You probably never expected to see an article here with a title like this. Sure, it is a little provocative… like attending a meteorology class taught by “Stormy” Daniels. But anyway, it’s hurricane season again, and time for a timely rant.
So what’s the bee in my bonnet now? It is how our National Hurricane Center (NHC) chooses to depict the forecasted paths of tropical storms, and by this I mean their stupid hurricane cone (click to zoom). Before we go on, please take a look at the top of the figure and read the disclaimer that the NHC feels obliged to attach to their diagrams:
“Note: The cone contains the probable path of the storm center but does not show the size of the storm. Hazardous conditions can occur outside of the cone.“
It’s a bad sign when the NHC (or any other arm of the government) has to warn you not to believe what you see. Because, disclaimer or no, even the most savvy of us cannot help but view that white cone-shaped area as depicting how the storm will spread as it progresses. After all, our experience is that storms do dissipate as they move overland. But the NHC tech-nerds have sowed this confusion themselves, by trying to present statistical data on a geographic background using ambiguous design elements.
Here is what the NHC intends to present: (1) the expected path and arrival times of the center of the storm; and (2) the range of possible paths and arrival times of the center of the storm. The NHC does not intend to depict the breadth of the storm — but that’s the unintended effect of its failure to properly use the elements of point, line and area.
In the NHC diagrams, black dots mark the most likely day-to-day path of a storm’s center; whereas the white cone is meant to encompass other possible paths and timelines. I find that the semicircular end of the cone, and the fact that the area inside the cone is white on a gray background, is largely responsible for the diagram’s ambiguity. NHC may want you to interpret the head of the cone as showing the uncertainty in the storm’s arrival time, but your eyes and mine see it as the geographic extent of the storm front, being that hurricanes are, by their nature, circular.
This, plainly stated, is a meteorological graphics fail.
As one who doesn’t need much coaxing to put my mouth where my mind is, I here offer my ideas — to anyone listening at the NHC — on how to depict the time/space predictions and uncertainties of hurricane path forecasts, with far fewer ambiguities. Please refer to my easy-to-read diagram, below:
The first thing to note in my diagram is that color is permitted. The NHC seems to live in a black-and-white world — they use color to show coastal storm-surge levels and that’s it. In my diagram, the expected path of a storm having hurricane-force winds (75+ mph) is shown in RED. The color changes to ORANGE when the winds fall to tropical-storm force (39-74 mph) and then BLACK for lower-strength winds. This use of color provides an immediate idea how the storm is expected to weaken, something the NHC diagram lacks.
The expected arrival of the storm center is shown by solid black dots and day/time labels, just as the NHC uses. But my diagram differs in how arrival time uncertainty is depicted; instead of cones and circles, I display “time bars” along the expected storm track to mark the 90% probability window.
Finally, to show geographic uncertainty in the predicted storm path, my diagram replaces the white cone with dashed lines. Each of the dashed lines is clearly labeled 5% to indicate the chance that the storm center could track past that line. I also place a solid gray dot on each dashed line to reinforce the notion that the dashed lines are alternate possible paths for the storm’s center, rather than representing the outer reaches of the storm.
So I say, NHC, save the disclaimers for Big Pharma TV ads and the cones for Dairy Queen. If you want to more clearly communicate big-wind warnings, you need to take off your 60s nerd specs and put on your Elton Johns.
[Editor’s Note: Readers who just can’t get enough of articles that call out bad design may enjoy this one, Bad Signs, published here last year.]