ESA title

Five takeaways from the Agentic AI for Earth Observation workshop

The first edition of the ‘Agentic AI for Earth Observation’ workshop explored how Earth observation data and products can boost agentic AI decision intelligence systems and how agentic AI could transform the way Earth observation data are accessed, analysed and used. Bringing together more than 500 attendees, the event highlighted the field’s rapid move towards practical applications, while emphasising the still open scientific questions, the need for human oversight, trusted ecosystems and solutions that deliver measurable value for the society.

Held in Berlin (Germany) from 19 to 21 October, the inaugural ‘Agentic AI for Earth Observation’ workshop brought together more than 500 researchers, practitioners, end users and decision-makers from artificial intelligence, geospatial analytics, climate and weather science, remote sensing and space mission.

Organised by ESA Φ-lab and the Berlin Institute for the Foundations of Learning and Data (BIFOLD), this workshop promoted exchanges between academia and the industry. Participants explored cutting-edge research, practical challenges and emerging opportunities, with a focus on how autonomous agents could transform the way geospatial and satellite data are acquired, processed and used.

In addition, the workshop also highlighted how Earth observation can provide the observations and geospatial context needed by agentic AI to support decision intelligence.

Held in Berlin from 19 to 21 October in Berlin (Germany), the Agentic AI for Earth Observation workshop was jointly organised by ESA Φ-lab and the Berlin Institute for the Foundations of Learning and Data (BIFOLD). From left to right: Nicolas Longépé (Earth Observation Data Scientist at ESA Φ-lab), Volker Markl (BIFOLD Director and Professor at the Technical University of Berlin), Claudio Iacopino (Head of the ESA Φ-lab Explore Office), Henry Marx (Permanent Secretary for Science of the State of Berlin), Begüm Demir (BIFOLD Deputy Director and Professor at the Technical University of Berlin), Fatma Deniz (President of the Technical University of Berlin), Klaus-Robert Müller (BIFOLD Director and Professor at the Technical University of Berlin), and Giuseppe Borghi (Head of the ESA Φ-lab Division). Credits: BIFOLD/Michael Setzpfandt.

The strong turnout highlighted the momentum around this topic. Here are five takeaways from the event:

Agentic AI for Earth observation is moving from exploration to practice

    Earth observation is rapidly moving beyond individual workflows towards connected systems that are increasingly autonomous and capable of handling complex tasks. Combining Earth observation data with geospatial foundation models, large language models, AI agents and specialised tools allows for deeper analysis and longer workflows.

    Earth observation specialisation remains essential: models and agents designed for the sector can perform better on specialised tasks than general-purpose solutions.

    The workshop also showed that agentic AI is already moving into practice, with many working prototypes available for users to explore and test, namely in utility/infrastructure monitoring, semantic object discovery without labels, crop/vegetation inspection, satellite image discovery using natural language, and Earth observation image retrieval and similarity search.

    We need the ‘right’ autonomy, not the maximum autonomy

    The aim of agentic Earth observation systems is not autonomy for its own sake, but to delegate decisions to agents where autonomous action brings value, while retaining human oversight, rigorous scientific evaluation, and system-level constraints where needed.

    Autonomous systems must also know when to complete a task, use a tool, delegate a task to another agent or stop the task entirely. More agents and tool calls do not necessarily produce better results.

    As these systems become more complex, ensuring that their workflows are transparent and reproducible, according to the FAIR (Findable, Accessible, Interoperable and Reusable) framework, is an emerging challenge. Provenance and scientific scrutiny must therefore cover the entire workflow, from the original data to the result.

    Agentic Earth observation requires ecosystem-level trust and technological sovereignty

    Agentic Earth observation is evolving from individual agents towards distributed systems in which multiple agents work together. This requires agents to be discoverable by other agents and orchestration systems, interoperable, and reusable across platforms and organisations.

    Increasing trust in these agents will require clear safeguards for identity, data provenance, permissions and accountability. As with AI in general, technological sovereignty is also an important consideration, spanning data, infrastructure, algorithms and models. In agentic systems, this extends to the services, tools and external agents on which they may depend. For Europe, this means identifying and addressing dependencies and vulnerabilities within agentic systems.

    Bringing real value to society is what matters most

    Agentic AI could make Earth observation more accessible by enabling users to formulate objectives rather than manually defining every step of an Earth observation workflow. Agents could discover relevant observations, select appropriate processing and analytical tools, coordinate their execution, and translate the resulting information into actionable insights. This could strengthen the integration of Earth observation into domains such as disaster management, climate resilience, insurance, environmental monitoring and scientific research.

    Such agent-driven workflows should remain transparent, traceable and scientifically robust, particularly when their outputs inform decisions. The success of an agentic system should not be measured by the number of agents or tool calls, or by the level of autonomy achieved. What really matters is building trusted, scientifically grounded systems that can deliver measurable value for users, science and society.

    The future lies in building the next generation of Earth Intelligence

    The next step is to move beyond individual proofs-of-concept and develop shared frameworks that can support agentic AI for Earth observation in real-world operations. This will require community benchmarks to assess whether agents are trustworthy, scientifically sound and ready for practical use. It will also mean connecting data, models, tools, computing resources and space- and ground-based assets more effectively.

    Throughout the development of these systems, it will be important to distinguish emerging research from demonstrated capabilities and operational or commercial solutions. Human oversight and Earth observation expertise must remain at the heart of every workflow.

    In the end, the goal is to create a new generation of Earth Intelligence systems that can observe, understand, decide and act, while operating within clearly defined constraints and accepted levels of risk.

    The Agentic AI for Earth Observation workshop brought together more than 500 researchers, practitioners, end users and decision-makers from artificial intelligence, geospatial analytics, climate and weather science, remote sensing and space mission operations. Credits: BIFOLD/Michael Setzpfandt.

    Claudio Iacopino, Head of the ESA Φ-lab Explore Office, reflected on the outcomes of the workshop: “Agentic AI has great potential to make Earth observation more accessible and useful, and this workshop provided insights that will help us move towards real-world adoption. But we need to keep in mind that the autonomy of these systems must go hand in hand with human oversight and scientific validation. Our goal is not simply to build more capable systems, but actually create trusted tools that deliver meaningful benefits for society.”

    Begüm Demir, BIFOLD Deputy Director and Professor at the Technical University of Berlin, said: “The goal of this workshop was to strengthen the community’s skills and expertise in the design and development of efficient, effective, and trustworthy AI agents for Earth observation. The lively discussions and tremendous interest from participants showed us that we had achieved this goal. We are delighted with the outcome.”

    To know more: Agentic AI for Earth Observation workshop, ESA Φ-lab, Berlin Institute for the Foundations of Learning and Data (BIFOLD)

    Photo courtesy of BIFOLD/Michael Setzpfandt.

    InCubed-funded terrAIntel becomes fully operational

    terrAIntel turns complex Earth observation data into accessible information through the use of natural language. As one of the first ‘ChatGPT-like’ tools for Earth observation co-funded by the European Space Agency, terrAIntel has evolved into an operational platform that allows anyone to explore satellite-derived insights, maps and statistics without requiring specialist expertise. The platform supports applications ranging from journalism and storytelling to humanitarian and disaster-response activities.

    From tracking the impact of floods and wildfires to monitoring changes in cities, forests or coastlines, Earth observation data can help journalists tell stories with facts and evidence. Yet, accessing and understanding this information is often far from straightforward.

    Finding the right data, making sense of it and turning it into information that can be used in reporting usually requires specialist knowledge, dedicated software and a significant amount of time. As a result, many newsrooms and public-interest organisations might end up not using Earth observation data in their reporting, even when it could strengthen their stories and support informed decisions.

    Developed by GeoVille and Cortecs and co-funded by ESA InCubed, terrAIntel is a platform that makes Earth observation information easier to access and understand, helping data journalists and digital investigative journalists find the information they need. An early version of the platform is now available.

    Through a simple conversational interface, users can ask questions about places, events or environmental changes in their own words and receive answers supported by maps, statistics and accessible explanations.

    Visual representation of the terrAIntel user interface. terrAIntel allows users to interact with various Earth Observation datasets through a natural language interface. Users can ask questions about specific locations or environmental phenomena, and the platform translates these requests into the appropriate geospatial analyses, presenting the results through maps, visualisations relevant statistics and data provenance information. Credits: GeoVille.

    They can define the location and time period they want to explore, helping them focus on the most relevant information for their stories. By combining Earth observation data, visualisation tools and transparent information about data sources and methods, terrAIntel transforms complex data into evidence that can be readily used for reporting, storytelling and decision-making.

    Although it was initially developed for journalists, terrAIntel can now support humanitarian efforts and disaster-response activities: public health officers and emergency response teams can explore flood-affected areas, investigate changes in land cover, identify settlements potentially exposed to environmental hazards, or monitor wildfire activity.

    terrAIntel can also support the planning of vaccination campaigns in remote areas by helping humanitarian teams identify and locate villages, estimate their populations and use this information to plan vaccination requirements.

    To raise awareness about the platform and explore new applications, a team from GeoVille showcased terrAIntel at AI WEEK and at ESA Industry Space Days, demonstrating how Earth observation data can be explored through natural language and engaging with potential users from different sectors.

    The terrAIntel team was present at AI WEEK. From left to right: Donvan Grobler (Project Manager at GeoVille), Markus Tretzmüller (Managing Director at Cortecs) and José Manuel Delgado Blasco (Geospatial System Engineer and InCubed Officer for terrAIntel). Credits: Cortecs.

    According to GeoVille, the story behind the development of terrAIntel resonated with many attendees, leading to discussions about the future of the platform and opportunities for interested entities to try it themselves. Their feedback was particularly encouraging as the team continues its efforts to bring terrAIntel to a wider market.

    Donvan Grobler, Project Manager at GeoVille, commented: “Earth Observation has enormous potential to help us better understand our planet and address some of the challenges we face, but accessing and interpreting this information still involves considerable technical barriers. With terrAIntel, our ambition is to change that by making satellite-derived intelligence accessible through a simple, conversational interface.”

    Recognising the role of InCubed in the development of terrAIntel, Donvan added: “The funding and mentoring received through ESA’s InCubed programme have been instrumental in helping us turn this vision into a working platform, providing the opportunity to develop and validate our approach and explore how Earth observation can become more accessible and actionable for a much broader audience, ultimately amplifying its impact.”

    Gabriele Meoni, ESA Technical Officer for terrAIntel, said: “Satellite data can provide powerful insights, but extracting them often involves multiple processing steps and specialised knowledge. Within terrAIntel, AI automates much of this work behind the scenes, allowing it to deliver relevant information in a faster and more accessible way. This is an important step towards making Earth observation data available to a much broader community of users.”

    José Manuel Delgado Blasco, InCubed Officer for terrAIntel, added: “Through InCubed, ESA helps companies transform innovative concepts into market-ready services. terrAIntel illustrates how this approach can support business growth while making Earth observation information more accessible, useful and relevant to users across a wide range of sectors. terrAIntel is the first activity of its kind to be completed with InCubed support, and I am eager to see where commercial Earth observation innovation will bring us next.”

    GeoVille continues to refine terrAIntel, including expanding its analytical capabilities, integrating additional datasets and improving the overall user experience. In parallel, GeoVille is actively engaging in business development activities to grow its user base and establish a commercially sustainable service.

    With a fully operational platform now in place, terrAIntel is well positioned to expand its reach and the impact of accessible Earth observation information across new sectors and applications.

    Find more information and explore terrAIntel on its dedicated platform.

    To know more: terrAIntel, terrAIntel InCubed Activity Webpage, ESA InCubed.

    Photo courtesy of Cortecs.

    Introducing ESA’s Earth Virtual Expert

    The European Space Agency (ESA) has released Earth Virtual Expert (EVE), a platform powered with Artificial Intelligence (AI), to make Earth observation information more accessible and easier to understand. Built on trusted scientific sources and created to answer questions, EVE brings complex research, data and expertise closer to researchers, policymakers, journalists and the wider public.

    EVE is a platform that provides information in a question-and-answer format, making it accessible for all levels of expertise. Users can ask questions such as: How has vegetation in southern Europe changed over the past decade? or how do satellites help monitor greenhouse gas emissions?

    Earth observation data comes in many forms, including scientific papers, reports, technical documentation, models and repositories. Finding, comparing and understanding this information can be difficult, often requiring specialist expertise. This limits its accessibility to policymakers and the wider public.

    To help bridge this gap, ESA Φ-lab partnered up with Pi School – an AI research and innovation company – as well as with two private companies: French-based AI company Mistral and Imperative Space, a multi-disciplinary company that works in innovation, communications and education for the space sector. Together they developed EVE, a large language model (LLM), which summarises, translates and presents huge amounts of data in accessible ways.

    Read the full article on www.esa.int.

    Exploring EVE with two new humanoid robots

    Over the last years, ESA Φ-lab and Pi School have carried out the development of EVE (Earth Virtual Expert), a new intelligent companion for exploring the world of Earth observation and Earth sciences, with support from Imperative Space, Mistral AI and Wiley. Two new humanoid robots, Ameca and Frankie, make interactions with EVE more accessible and engaging.

    Navigating today’s vast influx of Earth observation and Earth sciences data and information can often feel like searching for a needle in an endless digital haystack. From tracking climate patterns to monitoring crop health, Earth observation offers valuable insights. Uncovering these insights, however, usually requires specialised, technical expertise.

    Two new humanoid robots show that exploring our ‘pale blue dot’ does not have to be complicated. Rather than sitting behind a computer and deciphering dense code, people can now explore EVE (Earth Virtual Expert), a chat-based platform designed to revolutionise how we interact with Earth observation and Earth sciences, simply by starting a conversation.

    Meet the team: Ameca and Frankie

    Developed by Frank Reply GmbH, Ameca and Frankie are two humanoid robots created to make interactions with EVE more accessible and engaging. Equipped with built-in cameras for real-time visual analysis of images on a screen, this robotic duo is able to interpret queries in natural language.

    While EVE does the heavy lifting of Earth observation data analysis, these two new guides offer a fully interactive experience.

    Ameca is the expert for satellite image searches. People can speak directly to Ameca, who interprets plain-language requests to locate and retrieve the most relevant satellite imagery from EVE’s database. Frankie is the enthusiastic expert for vegetation analysis. By looking at a satellite image on a screen, he breaks down Earth observation concepts such as NDVI (Normalised Difference Vegetation Index) into engaging conversations.

    Ameca, the expert for satellite searches, interprets plain-language requests to locate and retrieve the most relevant satellite imagery. Credits: ESA.

    Daniel Fitzpatrick, Senior Manager at Frank Reply, commented: “By using the conversational interface, we are able to democratise Earth observation operations. In the past, due to the vast amounts of data involved and scientific expertise necessary to execute successful queries, highly specialised knowledge was a pre-requisite.”

    “Using natural language, people can simply ask for specific Earth observation imagery (for instance of their hometowns) and then watch in real-time as their requests are rendered into machine language using the EVE toolchains. We are looking forward to continuing our collaboration with ESA Φ-lab and will be hosting a demo and hands-on session at the ‘Agentic AI for Earth Observation’ workshop in Berlin this October”, Daniel added.

    A look under the hood

    These robots offer three ways to experience EVE’s technology: an interactive image recognition game guided by Ameca, an NDVI-based crop-health inspector led by Frankie, and the EVE ‘Behind the Scenes’ mode.

    For those curious about how intelligent assistants process complex requests, the ‘Behind the Scenes’ mode shows how EVE’s ‘brain’ works. As people speak to Ameca, a dedicated screen reveals every step of EVE’s operational workflow in real time, showing retrieval, reasoning, planning and tool calls as they happen.

    Making data transparent, approachable and fun

    By fusing an advanced, low latency speech-to-speech AI model with EVE’s underlying agent capabilities, satellite workflows and grounded data retrieval, Ameca and Frankie transform a complicated data science task into a fluid, natural conversation.

    Ultimately, this ‘sneak peek’ of EVE shows that conversational interfaces are going beyond simplifying technology and into making Earth observation transparent, approachable and fun. By allowing users to talk their way through satellite data and get a real feel for space technologies without having to understand them first, EVE is making the wonders of Earth observation and Earth sciences open to everyone.

    “The real breakthrough here is how these conversational interfaces turn complex Earth observation technology and knowledge into a natural dialogue. We no longer need to speak the specialised language of remote sensing to understand what the data are telling us”, commented Nicolas Longépé, Earth Observation Data Scientist at Φ-lab and EVE’s Technical Officer.

    “Anyone, from curious citizens to policymakers, can now simply ask a question and explore satellite intelligence, just like talking to a friend. Besides making EVE more transparent by showing its inner workings, these conversational interfaces also represent a major step forward in truly democratising Earth observation knowledge”, Nicolas added.

    Do not miss EVE’s demo and hands-on session at the ‘Agentic AI for Earth Observation’ workshop (Berlin, Germany) in October. A fully operational version of EVE will be launched soon. Register for the public launch and read more about it on the EVE website.

    To know more: ESA Φ-lab, Frank Reply GmbH.

    Photo courtesy of Unsplash/Gabriele Malaspina

    ESA Φ-lab calls for collaborators to expand the Earth Observation Virtual Constellation

    Earth observation is entering a new era where space and ground assets are expected not only to collect data, but also to autonomously collaborate, reason and respond to evolving user needs in near real-time. Supported by ESA Φ-lab, the Earth Observation Virtual Constellation initiative combines onboard AI, ground-based intelligence and multi-agent orchestration to enable satellites, in-situ sensors and distributed processing resources to dynamically coordinate their activities. ESA Φ-lab is now seeking to expand the ecosystem with new space and terrestrial assets, and additional stakeholder communities.

    Signals that anticipate natural disasters often present themselves hours before humans can react. In times marked by rapid environmental changes, the Earth observation field faces an urgent demand for faster, targeted and information-driven responses to dynamic events.

    Today, even with advanced satellites in orbit, effective disaster response remains restricted by rigid ground segment workflows. Traditional data acquisition schemes, while powerful, lack the agility to react in real time to fast-evolving events like floods or wildfires.

    Even as onboard artificial intelligence (AI) begins to be deployed, it typically operates on static instructions, waiting days in advance to execute pre-scheduled tasks over fixed areas. This disconnection highlights a gap in on-ground coordination and real-time responsiveness.

    To meet these challenges, a paradigm shift is underway, moving from satellites controlled by operators to an intent-driven cognitive system. In this model, satellites do not wait for ground operators to send commands – they directly trigger an action. A system of satellites will then autonomously observe, refine its actions and confirm outcomes in real time.

    Meet the Earth Observation Virtual Constellation (EOVC), a novel, federated and software-defined approach that orchestrates different space assets and ground-based data sources to operate as a single and unified structure, and that is at the heart of this paradigm shift.

    Rather than relying on fixed acquisition plans from isolated missions, this new virtual constellation will treat all available sensing resources – including public satellites, commercial constellations and ground-based Internet-of-Things networks – as interoperable nodes within a cloud-native ecosystem. This will shift the core operational question from ‘Which satellite acquires the data?’ to ‘What information is needed and how can the system best obtain it?’

    By leveraging agile ground segments, cloud-native interfaces and agentic AI technologies, satellites can automatically share insights with one another, forming an on-demand constellation that not only responds to active events by recruiting specific sensors, acquiring data from different orbits and using various satellite capabilities, but also anticipates future events.

    ESA Φ-lab and the EOVC consortium, led by the CS Group, are looking for new collaborations to expand EOVC capabilities. This non-procurement initiative aims to identify key stakeholders, assess existing space and ground capabilities, foster cross-industry exchanges and collaboratively shape the future system’s architecture.

    Organisations interested in contributing to this initiative are invited to express their interest and share relevant capabilities, technologies, and potential collaboration opportunities.

    The initial demonstration phase of the Earth Observation Virtual Constellation will focus on flood monitoring across selected European river basins, concentrating specifically on the Danube River and the French national river network.

    Subsequent phases will extend these capabilities to new spacecrafts, new geographical regions and a broader array of time-sensitive use cases, including disaster and emergency response, maritime surveillance and safety, and extreme weather and infrastructure resilience.

    Nicolas Longépé, Earth Observation Data Scientist at ESA Φ-lab and Technical Officer of the initiative, commented: “For decades, we have looked at Earth observation as a collection of individual satellites, each doing its own job. But the challenges we face today, from floods and wildfires to other rapidly evolving events, require a more connected approach.”

    “Virtual constellations such as this one will allow different space and ground assets to work together as a team, sharing information about what is happening on Earth faster. In the end, this is about making Earth observation more responsive, more efficient and more useful for the people that depend on it”, Nicolas added.

    The video below explores how the Earth Observation Virtual Constellation will work and what it means for the future of Earth observation.

    More information about this Call for Interest and how to collaborate is available on the Earth Observation Virtual Constellation website.

    To know more: Earth Observation Virtual Constellation, ESA Φ-lab.

    Photo courtesy of CS Group.

    Eight new Earth observation-based wildfire applications co-funded by ESA InCubed

    As the result of the latest joint co-funding call launched by ESA InCubed and the Spanish Space Agency, eight new Earth observation projects have kicked off in 2026. These initiatives will enable risk anticipation, enhance early detection and optimise action in critical situations, contributing to saving lives, protecting infrastructure and preserving natural heritage.

    Wildfires are moving from being seasonal threats to some of the most destructive and unpredictable natural phenomena. Driven by heatwaves, prolonged droughts and shifting weather patterns, wildfires are growing in both frequency and intensity.

    In light of the devastating wildfires that have recently ravaged Spain’s territory, communities and ecosystems, the urgency to increase prevention, detection and response capabilities has never been more critical. As a direct response, ESA InCubed and the Spanish Space Agency (AEE) launched a call in 2025 to fund and accelerate Earth observation applications that span the entire wildfire management cycle.

    Eight new initiatives signed a contract with ESA InCubed and kicked off in July 2026, being awarded a total of € 3 million:

    • FIREAI-OPS — Intelligent Wildfire Operations through AI and Earth Observation

      Emergency agencies lack real-time data assimilation and interoperability across Earth observation, airborne and terrestrial sources, limiting their ability to anticipate fire evolution and coordinate appropriate responses.

      Implemented by GEOAI Analytics, Multiverse Computing and CATEC, FIREAI-OPS is an initiative that addresses the lack of integrated, rapidly updated information available to emergency teams managing fast-moving wildfires.

      The activity aims to combine multi-source Earth observation data, artificial intelligence (AI) and physics-based fire modelling within a common operational environment, helping authorities anticipate fire evolution, coordinate observations and strengthen situational awareness during emergency operations. 

      • FIREWISE — Foundation Models for Reliable Wildfire Intelligence on Board Satellites

        Wildfire alerts can be delayed by satellite downlinks and ground-processing loops, while narrow AI models (artificial intelligence systems designed to perform one specific task) may not remain reliable as sensors and environmental conditions change.

        FIREWISE is an activity that addresses delays created when satellite imagery is first transmitted and processed on the ground, as well as the challenge of maintaining reliable AI performance under changing observation conditions.

        Led by the ARQUIMEA Research Centre, FIREWISE aims to develop a software framework for adapting and operating Earth observation foundation models onboard satellites, enabling faster wildfire alerts with uncertainty information to support confidence in their use. 

        • EO4PYRO — Earth Observation for Predictive Wildfire Risk and Resilience Operations

        Authorities and infrastructure operators rely on fragmented, static and poorly interoperable information that constrains prevention, predictive risk assessment and post-fire impact evaluation.

        Headed by Centro de Observación y Teledetección Espacial, S.A.U. (COTESA), in a consortium with Hisdesat, Universidad de Alicante and IMIB-CSIC of Universidad de Oviedo, EO4PYRO focuses on the fragmented and often outdated information used for wildfire prevention, operational planning and post-fire assessment.

        EO4PYRO aims to integrate multi-source Earth observation data, AI, enhanced-resolution products and digital-twin capabilities in a cloud platform that supports risk assessment, vegetation monitoring, fire-behaviour analysis and evaluation of environmental impacts after an event. 

        • FirEOcast — FIRE weather instability and fuel estimation by Earth Observation for improved fire danger foreCASTing

        Conventional fire-danger indices depend largely on meteorological variables and insufficiently represent atmospheric instability, fuel properties and fuel-moisture conditions.

        FirEOcast is a fire-danger service that aims to combine satellite-derived fuel information with meteorological modelling to deliver more dynamic and spatially detailed wildfire-danger information for prevention, preparedness and operational planning. 

        This initiative is led by isardSAT, with Universidad de Alcalá and MeteoGRID.

        • Deeper Vision Self-Trained AI (DVSTAI) — Thermal Infrared extension

        Ready-to-use platforms for locally configurable wildfire-risk prediction and monitoring of high-impact areas are limited, while multi-source processing can be computationally demanding and complex.

        Thales Alenia Space España S.A. is leading an extension of the DVSTAI cloud service that includes multispectral and thermal-infrared support, semantic-segmentation models, scheduled monitoring, risk alerts and potential-impact mapping.

        This extension addresses the limited availability of configurable operational tools that can transform recent satellite observations into detailed wildfire-risk and potential-impact information at local and regional scales.

        • PiroSAR — Fire-Front Detection and Evolution Monitoring (F2DEM) Service

        During major wildfires, smoke, cloud, darkness and operational restrictions can prevent optical, thermal and airborne systems from providing timely and reliable information on the active front and its evolution.

        Telespazio Ibérica S.L.U. is developing PiroSAR, a project that aims to use imagery from multiple Synthetic Aperture Radar (SAR) satellite constellations to provide all-weather, day-and-night information on active fire fronts, perimeter changes and fire evolution, supporting more informed tactical decisions by emergency management authorities. 

        • Horizon Wildfire Insights (HWI)

        Wildfire information from satellites, aerial systems and field reports is fragmented, preventing a unified operational picture and slowing anticipation, response and recovery decisions.

        Led by Airbus Secure Communications, S.A.U., Airbus GeoTech, Vexiza and B2Space, Horizon Wildfire Insights (HWI) addresses the fragmentation of satellite observations, aerial information, field reports and operational communications, which can leave emergency teams without a shared and timely picture of an evolving wildfire.

        HWI integrates Earth observation data, critical communications and intelligent analytical capabilities within a unified environment, supporting risk anticipation, active-fire monitoring, coordinated decision-making and assessment of affected areas across the wildfire-management lifecycle.

        • FireTwin — Operational Wildfire-Risk Forecasting

        Decision makers receive fragmented, static and outdated information that does not adequately capture changing vegetation condition, fuel moisture or evolving wildfire risk.

        FireTwin is an initiative being developed by Agresta Sociedad Cooperativa that aims to combine multi-sensor Earth observation and meteorological data through AI-enhanced models, producing regularly updated wildfire-risk information and developing short-term forecasting capabilities to help users identify critical areas, plan preventive action and allocate resources more effectively.

        Carlos Urbina Ortega, End-to-end System Engineer at ESA Φ-lab Invest Office and InCubed Officer for the management of this national campaign in Spain, commented: “Wildfires are becoming faster and more dangerous, threatening lives, homes and ecosystems. Through this InCubed call with the Spanish Space Agency, we are turning satellite data into real-world tools for every stage of a fire. These eight new applications will give emergency teams the exact information they need to react quicker, manage crises better and ultimately save lives and protect nature.”

        Learn more: ESA InCubed, Spanish Space Agency (AEE)

        Header: this image from Copernicus Sentinel-3, captured on 17 August 2025, is a wide view of France, Spain and Portugal and is dominated by smoke billowing from fires in the northern part of the Iberian Peninsula. Contains Copernicus Sentinel data (2025), processed by ESA.

        ESA and UKSA launch a new InCubed call for the United Kingdom

        Building on the success of previous InCubed calls, the European Space Agency is launching a new call for proposals, in collaboration with the UK Space Agency, through the ESA InCubed programme. The deadline for applications is 21 August 2026.

        The European Space Agency (ESA) and the UK Space Agency (UKSA) have witnessed the success of its joint 2024 InCubed call: more than £ 2.5 million were awarded to five innovative projects that use satellite data to improve public services, ranging from infrastructure monitoring and tracking methane emissions to supporting sustainable land management and safeguarding biodiversity.

        In 2026, ESA and UKSA join efforts again for a new round of applications. UK-based organisations are invited to submit proposals that demonstrate a clear route to market, credible commercial demand and a strong plan for operational deployment.

        Consortia are also encouraged to apply, including non-UK organisations that are supported by relevant national UK agencies or self-funded. Academic and/or research institutions are also welcome to apply, as long as they are part of an industry-led consortium and have no commercial interest in the product.

        Activities across the entire Earth observation value chain will be considered, including space segment, ground segment and data applications. They may align with either a de-risking cycle, supporting earlier-stage developments, or a product development cycle, supporting more mature activities that aim to deliver a commercially viable product or service.

        This call will support projects that are expected to last up to 18 months, with funding levels ranging from €300,000 to €4 million, depending on the scope and maturity of the activity. ESA’s support will be provided on a co-funding basis with up to 80% co-funding for SMEs and down to 50% for non-SMEs, which will once more depend on the activity’s risk and maturity.

        The process application consists of two stages: for the first stage, applicants need to submit an idea pitch, or outline proposal, via ESA’s ORBIT platform. ESA and UKSA will jointly assess submissions and invite selected applicants to a pitch session. For the second stage, those selected applicants will either submit the second part of their proposal through ORBIT or be told they were not successful. ESA will evaluate the proposals according to the InCubed criteria and UKSA will determine if a national delegate authorisation of funding can be provided.

        Proposals that show a clear path to commercialisation and demonstrate evidence of customer engagement will be preferred. The total funding amount is up to € 7.26 million and the deadline for applications is 21 August 2026.

        Michele Castorina, Head of the ESA Φ-lab Invest Office and InCubed Programme manager, commented: “Earth observation is undergoing a profound paradigm shift driven by rapid commercialisation and advanced technologies. This new 2026 InCubed call, delivered in partnership with the UK Space Agency, is essential for bridging the gap between innovative engineering and commercial market success. By focusing on de-risking and validation of real customer demands, call such as this one ensure that British and European space enterprises don’t just build remarkable technology, but can successfully establish world-leading, economically sustainable market positions.”

        Beth Greenaway, Head of Earth Observation Delivery at the UK Space Agency, said: “This call provides an exciting opportunity for UK organisations to turn innovative Earth observation ideas into commercially viable products and services. Through our continued partnership with ESA, InCubed helps businesses de-risk development, engage customers and accelerate their route to market, strengthening the UK’s position as a global leader in Earth observation.” 

        More information about this InCubed call can be found on the UK Government website.

        To know more: InCubed, UK Government website

        Photo courtesy of ESA

        ESA’s first quantum computer will shift computing frontiers in space

        ESA-ESRIN, the European Space Agency’s Centre for Earth Observation, has welcomed its first quantum computer. The computer’s quantum processing capabilities will be merged with classical high-performance computing capabilities to research the potential advantages of hybrid classical-quantum computing for complex Earth observation problems.

        The daily monitoring of our planet is generating an unprecedented amount of data, essential for climate modelling, weather forecasting and disaster response. This deluge of Earth observation data, now combined with the power of Machine Learning, requires a move beyond traditional computing paradigms, as classical computers are being pushed to their physical limits.

        For years, Earth observation researchers have been studying the potential advantages of quantum computing in terms of speed-up, utility and memory use for the processing of Earth observation data. Unlike classical computers, which process information using bits that can take the value 0 or 1, quantum computers use quantum bits, or qubits. Through unique quantum mechanical properties such as superposition and entanglement, quantum computers can explore many possible solutions simultaneously, potentially offering significant advantages for specific classes of problems.

        However, the demonstration of such advantages for Earth observation remains, in practice, a significant scientific challenge. Earth observation applications typically involve large, noisy and highly complex datasets, while the quantum computers available today remain relatively small and are still affected by hardware limitations, such as data coherence and error rates. As a result, researchers worldwide are actively investigating where quantum computing can deliver measurable benefits, and which Earth observation problems are best suited to hybrid classical-quantum approaches.

        A recent collaboration between the European Space Agency and Equal1 is taking an important step in that direction. Following the agreement signature in November 2025, Equal1’s Bell-1 quantum computer, a 6-qubit system, is under installation at ESA-ESRIN’s Data Centre in Frascati.  

        Bell-1 will provide ESA researchers with direct, on-premises access to quantum computing capabilities. Although limited in size, and in combination with ESA’s High-Performance Computing (HPC) infrastructure, this computer will serve as a testbed for the development and rapid prototyping of hybrid quantum-classical algorithms targeting Earth observation challenges. This unique environment will enable researchers to explore potential quantum advantages, benchmark emerging algorithms on real Earth observation datasets, and accelerate the transition from theoretical studies to practical demonstrations.

        The easy integration of Bell-1 into ESA-ESRIN’s Data Centre has been enabled by its specific technology: Bell-1 is based on silicon spin qubits that are built using a Complementary Metal-Oxide-Semiconductor (CMOS) manufacturing process similar to what is used to make the chips inside everyday smartphones and laptops. The Bell-1 chip operates around 0.3 kelvin, and while that is still incredibly cold, it is significantly warmer than what other popular quantum computers require.

        This temperature difference changes everything logistically. The entire computer is housed in a compact, rack-mounted chassis that can be directly integrated into more conventional data centre infrastructures. It relies entirely on an integrated, closed-cycle cooler, eliminating the need for external tanks of specialised coolers, and drawing only 1.6 kW of power – roughly the same as a single high-end enterprise server.

        Bridging the gap with Hybrid Quantum Computing

        After its installation in ESA-ESRIN’s Data Centre, Bell-1 will be available for internal research at ESA for one year after its commissioning. The goal is to operate Bell-1 under a hybrid quantum computing framework: ESA Φ-lab is planning a pilot demonstration of its use by the end of 2026, targeting several use cases investigated during the first phase of the project. Use cases include hybrid quantum neural networks for land use and land cover classification, satellite mission planning, and others to be later defined.

        If the pilot phase is successful, it will represent a shift in how the Earth observation field views and uses quantum computing. By the end of the pilot phase, the practical findings from these use cases will be shared with the broader scientific community in a joint workshop hosted by ESA Φ-lab and Equal1.

        Recognising the importance of using quantum computing for Earth observation data analysis, processing and modelling, Giuseppe Borghi, Head of the ESA Φ-lab Division, commented: “Earth observation will be entering an era where the scale and complexity of the data will challenge even our most advanced classical computing systems. Quantum computing promises a new way of approaching some of these problems, with the potential to complement the existing high-performance computing rather than replace it.”

        “By bringing quantum computing into our Earth observation research environment, we can rigorously test where it delivers real value, develop hybrid algorithms for practical applications, and build the scientific foundations for the next generation of climate, weather and, in general, Earth intelligence”, Giuseppe added.  

        The arrival of the Bell-1 quantum computer to ESA marks the beginning of a transformative chapter. Besides boosting ESA’s data processing power, it represents an important step towards a future where quantum and classical technologies work in unison to unlock solutions for our greatest global challenges.

        To know more: ESA Φ-lab, Equal1

        Photo courtesy of Unsplash/Laura Ockel

        ESA InCubed Ambassador Italy annual event in Rome highlights cross-sector collaboration for Earth observation innovation

        The European Space Agency (ESA), through its InCubed Ambassador for Italy powered by STAM, brought together key stakeholders in Rome for the annual event “Co-create. Connect. Innovate. – Shaping new project ideas between space & non-space industries”.

        The event aimed to foster collaboration between Earth observation (EO) actors and non-space sectors, supporting the development of commercially viable solutions by aligning technological capabilities with market needs and industrial challenges.

        InCubed stands for ‘Investing in Industrial Innovation’ and is an ESA co-funding programme, managed by Φ-lab, that supports the development of innovative and commercially viable EO products and services, with the objective of accelerating their path to market. In this context, the ESA InCubed Ambassador Italy initiative, powered by STAM, promotes access to the programme and supports companies, research centres and startups in identifying opportunities, shaping business cases, and preparing high-impact proposals.

        Read the full article on www.commercialisation.esa.int.

        Photo courtesy of STAM.

        SHRUG-FM wins ‘Best Paper Award’ at the EarthVision 2026 workshop

        Building on the success of the 2025 FDL Earth Systems Lab (ESL)’s research sprint, SHRUG-FM, a framework for reliability-aware prediction that enables geospatial foundation models to identify and abstain from likely failures, won ‘Best Paper Award’ at the EarthVision 2026 workshop that took place during the Computer Vision and Pattern Recognition (CVPR) 2026 Conference.

        Geospatial foundation models are very useful tools for Earth observation research since they allow a single, pre-trained AI model to be quickly adapted for multiple tasks, using drastically less labelled data than more conventional AI models. However, these models can be ‘confidently wrong’, failing to perform reliably in environments that were underrepresented during their training phase.

        This becomes particularly problematic when these geospatial foundation models are used in critical, time-sensitive situations such as disaster response after an extreme weather or climate event. But how is it possible to increase model transparency in cases where the model output has a high degree of uncertainty and requires human validation? A team mentored by Ruben Cartuyvels, Internal Research Fellow at ESA Φ-lab, and Patrick Ebel, former Internal Research Fellow at ESA Φ-lab and researcher at Google Research, decided to dive deep into new ways of making foundation models aware of their own uncertainty.

        During the 2025 FDL Earth Systems Lab (ESL)’s ‘Foundation Models for Extreme Environments’ sprint, the team developed SHRUG-FM (Systematic Handling of Real-world Uncertainty for Geospatial Foundation Models), a framework that helps foundation models flag when they may fail, addressing two main types of uncertainty: data-driven or model-driven.

        SHRUG-FM was developed as mechanism that can either provide a prediction, raise a warning or ‘shrug’ – indicating it doesn’t know the correct answer – when a foundation model is highly uncertain. This adaptable framework aims to make geospatial foundation models more transparent and reliable, showing a consistently reduced prediction risk for critical, climate-sensitive tasks like burn scar segmentation, flood mapping and landslide detection.

        On 4 June 2026, SHRUG-FM was presented during the EarthVision 2026 workshop. As part of the Computer Vision and Pattern Recognition (CVPR) 2026 Conference, this workshop aimed to strengthen collaboration between the Earth observation, computer vision and machine learning communities, fostering innovation in automated geospatial analysis. SHRUG-FM won ‘Best Paper Award’, and its paper and code are now available online.

        Vasileios Sitokonstantinou, member of the SHRUG-FM team, holds the ‘Best Paper Award’ certificate. From left to right: Ronny Hänsch, Vasileios Sitokonstantinou, Nathan Jacobs and Hannah Kerner.

        Claudio Iacopino, Head of the ESA Φ-lab Explore Office, commented on the importance of developing frameworks like SHRUG-FM: “Winning ‘Best Paper Award’ proves that the Earth observation research community is shifting its focus to a crucial point in the use of geospatial foundation models: trust. For a long time, the goal was to make these models smarter, and even though that is still an objective, it’s very important to have frameworks that give these models the ability to say ‘I might be wrong here’. Building this kind of reliability is exactly what both research and industries need to see before they fully adopt these models for widespread, everyday use.”

        A more detailed explanation about the development of SHRUG-FM is available here. 

        To know more: ESA Φ-lab, FDL Earth Systems Lab (ESL), EarthVision 2026, Computer Vision and Pattern Recognition (CVPR) 2026 Conference

        Photo courtesy of ESA, contains elements from FDL Earth Systems Lab.