Monday, November 11, 2013

New Yorker article fails to fact-check

A recent article in the New Yorker magazine features a profile of a the Climate Company, which offers individualized weather prediction services to farmers.  Unfortunately, the article makes a number of unfounded or vague assertions, and in some places is so boosterish of the new company that it veers into puff piece journalism.

A balanced review of the company's claims would better serve readers, and a more in-depth review of the science might help explain why.

For example, the article by Michael Specter, claims that

"If you are trying to decide whether to take an umbrella to work, the National Weather Service provides the kind of information you need. But the data, often taken from readings at local airports, are nearly useless for anyone who needs to gauge constantly changing conditions in the soil and the atmosphere."

But the NWS does offer zip-code specific weather readings and predictions.   While it would be great to have even better location-specific data, such a monitoring system is yet to be implemented.

"One of the company’s principal sources is Nexrad, or Next Generation Radar, a network of a hundred and fifty-nine Doppler radar stations operated by the National Weather Service. Using data from the system, the Climate Corporation creates moisture and precipitation maps so precise that in some cases a farmer can determine whether the field on one side of a road is wetter than the field on the other side. "

All private companies use NWS radar and satellite information, and are limited by the resolution of this data.  The highest resolution data available is 4 km grid boxes.  

According to local meteorologists, it is not possible to distinguish accumulating precipitation at smaller scales without installing individual weather stations on either side of the road.

"Soil type and quality can vary widely within a county, and even within a single farm field."

This quote is used to imply that the Climate Company has such intra-field soil data, but cannot account for every possible difference in soil texture.  It uses NRCS soil survey data compiled in the 1960's and 1970's for every county in the U.S.  

On their webpage, requests farmers fill in their specific soil type.  There is no high-tech substitute for good old-fashioned soil testing.  

The article also features such gee-whiz promotional quotes as "the algorithm divides the country into nearly half a million plots, then generates ten thousand daily weather scenarios for each of them... It matched that information with reports from two million locations that the National Weather Service scans regularly with Doppler radar."

Again, I spoke to several practicing meteorologists who were not sure how these absurdly large and contradictory numbers were computed.  Their best guess was that is counting the same location more than once, for each radar beam, or that they are counting different layers in the atmosphere as different "locations."  

Alternately, these discrepancies may be misquotes on the part of the author that were not picked up by New Yorker fact checkers.   

No comments: