AUM2020 Session 13 - Urban modelling and the planning of the built environment
Duration: 1 hour 59 mins
Share this media item:
Embed this media item:
Embed this media item:
About this item
Description: |
1. UK2070 Futures Modelling: Results from pre- and post-Lockdown scenarios
Dr Ying Jin, Mr Steve Denman, Dr Li Wan, ( University of Cambridge) 2. Using Machine Learning and GPU Processing to Build Faster and More Accurate Integrated Models Prof. Paul Waddell, University of California Berkeley Discussant: Prof. Adam Dennett, UCL Session chair: Dr Jamil Nur, University of Cambridge |
---|
Created: | 2020-12-02 08:36 |
---|---|
Collection: | Martin Centre AUM2020: Modelling the New Urban World |
Publisher: | University of Cambridge |
Copyright: | The Martin Centre |
Language: | eng (English) |
Keywords: | AUM2020; Applied Urban Modelling; Urban modelling and the planning of the built environment; Martin Centre; Architecture; session 13; |
Abstract: | 1. The UK2070 Futures Study aims to investigate distinct scenarios regarding spatial rebalancing and development in the UK, with a focus on the growth and change in jobs, supply and demand of housing and provision of transport infrastructure and services. The distinct scenarios are designed to explore a wide range of potential economic and demographic development trajectories that are cogent for policy purposes, both pre- and post-Pandemic. The modelling work is to examine the effectiveness of existing and potential options for intervention in a long term policy programme in the next 50 years – to 2070.
The study is intended to fill a persistent gap in the available evidence for making decisions on future developments across the UK. Policy makers, business leaders, community activists and academic researchers all aspire to coordinated interventions on jobs, housing and transport. However, the theories and data regarding the interactions among those sectors are complex, and there are few current studies available examining how these sectors actually connect and interact. This study builds on the research work from a wide range of institutions and individuals taking part in the UK2070 Commission research, and connects the insights from those distinct disciplines through a theory of spatial equilibrium that articulates the interactions among the sectors of economy and society that are key to spatial development decisions. In particular, a computer simulation model is used to understand and represent the multiple interactions which shape choices on jobs, housing and transport. This computer simulation model is first checked for fidelity in its predictions of business and consumer choices, and then used as a digital laboratory to test a wider range of policy and community interventions than what could be possible through thought experiments or single-sector analyses. The findings from the model are reported for comment by both specialists and non-specialists in an interdisciplinary context. The model findings have thus far supported the UK2070 Commission in its deliberations on the options of policy interventions and prioritisation against a broad, strategic understanding of the major opportunities and challenges facing the UK. 2. Integrated urban models have made numerous compromises while seeking to remain practically applicable to support actual planning decisions. Two common stress points in modeling are computational speed and accuracy, ideally measured longitudinally as well as spatially, and with data held out from the model building stage. Computational performance is critical in order for models to be practically useful in planning contexts, and to allow running sufficient number of scenarios, or to run uncertainty analysis, or both. Computational performance improvements are also necessary in order to increase the spatial resolution of of locations and networks within models to better support modeling of active modes and transit access. This talk describes ongoing work to overcome existing barriers to improving computational performance and model accuracy within a suite of integrated models, leveraging machine learning and GPU computing. The goals include enabling much more rapid construction and calibration of models that attain high quality longitudinal validation, and reducing run times of models dramatically, while increasing their spatial resolution. We leverage GPU processing to parallelize microsimulation of traffic flows on large scale networks at the metropolitan scale, achieving high computational performance along with promising validation results. We have begun using machine learning to improve predictive accuracy of price models to seed the structural microsimulation models of demand and supply in housing markets. And we have been developing a differentiable programming approach to building microsimulation models that enable calibrating the models longitudinally using machine learning algorithms, achieving high quality longitudinal validation by tuning the model parameters. |
---|
Available Formats
Format | Quality | Bitrate | Size | |||
---|---|---|---|---|---|---|
MPEG-4 Video | 640x360 | 1.33 Mbits/sec | 1.16 GB | View | ||
WebM | 640x360 | 457.18 kbits/sec | 398.48 MB | View | ||
iPod Video | 480x360 | 487.19 kbits/sec | 424.63 MB | View | ||
MP3 | 44100 Hz | 251.53 kbits/sec | 219.23 MB | Listen | ||
Auto * | (Allows browser to choose a format it supports) |