Advancing large-scale analysis of human settlements and their dynamics

Due to the importance for a range of sustainability challenges, it is important to understand the spatial dynamics of human settlements. The rapid expansion of built-up land is among the most extensive global land changes, even though built-up land occupies only a small fraction of the terrestrial biosphere. Moreover, the different ways in which human settlements are manifested are crucially important for their environmental and socioeconomic impacts. Yet, current analysis of human settlements heavily relies on land cover datasets, which typically have only one class to represent human settlements. Consequently, the analysis of human settlements does often not account for the heterogeneity within urban environment or their subtle changes. This simplistic representation severely limits our understanding of change processes in human settlements, as well as our capacity to assess socioeconomic and environmental impacts. This thesis aims to advance large-scale analysis of human settlements and their dynamics through the lens of land systems, with a specific focus on the role of land-use intensity. Chapter 2 explores the use of human settlement systems as an approach to understanding their variation in space and changes over time. Results show that settlement systems exist along a density gradient, and their change trajectories are typically gradual and incremental. In addition, results indicate that the total increase in built-up land in village landscapes outweighs that of dense urban regions. This chapter suggests that we should characterize human settlements more comprehensively to advance the analysis of human settlements, going beyond the emergence of new built-up land in a few mega-cities only. In Chapter 3, urban land-use intensity is operationalized by the horizontal and vertical spatial patterns of buildings. Particularly, I trained three random forest models to estimate building footprint, height, and volume, respectively, at a 1-km resolution for Europe, the US, and China. The models yield R2 values of 0.90, 0.81, and 0.88 for building footprint, height, and volume, respectively. The correlation between building footprint and building height at a pixel level was 0.66, illustrating the relevance of mapping these properties independently. Chapter 4 builds on the methodological approach presented in chapter 3. Specifically, it presents an improved approach to mapping 3D built-up patterns (i.e., 3D building structure), and applies this to map building footprint, height, and volume at a global scale. The methodological improvement includes an optimized model structure, additional explanatory variables, and updated input data. I find distance decay functions from the centre of the city to its outskirts for all three properties for major cities in all continents. Yet, again, the height, footprint (density), and volume differ drastically across these cities. Chapter 5 uses built-up land per person as an operationalization for urban land-use intensity, in order to investigate its temporal dynamics at a global scale. Results suggest that the decrease of urban land-use intensity relates to 38.3%, 49.6%, and 37.5% of the built-up land expansion in the three periods during 1975-2015, but with large local variations. In the Global South, densification often happens in regions where human settlements are already used intensively, suggesting potential trade-offs with other living standards. These chapters represent the recent advancements in large-scale analysis of human settlements by revealing a large variation in urban fabric. Urban densification is widely acknowledged as one of the tangible solutions to satisfy the increased land demand for human settlement while conserving other land, suggesting the relevance of these findings to inform sustainable development. Nevertheless, local settlement trajectories towards intensive forms should also be guided in a large-scale context with broad considerations, including the quality of life for inhabitants, because these trade-offs and synergies remain largely unexplored in this analysis.