Sandy was a painful reminder of extreme weather events becoming more frequent all over the world. Our September issue featured a story on the topic with this original chart by Senior Graphics Editor John Tomanio (click on it for a detailed view):
The chart shows all weather events since 1980 that caused at least one billion dollars in damages. We chose to measure all events in terms of economic damage to have comparable data for very different events, which are color coded by type (drought, flood, etc). The events are represented by half circles, an unusual chart type seen more often in election charts. I anticipate professor Alberto Cairo (see his terrific blog The Functional Art here) may not be thrilled with it!. As he explains, readers are used to compare visual differences with charts that grow in one dimension only (as a bar chart) rather than two (as any area chart like a pie chart or this one). I fully agree with that but in this case the overall pattern of distribution over time is more important than the individual values: We are seeing more extreme events, and more intense, in recent years. That is apparent in a quick look and the quick take-away message the chart tries to convey. Area charts tend to visually de-emphasize the differences between different values, and that was helpful in a graphic that needed to concentrate so many data points in a small space.
The half circle format provides a common baseline to all events in each year, a terrific idea John had when full circles proved to look too disorganized in the page.
EARLIER VERSIONS: John’s analysis of the 32 years of $1 billion weather disasters
from NOAA led to this first approach: the first 16 years of data (from 1980 to 1995) showed 46 weather events, while the next 16 year period (1996 to 2011) showed 87 events. John roughly plotted the events on two maps, each representing the two time periods. Although subsequent visual approaches would vary quite a bit, this original analysis informed the structure of all versions, comparing the two
time periods. (The “Why is weather extreme?” panel eventually moved to
another part of the story.)
This rough is a bolder, more graphic approach that still geo-locates the weather events on the small maps at bottom. While this proved to be the quickest read of all the versions, John wanted to explore another idea that added a time element to the weather events.
Below is the first attempt of roughly plotting all the weather events by the month and year which they occurred. We liked to see how the frequency of events builds over time, but while the center of each circle corresponds to the month and year of the event, the range of circle sizes made it hard to see that. It needed more order.
By using semi-circles, the final version (first image in this post) makes much clearer to see how the weather events build year-over-year. We called out the ten most expensive events, and added a bit of analysis. The graphic proved popular with the readers, and John keeps receiving requests for copies. A detail is shown below.
John added the exquisite sense of typographic design, color balance and superb editing that is typical of his data graphics. The clear structure of the graphic, comparing two consecutive time periods, is the basis for its success.
IPAD: Our iPad version (see screen grab below), took full advantage of the device’s ability to expand on any given data set. Readers can touch each of the vertical columns and a full scrollable list of all the events that happened that year appears on the left side of the screen: