Including Human Behaviour in Models to Understand the Impact of Climate Change on People

By Megan McGrory

In 2020 56% of the global population lived in cities and towns, and they accounted for two-thirds of global energy consumption and over 70% of CO2 emissions. The share of the global population living in urban areas is expected to rise to almost 70% in 2050 (World Energy Outlook 2021). This rapid urbanization is happening at the same time that climate change is becoming an increasingly pressing issue. Urbanization and climate change both directly impact each other and strengthen the already-large impact of climate change on our lives. Urbanization dramatically changes the landscape, with increased volume of buildings and paved/sealed surfaces, and therefore the surface energy balance of a region. The introduction of more buildings, roads, vehicles, and a large population density all have dramatic effects on the urban climate, therefore to fully understand how these impacts intertwine with those of climate change, it is key to model the urban climate correctly.

Modelling an urban climate has a number of unique challenges and considerations. Anthropogenic heat flux (QF) is an aspect of the surface energy balance which is unique to urban areas. Modelling this aspect of urban climate requires input data on heat released from activities linked to three aspects of QF: building (QF,B), transport (QF,T) and human/animal metabolism (QF,M). All of these are impacted by human behaviour which is a challenge to predict, as it changes based on many variables, and typical behaviour can change based on unexpected events, such as transport strikes, or extreme weather conditions, which are both becoming increasingly relevant worries in the UK.

DAVE  (Dynamic Anthropogenic actiVities and feedback to Emissions) is an agent-based model (ABM) which is being developed as part of the ERC urbisphere and NERC APEx projects to model QF and impacts of other emissions (e.g. air quality), in various cities across the world (London, Berlin, Paris, Nairobi, Beijing, and more). Here, we treat city spatial units (500 m x 500 m, Figure 1) as the agents in this agent-based model. Each spatial unit holds properties related to the buildings and citizen presence (at different times) in the grid. QF can be calculated for each spatial unit by combining the energy emissions from QF,B, QF,T, and QF,M within a grid. As human behaviour modifies these fluxes, the calculation needs to capture the spatial and temporal variability of people’s activities changing in response to their ‘normal’ and other events.

To run DAVE for London (as a first test case, with other cities to follow), extensive data mining has been carried out to model typical human activities and their variable behaviour as accurately as possible. The variation in building morphology (or form) and function, the many different transport systems, meteorology, and data on typical human activities, are all needed to allow human behaviour to drive the calculation of QF, incorporating dynamic responses to environmental conditions.

DAVE is a second generation ABM, like its predecessor it uses time use surveys to generate statistical probabilities which govern the behaviour of modelled citizens (Capel-Timms et al. 2020). The time use survey diarists document their daily activities every 10 minutes. Travel and building energy models are incorporated to calculate QF,B and QF,T. The building energy model, STEBBS (Simplified Thermal Energy Balance for Building Scheme) (Capel-Timms et al. 2020), takes into account the thermal characteristics and morphology of building stock in each 500 m x 500 m spatial unit area in London. The energy demand linked to different activities carried out by people (informed by time use surveys) impacts the energy use and from this anthropogenic heat flux from building energy use fluxes (Liu et al. 2022).

The transport model uses information about access to public transport (e.g. Fig. 1). As expected grids closer to stations have higher percentage of people using that travel mode. Other data used includes road densities, travel costs, and information on vehicle ownership and travel preferences to assign transport options to the modelled citizens when they travel.

Figure 1: Location of tube, train and bus stations/stops (dots) in London (500 m x 500 m grid resolution) with the relative percentage of people living in that grid who use that mode of transport (colour, lighter indicates higher percentage). Original data Sources: (ONS, 2014), (TfL, 2022)

An extensive amount of analysis and pre-processing of data are needed to run the model but this provides a rich resource for multiple MSc and Undergraduate student projects (past and  current) to analyse different aspects of the building and transport data. For example, a current project is modelling people’s exposure to pollution, informed by data such as shown in Fig. 2, linked with moving to and between different modes of transport between home and work/school. Therefore the areas that should be used/avoided to reduce risk of health problems by exposure to air pollution.

Figure 2:  London (500 m x 500 m resolution) annual mean NO2 emissions (colour) with Congestion Charge Zone (CCZ, blue) and Ultra Low Emission Zone (ULEZ, pink).  Data source: London Datastore, 2022

Future development and use of the model DAVE will allow for the consideration of many more unique aspects of urban environments and their impacts on the climate and people.

Acknowledgements: Thank you to Matthew Paskin and Denise Hertwig for providing the Figures included.


Capel-Timms, I., S. T. Smith, T. Sun, and S. Grimmond, 2020: Dynamic Anthropogenic activitieS impacting Heat emissions (DASH v1.0): Development and evaluation. Geoscientific Model Development, 13, 4891–4924

London Datastore, 2022: Greater London Authority, London Atmospheric Emissions Inventory 2019.

International Energy Agency, 2021: World Health Organisation 2021, (Accessed January 2023)

Liu, Y., Z. Luo, and S. Grimmond, 2022: Revising the definition of anthropogenic heat flux from buildings: role of human activities and building storage heat flux. Atmospheric Chemistry and Physics, 22, 4721–4735

ONS, 2014: Office for National Statistics, WU03UK – Location of usual residence and place of work by method of travel to work (Accessed August, 2022).

TfL, 2022: Transport for London timetables, (Accessed July 2022)

This entry was posted in Climate, Climate change, Climate modelling, Urban meteorology. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *