Conference Agenda

Urban expansion changes local climate conditions through modifications in surface properties and energy balance. We developed a multitemporal Local Climate Zone mapping approach to systematically track land use transformations across European cities from 2000 to 2023. Local Climate Zones represent a standardized classification system that categorizes urban and natural landscapes based on their thermal, radiative, and morphological properties, providing a framework for understanding climate impacts of different urban forms.

Our methodology combines neural networks with change detection algorithms to monitor seventeen Local Climate Zone categories. Deep learning models incorporate batch normalization and dropout to classify urban morphology based on building structure, surface cover, and vegetation patterns. Feature importance analysis using separate models provides insights into the most discriminative characteristics for Local Climate Zone differentiation. We apply Multivariate Alteration Detection to identify significant land use transitions between time periods.

The analysis relies on Landsat imagery processed through cloud masking and temporal compositing using percentile reducers on the CODE-DE platform. Additional input features include elevation data from Copernicus Digital Elevation Model to capture topographic influences on urban climate zones. Training areas combine manually digitized zones with existing WUDAPT (World Urban Database and Access Portal Tools) datasets, validated using building height information from global settlement layers. This approach ensures representative coverage of Local Climate Zone types across different European urban contexts.

Several technical challenges required careful consideration. Harmonizing data across different Landsat sensors while maintaining temporal consistency proved critical for multidecadal analysis. We implemented spatial cross-validation to address autocorrelation effects in training data. Class-specific Gaussian filtering improved spatial coherence while preserving important urban boundaries. The processing workflow incorporates spectral indices and morphological features to capture the diverse characteristics of European urban environments.

Change detection employs iteratively reweighted Multivariate Alteration Detection to generate probability maps of land use transitions. This approach quantifies conversion from natural and agricultural areas to various built-up zone types, particularly around urban peripheries where development pressure is highest.

The resulting continental-scale dataset will enable detailed investigation of how building density and height variations within different Local Climate Zone types influence urban heat development. Urban planners will benefit from data-driven insights about optimal building arrangements for heat mitigation, while policy makers receive quantitative evidence for climate-sensitive development strategies. This comprehensive monitoring across all European cities supports comparative analysis of urban development patterns and their thermal impacts, contributing to more effective climate-informed planning throughout Europe.

Dynamics of urban growth have been extensively studied from the demographic and economic perspective. However, little is known about urban spatial expansion, despite its crucial environmental, social and economic implications. Recent modeling efforts, based on various approaches such as reaction-diffusion equations, cluster growth or percolation, lack robust empirical grounding due to i) improper definition of city delimitations, ii) difficulty of choosing an appropriate time parameter and iii) little data availability. Remote sensing techniques have made available new urban artifical land use datasets, at a yearly frequency and fine-grained spatial resolution, such as the World Settlement Footprint Evolution dataset. Combining this dataset and historical population records, we propose a surface growth framework for the analysis of urban sprawl. We focus on the largest connected component of urban fabric and study its radial growth. We focused on 19 major cities spanning 3 continents between 1985 and 2015. We analyze the growth speed of the urban area in function of population and find generally linear scaling, resonating with recent Lemoy and Caruso cross-sectional findings. Moreover, we quantify deviations to isotropic growth through scaling of directional growth. Through the measure of the largest aggregated cluster, we can measure whether the growth of urban fabric resembles diffusion or aggregation processes. We notice a striking relation between demographic pressure and large cluster coalescence. Finally, we apply techniques from surface growth physics on the urban fringe. Surface growth theory has been developed in the last decades and has been applied to a variety of physical processes, such as molecular-beam epitaxy or tumor growth. Types of interfaces growth are classified in universality classes, a finite set of exponents characterizing the interface, both statically and dynamically. In this work, using recent radial surface growth techniques, we were able to measure the local roughness exponent, the growth exponent and the correlation length exponent. We identified an unique value for the local roughness exponent while the other exponents vary across cities. This work is, up to the authors knowledge, the first longitudinal analysis of the dynamics of urban sprawl and provide new empirical findings, necessary for modeling and simulating urban sprawl.

Accurate demographic predictions at fine spatial resolution play a critical role in helping urban planners allocate resources efficiently, ensuring better housing, transportation, and services for diverse populations. Traditional data collection methods, such as censuses and surveys, suffer from limitations in frequency, spatial resolution, and cost, which make it difficult to capture dynamic demographic changes in urban areas. This research proposes a solution to address this gap by leveraging machine learning (ML) models to enrich building data with predicted demographic characteristics at the building level. The primary aim of this study is to evaluate the effectiveness and adaptability of machine learning models, specifically Random Forest (RF) and Extreme Gradient Boosting (XGBoost), for predicting demographic patterns, including population count and age group proportions, in the German context.

The study uses Stuttgart, known as an economic hub with a strong industrial presence and a mixture of industrial and modern urban developments, as the training area. Dresden, a historical city with a mix of heritage and modern developments, is used as the test area to examine the generalizability of the models across cities with contrasting urban forms. The dataset includes detailed two-dimensional building attributes derived from 3D CityGML data, 100-meter grid census data, OpenStreetMap points of interest (POI), block data, house coordinate data, and accessibility measures. Through extensive feature engineering, over 50 predictive indicators were generated. These indicators fall into two main categories: building-level attributes (e.g., area, height, volume, shape complexity measured by Equivalent Rectangular Index (ERI) and Proximity Index (PI)) and urban-level indicators (e.g., walking distance or time to the city center, built-up density, number of neighboring buildings, and accessibility to various POI types within a 15-minute walk).

A key challenge in this research was the disaggregation of demographic data from the grid level to individual buildings, which significantly influenced the model outcomes. Despite these challenges, the results demonstrate that population prediction achieved good accuracy using the RF model, showing that building-level data combined with ML can effectively estimate population distribution. However, the prediction of age-related demographics was less precise, primarily due to inaccuracies and limitations in the input datasets. The analysis also revealed that POI data had an impact on age-group predictions but played a less critical role in population estimation.

This research highlights the potential of integrating AI with urban data as an alternative or complement to traditional demographic data collection methods. It demonstrates the critical importance of high-quality, detailed building data and carefully designed feature engineering for achieving accurate demographic predictions. Future work should focus on refining model hyperparameters and incorporating more accurate building-level demographic data. Addressing these challenges is essential for enhancing the generalizability of the approach and ensuring its applicability across diverse urban environments.

Much research about local land governance is currently focused on understanding the drivers of local land policies, their effectiveness and priorities, chiefly in light of the housing, climate and energy crises. However, we know little about how local governments start building capacity as well as develop policy directions, setting future path dependencies in land use systems.

With a newly acquired land management and spatial planning autonomy as a result of the land and administrative reforms in 2019-2021, Ukrainian municipalities are now the key political bodies responsible for land use. Lacking previous experience, they are trying to find their own way of governing lands in the context of the war.

In this paper, we explore organisational learning in the use of public land (land portfolio management) and control over privately owned lands (land use management). By doing so, this paper addresses the question of how municipalities develop land management policies, lock in certain modalities and invent new ones. Building on interviews and questionnaires, we construct an understanding of the emerging local land governance regime with a view to enhancing and supporting sustainable local land policy and land management practice in Ukraine. Findings on institutional learning in municipalities would also inform future comparative analysis and support further research venues across Europe and beyond.

Biotope networks aim to connect fragmented habitats – such as forests, meadows, rivers and wetlands – by creating corridors or stepping stones that support biodiversity, ecological resilience and exchange. These networks play a crucial role in meeting legal obligations set out in instruments such as Article 10 of the EU Fauna-Flora-Habitat Directive and Article 4 of Nature Restoration Law for the connectivity of habitats. In Germany, the federal states are mandated to designate 10 percent of their land area for biotope networks. In response, the federal states have identified large-scale areas deemed ecologically suitable.

However, biotope networks can only fulfil their intended purpose if they are legally secured. As early as 2017, the German legislator emphasized that the biggest deficit in the implementation of the biotope network is its legal securement. In Germany, legal securement can take the form of nature conservation sites, land-use plans, contractual agreements, or land ownership (§ 20 sect. 2 Federal Law on Nature Conservation). Among these, land ownership by public authorities offers a higher effectiveness for nature restoration measures, whereas voluntary cooperation between landowners and the public sector increases acceptance of measures. Yet, the spatial distribution of biotope networks and their legal securement has so far not been assessed across Germany.

