TPT January 2016

76 / 106

74 January 2016
Global Marketplace
In brief . . .
› Benteler Steel/Tube has announced the start of operations
at its tube rolling mill at the Port of Caddo-Bossier in
Louisiana, USA, with an initial workforce of 350.
As reported by WorldNow (4 September), this marks the
completion of Phase I of a project that will eventually employ
675 workers.
The location, on the Red River Waterway just south of
Shreveport, offers ready access to barge, rail, motor freight
and air transport.
For Benteler, which makes seamless hot-rolled steel tubes and
seamless cold-drawn precision steel tubes, the $975 million
project is the first of its kind outside Germany.
In November 2014 – on the campus of neighbouring Bossier
Parish Community College – the company opened the Center
for Advanced Manufacturing and Engineering Technology, a
$21 million, 65,000ft2 manufacturing training facility.
Duelling weather models
Having invested heavily in
computing power, Europe
outshines the US in storm
forecasting
“Maybe it will be Joaquin’s false alarm, not another Hurricane
Sandy, that gets America to make the GFS great again.”
The Joaquin referenced by Nate Cohn of the New York
Times was a 2015 hurricane that fizzled out; Sandy was
a 2012 hurricane that, famously, did not. The GFS is the
Global Forecast System, which is run by the US National
Weather Service. Mr Cohn has serious reservations about its
effectiveness.
Writing in the watch-and-wait period that is a feature of every
Caribbean storm system, the Times’s “Upshot” blogger called
his shots. Hurricane Joaquin would yield one clear winner: the
model from the European Centre for Medium-Range Weather
Forecasts (ECMWF) – or simply, the European model – which
consistently forecast that Joaquin would head off to sea.
It did, avoiding the direct hit from which, post-Sandy, the East
Coast of the US has not yet fully recovered.
The ECMWF forecast this time – the mirror image of what
happened with Hurricane Sandy – would not mark the first
instance of Europeans leading the pack.
In 2015, the US model and others called for Joaquin to turn
left. The European model dissented.
Wrote Mr Cohn, “It’s a familiar story for meteorologists who
have been calling for vast and attainable improvements in
American weather forecasting for years.” (“Hurricane Joaquin
Forecast: Why US Weather Model Has Fallen Behind,”
2 October)
In early 2013, the European model had nearly ten times
the raw computing capacity of the GFS. Mr Cohn noted the
“obvious and irrefutable” cumulative effect of this and other
problems with the GFS: the inferiority of the American system
was playing out in high-profile cases.
After Hurricane Sandy, Congress gave the National Weather
Service the money for more powerful computers. In January
2015 the National Oceanic and Atmospheric Administration
announced that it had increased computing capacity and
begun running an upgraded model with higher resolution.
The improved system in fact scored an early victory in the
matter of a blizzard that largely bypassed New York City. But
in the view of experts the upgraded GFS remains second-rate
or worse.
“The GFS is still quite inferior” to European and British
models, Cliff Mass, a professor of atmospheric sciences at the
University of Washington, told the Times.
Steady data collection vs ‘snapshots’
Additional upgrades – which will yield a nearly tenfold increase
in computing capacity – will help, but not enough.
According to Mr Mass the problems of the GFS run deeper,
all the way down to the description and modelling of the basic
physics of radiation, clouds, precipitation and turbulence.
Data assimilation is the process of taking all available data and
building an initial description of the atmosphere. The model
runs from that. But a perfect model of a “wrong” atmosphere
will produce a wrong answer (ie a flawed weather forecast).
“It is clear that our initialisations are inferior,” said Mr Mass,
who believes that flawed initialisation was probably at play in
the GFS forecast for Joaquin.
“That’s the real problem. [The Europeans] have a lot more
people and have taken a more sophisticated approach . . .
There’s a subtlety that the European centre is getting right that
we’re not.”
› Specifically, the European model relies on many more
observations than does the GFS – including satellite
measurements of radiation from clouds, crucial in the absence
of land-based observations. It also assimilates the data over
time, starting with the weather heading into the forecast period
and monitoring its evolution.
The superiority of the European model is scarcely to be
wondered at, since the supercomputer complex at ECMWF
headquarters in Reading, England, is linked to the computer
systems of the national weather services of 21 European
member-nations and 13 cooperating states. Its archive of
numerical weather prediction data is the world’s largest.
› The GFS is making strides in the area of data assimilation,
Mr Mass acknowledged to “The Upshot”. He added –
unnecessarily – that the US model has a long way to go.