Stockholm university

Research project AI-powered knowledge integration to Carbon-neutral Cities

Transition of cities towards carbon neutrality requires an improved understanding of the interlinkages between land-use and energy planning, and of the complexity of coupled socio-ecological processes, and feedbacks in urban-regional systems.

WP5 Knowledge transfer and communication
WP5 Knowledge transfer and communication

Artificial intelligence-based decision support systems (AI-DSS) can help develop and improve this understanding through learning, mapping, and forecasting complex urban interactions in a scenario-based analytical framework. The general aims of this project are to initiate a collaborative learning process and develop an AI-DSS for handling the complex interactions, and their effects and feedbacks, in urban transformation and associated land-use changes and climate change impacts. The project will provide new scientific knowledge on the interactions between human and natural systems and support sustainable urban spatial planning and carbon-neutrality goals in urban areas. The highly replicable decision support system developed (including a deep transfer learning approach) will increase the ability to improve climate change mitigation and adaptation strategies, by helping different stakeholders in different places to share experiences and learn from each other. The AI-DSS will enable a wide range of users and stakeholders to access, operate, and feed back into the models relatively easily, to test various urban planning and policy scenarios, and to assess effects on GHG emissions.

Project description

The overarching aim of the project is to initiate a collaborative learning process in order to develop a replicable and adaptable, artificial intelligence-based decision support system (AI-DSS) for handling complex socio-physical interactions, their implications, and potential feedbacks. The AI-DSS will couple socio-ecological systems models, including: i) a spatio-temporal regional land-use change model ii) a socio-econometric forecasting model, and iii) dynamic, spatial-explicit impact models for assessing the socio-environmental implications of socio-economic and geophysical activities in the region, including models for greenhouse gas (GHG) emissions assessment (including carbon sinks, storage potential, and the building and transportation sectors).

The project will:

  • Develop and apply an open-source, cloud-based spatial multi-agent system for the Stockholm region to explore scenarios of urban-regional development (urban transformation) over a long planning horizon (to 2050); and apply the approach to other urban centers across Sweden and the world.
  • Build an interactive user-friendly internet interface for systematic data collection and feedback from various stakeholders. The goal is to increase the applicability, local viability, and use of the decision support system, while also promoting collaborative learning.
  • Project GHG emissions from human activities and developments in complex urban environments, such as residential/commercial buildings, transportation, construction, for different scenarios of urban transformation, through ML-based prediction approaches.
  • Map and inventory carbon sequestration potential in the study areas to facilitate use of nature-based solutions (NBS) to increase carbon storage and/or reduce GHG emissions.
  • Use AI-based deep transfer learning to enable easy replicability of the coupled socioecological system model and interface to urban areas across Sweden and the world.

The project will develop, test, and apply coupled complex systems models in a novel AI-DSS developed collaboratively by a network of researchers, urban planners, and other stakeholders (local NGOs, residents, and businesses). The Stockholm region in Sweden will be used as a test case to apply and train the models. The trained modeling system will analyse various scenarios of urban transformation over both short- and long-term time horizons. The trained models and AI-DSS will then be prepared for application to different test cities (Gothenburg, Sweden; Chicago, USA; and Nanjing, China) for comparison of models, comparison of potential climate mitigation solutions, and testing of deep transfer learning techniques. Feedback from a wide range of users and stakeholders will be solicited from each application region. Social learning across regions will be measured and evaluated. The work is organized around five interacting work packages (WPs), each of which oversees and/or is responsible for various parts of the work during the four-year project period

WP5 Knowledge transfer and communication
WP5 Knowledge transfer and communication

 

Project members

Project managers

Zahra Kalantari

Forskare

Department of Physical Geography
Zahra Kalantari

Members

Georgia Destouni

Professor i hydrologi

Department of Physical Geography
Gia Destouni

Zahra Kalantari

Forskare

Department of Physical Geography
Zahra Kalantari

Haozhi Pan

Associate Professor

Shanghai Jiao Tong University

Brian Deal

Professor

University of Illinois Urbana-Champaign

Helena Näsström

Region Stockholm

News