This research, funded by the German Federal Agency for Nature Conservation, assesses the extent of legally secured areas designated for biotope networks. We seek to understand the function of both regulatory and cooperative instruments in achieving robust land securement. In a first step, we developed a consistent framework to assess the legal status of areas within biotope networks and collected spatial data from state authorities and large-scale landholders engaged in conservation—such as land agencies, NGOs, and foundations. Second, we compare the legally secured areas to state-level concepts and policy objectives. The results will be discussed with public and civil society stakeholders to identify implementation deficits and develop policy recommendations.

Urbanization is associated with increased sealing, habitat fragmentation, rising temperatures, and anthropogenic noise in densely built-up areas. Since these impacts are linked to the deterioration of human health as well as biodiversity loss, new solutions need to be found to make cities more livable and resilient.

Various urban planning approaches already exist to account for and mitigate these impacts by transforming urban structures while recognizing the value of nature beyond its ornamental purpose. However, such approaches often focus on isolated aspects and prioritize selected disciplines over others. For example, Nature-based Solutions are not always considered in planning decisions, and the concept of ecosystem services itself is often applied in a narrow sense, e.g., when specific trees are selected for cooling effects while their broader ecological role, including benefits for wildlife, is neglected.

In this way, urban planning is not yet a multidisciplinary, integral process, but rather a mosaic of multiple different disciplines whose coordination is often only attempted afterward. This later harmonization often fails because individual disciplinary solutions are not compatible. In particular, human-centric approaches such as traffic planning often lead to compromises at the expense of nature available for humans and other organisms. In contrast, early data-driven integration of multiple disciplines can uncover synergies between the different disciplines. Meanwhile, multidisciplinary, collaborative, and participatory approaches have proven effective in delivering resilient solutions and have already been applied to the urban environment.

In this paper, we advocate for the early integration and equal importance attribution to multiple planning perspectives into a single, holistic approach. We point out the conceptual, but also technical challenges that have to be overcome for interdisciplinary integration, in particular with respect to the further development of disciplinary models. We argue that integration can best be achieved when disciplines come together by jointly developing urban planning scenarios. This will also facilitate the identification of interfaces between disciplines with respect to data formats and modelled variables. We also argue that such scenario development should include stakeholders in a co-creative approach to make sure that scenarios will address real-world planning questions and challenges.

We illustrate how integration could proceed by considering the integration of a traffic planning and an ecological connectivity model based on multiple scenarios, to highlight the interactions between both disciplines and the benefits that accrue from early integration in the planning process. To illustrate how the co-creative process can be conducted, we report on early steps of the CoCoNet-project (Co-creative Cohabitation Network), which aims to develop an integrated, multidisciplinary, and participatory modeling-based planning approach further integrating microclimate and noise models. By pointing out both the advantages of multidisciplinary integrations as well as suggesting a pathway on how this can be achieved, this paper aims to encourage planners from different disciplines to collaborate in the development of desirable urban planning pathways while considering the needs of humans and urban wildlife.

Urban transport nodes such as rail stations are vital enablers of mobility and accessibility in metropolitan regions. Understanding the dynamic spatial patterns of activity around these hubs is increasingly crucial for guiding equitable and sustainable urban development. This study explores the urban spatial dynamics surrounding the Marmaray mass transit stations on the Anatolian side of Istanbul, which employs venue-based data from the location-based social network, Foursquare. By analyzing venue distributions within a 500-meter radius of 29 stations using Kernel Density Estimation and K-means clustering, four distinct clusters of stations were identified, each characterized by different configurations of retail, dining, recreation, and service-oriented functions. The findings highlight significant variation in activity density and functional mix across the corridor, which reveals spatial inequalities and differentiated urban roles of transit-adjacent areas. By integrating geo-tagged data, this research contributes to the growing body of work on data-informed urban analysis and offers actionable insights for transit-oriented development and urban policy.

The concept of the 15-minute city has gained considerable attention in recent years as a promising solution for promoting sustainable mobility and improving residents' quality of life. The idea is to provide residents with easy access to important services and facilities within 15 minutes on foot or by bike. Different types of urban neighbourhoods, in terms of spatial density, infrastructure provision or requirements and opportunities of local people, might have different pre-conditions and needs to apply the concept. Large housing estates (LHEs) on the one hand are often planned as compact, people centred neighbourhoods, providing all facilities for daily needs. On the other hand, in recent years they followed different trajectories, causing several challenges, as limited and ageing infrastructure, changing socio-demographic situation and limited public awareness.

