Conference Agenda

Detailed knowledge of the age of the building stock is crucial for a wide range of pressing issues related to energy efficiency, sustainable development, disaster risk, and vulnerability of buildings and their inhabitants. Recent advances in Earth observation data collection and processing have boosted the emergence of (quasi-)global building footprint datasets (e.g., Microsoft GlobalMLBuildingFootprints, Google Open Buildings, Overture Maps buildings, GlobalBuildingMap), complementing volunteered geographic information such as OpenStreetMap (OSM) or the OSM-derived OpenBuildingMap. However, while geometric information on buildings is abundant, thematic microdata on building age or function are often limited or entirely unavailable.

Global or continent-level data harmonization and curation efforts such as OpenBuildingMap or EUBUCCO partially provide building age estimates using authoritative data (e.g., cadastral data sources). Model-based inferential estimates, and country- or city-level crowdsourced and community-based efforts (e.g., the “Colouring Britain” project) effectively fill data gaps at country- or local scales. However, the potential of historical Earth observation data (e.g. from the Landsat archive) for explicit modelling of the age of building stocks and for enriching building footprint data has largely been unexplored.

Hence, we present and evaluate GHS-AGE, a global, free and open gridded dataset modelling the age of the built stock. GHS-AGE has been derived from global, multitemporal gridded estimates of built-up surface from the Global Human Settlement Layer (GHSL), measuring subpixel built-up surface area on a continuous scale at 100m spatial resolution, in 5-year intervals from 1975 to 2020. From this multi-temporal global gridded data stack, we identify the year in which 50% of the contemporary built-up surface area was exceeded for the first time. This year represents, in the absence of further information, our best guess of the median building construction year per 100m grid cell. Mapping this year (or epoch) to each grid cell reveals not only striking pictures of urbanization and settlement expansion, but represents the first, global, statistically formalized estimate of the construction epoch of the built stock.

We compare these gridded estimates to authoritative (i.e., census, cadastral, tax assessment) data, as well as real-estate related data sources reporting on building construction years and find encouraging levels of agreement across different countries and continents. GHS-AGE complements existing, global gridded datasets mapping the evolution of built-up surface area (e.g., GHSL), settlement footprints (e.g., World Settlement Footprint Evolution), or impervious surfaces (e.g., GISA) by providing a well-defined, statistically founded semantic, estimating the median construction year of the built stock, consistently enumerated across the globe. GHS-AGE is provided as a global, gridded dataset at 100m and 1km resolution and enables large-scale raster-based analyses of the built stock, facilitating the integration with other thematic variables from the GHSL such as population counts, building height, functional (i.e., residential / non-residential) or settlement classifications, enumerated in the same grid.

While being of gridded nature, GHS-AGE constitutes also an important input to the Global Human Settlement Open Building Attribute Table (GHS-OBAT), a recent data integration effort to enrich global building footprint datasets like the Overture Maps buildings with remote-sensing derived attributes in a flexible and extendable framework.

Knowledge about the functional use and exact location of buildings is essential for a wide range of applications, including urban planning, natural hazard risk assessment, and mobility analysis. However, comprehensive, up-to-date, and large-scale building datasets that contain both precise geometry and information on building use remain scarce, even in data-rich countries like Germany. Existing data sources in Germany, such as the official cadastral data (LoD1 and LoD2), provide semantic information on building types, but they are often incomplete due to data gaps, inconsistencies across federal states, and outdated information. Similar problems exist in building data from OpenStreetMap (OSM): while offering crowd-sourced building types, both the data quality and completeness vary greatly by region. Consequently, parts of the built environment remain unmapped and lack functional labels, impeding data-driven urban analysis.

To address these limitations, this study presents a scalable workflow that combines AI-based remote sensing techniques with multi-source data fusion to infer building use types across an entire metropolitan region. High-resolution aerial imagery and digital surface models are used to detect building footprints using a deep learning model based on a fully convolutional encoder-decoder architecture. Aerial imagery is used for the building detection because of its large-scale and up to date coverage, and its high spatial resolution. The model is trained and evaluated on annotated data from North Rhine-Westphalia and Berlin and achieves a mean intersection-over-union of over 92% on validation data. By experimenting with different training data sampling strategies, the model generalization is improved, and the best-performing configuration is applied to the full area of Berlin to extract accurate and up-to-date building geometries. This approach enables a systematic quantification of missing buildings in both cadastral and OSM building datasets.

Since most building usage types are not discernable only based on aerial images, additional data are necessary to assign functional attributes to the detected buildings. These include both previously mentioned official cadastral data and multiple OpenStreetMap layers, such as buildings and land use polygons. The heterogeneous usage categories from both sources are harmonized and mapped onto four generalized functional classes: ‘residential’, ‘commercial’, ‘work-related’, and ‘other’. These categories reflect the primary daily travel purposes observed in urban areas, making them well-suited for urban mobility modeling. Adjacent buildings are spatially aggregated into building complexes and proportional type shares are assigned based on areal relationships from the external building datasets. For building complexes with no matching semantic data, functional approximations are derived from land use information.

The resulting dataset provides proportional use shares across the four target categories for all buildings in the federal state of Berlin. This enables a nuanced representation of mixed-use as well as single-use structures. Beyond quantifying data gaps in existing building datasets, the study demonstrates how combining Earth observation data with volunteered and official sources can fill critical missing information on building location and use type. This approach offers a scalable solution for generating semantically enriched complete building inventories that reflect current urban conditions by bridging the gap between building geometry and semantic usage information.

To address complex challenges in urban sustainability, Austrian cities are increasingly relying on powerful digital solutions. However, the efficiency of these solutions is limited by the quality of the underlying urban infrastructure data, which is often fragmented, inaccessible, inaccurate, or outdated. This paper describes the implementation of the Colouring Cities Framework (CCF) in an Austrian context, laying the foundation for Colouring Austria, a national-scale initiative for building a standardized, extensible platform for building-level urban data. Originating from the Colouring Cities Research Program (CCRP) at The Alan Turing Institute, the CCF offers a modular framework for collecting, managing, and visualizing urban data using open-source technologies.

Using the city of Graz as a case study, we benchmarked the platform’s deployment, data integration workflows, and thematic enrichment. The broader research scope investigates how Austria’s urban data landscape can be integrated into a modern, flexible, and scalable digital solution, transferable to other local contexts such as the city of Linz and Salzburg, establishing a cohesive national approach to urban data infrastructure.

Following the CCRP documentation and standards, the platform was first deployed using a PostgreSQL database configured to store detailed building information. The database was then populated with national open geospatial and semantic data, including building geometry, height, use type, and other attributes. Additionally, to demonstrate the platform’s extensibility, we developed a test case focused on classifying roof characteristics and assessing the presence and potential of rooftop photovoltaic (PV) systems. The analysis aimed to identify roof geometry, materials, and solar exposure using publicly available data retrieved from federal GIS data portals, including airborne laser scanning and high-resolution aerial imagery. Based on these inputs, geospatial processing and classification techniques were applied to assign attributes such as roof shape (e.g., flat, gabled, hipped), material (e.g., tiles, metal, green roofs), and PV suitability to individual buildings. This detailed information was integrated into the database and CCPR web interface.

This test case highlights the framework’s capacity to support advanced spatial analysis workflows and integrate derived data within a centralized, queryable platform. All components of the deployment, including system configuration, data pipelines, and classification outputs, follow open standards, supporting reproducibility, interoperability, and transparency. The platform enables both experts and non-experts to explore urban data through an interactive, map-based interface.

By integrating high-resolution geospatial data with a flexible, open-source platform architecture, the project supports a more granular understanding of urban structures and spatial dynamics at the building level. This work contributes to the creation of a single source of truth for urban data, aligning with broader goals in sustainable urban development, data-driven decision-making, and digital twin applications. Importantly, the experience gained through this implementation offers valuable insights for enhancing deployment workflows, platform scalability, and data integration practices. The lessons learned from this study are actively shared with network partners to inform future iterations of the CCF and contribute to the continuous development of the CCRP.

Urban landscapes undergo continuous transformation, yet traditional monitoring systems, e.g. based on topographic maps produced by public administration, often lack the temporal granularity and semantic richness needed for comprehensive change detection. This study explores a novel approach that integrates crowdsourced OpenStreetMap (OSM) data with satellite imagery to enhance automated landscape monitoring capabilities. Building for the LaVerDi (Landschaftveränderungsdienst) operational service framework provided by Bundesamt für Kartographie und Geodäsie (BKG), the aim is to develop a machine learning pipeline that leverages the complementary strengths of both data sources for improved land use/land cover (LULC) change detection.

Our methodology begins with a quality assessment of OSM land use data in the study areas using the ohsome-quality-api, ensuring a reliable foundation for subsequent analysis. For efficient and scalable temporal analysis, we utilize ohsome-planet to access OSM data in Parquet format, significantly accelerating the processing of large-scale, time-series OSM datasets. We address the temporal offset between real-world changes and their representation in OSM by implementing a one year lag analysis between satellite observations and OSM updates. Using heterogeneous test sites in Germany, we generate OSM-derived land use rasters mapped to standardized classification schemes (Dynamic World) and compare them with multi-temporal satellite data through confusion matrix analysis. The approach uses high-resolution Planet RGB data to validate detected changes.

Looking ahead, we plan to develop machine learning models that can recognize patterns of valid change in OSM data by learning from historical satellite-OSM alignment patterns. The envisioned system would help distinguish between meaningful landscape transformations and mapping artifacts, and could automatically trigger validation workflows when significant changes are detected. Such a model would aim to validate OSM-detected changes against satellite evidence and determine when new LULC maps should be published, creating a more responsive monitoring system that combines the semantic detail of crowdsourced data with the objectivity of satellite observations.

Preliminary results demonstrate that OSM data can effectively capture land use dynamics, particularly for anthropogenic changes, while satellite data provides crucial validation for change authenticity. This integrated approach holds great promise for supporting sustainable urban development by providing more timely and precise landscape insights. This research contributes to advancing AI-based spatial data analysis, offering a scalable way to monitor landscapes by combining detailed crowdsourced information with objective satellite observations.