SearcharxivSearch

arXiv subjects

Sue Grimmond

Publications and source records attributed to Sue Grimmond.

2 recordsLinked to original sources

Modelling human activities in a system of cities

Cities host most of the world population with diverse services and activities. One key challenge in urban modelling is the quantification of intra- and inter-city mobility patterns and the associated space-time dynamics of population density and anthropogenic activities. To address this, we apply the novel agent-based urban model DAVE (Dynamic Anthropogenic actiVities and feedback to Emissions) to simulate population behaviour and mobility in the Vaud and Geneva Cantons, a system of small- to medium-size cities in Switzerland. Simulation results provide detailed temporal (10 min) and spatial (500 m) population dynamics for different age groups and day types. DAVE further models the time-varying population distribution in 11 different microenvironments (e.g., home, work, leisure, outdoor) and the travel flows by different modes. Simulation results align with observations, confirming the possibility of driving urban system modelling with statistical information on residents' behaviour. Sustainability and health indicators like daily driving distance and walking time for each neighbourhood are also reflected by the model with urban-rural gradients displayed. This work serves as a foundation for future applications of DAVE to study bottom-up human-built environment interactions, from anthropogenic emissions and building energy to urban climate, exposure, and health in cities around the world.

physics.soc-ph

Machine Learning Emulation of Urban Land Surface Processes

Can we improve the modeling of urban land surface processes with machine learning (ML)? A prior comparison of urban land surface models (ULSMs) found that no single model is 'best' at predicting all common surface fluxes. Here, we develop an urban neural network (UNN) trained on the mean predicted fluxes from 22 ULSMs at one site. The UNN emulates the mean output of ULSMs accurately. When compared to a reference ULSM (Town Energy Balance; TEB), the UNN has greater accuracy relative to flux observations, less computational cost, and requires fewer input parameters. When coupled to the Weather Research Forecasting (WRF) model using TensorFlow bindings, WRF-UNN is stable and more accurate than the reference WRF-TEB. Although the application is currently constrained by the training data (1 site), we show a novel approach to improve the modeling of surface fluxes by combining the strengths of several ULSMs into one using ML.

cs.LG