Bad science journalism: Linking hurricanes and dust

Colleagues here at UW-Madison have a paper in GRL on hurricanes and dust. They showed that

...mean dust coverage and tropical cyclone activity are strongly (inversely) correlated over the tropical North Atlantic...

That is, higher hurricane activity occurs when there’s little dust in the atmosphere, and vice versa.

The research has been a hit with the media, but the reporting is not always accurate. Here’s the BBC’s headline: Dust 'affects hurricane activity'. Eh? Dust has been proven to affect hurricane activity? Not so. And even text within the article says as much:

"What we don't know is whether the dust affects the hurricanes directly, or whether both [dust and hurricanes] are responding to the same large-scale atmospheric changes around the tropical Atlantic," said Dr Foley. "That's what future research needs to find out."

Whoever chose BBC’s title did some bad science journalism. The blog SmartEconomy does the same: Dust dampens Hurricanes

Most articles appear to be transmitting a more accurate message: dust “may” dampen hurricane activity. For example: Study suggests African dust may quell hurricanes in Atlantic; Study: Dust might dampen hurricane fury; African Dust Storms Stifle Hurricanes, Study Suggests; Dust may weaken Atlantic hurricanes; Do Africa's dust storms really quell hurricanes?.

The use of "may" can be traced back to the UW-Madison press release: Dust may dampen hurricane fury.

Furthermore, when media articles were long or short, it seemed popular to mention a "robust link" between hurricanes and dust. For example:

The lead researcher at the University of Wisconsin-Madison says he and his colleagues looked at more than 20 years of satellite images, and found a "robust link" between tropical systems and the amount of African dust blown out over the Atlantic.

And again, from the BBC:

US researchers have discovered a link between Atlantic hurricane activity and thick clouds of dust that periodically rise up from the Sahara Desert.

An image this creates is of a causal link between dust and hurricanes, but this is wrong. The proven link is purely statistical. All that they know is that there is a link, but they don’t know what the link is. From the research article itself:

While no direct causality has been established here, our analysis suggests the variability in dust (and variability in the presence of the Saharan Air Layer) is strongly linked to changes in North Atlantic tropical cyclone activity. Whether there is a direct or indirect link remains elusive, but at the very least a strong argument is made for new field observations and modeling experiments designed to further explain and understand the observed relationships between the Saharan Air Layer, African dust, and tropical cyclone activity over the North Atlantic.

"Correlation" is used by what appears to be a press release from the American Geophysical Union (AGU), the organisation that published the research. The blog Muck and Mystery does the same. And that’s all we have to go on for now until more research says otherwise.