General Context
SPACE4ALL aims to unravel the climate vulnerability of slum communities in large and secondary cities by combining Citizen Science and Earth Observation methods. Earth Observation methods have the advantage of producing physical data for large areas and have shown their capability to map the physical aspects of slums. However, the hidden aspects of vulnerabilities in relation to hazards are insufficiently captured by these methods. The proposed workflow will combine Earth Observation data with qualitative, rich and diverse data from Citizen Science methods to capture local vulnerabilities. We will develop methods suited to collect data in a typical data-poor environment. The collected data will allow training state-of-the-art deep learning models, which normally fall short because of insufficient data in such areas. The developed novel workflow will assess climate vulnerabilities adapted to the local context of slum communities. Results will inform cross-disciplinary research in terms of methods and access to data where typically large data gaps hinder scientific progress and contribute to innovation in the EO field. All results will be open-access to stimulate cross-disciplinary research. Results include the prediction of the locations of slums, deep-learning process chains, transfer-learning strategy to train deep networks that provide solutions to the common problem of limited training and data citizen science methods to capture the vulnerabilities of slums regarding climate-related hazards. Thus, the results will allow combining climate-related hazards and multiple dimensions of poverty to prioritize risk hotspots in support of local information needs.
