Thursday, October 22, 2020

Phosphine on Venus

The announcement that phosphine (PH3) might be present in Venus' atmosphere generated a lot of skepticism, though you might not know that from the news stories about "life on Venus."

Re-analysis of the 267-GHz ALMA observations of Venus: "No statistically significant detection of phosphine."

To be precise: ALMA looked at Venus in this region because another observer had suggested phosphine might be there. Nothing in the above report deals with the other experiment at all--just ALMA's result.

But oh my... The ALMA spectrum is, of course, horribly jagged. How do you figure out what the background is for a given feature? For that matter, how do you tell what's a feature?

The upper left plot is the ALMA 15-σ signal. The other 5 are from the skeptics performing exactly the same procedure on other eyeballed "features." The black curve in each one is the result of a 12th order polynomial(*) fit to a region around, but not including, the feature region. They get, as you can easily guess, apparently quite significant signals from noise. A more conservative background estimator gives a much less significant PH3 signal.

Results: We find that the 12th-order polynomial fit to the spectral passband utilised in the published study leads to spurious results. Following their recipe, five other >10 sigma lines can be produced in absorption or emission within 60 km/s from the PH3 1-0 transition frequency by suppressing the surrounding noise. Our independent analysis shows a feature near the PH3 frequency at a ~2 sigma level, below the common threshold for statistical significance.

(*) John von Neumann said "With four parameters I can fit an elephant, and with five I can make him wiggle his trunk."

No comments: