Paolo Mazzetti

Geospatial foundation models are a new frontier in artificial intelligence, designed to understand and analyze spatial data at scale. Trained on huge sets of EO data, these models can support a wide range of applications—from monitoring natural disasters to guiding urban development and tracking climate change. To this aim, researchers and practitioners need to fine-tune the foundation models for specific tasks, utilizing a relatively small amount of additional data. As a result, geospatial foundation models are reshaping how we observe, manage, and protect our planet. Fine-tuning a geospatial foundation model requires carefully curated training datasets that reflect specific regions, time periods, or tasks—such as detecting deforestation or mapping urban growth. Yet preparing these datasets is often labor-intensive, involving steps like selecting relevant imagery, aligning spatial formats, and generating accurate labels. In practice, this means that the effectiveness of GFMs hinges on the availability of AI-ready data. This bottleneck limits the accessibility and scalability of GFMs for scientific and operational applications. In this work, we introduce a software library designed to automate these preparatory steps, streamlining the transformation of geospatial datasets into consistent, high-quality inputs for GFM fine-tuning. By reducing technical overhead and ensuring data readiness, the library enables faster, more reliable, and more inclusive adaptation of foundation models to local environmental challenges and specialized domain needs.

1. Public space has an essential role to play in making cities livable and is interlinked with various other development issues such as environment and climate change, economic development, urban p...

The I-CHANGE project addresses the critical challenges posed by climate change, focusing on active citizen participation and the enhancement of public awareness through evidence-based methodologies. This paper presents the key achievements of the project, which include the organization of diverse community-oriented initiatives aimed at fostering environmental awareness, the deployment and testing of advanced environmental monitoring sensors, and the development of cutting-edge digital tools, such as an interactive dashboard and the ChallengeYeti mobile application. Additionally, the project analyzed extensive data collected through awareness raising campaigns with surveys on individual environmental behaviour, offering valuable insights into the drivers of environmental consciousness. The results underline a significant increase in awareness levels among participants, the effectiveness of technological solutions in promoting engagement, and the relevance of comprehensive data analysis in understanding and addressing climate-related challenges. I-CHANGE proposes a scalable and replicable model that combines technological innovation and inclusive citizen engagement to support climate adaptation and mitigation efforts.

The I-CHANGE (Individual Change of HAbits Needed for Green European transition) project is a 3.5-year Innovation Action initiative, funded under the European Union's Horizon 2020 programme, concluding in April 2025. It aims to engage citizens actively in environmental monitoring and climate action, demonstrating that individual behavioral changes, facilitated through citizen science initiatives utilizing sensors and monitoring devices, can significantly reduce environmental footprints. The project establishes Living Labs (LLs) as collaborative spaces where researchers and communities engage in scientific discourse, sharing insights and outcomes.Central to I-CHANGE is the Environmental Impact Hub (EIH), a comprehensive data infrastructure designed to collect data, provide tools, and support initiatives facilitating citizens to participate in environmental monitoring and action. The EIH is a data hub that effectively shares, manages, and processes diverse datasets, ensuring interoperability and usability of heterogeneous data. It supports machine-to-machine interactions to facilitate the development of desktop and mobile applications through dedicated service interfaces and APIs. The EIH is built of discrete components, including the Data and Information Broker - based on the pre-existing Discovery and Access Broker (DAB) - which facilitates seamless discovery and access to diverse data sources, including in-situ measurements, citizen-provided data, and European infrastructures; the Citizen Observatory Archive which ingests, quality-checks, and stores observations generated by the LLs; the Dashboard which supports visualization of environmental data and empowers citizens to monitor and understand their environmental impact with intuitive and user-friendly interfaces, near real-time analytics, and actionable insights, enabling users to explore the environmental consequences of their actions and track improvements over time. The EIH offers multiple interfaces to support a wide range of use cases and interaction types. These include geospatial interfaces adhering to standards (such as those from OGC and ISO), a Web API for easy web development integration, and a RESTful API for exchanging JSON data across various platforms.I-CHANGE is predicated on the belief that citizens and civil society play a central role in environmental protection and climate action. As direct involvement of private citizens is considered essential to drive meaningful shifts towards more sustainable behavior, I-CHANGE presents a comprehensive effort to engage citizens in environmental monitoring and action, by providing both advanced technological platforms and participatory Living Labs. The EIH supports this vision by providing the underpinning technical infrastructure. By facilitating access to diverse data and tools, and promoting citizen involvement, I-CHANGE aims to empower individuals to make informed decisions that contribute to environmental sustainability, mitigating environmental challenges, and advancing climate action.

