Conference Agenda

The anthropogenic footprint on Earth, including transportation networks and settlements such as cities, towns, and villages, has been shaped over the course of centuries and more. However, our quantitative knowledge of these long-term landscape changes is scarce, due to a lack of digital geospatial data reflecting land use and land cover (LULC) prior to the era of operational remote sensing and digital cartography.

We argue that for an enhanced understanding of longer-term landscape transformation processes such as urban growth, conurbation and (sub-)urbanization, as well as urban and rural land use dynamics, commonly used remote-sensing based LULC data from recent decades only provide a limited picture. As a consequence, derived trends may disregard important long-term context, and extrapolated future development pathways may lack statistical robustness due to limited historical data available for modelling. To address this shortcoming, researchers have started to use geohistorical data sources such as scanned and georeferenced historical maps, historical overhead imagery, authoritative building and road register data providing information on construction epochs, historical gazetteers and other structured and unstructured data sources. Such data are increasingly available in openly accessible, digital data formats, and recent advances in Artificial Intelligence and data processing and integration capabilities have further catalyzed such efforts.

Herein, we summarize our recent advances in curating, integrating, and providing analysis-ready geospatial data that enables researchers, planners, and practitioners to gain quantitative knowledge on long-term landscape changes during the 20^th century, ranging from national-level to global geographic scope, encompassing long-term changes in built-up areas, road networks, land cover, and populated areas.

We discuss novel data sources such as HISDAC (Historical Settlement Data Compilation) that leverages building and property-level microdata including construction year information to model historical built-up areas, building densities and settlement extents using spatial aggregation techniques. HISDAC has been produced for the conterminous US (HISDAC-US) and Spain (HISDAC-ES), and has recently been extended to the Netherlands (HISDAC-NL). We briefly discuss ongoing efforts that apply HISDAC data, including long-term population disaggregation and rural-urban continuum modelling, as well as modelling of the age and expansion patterns of urban street networks.

Moreover, we outline how we employ scanned and georeferenced historical maps and historical aerial imagery for large-scale, retrospective change analysis of forest and settlement extents, integrated with modern remote-sensing based data, and showcase data integration efforts of gridded population data and long-term land use models for refined historical population modelling at 1-km resolution, at global scale since 1900.

To align with the principles of sustainable development, it is essential to understand the dynamics of urban transformation, particularly the mechanisms of building renewal, renovation and new construction. Monitoring the evolution of building volumes and the expansion of constructed area enables a better understanding of urban densification dynamics and supports evidence-based spatial planning. To address this question, this study integrates cadastral and LiDAR data to quantify residential building volume changes in Wallonia (Belgium) between 2010 and 2020. The primary reference data is derived from official regional cadastral records, which primarily describe parcel-level attributes rather than building characteristics. For residential buildings, critical data gaps exist, such as missing building types or floor counts. To address these gaps, we integrate LiDAR-derived height data (Digital Surface Models and Digital Terrain Models) with calibrated cadastral records for 2010 and 2020, ensuring accurate classification of building typology and volume estimation. The volume is estimated for each building by multiplying the footprint area by the building height. The 3D building density is calculated by dividing the sum of the building volumes within the zone of interest by the area of the zone of interest. In addition, the 3D density index is mapped at multiple scales to compare the differences in spatial development of the Belgian territories. Lastly, economic factors such as land values and house prices, as provided by the Federal Public Service Finance (Belgium), are included to study their influence on the expansion and renewal of the urban building stock.

Spatial and temporal characterization of human settlements over large areas is essential for a wide range of applications. Settlement classification supports the monitoring of sustainable urban and rural development, while the analysis of settlement trends enables cause-and-effect assessments that link territorial management and spatial planning to the impacts of implemented policies and activities. This work therefore aims to leverage novel global, multitemporal data on built-up surfaces and population distributions within an integrated framework for detailed characterization of human settlements over time.

Specifically, we use Earth-observation based gridded built-up surface and census-derived gridded population estimates from the Global Human Settlement Layer (GHSL) project of the European Commission’s Joint Research Centre. The GHSL data offer multivariate global data on the built environment and population, consistently enumerated at 100-meter spatial resolution, covering almost five decades. The multivariate nature of GHSL data facilitates the monitoring of large-scale changes in the distribution of population and built-up surface in a joint manner.

In this study, we use GHSL data representing estimates of built-up surface and population for the period 1975–2020, at five-year intervals, and Local Administrative Units (LAU) as a spatial unit of analysis. We first discretize the continuous built-up surface and population densities into quantile-based classes, and categorize the area of interest based on their cross-tabulation. This way, we identify for example highly built-up areas with high, or low population density, modelling the interaction between population and built-up surface. For the areas covered by each of these classes, for each point in time, we observe the absolute changes in built-up surface and population, and derive a wide set of landscape metrics to describe their spatial structure, i.e., their composition and spatial configuration across space and time.

This two-fold assessment, which considers quantitative and spatial changes in human settlements enables the identification of areas undergoing substantial changes in land development from a novel, multi-faceted perspective. The proposed framework allows for the integrated quantification of these developments for each local government unit (LAU). Assessment at the LAU level facilitates data-driven classification using clustering methods, linking derived classes to socio-economic data, and ultimately, a joint assessment of settlement trends using additional indicators.

By using global, spatially and temporally consistent data, the proposed approach is universal and allows for flexible selection of case studies. We tested proposed approach on the example of Poland: a country with strong socio-economic development dynamics in recent decades and the complex settlement structure of urban and rural areas; and are currently working on extending the proposed approach to larger geographic scopes.

Urbanization is an ongoing planetary process. With the ongoing rapid pace of urbanization and the conjunctural environmental and social pressures, better understanding of cities at a global level becomes an ever more pressing matter. In that research agenda, urban structures (here understood as the patterns of composition and configuration of the built and natural urban space) plays the role of a cornerstone. On the one hand, urban structures are a by-product of (even unconscious) societal choices accumulated over time. On the other hand, urban structures have direct impacts on both cities’ inhabitants and the environment. Therefore, the study of urban structures lies at the core of understanding how these have been shaped and how in turn, they shape urban phenomena.

In this context, the constitution of a globally consistent repertory of urban structures is of prime interest to promote systematic analysis of their relations to urban planning and urban phenomena. While remote sensing and geospatial data support an increasing body of research on this topic, these studies have been generally limited in their spatial coverage due to different data and computational bottlenecks. For these reasons, the current focus of research is on urban structures that can be considered salient due to their functional, cultural or aesthetic value. Yet, beyond these types of urban structures lies a vast typology that could enable a more complete and more detailed picture of how urban structures affect their populations and our environment. In the present study, we aim at constituting and analyzing a typology of urban structures as comprehensive and global as possible.

A novel unsupervised clustering approach based on deep learning, the SCAN+RUC framework, is employed on a global Land-use-Land-cover classification product, namely the Local Climate Zones for more than 1500 cities. This dataset describes the urban fabric in terms of morphological characteristics and land cover types (built-up, vegetation, open land, water) at a resolution of 100m. In this study, patches of 3.2km*3.2km of this product are used as a proxy for representing the composition and the configuration of the urban fabric at an aggregated meso-scale. With this, we extract a typology of 138 intra-urban pattern types representing statistically different structures of urban space at a meso-scale.

Analyzing the compositional and configurational specificities of these types, and their global geographical distributions, we identified urban structures associated with specific urban planning practices. From these, we analyze the regions where particular urban structures are present and discuss the dynamics leading to their geographical spreads. With these results, we aim to contribute to debates on the viability of urban structures, urban planning practices and policies across different geographical and cultural contexts.