Mapping Long-Term Dynamics of Population and Dwellings Based on a Multi-Temporal Analysis of Urban Morphologies

Abstract

Information on the distribution and dynamics of dwellings and their inhabitants is essential to support decision-making in various fields such as energy provision, land use planning, risk assessment and disaster management. However, as various different of approaches to estimate the current distribution of population and dwellings exists, further evidence on past dynamics is needed for a better understanding of urban processes. This article therefore addresses the question of whether and how accurately historical distributions of dwellings and inhabitants can be reconstructed with commonly available geodata from national mapping and cadastral agencies. For this purpose, an approach for the automatic derivation of such information is presented. The data basis is constituted by a current digital landscape model and a 3D building model combined with historical land use information automatically extracted from historical topographic maps. For this purpose, methods of image processing, machine learning, change detection and dasymetric mapping are applied. The results for a study area in Germany show that it is possible to automatically derive decadal historical patterns of population and dwellings from 1950 to 2011 at the level of a 100 m grid with slight underestimations and acceptable standard deviations. By a differentiated analysis we were able to quantify the errors for different urban structure types.

Cartographic reconstruction of building footprints from historical maps: A study on the Swiss Siegfried map

“…The question remains whether DL is superior to other extraction algorithms that rely on more traditional computer vision techniques, such as color segmentation, morphological operators, and template matching, or their combination. While the literature rarely provides enough information to re‐implement existing approaches (see for example, the vague workflow explanations by Frischknecht and Kanani and Hecht et al ), a comparison on a higher level is possible. In fact, O'Mahony et al () provide a thorough discussion on this subject.…”

“…Concatenating each pixel of a footprint contour is probably the simplest approach to obtain such a polygon. An example for such a result can be found in the work by Hecht, Herold, Behnisch, and Jehling (). Subsequently applying a generalization algorithm reduces the number of vertices but still does not yield a footprint of cartographic quality (i.e., one that is comprised of mostly parallel walls and perpendicular corners).…”

Tools for mapping multi-scale settlement patterns of building footprints: An introduction to the R package foot

“…Specifically, within much of the work involving building maps there is often a goal to identify or classify similar patterns of morphometric characteristics. The identified classes can then help distinguish intra-urban neighbourhoods, differentiate settlement types or periods of development [ 3 , 8 , 13 , 14 ].…”

“…Accurate and complete maps of buildings are a foundational data layer for researchers and practitioners seeking to understanding cities and characteristics of the built environment. Identifying and mapping the footprints of structures and their agglomerations into human settlements is a first step towards improving our understanding of local population patterns, providing services and utilities to all areas, and mapping building stock and urban extents [1][2][3][4]. Within the field of planning and particularly the area of urban morphology, the form and the patterns of buildings in space have been a means to explore the history of cities and the political economic and social forces shaping their developments [5,6].…”

Dasymetric Mapping Using UAV High Resolution 3D Data Within Urban Areas

“…The dasymetric mapping is essential for the multi-temporal analysis of census datasets, when the shapes of the census tracts do not coincide [9]. The application of dasymetric mapping for the estimation of population size can be improved by using ancillary 2D or 3D data related to the buildings.…”