A new approach to modelling the energy balance of a complex urban geometry

Urban areas are characterised by a complex three-dimensional geometry of buildings and trees. Urban climate models can operate at the mesoscale (100 m to 2 km resolution; urban canopy models) and greatly simplify the urban geometry (e.g. by assuming an infinite street canyon), or at the building scale (about 1 m resolution; micrometeorological models). The building-resolving models use the radiosity method in combination with a matrix of view factors to compute the exterior radiative exchange and separate algorithms to compute heat conduction in the buildings. The disadvantages of these methods are their strong dependence on the number of elements in the scene (in terms of computational time and/or memory usage) and the challenges of representing different physical processes such as spectral material reflectivities or specular reflections.

This study presents a new approach to calculate the combined radiative (solar and terrestrial infrared) exchange, heat conduction, and convection in a complex 3D urban geometry. It is based on a Monte Carlo solver (stardis) for combined radiation, conduction, and convection, which is used to calculate the urban surface temperature as a function of the meteorological forcing provided by the meteorological model (Meso-NH) representing the buildings with an immersed boundary method. Stardis solves the physical equations without a computational grid and is therefore very well suited to deal with a highly complex 3D geometry. A very important limitation of the presented model is that moist processes are currently not included. A validation of the sensible heat flux, the near-surface air temperature profile, and the skin surface temperature simulated by Meso-NH-stardis is performed for the COSMO outdoor urban test site for dry heat wave conditions. The new model and the validation results are presented.