Famine Early Warning Systems in the News

The Famine Early Warning Systems (FEWS) recently announced an emergency for the Horn of Africa. The model now reports that 18 million people are facing severe food shortages. Most of these people are in Ethiopia.

FEWS is the best known instance of computer models that predict potential hotspots of famines. It is also an example of the extraordinary difficulty in creating reliable output at a continental or global scale. These models are very data intensive and therefore depend entirely on the quality of the data. Poor data can result in massive under- or overstatement of a crisis. The models rely heavily on remotely sensed images from which the modelers infer vegetation levels, water/rain availability, and crop conditions. The temptation is to rely primarily on the remote sensing instead of visiting the sites, which may be difficult or dangerous to reach and therefore expensive to monitor. Sophisticated models like FEWS are calibrated with ground based data. The availability of ground based data over areas like the African continent is uneven and local data can be suspect. The Sudanese government, for example, has been known to control the availability, accuracy and interpretation of datasets characterizing their country as a way to play politics with humanitarian relief agencies. So, even with the most careful methods, 18 million is a rough estimate at best. At the same time, even rough estimates can save innumerable lives.

On another matter: Reuters, which carried the story as part of its alert system for humanitarians, has an associated interactive map that I find quite wanting. When I clicked on it I expected to see some numbers related to potential famine. Nothing. Indeed, a pulldown menu, with items like the Indian Ocean tsumami or AIDS in Asia, has at most standard map layers (roads, river, city locations). No information related to the subject. Also, the legend is broken for most of the links. Come on, guys, if you want map technology related to your stories then implement something. Don’t give us a standard atlas! Actually, this interactive map contains less information than an atlas. For a much better interactive map, see the Famine Early Warning Systems site.

Comments are closed.