Are detailed urban canopy parameters necessary for modelling the urban climate in Africa?

Local Climate Zones (LCZ) have now been widely accepted and used by the urban climate community (Ching et al., 2018). However, their use over Sub-Saharan Africa has still been limited because of data scarcity in the region. Brousse et al. (2019, 2020) demonstrated the added value of applying spatially variant urban canyon parameters derived from LCZ in the urban climate model TERRA_URB – embedded in the COSMO-CLM model. Despite its promising results, thermal and morphological parameters extracted out of the ranges proposed by Stewart and Oke (2012) are mostly derived from Western cities. Hence, uncertainties related to the use of unascertained urban forms and functions of African cities for urban climate modelling have not yet been evaluated.

To quantify the sensitivity of the model to more representative urban canopy parameters of African cities, this study sets up a methodology for: (i) obtaining from in situ measurements archetypal parameters of LCZ classes for Kampala (Uganda); and (ii) simulating the potential effect of the newly defined urban structure on the local climate.

In situ data were obtained during field work held in the summer months of 2018. A representative sample of 1300 measurement points was selected throughout the city of Kampala, for which both quantitative (road width, distance between houses, heights of buildings) and qualitatively estimated (vegetation fraction, road-wall-roof material) variables were collected.  These variables enabled the development of an updated LCZ map of the city of Kampala.

To evaluate the model’s sensitivity to the new spatially explicit urban morphological and thermal parameters, this new information was fed into the TERRA_URB scheme at a horizontal resolution of 1 km for a 3-months period (December 2017 – February 2018). The run was nested within a 12 km simulation forced by ERA-Interim reanalysis data. Results show tangible effects of the updated parameters on the 2-meter air temperature, land surface temperature and surface energy balance components. Still, no major improvements in model skill compared to the default LCZ framework proposed by Brousse et al. (2020) were found. [1] [WT2] This study is among the first studies to test the sensitivity of an urban climate model to more realistic urban parameters in Africa and aims at triggering more research to be done in the area with a variety of urban climate models.