This study investigates the suitability of existing methods for assessing 15-minute city maturity for application to LHEs. The gravity-based approach to assessing accessibility to basic essentials (in the areas of health, education, recreation, commerce etc.) is applied to European LHE case studies, using a combination of official municipal data, OpenStreetMap data, and data from participatory, co-creative approaches. The results show that the complex challenges and limitations of implementing the 15-minute city concept in LHEs must be carefully considered.

One of the main limitations of the study is the use of gravity-based indicators, which are difficult to interpret and may not accurately reflect user needs and preferences. Additionally, the weighting of indicators in the calculation of the 15-minute score may need to be adjusted to take into account regional and user-specific preferences. For example, the willingness to walk or cycle can vary significantly between different cities and regions. The study also highlights the importance of multimodal considerations, including public transport, walking, and cycling, for convenient and efficient access to essential services and facilities. Furthermore, the concept of walkability and cycleability is crucial for assessing urban environments, as it takes into account the safety, comfort, and attractiveness of walking and cycling routes. Moreover, the study raises questions about how the issue of work and employment can be addressed in the context of the 15-minute city concept. While the concept focuses on the accessibility of essential services and facilities, it is unclear how the needs of commuters and workers who may have to travel longer distances to their workplaces can be taken into account. This highlights the need for a more nuanced understanding of the complex links between transport, land use, and employment, as well as the development of more effective strategies to promote sustainable mobility and reduce the need for long commutes.

The study is part of the 15minESTATES research project, which aims to co-create spatial strategies for just and sustainable mobility in large-scale housing estates by investigating the spatial conditions and mobility needs of different social groups in these areas.

In summary, this study contributes to the growing body of research on the 15-minute city concept and highlights the need for differentiated and context-specific approaches to assessing and implementing sustainable mobility solutions in LHEs.

The building age is an important indicator for sustainability planning, for example when it comes to retrofitting potential or risk assessment. The EUBUCCO project compiled all available building stock data with building ages, building types (residential or non-residential), building heights and footprints across Europe in 2021. The project revealed that data availability and definitions differ greatly between states and regions.
In this study, we evaluate the required data quality to predict building ages based on morphological features using Switzerland and Germany as case studies. According to ISO 19157:2013, data quality comprises four elements: completeness, thematic accuracy, logical consistency, temporal quality. We investigate the effects of data completeness and thematic accuracy (referring mainly to the building age). Switzerland provides a dataset with nationwide coverage and detailed building age entries (up to the month). Germany does not provide a
joint dataset. Only few federal states provide an open dataset with the building age.
After preprocessing the data and deriving the relevant features such as proximity to the next building or building height, this work applies the same machine learning approaches to predict the building age. The algorithms used are gradient boosting and graph neural networks. Both algorithms are known for their ability to handle spatial data. A separate machine learning model is fitted to the available data for each of the countries. To ensure comparability, the same amount of hyperparameter optimization is performed. Comparing the two set-
ups provides insight into how quality affects the prediction. The results can inform further model development and potentially influence policies for effectively sharing building age data.

Many Earth science disciplines rely on long-term time series data. Historic land use is a valuable information source and proxy for various applications in biodiversity science, climate protection, urban analytics and many more. However, the acquisition of this data is not trivial and on a national scale extremely labour-intensive. In this study, we investigate the feasibility and possibilities of extracting this information automatically from its historical sources. We present results in terms of both the temporal data source availability and methodical approaches for a national application.

The IOER RDC (IÖR-FDZ) offers cross-scale and cross-sectional, high-resolution object and spatial data sets, indicators, models, simulations and digital tools on the topics of land use, landscape quality, settlements, buildings, ecosystem services, urban greenery and environmental risks.

Our services address stakeholders from research, politics and administration, the media, business and civil society. They are further developed and made available in an application- and user-orientated manner in order to address the various needs of the user groups and increase the usability of the data.

In order to achieve this effectively, the IOER RDC (IÖR-FDZ) pursues a modular approach: data and technical functionalities (so-called components) are developed in such a way that they can be combined and used flexibly in different contexts, both internally and externally.

The poster or short presentation will illustrate this modular approach and present new data offerings. In addition, the poster/ presentation provides an overview of newly developed options for accessing, linking and (re)using IOER data.

The Colouring Cities Research Programme (CCRP) is a global initiative designed to advance the collaborative mapping of national building stocks through an extensive network of interoperable platforms. Focused on facilitating progress toward the United Nations Sustainable Development Goals (SDGs), CCRP leverages open data methodologies, engaging a broad consortium of over 100 researchers from more than 30 nations. Experts from diverse fields—including computer science, data science, urban science, and environmental science—convene regional hubs across North America, Europe, the Middle East, Latin America, Africa, and the Asia Pacific. This multidisciplinary approach enhances cross-sector knowledge sharing, fostering a rich collaborative environment aimed at sustainable urban development.

CCRP utilizes open-source platforms that standardize data interfaces, allowing for the effective pooling and sharing of micro-spatial data. Through an innovative integration of crowdsourced, bulk, and streamed data, the program employs artificial intelligence (AI) and machine learning (ML) techniques to develop large-scale datasets, which are carefully moderated by academic experts to ensure accuracy. Subsequent crowd-verification processes further enrich these datasets, promoting the principles of openness, transparency, and democratic decision-making within the program. By emphasizing societal engagement, CCRP seeks to involve diverse stakeholders in the data collection and visualization processes, aligning academic research with community needs and sustainability objectives.

The platforms developed under CCRP provide an interactive map interface enabling collaborative data collection and visualization. With over 150 standardized spatial data classes, the platform categorizes data into 12 main areas, alongside bespoke data tailored to specific community needs. Ethical oversight remains a paramount concern within CCRP, as academic supervision governs the ethical collection of data, ensuring user privacy and preventing the publication of personal data. Multiple data capture methods enhance accessibility, allowing citizens to participate as "scientists" in the mapping initiative. Bulk uploads, API streaming, and computationally inferred data contribute to the robustness of the datasets gathered.

Moreover, all CCRP resources—including code, datasets, and methodologies—are made openly accessible under liberal licenses, reinforcing the ethos of transparency and facilitating widespread collaboration across diverse contexts. This commitment to open access ensures that community members, researchers, and policymakers can engage with and leverage the data effectively for local and global development goals.

Participation in the CCRP is possible at various levels and across different areas of expertise. Interested individuals and organizations can explore the available platforms online. At the local level, individuals are encouraged to support Colouring Cities initiatives by participating in various citizen science formats, thereby involving the community in mapping and sustainable development.

The CCRP offers a transformative framework for the collaborative mapping of urban environments, aspiring to catalyze significant advancements in sustainable development through integrative, ethical, and inclusive practices. By leveraging the strengths of a diverse global network, CCRP seeks to create invaluable resources for cities worldwide, thereby contributing to a more sustainable future for urban living.

Urban sprawl, landscape fragmentation, and habitat network connectivity have long been recognized as interrelated processes that pose significant challenges to sustainable land-use management, particularly in rapidly urbanizing regions. Globally, accelerated urban expansion has transformed natural landscapes, leading to habitat loss, landscape fragmentation, and increased competition for environmental resources. In West Africa, unregulated urban growth, deforestation, and infrastructure development have further amplified these transformations. However, previous studies have often examined these processes in isolation, overlooking their interdependencies and feedback mechanisms, thereby constraining coherent, cross-sectoral planning and sustainable resource management.
In response, the present research applies a Resource Nexus perspective to investigate the interlinkages among urban sprawl, landscape fragmentation, and habitat network connectivity. This approach emphasizes a system-based view for the integrated management of environmental resources and the interconnections to promote sustainable development.
Through a systematic review of geospatial modeling approaches, the study identifies and evaluates suitable analytical frameworks for assessing the spatial and temporal dynamics of these processes. The selected approach will then be applied to assess urban sprawl, landscape fragmentation, and habitat network connectivity in Sierra Leone. A model will be established for integrated assessment through a Resource Nexus perspective, enabling a comprehensive understanding of their interconnected effects.
The study will develop a comprehensive modeling framework that links spatial patterns to resource-use outcomes, providing a scientific basis for sustainable land-use and environmental policy. The findings will provide an integrated scientific basis for policy and planning interventions that balance urban growth with environmental resource sustainability, offering critical insights for applying the Resource Nexus approach in rapidly urbanizing regions.