Currently, 24 people are confirmed dead after a tornado swept through Moore, Okla., on Monday, leaving devastation in its wake.
|Scooped by Nathan Phillips|
Slate's Chris Kirk (@cperryk) mapped the locations, paths, and death tolls of each tornado that killed at least one person since 1950 in the U.S. I continue to be a bit confused by this map. When I've worked with young people while they read maps, I often ask them to talk about the map. What do they see? What do they understand? What makes sense? What doesn't? What questions do they have?
As a way of thinking about my confusion in reading this map as well as the possibilities for spatial analysis that are here, I'll try and produce that kind of a reading (answering the questions above). Here, then, is a bit of a close reading/think aloud from my reading of the map:
Kirk writes that the "area and transparency of each circle reflects the number of people killed." With a quick glance, it's easy to see the relative differences in circle size. Zooming into diferent circles and selecting them (by rolling over with the mouse) reveals three distinct data points for each tornado: date it occurred, Fujita scale score (which I understand as a relative measure of wind speeds), and number of people killed. The size of the circle does seem to correspond to the number of people killed, though Kirk doesn't include a key, so I'm not sure what the scale is or how to think about relative sizes at a glance. In other words, at a glance, I can't get an impression for an approximate number of deaths based on the size of the circle. With a key, I think it would be possible to do this.
But the real confusion for me comes with the use of transparency to reflect the number of people killed. I'm confused because I see almost no variation in the transparency. Again, without a legend, it's difficult to determine what scale Kirk might have in mind here or how he's shifting the relative transparency based on the number of deaths. But to my eye, I see almost no variation. The one exception to this is the 2011 Joplin, Missouri, tornado, in which 158 people were killed. The circle for this tornado is visibly less transparent than others on the map. At a glance, the circle for the 1953 tornado near Flint, Michigan, which killed 116 people, is also slightly darker than those around it.
As I've thought about the map since I first saw it, I realized one way in which relative transparency does reflect death toll: the layering of transparent circles over multiple events indicates areas of the United States in which tornadoes have been more deadly on aggregate over the last 60 years. For example, at a scale in which several states are visible (but not pulled all the way out), the area around Birmingham, Alabama, looks particularly dark due to the layering of circles. And zooming in confirms that this has been a deadly area over time. Over the last 60 years, about 150 people have died in this area from about 10 tornadoes.
One way in which I would like to be able to better read the map is to use the length of the tornado paths to think about the connection between length of path and death toll. Is there a connection? Using the map, it's hard to answer that question in any systematic way because looking at path lengths means zooming in quite closely. However, one argument seems clear: some storms have very long paths and kill very few people. Other storms have short paths and kill many more people. I suspect this is largely related to the population density of the area where the tornado hits--and this is something that's not represented at all on the map.