ADRELO: Advancing Resilience in Low Income Housing Using Climate-Change Science and Big Data Analytics

The project aims at enhancing the resilience of low-income communities living in disaster prone areas. The focus is on low-lying coastal zones that have a high risks of droughts and floods in selected parts of East Africa, Brazil and North America. It develops the geographic and socio-economic knowledge of persons living in slum and riverbed areas by gathering georeferenced data on infrastructures and information on the natural heritage of project sites. The project team will also investigate technology adoption barriers and diffusion drivers through designing and prototyping an affordable, disaster-resilient, low-income housing system that use sustainable locally-resourced materials. The development of urban spaces is a function of geographic location, economic history, urban development pattern, and therefore governance will have a bearing on resilience. Still, given that development (or lack thereof) of an urban center is an outcome of existing social, economic, and political inequities political inequities; policy packages for disaster preparedness that do not consider the unique circumstances of vulnerable populations can inadvertently cause harm to low- income households. Furthermore, policy packages will include environmental sustainability and public health considerations. The research will also contribute to accurate modelling of climate and extreme weather events at spatiotemporal level to increase the understanding of climate scientists while empowering policy makers in disaster related decision-making. Machine Learning and Big Data Analytics will be used for climate modelling and to identify optimal disaster resilient-housing urban design and planning policy packages considering projected climate change- related extreme weather scenarios between the current time and 2050. Whilst Big Climate Data is amenable to long-term climate prediction, data for localized and seasonal predictions is still uncertain and sparse. Machine Learning has potential to handle this uncertainty and data sparsity as other applications have demonstrated that it can work with either big data or sparse data.

The following are challenges to be solved to reduce disaster risks and enhance resilience. First, difficulty in obtaining accurate localized predictions make preparation and mitigation difficult and makes decision making uncertain and misinformed. Secondly, the inability to more elaborately link populations, climate change, disaster risk, policy actions and interventions with impacts on disaster risk reduction and resilience mean that the impacts of the actions of policy makers cannot be quantified. This project seeks to use big data and machine learning to produce models that perform localized predictions. Also, socio-economic modelling will be applied to link climate change, disaster risk, policy actions and impacts on disaster resilience. The resulting models will be integrated into visual analytics tools as part of a decision support system that can support policy making. The idea is to generate actionable insights that can be encourage decision makers to consider various policy choices while simultaneously evaluating the associated impacts to disaster risk reduction and increased resilience. Still, the project will develop technology for low income housing based on affordable and resilient materials and technology.

The application of social-econometric strategies to design, appraise and monitor the implementation of the structures proposed especially for the economic and social impacts is highly desirable. This will require an analysis of various variables such as the cost implications and access parameters of materials to the local communities, the demographic sub-contexts of post disaster resettlement as well as the possibility of ongoing sunk costs associated with post implementation and maintenance. The models tested will be applied locally and these can then be adapted for specific climate change-related global phenomena.

In dealing with specific risks such as drought and flooding, the data already generated can help understand the current and future socio-economic impacts of such risks at aggregate and dis-aggregate levels. This can then be evaluated against the backdrop of installing risk reducing structures such as shelters with access to food and medicines to ensure the transitionary period from disaster to reinstatement- especially where the vulnerable in the population like the elderly and children are concerned- renders itself with as little consequence as possible. In efforts directed at informing policy, expected future economic impacts such as increased output due to continuity of occupancy in the region and a constant or increasing population as opposed to periodic periods of prolonged evacuation will be examined to inform the implementation of longer term infrastructural adaptations that not only reduce the effects of any risks experienced, but also allow for shorter recovery times and benefits from economies of scale.

The inter-disciplinary nature of the research will allow the project team to estimate not just the feasibility but also the sustainability of the affordable, resilient and affordable housing that emerge from the research. The socio-economic models will be integrated into decision support tools for policy makers. Furthermore, this will culminate into accessible policy reports and briefings towards ongoing feedback for the duration of the project as well as for suggested policy options after.

The project will have the following impacts: 1) A large set of data on weather, social economics, demographics, natural heritage, infrastructures will be available for the three sites in North east of Brazil, eastern USA and East Africa that at present do not exist. 2) The dwellers of low-lying areas that are vulnerable to humanitarian disaster will be assessed for each project site. 3) Climate models and visual models for the sites will be available. 4) Improved housing that are durable and affordable will be made public. 5) Visual model tools for policy making will also be made available.