Part IIIHumanitarian Action
In the wake of climate change, the frequency and severity of natural disasters like hurricanes and floods have increased. But these types of disasters—and others like earthquakes and avalanches and wars—have existed since the dawn of humanity.
Humans have gotten better at responding to the emergencies that disasters create. But emergency response is often challenged by difficulties obtaining accurate data on, for instance, the extent of damage to a particular locale and the food needs that communities who are impacted by disasters have. Those challenges can both delay emergency response and misdirect scarce resources.
In this part of the book, we provide several examples of how a variety of datasets—from satellite data to household survey data—can be used to rapidly and accurately identify humanitarian needs. Generally, the methods used rely on longitudinal data—to assess the extent of damage that we see today, we need to know whether that damage was present yesterday. Further, the models described herein attempt to provide nuanced data on damage. Many existing models use a binary classification scheme to determine, for instance, building damage as damaged or not damaged. A more nuanced approach attempts to determine the degree of damage that a building sustained, and, further, to estimate the number of people potentially impacted by that damage. This level of information can help emergency response efforts focus on areas where there is the greatest ...