Changing urban form: detection, transformation, and implications for mental well-being

Publikation: Bog/antologi/afhandling/rapportPh.d.-afhandling

Urban land cover grows at the fastest rate amongst all kinds of land cover globally. Not only does urban expansion influence the environment by competing with other land covers, but the fabric and form of the urban realm also shape people's daily experiences and mental well-being: perceived temperature, air quality, commercial and recreational activities, and landscapes and spaces, which provide a sense of belonging, stress or stress-relief.

In the studies of urban sustainability, time-series data are essential for scientists and planners seeking evidence-based insights and sustainable solutions. Although scientists have measured environmental factors, such as air, rainfall, and temperature with multiple time steps and high spatial resolution, the same has not been done for three-dimensional urban form that considers building height and density. One-time urban form data rests on an assumption of the built environment as static. However, the urban densification occurs at a variety of speeds and dimensions around the world, which points to the need for more spatiotemporal details.

This thesis contributes to improved assessments of changing urban form by answering: (1) Why is long term monitoring of the three-dimensional built environment challenging? (2) How to overcome those challenges? (3) What are the implications of changing urban forms for mental well-being? The dissertation includes one review and five research articles (three written as first-author, and two written as co-author) that cover methodologies and applications.

Article 1 systematically reviews urban remote sensing studies and explains why they predominately focus dichotomous urban/non-urban classification by pointing out a spatiotemporal dilemma in this field – high-resolution imagery, able to detect nuanced urban attributes, has a relatively short temporal span. In contrast, older image-series can detect long-term changes, but has a lower spatial resolution. The article concludes by proposing three pathways to overcome the dilemma: data fusion, artificial intelligence, and selection of meaningful feature scales.

Article 2 addresses the challenge of using Landsat data for monitoring three-dimensional urban densification. Particularly, Landsat’s 30m image resolution makes it difficult to capture 3D structures because the pixel is larger than many urban objects. The study proposes a semantic segmentation framework that uses multi-scale spatial features and shows how this framework improves urban form mapping compared to a simple fully convolutional network (FCN) and random forest (RF) in terms of accuracy as well as spatial and temporal transferability. Urban form time-series data reveal that Danish provincial cities have been densifying horizontally, while the capital suburbs have been growing vertically since the late 1980s.

Article 3 tackles another challenge in urban mapping, namely that of distinguishing between landslides and urbanised areas, because human constructions and landslides are spectrally similar. The study demonstrates a multi-sensor methodology combining nighttime light imagery with multi-seasonal Landsat imagery, and shows how this approach can improve landslide mapping significantly.

Article 4 applies the results from Article 2 to link dynamic urban form patterns to two million Danish people’s addresses, relocation and mental health records to explore associations between urban form and mental well-being. The study shows that mood and neurotic disorders are associated with sprawling low-rise areas and also with transitional experiences associated with moving to or the development of new dense, low areas. In contrast, sparse high-rise areas are associated with lower relative risk of mental problems, implying a positive effect of the intertwined intense built environment and green spaces. These results challenge the general concept that either urban environments or natural environments are uniformly beneficial for mental well-being.

The last two papers are co-authored, and tackle the same issue of harmonised land-cover monitoring. Article 5 provides a harmonised approach to monitor urban green spaces in Denmark using a seasonal filter, three-year rolling window, and between-sensor (Landsat) calibration. Further, Article 6 advances the harmonised land cover monitoring by proposing a deep-learning approach to make images cloud-free.

Taken together, this thesis sheds light on three-dimensional urban form monitoring by proposing methods that provide spatially explicit maps with harmonised time series. The results of horizontal and vertical urban density have been applied to connect people's individual mobility and their mental health, ending with an argument against the simplistic urban-rural gradient of mental well-being. Advances in the spatiotemporal details of urban form are promising to generate significant new insights for the well-being of people.
ForlagAarhus Universitet
Antal sider209
StatusUdgivet - 2020

Se relationer på Aarhus Universitet Citationsformater


ID: 195793553