Overview

Climate change is one of the defining issues of the twenty-first century and will have profound implications for poor and vulnerable people. In particular, extreme weather events will lead to significant increases in displacement and essential needs including protection, food, water, shelter, personal safety, health and transport in Sub-Saharan Africa. We seek to answer the questions: What do poor people need when they are faced with extreme weather events and other climate change impacts? And can we anticipate resulting essential needs to minimize economic disruption?

In a first case study for Somalia, we combine survey data on essential needs with climate data to assess how different extreme weather events affect demand for essential needs. We plan to use satellite imagery to extend our analysis in two ways. First, we can use satellite imagery to extract more information about the circumstances of displacement and use model selection procedures to find features that help predicting essential needs. Second, satellite imagery allows us to scale-up the application of the model to other countries and regions. Except for Somalia, detailed survey data on displacement is rarely available but satellite imagery might allow us to measure migration using changes in the land surface, such as the physical expansion of displacement camps. The Somalian survey data can be used to validate any potential proxies that are based on satellite imagery.

Providing predictions of the likely essential needs arising from extreme weather events improves the capacity of local decision makers to respond early and prevent more harm. We will develop an end-user tool that generates these predictions and that can be accessed freely: the Oxford Essential Needs Action Tracker (OxENAT).

Approach and Model

OxENAT provides a systematic way to track and forecast essential needs as a basis for economic policy options for governments across countries and sub-national jurisdictions over time. It collects systematic information from climate datasets, satellite imagery and socio-economic indicators to track, compare and forecast essential needs (including protection, food, water, shelter, personal safety, health and transport) rigorously and consistently. We transform this data into a model that aggregates various measures of economic policy responses. This model is used to describe variation in essential needs responses, explore whether the economic policy response affects the rate of essential needs, and identify correlates of more or less intense responses in the context of extreme weather events.

In this proposal, we propose to develop a web-based action tracker for a case study region to make vast amounts of data and tools accessible for early policy actions to policy makers and people in need. Our conceptual framework is straightforward: we begin by compiling a large amount of spatial and temporal data. We obtain this data at the subnational level and assemble a multinomial choice model. We use the surveyed reasons for displacement, additional climate data, and attributes of the journey's origin and destination to explain the most important essential needs. Satellite imagery can be used to construct new features that might prove helpful in explaining the data.

Dashboard and API

We then use our model to produce predictions of future essential needs – potentially in real time – and wrap all this behind an application programming interface (API). The final product will include both a public dashboard and a decision support web app for policy makers to use for early economic policy action. The web app will show the results of our analysis, which could be illustrated on an interactive map.

Data 

As a starting point of the OxENAT project, we propose a case study on Somalia using their rich survey data: First, we need historical and predicted meteorological forcing data such as temperature, and precipitation up to six months into the future. We combine this data with observed essential needs data at the sub-national level. We also condition on socio-economic data including population, irrigated agricultural area, and GDP. The data will be aggregated using district-level geometry which allows us to take advantage of the granularity of the data. We use all this data to predict essential needs over space and time for various extreme weather events. After validating proxies based on satellite imagery, we suggest broadening the geographical scope by using satellite data to supplement ground-level data collection in areas most at-risk of climate-induced economic and humanitarian havoc.