It was one of those “big, if true” stories. In September, scientists reported that Venus’ atmosphere seems to be laced with phosphine, a possible sign of life.
Now there’s increasing emphasis on the “if.” As scientists take fresh looks at the data behind the Venus announcement, and add other datasets to the mix, the original claim of inexplicable amounts of phosphine is being called into doubt. And that’s a good thing, many scientists say.
“It’s exactly how science should work,” says planetary scientist Paul Byrne of North Carolina State University in Raleigh, who studies Venus but was not involved in any of the phosphine papers. “It’s too early to say one way or the other what this detection means for Venus.”
The big claim
On September 14, astronomer Jane Greaves of Cardiff University in Wales and colleagues reported that they had seen signs of phosphine in Venus’ clouds using two different telescopes (SN: 9/14/20). The phosphine seemed to be too abundant to exist without some kind of source replenishing it. That source could be strange microbes living in the clouds, or some weird unknown Venusian chemistry, the team said.
Greaves and colleagues first spotted phosphine with the James Clerk Maxwell Telescope in Hawaii and followed up with the powerful ALMA telescope array in Chile. But those ALMA data, and particularly the way they were handled, are now being called into question.
The key Venus observations were spectra, or plots of the light coming from the planet in a range of wavelengths. Different molecules block or absorb light at specific wavelengths, so searching for dips in a spectrum can reveal the chemicals in a planet’s atmosphere.
Phosphine showed up as a dip in Venus’ spectrum at about 1.12 millimeters, a wavelength of light that the molecule was thought to be absorbing. If Venus’ spectrum could be drawn as a straight line across all wavelengths of light, phosphine would make a deep valley at that wavelength.
But real data are never that easy to read. In real life, other sources — from Earth’s atmosphere to the inner workings of the telescope itself — introduce wiggles, or “noise,” into that nice straight line. The bigger the wiggles, the less scientists believe that the dips represent interesting molecules. Any particular dip might instead be just a random, extra-large wiggle.
That problem gets even worse when looking at a bright object such as Venus with a powerful telescope like ALMA, says Martin Cordiner, an astrochemist at NASA’s Goddard Space Flight Center in Greenbelt, Md. Cordiner uses ALMA to observe other objects in the solar system, like Saturn’s moon Titan, but was not involved in the Venus work.
“The reason those bumps and wiggles are here at all is because of the intrinsic brightness of Venus, which makes it difficult to get a reliable measurement,” Cordiner says. “You could think of it as being dazzled by a bright light: If there’s a bright light in your vision, then your ability to pick out fainter details becomes diminished.”
So astronomers do a few different things to smooth out the data and let real signals shine through. One strategy is to write an equation that describes the wiggles caused by the noise. Scientists can then subtract that equation from the data to highlight the signal they’re interested in, like fuzzing out the background of a photo to let a portrait subject pop. That’s a standard practice, says Cordiner.
But it’s possible to write an equation that fits the noise too well. The simplest equation one could use is just a straight line, also known as a first-order polynomial, described by the equation y=mx+b. A second-order polynomial adds a term with x squared, third-order with x cubed, and so on.
Greaves and colleagues used a twelfth-order polynomial, or an equation with twelve terms (plus a constant, the +b in the equation), to describe the noise in their ALMA data.
“That was a red flag that this needed to be looked at in more detail, and that the results of that polynomial fitting could be untrustworthy,” says Cordiner. Going all the way out to the power of 12 could mean a researcher subtracts more noise than is truly random, allowing them to find things in the data that aren’t really there.
To see if the researchers were a little overzealous in their polynomial fitting, astrophysicist Ignas Snellen, of Leiden University in the Netherlands, and colleagues reapplied the same noise reduction recipe to the ALMA data on Venus and found no statistically significant sign of phosphine, they report in a paper posted at arXiv.org on October 19.
Then the researchers tried the same noise filtering on other parts of Venus’ spectrum, where no interesting molecules should be found. They found five different signals of molecules that aren’t really there.
“Our analysis … shows that at least a handful of spurious features can be obtained with their method, and therefore [we] conclude that the presented analysis does not provide a solid basis to infer the presence of [phosphine] in the Venus atmosphere,” the team wrote.
Source: Science News