As urban areas face increasing threats from climate change, citizen science has emerged as an important tool to engage communities in monitoring and responding to environmental challenges, thus filling in the gap which existing tools are not addressing appropriately. Citizen science initiatives are essential for engaging citizens in climate action, involving them in environmental observations and monitoring human impacts. These participatory initiatives between science and society have gained popularity across various fields, including sociology, astronomy, and environmental protection. By involving students, citizens, and stakeholders, these initiatives foster a sense of ownership and empowerment, encouraging continued engagement and collaboration.This paper introduces a dashboard developed within the Horizon 2020 project I-CHANGE, designed to involve citizens in the collection and analysis of environmental data. The dashboard empowers urban residents to use low-cost sensors and crowdsourced observations to gather vital information on air quality and climate variables. Co-designed with scientists and stakeholders, the dashboard provides an intuitive platform for citizens to view, understand, and interpret complex collected data. By presenting crowdsourced data in a meaningful manner, the dashboard bridges the knowledge-action gap, fostering greater public awareness and environmental consciousness. Such participatory approach increases the level of understanding of urban climate risks and strengthens adaptation strategies by integrating local insights and vulnerabilities.Through the active involvement of citizens in data collection, the dashboard promotes hands-on experience with the real effects of climate change, leading to increased awareness and climate-friendly behaviours. This engagement is essential for achieving climate mitigation goals and advancing Europe's climate adaptation strategies. The paper discusses the design, implementation, and data integration of the dashboard, highlighting its role in combating misinformation and supporting community-driven climate action.

The Discovery and Access Broker (DAB) technology is implemented and deployed in the context of several European Union research projects and international initiatives, such as the Global Earth Observation System of Systems (GEOSS) and WMO Hydrological Observing System (WHOS), to enable discovery and access amongst distributed geospatial data provider services, along the line of fair and open approaches. Lately the DAB has been employed in the context of two European Open Science Cloud (EOSC) EU-funded projects FAIR-EASE, and Blue-Cloud 2026, characterized by a strong synergy between them. In FAIR-EASE, a customized DAB instance enables harmonized semantic searches amongst 15 data sources: EuroArgo, the Copernicus Marine Environment Monitoring Service (CMEMS), EasyData, ELIXIR-ENA, EurOBIS, European Environment Agency SDI Catalog, EMODnet, ICOS (Data Portal and SOCAT), Joint Research Centre Data Catalog, SeaDataNet (open datasets and products), US NODC collections, VITO/Copernicus Global Land Services, and WEkEO, totaling 156,000 datasets available for search. These services publish data using different service interfaces and data models. The DAB can seamlessly connect to each of them and map results to the interface and data model required by the users, leveraging a flexible and extensible harmonized internal data model based on ISO 19115. The broker discovery services are accessible through multiple interfaces, such as OGC CSW, OpenSearch API, and SPARQL endpoint. In particular, the SPARQL endpoint interface follows a linked data approach that is further leveraged by the FAIR-EASE architecture. To realize the SPARQL endpoint, a mapping from ISO 19115 to the FAIR-EASE DCAT profile was implemented. Concept URIs found in the original metadata (e.g., linking parameters, stations, instruments, keywords from online vocabularies) can be easily represented as properties and relations of the mapped FAIR-EASE dataset, supporting further reasoning and harmonized semantic query. The connection with the Semantic Analyzer component provided by BODC has been explored and will be the subject of further research. This tool can analyze FAIR-EASE datasets by connecting to the DAB CSW interface, extracting a list of terms and scanning known online vocabularies and ontologies to identify possible concept URIs matches. These URIs can finally replace natural text occurring in the original metadata, enhancing overall semantics and quality.

Land degradation processes have experienced a significant increase in recent decades, a trend that is projected to escalate further in the absence of any intervention. The need of adopting practices to contain, mitigate and restore degraded land have been stressed also by the new European Mission 'A Soil Deal for Europe'. To guide and support restoration actions, through a common and effective framework, an efficient monitoring approach and an adaptive ecological restoration process is needed. The NewLife4Drylands LIFE project provides a Protocol for design, implementation, and maintenance of restoration activities based on Nature-Based Solutions (NBS) within drylands. The Protocol, developed following the principles and inputs of some international restoration standards (SER, IUCN), is based on the identification and monitoring of degradation processes exploiting remote sensing capabilities, with the aim to integrate data derivation procedures into ecological restoration and maintenance activities. The Protocol is supported by a Decision-Making web-tool guiding trough the degradation processes, NBS along with indices/indicators with the aim to reduce the knowledge effort and helps in prioritizing options. The Newlife4drylands experience highlighted the heterogeneity and complexity of degradation processes, as resulted from a selected set of degraded pilot sites within Mediterranean Protected Areas, together with the issue for harmonization and standardization of ecological/physical indicators, especially those derived from satellite observations, when used as proxies of land degradation. The integrated use of both available field data (for short-term monitoring) and satellite data (for medium and long-term monitoring) have been explored to identify indicators for evaluating the effectiveness of planned restoration actions. This approach is geared, towards fostering adaptive and collaborative management of the ecological restoration process. Therefore, the Protocol acts as support tool for decision-makers, including public administration of Protected Areas, as well as technicians and planners. The proposed approach aims to raise awareness about the needs of drylands and opportunities provided by NBS. It serves as a guide for the identification of specific/local NBS for the restoration of drylands, beginning with the identification of degradation processes.

In the framework of the NewLife4Drylands LIFE Preparatory project (LIFE20 PRE/IT/000007, 2021-2024) the estimation of SDG 15.3.1 indicator [1], adopted in the UNCCD’s Good Practice Guidance [2], was applied for evaluating Land Degradation (LD) in different Mediterranean Protected Areas (PA). To effectively support PAs managers, joint effort was made in the evaluation of SDG 15.3.1 indicator at the local scale by using satellite Remote Sensing data in the computation of the three main sub-indicators as trend in Land Cover (LC), Primary Production (PP) and Soil Organic Carbon (SOC) stock. Where feasible, local scale sub-indicators were not sourced from open-access global/European databases due to their lack of accuracy at the site scale [3]. LD was estimated not only for the whole PA but also for specific LC classes of interest, considering additional sub-indicators related to pressures and threats affecting the class. This study focuses on the dryland Alta Murgia (IT9120007) PA, in southern Italy, and the wetland Nestos River Delta (GR1150001) PA, in Greece. For Alta Murgia site, featuring semi-natural dry grassland habitats of community interest that are frequently subjected to fire events during the summer season, the Burn Severity (BS) index was included. BS trends were measured by assessing the difference in pre/post–fire Normalized Burn Ratio (NBR) index from Landsat data during summer. Baseline data from 2004, coinciding with the establishment of a National Park within PA, was compared with 2018 for validating field data availability. Nestos River Delta hosts the largest natural riparian forest in Greece and is frequently subjected to hydrological cycle modifications, involving water scarcity due to both inappropriate river management and climate change, in turn hampering the transport of nutrient-rich sediments and the enrichment of soils being at risk of aridification. Within this framework, Hydroperiod and Soil Salinity indices were considered for LD and specific impacts on aquatic vegetation LC. Baseline data from 2017, after the dry climate conditions of 2016-2017, was compared with 2021 for validating field data availability. Both in Alta Murgia and Nestos, LC mappings were obtained by a data-driven pixel-based approach considering Landsat/Sentinel-2, respectively, multi-seasonal imagery and a multi-class Support Vector Machine (SVM) classifier trained with data from in-field campaigns and historical orthophotos interpretation. Time series of MSAVI from Landsat (which replaced standard NDVI for its soil correction benefits [4]) and PPI from Sentinel-2 by Copernicus services, respectively, were used to track grassland PP trends. Lastly, for SOC stock trends, the open-source Trends.Earth QGIS plugin [5], incorporating customized LC data and global SoilGrids product, was adopted to supplement local data limitations. According to its specification, the SDG 15.3.1 indicator was computed by integrating all the sub-indicators according to the principle “one out, all out” obtaining the 3-classes output mapping (Degradation, Improvement, Stable). The findings can support the monitoring and evaluation of LD, guiding protective measures aligned with the Agenda 2030 for Sustainable Development. They, also, highlight the importance of the integration of local scale data and sub-indicators within the UNCCD methodology.References[1] https://unstats.un.org/sdgs/metadata/files/Metadata-15-03-01.pdf[2]https://www.unccd.int/publications/good-practice-guidance-sdg-indicator-1531-proportion-land-degraded-over-total-land[3] https://doi.org/10.3390/rs13020277[4] https://doi.org/10.3390/rs12010083[5] http://trends.earth/docs/en