Original story available on NYTimes.com
Original text available below:

Flawed Snowfall Data Jeopardize Climate-Change Research

February 11, 2003

When it comes to snowfall, even a flurry is likely to
provoke a blizzard of reporting. 

"People love hearing about snow," said Nolan Doesken, a
meteorologist at Colorado State University. But too often,
he and other experts say, the nonstop coverage of winter
weather masks a troubling decline in reliable snowfall

"The loss of snowfall data globally is a major concern,"
said Dr. Barry Goodison, a climatologist in Toronto. As
chairman of a scientific steering group in the World
Climate Research Program, Dr. Goodison depends on snowfall
figures to help predict climate change. But too often, he
said, the information he needs does not exist. 

In Russia, social and economic shifts have undermined data
collection across that vast and snowy country, Dr. Goodison
said. In the United States, after the National Weather
Service closed many of its weather stations in the 1990's,
snowfall measurements were left in the hands of untrained

And training is critical, meteorologists say. "Of the basic
weather parameters, snowfall is the most difficult to
measure," said Mr. Doesken, the author of a book and a
videotape used to teach measuring techniques. 

The basic method - sticking a ruler in the snow - has not
changed much in hundreds of years. But decisions have to be
made about where and when to measure. 

"If one person's measuring on the roof of his car and
another on a snowboard," Mr. Doesken said, referring to a
surface that is cleared of snow at regular intervals,
"you've got a problem," Mr. Doesken said. 

Snow reaching the ground can melt, turn to vapor or
compress snow that fell before it. If two people measure
the same snowfall, one every hour and the other at the end
of the storm, they "would come up with different numbers,"
he added. 

Even the precise location matters. Dr. David J. Nolan, a
meteorologist at the University of Miami, said: "Drifting
has to be accounted for. And think about walking around in
the backyard with a ruler. You can get large variations
just from the effects of a tree." 

For much of the 20th century, the Weather Service operated
hundreds of stations where meteorologists used established
measurement techniques. In the 1990's, the service closed
nearly half of its 300 stations, including offices in
Stratford, Conn.; Newark; and Wilmington, Del. 

"When you look at the historical weather records," said
Michael Schlacter, chief meteorologist at Weather 2000, a
forecasting company in New York, "you see this beautiful
tally of snowfall going back 40 years or more. And then in
1997 or '98, the data just stop." 

Some Weather Service offices were consolidated. The station
at the Brookhaven National Laboratory in Upton, N.Y., on
Long Island, now controls data collection for the entire
New York metropolitan region, which once had six stations.
The meteorologist in charge of the office, Michael E.
Wyllie, conceded that the approach left "gaps in snowfall

At the major airports, he said, Federal Aviation
Administration employees have continued to take readings,
"although they're not required to." 

In Central Park, zookeepers, not meteorologists, take the
measurements, according to Mr. Wyllie, who said he had
fought to keep the lone weather station in Manhattan open. 

Mr. Schlacter said the last 12 measurements Central Park
that totaled at least three inches ended in "point zero" or
"point five." 

"It makes you very suspicious that they're just rounding,"
he said. 

"We're working with them," Mr. Wyllie said about the
zookeepers. "We're doing the best we can." 

At airports, meteorologists said, F.A.A. employees may not
take measurements unless airport operations are affected.
As a result, small snowfalls are reported as "traces"
rather than actual amounts. Heavier snow may not be
measured until it stops falling, meaning that there is no
information about the progress of a storm. 

"And that isn't much use if you're trying to compare a
snowfall of 1889 with a snowfall of 2003," Mr. Doesken.
"The numbers you hear on the news are typically from
helpful spotters, who love to call in their reports. The
media have learned to tap into these people. But they
aren't the kind of measurements that can go into the
official record." 

Paul Knight, who is the state climatologist for
Pennsylvania and a meteorologist at Penn State University,
which has provided weather information to The New York
Times for 17 years, agreed. "The quality of snow reports
has certainly diminished," he said. "The Weather Service
can't go sending out their personnel to Ronkonkoma and
Passaic to see if the volunteers are measuring the snow

The loss of information has real-world effects on fields
like insurance and litigation. Mark Kramer, a meteorologist
who often appears as an expert witness, said he testified
in a slip-and-fall case in which the amount of snow on the
ground was an issue. Mr. Kramer said he could not find
snowfall figures for Manhattan for several weeks before the
accident. "The judge," he said, "was stunned." 

The Weather Service moved heavily into automation in the
90's, replacing many staffed stations with electronic
devices. "For most weather variables, that's a story with a
happy ending," Mr. Schlacter said. "When it comes to
temperature and humidity, readings were no longer impacted
by human error. And you could get the information on the
Web in real time." 

But that is not true of snowfall. "As easy as it seems,"
Mr. Doesken said, "science is not producing a quick answer
to the question of snowfall measurement." 

In Canada, where vast amounts of snow collect in remote
areas, hundreds of electronic measurement devices are used.
Typically, a device sends an ultrasonic pulse from the top
of a pole to the ground. The time it takes for the pulse to
return, adjusted according to temperature, indicates snow

"I have one of them right here," Mr. Doesken said. "It
works once the snowfall has stopped and the snow on the
ground has a firm, solid surface. If snow is still
fluttering down or if the surface is fluffy, it's almost

Dr. Goodison works for the Meteorological Service of
Canada, where he helped develop the device. Although it is
imperfect, he said, it fills a need in a country where snow
often falls in remote areas. Over time, he said, scientists
will develop formulas that will allow them to come up with
accurate snowfall numbers from imperfect readings. 

For an automated approach to work, in Mr. Schlacter's view,
it will have to simulate the actions of a human measurer.
Such a device, he said, could take decades to develop. In
the meantime, scientists hope to rely on more trained
measurers, including volunteers. 

In places like Stratford, where the Weather Service closed
its office in the 90's, "volunteers have kept the snowfall
records going," Mr. Wyllie said. 

"We've been lucky in the New York City area," he said, "to
have good people who have continued to take the
measurements for us."

Copyright 2003 The New York Times Company