxes exercises and remote sensing products calibration.

Daily gridded meteorological datasets are an important source of information for analysis of historical weather and many other research areas since they have no gaps in the spatio-temporal domain they cover. Most of the daily gridded meteorological datasets represent reanalysis or estimations from different remote sensing sensors or are generated by downscaling procedures. A daily gridded meteorological dataset for Europe at 1 km spatial resolution, named MeteoEurope1km, is created, covering the 1961–2020 period and consists of five variables: maximum (TMAX), minimum (TMIN), and mean (TMEAN) temperature, sea-level pressure (SLP), and total precipitation (PRCP). Spatio-temporal regression kriging, an interpolation method that combines multiple linear regression for trend modeling and space-time kriging for the estimation of the residuals, is used for interpolation of daily temperature variables. Ordinary kriging is used for SLP and PRCP, except that for PRCP an additional step to predict PRCP occurrence is applied using Indicator kriging. Combination of GHCN-daily, ECA&D, and SYNOP observations from OGIMET service is used as an observational dataset, with previous removal of duplicated stations and outliers. Geometric temperature trend, digital elevation model and topographic wetness index are used as auxiliary variables for temperature datasets. Accuracy assessment (leave-one-station-out cross-validation) shows high accuracy of the fitted models. Coefficient of determination for all temperature parameters and SLP is greater than 96%, while for PRCP is greater than 76 %. Root mean square error is 1.3°C, 1.6°C, 1.8°C, 1.5 mbar, and 2.5 mm for TMEAN, TMAX, TMIN, SLP, and PRCP, respectively. MeteoEurope1km is available as cloud optimized GeoTIFFs, and are accessible through dailymeteo.com portal, ZENODO, and R meteo package. Future work will be oriented towards increasing the spatial extent to other continents besides Europe, interpolation of other daily meteorological variables, and improving models performances by applying spatial machine learning methods, such as Random Forest Spatial Interpolation.

Satellite radar remote sensing utilizes long-wavelength energy that can penetrate clouds and is sensitive to changes in the physical structure of vegetation. These characteristics, in combination with the high spatial and temporal detail of new and near-future radar satellites, provide major opportunities for monitoring forest disturbances and regrowth dynamics.

We provide an overview of recent research activities on the use of radar remote sensing to monitor forest dynamics and present key results achieved in the Open-Earth-Monitor project. These include forest disturbance monitoring, monitoring of forest loss drivers and carbon, and assessments of selective logging intensity. We will highlight how the near-future availability of freely available multi-frequency radar data from Sentinel-1 (C-band), NISAR (L-band), and BIOMASS (P-band) will improve our ability to assess forest dynamics. We will also discuss our open-source initiatives aimed at facilitating the adoption of radar data and change detection approaches by both the scientific community and country stakeholders.

The WorldCereal project, funded by the European Space Agency (ESA), aims to provide a comprehensive understanding of global cropped areas, irrigation practices, and the distribution of major commodity crops. WorldCereal has developed a dynamic open-source system that generates a range of products, including temporary crop extent, seasonal maize and cereal maps, seasonal irrigation maps, seasonal active cropland maps, and confidence layers. These products are based on the analysis of Sentinel-1 and Sentinel-2 imagery at 10 m spatial resolution, complemented by Landsat 8 imagery and AgERA5 meteorological information, and are updated at seasonal intervals for each agricultural system. WorldCereal has demonstrated the feasibility of global crop mapping by producing the first global, seasonally updated crop and irrigation maps for the year 2021. WorldCereal has also released a fully open, harmonized database of in-situ reference data related to land cover, crop type, and irrigation, enabling a broad community to access and contribute to this growing resource. WorldCereal is now entering a new phase, in which the system is being implemented as a cloud-based processing service in the new Copernicus Data Space Ecosystem. The system will offer more flexibility and customization options to users, allowing them to generate tailored crop type products for their regions of interest. Moreover, the WorldCereal product suite will be extended with eight new crops, and the in-situ reference database will be updated and expanded. WorldCereal will also conduct a series of regional use cases and capacity building activities to demonstrate the system’s capabilities and to boost user uptake by the broad agricultural monitoring community. WorldCereal provides a vital tool for policymakers, international organizations, and researchers to better understand local to global cropping patterns and to inform decision-making related to food security and sustainable agriculture.

Air pollution, particularly fine particulate matter (PM2.5), ground-level ozone (O3), and nitrogen dioxide (NO2), poses a significant global health risk, contributing to early mortality. Measuring population exposure is crucial for understanding and mitigating these health impacts. This paper leverages recent advancements in air pollution data to review various global air pollution datasets based on a criteria set. The framework facilitates comparisons between various hybrid datasets (combining ground-based and satellite measurements) and offers a methodology for constructing air pollution exposure indicators for PM2.5, O3, and NO2. It uses the Global Burden of Disease data to update the indicator set on the national and subnational levels across the 1990-2020 period. Results reveal that most OECD countries fall short of the World Health Organization's (WHO) 2021 air quality guidelines for PM2.5, O3, and NO2. Countries such as Chile, Korea, Poland, and Türkiye exhibit PM2.5 concentrations (population weighted) exceeding safe levels by a factor of four. Similarly, several OECD countries such as Korea, Italy, and Slovenia experienced severe O3 exposure in 2020, while non-OECD countries such as India displayed even higher population weighted O3 concentrations, exceeding safe levels by more than double. A sensitivity analysis further indicates that despite similar trends observed across different air pollution datasets, considerable differences are found between global datasets and national statistics. This highlights the need to further examine the accuracy of the various data sources and help guide policy analysis at the national and subnational levels. Given the widespread failure to meet safe air quality standards, our findings emphasize the urgent need for global policy actions to reduce population exposure to air pollution and safeguard public health.

There is currently a lack of high-resolution pan-European information on land use management, especially in terms of how forests, cropland and grassland are intensively and extensively managed. This is partly due to the lack of ground-based information, which is needed to downscale these types of management practices (some of which are captured in different types of agricultural censuses and surveys and National Forest Inventories) as well as the inability of remote sensing to capture different kinds of land use. This type of information is needed for economic land use modelling and for assessing policy impacts, such as the latest reforms from the Common Agricultural Policy (CAP) and other European Union (EU) Green Deal targets. These types of analyses are undertaken using economic land use models such as GLOBIOM and CAPRI, which is one of the main aims of the Horizon Europe funded LAMASUS project (https://www.lamasus.eu/).

One of the main inputs to the development of a land use management map is Corine land cover, which is a remotely sensed product developed by the Copernicus Land Monitoring Service every six years. First, we produced an annual time series of Corine from 2000 to 2018 by using the high-resolution land cover times series produced by OpenGeoHub and the BFAST algorithm applied to MODIS data to determine the year of change between the six-year production cycle of CORINE. Any remaining changes that were unaccounted for had the year of change selected randomly. Transition rules were also applied to ensure that the land cover/land use transitions were reasonable.

Land use management classes for forest, cropland, grassland and urban areas were then devised in collaboration with the modelers in the LAMASUS project as well as around 30 stakeholders who participated in the first LAMASUS stakeholder workshop. Using different input data sets from remote sensing, in-situ data (from LUCAS), modelled data from CAPRI, and statistical information from agricultural censuses, surveys and other sources, rules were developed to allocate the Corine land cover classes to more detailed land use management classes. Here we will present the results of this mapping along with a method for how the map has been fit to official area statistics so that this information can be used by the economic land use models in LAMASUS.

This workshop equips participants with hands-on experience managing their EO data on Zenodo using the zen Python library. Participants will learn to customize metadata and automate data management workflows using zen scripts.

Improving the sustainability of the European livestock sector requires high resolution spatial data. Otherwise potential negative impacts of livestock related to local ecosystem degradation, as well as positive ones such as preserving cultural landscapes through grazing cannot be analysed. Data on livestock numbers usually used in scientific analyses are collected and provided by the European statistical office, but are provided on a rather coarse spatial resolution of statistical regions. In addition, data on the actual use of grasslands, whether grazed, mown and the intensity of their use is not collected systematically or not at all. We provide an approach for mapping grazing livestock (cattle, small ruminants) density and the use of grassland for Europe. We first collected livestock numbers on a local level for all EU countries, which we harmonized, and supplemented it with statistics on actual outdoor grazing of animals. We then mapped areas that are grazed by combining EU-wide in-situ data on grazing with a set of socio-economic, terrain, soil and climate characteristics using machine learning. We then allocated grazing livestock on two different earth observation derived land use and land cover products: corine land cover and the high resolution grassland layer. Our approach enables identifying areas that are grazed, and combined with livestock statistics, also how intensively these areas are used either for grazing or mowing. Such information can support tracking the state of european grassland ecosystems, landscape conservation, as well as other environmental dimensions related to the livestock sector, such as nitrogen deposition, with a high spatial detail. Finally, by using regularly updated systematically collected data, we can update the data in the future.

Findable, Accessible, Interoperable, and Reusable (FAIR) data principles are composed of a set of guidelines focused on efficient discovery and data utilization, which are crucial for sharing scientific data effectively. Hence, adapting to the FAIR principles benefits diverse environmental applications and supports a diversity of policies. This study presents the findings of an extended user survey conducted within the Open Earth Monitor Cyberinfrastructure (OEMC) project, exploring user perspectives on FAIR environmental data. For this purpose, an existing survey targeted at both users and producers of geospatial data was extended to enhance the representability and have the widest feedback for understanding users' and producers' needs, expectations, experiences, and understanding of FAIR principles.
The survey included three blocks. The first block addressed the background and general information of the survey respondents. The second block inquired about the characteristics of the geospatial data that has been primarily used or produced. The third block investigated how user and producer group participants are familiar with the FAIR principles and which of those seemed most relevant to them. In addition, we fostered a target-specific participant selection strategy to cover the main institutions and relevant user groups.
The survey revealed a discrepancy in the preferred observational scales between data producers and users. While producers primarily focus on generating data at global scales, users frequently require data at local and regional levels. This finding underscores the need for improved communication and collaboration between data providers and users to ensure data production aligns with user needs. Furthermore, the survey identified findability and openness as the top priorities for FAIR environmental data, alongside clear licensing, comprehensive metadata availability, and detailed documentation.
These findings emphasize the crucial role of robust data management practices and user-centric approaches in promoting the effective utilization of environmental data.
Further key findings from user responses will be presented, highlighting user perceptions of FAIRness in environmental data, current gaps in FAIR implementation, and identified challenges. Based on these insights, we will discuss the implications of the survey results and propose recommendations for advancing the FAIRness of environmental data in the future.
This research contributes to ongoing efforts within the OEMC project and beyond, informing strategies for improving the discoverability, usability, and overall value of environmental data for various stakeholders.

Earth observation data provides an invaluable resource to assess the state and condition of the environment. However, many domain–specific applications, such as the mapping of species-specific habitats and vegetation for conservation, often require specific spatial and thematic resolutions, rather than off–the–shelf products. And although remotely sensed data is critical to assess actual coverage, particularly for the assessment of restoration opportunities, knowledge on the potential distribution of habitats and vegetation is usually required. Here I will provide an overview of ongoing efforts to estimate current and potential vegetation types across global, European and local extents. I will focus both on approaches to integrate existing data sets for global and European habitat estimates, but also demonstrate the potential of earth observation data and deep learning to identify vegetation types at high resolution. Finally, I will highlight opportunities to bring the Earth observation and ecology community closer together particularly in the light of data gaps, harmonization and standards.

Integrity of natural ecosystems is one of the main concerns of current European and Global Green Policies, e.g., the European Green Deal. Public administration managers need reliable and long-term information for a better monitoring of the ecosystems and climate evolution and inform decision makers. Data Spaces are intended to become the EC comprehensive solution to integrate data from different sources with the aim to generate and provide a more ready to use knowledge on climate change, circular economy, pollution, biodiversity, and deforestation. This workshop aims to discuss pros and cons of some technological solutions in terms of Data Spaces, OGC standards, semantic descriptions, datacubes, FAIR principles and sovereignty of data. It also intends to share lessons learned from main EC projects dealing with the topic: AD4GD, GREAT, B-cubed, Fairicube, etc.

Recording of the session:
https://youtu.be/JH14NmIazpc?si=I8jQUkY5uWhLfzE8

nrt is a Python package designed to streamline environmental monitoring efforts by offering a unified Application Programming Interface (API) for a diverse array of forest disturbance monitoring algorithms. This unified API simplifies the process for users, enabling easy comparison and integration of different algorithms that are optimized for rapid computation and scalable deployment.

Beyond its core functionality, the nrt ecosystem encompasses additional tools that enhance its utility and versatility. These include diagnostics, time-series simulation, generation of reference data, and computation of accuracy metrics. Collectively, these features make nrt a valuable resource for environmental monitoring and analysis, catering to a wide range of research and operational needs.

During the workshop, participants will engage in hands-on demonstrations covering the various aspects of the nrt ecosystem. This practical experience aims to equip attendees with the knowledge and skills necessary to effectively utilize this tool in their projects, enhancing their capability to leverage any of its components for their projects.

Biodiversity loss is a critical environmental concern, with habitat destruction and degradation identified as key drivers. Recent advancements in computational methods and the ever-growing availability of Earth Observation (EO) data enable detailed analyses of land cover changes at unprecedented spatial and temporal scales. This paper develops a set of indicators of land cover and land cover conversions to assess potential pressures on terrestrial biodiversity and ecosystems. Key land cover conversions include deforestation/reforestation, cropland expansion/contraction, and urban/infrastructure development. We leverage two high-resolution datasets (i.e. the Copernicus Climate Change Initiative Land Cover [CCI-LC] and the Global Human Settlement Layer [GHSL] built-up area) to develop national and subnational indicators for all countries globally, spanning 2000-2020 for CCI-LC and 1975-2030 for GHSL. The analysis reveals a continued decline in natural and semi-natural vegetation cover in many OECD countries and partner countries since the 2000 baseline. For example, Brazil experienced a substantial loss of tree cover (200,000 km²) between 2000 and 2020, equivalent to an area exceeding Switzerland's landmass by a factor of six. Meanwhile, most OECD countries exhibited a net gain in tree cover during the same period. Urban development is another key reason for the observed decline in natural and semi-natural vegetated land where countries such as China and India displayed a significantly higher increase in artificial surfaces compared to OECD countries over the past two decades. Results currently only account for the ecosystem extent and do not account for the ecosystem condition. For instance, some grassland land cover may have been significantly modified by long-term grazing and is in fact intensively managed grassland (wild prairies vs grassland pastures). Therefore, these results should be considered alongside complementary data sources to provide a more comprehensive picture of biodiversity pressures and highlight that current global land monitoring EO products do not adequately meet the needs of policy analysts who require data at the interface of land cover and land use.

Accurate and timely flood risk assessment is paramount for effective disaster mitigation and preparedness. Traditional flood susceptibility maps (FSMs) often fall short by providing static representations, failing to capture the dynamic nature of flood risk in a changing climate. This study presents a novel dynamic FSM framework that integrates high-resolution climate data and temporal analysis to address these limitations. Developed within the context of the Open-Earth-Monitor Cyberinfrastructure (OEMC) project, our methodology offers a significant advancement in flood risk modeling.
To generate dynamic, high-resolution (1 km) FSMs for the Mediterranean region, we utilized the Random Forest algorithm. These maps uniquely adapt to varying seasonal conditions, precipitation intensities, and post-drought scenarios. Our model's adaptability stems from its training on a comprehensive dataset that combines flood and non-flood locations from the Copernicus Emergency Management Service (EMS) and the Global Flood Database v1. Additionally, we incorporated crucial factors influencing flood events, including elevation, slope, land cover, drainage density, soil moisture, and precipitation. Model evaluation employed cross-validation techniques utilizing both training and testing datasets. This comprehensive assessment confirmed the superior performance of the Random Forest model, solidifying its effectiveness as a robust tool for flood susceptibility mapping.

Tropical vegetation dynamics and ecosystem carbon (C) stocks typically vary with local topography and forest disturbance history. Yet, neither remote sensing nor vegetation modeling captures the underlying mechanistic processes determining ecosystem functioning and therefore the resulting estimates often do not match field observations of vegetation C stocks, especially so in hyperdiverse tropical forest ecosystems. This mismatch is further aggravated by the fact that multiple interacting factors, such as climatic drivers (i.e., temperature, precipitation, climate seasonality), edaphic factors (i.e., soil fertility,
topographic diversity) and diversity-related parameters (i.e., species composition and associated plant functional traits) in concert determine ecosystem functioning and therefore affect tropical forest C sink-strength. Here, we propose a novel framework designed for integrating in-situ observations of local plant species diversity with remotely sensed estimates of plant functional traits, with the goal to deduce parameters for a recently developed trait- and size-structured demographic vegetation model. Plant-FATE (Plant Functional Acclimation and Trait Evolution) captures the acclimation of plastic traits within individual plants in response to the local environment and simulates shifts in species composition through demographic changes between coexisting species, in association with differences in their life-history strategies. Our framework allows to project the functional response of tropical forest ecosystems under present and future climate change scenarios and thus should have crucial implications for assisted restoration and management of tropical plant species threatened by extinction.

Among the many services in-situ datasets can provide to society, one of the more pressing interests currently active in the Earth Observation (EO) sector is the integration of in-situ and satellite datasets. The remote sensing community is actively using ICOS (Integrated Carbon Observation System) outputs for calibration and validation activities of satellite products. However, there are additional measurements currently excluded from the ICOS portfolio that could be beneficial for calibration and validation opportunities: for example, fraction of absorbed photosynthetic active radiation (fAPAR) and land surface temperature (LST) from thermal cameras.
An experimental setup was implemented on a subset of ICOS stations for estimating leaf area index (LAI), strictly related to fAPAR, from above- and below-canopy measurements of photosynthetic active radiation (PAR). The first longer-than-1-year datasets being available, we present some relevant preliminary results and the future direction of this activity.
NASA recently published some best practices on LST measurements for validation of satellite products. At this scope, a single thermal camera of high accuracy is deployed on a network of measuring stations. We intend to check how this setup relates to different configurations, such as different camera models, or the deployment of 3-4 lower-standard sensors looking at different angles, thus increasing the spatial resolution.
Additional points under scrutiny are: what is the heterogeneity of these variables in the eddy covariance footprint, and how can these measurements add value to the net ecosystem exchange (NEE) and its derived products? And how can the integration between satellite imagery and ground observations benefit from them?

Air pollution is a health risk to millions of citizens in Europe. Critical concentrations of nitrogen-dioxide (NO2), ozone (O3), and particulate matter (PM10 and PM2.5) occur predominantly in densely populated areas affected by high volumes of traffic or industry. Although several thousand air quality stations scattered over Europe record hourly measurements, the EEA publishes continuous maps on an annual basis with considerable time lag. However, there is a public benefit in accessing such maps more timely.
With the OEMC Air Quality Monitor we design tools which streamline the mapping workflow building on top of the EEA methodology. The process includes gathering and pre-processing data (both measurement and covariates) and making spatial predictions for the four mentioned air pollutants. We leverage public station measurements, gridded climate and atmospheric transport model outputs, and land cover and traffic information as well as open source software. This combination facilitates a transparent way to map air quality in Europe at one kilometer spatial resolution for daily, monthly, and annual intervals.

The pervasive presence of microplastics in terrestrial ecosystems has emerged as a pressing environmental concern. Recent studies have identified soil as a major sink for microplastics contamination, potentially surpassing oceanic levels by factors ranging from 4 to 23-fold. The small size of microplastics and the complexity of soil matrices as sink substrates pose challenges for quantifying soil pollution. As a result, current analytical methods are limited in efficiency, making large-scale environmental assessments unfeasible. The vertical incorporation of microplastics into soil, along with the challenges of recognizing microscopic objects in satellite images, restricts the practicality of using remote sensing for direct large-scale environmental assessments. Hence, a more comprehensive approach is necessary to tackle these challenges. One potential solution involves utilizing satellite imagery combined with a maximum entropy model. By integrating locations where microplastic presence has been confirmed and extracted from soil samples, the maximum entropy model can establish a connection between satellite-derived environmental predictors and the presence of microplastics in soil. The aim of this research was to assess the practicality and viability of employing this approach in a real-world setting.

To test our approach, we designed a case study covering wider administrative area of the City of Osijek, Croatia. For training data, we utilized 31 sampled locations where soil microplastics have been confirmed through previous research, along with environmental variables primarily derived through signal enhancement of Sentinel-based imagery. After literature review, a preliminary list of 31 environmental predictor variables was generated, covering various facets of microplastics input to the soil and their dispersion in the environment. These were tested for variance inflation factor (VIF) and spatial autocorrelation to identify statistically significant variables for model calibration. To relate environmental variables to microplastics presence, we leveraged maximum entropy model. The best-performing model underwent additional testing using various permutation tests to evaluate its robustness. We identified 4491 different sets of three environmental variables eligible for further examination. We employed each combination to train maximum entropy models using 5-fold cross-validation to identify the most robust model. Additional testing included jackknife cross-validation to identify and remove outlier samples.

The best performing model, with an AUC under the ROC of 0.863, was the one trained using combination of environmental predictors including land cover (CLC+ Backbone raster product), soil moisture derived from Sentinel-1 imagery, and catchment areas determined through hydrological analysis of the digital elevation model. The output prediction map clearly delineates areas that highly likely represent pollution hotspots. This research demonstrates the feasibility of utilizing satellite imagery, in conjunction with topological analysis and maximum entropy models, to conduct large-scale environmental assessment and accurately pinpoint hotspots of soil microplastics contamination. This approach could significantly aid future stakeholders since the EU has taken proactive steps as of 2018 to tackle soil microplastics pollution, by implementing regulations, action plans, and initiatives to prevent plastic pellet loss. Furthermore, the European Commission has incorporated impact assessments into its decision-making process regarding microplastics. Advanced environmental monitoring techniques offer potential in tracking progress and quantifying effectiveness of forthcoming measures.

The collection of representative observational datasets in environmental sciences is crucial for advancing the understanding of the phenomena under consideration. The integration between in-situ datasets with remote sensing and machine learning techniques makes possible reliable predictions and analyses with enhanced precision and resolution. The OEMC project aims at supporting informed decision making on environmental policies for the benefit of the whole society, by combining in-situ measurements and remote sensing datasets. Here we investigate the impacts of some of the OEMC in-situ datasets on society and policymakers: how are the in-situ datasets supporting the use-cases of the project? What is their combined potential in terms of technological advancement and knowledge boost? The following categories of OEMC in-situ data, their benefits, and relation to sustainable development goals (SDGs) are scrutinised.
GHG fluxes: GHG fluxes ground observations, combined with satellite data, can be proficiently used for calibration and validation of models, with benefits in terms of better predictions, development of early warning systems, better understanding of climate change impacts, ecosystem services, etc. Current and potential stakeholders are the Intergovernmental Panel on Climate Change (IPCC) and international projects such as the Global Carbon Project (GCP) and FluxCom initiative. UNFCCC is also using GHG flux data. Related SDGs include 11, 12, 13 and 15.
Forest biomass: in-situ observations of forest biomass are fundamental in refining the assessment of global forest carbon stocks and their change under natural and anthropogenic drivers. These data serve the needs of a wide range of stakeholders, from both the scientific and the policy making sectors, interested in quantifying the actual carbon sequestration capacity of forests and refining estimates of forest inventories. Policies such as the European Forest strategy and monitoring of SDG 15 will benefit from such datasets.
Marine and terrestrial biodiversity: these datasets support projects and activities of biodiversity conservation, a fundamental branch of Earth science and a crucial aspect for the survival of humanity. Potential stakeholders include the European Environmental Agency (EEA) and the Joint Research Centre of the European Commission (JRC), and policies such as the European Biodiversity strategy and SDGs 14 and 15.
Ocean and coastal datasets: the importance of ocean and coastal organisms for the balance of the biosphere becomes more and more evident, but scientific knowledge is still limited in comparison with the terrestrial counterpart. Increasing the monitoring of these ecosystems is crucial, in particular for human communities living in coastal areas. EEA and JRC are included in the stakeholders interested. Related SDG: 14
LCLU: in-situ land use and land cover information derived from processing land surveys data and satellite imagery support land degradation alert systems and EO mapping. Potentially supported SDGs are 11, 12, 13, 14 and 15.
Automated ground observations: automated measurements of biological processes support the validation of EO products and provide input for ecological modelling. Data consistency is enhanced by the availability of a continuous dataflow from field sites where sampling is logistically or financially constrained. Possible applications include early warning systems in agricultural, forestry, and urban greening sectors, improved agronomic and silvicultural practices, monitoring ecosystems productivity and biodiversity levels. Potential stakeholders are the EEA, the JRC, entities involved in mandatory and voluntary carbon markets (UNFCCC, UNDP, private companies), national governments and local administrations. Related SDGs are 11, 12, 13 and 15.
Citizen science: citizen science in-situ data for training and validation of EO mapping models can play a fundamental part in supporting environmental policies, covering a wide range of topics, from deforestation to aboveground biomass assessment, from crop type to land use and land cover distributions. The European Green Deal is expected to greatly benefit from this type of in-situ datasets, and SDGs 13, 14 and 15 will potentially be supported.
In-situ and gridded integration: although the combination of in-situ and gridded datasets is common, their spatial resolution often differs. A case study focusing on eddy covariance data tries to shed light on the overlapping degree of ground and satellite footprints, with benefits for society in terms of technological advancements and a deeper understanding of how ecosystems react to climate change, with potential benefits for SDGs 13 and 15.

