Keywords
land-use change; urban growth; climate change; urban heat island; local climate zones
Start Date
7-7-2022 2:20 PM
End Date
7-7-2022 2:40 PM
Abstract
The urban heat island effect leads to higher temperatures in cities compared to rural areas. The combined effect of climate change and strong urban growth will as such have a massive impact on human health in growing large cities, especially in tropical or humid regions. Climate models usually do not take urban growth into account or do not have detailed data on the specific characteristics of the urban area. Most land-use change simulations are driven by functional land uses (e.g. residential land, commercial land, agriculture), while an urban climate model rather needs more information on climate-relevant surface properties such as the density and height of the built-up land. The Local Climate Zone (LCZ) scheme is a land-cover classification that includes several built-up morphologies and that can be mapped from remote sensing with a workflow defined by the “World Urban Database and Access Portal Tools” (WUDAPT) project. We have applied this workflow to four Indian cities (Bangalore, Lucknow, Guwahati and Ahmedabad). A main limitation is that confusion between LCZ categories can be high, but we tackled this by combining results with other data sources (urban masks, open GIS data) to make 30 m resolution maps for two time steps (present and 10 years ago). These maps are then generalized to 100 m, and used as input to calibrate a cellular automata (CA) land-use change model. Special attention was paid in the semi-automated calibration (with a CMA-ES algorithm) to gradual evolutions between land covers with different densities. The calibration can directly be used to simulate business-as-usual (BAU) urban growth towards 2070. In addition to BAU growth, two other policy and demographic scenarios were defined with different population growths and urban planning regulations. Finally, all resulting urban growth maps are again generalized to a 200 m resolution, and served as input to model present and future urban climate, and to calculate different climate and health indicators for all cities.
Modelling urban growth and future urban climate in Indian cities with local climate zones as land uses
The urban heat island effect leads to higher temperatures in cities compared to rural areas. The combined effect of climate change and strong urban growth will as such have a massive impact on human health in growing large cities, especially in tropical or humid regions. Climate models usually do not take urban growth into account or do not have detailed data on the specific characteristics of the urban area. Most land-use change simulations are driven by functional land uses (e.g. residential land, commercial land, agriculture), while an urban climate model rather needs more information on climate-relevant surface properties such as the density and height of the built-up land. The Local Climate Zone (LCZ) scheme is a land-cover classification that includes several built-up morphologies and that can be mapped from remote sensing with a workflow defined by the “World Urban Database and Access Portal Tools” (WUDAPT) project. We have applied this workflow to four Indian cities (Bangalore, Lucknow, Guwahati and Ahmedabad). A main limitation is that confusion between LCZ categories can be high, but we tackled this by combining results with other data sources (urban masks, open GIS data) to make 30 m resolution maps for two time steps (present and 10 years ago). These maps are then generalized to 100 m, and used as input to calibrate a cellular automata (CA) land-use change model. Special attention was paid in the semi-automated calibration (with a CMA-ES algorithm) to gradual evolutions between land covers with different densities. The calibration can directly be used to simulate business-as-usual (BAU) urban growth towards 2070. In addition to BAU growth, two other policy and demographic scenarios were defined with different population growths and urban planning regulations. Finally, all resulting urban growth maps are again generalized to a 200 m resolution, and served as input to model present and future urban climate, and to calculate different climate and health indicators for all cities.
Stream and Session
false