Doing transformation!? The aspiration and reality of governance for transdisciplinary regional sustainability transformation, using the example of the development process of the Ruhr Academy

Markus Gornik, M.Sc., International Urban and metropolitan Studies

Based on the ongoing dissertation project, the presentation highlights the aspirations and realities of implementing regional sustainability transformation using the example of the Ruhr Academy. The focus is on the question of how transformative governance capacities (Wolfram et al., 2019) can develop in the regional innovation ecosystem of the Ruhr area and what challenges arise in the process. To this end, the science-practice format of the Ruhr Academy—a flagship project of the Ruhr Conference—which failed in the development process, will be evaluated as an example of “doing transformation” based on indicators. Theoretically, the work combines systemic approaches to transformation research (Fastenrath et al. 2023) and mission-oriented innovation policy (Kattel & Mazzucato 2023) with action- and practice-theoretical perspectives from human geography (Reckwitz 2021; Werlen 2020). Empirically, the Ruhr Academy is analyzed as a governance episode (Healey 2006) in order to highlight the positive and negative tipping points between strategic control and exploratory testing of transdisciplinary cooperation in the sustainability transformation of post-industrial regions.

Analysis of the transferability and limitations of AI-based methods for automated detection of impervious surfaces in bi-temporal, aerial remote sensing data

Jan-Philipp Langenkamp, M.Sc.  (Geomatics)

To reduce negative effects such as the loss of soil functions or the emergence of urban heat island effects, the amount of new land used for settlement and transport in Germany is to be reduced to less than 30 hectares per day by 2030. To achieve this goal, spatially explicit and reliable monitoring of impervious surfaces is required. Currently, this monitoring is based on statistical analyses of cadastral data on land use. This thesis investigates the automated detection of impervious surfaces using artificial intelligence (AI) methods based on bi-temporal aerial image data. A central focus is on analysing the transferability and limitations of these methods in seasonal contexts.

Synthesis of multispectral Sentinel-2 data from land use and land cover information using generative deep learning approaches.

Torben Dedring, M.Sc.  (Geomatics)

AI-driven image generation has quickly become integrated into daily life. Remote sensing images have not been spared from being artificially generated, having coined the term “Deep Fake Geography” (Zhao et al. 2021). This research focuses on the generation of multispectral Sentinel-2 images using different generative deep-learning approaches guided by land use and land cover information. In line with the technological development of generative artificial intelligence, this study demonstrates the potential for generating images using Generative Adversarial Networks and Diffusion Models. Those models open up the new fields of time-projected satellite imagery and artificial intelligence for detecting manipulated or generated images.

Zhao, B., S. Zhang, C. Xu, Y. Sun, and C. Deng. 2021. “Deep Fake Geography? When Geospatial Data Encounter Artificial Intelligence.” Cartography and Geographic Information Science 48 (4): 338–352. https://doi.org/10.1080/15230406.2021.1910075