Abstract/Summary

Radiation exchanges in cities are impacted by their heterogenous structure, driving changes in within-canopy heat fluxes and temperatures, and increasing vulnerability to extreme weather events. However, high computational costs and a lack of knowledge on city structure constrain the representation of cities for numerical weather prediction often to single-layer models with simplified urban form. This thesis assesses the abilities and benefits of a multi-layer urban radiative transfer model “SPARTACUS-Urban” against a complex building-resolving model “DART”, and a single-layer infinite street canyon approach to urban radiation. The SPARTACUS-Urban shortwave and longwave radiation evaluations demonstrate good agreement with DART. Agreement is best for random cuboid geometries that fulfil the SPARTACUS assumptions that buildings are randomly distributed in the horizontal plane, with normalised bias errors in bulk albedo < 6%, and within-canopy facet absorptions < 15%. Shortwave performance in real-world domains is improved with a modification to building edge length to account for large open spaces, leading to better prediction of radiation penetrating to ground level. The infinite street canyon approach predictstop-of-canopy fluxes well, but SPARTACUS-Urban model performance is better due to the ability to represent realistic descriptions of within-canopy roof and wall variation. To apply SPARTACUS-Urban, vertical urban form data are needed. Statistical relations are developed using data from six cities to obtain profiles from surface building fraction, mean building height, and the effective building diameter. Reducing the needed inputs has little effect on the bulk albedo, but produces considerably larger within-canyon absorption errors, particularly if total wall area is not used. Potentially, the statistical relations can be used globally. SPARTACUS-Urban coupled to the SUEWS urban energy balance model, is used to investigate impacts of real-world geometry. The results are similar for older top-of-canopy net all-wave radiation approaches, but the new system enables simulation of spatially variable surface temperatures.