Assimilating Leaf Area Index and Soil Moisture from Optical and SAR Data into the WOFOST Model to Improve Maize (Zea mays L.) Yield Estimation

Gebeyehu Abebe 1,2, Odunayo David Adeniyi1,3, Amazirh Abdelhakim1,4, Zoltan Szantoi1
1European Space Agency (ESA)/ESRIN, Frascati RM 00044, Italy
2Department of Natural Resources Management, Debre Berhan University, Debre Berhan, Ethiopia.
3Department of Earth and Environmental sciences, University of Pavia, Italy, Via Ferrata 1, Pavia, 27100, Italy.
4Centre for Remote Sensing Applications (CRSA), Mohammed VI Polytechnic University (UM6P), Hay My Rachid, Ben Guerir 43150, Morocco
Abstract
Crop Simulation Models (CSM) are commonly used to estimate crop yield at a local scale. Meanwhile, Remote Sensing (RS) data provides valuable information on crop parameters like soil moisture and leaf area index (LAI) across different spatial scales. Data Assimilation (DA) is a powerful technique that combines CSM and RS data from satellite imagery to enhance simulated crop state variables and model outputs, such as total biomass and yield. In this study, we aimed to implement a joint assimilation strategy for LAI and soil moisture data in the WOFOST model. The goal was to simulate rainfed grain maize yield at the field scale and evaluate its performance at both the field and administrative zone levels. The Ensemble Kalman Filter (EnKF) algorithm was applied to achieve this integration. The LAI and soil moisture data were sourced from Sentinel 3 and Soil Moisture Active Passive (SMAP) L3 Radiometer Global Daily 9 km Soil Moisture, respectively. The study tested various assimilation scenarios, including deterministic modeling, independent assimilation of LAI from Sentinel 3, independent assimilation of soil moisture from SMAP, and joint assimilation of both LAI and soil moisture data. Ongoing validation involves comparing the simulated grain maize yield with field observations and independent grain maize statistics data in the major maize-growing administrative zones of western and southwestern Ethiopia. The expected outcomes include improved accuracy in grain maize yield predictions at the field scale and enhanced crop monitoring and forecasting at local and national levels.
Keywords: Data assimilation; EnKF; LAI; soil moisture; WOFOST; grain maize yield

The ‘Exploitation Platform’ concept derives from the need to access and process an ever-growing volume of data. Many web-based platforms have emerged - offering access to a wealth of satellite Earth Observation (EO) data. Increasingly, these are collocated with cloud computing resources and applications for exploiting the data. Rather than downloading the data, the exploitation platform offers a cloud environment with access to EO data and associated compute and tools that facilitate the analysis and processing of large data volumes. The Exploitation Platform benefits users, data providers and infrastructure providers. Users benefit from the scalability & performance of the cloud infrastructure, the added-value services offered by the platform – and avoid the need to maintain their own hardware. Data hosted in the cloud infrastructure reaches a wider audience and Infrastructure Providers gain an increased cloud user base.

Users are beginning to appreciate the advantages of exploitation platforms. However, the market now offers a plethora of platforms with various added value services and data access capabilities. This ever-increasing offer is rather intimidating and confusing for most users. In order to fully exploit the potential of these complementary platform resources we anticipate the need to encourage interoperation amongst the platforms, such that users of one platform may consume the services of another directly platform-to-platform.

EOEPCA (EO Exploitation Platform Common Architecture) is a European Space Agency (ESA) funded project with the goal to define and agree a re-usable exploitation platform architecture using standard interfaces to encourage interoperation and federation between operational exploitation platforms - facilitating easier access and more efficient exploitation of the rapidly growing body of EO and other data. Interoperability through open standards is a key guiding force for the Common Architecture: platform developers are more likely to invest their efforts in standard implementations that have wide usage; off-the-shelf clients and software are more likely to be found for standards-based solutions.

The EOEPCA system architecture is designed to meet a set of defined use cases for various levels of user, from expert application developers to consumers. The architecture is defined as a set of Building Blocks (BBs), exposing well-defined open-standard interfaces. These include Identity and Access Management, Resource Discovery, Data Access, Processing Workflows, Data Cube Access, Machine Learning Operations, and more. Each of these BBs are containerized for Kubernetes deployment, which provides an infrastructure-agnostic deployment target.

The exploitation platform is conceived as a ‘virtual work environment’ where users can access data, develop algorithms, conduct analysis and share their value-adding outcomes. The EOEPCA architecture facilitates this through a Workspace BB that provides a user-centric platform experience in which the standard discovery, visualisation and access interfaces are re-used for user-owned resources maintained within the platform - including data, applications, added-value products (from processing), etc. This is supported by an Application Hub building-block that provides interactive web-tooling for analysis, algorithm development, data exploitation and provides a web dashboard capability through which added-value outcomes can be showcased.

Our presentation will highlight the generalised architecture, standards, best practice and open source software components available.

Air pollution has emerged as a critical global concern, exerting adverse impacts on natural ecosystems and exacerbating the pace of climate change. Despite the existence of mitigation strategies, the accurate quantification of methane emissions remains a formidable challenge, impeding progress towards meeting emission reduction targets set for 2030. This study is dedicated to addressing the urgent global issue of air pollution, with a particular focus on methane emissions, known for their significant contribution to climate change and associated environmental and health hazards. Conventional monitoring techniques have proven inadequate, leaving millions of abandoned oil wells unchecked in their methane emissions, thus demanding a comprehensive solution. In response, we present a novel technological advancement based on satellite data, to facilitate the precise measurement, detection, and ongoing monitoring of methane leaks. By harnessing breakthroughs in deep tech disciplines such as Earth observation integrated with machine learning, astrophysical methodologies, theoretical chemistry, and computational fluid dynamics, this technology enables the identification of methane leaks across diverse geographical locations worldwide.
Furthermore, this study underscores the critical importance of fostering collaboration and information exchange among stakeholders to optimize the effectiveness of emission reduction endeavors. Through its innovative approach and interdisciplinary collaboration, this work aspires to deliver a significant contribution towards mitigating climate change impacts and safeguarding natural resources for the benefit of future generations.

One of the major challenges in data management is (and in the project OEMC) is demonstrating the correct implementation of the FAIR (Findable, Accessible, Interoperable and Reproducible) principles. To make data accessible, it is required that “data is retrievable by their identifier using a standardised communications protocol that should be open, free, and universally implementable”.
OEMC has produced a list of datasets that are exposed to the public with and elegant Open-Earth-Monitor App. Our talk will focus on demonstrating the interoperability of the taken approach, showing an alternative web map browser that gives access to the same OEMC datasets. This web map browser was deployed using the original MiraMon Map Browser technology without any customization and using only Open Geospatial Consortium (OGC) standards web services calls, demonstrating the technical interoperability of the OEMC services. The presented visualization portal goes beyond a simple visualization by combining the OGC WMS standard with modern web browser capabilities. During the talk, we will demonstrate how to access OEMC datasets through MiraMon browser functionalities, such as query by location, multiple projections support, reading storymaps, and data multidimensional support among others. An important feature of the visualization portal is that it allows the final users to provide common feedback about the data (such as star rating and comments) that are shared with other users as well as to produce and share their own storymaps and this way share the knowledge gained by analysing the data.

Hydrological planners need accurate and up-to-date information on snow dynamics. The OEMC project aims to improve the measurement of the Snow Water Equivalent as an estimation of the available water stored in snow covered areas in the Alps to support planning activities such as hydropower, agriculture and drinking water. To reach this objective high spatial and temporal information is required. Several remote sensing sensors exist with different spatial and temporal resolutions and hence different potentiality. To produce optimal results, an integration of different data sources is necessary. This requires large computational resources as well as huge data amounts. In this context, standardized cloud processing APIs such as OpenEO serve as powerful processing tools that can promote openness and reproducibility. In this talk we will present how we exploited cloud native EO to improve the development of snow products, such as snow cover fraction and snow water equivalent maps.

At the OEMC Global Workshop in 2023, we presented a community led initiative part of the wider Open Innovation framework at European Space Agency that worked to implement an open, interactive, user intuitive platform for a constantly updated, comprehensive and detailed overview of the dynamic environment of the open source digital infrastructure for geospatial data storage, processing and visualisation systems. Today, we have over 450 documented geospatial FOSS projects, interconnected into the FOSS4G ecosystem.
At the OEMC Global Workshop of 2024, the team presents the work done within the next steps, identifying quality metrics for open source software and assess the connection with the health of the associated project and thus paving the way to understand the benefits as well as the pitfalls of certification in geospatial open source software.
The work is supported by ESA, under the Permanently Open Call for Proposals for Future EO-1: EO Science for Society.

Over the last decade, UAV systems have enabled high-resolution data collection for various applications at relatively low

costs and with great flexibility in acquisition time and parameters [5]. This data can serve as a valuable reference for large-
scale space-borne applications. However, the flexibility in image acquisition presents challenges related to varying data types

and quality, which are affected by environmental conditions, sensor specifications, and radiometric calibration. Capturing
comparable reflectance values with UAV systems is particularly challenging, and many early studies relied on minimal
preprocessing or raw digital number (DN) values [4]. Given that some datasets' spectral information (reflectance/DN) may
not be directly comparable, a classifier that emphasizes generalized texture information is needed rather than relying solely
on spectral data. Among common machine learning (ML) techniques, the convolutional neural network (CNN) of deep
learning (DL) has proven to be a successful tool in classifying images of land use from remote sensing data [1, 2]. CNN
allows high-order representation based on generalized texture information already used in crop classification [7, 3, 6].
Our research explores the potential of using low-to-moderate-quality UAV data for agricultural pattern classification,
focusing on how color-balancing techniques can enhance data consistency when images are captured under variable lighting

conditions. We evaluated the performance of CNNs in classifying agricultural patterns using moderate- to low-quality, high-
resolution (0.07-meter) optical multispectral data collected from three agricultural test sites in Germany between 2019 and

1. We used models trained exclusively on samples converted to reflectance values and applied them to images impacted
   by different sunlight conditions, including digital number (DN) and reflectance data. The models were trained to classify
   small-scale agricultural patterns, such as damaged and undamaged canopy, weed-infested and bare soil areas, across four
   crop types: winter wheat, rapeseed, corn, and spring barley.
   This study, funded by the German Federal Ministry for Economic Affairs and Energy (FKZ: 50EE1901), is carried out in
   collaboration with CLAAS E-Systems GmbH to develop an application for crop monitoring based on Sentinel-1 data.

In the face of the escalating global climate crisis we are facing, the integration of open data, open science, and open platforms has emerged as a transformative approach in Earth Observation (EO) and its applications. This abstract explores the pivotal role of these interconnected principles in addressing these climate challenges in the European Space Agency (ESA).
Open data initiatives have democratized access to valuable EO datasets, fostering collaboration and innovation across a wide range of stakeholders from policy makers, policy owners to scientists, and end users as farmers. By facilitating transparency and accessibility, these initiatives enable a deeper understanding of Earth's systems, crucial for informed decision-making amidst climate uncertainty.
Coupled with open data, open science practices advocate for transparency, reproducibility, and the sharing of methodologies, results, and findings. This collaborative view not only accelerates scientific discovery but also cultivates a culture of accountability essential in confronting the multifaceted complexities of climate change and the climate finance behind it.
Furthermore, the integration of open platforms also in ESA provides a dynamic infrastructure for EO research and application development. These platforms not only streamline data management and analysis but also empower communities to co-create solutions tailored to their unique challenges, fostering resilience in the face of environmental threats, which ESA is supporting through many of its projects and programmes.
As the climate crisis intensifies, the synergy between open data, open science, and open platforms offers a promising pathway forward in EO endeavours. By fostering inclusivity, innovation, and collective action, this integrated approach holds the potential to catalyse transformative change, safeguarding our planet for future generations making good use of all ESA’s EO missions and options.

Due to their ability to observe the land surface irrespective of weather and lightning conditions, radar satellite constellations are indispensable for monitoring of highly dynamic land surface processes. While in the past only scatterometer missions allowed consistent monitoring at global scale, albeit at very coarse spatial scales, this has changed fundamentally with the Copernicus Sentinel-1 mission that stands out as one of the most successful Synthetic Aperture Radar (SAR) missions. With its novel combination of high spatial and temporal resolution, long-term mission planning, and open data policy it has served as a role model for the conceptualization of future radar missions. With the upcoming launches of the Japanese Advanced Land Observing Satellite-4 (ALOS-4) satellite, the NASA-ISRO SAR Mission (NISAR), ESA’s Biomass mission, and the Copernicus Radar Observing System for Europe in L-band (ROSE-L) satellites, there is now the opportunity to monitor dynamic processes at high spatial resolution (10-20m) with short revisit times (1-3 days) at multiple frequencies (C-, L-, and P-band). In this presentation I will discuss a collaborative effort of the Vienna University of Technology (TU Wien) and the EODC Earth Observation Data Centre to build a global multi-frequency SAR datacube suited for applying hybrid algorithms combining physical models and machine learning. Furthermore, I will show examples of how we use this tailored datacube for the monitoring of soil moisture, floods, vegetation, and soil structural characteristics.

Please provide an abstract and exact title as soon as possible

The primary objective of this study was to develop and evaluate different remote sensing techniques for mapping Other Wooded Lands (OWL), while also assessing the accuracy and uncertainties associated with classifying OWL class compared to forest and grasslands. Additionally, we aimed to design a scalable process for large-scale OWL mapping. As defined by the Food and Agriculture Organization (FAO), OWLs are areas with 5-10% tree canopy cover for trees reaching a height of 5 meters at maturity, or with a combined cover of shrubs, bushes, and trees above 10 percent. Also, OWLs must span a minimum land area of 0.5 hectares and exclude predominantly agricultural or urban land uses. Three diverse landscapes were chosen based on expert input, encompassing natural regions globally and representing the three main land cover classes of interest: forest, OWL, and grassland. The selected areas were (1) Cheringoma, Sofala, Mozambique; (2) Cerrado biome, Goiás, Brazil; and (3) Albacete and Jaén, Spain. For each Area of Interest (AOI), we selected a Sentinel-2 MGRS tile that entirely covered the area. A stratified random sampling approach ensured robust sample collection across all land cover classes within each scene, resulting in over 1.7 million samples per scene. High-resolution imagery from Google Earth/Bing was utilized for visual interpretation. The mapping utilized data from 2022, encompassing a six-month window before and after the year of interest (totaling two years). A total of 174 metrics were calculated on data from various sources to characterize land cover for OWL modeling. Data processing was conducted using Google Earth Engine (GEE), and a Random Forest algorithm was employed for OWL land cover modeling. The resulting maps exhibited a global accuracy of 74.5% (Mozambique) and 76.5% (Brazil), Spain is currently under analysis. In Mozambique, the producer accuracy for OWL was 42.4%, with omissions associated with grasslands and forests at 34.5% and 21.5%, respectively. For the Cerrado region, both user and producer accuracies were notably higher, at 71.6% and 74.7% respectively. Mapping results were combined with ICESat-2 satellite lidar, where available, to investigate the vegetation height and structure of land cover classes. Top of canopy heights, median heights, and percent forest cover decreased between forest, OWL, and grassland classes. This methodology offers a scalable approach for mapping OWLs, contributing to improved deforestation monitoring and environmental protection efforts.

This hands-on workshop will present the use of big data analytics to work with data available at the Open GEO Hub cloud service

In this workshop, the participants will apply an automated machine learning framework suitable for EO data.

The workshop aims to present new data sets related to tree management available on the GFW website and tools for collecting feedback on these data sets and tools to collect training and validation data.
The list of new data sets includes the new version of Spatial Database of Planted Trees, the
Natural Lands Map, and a new version of the Forest management layer for the year 2020. Discussion will focus on current challenges data producers face such as dataset definitions, data gaps, and quality assurance of the presented datasets.

Climate change poses a significant threat to the distribution and composition of forest tree species worldwide. European forest tree species’ range is expected to shift to cope with the increasing frequency and intensity of extreme weather events, pests and diseases caused by climate change. Despite numerous regional studies, a continental scale assessment of current changes in species distributions in Europe is missing due to the difficult task of modeling a species realized distribution and to quantify the influence of forest disturbances on each species. In this study we conducted a trend analysis on the realized distribution of 6 main European forest tree species (Abies alba Mill., Fagus sylvatica L., Picea abies L. H. Karst., Pinus nigra J. F. Arnold, Pinus sylvestris L. and Quercus robur L.) to capture and map the prevalent trends in probability of occurrence for the period 2000–2020. We also analyzed the impact of forest disturbances on each species’ range and identified the dominant disturbance drivers. Our results revealed an overall trend of stability in species’ distributions (85% of the pixels are considered stable by 2020 for all species) but we also identified some hot spots characterized by negative trends in probability of occurrence, mostly at the edges of each species’ latitudinal range. Additionally, we identified a steady increase in disturbance events in each species’ range by disturbance (affected range doubled by 2020, from 3.5% to 7% on average) and highlighted species-specific responses to forest disturbance drivers such as wind and fire. Overall, our study provides insights into distribution trends and disturbance patterns for the main European forest tree species. The identification of range shifts and the intensifying impacts of disturbances call for proactive conservation efforts and long-term planning to ensure the resilience and sustainability of European forests.

Earth Observation (EO) biomass and carbon datasets are increasing and their potential as inputs to the environmental-economic accounting framework based on SEEA was assessed in this study toward accounting for all carbon pools: above-ground, below-ground, deadwood, litter and soil carbon. This demonstration allowed the compilation of carbon accounts in four accounting periods 2010-2017, 2017-2018, 2018-2019 and 2019-2020 for six case countries namely Brazil, Mozambique, the Netherlands, the Philippines, Sweden and USA, and later on compared with the accounts from a counterpart carbon accounting framework based on UNFCCC. The compiled carbon accounts revealed the above-ground component being the dominant carbon pool in Brazil and the Philippines, while soil organic carbon outweighs other carbon pools in the Netherlands, Sweden and surprisingly Mozambique. We found decreasing carbon stocks especially for Brazil even in shorter accounting periods i.e., 2018-2019 captured by the EO dataset. This is in contrast to what has been reported by countries to UNFCCC mostly reporting stability in the carbon flows over the years. Part of the discrepancy is the country definitions of managed forests which can be inconsistent with forest management datasets from EO (this study). Another reason is the dependency of countries on national forest inventories which are rarely updated on an annual basis. Moreover, our compiled accounts showed minimal carbon emissions from forest degradation mainly driven by the choice of ecosystem extent input, and lower soil carbon emissions than UNFCCC reports, potentially underestimating peatland emissions. The findings and outputs from this demonstration echo the potential of EO datasets for carbon accounting especially with the advent of time series biomass data, higher spatial resolution of ecosystem extent maps 5-10 m and online ecosystem accounting tools for efficient use cases.

xcube is a mature and capable Python software package and framework for EO data ingestion, processing, analysis, visualization, and dissemination. In the scope of the Open Earth Monitor project, xcube is being updated and expanded to support new data sources, improve on-demand cluster processing capabilities, and run seamlessly on the new Copernicus Data Space Ecosystem. Recent work also focuses on providing a maximally preconfigured turnkey distribution of xcube, increasing its suitability as a drop-in compute engine for cloud infrastructures such as CDSE. xcube’s features are complemented by the new zappend tool, which provides robust creation and updating of large, slice-structured Zarr datasets.

This talk will describe and demonstrate a typical large-scale processing workflow using the xcube framework in the CDSE ecosystem – running the gamut through data ingestion from multiple sources through the xcube data store subsystem, data cube construction and normalization, data synthesis and processing to export, dissemination, and seamless visualization via the server and viewer components. Scalability and big data capability is accounted for throughout through approaches such as object storage, parallelization, on-demand cluster processing, dataset pyramidization, and lazy computation. The newly implemented components and improved integration make xcube an ideal tool for the realization of typical Open Earth Monitor workflows.

Here we propose a planetary health diagnostic framework, which aims to track, understand, and characterize the Earth system during the onset and progression of both chronic change (such as climate change) and abrupt disruptions (stemming from climate extremes and socio-economic shocks). However, monitoring a single component of the Earth system to guide policy, but ignoring other essential components, could lead to misleading diagnostics and maladaptation of global sustainability. To gain insights into the integration of climate, biosphere, and society, we apply an interactive dimensionality reduction to the annual variability of multi-stream global data from 2003-2022, including data representing the biosphere and climate combined with national socio-economic indicators.

We find that the interactions between biosphere, atmosphere and socio-economy can be captured by three principal axes, which cumulatively explain 17.3%, 22.8% and 24.5% of the variability condensed by non-interactive dimensionality reduction in each individual domain, respectively. The 1st and 2rd pairs of Biosphere-atmosphere-socioeconomic interactive axes describe terrestrial vegetation and land surface water syndromes. The first axes positively correlate to terrestrial vegetation productivity, air temperature, and technology and public health. The second axes negatively correlate to soil moisture, potential evaporation, and reflect several combined socioeconomic aspects such as land use and inequality. We find distinct trajectories across countries with high-income countries more resistant COVID-19-induced economic shock. High and low income groups show contrasting trajectories that are related to poverty reduction and methane emission in the low-income country group. This study advocates for a data-driven paradigm to jointly monitor the recent trajectories of the biosphere, atmosphere, and society that could provide a better understanding and early warning of the state of the Earth system for human well-being.

The workshop aims to exchange on recent policy requirements, progress in providing EO-based data and products and equip participants with better knowledge and skills to analyze the drivers of deforestation and associated carbon emissions using remote sensing and Machine learning. The workshop aligns with recent European Union(EU) regulations to curb the EU market’s impact on global deforestation and provides valuable information for monitoring land use following deforestation, crucial for environmental initiatives and carbon neutrality goals.

Climate change is profoundly affecting the global water cycle, increasing the likelihood and severity of extreme water-related events. Droughts are becoming more frequent and intense. Extreme precipitation events are more localised and of unprecedented magnitude, causing widespread flooding and severe impacts on our lives and assets.
Accurately predicting and monitoring water-related environmental disasters, as well as optimal water resource management, require better decision support systems. These systems should integrate remote sensing, in-situ and citizen observations with high-resolution Earth system modelling, artificial intelligence, information and communication technologies, and high-performance computing.
Within the Digital Twin Earth for Hydrology and the Open Earth Monitor Cyberinfrastructure projects, we have developed advanced interactive tools for building what-if scenarios for flood risk assessment, drought monitoring and water resources management. The workshop will describe the developed tools (current version here: https://explorer.dte-hydro.adamplatform.eu/) and the recent advances developed within the Open Earth Monitor Cyberinfrastructure and related projects. An interactive session will be held to demonstrate the potential and limitations of the developed what-if scenarios.

Spatiotemporal prediction of SOCD for Europe (2000–2022) in 3D+T

While various imputation methods are available to reconstruct gappy time series of images, most of them are inadequate for large datasets like the full Landsat archive.
To address this need, this work proposes a new methodology called seasonally weighted average generalization (SWAG). SWAG works solely on the time dimension, reconstructing images by employing a weighted average of available samples in the original time series. It prioritizes images collected at integer multiples of a year to enforce annual seasonality and gives higher weights to more recent images to avoid propagating land cover changes. The method is implemented as part of the open source Python package scikit-map and optimized for computational efficiency.

The UN has declared this to be the Decade of Ecosystem Restoration, which should foster the development of restoration projects in many parts of the world suffering from land degradation. In parallel, there is growing demand for deforestation-free and sustainably produced products, as reflected partly by the establishment of the new EU Regulation on Deforestation-free products. The combination of these trends will likely lead to local land use changes resulting in increases in landscape heterogeneity. Here we place an interest in the effects that such changes have on biophysical variables that directly impact the Earth system and the local climate, such as short-wave radiation, land surface temperature and evapotranspiration, as estimated diurnally from geostationary satellite observations. In this study, we explore how the tree density and tree spatial arrangement in different ecosystems of the African continent have an impact on the energetic budget at local and regional scales. We perform a space for time analysis where local changes on vegetation are used to disentangle the effect of land cover transitions on biophysical variables. We expect the results of the study to provide insights into where increasing landscape complexity may provide additional benefits in terms of ecosystem services and thereby contribute towards guidelines in sustainable land planning.

Please provide an abstract of your talk as soon as possible

Access methods and processing pipeline of Landsat bi-monthly, complete and cloud optimized collection

Machine Learning is commonly used to map environmental variables in 2D, but what about generating predictions of dynamic variables such as above ground biomass, forest species, soil carbon and similar? The difference between spatiotemporal vs purely 2D / 3D mapping is in the three main aspects: (1) points and covariate layers are matched in spacetime (usually month-year period or at least year), (2) covariate layers are based on time-series data and include also accumulative indices (e.g. cumulative rainfall, cumulative snow cover, cumulative cropping fraction and similar) and derivatives, (3) during model training and validation, points are subset in both spacetime to avoid overfitting and bias in predictions. The rationale for using spatiotemporal machine learning is fitness of data for reliable time-series analysis: the predictions for anywhere in the spacetime cube need to be unbiased, with objectively quantified prediction errors (uncertainty), so that hence changes can be derived without a risk for serious over-/under-estimation. We have tested this framework on local and regional data sets (e.g. LUCAS soil samples covering 2009, 2012, 2015, 2018 for Europe) and can be now potentially applied using global compilations of soil points (https://opengeohub.github.io/SoilSamples/). Spatiotemporal machine learning could also potentially be used for predicting future states of soil, e.g. by extrapolating models to future climate scenarios and future land use systems (Bonannella et al., 2023).

Drought monitoring across scales is increasingly feasible with the use of open data. Multiple missions dedicated to monitor specific variables as indicators of the status of the earth system contribute to the growing availability of earth observation datasets. Soil moisture is one of these key indicators to monitor the status of drought.

However, drought, as a process dependent on multiple conditions from the atmospheric scale to the local land surface scale, expresses itself as a pattern of patterns. This nested nature consisting of vast anomalies conditioned in fragments, frequently complicates the characterization of drought from only one type of observations (e.g. ground data or only certain scale of remote sensing observations). Therefore, soil moisture data at multiple spatial scales are needed.

Currently, soil moisture datasets cover a reasonably wide range of scales to enable the monitoring of drought from continental to local scale. Multiple products exist to cover the monitoring of soil moisture anomalies with resolutions in the order of tens of kilometres either from active and passive radiometric technologies like ASCAT (Advanced SCATterometer) and the European Space Agency - Climate Change Initiative (ESA-CCI) products. Similarly, the pursuit of high-resolution observations is already evidencing the advantage of high-resolution data such as that of Sentinel-1 mission for dealing with the small-scale heterogeneity. Evaluation of these two scales of available data over Europe and Italy serve as examples of their suitability for multiple drought applications, also in an operational context

For this study we benefit from the Open Earth Monitoring Cyberinfrastructure project aiming to democratize the use of earth observations open the path to generalize the integration of open datasets across scales. Overall, our goal is to support this initiative and improve the comparison and combination of open data sources. This is crucial for addressing the multi-scale challenges of earth system sciences.

The achievement of ambitious LULUCF mitigation targets for 2030 and the EU 2050 climate
neutrality goals strongly rely on forests. Currently, there is a large discrepancy in data for monitoring
the status of EU forests with large differences across sources of information. In particular, remote
sensing data and national statistics are not sufficiently detailed and consistently integrated to allow
for comprehensive monitoring of forest status and consistently modelling biomass and carbon over
time, by showing a latency in capturing changes in forest cover and forest biomass.
ForestNavigator aims at modelling a series of forest sector policy pathways aligned to EU climate
neutrality goals. These pathways rely on integrating various data sources, including high resolution
remote sensing derived datasets (forest area, disturbances), ground data sources (NFI structural
data) and national statistics (forest harvest and products). In ForestNavigator, we consistently
combine these sources allowing for for a consistent representation of forests and forest sector
status featured in forest biophysical and socioeconomic models. Additionally, ForestNavigator
develops workflows that enable to timely update mitigation pathways according to near-real time
detection of changes in forests and in the forest bioeconomy. This near-real time update of policy
pathways, according to the continuously changing conditions, enables to timely correct efforts for
achieving policy mitigation targets. We present recent developments ongoing in ForestNavigator
project for a model-data fusion towards the assessment of EU consistent forest policy pathways.

GEO now since 2 decades has been working, to advocate Earth Observations Open data and Open knowledge, it is urgent to make sure that users are able to discover, access and re-use the available open applications, enhance knowledge sharing and solve most urgent countries socio environmental issues. The GEO Knowledge Hub is a promising tool to enhance knowledge sharing among the scientific community and accelerate the impact that EO Data and EO Knowledge can have.

Satellite imagery is traditionally stored and processed on rectangular grids. However, the widespread usage of such grids has normalized their inherent distortions, particularly near the poles. Previous attempts to address this issue, such as employing multiple local projections like the UTM-based Sentinel 2 L1C grid, have led to inefficiencies, including a significant increase in data volume (~30%) due to overlaps that need to be stored, downloaded, and processed. Additionally, there is a lack of a unified global indexing system and the choice of pixel cell shape, which further complicate the analysis.

In this keynote talk, we advocate for a paradigm shift towards Discrete Global Grid Systems (DGGS) to mitigate these challenges. DGGS tessellate the Earth's surface with hierarchical cells of equal area, minimizing distortion and reducing loading time of large geospatial datasets. This approach would greatly improve spatial statistics and convolutional Machine Learning models, where accurate representation of global phenomena is paramount at a global scale.

An abstract needs to be submitted as soon as possible

Evidence-based policy is gaining importance, also in the environmental policy domain in Flanders, Belgium. However, the most prevalent source of policy-relevant information still remains ground sampling, with limited spatial and temporal detail and coverage. The ease of access to freely available (Sentinel) satellite imagery from the Copernicus program through the new OpenEO API provides a golden opportunity for filling this information gap. During the GEO.INFORMED project, remote sensing and deep learning researchers engaged in a co-creation trajectory with regional environmental policy makers to develop machine learning workflows for transforming Copernicus satellite data into policy-relevant geodata. The main challenges encountered in the project where associated with ensuring mutual understanding between scientists and policy-makers; and with the technical implications of non-standard model inputs and limited reference data availability. Within the project, a range of strategies for overcoming these challenges were tested, and the lessons learned will be the main focus of this talk.

Policies on opening satellite image archives have shifted earth observation to the big data era. However, due to the associated data-hungry analytics, such as deep learning, satellite observations have to be combined (trained and then validated) with a large amount of in-situ data to get meaningful results. Yet, the collection of in-situ data is often laborious, and the resulting observations are rarely open for others to use. To bridge this in-situ data gap, this workshop will analyze the suitability of a citizen science mobile app for measuring biomass and tree species of individual trees and forest plots, i.e., the TreeQuest and ForestQuest modules, respectively. The app has been developed by the International Institute for Applied Systems Analysis (IIASA) and will be freely available for Android and iOS phones by the workshop. We will first present the app and then initiate a citizen science campaign motivating the conference participants to take part by testing the app and surveying selected trees around the conference center. Members from TU Wien will measure and model selected trees using a terrestrial laser scanner. The resulting 3D point cloud will allow the extraction of detailed information on vegetation structure, which will be used for comparison with the mobile app and forest inventory measurements acquired with traditional forest measurement tools (e.g. caliper, vertex). Finally, we will present the results and discuss the performance and potential further development of the app with workshop participants.

The workshop will also discuss the relevance of collected data and the approach for the two ongoing initiatives such as (a) the Citizens for Copernicus project that is funded by the Austrian Research Promotion Agency, application No. 47907528, and (b) the Open Earth Monitor Cyberinfrastructure project funded from the European Union's Horizon Europe research and innovation programme under grant agreement No. 101059548.

Global Trait-Based Vegetation Monitoring: Leveraging Multispectral Imagery for Restoration Project Assessment

Restoration projects are crucial for ecosystem recovery and biodiversity conservation, but their large-scale monitoring poses significant challenges. Conventional approaches often rely on intensive manual work, incur high costs and need help with standardisation, making monitoring on a global scale impossible. Public satellite missions such as Sentinel-2 have great potential to transform ecosystem monitoring due to their high spatial and temporal resolution when linked directly to ecosystem characteristics. Here, we present several global, high-resolution (20m) maps of vegetation traits derived from Sentinel-2 multispectral imagery, reflecting the mean trait value during the vegetation period at annual intervals from 2019 onwards. Using a hybrid inversion approach of the physically-based radiative transfer model PROSAIL, we estimate leaf functional traits (e.g. chlorophyll content, equivalent water thickness, or leaf mass per area) and canopy structural traits (e.g. leaf area index). Validation using in-situ data suggests that the trait maps can effectively track local temporal changes. Further, we show how the generated trait maps can map functional trait diversity at a coarser resolution. Altogether, these products provide deeper insights into ecosystem health, biodiversity status and restoration efforts.

Various modelling techniques are available to understand the temporal and spatial variations of the phenology of species. Scientists often rely on correlative models, which establish a statistical relationship between a response variable (such as species abundance or presence-absence) and a set of predominantly abiotic covariates. The choice of the modelling approach, i.e., the algorithm, is a crucial factor in addressing the multiple sources of variability that can lead to disparate outcomes when different models are applied to the same dataset. This inter-model variability has led to the adoption of ensemble modelling techniques, among which stacked generalisation, which has recently demonstrated its capacity to produce robust results. Stacked ensemble modelling incorporates predictions from multiple base learners or models as inputs for a meta-learner. The meta-learner, in turn, assimilates these predictions and generates a final prediction by combining the information from all the base learners. Our study utilized a recently published dataset documenting egg abundance observations of Aedes albopictus collected using ovitraps. This dataset spans various locations in southern Europe, covering four countries -Albania, France, Italy, and Switzerland- and encompasses multiple seasons from 2010 to 2022. Utilising these ovitrap observations and a set of environmental predictors, we employed a stacked machine learning model to forecast the weekly average number of mosquito eggs. This approach enabled us to i) unearth the seasonal dynamics of Ae. albopictus for 12 years; ii) generate spatio-temporal explicit forecasts of mosquito egg abundance in regions not covered by conventional monitoring initiatives. Beyond its immediate application for public health management, our work presents a versatile modelling framework adaptable to infer the spatio-temporal abundance of various species, extending its relevance beyond the specific case of Ae. albopictus.

Cocoa cultivation serves as a crucial source of income for countless farmers across pantropical regions. However, this agricultural practice often leads to deforestation in tropical forests. While previous studies have highlighted the expansion of cocoa farms, particularly in select African countries, there remains a significant gap in comprehensive data regarding the location of cocoa farms on a pantropical scale. To address this challenge, our study employs deep learning models trained on Sentinel-1 and Sentinel-2 satellite imagery, coupled with annotated reference datasets, to map cocoa farms across pantropical regions.
Our findings provide valuable insights for governments, cocoa companies, consumers, NGOs, and international organizations striving to mitigate the challenges associated with escalating deforestation linked to cocoa production. Of particular significance is the utility of this dataset in addressing the recent European Union Regulation mandating companies to refrain from importing commodity crops associated with deforestation. By providing a comprehensive understanding of cocoa farm distribution across pantropical regions, our research contributes to informed decision-making and sustainable practices in cocoa production and trade.

This workshop will explore the high-resolution mapping of gross primary productivity (GPP) using light-use efficiency models. During the workshop, we will cover how to access the bi-monthly GPP maps and assess the accuracy of the maps via eddy covariance flux measurements. Participants will gain insight on how to exploit high-resolution GPP maps across diverse ecosystems.

The EarthCODE (Earth Science Collaborative Open Development Environment) vision provides an integrated, cloud-based, user-centric development environment which can be used to support the European Space Agency’s (ESA) science activities and projects. Building on activities that developed the European EO open-source ecosystem and the Open Earth System Science community (e.g. EOEPCA - Exploitation Platform Common Architecture, DeepESDL - Deep Earth System Data Lab, openEO Platform, ESA Euro Data Cube, etc.), ESA is implementing EarthCODE as a collaborative platform for conducting Earth System Science sustainably and adhering to FAIR and Open Science Principles. EarthCODE will enable the long-term persistence of research outputs from science activities.

EarthCODE looks to maximise reproducibility, reuse, and consumption of research outputs by the wider community, promoting a flexible and scalable architecture developed with interoperable open-source blocks, with a long-term vision evolving by incrementally integrating industrially provided services from a portfolio of the Network of Resources. EarthCODE platform collaborators will participate in creating integrated architecture, with interoperable solutions and federated capabilities.

EarthCODE will use EOEPCA Open Standards to help support Open Science, and help drive these standards. Open science principles are increasingly being embraced in the field of Earth Sciences, promoting transparency, collaboration, and accessibility of research. This is being done by promoting open access publications, preprints and open review processes, sharing data/methodologies for verification, reproducibility and reuse. In software development, these principles allow inspection, modification, and code contribution, encouraging collaboration among researchers through various platforms (i.e. GitHub, GitLab, etc.). Sharing of educational resources openly allow for global audience, and involvement of the public through citizen science for scientific research.

EarthCODE will provide an Integrated Development Platform, giving developers the tools needed to develop high quality workflows that allow experiments to be executed in the cloud and the reproduced by other scientists, following Open Science principles. Our solution is built around existing open-source solutions and building blocks, primarily the Open Science Catalogue, EOxHub and EOEPCA. With it’s adopted federated approach, EarthCODE will have the capability to facilitate processing on other platforms, i.e. DeepESDL, ESA EURO Data Cube, Open EO Cloud/Open EO Platform and AIOPEN/AI4DTE.

Collaboration and Federation are at the heart of EarthCODE. As EarthCODE evolves we expect providing solutions allowing allow federation of data and processing. EarthCODE has ambition to deliver a model for a Collaborative Open Development Environment for Earth system science, where researchers can leverage the power of the wide range of EO platform services available to conduct their science, while also making use of FAIR Open Science tools to manage data, code and documentation, create end-to-end reproducible workflows on platforms, and have the opportunity to discover, use, reuse, modify and build upon the research of others in a fair and safe way. EarthCODE thus aims to make possible the eight enabling elements of the EO Open Science and Innovation vision: open data, open-source code, linked data & code, open access documentation, end-to-end workflows reproducible on platforms, open science resources, open science tools, and a healthy community applying all the elements in their practice.

The workshop starts with accessing a global ensemble digital terrain model, cropping to tile and reprojecting to Equi7 projection system. Secondly, the attendants will set up a docker containing GRASS, SAGA and Whtieboxtool with R/Python that enables to script land relief parameterization process. Lastly, attendant will follow a workflow that produces different land relief parameters by tiles and mosaics with consideration of boundary effect, in order to achieve high-resolution global scale terrain parameter mapping.