Modelling the urban heat island of London, and implications for heat-related mortality during the 2018 summer heatwave

The impact of cities on the local climate is a well-known and -studied phenomenon. In particular, cities increase local air temperature, particularly at night, and creating what is called the urban heat island (UHI). The UHI in the UK’s capital city of London was one of the first to be quantified by Luke Howard around 1830. Since then, many studies have measured, or modelled, the impact of urbanization on local temperatures, and considered the potential impacts on heat-related mortality. Nevertheless, these studies are often: i) focused on short-time periods – e.g., constrained to few days of heatwave; ii) lack spatial density and/or representativity of measurements; or iii) don’t report a method that would make their results and outcomes comparable to other cities.

Our aim is to make coherent spatio-temporal estimations of the burden that cities bring in terms of heat-related mortality. To achieve this, we ran two 3-months (June to August) regional climate simulations at 1 km horizontal resolution using the Weather and Research Forecasting (WRF) which consist of two simple scenarios: with and without the city. For both scenarios, the model was parameterized using the new standardized WUDAPT-TO-WRF python tool. In the natural scenario, surrounding natural pixels from MODIS were considered most probable land covers and replace the city. In the urban scenario, urban canopy parameters were obtained from the European Local Climate Zones (LCZ) map. We used the complex three-dimensional Building Effect Parameterization urban canopy model with its Building Energy Model (BEP-BEM) to represent the urban effect in the urban scenario. The simulations were run for the 2018 summer and its 4 heatwaves over London and the south east of England. The model was evaluated for its urban scenario against a variety of earth observations and meteorological measurements from official and crowd-sourced data. Finally, we bias-corrected the urban and the natural scenario using an innovative method that relies on official automatic and citizen weather stations. This way, me make sure that the calculated heat anomaly induced by the city is as representative as possible , and allows us to quantify the proportion of heat related mortality which we attribute to the urban heat island in London.

Our study is considered one of the first to model a whole seasonal impact of a city on its local climate using a highly complex urban canopy model and a standardized method of parameterization. Our bias-correction method is also expected to provide key perspectives on the joint utility of modelled and crowd-sourced weather data for heat-related epidemiological studies.